Lab7.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# TensorBoard進階技巧"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## tf.summary\n",
    "\n",
    "tf.summary是TensorFlow提供TensorBoard低階API指令，主要是用來紀錄log檔。\n",
    "以下整理了幾個常用功能：\n",
    "- tf.scalar：儲存顯示如損失、指標或學習率等的變化趨勢。\n",
    "- tf.image：儲存顯示影像。\n",
    "- tf.audio：儲存顯示可播放的音頻。\n",
    "- tf.histogram：儲存顯示模型權重。\n",
    "- tf.text：儲存顯示一段文字。\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 創建TensorBoard log檔"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "summary_writer = tf.summary.create_file_writer('lab7-logs-summary')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Scalar"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 在0~2π之間產生100個點\n",
    "x = np.linspace(0, 2 * np.pi , 100)\n",
    "# 將100個點帶入sin函數中\n",
    "data = np.sin(x)\n",
    "with summary_writer.as_default():  # summary_writer作為預設寫入的紀錄檔\n",
    "    for i, y in enumerate(data):\n",
    "        tf.summary.scalar('sin', y, step=i)  # 存入數值(y為數值，i為時間軸)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Image"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "1. 儲存一張影像在紀錄檔中並顯示"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f64384ffb70>"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQEAAAD8CAYAAAB3lxGOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjAsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+17YcXAAAgAElEQVR4nOy9TcyuWXYdtPY+73erqqvbtH/bju38DKygSAhETMIgAyILFJg4o4hMsFCknsCIAfKMKSMGTBAeRJiBA5lYsYQDWJ4wCAkOCJCNbdw47nS32+60q+J09U/Vfc/eDNZa+zzfrXur2m5XfKO+T3fd+933e9/nPc85++y99to/J7obr65X16vr2/fKP+4BvLpeXa+uP97rlRJ4db26vs2vV0rg1fXq+ja/XimBV9er69v8eqUEXl2vrm/z65USeHW9ur7Nr49MCUTEX4mIX4+Iz0TET35U3/PqenW9ur61Kz6KPIGIWAD+XwD/NoDPA/glAH+9u/+fP/Ive3W9ul5d39L1USGBvwDgM939m939HoD/DsCPf0Tf9ep6db26voXr9hHd9wcBfO7y788D+IsvevObn/ye/s4f+FMIAMYlrT/i8r7WP8JvCr+/AcS8dz57/fCzFz9y+QCe+bLLz3F9Md7347Pjjmc+Ord63vd82PD0w7kPX+jrMz/7pR94ww986Zu4zqf6eXN8He8HfMELf/W+X7x/0q7zGXGZYX32/fd+/rpdb9bPGcyL5uaxaPwBFvVFY5pXGoHgvM4tz0O1JnUkPZ5zmz73eXTvAH771/6PL3f39z47ko9KCXzoFRGfBvBpAPjO7/9h/Kc/8/cQFUA1OgLV0AMDyAAK2MkJX00pqy40gMxERqM3AASwgC5gRQNIdDUigEIAVUADt5UoTXIDqGpOaFrJBLobay0uSjc6mu8OIDtQXDI0CoUGIoHm77yAgUBIMHuf1yuArfE/ZGLpGctjrUZIyOd20cgFbADYUgSN2Qw7GhGJqEIAyJWoKnRxs0YGIgJVxZsk/7rFGVd3zO6NCCSAQnM8GeiSKkqguvGQCwCw7xtcqgQC2AFkNce+NhKBp5sTfAuJaATHlsE57EZiobok98Up7eZ9O4ANVAId3CyrgQWgQnNQxc1SjRWBXJz82o3MxZXSe7HvCARy8fUCP5ednNbixGUGshO1NudoByIT3Q0JKgKJQiMjEA08lUysCGQHwouo6Y1V/FQXyvJefF8Ef4cAEo1qzldEortQ0bxncEpWBwqbi0yhRiU/U3L3I4D/5N94/bPP24sflTvwBQA/fPn3D+m1ubr7p7r7R7v7R9/85PcCFWhpwZImzKCeLQRCI43L71YEFjjpVUABqGgUvGG1gfgBoKlgkIHqRldzIbXpYyaRH4oIdBe1ffR8L7cGNwSVRyA6hVz4OS4aN+pYjtSOBWYTWGj82auZGeTjn4OLWlubvPjb1ibubG6g5sYahBThbzyXNhFAhdPV3JA4rwOajm4pEOo5bloLZUvJXgbrsQbXAy1hTG0uzTeKc3rvjQKVH+ebFnY2fie6AOi1aOiznLjiF6DLnw0kqFRnLqKBOL+P5phD89AUNG5WrVqEhCckeBQwzpGMRgPY/cz6oJFJ2fQcdgKxOKaMQHVw3lvfzxUfAadObtQzK9c2KBZTSCH6TU2lGVTd86z5AbD4o1ICvwTgRyLiz0TEEwD/PoCfe9GbI4BoWuzORGYIzgQyE4t7jg/TVhaNJauC5mZuWaB7FSdWD94B7G50FnZs3LNpNTMQ1i4IWsFZjrK0PpKD3gCqUVVEIJnIprbPomntLlQXIogSrP27uTAjPf6ubvTWBs4Doym8gneS6QKo0PgGLv4KPJWm0FSNYkvD5QC6AvtpCV0UGs33h589Nc4eS1y1j5JBUZFFS3HxOUtQIxdRQRTQ1dga+7bBREqxG+nxecIbEMC6AUgqcOwGNnCLRZS4AxGNWwO3DqIAbfAQ3FrZWFl4iIWbkFntklLUStcelBlo7L1pnStm3uxyBhrYXO+oQGzLbMli2wAXWjqr0EgEHpBY88y8B7Kx645G6rMLYfm5AfFQ2AFsFJALETcIJEhBBjIFOqtRDey9kaAs9gp0BqFw9AWFXp2Yx9dH4g509z0i/mMA/xOI1v5md//Kiz8gaN+Es131yO/3DzH+D61AWANrlwUaS5uZFtabDNyQQZUfDVoKYCxTxvm5+8D4uFjMDk48t1qPpm0LmKyYbDqdgQDueyOJWTgGEPYt2KOklSwcS4WWBY9GNGFzoxFL2l82IppjSsNMwVd0I2Q9j0U7Sg3gHATiAhkvSrCNWiT1FeN6xfxeG6AKmWsU3iifkAWS6SIs5s98W49y1UPNXA/6sdUFNxd3jq1wy2WJgU0VdyAC5Q1QrQfui4Izl9LjGoRQI/R26+iQwQndkyDRborHzgdotKzyUbwRyd8YNYWQYTdSvmeUoDu2kAhQMm7L4wTGSHkeclaL87Q7KEMBVAW6ArfgOlf9c1YCHHf/PICf/6beLE2aK5FVZzK7kFhGtfRx5fp0BzYKLW2/YiGiEfKDAci6BHZtCngmsum3ZjdKi5uZ2JsCnPb1LGStjWsLQQMMuouBDWB7UVZS8cjYrw4EbnhIbzQqORTQQWibaPqXoLbsNhRdADjGBgAhD1SczdA94xVCpgVO+aLauLSsW3tR85ULe9cIOJWdNzYFlzsl9dzXzR/iSrhWt9u6uE2J7Y3mzR9WbEQ3G+9xU2x+j92KQKDuLYVCpFacrAFPUn10Z9C4F+fo1kQ4USlXZ5PfWXZZuG5WNIFARYLeYSGavjeEVCTF5COWUKoM0GqiqjRfE4mOIm/Q80ncAQQ20avGh+Lce97vTwtrcW6TW55rHQBW4b4LNyysEABEYG/zAdRCtTjnJdlZIW4BgdobnYG9Xrz9XoqMwasPSl8nJASJXZuW5upz8p1j0truQPexJBLCqyfkiRnTr1+aiArYGmpDX3z68RvDm5SDCYAaARSkaFwgOd+3a1MocayeYXShBwXx/onM9QjioqnAIil46c2Vh4cwhJekUMhSiqwIdbsPlzGbPw/3MfNk6xgJqqZ1UTQk8KwU0RilW92A+IpxdMtWEppLP+9xi2zRjWQmd0XKI8AdEBcOJa0IDMK4ixCDuIKuk9dqfOIYZVbV2GUYQrnoqot1PePxWlABrTMuKcUt3Z9lsjKRdcYfIELvEHKUq2RFM8/cgRULKUR83kdPbDXwgMQNVDzIginquMipaaDIHKTzouuPLTpwvYgcg/40HPySIC8+zX3f5YOLkOvAWiFr6CcG0IVbJKIIwSqoyXHkihpYLkfT4A48bSmc+95EBm2hJCLIWXggJDA38PtSm1SxAi5M69n0uwDZ+dUpopCIYbesShH2UvD6WFIxwxb6tAsSnKdMjsHophN4SohEv76BFjEZnbSogpwbmlsE9ZlHH417PwUisOIm1p1uVqNE4hZut4WqjSCuGbdrIWnxqwd6ZwGIB7oEbacJuLCjx/KLcKwq5EPz+TZEMG65dFSQuwpdjSVZCpFqGyXStgdq89kIwCuIHFMRpcgkhxA4ShiNOzbWSupYKZAQWmk0brWo81J8SRUSgV3APSiTAHDHRrfuswsrUxs9sO/83iU5pUxQaWSAZGKQTMxurLgB942WY3DYLaM5YJee5ZFqe3y9FEgAwPi/SBnp5KboaGo8w2lIOcjtemTrG4f1hxSKLCKyR1vql5ePiS8wAbYLCW607kaF2PD5jKyQw0NJHzGTm73kt4kWHIs236N7hzZCg2EfQzpA328kItSBEmx09AEcVEcBsTWsi2kcaGqElWgrom5UFGL1kJHjM4PI4DD1itpoXjkftORm9O0q6QWRj41h5DW3AUJirpGUzfjBhcrimmdw0BUTmmwpfLoQhMIrqHq41vAOIGroFKP2GBHyPXJrxJxXORR6FIDHzI9Q0exuKtVsdGp2L1AxrWAkM2leBGfNHNrNYZz9PZZXfmkoBOCQ+XxN8LsXWWEpQ87HwR2P0WA+noFH10uBBACjtZiFNATLInlCBp0CseRDb/lZzgMIx0bHWFNYLIQlNtcwfQSWbyWjj5SlKfSmI9lcsQm1EL4menFyKYCFjqXcBerW6kJGI7LlvazZHB3chPRRc8Jlg+NEheQwlluyIl9008pv+Txp7VnaDOY8khwI/UPFyuW/l2OxlFnUPi4BSa/ELddswFsytAXQYnYRVewq23PcomS9QSuXVB7zXRlYmzkaO45bxY16GxZ7iF80omlVO2ltUYW1bkDQ0qEbD7KaVYXAhl2klAWmZ8IxOOx2C/NH3ERLSj80ASTyZIm1IZ0vQsWE+ezOo7SjaDgiWpGtQm0amdWJB4gE1tpWG0UuiyS6gYXAEpdREcwr6cZt0R3JbiIDgJGNCwqDuK6UC2Ly93nXS6QETqZUKzElI3CXMC+QtNvEgyR0pPXK2lo+ZXkBIStmqHhL+aPAQ1xj1jgET1lpEGKZVxjfE9xUVYW1FoyejRqG9QVRQWQhemH3Rpf4BxQXGE44wYlcRE68nvMCxdAZk4BItC3YuZphPcbjS+EpCn4Kaho1JaQwM/VcG96AVy7Fk9kR6LuQy43feysAyhOAyds8YyyNfQkRRY2ewVIiUSdDvYbaKTza3ihJxVtJniWT1q5k7SMSVc1opTZejc+fyinhPLT4n6qiQutmgk4RQdFnpqWv2uJZlJ+iNckWjhPPEXlTjF8CVI24BUJh5YjALagw7V5aRpfG1rWljBLZiShu6q4a/mf3ZjJQLSYNyQDUFtEdCw5pbEH+RCqniqaoopQ492Jm8OVQAi3fZQTS4RNq09QmoyHJseaw1px/H4KMBkYZfEhaX1kjW9S55Apo28mC2Ce1hArq9fs/n/qeiSo4bg7C2W6gy7F5WQxviHkOZfshJtFnFKM2y9UF8aY4QPf9cG/mwm7TVTnKSjpkJk0GRegHVraz9+RCdXJ8pZj1TGGc/0CAI6t2cbNg9Ms3ppTfowgFFCOHYKwyE+nSxIwPkEUP8htGhdkAQ2nF1MKkBUUUSq5BrkQqj8MQ3DkfVGTknMijxkk6spGpQivis8LuynGPwi5hgNwUGog1IVtJjRAHF6gpSEDn5Bl4WUwapiaaSh+HRolQxKrtGIzmJYJinsyLrpdCCTDJh0JBX1xenpjmlbSOtTdiybfKmLAeMkioSZHEA53c1K83jDRIVtnP2mJfuYm5pbzfGk67lTUI01B9fGOhimgLMK1DrmvaKR+QYaEteKy0n05aIJNXFz+0/TwZJEc7sIv8Q0IhSkDhvx6ocoOQUNCy+Fm9GeVmi5jUPQrcWnZ3DJv3MeMCLMguZDFhpZRW2+I1nEy1VpJv0H0yU2tzEAeAi4Ki4Hsz0WLr9Sq5U9YifM/MdxH638wBBMOBtYC4AcCmEUHKHdDmQpIEbbsv2khbaONi1VOEavWib97EUJmJVSD5uo/67E3l5k0eUFo7+FnH7CMTO0qJU4UVSaMjmcla44rS4ZIiRyjaU6M4o0QoyjV+iqZLBuux/cL991IoAUCWlbMGZwEFTPTRKujxL369SRNCLoaPjl83gi+reiaRvzOEtYWy+wClxtpHPglD2lQNdBaJm6IF3WEiUAJbhxjMNn9L+EdmvZET9RakkxglcsKNuzj+ZQWBllbnfAWT+4dJ7wtpaLTaMHMufgNxFE/3Y1SkuRlFOwQer601yot/RL21xl8ufR9GcDHczQzdyTwdFyvF57Tl86tUeJfPZhyOQZaU/rzz6L0B65B7yt23XNitAOTbB9fCz9pdF8pN8uQkJZto/a4bQNLd2RMi5djT9yiMdd9eszN5yDJhe8F25odwfYQT+vaeyM5JdMp5b1xqZZw5+vzrJVECDWgjr1TqKpgS2SvwtDaTJRa5UG68Q4oBFKRA0moUsKtZPBJgSExZbSoTAT9Fq8n1P/lXkYpFT3YclKkXuN858Qu0zmvlhPS6ucjtYg4EYaisdVQD8jtnY5UU0rIyS3TRV49UyqkUUDkMmevAz2aoaq+Uh9+Izfnx9xSkQMSBJUiK7WAcI1vpvhKUE6uv435ZgXjzVyFjMXANEaOc1HEVfL/dJMaWYGuXvTe5eG1EwV2+5Ucf164vcsGwLTN2GH7zREY2npYiA/tG5bKkWFAnrIqDeBpM1zUZhzYSFQqJQAdzO4cnApN+s4FKpefWUj7LVKDRXNX1+6DQ8zFYRKp6fkgn2J0Jx5TiQNRqrKYcM1GMcSjns9VNSKeKEWitgQnS510vTYiQWjJFuBiSeQPQgoaoQPtxk4lvf91YQRqR0MtWCQAcw8XE76VD5p5ejNZihxaCqKTGKqQCUyWruZRlF6DSCVysHVSjFnxGRiwU2x+DeqIUDKgDQrfyc/l+7bJjDVQVyboKwkGkrd0h5mh9lUojH9czG3EUwMkatAXSd3U7DQf+8o6a7xnLqo006ELym1e+4ZrjIIWaXgtbXY9OWX2T9ho4PILWh5sjELUQK4Hc4l80PihExsWZm0+C1iWJzHF/XIwFQ3kJ6xvLTiKECvJYbZFGKSUciut7cXOtEwKUMnARXLQMhQzgVgIXM0sxipWk+bXmRXIbwB38DzObZ01fdL0USIAQJ7DveyBTrhiLFKUHDsJY/rME45isk9a6B0ENrKXvK80/YSCF/BxelLtgeCdlirU4gaVKt8yFaFW+aVgPYFJSrFOwAtBKODPNnII5BL+vq7HWDdmNezBdNNbJS6jaDIuuNanH+94Xa8JsJydAdoZBOL97Fy1HyGo2sG70ubMtSP0+6w2lNfO7RDANagLu6TRYAt4U2edCIZl4fs/FJSP85vh314TgohtP5FLw9UaoIGmLLGPmI8NiIRY8ciOisfYN6CCKihwFa9DOSkVFRTq5BrYInYeY7Fakx/JmxU5UlxG4lzg8NFanci5CUQ9Cim6S2UNxbusBugssH954uCTAofm9DmqGkuC2ogHOzoDT4OUetMabDcSm929VWnoPPkAPvDRIIOKEtEz+lLKuDoyTXy1hshZN2DJh4q+OjTvcw02lTCspgAPx4lhOb64h6SQgzZh4t1jYkCBtwcFu5mkr7hxJj7/HKmgTBeAMXaLgg2wCgY0+PvIzCq2c0prmPMBwnaIJTl6xJW3hc6OkAobVNoJykVV4XiNOGjDoVhy+xnPMT9T19SqlT+s+DUGZYwXnTyVJEZCcZ7f2HeRbG05Y8mc7L7yHEJXRAaSYGPUQseocE4ULnVE5vgViNnt7y/o9HrrmxSiRH0lEHIs/4/Z7ZlZ5Zerz7RWClGQPV2IZNIGJPpmuYhD4++gp0fbreRmjeYGAcg1mzzz/eimQAOAwEAkewj+Fx2pU4+TYZ96kqQ3cATlCAESuJE64qJkjHuH3yfojpiDpmkMfUMp7EpZFBPq+jpUO+u8Pyc2+sekfKxQJkWAr6d4sKaXOxsCXzhkvome8qwMo8wja4BLojABqy2+l77ydWttA7rZMK98hFD0gc//EVhjgvAUtWEa4ekm/t1LkO2mRniKCMW26X0wi2nDhChXNUgVjC8IzctJTpQm7Ewjcli0sv+MuGRDdQaKtMeNEqFa/YzI0c9+05lSQKwI9oUD2nPB6oZypeacyqQvaDBcKsYJ53LBgIVZ1MUoTQGyFtDPE89AorAxg65nXcVMbOFWgF0XNkLNQZHDOKLMUwMhnlRbTh9MljJrOW3NvuHT7ZlegAaeu5QdAgZdDCbS2csicwxBGYtPH8jL5QcrgJuDkOgBNeDUFxbXmDNGBPp/u7oKZTIOvUmKHfTX6cobJzlOAFJEtZCTYMaiBji2/WJtRvqb9OW9QM8MTDtTCs/qu0bGZBBRUYtJjSpqK8VuZDagwUuUlW6xnE5Q22STWtC29/EobJYUqz5oM+0IUhQVyGhi/lPlJxeYhOJYnAqeSUAqlxGmcsGkjlKHZQ8py3KF7GFZvj0HLeSo7Q++/IqoYeUpT5tFIz4nnEzioU25WaI5bJKjzK9xhyLIaXXoPF4Y5GydE3XJNZW4k23qupADcnZ0q+aYCOC7Ts3bb6AfNDW200MEo1rV0Hgq1GlB4DV50vRxKICg8fEpD3kOCoULIMrCwgCxGPsTUGwpZhwxZtXteC/lsVBgiV5JFG4QenPyn8je9acJmaeLxTkDZVE53ABm4w4VKB2aSQzjk31idZZ+df7Qz7pDo3ehafP7N741S26oAk1iCcenVwI56dG/CGGao3auQ2VhdCj3dOF/JbAlbs2orUY774baMr6YMu0NWDkIGxYBmVCCWSNFuRFC4h7CDhL39+yW0BLjCMOLkXhhWG9IjWqnLavwBFd+UcjHAug3OZ0wHoqn/b4xhuYWjJX2UGGoQYolcG38qGKrtJUWmEt5Yi9l9XWN4RtGuGE6JHAXHcleq+MMWuuhlyMY1SKDv/PIqFiHbLW0wezErpp9MZrKobkmJNonhFrLMRW6opHTXB0QHXg4lgOOzuKecNSe1nSOiMYvD0NQ+71GA1Y0rRPHpM5oY5RBAyKJdoGHWF1wUpl0WbqGS3oyB1A35dRXoTCi6Rquj5Bov3E0+3TSBCEuYfTsAgmuOT4V+NzFgWY+rRi+HIoVICpwO6sKE+2iYpBqb1FQmu21ZxX0YJcjinQYvQjLiCdwQhRZXasv+uZ/LoL/t2oU8mGnBMpukYSG/KC+90yG19rhBxbY0Xhdiwd/aQhPiADgMrzPlwJmHztnHZU67jvvgTEwrq+7Lk4kbgvmFJQh/lWErzEhAXQUGdYSQCVy+Wuhe46bRej/esAFmQhrF2g2OCXUdgtf1HCH4mCuVLv4stjjXS6MEkHeMTxremDkQCxG4K54cm/DOMLCskZu+5ymzDbgbSzfwFJt9AhC4iaSrOEVB3Wxl1UVLVHdJSgWcp2034kGFOAUyxBRBWqc0UdctYyfYX6UeCY5OQLuhBEWLfqrDaVdLoGcjHARm8RuEugi5IjXppCslrPDvhaz4aTZbbcb7C4V8uF3G0cpfbzHutOnuJbA0H0+X1uHOL+1F5HMDiGoiNODE7g2GRkPPsc5GB1226sZ9KbcgEqtd56CGopt5FoWnRDS9lNkozmf4eP5YmrAKADfJSGm94kD+iJ6o4Moz59UsAmqtXxuddoj9Z9SoK+HMQIVEgLqjb0YGShhLygBRlQqAhE6nQUzjRJMCypxltIS1JuJlcl3KoJt1EGlORQh35vdldwdss6NVRHFgvS0MBSFm44eg1GC3AB5ZWRE5suEwzrTQudLvOjWqBNZtRXWFvtyW4QyZ7kkoeSni0UQHCAHt4kQfVhtFaDhNM1tdkeY7Y55n3BsAdooHJZinOE8uIgnyYMx3Y+YlM4HNFNIuukx4wGRaziNI+ThhCVUaly2tfV+G2M5z63dxFObYfRlgswyYn+efygEgenMDlVZvPnSMlQyVWdqNyDiryenL+UxVIW4MMyIcGcLIg/01bpYc//mQeo3OM6+Arb54I4CNYDjJcBIUNzC/M4PJ0az6O5WmkQfd4DL9sEsbFMqsk6Yumzhu1jyHCwqGb4gzry+7EiDMuhTUeLtrY6Us0G0J8nkzB2PNaa3szbqbZZndYOYMLUWKXHTiCLW8Y8dKPhJfgICEqADkKIZ0FEHW4KqIiOiD7b0QU+RjOvK21LRDhSlZevagNY8C/U+tZSTQ4iwMYbc6FOUqVN+xcMOKJSvLij4jIocaMxZdHLVuM6FKttxJTzUwPUMkI5TlpzVaViIWqVKhV/VETSY8uGmZyBJOocPZ2C0lH0AoI4/Wf7EtWzVMVMQS9A0AdzsfSxYXE6lotFj5mHh5gv5xQKgMsqrweC1SbkPHxiR2cDJDjdnIE9Vu1iToM9GL87KMsmSotIHNt4RyHbCAVUroCWDtJaSkSQm5LkIpFUp0utnYtBQllJoOJkplI8y/ZLDdO9QJuQq39ZJXEQYuwnPMBdwLwOwtIZatRSlioHt4wsMwWrBoiB9g2GT7+PNdCnPZPMSx6XHxt3xNbru5izipsrZxtKIQAXcsX6RDWHYV9Fh6xm2tIMTi+xsNpIttIscal8lLxDDoAxGMfJpW3WnSkbj03JOymDDpBRBA42gMI09ruQel9VgmONsVNrUNb9SY2gijAYfLADPZgunXBiJJ0e649ISEUeJlhAGB3hDO61nfmJXR++MghkdyGMqp0OfcrHSguYhmpHgOff7KzE9PSArIRa4vYtQJpx+OFzLcl5EKi7RghYieeT0AS5WgQiqcahOVUnDK3Tih4fdfL4USmCtwynfba8UN23laeO3d49+GNsK9XBpaiFy0BmbjPXFalG2iqcUvAifRJpSldyF5TiotTiLNbuRtPXJP/F67HLjL112p3Pca4Y00kcXvT8HKm5l4aj9M7X4RISwz0/uOFZfd10eiXe4asuhNLULSS3zLKau2k4VBBCVeInDSroetb8JZj7tCaEIcAoUZSpO2GriGVy/JX8phwDQH5bgyuSG42ZW3gRhXu7tIIi6ik9IzZ1M2zPSje3oIaBvNGuuBJV3ayGpKuu+UIyzmYjjD86ZIT92V92Bs3nyEBNTYpjkOiJVPtSOTdt6oOWQGuZjfkUDnhs/fMCptVh5xg2scLnluMCR919+IVrYq2+fBxK5DCi+4Xg4l0Lg0YHBoiDuXGW5giSSOFmQduXPOOEEZN05a5Pij08I85DI0G3y26rt3DvADACoYAAj3DALMDXT7RCLD7QnszcYDAo8Og0nFdfd9EpDc7IIlqhRupMfsvgryf6MGpVy7EWfeRodIv0m4KWSxznjcpqrKcX7DSM59r+OlW/FWE54vHAafKRWC0RfiSaZsQnp8fgozNE8IICugtpBSyoFu5tIb7IUGEUhU7FFP9r70RCjXyGruw+6FZGUtZTTy00QD7XMCjA1O+Xg15hQqpn8TlUxegdEiGiWuwYiDbdVNJjqXfSP6xnHL7UvVlKzUfUf2i2RlMxq1J2VesilFSFdPisFZqyEDon6Ffp4WAR6uxckrO/T4ejnShuP4okMEAlb7spyY8FFkATo8g3DaFrjZIwCWTwFPk2kNkMHNuad30lj0+XbBxmAbsBGC7vlZ7IJHKzeiMZzMs1DQ0DJd2umQkeomXCoAACAASURBVAgfYX/6yxBywKg690+01e/59xW/Mye9FfV4nCjSA//5EXMWSiOmDhro7A0DHER0YH9M6AwoQXimevd5AKBJfAEYeA/9ytNsNWwtVArThrsjt5Ul2XhHTtA1peYY8TeIci7GeeZTgHby5+YZNVeRx03wGJcPxElMw5cMRg0GfksuWNKeKCWQMVsPbIAax10I2OjFRKf8nVQmR9ayGbr0oMNPq+hHZtDFsxIOr7WUbyaeDTter5cDCeDA4xMvdXqtfTS+br8ywPr4SLWFutEaLLfHphMMQMQKLtAuRcQJdXBxFHoU8jDhF3B5qDasN10EmKdwoDC5jaN8DgAj6cmTeS6/FzRf7obcgft+nNzxVJWLDkJ6MX33MltvN+deuK0bWoU2LUtBwZEiSTceOWHWbifQHORAIyToH6OKKKgkMmZDMjml5+y/brlvev6bRn3sEc+UaDAEmcpyc9FNR6FdoSWadaFPVWSflmLcl04QOkU7KfTkwhyvG7sJh60AJqlHoUxoM1MwSaiaEGWkBCIP6Wp1XLosdQOxsTRmlyM73wO1GboUoV27AVlthyA5VFdXisQsTLTrvU0Juq0T3o5MldNzfpZkjhzQi1EA8C0qgYj4LQBfAZv33Lv7RyPiuwD89wD+NIDfAvDXuvvtD70Xrj5bjM90/Gxu6g6IObb2nEwfWNTLlsP3le9nS8z6f5Npg4oH3YYXGIBTyOJRuEuaXLD5uDAaQ4jsq0uySh2rChzewKm8Rjw+f5FhRCCXFFQMFuYYzXXYyqUViUJsMxv8KWz1FYY8Ckqxc4jXgAlV3V8W7KKHuZG0Gce/bz3fJcV2FHlO4rRe1xqjhPa4cclD6Hfd2lxWI9rwEIlqaZ9BCf6OGxiTxMNftxq01CV77syOrSef87hA5CGoGFoQxl877o9uxXmUzEhbOcritMbQ/A1RJzlx/pITy7oKsW7KGD3o4EAxy1MP58HI1FlvzvWer3/R9UfhDvzl7v7XuvtH9e+fBPCL3f0jAH5R//7Qy3X583OcsBwClzpvDruVl8/3UCKqJcBjyLxxD2/AasC2FzAVV4agPOE4xzWZAiRZ1SEAE0NIIWiR79i47ztQxf743liGZyFro+eZ/gDgvUlYikTUno5KrCbvsAxJpcygzckzA5hjkZlKGlEikUOe8zkgq3FD4ta00KtCKIW8AcJseuv7AhELjdRGCp2IcyCsIX5r3hLAeljAYkq1E2+ilcQEcPM/G1YDC4kYziwAW65/cV6BmQNmOrKN+gY30MrGTTxMtJSH0Q1fGMXJRsV9XBhAlatar9lIPR2Vt/Ip2HU5cfdrUpiVKbSqEuMM7FTTUZF2DKkm7nmQrQnjtJ+/2IEKTdLxpEDL5chkGzOhiZB+SHCNowrTpTmA9z5g730UnMCPA/hp/fzTAP7qH+TDw+bbRxr34CgJg+0ShLZadkMSWyUXB11ry1kiwM+yz//5DtuNlnbl91tAzu/OOK4Wg4jgMLdgxtyuMacXN3n8e8NxQNbW8D4Z7zWf4dReALKW2jwXHmPm7Zi/mUMfdd7qygQwCoI+QmCwHpqj8ZOvYw/PgZRzGxXZSua4GsZ2oQ/HzL0RUCkPwWQs/8s+SnhqEMIzr89GqIcC723SsDm5Urx1rC/MBgTMLttIX34gB9PH3BAZuLORntno1IiuOdeNFMEIuJCJy8Hd2V6Lvs5LTWjWmxhGg7vnmPtrlCoHH3HVLHPsdmW58nccxPCi61vlBBrA/xyUyP+6u38KwKe6+4v6/e8A+NTzPhgRnwbwaQD4zu//YW5iwVqD2Q6Wqnar9VjEdG89yPrkma+8NHGonsV07oB95GsFIrQRb5cONlZEHCgXOfJyNmAz9fQm5bGVlvrQZ7GmO7JOx8kCoEIUgzWTdkwMohRvABUbyy5N2JfkPNAYNGo/1SbNI8ieu+vYJdj3bllvZdp1oRd7/2srDcJO+Z/ecGiHTQFj7K4Y92eSJluRGLD1m5OwbqmGKH0Ug7MXKfjyX4vtMFaKKQcVBXkNnChoqVNj22KeHo6FOApAm8XRDHeFdtm06y7ceT2g/hWawJOhpxyHlLvRogPbhJ/4DD1dxpKCOx2BWsqgjVDbfvtRipb5kFtxWw8olNyBvhiYslBybyR5jmOUMIlbDMEyvPmi61tVAn+pu78QEd8H4Bci4teuv+zuloJ43yWF8VMA8Cf/3J/v0XLWgvbHHHuuDWBNXzUAsMRPCrFy51PdiSELak0a5ZCTXcCrVbdWP6z3YcQdUsL4blHHWrhXAMnYsYvHcmcISiqioQIgvq8ItRX7Dh8Vpv73sCshs2trDLP/fo+got8NWyspH/uhjm/b96ZIOnIi5eSeADNH/o03MLFvo5lOa+rVpcJoFVi1NkHPfAKXOdVYnO3qWHc7phZp+0xXKuIUwoTZfx1EeggLrXdOs1grxbNiIzwa90Fhz/rOAVwO+NChLeh5hiV3sCWjHAaVt/17ybs6ANUof9+jmxjMCgXN/Acmu6oMe5a4p6CMN9a99Lm4yIARZRpZvuD6lpRAd39Bf38pIn4WwF8A8LsR8QPd/cWI+AEAX/rm7gV1aAnmqbcsmqS+oXJd7XULJwAy6tWIVWdS5Tsd3MQ49/xTVnI6Dssftd+fGejNE3cBwKccbS14RnMDFnSSsjaFk2ZkyTd4Oq1LkQ2lp9mGGOTspahFYQFzVqJdCRqxwOHhblrwA5OdbzwJSbIEw5MUfVSef9eDgpzgHMFknl33mduIwMZ2VTOy6IP3ahYRqbqQqGuzslL3XUILbAyiDacUuYaJ09DQ5EqYKwTXrwKX6kZ2yckA7mo1bmIycLY1i2y4qXNdNonXsEcElNbbOlHaZxde3CDITQoZCG1HW2UIlc6zREOHajIZLU77N1RjR+KuFUshHaUG0jWV+5i5WCosBVzKFbByJzgslRQHKtjO/omyaB1lSkPDUZLvv/7QnEBEvBkRn/DPAP4dAL8M4OcA/ITe9hMA/s43dcPGaMfHL592WITiR5PyDRLXsYa4+PA993Y7KI1X1ubEtE0YphaV+Rj9aEw1/qJDQ2CSj62BEO9hfgPXw0vwzHctC5utqtidbteSu7XZ8eiOiyxr7HmTGzOdhc+DC3LHCO426gHm+a+FSfyihknZVCgWIuHG/0ZfZrln3h13N5KwpYNvKeVuhGGLOMuoRCluZm4XLsHJQjRkJ99RM1d2xQeTGTWOv+OyXIV89ZrnI6x0JScxzyalAUetjmyUmr22UPoUKo1rcpHGyz3h+afQaj243j5zAnK50u3M5vN60HLlbOpZOFA2nOG/nb35outbQQKfAvCzWrwbgJ/p7v8xIn4JwN+OiL8B4LMA/to3c7O1UhvpwKTrwkE+5BTIXA60yHUTLIRSXgkPbRF7AzYx7Ztq7m2088IJECon1srzXRuPctblbcF5aVGYxA+XP2/1LgtB+uwlN4CWfYoHgEtJuPsccJOswDD00Pc7DXjmyAKdGAbeQjxuA+x8lKykFQmwJvGKadBYmPi2hdWkUwlBRSguHaGwHTf9A3A6DXl8mRORac2c5wHNpSkU/fuL+wWAUDZUqRdqwS7k5Yag0co1CG7TMCzA2XC7+lEqcaLRvZFqUIPOKR6jC0nk0MXj5viSU5kBW505SESGAHCUVAhH+Qe7MWXDKxO5zW2F+lx4zHreLsRmFOoa0Cwpz9J90EDhLsNxXIGruzVk0wuuP7QS6O7fBPCvPuf13wPwY3/gG4Y3d4hdhUyG2ntRQeOI5Zl4RwSuZbTP3ntw9KOx6u/La4SXhe593AQnGYV62AnaRZLE82GQAW46pMdE39Drm0Hg7d7z5wAOQmFbTXI+xzIClxCqlda8PmfTjNJ0/QWNjLSdrB2nlJ1sONxLYatCqJO/AIXzqhWyDGWmtc4JBEk/cEDRbj7CgfIe5Glk0o4Sk9tn1JCuZrqgmGv0JRCP+k12ABU1GXtGMd4khubesNy8em0SiuH6GuaP2G2x8bDcaI7WyIoRpR9LQtdQVSaRwZG4mO8k8mulSh+XNtOVm3z24caMbvrIv+85XIE2/Jz1GOyYlRLyaUH3guulyRisbd8KFApIdjRZt7Ww92ax5FS7tVaqRzucFFPFCeqyWH2E0PkFIXg3x5DbEnF24aw+3+Pamrtqk3gx/PUGNgQchv0ICvOcDlT0+XG29OWDR1pjcIZe+/cS9liIrdCVrPqp0uvxIR3eg763q5G3II8R6ozUC/cuPAE3w93lygG1p0prZSAKSymzoWapiaM8GM6MS18GQW+Nywth+re61OjluBX803PvY8gwExu4ZAJCSImvIK/+e1+UIRo31WpUN+7dQCd99gD2vVlvsVtxfR/hRbRyEyNs6z5uQR4F3OjJ2kOfWH8E8MSFTJsqoVNKTe5DXNbOqcdTKJeNrMIcNqp0aR4Mo8ai8ldK5KVzaSZ68Xx+HsBLogS6vQn6+Oip1l4XC+JEHV/DNu9+jJkATMPPFGwEJjFlMsm6WCzyaCw9FoN7e/FcPU1mGKqHQ0Gy+LNpuQARchca8CEnFNg41vmyLq7qs/vAdM9nIxQan+473wX1QgC4x8Z3bDgSceL43IQrVHGHnn4DAIuPjMgghRnwVB7+JYwyJPFXy+UNO7/G+bvzdDqaxBYpEIdyrKzEww8ENzPu6UMwmmLFD5zwXuNZ2AjQDyK3EdHovVGlnoVhtJZT8FSO9khhnLUrnNOItA5KG992URGjGM7FDX3BDjJKVFpLNSVbG5nNb2308swhhHBC8yHkd7gRi1ICzVqL/AhDhH8kV4R8Qj0AF1zBWwtF+0EbjjpODDsuxJOsOhGZMq6Kws98glI+QiB6TdlwtQGoEmyGrHJY7Ux+wJ2AVREIHIXV1ro2hcpekwXCfauPnbLAdE1kwRxCUMTo4+e8JxUpYdoCd+PwZ1CGoJTIkkUug6UIONbUdkcQSG0ktktjFttrei5zHANpleZrAfdhJLbtRlvnzAJgScG1LFbg5HKetJfSWmvtha4CYA1BTAyEY1EWnQ+dtQIuad4KHdMeSwk3Bw6W1mXljY1CvMbhasOAW6+jmV+Syg+gcopRLkuNRe1S3aJx17Pv2lMTs+81Roxdl+QSPhAdZMepkEXTQAlFRZhPkN7txQKmlfO903BW/1s+mwIc377/C3AgqU8esjVDY/wmtImiA7X8w2yjbmwjBRtVHCJwPmJ9OpknujTh1UfBMLnO0EKvexOo0YPPuJ9NFo9vGiDktZ/g5BGP/xCgj0aHIf7mNUtdDogYnlgJJsOjJMNeJwB+rOhcg5g0R3avQAtiP/QRWgEehSWnZmGeldrI71gX0uwQX0IqekbHx3HuiszFla4Ttn12Vof01R8nm+7YZzTLfgMhghHzOSucsEAJrZmBuB7sGj7BWWRV4wo8hTIvc+fy4pm+ASqWZ6AWBuVMmrLRzZh8DtAH7rbWep7BHFhAfRiYW9L+rJ6TIvvMJF6ul0YJdMqfiiv0PQ1EG2y4wQcWVB225BT1olrHYrN9Fyzk1CqCysmjndJk3pmg280n1UAFg/yetZasksMtMfFkaFEmiQQeCrX5SnBzKPX0KWiNPGb60dCiPibITF7aiuxiA06HSlOfW4sbMHBSTSFAjbB/f3Z1KEvvustDdnnVeaUvjUDGFQk+bUQwrp2jjub7AgF3APPW2s6t8M4Y2BrezWLWOc4QLLxuurqMlwVRRF4L5pFaS02upvRdHYVqNZlVOBRdygRULcpe6CikDxlVRSoE11m220QPQzrmqXVpR1VO3sNIVjTc2JQy5cQxHmu3w3kDR1sauTbcpITuz/JaKA8DGVh7oe+S/Xi8Vg0T3s+/PoragT/UVW1IbksrkYxwIZ82RksBnI3hqkEdkCuWt8eXcgMLEnF1YrB8O4VJwmOFYTLJTDFPmjnj7T5KhQpLf0jTG9oTgcZ85tQs9IkDjzWmW1S26n7k6/pdLPegUg1gBRRCxWTWGSvRp2fWuV2Oqe+HyEbVSTgHP0YtaKLaVlg4S2tybKaWqXxYMS3dtf1ZNhXXlBlpU/je/qrr3+JoR1ptwUlIYsKPmHFQedCV4vc4ZXzu2ECEswnpGil/W/MVz6SQi/ZMbuZULofdm9THswMLeYnb8z/KIpOSklwkbpZfjNiMirYuYPkwb5IZcCLyQbTnvws1MgVq44JcId0z10uDBJYGGw1l+nHQ5/y9VGJEom8UKh8GkQBC6aM8e00QmLsDNx2mwTTKliCD37FFeCmtvJUJExl4AGCFwI211fQy0JX0pYELkQYx40wjftrM7xf/BjcUCeV5O8tsoL8tX58Fq4GlRh79SAwsxK2DSU97tgNxHfIDrq4Ixhpam/gATHcQwny2cW/72Gy4mcrFCCmxLYsZjgjywxfBZirsyYeQQk9HPLwOa1K9N2Jy7bN9noLRIHPzFwLtbsdduKeVQ+DWFA6iozW5DABhewl5FNjPYEkhEq3UHAFH1Nd42kBgYYWPBQ+dLyArfVXQmoBATO5KgwejpjIC9W0kBKWEIUI6kZMJzx6D6rQUQG+5vcl9sKDTmGPjLv6FB9NQM2Wmt9Nzr5dGCaAKN3W+7cs83pLtp6rvRD7tbjjUC3R5L/HgJNSnb1VCCoeUsbWZZAxZIrZ9VtswwQknw0QS9rp/nNfZPniLUIvZsNqAE4znPeKubEfH3GXJEGbSpXC0+K6b90baDfj8AcM8wPDccDGmyMphRfMCzgociztlvPx8KiSJIAFWmm8g2PtQ5iZCzTCAadnmnvkLbPvWl7TqsOmTwnFil83XVP/Zzo12K7jHIFNlY3iiKcxSLQktqlWeXDB4XvjXMtxOk5iG1soTULORBpWR29LVZt/ih7hpHHTZTu4/v5VjzxPpAWh0AujyWRKpjFEwiS02VL2NpQ7ZHQU3HXXPxAR5gd2NWF5LPvFWcRnQyBsFKPsGdzwmN/Ji0P/SuAMHx0jTm0Cpoh+uCro9rXIPlAVAwgSH+YU2gTO6jqLQps8k1DIRo43lrCTWD1xIL7sPnUCf3vQnisCNNi3G7ZPpv9Fs3pfAIcxA4TkZgByxLRqLxuhTWwH6HpPajOOt1NwmILLhxIkv/MeZD763ry94nGm1ZrMeB3rqO5wmPatxzXAMHBiuQZ7v0Xi0nru2QTcmnflouvO6UY2VbsTM34Dn4VQuyCquiGgcpdkEKY7IpdfTXUjWholZTLf0YaajY8wzXNYmDrxkboUQBcHYMUgzPSMjCagvgZsrOU3bH+hm1IiHnlzl5hiI6cbddsmef708SEChG0PZw/LbcvPvjWSfd7kOA8O65uHTdPMsEo+Caqx5jfcvYMVMlkN7K5gI1LrHtRIN5iAaGOtrv0tJRVsKK8e/1FGfybRhbJ+qBC7oxiXefXTFQORg3mSiVbxE1wipDEZAUPWa7HRRkPB+cuIVx8WUWXdeLqTJz9nhQmXOgAxwFA2RacopuC0Nh8VGDHcK0irrq7rwsHjiUIqXaKV+b4fsrBSNEsATfa4uBcJuCGm37XAiGBG5aS1d4zHpu200YQOjddcC9HY4z7opJlEthaIgBLcpLBrQaV7qVXPS2PUI+brUKRD9kBNK8GzGANvBbVYxoaMAoxGhR0Dk4ea+cGOdgo7Ii9MyvcOJQzrYpv9FUAKQNR/LDSoDpdMmliBQwG2rvOm5WfOi8R1CMgy+3N8QW1ZtTvh9Ziwk9K1RrV17jgJHtophQnUAuNxH4ScJxT3UMRltNw6n7ZS+X4LvIOLjRKljLSABRzfdjXlEcwHud0eluKuxpu12U9nuHlgtTXus3vkaugN6hsgULL5OaI5V2mqHFlLkogQdWT1rev1Za7RsF93GS8/g+L01cAM6oj2VYIQ50YfCT8XhvhS7nQ5tGzq4bFCDm7j48a+JWYBDr7LaMY4o79FHWft5XE5OsT18y0xZHzRCdGVyUNEDybUXnp2pN1oZgCSZGbHhc8lwpvkHcRRtma153hddL4cSaEEn4PEEBSasEhkkPeBCHUD69DRzvPg9IQvONy7M4Y2zKNT0NX4nN9/Vv29Z+8lUbDAU1zwkhFaxXTnKB9GCO/PQY1k2CZGI22NI7Aw5Dk0C6Q16JfcAIFxf0VKa/PwNgKFIAoCs7C2pGNYkzGBchLBlV33GlUM+Kbdugd7I5YMz6YsunalXILoINEkwxKT1AkBpnpnrHvJbSm3RnOjUtMaBSbBJwYJWIc9xc06YzHA7l9wpkbpVNS3aIXm5dwn8HffGp1OHLWeEckCsjG28mSwUbjSCkgJXUVNs6ZgkmWvlYQNio6S1nszGZJgSEdMx2OuQSkV7GlI7Et/c1L9IIHQ2wmSLybixaU4rZN4sLX7B9XJwAhdtPxtBDJH9u4l7AkCTcTXEAzAE3/U/tvb25+k/TeWMN1UH0PK/rDXh5I+ecblHXoOVdNfW0dbaZvgjrc15Ta8k8w+2fCNkgQ77nbZZ13v4O3i3uhJPwGUzxFgWaN5MdJWNoB8fjxWv9pJ/oW/tU1HnSRbpx3CVhNuKyoKOw6GcNcxZF8R5fqsd9kZ0r4NDinKdYioIXeV3Hp8I4rT16otVrjPPuCQXgUk1CCMLyscja2lFPC7amdO8jn+eibfcTuCOwRzzvJxDPPrZn7cCiiC62fruBrkKd2Gui1wzrHveN7KEVnMXGMaeNXzO9XIgAYCloJkDfWfh2rFzWrtpqlsuGW0JQR+8eVEo83o3fPQTV+wcSgLli9O68bOjHeXDsfU4fW8AeLjA+OkCOwIuSKYQU8tfPfndF4g4wn7SXAcLyR1qKNln3I8LGWcro7RZyBoDPPwyytD10FD2lVkHf8Zi5eWGLEMYbpVT9znuhYfBMFuSMP1MLVqt2LWJ0UydvgX9WCv5EnKJMPJobABP5XKtR+5Sj1wMIjCysNvmnaDW5Y+6ROGkf3tZqxVluIv7UzbmzgJwsvScCmaYLqwIu1cIAE8pH4lkcdIyJCSKsJvQTrIq3oGdbhTy1hizA1hAKEHooYD7/Y5+4Fma3T1nFFbRpQnJUGfp6PnbnJ3oOX7R9dIoAQCPBNtKgPCJPdrbxSOB6YFnf07/4p46BlxWnYtGwKpjKr2hBu4BPgPusFMm647mdubV1BO4zViehJFTC6RFH+2Ei8dwkUiOiO9U2W5qA1GHHYLucTalN7ctql0Ghz7PHMAbDbikDxsp9GxEvco7hrOjzzznJA/1ZOsd7iC8E2feQjzONOu4kGrTeU66aBSasvE6wvU2Ir+YiFPNvnscMzdaqo7iCLvdGYfxPG+SjTgoLsQXO5sPQk+uRGVFpsjIhioQpai0Vl3EkqEDU7FuQzRLkjEemXM5tBY8UFbHwDQTgkKKknLJcx22naFTsAEEcx3YAEvPE/AxERcEgxdeL40SOIsjcpASh6qnZNUvexE4ApOZuN+3LKTj4Gs2ud/Ln3PgMothRLAozDKnETd/Xz0yzXPqUHCvrsYiQuEXwHX5DfpsAKFhKkuD6mXPjiwoOaWP/504GzSuAjRKjaxvb5YBm0zKYGJMNVENm0+uUWrdYpWj1QtRO2gs8NWtOhwIwUyz5NjukUcVh1Tj84jEMwudqVillflxWyI3rt2DnBEqZwvaocofaMAlWBEI5dGnzvfzkW4TegtD41AZsnxjbQ5XDNrT6958PuMvzVmhsZWxuhy12k+BbOz7u/j9r3wN7z59in/2ztfxja99He++/RU8/cZ7+OwXvoDv/8EfxI/95b+Iut1wi0V0JaQWwdJpa74qo0cOKyNQd3FZkke2G09kOTnLzW7U0DRP4lzEwi0wIeFQDsf1+Phnr5dGCYQX0BaGWF7+co9lc4YdDRQFxEJLX40kiaG+LQy/wxAtxtrAm8OMtMgZKhOFwaImR38snsS1pbgCcVDAWDR/Tzx6zmGbCaRhRz2Aiwth02TlyHuwmESP0adwZWtjPi8U5Kq+RwVEwCSTXHmUIVfbTym4/kgBnA3t941h8vv9e7tTF77juDoLCUz+A2kgrdWsQUwadVKDqLJRr+mZytv4gq40LEToFKGrFUGriactdA8vsUupyrpvalN1sJHo//ePPodf/43P4mvvfBW/99ZX8LV3vo733noH+7338NV/+g7iL72BpbCf1zl1DiSrK3vyJezKGmW0lX03zuGsgA/Qpe7gTDtvIiYlelYMSMn3JVfmRddLoQS4uXN8RUKxA3NtC0oMsev4D6RKfZ7cQQbh2MZ+ZuED50htPzqtyAqtTas1R+j47XT4rAH3DwCAUrYcWtCUk38LNQ7RaTot62xfPSOnA0z107HwQIjRvzOVVeXQN323kcaWojFo8Tl/joFL0tRokooks7UBlfJrhaI/zAUMJyMlW+jZXPZjcXElIAtslHSstgRQaauRag8ednpMgrp7k9yXOP+llc2UCp823+QfEgidfRg8eSdbRObY9ZwDkdLBzqO7uC6KlNzRmHMFQXmMZo5ABDv2fuz1G776+1/B3/rZv4uvf/kbQG+sW6PrzhDqLXH7E5/Ed/yJNwfJcJ0S62ZXzWNy+LLgcyl3hLonhaIYS8eUWSEai5HP2qEy+FkSIaIu7HaLUhqOlS9GAi9FdMB7rLXA7XoTs54K2ay1BIEwRTkAcA7lpD+19xY7erVyJwTk0t9jrZMbPBJTfourf3k5a+4y5m7GwZcLNZRt+NgvhjLNHseHyVMk3AGHaegkB+fcKZo2/d8+eRscIwEVrMSEkLQP+T1xxup8iJKyY1d/blhL/rVm4RpfjsCJLnje9Du65HpvnujHGT8H5fcD3vS+oWAwhDKuLpgU2ySGGXXMXIYSeDAm/2xijHvgZzDB6Iq+1piyyM3xBKurW8SBRW8sFP6l14D3vvE29n4X+WQDT+6I2IjU+RgJxP2OuIvwhE8JbrgasOF2fyyXtgIo8OSle2u7B229+S5beCuCiETG0vz57zDQvTRXsQDgVwQ0bwAAIABJREFUhddLgQQAWYBnctmJiJ2uIk3Y7tOPydDLlaMIGsHGkFFqSJmHIBoIpdRPnB0z8V8d4OG9cbAIv/+pMhNvefgF19gPrAtHCEgusdoMB9KbAF3nsJAriu8Abu4J0DmfQzPuH1BzC1tOk6O20vty2m23z1/ndxVwTzfOlk8dfYHRGpuUHMtrADiE5rTh2an8Y6ArehRhHE0A3zT0Xn+f/wzrTzBXP6GaEQhWC6UcPdVz54hWURbmSDYiPncAikE0NCoxrbkSm5l5UF6+kUTqABU0dtzx8TcWvm+9g5/+B7+EuD3Fx777TUQCX33rHdS9x6XAQ+L1j70BxOkd0QgeXdvgqUKSKx6xCpyioUBHyHIDzseYE40kG9OJSYq07zVu0d3gxjk1yZDyB7kELwUSONdIAq/uCa14Q1E1W+AvPiZMUh0n9VGrJWDcB3/GSTpGCvZ7WfLbyt3mwmy4P6CWsE4kwdr2nJ4kaBwAXG02KcQY6P0ov13CDvfEK7kZiFF4jmrs7jlOzV2VYTIPGEU0ZJ+UWAZZbse5PW/+jMcjPKIluW7ig0ic238VLcvZSX92Alei5PvOXM19Dx9h5LJi6dCZx/e7jmOUlV5yAtnMR19+78093JD6Jmjj+qy/6JOmbfetiucbfu8nCp/9/Ofx21/8IoDE7o1dDSyWDefD4vovF1sdUnfCsHrtcB+XPAn9K86gaQTmVGhMfUIsulhQPoILsFoo6jrHs64vOxJoHEu4JaS38FmDJj/iIIBHQmEBTwV7ZelhLViT4gvcp2qsi5t2bzfkFNdQbr9NpVN3JrAkEhsFNd+mromzuCbs2kpHhKDPrOG7pONMSm5Z8yrVqISOEb8B7SMqpKzEmpcQxzKgd7+8Zuqup8THfDM1GsfyrppsttmMTavvY85jyco0IW22SMVORNSF4AtZ+xr9Pd/fIi0ZPlHGX4zi5MeP4io0C4gysJypGCeFepJfdOVaQG908Yh6x2yOsC9uMaczZ55xl+ZwJaoWanHLUjZ4CAuazW2rgO/7ztfwQ58o/Bf/w9/DZ7/8Dva7hffiXcTW2B6AWsDDWoguFN6Vu6rIAEpHjAM+Q9HyFZHKbDzHkhsNU+8HCqygXepk3AiVbCsZ6lZEzK6shA0m13atZPv7F1wvhRIY7deERCam+EuzzcBJdz1qje57HlWImMy+RE6Jq6XjJBWpJkEbhPKhhh4jnIcJmEaPbf+yVPByrN4BnDFKQsEgOJ+7vQma0Qd+mNCUp+HkwOa5F+rRKVLDCcSVIQ44BO9nbClRKLXYbfO2uuMkSEDZ4bkigtNHUfarIEV03IXH62O8ofskoEP/OFYLpriTvvg/J2hjy9XqCiR7GkfdzHWZd9hN4S8wAnNBgs9DNlyGUL4PPxNOPpNLgS5898du+O0vvYXf/N238e67C5Xs9PMAuwGBJwjmbDxJPHn9NdiDP5mPI54ajDiHroMWy+E8u5K6v0sJA4h2gpBDzKHk4tMLMmQcHL0a0vgF14e6AxHxNyPiSxHxy5fXvisifiEifkN/f6dej4j4LyPiMxHxf0fEv/5h95/J0fosTeaIVAPVJ45qYSG6kx3qTRmQD7kzFfudWwwRh+bvqkt12DhQHYVKHnG11WrbITSe6hv0IfucDnPOFRDxptWKlCChh1HfKHUVEhSUcB94TorPTSRY85Bg7txCYaF7IfCaeIQaKL1LFWjalLG0eVbzKC5RC1mJrhvc+2sIx2sCUdPt8EZ0Kmy0U1jPeqF9zFjqcA+DXnvDxyUxzB7Y61RY7dTVgXU/6+riMS3j3NWiwXyQ1M9Gb84SFIcQXm9gRyibL3g4LPlgLM0DWfZU+7rAQwW+6/WFP5Pv4O//g/8db7+z8V4lYgduQqR3cN7vzaPpFwKvr8UNKn/AroiLtuwiWA+6o1MuNYZRyHSnyMpgopy5AZOIuwM72C4dDSWOLSDWELnVd3TsS4n9+69vhhP4bwD8lWde+0kAv9jdPwLgF/VvAPh3AfyI/vs0gP/qm7j/xfJBUPzi4xo+ywqxaefV8l4UhP60lUO3mHk89pFGUDFWuVFjTBwy61YjyADYoKQVMuyx7Oc+EFtPc3xl0FkYwiCVW1KNaNvPG0vOhbXbQ2tw/OZjGRUfBj8DQ/6BgTFugmFBGl+YA7G7pdcY4fDP5++QtbNwXuwqc/r1StQh1vKRT4ppAEPFEVKoVviYdNkw/MBjZHId8+E6uFaXQAOvgrgeRWbsmuh+zzyCjEUrkHGQJLDwXa/f8Buf/yJ++bc+j3x4DbEWnPobl/ZdWAmsxO12o+IWaj1/HiV25tqk70F18HsiXNYyBDAVx2NYQTfm4qJKCdoVm7q5D9jqH6oEuvt/AfDWMy//OICf1s8/DeCvXl7/b5vX3wfwyeChpB94ne18tSLc/LkakTUCvRwTakxGXLh3Hgh7bsHN2NqQ1c0GnZtltPvOzRfZE9aKzFECqTh+otG71P5KaqbJTVSzemtywYELOulnHxDd9NVuOBVk7nd4UmyZONPJDLOaQiRvJLLZMdENE00Mc92QU4VGgxknpAqMVadff6eyyZg5r+rHUQo4nEiT32BWYqsHP+vhqVzWWDw9r8bpT45lB9Qu3CFRJmJFqaJzUWobjXuUPL0790qCefKZTCEPKfm+JBUh5AaIeA2WFK9cPCpuUKcQXgP3vuCVVlekBD75sffwpz95x8/83f8Vv/qP38Ibn1h4/Y1CPnlAP9xQD44sJR468QTkPr6OwrutqIrQhTd5pKJZTXlkpKRwR+G9fcfGRfkX5S8awL21mVnf8rASDyjcnJYuNFc+zFdIDLHAE2b+6DMGP9XdX9TPvwOeSwgAPwjgc5f3fV6vfREfctXE4qU71Um1Eyq9XFNSbIF2x97rYSHaAhMPhwpPuAisG8DinGZzAkze8u1imfW/Srcoo4ARhitf3tEAWXBIqCqVy64YucNHLGG13cSFZ4D6y3HDVDUewtZBl6wkZaYmW2gal0BZc+p7yMaoMWcOHLdE2GU6kkJj0XgC04WGCEr8SviMhZRGUSWllaiRjBRBl2VWreDhcQBOnz4dj7ledCcKD4bBspg56kPul9BdlxKQ+oLKMhRalaLtO5NvtHZMtRbygBGCPXhyMyu5kb7vjRt+7R99Hr/yO2/jnUh87GOvIxaQ+S6+8bWn2AXcdO5A4j1gB/LJa3h9PWDhBhh1GJENf9HjKvDYsq3xLIUI+ZaHYPi2t0PAUqg6maimClTUoRHGxTBpS4zL/bzrWw4R9jWp/A9wRcSnI+IfRsQ/fOftL8NhLeBCirQy0vxd9TiUZws2iRGG0PyC4wLwt2dCQIGjQNnGAXO6EGIq0iYlGY/v5UWcRroXyLeaSETcNg4xdqDsFcFeFyjHrTiz2mD4zecRXkM+bnPuU33udd7rZ9WEwyE7NujMcQ/OPCk3IRkS3U2StMBmqc7ln1OXrbQvSEi+BeDnJaSaOgfKrVRsXLkI3jPD0XWO88oLXL/EiGf6FYbdo76M6+KWoEdZwSjxKkeBQZLdwGsZePPJHZ9/6/fwjc3oU6Dx2ms3fPxjr+OGwkMAqM3ioag5AOR2i3GzIOXakl8/yzlYVHLZ7Gi19FxDllqUL8xepteN/1VdCGoEJgdmuBsdmfeC6w+LBH43In6gu78ouP8lvf4FAD98ed8P6bX3Xd39UwB+CgD+5J/78x0K5RC+1/hbHQyVFSCyq+DuNVasLJcUTKzC3TLatMgVxxcXzU0CUqy8vDA2G5VAUIBZudZO4As2IMkIZMWxSk48ubgC5Uw1yCrK5QgQ/kUEeh+C02lyS7C+QIHpPIqN4SQLMxHFbocZxcVLsTmJKQWjfSyWP+MwFTc+NZmpTXfdi7CVCyAWHqX/9Blzl0qM1Q06KrQptQEvVXO02v65joC3UqS72ck5DmSHoX57M/jk5lJZtXx0PWPDxUxMt1mwBbZyVqpUs0tRbynM4D1fuyX+7CcLX/vyF/F//epv4msFvP7GGyg0nqxEP0m8+bEb9m48rRvuu/He08LehSeffB3xcBO6bKFW4txMo8cDlSICT7c4E6YUUS4TOkl54yGJmnjysM6e1HO35vHW5Crc0sydthqbhVYsM3zu9YdFAj8H4Cf0808A+DuX1/8DRQn+TQC/f3EbPvgaAkjEFgAuFDO5hpDqa4hQTLDMJo8uOxo/YCt+YrTu3D594y/oA+CEsvoMcO57E6+fbj52zvy5waoQv3QqCrfcgx0+SBLnzRftzg3zuNJB+mjIrLxAhrj8N3864cXvDRN8ciHiIBGTgqMYNeaC4beITNVwcCpapwWde2fYAesTpZOZMmCRoT6NWMLj5W9HZxv+K9djt9OrvU6XOQdwzgbgz1fUQ3IMVNDPFNH4VoPM1BHYd/iOh41PfTzxmS/8E3zp7a8QUUAs/m7kLfHkjdfw8PpNnERhYeEhWDX4EOkpHWvujlXuywhAUeo+CswGCFaozJnY7dxBuQN9mpAgSCR7zkkiNs52ibPML7g+FAlExN8C8G8B+J6I+DyA/wzAfw7gb0fE3wDwWQB/TW//eQD/HoDPAPgagP/ww+7vi6Rw4aY01pqaaSZTRCv1MxWSMowP9W8AHz7BkksYMnXPeXEPAJ52oHMp7ccHjUpA4whYOCNQ89u35Hl0YUa7h9kGWMjjduATBAsHDnkPd+JxLSJgKMtvCQC1nL4qf1FJPzzhFpcejCc2kmXLD5W9isEWwkEHfKwWANwuVjlyqVWaFcQhaS1w3rgpn525D3IDhESsFgUwMK3XAgCUzOOQWDApKNQj1G9zKXJjI5OWtqFQHoQ8pNyoDKiuDpt/XBK2Qucz7VZ2IKAqQef2syAsIYrk3njIjX/5uxee7K/iF/7PX8Hb33gP3/XmJ9Ad+Ke/9Vm88fE38eZ3fAKf+NjHcO+Nr33166hdePJAwvS1CHx83XhOAlMDEeCZhVAeSPeGT0faGis0Z36GjIVSAVypoYs5i5AM0jg6t0DbpYVwNamBwFo3nNV///WhSqC7//oLfvVjz3lvA/iPPuyez/8i2SBnrY2WzEkFbS3sgmKnZqGy1SaaG8/xd8gPxWwGj1Nb0L4icKyY3l0WpLGggJiZ8fdam9d/ekPbfUGbOVfHWL3h5JTfZeQOAjAo13zq77PpTpdlfWE5ZDqIFxt3OIQaIk2PJdW8qYjGsDH6WOjLdsKJdhxE0AO1qVCmjHeeIEahOFTpSXpktS+JWg1VRDaPWfOMxCAPfpacwVGeoRLdo/Rm5fkZZ0hZO+BEXLxJ0HTfVgRuGfjEQ+Fzv/02vvT2V/GuDvrY98I7v/1l1BtfxRt/6nU8fPycSxBr4bXXF6ITrz95HU9ur58Cpc5LZaOQTyjzE6cwK/q0Ph831QRnWl4uJddniSS/Wi/nEFuOob4L3woS+Od3NTob95YVsSXOAZt4gI9s4gSxmSXh3CSEDInEbowBAE4KSh4wkYmjQGTBb16g8/KjVmBx2fDUB0YDlODJbwDgltr2/0tWbElTMOTotFgM54BWwsoklYgFxqw3fWKVCRds0Q6YR0hhIJVSvEmuxkLtA3/tjXRteSUnA6jds8BJWpl4+pQlu5kWZAr5e7WZK6CJ4wlNfCa6JrTSV4V5c8EXGj0FW1T8pajQ02KSU6q2496n1FeYCdF3ADeuj0w9G7oIQYB1COiTCj4bJA7R2mhgB25xx/d/IvHe19/C//arv45/8pWvY3fik2+8ia987vN4+hufwT+7bzx5+218z7/yZxEPT7DiBtwab37i43jy2gO+45NPcPt4Y69CNBOyxtXrnvLpUNPSFc7/PMoyJBSaOaTus+uiqFt1DqwzxjTLjTP/0oNw6vKLrpdGCeyQfznWGtL+xy65riBjXSq+zCPwQytscXFwJvHsHPtlaO77Qj7ZYb0PBPakEzKbycWEKx+9HwFnH054Uk0q5plcrehLxQZ+Bt4xR2j8Wut+9uUHDchCZJgs45Hk57f+U+6JkZEtlQbmJmHDasNKws+Ki4LxnPEXk4SkubnMxswNgCnWui0dC2foIjR0xnwQ4WRVIsaiUkE6ldrazKRua04AH5N2LL5R0GX6DSbBjsmfejPwuS+9hd/6nS/h6dONp1XofIr19Xdx24XujW+89Tbe+se/g9e+73vw2muJe22sDty/cUeuhbqrE1YfniTOyqKBUfZDfsrwRwA8Il3clPIdOOwayNoFZj22Wq2taQ4vdHVcJC//i66XRglEC9qNjF3gfhxhhQtY2luZVg7gKS9TsBOAs/pOs7Y4m18nGZF6iHNgpcxyrsWjCputsGxFrDQsb0YFlqbwZo2YRbzFQnXjqT7rctBlHeXdE2a1G67417IOi36/38d9iGDZrd0MGls9q/IrSu3WncF4tJiP1yJyoYKwheEaVBVy3XDXho3uidpUCxno8ySjCuhTqDN6Si4AM+uAd/dT/rw9lZpMxcPvTpBROBGttuBL64NAxG2anLCV+cLskQy14romceVlwfigmQWfKPz/U/cmMZttWXrWs/Y+52v+JrobEbfPm3lvVlbeyjRUleWSLSywxMQCJMQMJgxAmAGICSOYgGR5RjNBQjKiEQNADOmEJcuWEBLIoqzCdtlUpdOV6czbR/s3X3PO3msxWGvv80U6b1bhAhR1UpE34v+/5jR7r+Zd73qXmPGNNyof3B/4H//qT3lyI5zdveTmescPf/O30K++4CIdqTpzfPYZtkncf3TO2eYNalGuf/IFx2nm/N13OB9Gkg3O/oxb7hLoLYZp92rhP0gSag25u4jdXeloGY/WxvRVZQncEETcz7coWYyOIdFKqH8U5MVSY1W1jda12VuwC97gQ/d0jQDS4h6xFE0Y/v5mQZtkYctVWxkQ05iQg3uSWBD+a+tlur7HCXJKO2nxRwpeNdAwKM1aN+PUHszCSAxcIAwFlvtnNZqtxoKQ+J7UO+kaKSouOzXPEAq/7V4IVCudMdiNYTvt6BBsHrmlHo055bLioWWHDyTNrWYf39suv/PS5QShX0Kwfv/9vCTAMJ8u5Nbs1Ng3MlCU9iL1aJFAm0TVz9PaRrC+vzt12gPk0ENop9iuj8jJoyKhcH80bncHvnp+RVW4WGdyXTNfrtk/t1AQMgaMsyRss3D9/JrbqytefPIZ43ZLyolJwKJi1fQwgB7ZGJ23RQQMNIVk8B4BkveauPJQjtc5jrNQiP3+WYjZtNc0HGdhm8ormNfPHq+REQhCQzeN/qchygJI8puhWK8WBAzlobOIU4wDLW16f7Awxpo37LajeXkMs/BoWF9cQ/IxUbVGi2usPOce+ENSDX269qEsQywMTwlUNWTKe0ruD17Ewz8RoMmXtdA9rj1OWg2yDMFGtL7QPCrOS4MKhlVIKXeyjGciyxbtU5yiZr4MbmkbM8BRYmGllmOfAJO1JxQsP3RDeroYNQWgWRUrMWTWWuxiEf7HpkSgia3E/xo1GFrUsAC2jQTkzrNhOj4Utdgc3tOfc3sObUdkEitLkI07K6OUif/+//y7/PW//SNePtmTdE9erfn+dz/i93a3PH/2BKqyGQeGw4R98ZLnT55x9cUTylyoF3ewg7AafBho4yiItjRKIHpISL52mk4F4lUu77Rs0HGJKI3utNqUpQYWJ4U6LKlYAx89yvNr5uSaf97x2hgBt+I5bkIvsoUL9w11ajp7ft12drPwRq8mRCbZ2WVtIUNfuiwbPrxn84wYWQJsiYfTEHqI/LbNFJM2BamRUMILyKITf8oJMLHgLC2RzrKdpFcv2nm2n0tbMLT6fMtxT17fnjnuB08ixzBg0Jh1KS1Ms/YMmlk9vUP9TG3xKMvwlTC07dFoDzt69NQ9YfLnuKRkC6GotyKbLca6Vw5aRWWh+Xbb0599pCC23FFpAi3tymxJvYTwqrhSxNt3BasTL/Y7yjHx5KdfMX31KZuLM975tV8hpxHv4ARMmA43vPjkE6b9AXRGpkrdH4L1t1RbNPl69EykXfsp16GdXRgMoTuiLicvjXR0siKEHsH1co1INHH5U9SgUP/s0/zZ4/UxAtby6ZbTBhraQnlJ1Fr6hBpjEQNpi1HDM3lUvrTKhgQbfUZApx5pv/F9I0f45kSZqIGLK3QpxhChaw9VkxNqLLzSQv6IpSZgMbwzaWzCIWNpKQk5raHtorj8qD6kNgE5NABUvH11cCi9R0e03guDwX/lrlW84nBaJXS7EXlQc8H4TfRRYz5XzXCjKMmbmT1qbRHEUjJcCFWcgKnS81QAqQHapua1lkaqqosEW6+qEF4vOzHESMsFtNo4zfjG90hUBcLoNup1CiEZR+qX9mtH1pWP7iR+/Z2B//yv/IAf/PgT0sWKdDYyP3tJef6cv/7VE6xOqHi7zogyTEdun3zpG7Uqcxq5fOM+9+5uyDlHfi7RZ2K9cawbserNXcRsRd/HjdwVZjjLSQu8YSGq09ab4zK43LwQJLdupRdg9FU/8Q8cr40RaGo7rb6eNZRpG+CWJfrsTxmFETa3yEAWLnoLHrpHOUkH2hScJUde5LBTvDlFKqKxaCSIQj57zr+rkVFK4A+ubRcodWreMMqB4V+J6cMOZp0cduIVpMUGze4LxZzC3OYHtPQHogZ80gWpPQcMr2qtIuLfURHPNXOkLJ32HKFz3A8VL2E1N7+U2V59Zt0MNxrxSYXBPS7LHzt9T7vC9qzCEPTPc4uRxU4kxk9uGEKvKsSmkDCabv8cS6kmneloJ960Uas/uCesBX70yZdc7ScsC+M6wGaD/fVLUjJqBanVDVkGpHqfRnxWKTN1dvZg0hBlCWMrKULyKOGp+p/WU5Iwmkx4S//a/egPLn7eDKjbuiinW9QdpBnSpTsVoPTpW//g8doYgYRbs4T2OrkAmlwbT6sPW/CNpbxCsY0bkhMgGmFf6lbQwsOYQso1rO1SMWjagW2ghUtShT9qUlRp9HNq5b4Ix4kHGcG6nw5CU0NOJgsyJfRwr02n7cDWSalQOAmRY+OotoEiXntXcYMwqG9wb0FVepXiNAYMZFP9RveQFjWSGjVZ3/gCoI5/NJUjN1phXK2hMLEp81K+asG1nnog6IalL0q1PhSkgVw99BCjiBNscozYanfY76L1jeE08X7HkVM1HbzzsbPwupf0f6u6QMxqrLx/d8Xf/PFP+XufPOOmJh49esCdszP++g9+TH35DDHFipEtYQwU9VJg1cpUhKqJ8c4Z2zsX7A5HGg7kg2HimZg/kBTRQVO06mP12uWnFtY0xyZY9f/2WkdyA9X6YWan1C50cWvVsTCQRsjRfd3ee12OqKmDe4IGBUjUgl3jf0GxG8rfFoeEBWwepoX4Lfz0B9PCxlZTXhDVU++QUkPNg4YcrK8WXnYfJuaaBNIQe+m/V3xjaqQTxEZvkUJrAT7tuQc6wNhz5WD2tapB84Ctc1DFlhmr7dri+tpr2v/8WzSYgv7yxqn4WWzAUyoJL0OMWFvKjPLK+9qffuM5GSHcr6V7+h6ttVzfFsN38kfRk5tNs6A9tm33zYFVOjaBtKqKn1djGzdBDklKdo1x3jgXXh4P/I0ffcqORJHMjFFqpWShJJf29qpCRYuXZNWiguOhCvfefsT7H3/EcLZ1g6rFI1A76e8Po+/RWKyLZjD7Woxn0JuD2to4McJm/XkJ4jMLdRmk0n/TnlGs5a87XptIwMNxZ6o5Bz9yz/DixXdv8PWXkoGx9J3X6rx5iRzcw2EvHwme05u5yMTYtkUHlBSLlkxJnLTbKzmn/nAI0LAtQ2kbrHngBj5KFBkjvGvdfoIgkQNLe9BitAGpLTx0yTRf8Ck+V3FRTcyYpfUpxLk3q9nDRpbNFPnhGHjKrIU++FQCQGPZ/Jac+CzQy7TBXqW7d5Y3uFePQSKoMzNtSa04SQ+WrbtEty3V0EhJksUA185sDLBUlvSm4TttfLoTwegRin92BimARJnN0zUMlML7dwt/6p0z/upv/k3+19/+EftJuakzLychq/Lur/4y1599yfXv/IBRZvKwZ87KXJpSMeSkbDcrfu0f+zWmixW3UiipktPyXFMoCGmsLTO8ESkH5TqwG4u7kMiOzdhygxpWAG10WQsdQKSSdcAwJ90RFShTLClLB+3PP14bI1Cjfw0ASZS2CPEFkKV5WeeWm++U7k2AJWGMDWmxME4n7jRqdSPstLoxLRMNzz5rQSS7wqs2o9LjhYhWUnj2KAVGhHKatnpk5+dfcUOUcY+mvc9cGZKHPb32e5LHtxKfP3HHPUZti9k9e7LGLGvA6YJFt814bLiCDAFZCEQk0okmtKlGS4TVrsUQqimJwCEaAQfBLPunSITh1kq0/cy7IWgGVonINSor1hWBo2Yehr8RDRxP+JnzUq+jt+giR0rhrzEgkyWY1jhuMwR56k++u2WbK3/5t36XL24LcxKSZaoWJqms333EWx+9z7O7D/j8b/5fyM1nrJNyKMqcMoMI2wTKxA9++3d48OEH/PI3vsNoA4MkakyWtYgaoCH2MGb8WXfBRsK+Jkxq3JsBwUjZyUDtvknMvIjaMNlajkfHwiLZCrOikUr//OO1MQKdsxPhXiuFSZO/CvN+4pB73Tv10l97E7Ex2+Gbpw1tcO/bvKy/zzdxs7ihZBSRSOPuvxqEtn8H8CfGMtnIQaCWqngI3Dxpq7PbCXEkdQPSLHwj4/Ta/kk5T8A7GokSVDuTkxBxaSeVjuU1g9p0EImog7REJT2Zb7/r+L6vwGTL8iKimxZxNXDPN3gY2KgYnLI/23Ppsmr+D1oI2/CUbqf6HQ88wk6f7gLuniwMTj6yg77emFQxM1ZJubsyPn36kpf7wlT8vRLgnIlwsMKuFPJ2CysvEQ6WsDFR0wBSEcnYMHD1YgdfXiGo4yIBYKeUo1ekheUORjYxmkyrAEjvNXAWZMWkhPPzak0hIouW7rTyNc1ZJGd3Jr8Fqd84loV4AAAgAElEQVSxk5v+c47XxghIhKep0V2b548+15bz5e7hxXtHGmpvDgL58669NBeWhVZtKiJBB7Z+Y7xnwVHfBoC5R1ZMMyKlFRJokX/7rsbXNwSrbepwmIp4yI0uKFifHSjWognc6LXSmziy23J7xzROQlyi/BehPW22YkfJF0PgKkUW52LdMEpgAp0yS/Mu9g+slTb5tkWT7bolaSgYhVGKVKcNxkvR/rpAcyd05BOjgSiS5QTUooO4HQPAAgyOdnJraQCddaihwd/O388rypoW320VRdkI/PKbAz/96jl/6Tf/FsXWUQ2ayCkza0VF2Qwr5nnmJ199xrw5Z/Pm+wzrzPvvvUUOduCwXnFxccF4vmI3F26PrjDU8KquJByGtZbq9zwTgGuULIPwLglSLSjKbn9FnStPnr5kLsr53Ttst2fcu7wXUWNgIdkXRW1zCMKJLL0Zr/JBfvZ4jYzAYv2bd1nchoeD3cNF7unZcrR0tqVmQPYFqKY+q6+pEBm+wZp3koVC3LJ8/wxfum0ajaUmdCF0DTuxboysh+Je8/XNJMFlt0UOKv6/6RX2SCCuvxmOFuuaLoBeA9xaWqLQU4cUxqVTf0OyKp18H3iMmLocVzQ2tQ3U9O79YbQMidascyrL7v8Ng9EbiOzkPT1LaDHWcm0soX2Ywf7YmrEykyV3jg9tTqAPEwmj05dKPMcGwjQFKAsKuT+HgplxZ1DeOh/54Y+f8+nLHYfZKFV9+AjenZoksxoHjCMPv/Uu+T1jK8Y4ZGy1Yqczm7NLpizcpMTxZo8hPJZMkeSiNS11acNpgT5CHd+o2sgwGMYMppT5wDxPPPvic253R548u0WGFef37jEMQ9ypRVYPbaI7hAiM9X3TaAavTt969XhtjIBxEvqeLmERLAUSboZUt5adDRbhowuROAccXaTC/Gb7p1YzcpTUmgfT5sEiTPPvTP0cNMp5c9O3T17yGZAuY5YtodWcuNPTFsNy5PW6TI1RgZTjWjSmz0QzU1sKEp7S74mXvrwN1bo9yKlN6XGIs7U9qzmxycwfrmMiTWoqjJNBEzwN6VWwpcXaCPquWYT20RqdDAnGnwlOo9bTqEIaPcKrFvaq0kDT/GptNdmaFkErsDYmXcx5IAWCHi23ZifGgC7vAIvxsFAoTlF3kBrGQiduDzdMN1f86e9ckOc1f/Vv/IDPnh44ZGXWaPEtpWNGFcPyiN49YyqFNK44qFIm4/r6lmlYc0BBBsbVABU2m5FNHvwets1u6lOPRfpdN3NwUTkixUh14vrqJfv9LU+ePuFwnKlp5PLykt/4E99ls9mig6sM+SSohMX8RRH8occ61/B4SZpTUPgFJcLXxgg0oOoVlRjzDWIWoX14AQeNDcSBpByW1BWGPApYKsv8zEJt5T7nbXevFBFGCzXjnd07vRIo4K3P1hYkxNxAWt7i/6gLmNg6C1t7cFu9LWRrjLHe7RdpTLg3Wrsp7RPt5Nziay1uWg/qm6dokQC9ltFD/xYhnUYZ2DKwVE4u3Bdui2lkMcLLWfWIrvEJutBnc+sW5SrPKQKcbdcQYWu/37Z8sb0aLS7P1f+lEdmZOS2/vdKoJDPKdOD5pz+G/RWXf+xjfvrlE756ued2cslxcg6gObuRVpjmggL7WqmmSKlkYqaA5nYJ1BwVDPF0SRXaKHnpz7lVXJqxql4FqzO72yvm/Z7Pv/qC/e7IoQibswvee+ddLi7O2Fyeg7kiknLSoxLeXhs4HV7fOSRLOVdP193POV4bIyBR5uqjqmIjpcimSuipI0ELRhArwR+wuCf+0CUl1x6wVjZqEYHvFMVI5ioyFuUZaTvxJFQNM07zm17/9fP1fLT1CnBCcJJevaltC8lpXt4WsSwDOSzTBUJi1VSLluE4Z5f3Ag3iSI1PcWEVfEPFth2aDWiLRJpB9WhpCHzAXyMNIwTz6MKxEfNooy5VD4AS59nKLEPOPXroRgVfdC6xvhilNhgjpUy16sZGhKhxkENyuw3lUHUpM8lDGL3ASULYtDMUVRlSBhNKC7zrTJ6P3N4+5erlDb/347/DO4PyG9/7Jr/5w0/5rR99ycspM1HxukYAggijDKgIByqqymYcqTUhtmaaK0mMebfnplQ2j+6jc/X2ZxXWKRN5Zyyfdv3KbMaqHFhZYXd7y2F3xfMXT/jq+RXTrFzef8SdR2/znbfeZbPZkoYRNZiqOy/SEGsrummzP7uirQHMPGqK56KmzC10/AXHa2EEDKc1Cilq/AsrrXmBlHz0Ugrz5mpAEQ/KkhN5rt7yfP93A2UkWIIJOsXW26wXYkU/p9hASCtDxk4xC5MQA0gtBoXQnz1NTdjzdEjpJLLBvYWcfE9rmjqxEOE9Y+O0RMGWjdaROv9Nr6W3kLnd1/aqiLLx3nOvEOQoMak1+C7RVWojFaBFO3biWcxJVCJtAEv7Lq9xN6/j5JV4krKIllZVNBid3auf3H4F6ENZwsvi7M2c27OKyEarn0tEEDkpsxWmm1t0d82PP/0Rty9uubp+wa9//wNss+Vv/P2veHZbkTTEGDINxqqrP1nxTTUM5o7GkhvE2UfTH48zMz7MlsPkT3S1RtX8Z7WgdfYoNvJ1TQWsUPc75uOBL559ye3VFddXN9h2y/bOmrfe/wYXF3fYrLYIiVJpxM5uVJv9RRzLSRoVja6rGaXUqIdLblTw1xwTaAFm7xunxagROqXgjydHPwcZ/ffWqKqZnCwQ9tS7rki5h+Umyw1yyW0DyTTRcW/ZtN4P0JDsuW/ZCNVTSGl1KqKXhJb3BEcv9vQSuVnfuEm8PXmOPD5FWQkzpsitxwCWOnVY3AC1vL13EeK/K8E1EDnx3BK6+y2dSA4iamz22g1l0G+1dh784HGkG2WWKT+ZHBGWn09t5VmzAEVbRBbeWmUp80bInowgP+E6BYECTo1c1WTEzVWPNbX+DYnpui20sZgkbJgeYSo8ffaUFy+e8eXf/xHH/S0v5j0IfP+X3mW884jf/vyaT27gKJkbMWpODLL2VGf05znPlVqNMgxQZpgLCeNQj1g1bq4PHPeVYTVg1zN5GHBug/L3P/mM/eGIrFbknBlk5eVcPWI6ofs9h/2B6+OB7XrDt7/zXd5+/x226w3omgoUdRGVUKNwY230ZqkhuAc+NUsYTJxno74WLCUvS0ZLtYBXrr7meC2MAJzk00bfrH0XheEfcu4bz9fg0pjTF5lG2E/zvtbzbWjlmNik1kQ8li9pebFbz5ZZLrloGwF22hpMfH+X4NLu8ml0WTGjBiCm0W7bpc5aR1/7JAvL3dDwFsnIQjttW61nMe29cVqnkY3jJyfGtiHsRK8+GoM+ItKRdk50+mq7y/7d0lPSlgrIyd8hIoB+DxaxzdPoyrlF/j1JBKsWqry0vIIG0baKYQOIs+A5vBrUyu7mOdcvX/KTn/yU3e0NL18+xQTGdeL+3bt88N47vDxM3EwV9ekNVIrX00Uwq+Q0UIuitfosgVJQVY5zQQzmCpIym/WWui0kMc7PzxnWmZRXWK18+uIlP3nxgiGHEUgjmUqyGaGQDea5sPvyBQ/efsw/8d775NXah+PGfWvRWR/GQ5Q529yKtlLa+uzR35KWtfvVZjv+oozg9TECYdzbojZ39P3vmAQDNLzMCVtQohTXfLrn6wtjMHgg3QP10oq1VMHQmMRLlKeaHUk4kJUdSaBJaVeUMQhFXfGFNoo6RcJA5LIJNcfEwRhSgGYt7za6Ok/CUfdGcGmGpN2YFqIbEnZooTD3MJ7FqAp0Q1ejl2VI9Oggi3nEFPXCFIKtbcZi+/xEGNcu2xUpjzmO05ICV1MSMGWQZrO1Gw8l8lcTxDznD0IvKbeUqXE+mkBoAfFxZuDPIqlhWnjy9HOur695efWClCp33xh5+OgBm9VjLs625HFgqpWnh4nDpGgaWI+FwYxx2CBAmQtqPjykrtdkSRSDfJh9FuV4RjVjOjNEK/vNLWeXK8Y8cn6xIQ+Z/X7CbMX5KrFOK253e253t9g6k9eZMfl6rHPl+sVLPvkrf43fqxN/9p/8s9x566w7QG+uahFaM754hBeMzBYvaweo/LNTTr5OuyU2hgbMfn028PoYAVPzENHdi3tUmn/0ZpIo+sQ7whtJ4w5Y9FkbVlmYdEbMcQ+Gpi35toN5S6MGRChqraQnjBr8l0jbW+6arMFwvuWyNFUifAKPhWEKIBMJrxXn3r6vs+ZaNBKdgGacgPvxmgjDrYmdvOIVmkeQrrjcJvqoNkAtvHxnARIb3LXtI+9wZLmHFHHPoHv2FlWceqVX6M3WWsCr30cZejS2BF3aP9h630G7f+6nF1r35FUjAan+2WWeKPOMzTvOVnD55j2PZrI37+SckCGzOxT204G5JI6mLGMcM8Uc9TeFWpWpKsWUuYYhTcIgnlKaJHIxsgxY2ZPSSJaB9XoFklgPzkvYjkOMeVO0HNFxRIsiq+RzL21GqOgqUW4ndrfPuWMPWEp4r0ax/hgCgJZWTg3jHM/UX7SkpCJRTlZBsrujPxpkIRzY8r/TgaiGhmJGyp57tWaSpjickguGSoB87vKbPDaOCcArP3OZqZYqCEhMdyV1pFU0FmQ1NCStLXjtzWM1Wm5toXmH2uPcm+tVjcXXBkMsIXXjALTR363pyTQqJj2d8Y9CCZ0Dz9n7Zon7prCQh+KOOguTvpDasBRT/57eSCQRWbUHEziEa/FFNIKHaWrNJHtuTsvVk4LOHHc7xvXKJw0HRiEGOSocZjUIVTNz8SnGqooVp2FrrZGOVKqap2J19g1bFKvFnWdVD/PNmKdK1cK+HEGFw9E4Vp8VeKjVDbd6OfCoSjUYgxhWVahBHc8CmzSCVpK57N06SoHDesMhJcbVyGpckTDqegVagIrpzPkazlZ3uD1ck6hwrNRZmQ++noZ33+LenY9ZJwdSrXmYE2PuoiT42mwVJmnGuEmytV4Vj+QiHnPsKEHVGhT0PwQmICL/GfDPAF+a2ffjZ/8u8K8AX8XL/m0z+5/id/8W8C/j3ZL/hpn9pd/vO+J9tLJQz1/bXjFCUjksXOS+biR8J6m5d27rUE7y0x4/9IXr4ZEzkj090L4RJEqB1Tdk06iT1hIKDbQELwdZfG/bEB6CR3QDzm+I8+hRRytv4ZUPFzCxPjxCkusrtHJo+4RqRFIStNATbn770z5Y8uI1ek3Dgi+RxMdeyzLVhmYAeumN3hXpKU/L0064CKSQJvfSYWoRQJ2xOpE09WnDSlyneStVtdnPTGfHd4rfYxcQUYpWKpVaK6kaqDLXghocZ0W1+JwCVQ6lMpsxHSYOZeJmf0RVmGZjPxe226Gnhdk8qrIkkBJzrQySkASDuKxcSzesVjQlbD46OScJh/0+0jtlsoMLlxwLSKgEaUXLFDyB4k5EK1aVKgPjdsP3/+Sv8633PuDi0RuY+OhwixUibQF1IRXrz7AtoHb3c4+EWVKFFgnE79qz/7rjDxIJ/BfAfwT8lz/z8//QzP690x+IyK8A/zzwPeAd4C+LyHdsUX/42iMyc1IO2m1cRAsTG6vNvXcr2NUu990R8QRkDQ8R8tjhgTxqWNKJ1IExQBIVJZmr9mLRp0/wDdptbN+XcCaWpd6RGALoYRjoobNawcO3SGhaM5B4jlzMvyOHcTJT1EI2LOHe2ARLvvAq7mwzflHaRFaSICaMxWNeN0Jh2ALlT2pI1PbzkJeIJNqNSc5uqOrnIjgoNoTnTqZUrahOkVY5Qq1FEQpVKgnF6gGpB8phRzGvWXueZphOFINZ/ZmbVg/Ji1ExSuS7Wr1OT3XDpaocVbFamdS5I2V2YzaZRwv7aWJ/OJBTQhGqDuyqMs0zm3EgM+ACMu36lDQKqjO5KrUWXl5dsb85cNgfGFcDeRzC+HrkcnN1YK4zKsImJYYxk5JQdeZ8MzKMgqTidN7ZjdN63LJaXfDd73/M48fv8N6H33ZDpJ6aNG/Xk8xao2+jRXnSHWIzyu1Qi1F7gSeArxVtMzUR/lCS42b2v4jIN3+/18XxzwL/jZkdgd8Tkb8L/Abwv/2+39NrvxHQtHyYV//bwS6gz1yzloNz4qVi85+8t0man3zpSceada/oLnG5cd5xGBJctJp1y9mbh2zlulfz9Ibw+wOUXv461SL0ICgeco8ZpBuX1mzUw8QWdQTzTl65LmlJb/cYnQiVWl5pvqF7G6vi02z95/6eFCW/QpufkAQoE+V45FgmhmFwzUeAqt75xoxYBS1UVeZaKSFyWOuM1UpVpZowF48KWk99jSihSvZ6uxpanRhTSqWYMqkiquxrpVRAM6qVKc776nhgmgv3Ls4YotaeirLerHxkfFrAYB9yXV0ARJXDbs98PPLk8885XB/ZjGvWrDhfZ4p6JCJkploo84SpcrTKLB5CGZWNbpHN6CBnrZiNqCTu33vE4zff5rsff4/Ly0vMVtR6hJYOWDiJhp2kJh/vTs8sel4coAhYxSLUD2xFWvrs6Vpb+0mkN3b9vOMPgwn86yLyLwL/B/Bvmtlz4F3gfz95zU/jZ7//0WoatNBnwQYIsEsi1wmIoHv1ZfknpOL6bxZswwAMaTa27Y+Tz27hsX+Wo+PZArNO3kPfZsV5p5YboIQ6R16jxbazjaMzJyDZttksPC6IP1hxnnqKXMLZzRnBJ/L2YMJgFGcVOugXIThtU8e9sDbyzH+XUiJZZVRH2WvcD61e9xYLb0FhyMJht2OeZrCZnMbwTNUHX4qQkpKqUkulRvjdxmp5RJU4TpNHMtWYzdeeE2gcjCwRzpsJUzD/Si2eklVFU2IuIZhShaqFjKPqs4qLqShMGAUlJwuPGilM8Xs4pAFLsBmEasI2Z8bBTXgOrKGU2YHGeaaWien2irXAw3VCZUOphVH32G6iHI48f/oCrcLLFy+RlBjWmfX9LeOQGNPoJefpyHycKevM+uyMX/ljv8rjx2/zrfd/idWwZk4rDqZkLQx57AIhfeaGOfbSui5cUMexj9wcSZRzfWxfJIwuduiCMAZI7mudugDRP+/4hzUC/zHw52Mv/Xng3wf+pf8nHyAifw74cwD33/oGLaP23xGgWHh7kZN2UOmbYxk37Zu41mB+tc+J/H+pnnqo1erhiPTfLNFFu2FLya6bDRPawI5kFvz4IKyoQRs0EZ8hJ9fU/t1KZe71XexBzUP7Vk93ZV18cbRzEPp5N2zErTwdw0jJr1ljISWyg1rl4BtttfJIRmOjkpx8Iy6uWaYdNWixVdXnNkqlYtFvX51GjDDPc9Ts/T4dZycaaS0YiVI8bTHwXFqDPWfRgoywi+lJVheMQanMEfKK+ntHE6ymmEbtClCSYEhLKTbHuknJGJM7gpRgbYmahVXywbDJoM4V1ZkyT4gmRJUssMpwnrykWI5Hrl68AGAuhbnMXB92mAnZ4PLyLquzNeN2JGGsU/boMAnZMm+9+x7333ybjz/+R7i4vEvOG2rQ0pN4H0xrYmvrY6meWIC/jnO5GnFCtLa9E+9pTiBFlOCOEAGtteM8mONNX3f8QxkBM/ui/V1E/hPgf4h/fgK8f/LS9+JnP+8z/iLwFwG+8St/3MylV7uHT9k9kQVQ1jZDGzxCdBFKbA6NyUPukXKE7iUopxKraqGqtjQiUu5FmERdiLS6hC+DwWhecjyhwnhfgPkDSYZLRVmU9moOXOdkem4DBfEV7j/OMRdCqA37kCWgSC584N1slkgqDOKSUc4eMy/31YqZMs0Vqc6GLBRq8Rxe6oGpFOS4QoBaCybKXIK8VBdsxEG32SXIUoJ5RpHo3QiAywYmqx6qFqFaZdLoZCRHXm8w+7DSaQ4zHIBfK6xqS3FOjWUS5phP1jQdqebRHbi+RA6ERiJuEn+f4NqB4zAwJmWURN3dMNbCoB65lKpeWjRlyEYNT2pVKRX+7qef8OLLp+hc2O92DGnk/GyLGdy73CJ5YNzeZ7M11kMiyeB4Ra3k7Tlvvv0+mTV/9p/6p8ln95jFS86QI2u1zrUg1t4rDqN78aC393QQvKuV7nyG5F2YAVVgKVSpkT7u3AwkNYH+n3/8QxkBEXnbzD6Lf/5zwN+Kv/93wH8lIv8BDgz+EvDX/iCfmYYodJSwghFmth5+V8c+Qd5pSH5rOg6UO+r8ScTZZ5j3EpjzzGGxvhKRRevBB0d/Xe0F569LinbbZraXTdz4A0POC/NWQJPX2BrF11/r5agkxEy6NlHIF65qUJkbq9G8dJeDfejlvOrewGC2gllFbO7dZVYqVC+VFVFSFbDCbJ5GTHWPoNisqCjH4gYyUVGFWd24FK0gGTOhVC8zKVBFoRhikxsFCXBV3bMrcNSjG95SXNdBZ4wcbLcQ/tCQREuRxqiiIpTkfBDHymFlvjzrUP0eUthuEtNxDt0+WAWQWsR3wnYQpnogHXw/3Vy9QDCOY4qaOSRRN0a1OUr3wENO3Ht0l2GTqMdK3Re0Ok6Sc+Jsm0nDQE0Jm6pPeV5tGddbPvjoI955/wO+9e1fRmpCR6fuanSTxVSIILFFphg1/i5/14VlrMezIhIRVV1mUIinBBrr2aIJTM3L14sYqXT14lNuyM8ef5AS4X8N/BngoYj8FPh3gD8jIr/qO4IfAf+qn4z9toj8t8Dfxol6/9ofpDLQj5OyRisZOmAFjSThDnzRXT8tl/QSlQVA0vvo6RoE2ClYGL8L69oSAzX1kdZJe1lMrfHyW8jtHqghtmEW3GpLKPq2JWb+XUPj/UcHiC/JClrJ0TGnNvtGrIrguStE374Vr0VbYjYNI1CYVamVqP9VZq0UMfIsiCpHnVFxND6ZUg4zJGEf9FmLnLpWP++iM1BRE0fwzRtnLO7fYObNLbZwHzQINjU4Gs6StJPqNb2zMw3Z82cqgzlmUtpzrMpq8DFajo54SlJKxbSQ05rd7Q3bszO2qxVDMkq1znmo88Q8TdR1poqxXQ1M88ykEyOjR2hanHB0KkYTlaHNaiRdnMFGqVsHN6fjhIhPLVai92McIGXeevNd7r/xJh//6j/KnXv3SXlFHkcXvY3qkpN4lhBeWDgZ7q3dqzSn5v/XdC1P0lvoEVDHuJpxCe1CWNirrsMZDvDrA4E/UHXgX/g5P/5Pf8Hr/wLwF36/z331iHBeEjktXiXhOaEk5/EWdQ+WEEdrI+UZwBeendRHI8RKwQcAOiffS3BBCgk1mSaEJZyEUSSf6RbWVMSbOerxwKyF1fnWue/V+fcS/IHUEjTTqPUqYhWtlRp89CEl96Z4nbxGuD3roeMdYkatFhGDh+LH6p+jlvFOs+JevhRsN/nXrr0/4dkXz9DJ5bvOzrcMQ+b62XO++vET3v3wPXaXDzgmEJtpegWOSDufQhXGYQzD6yCdYJ6bN4OZklca8IjK8dXqpd7q4idJlKITq2Ggzgbm5UWbJ9KQuHq5YxzX7I5HSpkY7p3z6Zdf8c3Hb3upLMOqlRgL3L54Tq6V8WzDJEpVY5pnB8d2B/JcKNGDkESZbaIcJ3J25eEsnhJJspg67eZfywGpyqoKVGXOXkpMtkGzkdLIZr3h8Qcf8Pa7H7HanPHt9z9gHDaUGu5dFgq4igRDUUhW3KGkjLnc0UIPjpRIcvu3r9ck4lJlrRclGY2NKWFoEXeKEtGvmkd3jUjXPOIv8sSvCWPQN25jP7WGG7eg6aRcon2DWyw4N3BuDZ115wYkp+UzlkgicjAijWhpRSzoFqJDo/ECSQIviI4/UaZyIEkl2cpDb51BlByb2kttGh1vhpaJJEKZ59Cci1Q3wsWqE1MJn2kztZZoQhKKRghrM9UiZC8VmX0jXR+PkBxQfPqTzzASD969ixZ48tkztusNV7e33Ll/yflqy/Mnt5RnV8xv3lIv7/v0JPNQtDURFXMthKYghLUuyzB2RL+ADMHfoJcKR3FehGqm1NYxaNRaGC1xPBzJo1OoD/sd+WzLy5tr7pwrx+OBYznwsG5gnslaXXZcYCoztbgR2mw3TLWSp8k5GJF+eMW0MNeZ9Tz4+1J1E19LRJUOIGtxIJVIwVAjV2crquGplyTQShFI48C9ew95+MZjPvzu93nwxrvk1UjKI7OmcEK8QlJzhxRIULjztq6aRDot4o2I0avN0naFd1LKklYubdQtbk0RRfhPLJxZLx1DUMX/X8YE/r84nLnXgL1AyluohKHVfEIwi3ZAk7pqdXb/nDCYxklTC7ihSUvNHqIv3tod9JA0wtvkBTUys4fAkiNPPSDjkSyw3z1lVmWejpi5voGPIivOQdfqhKPJN041RVWiBu7lNq3GMVWO1UPTeXegTpXjNDOsBrarNfP+yPPnT9nc2ZA3Z9w8u+GL3/17PH70gFtVzu/cIWnmJz/4lKLG+eUZt4eJ+Qh3zgduDzPbAj968hVZ4Wol6H7i7jRjCOQYOx5U5CzJjSg40GhGluqzIAdQnGe/zrC/nRk9wWWeZzabFVeHA3mEq+s942oFZeb6uOeDB28y3b7k8uLMP1cruSp31iObAVabjNqWVAuP7l5ws7tGUI6zsa8zpgUtzjNY54zoSM4DWgvZfLOPwK5M7I7uCCabSabcPr8lnw1ApWh4/+K6gqthAHOjkMnsp4kqsMprzs7v8vF3v8Mbjx7xzQ++x5DXlCFK0RiTtqG3bqySLfWkQdx5mEFJEe5bCScV/QDRNyHdEDQTIlhgACnhWgHiqkUaXA93aq+W/4QWAUQZMVdEjWxfv9VfGyPAsrWbiYsjbswrKGm8JvIio13yYggcK45PlGYUIkoXQpI5Iam+YiNbn3wSH4PlKYMxSnEyohWGtbA7HDjsJ28BrXNMAvbznLVEX7h/YZ08iqimHt7X4jnK/kAthq4zx+Se99lXTylTZVLl/PIcK8buxRVf/uRz3nj3AZsHK65vbri9nfhs/4Ttw0vk3Hh5dcM8V65udxxL5fp44On1LQcK4/mGvF6zvifcvzgjP1+RtmdI6DOkAJqILr6s5oGleiIAACAASURBVJRcc7WfuVRWQ2aajliBcRBSUUaBm/2OYTU4hXY+slln6uQNPKvsFZphTBRGcqqsBsg6oyQ2SbA6sx0DI8ggmphqYcyJ69sJwdjNhWM5RticmY8zaaOsh0Q15yxIsug9MKBSpiOWM3OdocxM+wO3RViNArnNrfAHlIunnvNkWBbmBKwG3n7/Ix4/epuPfvm7rM8uSPnM6Q6+4BAThpgBqMlLs7VJ27V1BLRpSh6dS1u6dL7/qcMCx7nCNy3UuKXQ3WTLhr7+ZcEd2mLu4UjMLPjDYAL/fx01Lizm4gQXHTrKB0EYov8pAhGoRp08rF9rKz4pxTQRziZq4WI/CtoaOJQxGaNkhkG5WFdSUrJN1LmyP94ylYmX+4n9sfDy5Q3T5LjLdHPtNe2cyOOaQwq0dj8x3+4gZVarzP5QePLsBY/u3+Vwc8v158/Z73c8/PY3yBeX1P3Mlz95zvvvPeTsbIMh/M7vfkK53XP7YkcZzrmbb/ni+ZFvfvxtPv/kS3YMrEzYDYkP//j3+PLpCz7dHYDEWx+9y9nZiu3ZGiVxhkt6PTh7E3Ki1AqlsMoDz3c7zs9XHA4H1uOKZy9fsFpvuBhHnl5fcXn/DQ77a8Yhs8lrVqaMqojO2FxgPUCC43zLkJVUR87GoMNq5s644jAdGHNiKkcwGAQv15Gp6noLYo6/+OSkGVTZDolx9Hq8VSNtNogVTA9Y9dy5TMVTMDPm454vvnhBzhmTwioN6M0N27tbRDJZM02oUwEbvAq1ubjD6u4d/sR3v8e9x29y9833yAyAy643ZM5Zpto3unvvCpIWDUJzrQJvrQYsjG0O2XwzREa0lhPhFl+/SxdnRBeSIjV0LItIkTRSRYuJ1TXuj7RdYXj03PLgrzleGyOwhEMslvJnaL/goVAiEE8aWaKZBfrFttyqAy1xI3rqIMRDU0S8LJUFzreVsxFWg9vpw+3EcX/g5e0N0zRxe5w5zMb1/kjdF8pc2b14zmAwbUbGzTm2WmNi1Jc7dk+eYQgPHt3n6nrH0y+e8/DsksP1kelYOhlqUtgfZsrsyjb73Y48rNmZh/d1PnBD4UIEtmvSvQserUZuDhOHJMyDMK2E7cO7PL26YhiE+w/OWeUc3BNFtJIEXt7u2J5tuN3dUKsxnl1ws7vi4uwB8zQxpBjLPhdWq0wujqSPw0AWo8wzx6lgOpCHBrQqOXso4QpHscFwCSwwSj2SJfeUTK2GAXY/p1oin/WcdpXdaI/JOEgNDMi6wKiqR1auTBQKxmqsM+RUqeXIajWSk3F+uWGzHkjJMC0IMKWEiEuIWx758Fvf4f67b/PeN79NXm+xNNIatOgBuHtno6ld9/jVKx9dHWEO7r+DtBrqSsMrcaf1NZt6AHzKIRBaq7uYE66aMXEApAKZ4cQzWqPCS9tLsVf+SBgBAx8/5ht4MCeLlPh9xkUhI/XqIb9HAY66ZPHGiVDF6HiB9nSipQZhFQx0MDZD5t37ysXac9tymPnixROub/Z8+fSK/VS4utkzDBk0szsWrq4OvPj0M9b7mZeffULaZC7eege5rOQ796gKX/zwU9LhwFRm0jjw7HpPUeFmLnxxc2C8GHl5M/lcxDqzPcuUMfHDr55y/uAO55eJxx88YrvOPP72m1RVch746I17mBbWZ1um65HDPHGxXXOUmTI4PyIJ3E4Hvrre8cbdc4bsuf4gicN0y3olDGqoFUo5sh4crd8MCSnKve3Wufq1cu/uBYf5QIoKyDTPmAnHMqES4OtcyeoqzIpzFFSd9FNjg4oUDEFrxkSjndvn+VUrVIJ0Y+6Zc/bvK8U/22ncDqqVoCFr8RxbTNBSWSUYxXj3zQcUU8rkFZNRhWwzVhOW1zCOvPH4bR4+eps/9iu/zoRx/+Ieq+0ZJj6BWivdS0uTTlPrxLJTdeSupUhTBerTLkNExQLUtv53gJwzTa79FQohAEtfiIuxLGChmI/q83kYXk9rTAx6hyxQX+1J+XnHa2MEHKiN2mdYMO0RUVi+htjH9eQAZCSlqKe3De4FxhzRRUqNZHHyZjwFuNgWHp4PvHF3QFB2u0ophaubHS9vbjlOM6Uax+q90WWqXB9nrnZHjtdHrr96xu3nzxkv1tjDwtac8HSYKgeDzWrD7XxkV10ZJ6/XzAlWl2fcubdhvHtBTpmagSHx3gdvoVRsGEgrQSRTzFgNGS1w0MqZCNeHI2fbNUMG5srhuOfx2X2muXJVZxKJUL+nzpUVyZl4ObMahhDtzKzwOZCb9ZpaioOz6pUYD7+dcFLNSTMSvRNe5dDQs6t9AYqliL58g7v6s0+WzOK1hVorklp5VDx0FqVGTVvVqcdWAocprvykYoGrVBoL1BtrFFMv507VwbZqRlHrasm1VoaUSXnk/MGbXN69y7c//h4PHz3m4eN3mQ04WoBxsc7EeilZIk93l+8btkWZ3nimveHL39vq/Ordq5x0rNKyfFq5ajEqkQJY649pfJh4k8Ey1CTA8+DX041GRw9az8oSXfy847UxAhjBASA67dwQWPTrSyfivFoOgeDwNw/fegtEqVajdj8QBS8Uv+ghG/fO4Ve/tWFIxu1cOB6V/e2em93E1dWEkRnG0ck4ZeLF7Z6vdrNz4KeZQsa2Z1x+60N0k6gX59TNSM2CjsI7H77F2TjwxvwWSTKP3gCxAVLh4dmWY5043wxkTdzMR9JmzZPphrvbDdO8Z3eYeXB+zmdPnvL+o4eucisT63zG8xdfkfMbrCUxVWXQxOF2R6nGnc3gdN3DnjubEaEyVdhrxeaZcXNGDTIOw8AkRsqZUipzbN4hWnbH7CR2z0drhJweuQiQdIi7WqgaZBoKZoOvZDFqrRGcCTQClHqJ9PSoWhEc3MtAxVuMReFgNVR3PUSuVcEKpU4ky+To6pzVX1dDH27IAyLG+Rvv8tE3P+TBg4e89+HHrLbnSB6omjiUiBrz6BL2pmCJnBrHo52hxAZdFJZb115OXoRT6uL5jc6d0BYAsPSrmHp1yuKe5hjFV6100dYOZhs09aHWIJcbgBgJsbHMplyZtyHX2pzg12+918YIBLuy513dVkoXqXJARXrxI26Q97mLpNAkDG5hZ+wRHsMfSE4Vq3C+Ed67n7nMyhSfr+K9/bPW6OITDpMxFY8EDrNyOB5JOSGjYCtId84YVwM1K6vzNeN6oIqPqxpFOFJZbVyE8jhXMsbVfsf5esPt4ZaUhbNh4DgduLvdcDzusPVA0Zn98cjq4pIxe8fi2GnIyvl6hWjlWGZnMubM4ej8gjw4IWZXCmOEm1q9FVcB0YkWvNboaqzmQzoMHKwy9yNTncg5xexGj8aa/mErjeVkiPo911oCPHNOQTS3hWfzCMNDVYNoUfaegtgQqp39pupYgzcP1RjqERTueSKZMki4zKAux+N2Lf6cSdstdy4v+aUPv8+H3/llzs4vGDd33FD0JRJlvQD6fOW58ZMgoTng3ARuPVpQ2gRmcxyjJanmUWMjpzUyUKLhXdajDZEU05o6DBYGYJl+lZL3ErTTbVFJjx6qLZTohpV5TO1XEvf5647Xxgi0ygv9v9q9v8dFJxYvOtEkZL5aE0fFQ9ThlettD8cZXIMm0jDzpz9c83BMzBgrE251puxnpsPE/nhEpFI1MxXYT0pVyAxc5JVHGavE6u2HUJXddOD8bMNcKvM0M1hhtz8yXJ7x/Pqau3c2WFVeXt/yxp27HHe3nA2CHQ9ohnE85ywBdWK7drryJgusBqRW3r5zh6oFy7DRM+YjrFYXTJOym4+UubLNK0o5YAZHS6zyyO44s5I1q7RQVMGYa2k+2UHtds8tGoQESlCVs1ZECtlcuELnEkSXNpDVmRvJXDZ9SBUrRrGZgncj5uTiLlNNJFOMAhSETBtFbgpz8RzX2c/VhTGqMlqi2uSItzqFmeh3yOLYg+sdJMbNmtXZOW9+6yPuvvEW337/IzabDavVpasQNUwCnIhkGmpILNsmjF3fwJGLN5AuFBBoGgApyFo9t4dOaGuklJBfDXZfeH9JkXp5hBtJDllDaNZbROP6lnC+NxoHk8sb11qZ0PeJhlVJYRx+ASTwehgBw0ttbWbe0v4LhAVNpsECWpBOibnsHecL2KALM7f0DMFEyKqowPmQOM9wKIWdKXkY2O/3XF3fuDDlYWauTvk8ToVpdiLKkCGJI9rTXNBh4O7mjKvjDpHE7vaaYcxcrtaUeSKzJUezCqqu/IpxtsqIVc43q2CNOfJe6szZOAIwjCPbYegLvFZnFyaNVhTTKCUZmir7cmSQzDhkdscDxzJ7Y1PyCQCmYSQ5IVPFPcvm5anmecmCqashoSzlseD7J3PyFilRArDLQYkr1J4vOwJgHd019Wm7KQgu1Qdpe9SmSmrTlWpIgQ+ZQUHq5I1TyaiSQtPBH3bOztobz87Iw4rHj9/i8aO3ePdb3+Hs8i6bzaV/l6WoNlkP4X1btbF1J7k/4XEjjIcFTPbUtEaaaoEbeBLUKlKtluC8gahmdVgrGKoSjNWTVN0Nk3SVJ+m4Ah2biFBiOUf8RJcoukXQrdsmrvUXEAVeCyMghGACJVpqhUyiTblpeVSj5BMMq7bRJeL9AS+PVMCFGry+mnOMIs/KJivzy5f8z7/1nC8+/RRF+cf/1G/wxdPnvLy64fYA18cjnzy/RWfh+fMrdtPE5eUF29XA7tZlumspFC3MGcp8S6mZzTq78Kco63XiOM+MeYBSEMusVhv208R2M2BaGUYPEGt1eW6dbUkAa2UUR9tdVCRFa3KlamVjguTEuNpwLAMvdzNDGhGDzdmW2dwj72qE4LVQqvPsc3YpNQQoMCaX/yoFasVr8KkykJjEWEnqghxJR2SE4zQzVaHaTML1BoSZFPJkhRlRxaqjMKaGZJjnuFatvetNTVEqZfbGLSvFJ+ccPPn16/Z+EctCHs+4c3HJsFnx7jfe5/LiHvffe4/N+pzzccs4rLDkbdME4Utwz2+xkLoCD87U9PZz7U7Evy8GqLRQPUqbTawmi+MaDgxG81pVZ/edWBIzoURXrBsEl2pvHAOxRG8V0pZ6BF8gpeirEZ9bkcT1FCLC8PQYCjM1CUm9R7V1sAJkghPyNcdrYQTcUHpOHxQrOhcwONcaeb9Ky4ujPyA2TRLPA5t3w4SUPUfyPMsQBlb1hh9/8TnffGuFkTkej8zTxKSFKdpp91PhdlcoRRlT5mwYWadoN00gqqyTMJkDR3fOLhyzyNknD5s6i65O3gQ0a1f/VbMgKjnDzZeidgmvTHK0OxaCX7f2UVy+VARSDuTYQayUXAIrkYLV5j/LyRWJEpkVgmahpoCY1HPJFPxzlxw0xCrklde0xVilzDj6tCaJunkluyoZICEKYjr7VqqVSnGxTsNHcpkLfFK9tdm7BA1LIdUejD9MSOpgWjWliqsNjXlNGgYu7t7n7PySd+8/Znvvkrff/ybb7RlpuwFJ5JrCMy5VfWil4tQdBw1EjqYfCXmyKhI9CEtyqoGPpA5KN4PQ2nzbhzou0kVZI+pIQnSO+qbVkzMDL6nCMu0ZWlmQpdoAEYUsHl3R9rVh3KynLUmjOzSaO36ButhrYgR6PuUiHidxD1WDwjvkdgdp4VCb7T5kb/IZxOWWchoCPJxIyRfxYMYHD+HxdsCmke0gyFh468Fjjgb7Y+L2CJ882fHsxc5VbaUySiVRefr8CTl7qcnUQ9qMcTjsSbUyq6crOiuTAEUYfW17EJ4qhFiGmDfe1OLgWU3KIBlrykjm3pqWy5tzz1EYxjUJY3ecQaM3XR0NFqvUOpGGIbAUpZbCVGfnqofnWKeVg6SDQO9E9AehGkNWzLs3R0kkhVIESTlamSN7Td7PkUdX58EGxjlTy4yVmcNhwqpjAqaVw+xaBRbdeSsRUhUmEbJkVisH3FbbFZIyab1luz3n8duPefjWO2w359y/+wa23XCx2uCLfozrCi8fSHiWZgQWEK3VkyBAYqIZLbogLaeljBdCjh6Jxra1hgPQcQ4L6fqlsScAUmIUuSxTsmDpT0kBIFp4eDG8fb0bqGgFMg/zTc3ZiDTDr0sqIcrotDc38JKxxlnXNvfiNU8H/Gh909pvmQDDEOmAFYzsuc0JYqsmlFpJ6iDIIOLqteYo+XlWvvfYeOtCeLx2tdnv/ImP2Rk8nb/ND5/e8OTFDT/88VOeX+15fv2S3f7Ibn9kHBPXapS5MB0PrMaxs9tE3XNVMyR5iagcvTYtWX2BA4giuQ399CSzig8kc2DeDYdk71NfIp5EGjzcL1PhxfUNKSU21am2UymYGatkZDPWET7WpOTpSCmFHGq1gyo5ZaZ5ooow6y0pJ/bVI4Qyz1hO7I+Fm5e37L/4nNX5hos7d7n/xl1EhHkYFjCBAMnMWCVHGky9ZDiZi23UOTEfKsfbI/l8A6Ksz+9yNozc3OwYthve+fAbrIeRs/WWcVhz586Fe/vLu4w5c3Zxh3FceXifVy3EQyU5kUdC0wFFrCFBsT7Cuy6AWV7KcQR4FimRtOYa9fuhhCcXb9tdxydbjEhq26nV8YGOB7TNmxuWEKITGobBDVTTiPRFMDK49mMzAAApJjMF+t++o7cjsnQGigwu0qqKZjrYGShGCN38ETAC1ksgsjCxpD8xf5jxUJMsLZM5NXQ0SC4tN0tCksrDjfJLd0fOhsQy5AFc6115/uIln3/5kk+evGC3P3Czv2U6zpSiJEbXvQ/FHdXiYaVaVCC8PEnIRyVJ0ffvIayKW3gNJV5JQ6zjRWIrE51nNqGSkTy6Jc8gWSilcns4omaMSbi6ecEgifPR8eZNdgxCamtjVmZRauSExSrTXKnTBDq7Z6sSlLJEzitub6+pqhx2lel2x7w/UKZbpFbOLkZyGhGtDCvxHD+JqwuroVJBamcHTtOMictu5c05HBMP3nmLzXbLg7sPubM943d/9yfceXDO977/a4zjyHa9YswrztZnnuYMqT9zn7GQqQEmNjk3aSQ8A0jU5H0HJ4U6B0FTGAJvSAiPfgqtAU2EAwkkP2S7IhI4/UQ7ydNN8G7CJjYTRqa76FjDXQvQLQV9AlYQ4DQkwdoeCOqR4w/mF6txni32txj0akZIr4VBa2cbvAWfLSGdTPfzjtfGCHCComqL33BQy+9lXmSh1Pn2XZ9OmsBn6uBhRv5v6t4t1tYsOw/6xpz/2vvcqurUrbv6Vn2x3R3cCdjBspFMkFEkUMyDZSECPFgEIZmHRCISD4Q88RIpD2CUJyQjkIgUFCJhiSCFh4DgIRLkZiw7iWXcxpfu6up2VXfVue+915pz8PB935j/rq7TVZFFtL3sqj61zlrr//85xxzjG2N8Ywx88fYJf+wTG+70XlHfZ+j4zumEX/2db+Jb73wXv/b/fAPvP3yK7z69BGLgcqgQ5dkV2vlANqavHj1+hMPWcO8FjtQ6nnid1jY115yMQIOEp5wDRzVVGrEBCFXnZbVCn0mrdjYH+jzhzdfv4e7t23g0Ao+uOt59/ARPnjKGfvXsCk+fXuB4fIjbPcQ56Lg4P0M/29ADOOYRV5dXePrdx3h6eYnjcWK2jscPrzCeXeDqyROqiTyyb992QG8N2wv3WDXYeKjPto47t+9iHq9w9fghDufn6KcDxlXgoJz18XjCaQwcMXF22ICzjrN2wJtf+gG89PIr+NybX8B7j0/42j/6Gn76X/1RnN9/HaN1bBn49Oe/jW2c8MlPfxaIwClOHAueB1ZjukmJfPmhiHjNkITLw2152R2afjVlZ1acicxQfw/y0Y0VEE47A44CUgEEzMQbE+XrU1RnfScV/DvNFZsnrTgQDTidaDjCfSFaAHNgLEceK6oQSy/BDULIReghWxOqQ2gQ1Oc0YtdtTLnWzbEFxTEOu1jCB183RgmEsgLuD8imoIoM7+iPXniW5jIS3RRIqVSM1vLlex13zzucQkFOfPtq4u33n+H33voWvv3dB/jug0e4uNQAjCp5CTx48hiHywOOpwuMATx8+AB3bt3CrdsvcJMnuwyPcQQAtdea1QfRAsopRqLInhShVTCsz0DMgWOb+Mw58K/84Gdw58WX8BvffA+/8vVHePDgEVI189/51ju4eO8hDu0Cj8YRx0fPEAjcunMHsR3Qzjcck62wHn/rPRyPV2iHc5zfvoO4cxvb+T1cXj4Frk7A1ICV5ijyFbYtgN6BBrQeODvbME5AnE5s7NkaMDsbkEbiOCYyNvSt486927j/xifw4gsv40uf/xLuvHAP91+6j/sXgXbc8OTZxOEeFdLIxKc+/Sk2+VARzhSRxR2guirvVrArdkbQfHryPtyP0RH9ifU5ph9XooyGfpa/D0foIZlRe/neutKcS+ns/AgRhWbVLIRhO8jik14Akh2R3eJOt1ZHnr0CWgUDfbWWcmZUntzDk6JCesw8g7olIohQ+Xw6dOahsvtf/97XjVECREdZcM0x04heARmkR4IR9u6B2mKxAQ0dh0y8ce8ctzb5RZj4vSdX+NXf/ibeefQUv/377+LBgye4PA4cxwnjdKl22Ozd/+TyCS4eXuIMDAK9cv8uzs87clxhzs7+dyf6YcdUa7Rq+qC4xmRH3q2pFfcm/y4mek6OvoqGPHV8/fd+H7/0jd/Fb33tN/G77ww8Gfcxx8S9z76Blz/3aYzTERdPH+Py8jF6DrSNSdQnVxdop4F2ccBpKI12BKcv5SWeni7Qxxm2Brzy8i3cOtswj4kZHU8TaBsPT2v0LTOAOA/0wxm23NAO59jOzoCzW0Df8PKdl3Hv9jluvXwPL91/BW9+5rM4v32OF+/dB0OlYigeJ+6cBb7ylS/g8vIKpwn0dmB1JQZa33BS/8gt22LMNRlhWf90AUlDHfgEkEH3rkWDe/M3uQuhbkGeC0kLq3r7PbNPxqRa1wcVkF2bgB13EoakR0qZZDaOSNcBS2B1jYZciSkXt3xyAn63CHPas03VCSCZFkaizZPiSk18Bd6Au29BSm6CGZkWKqtvKsuOpqG0y0P5sNfNUQKZO4uvISNIuI+++c8TSeIQyOhyRxcuuIJBM3DExNa5XJcJPBkDb73zXTx89BCPHl/g0ZNneHbJNOBROX/6tjzAr7x8D4m7OGtM2fWmHnuD7sloEqYIzOPEbFnQKwd9kuE26KcTYxGTw0a2HNgi8eKtO2jR8OAZkLdfxTe+8w6+8fRF5L3bwMNL5OUFnr71NskCZxNXz54gjiz1PR7ZUjvnFSKfouWGMTj0cl5d8dCcdWyHMxxun+Fs63jx7DbOzze0AC5PAzhNTYJu6gW4YWRiu3WG7XAL57fv4dYLL+LO7du4d/8+7ty5i8++/mm8cPsuzu7exvn5Ldy6fc7AxmzA1KCXxmrOnBTM2+fnhPWTxKAuCM70mqLoZoZC8yJD+w716ceATfYMz1ZA1QyETylA2YBTf37XVlJQfyd3Rpge4GIeSghZpi/NL3C2wGw15wFKlaZ/Xwc0699CHAl+x+8mCXITICVZ9zNjMf+MTD14z321olASfZ9wpLGJxbgLYFbQ8jmvG6MEZianpyAQTsVB5a/g801bf09XSWDXrVGDLwPAEdEbno2Biwz85nce472HT/HNb3wTv/Pt38f7jy/x3oOnuLw84dnllSriTNygpW6NtNRq+Hni8p8uWQE3GgpK5hWopNoA2qx++S068jRwwiVGDPQY2LLhc6++iq98/tP47BufwG+89Q7+l7/3m3jvW+9i5gHj3puYl5fAnRP63ZeAqys8/s57yIuHmPOI7ew2LmYi3n+f0f0j4yRxfo68c4uNTW7Tar/4uU9iu91wq204O5zj7NCRreFw7w5utY4X58B26MjRgeg4nN/CoR/wwv2X8OUf+BJefuk+7r72GvM1wdr7OYZ8Y/IEJhx84qHg6C1ZNR9OBfVcV8PA26AyCCCnhsJCAVvVA3DSsVqeqTnGdA2CFMSIxGxZljXm8t09GYmHB8VxCGnrRQPmQU3B6KL4zgWncy6KMAemBmKKj1Cj5Uh6I7FAx3ZHr+bkY00Jhj4zyYaNEN8jiBAx2f4e2WUaJ4azIcmqzt6pFBoANOCUQI5Ua7iGMDkgEv35ccGbowRCfeRZYRaluU7iSTdtnKcQly9QwRWUjxfYkCPxztPAaBNvf/ddPH5wie88eIT33nuCJxdTGQCO92RzCzDlZ+LLccofo0ic5kmjrlKNKNeqBhojhW1oUHFc09SJI3qfuHe24fbtc3zlh76Ir37xc/i9dx7gnUdXeP3Vl/H6q6/i4vIKX//m+3j6+AJ3X9zQDx1Xjx7i8slTzHkFzI5xdoZ+5Eix2QL9nCoyoiM2HuazT34Kt168i0995hM4u9Vx79ZtnJ2f4c69F9G2jtdeuo+zwwGQ+zPiDDOBO2e3cNYPuPfCC3j1lVfQtw3HkBVS6fCUH08BXxYPKhaqsm3kmteobA0c6FXEnS5A1Joi6RKyAGmtnwfNyEMAFI9piiGlek0WXk9CbiujWfZYh8Lo37ElE9ICalgSlYmryICgugOBMWiWGPGfikNF/S59fix2ny5biAKqfIXbkdvfX6T5hGtj8rqLZIzh84LGQCN21GefD9MF9ufkA68bowQANnHoTZHNpC/HXC6XTVk5rxd6Qt6+/NqQlQgGdn75rUvcu9vw7N2Jd955iLfeeoR333uIy9PA06sjZks8eXaFmRPn2uxkP21spF5gno4AArntqtr02dOzC0RrGKeJq6dXONsCh7MDprr5nLKh9wM2THz+jRfxJ3/qT+DW3Xt4+9EV/v43HuHBoxPi9uv4zJv38ejxBcaDxzi8OHHn9l0Euc54/PY7mN95iDGOGBtHaZ3mRHv1VXRMzGdXyOMRcfc1nL/2Ku6/cR//5s/9LF5+5XW8fHaGvm2Ifg4o/mA3BVgsxFFZFaGeGTjOwBiE+CPJ7ItJVmSCE4/QAjGY96ZQxmrvDiBVwdisvLWvTukywzNIwUXt9wAAIABJREFUlPKqhjpLN9fNm+zDfWmC44yECzYrLkDFM5TGZa8D9omUVpGCkMToMC6k0OvOiTodkNwUqJg5K1jpnD+7CSUOij1Y3UyRrcLlyHoulxxH16Rm0YCdpYgkp6CCfgHFNpQhSHaCRiPZrLdNt8zf7eJyOC0JBRT74Q+BEnDUn2O4NDYp1w1m6KH0YelcmFfgiDGDRlADioaHTyfm9iKu+hW+++wtPLo64nQ64mqg5uohSfVtyaEXsiNl+QFWuEWAfneS4366HIieiMOGfnagtYyO2ToiA4dbd7FtG9544VX8xI99GZ/45Gt47+kV3n3vGR48vsTDR1d498ETnOKEJ08v8PDRMzw9Mt99flBe+dklcHXBrsdjsmKyd8zGmvuprj2f+Od+AJ//6lfw2c98Al/94g9hYEPKGrOfZmcL9ISGemh8F6b890R0QeXZ6GXNZd3WQZxydUALqLoDR65tXXlWpvZzp7kz6G7ZUjGUpUk5vifny6FdNhwPu9fy+Z2f56Ff6T7unTn4gMbY7Syy/XakA4ZYgWnLVrP5XYJVhTz+lN2LnRx7DUKXsVKyxHotEyoiMuoJf4LP50uTMNBUh7HbE6xbM6qY4ovQfcpyRT7QuuHa6+NMIPocgL8K4JO67i9m5l+JiFcA/A8AvgBOIfrTmfleUCX/FQA/DeApgD+Tmb/8Udex1M0MWZTUArp7CtR5aPfwVWeg6bpcfWpxmiIcTw397GW8+tmX8S+89Aa+9e1v4PGjh3jw4H08efYUF1dPMS6PwMUFqbxxhVMLtKEYQZe2V3edw52O8564d+suzj99C4jARe/o/QyvnN/D+dkBL772Im6dneOV+69h9g0/+5WXcTi7wP/89/9f/PbvP8Db73HOwKOLiQcXE4ct8PTqCjOAO7eY7719+xwXT55hXjzCmEcK1piYeeQMgidHRqcBzN7wU//aj+An/8SfwpzA0yPbcsX0OPUmBWvQTLTEdW1cV/KbEQBGV/S6umKxK1AER7JRCNW0QoKbUsKsxuMXe33WZJwpIpGCeJPoYUaqQLQ2WjaVA1aY5xcvP1i/gEBZxplmZAIAOSFdWQ/OqGykmJe88EULnVW9WGO+ZZB8bqYPNALsBeAuzVRgiWR6NeXLh0FH1votKwedfPd2SPk+Cox7Ha0MVT3LIiTGUDYh0RlNCmwCedIA2YM2TetR+/oHQwIncPT4L0fECwD+YUT8bQB/BsD/lpl/OSL+AoC/AOA/AfCnwBmEPwTgJ8AJxj/xURcJCWl62oo0aMjyeERTYglmsQolOwA1bjRzDqQ4cgMice/ei/jinR/A5eUlnj1+jKurE955+B08e3KB/vApMgeejgucGg/+GInzrXP67UYufzs/ILbA+fkd3L/9Ag7bAZfbht7PcG87w63zM5zd3bD1DcAZLsbEJ186x6Nnz/D+w0s8fDpwkYnROg7niVdeewFbnjCPz9DziDFIYzl04KoF0BKt2zI05JEZkK62X9kC4/gMX/+t38PjH7tAbwds6pGHkOWp5cmypAjUwWU6E0C5XgzANjjKP0vRVrm3ho3Y3J3GqO/6bVvYskIJHuTyyRXBd9LLpi+8b/K7WyCkfiC/WMi+goBN0XEWXq1D6BRghPP1171uBvZQN8gegk2xAgZBW5KcQ90m2ZLCsuixhfmy5kvT8B9HkMjqTdF4s+oIsHueJuYg5Ns3pZ5b891TSW8t6Nol13KRqRYlfcc+eO7Z+zhjyN4G8Lb+/Cgifh3AZwD8DICf0sf+OwD/B6gEfgbAX03uzv8VEfc/MMD0Q1+hSMymQ83R315FBnec7unNMCdYCdfsExqGWZiuT2cds6O127h3+zbunL2E7InPvPkmMhNbhCK/HBd+UioGJ1YOjmC2YqCpCpDXr5FiaSsVaJgK1TacI3BogYurifcfPMA8NtzeOo6qFCR7eUOMK3SwQy4QmBcT4+ICOa6QDRjHK8TpiImOUwZuRcepTeSde+jzNm7lfZxnx7Ai9XNzVeQnejoT758I2o0zGMXnp1mMtWjqfL9pYEZrPgCMHyQ4XzAn8/YdrZhriZWPV0lb7ZM/E9rvROx66kUpD1pexwakAvwfGpLSonHoq1p/9dRINFnKI6ZYuoWFlFak+6eOXWyoMke5mTN0oKbJR3QthpWPqL+03hysmpEsMgt+fikNZUBULcn3m9xSumBmRTm20NSmqGUD2gkTg4zvzBX9v4ZwqNRO4N/3MMj+gyGB/cZ9AcCPAvi7AD65O9jfAt0FgAri67uvfUPvXVMCEfHzAH4eAF5+43PlV3mQp6HlaKhATgvzCCBVv9O42KEE6WcGpAxGw0aEga5NvAP5kmOyJXeAcHP4EtkwZ+C0o/pGCmInt2oqyDZVyz2lybu0+nsXA++89wTjdEKegh13cmIc2W1jngYuLq4w5sDVxTPeQwYunl0SzeVAH+QoRGxANIyzW4jDhvNPfwr3Xn0Vb371B5lSmy5HlQ89o/zRBWuXVwxbR0FXSImO1Fo2wPRd5zq8/sF6XzBN1fmDjgcsAF3pV7bJQqVkXRPvO5rep1wMwHVk2+4zuYLEfhlW2xgYinvPOtQCLSqd6AdxN19ICZ0E6a2QqgyXloexEzat0CMTbVj+snmVtRLVXEAp5wiVlu92Y2k4Up0bDzDSrFPPgIxau20Xe/EczDmteD3LAGK3djzv9bGVQETcA/A/AvjzmfnwWrujzIxw1O7jvTLzFwH8IgC8+cP/Yukxako27jAtlMjA3YcnonrXexFJg21BIon9R0O1SEJnRk4F4qq9cxQKyQwWqoBjuIbcDfpuRBtdYjgnWO6bJCdlVYlxak+LBpOP/u9vPsPvfO1dvP3O+7iYqpGPwFBF4KNHz/Do4fts/PH0EQk29w44PzTg3h3kdzlFd7TA4fVP4YU3PoEv/fEfx+e++En8y3/0Kzi/+wJ6bLhKApCuvv0wYaYti1vnLrXaYaLLmlY3xmSfga5MzSQKm4LyMwF0KovonNTUpExnJk5IbMyVai+UPRhzpa4ARd6lXPRdphmJAjZ34RGKAaBSYfvP0JCa0KgCKv3W6FKl0UBrOMmo2MauqFuT8hJgTgXVkMt6eg3CrhMj9ROM1K9rNrhnBAOlvRRtdyuxVGxm2ovNogcDYgMqWjgiSfpSTQNis8hSIaZmMYi7AMh9qzOjtZwsYnve62MpgYg4gArgr2XmL+ntbxvmR8SnAPy+3n8LwOd2X/+s3nvuK/1vCxrItLOvw+IJw/22LItzt7ZY4d+TGfPvh0kjsihJISnqJQCnkPJEoSBbTflVCy12f26tLJ59WXecoXWVNWmJZ5cNTy8SlxcnjDYwOpHAHBzkMY/PMC6f4XQkgQZNUfWW2G4FZk9kso/di3/kS/jhH//n8eN//F/C6y+9iLtnG46nxFF+pl0oVo4BPt7Oby/ltyyQraMtmeiamKUaEkuEQj0P9sk1U7h77dmUO1fBRQQDfKkBrzRRy5jL93X2gVaN9fhGAe4BMDLZj09Bv0wHHlEoygchVdjTfdiRSimjmodMMU8DHPuOpudQ1qQowAU8aRyMXs1sdOtb339oJLZdVKOXYhfqs1ZHHYDb7i8cgTJ20JqepCTIaMwV+4CzF0Y7Rm0spX7e6+NkBwIcRf7rmfkLu7/6mwD+PQB/Wf/7P+3e/3MR8dfBgOCDj4oHANJYADuwRGNXGlt1ON0SVc/fmnxfBK0DAseT/Esm+OpALl+U7EKpDtJce+PxdUS7ETKytVRUhHuGSzUX1BuwP6pf71PwTUeF0Rw8nRva/U/j6fFXMdtEHp+xndflkW2944Q+j3RHNrbWmsdL9HbA/U+/gceHO7j41kP0F17Ev/1v/Rv4wS9/GcdxC5iJyxPvaZg3DsCNL0Jmw5HzOYcOIYV5pFeXYhnBLIF91tB3yzLmQhINOiB91wLO6CNRwpiTdfGrcSbKyrZgc5GTkBhptKRntxbY+gZ39q1zMCaHk4QCeLxzWt9g41HklKI3OpN7CPrm04c1zOpjeTIzJgxBDoADbxuRYhg1ZAOkzGSofVIoC1rNK3CdtuI0oNbYCo+1KllebUgLh7TFBrIIrYJTiKqlFR70HCFWdSJVWn/Q2RlDszb/IEoAwE8C+DkAvxYRv6L3/iJ4+P9GRPwHAH4XwJ/W3/0tMD34NTBF+O9/jGtU+TCHY3Lhc8rOBNSltl1r/RRSzTYeLh5ZraHky+4dFVmD5c/x7+dYH6pKtVDwLJN7D/txAUAzDYI2oA6Avj+1OS07TjFxfvsuDtstPL58TKE4HXH57BmAiX4WuHXoOCFxOgaQja3Az4Dze3ex3X0Bt778Vbz82kv48me/gAPOcIKQUMVFEq5Vr0fV+4SmtJAVGNsvjINUPoyC8Kk9Ma2f6KvofPxkFkv22l6uhvByzQK79GLWfQ377noOV9ZxTV3x5wg/ypfnWXKmQdZzh96xCwBac32vv7qQoAOVTf8rI6tUqb6fS7GmkYHH0eve4XVqgMwapa2t5Yass28hhSykq7SvXJ8JGRmhOlt5SLnlVKWtFCOggag7GSji0HNeHyc78He8rh/y+pMf8vkE8Gc/6nf3rwAYNEr2wc+d1WAp+EB2qMJvCVK1JEcsVhqigi+VFW/aGOR1YW7AGM4vt4qiJoKdg5qj5joJ+4MSUPeYiSHryC6xUcoJ6Bhgn/1+5ww//Ef/GH77t76Gt97+HczR2LQTibgcOPRbuDxeIHDA3Tu3cP+113Dv5Rfw5pf+CF557TW88foncNYPuBqB02xVeEJF5MAeFVFrBJdzDt2zhKeZVEXE0pzSi1Dv/4ULHFWLnb9qEpAhJpomLyOAsdBZi6guPC0aRp7qkCUSfezQlfdMw0h5CKncsly8WUw4tWAqqzvMnVcQ0Q02czJcSRdkcQoC8udF+KpckpTMnOTuW7lMKU9IUbnzsmXNhjg17pyZKlniNOYMqQKu49TyRbC9aUvLsQySh9zrEicpUmsiNys9mW4BGs0OYHOWQWgIrZEcds0SXn/dIMbglHfpXKv0AMOwSHiUWCtFWhF/B3kcLXWqRe9XzbcGOPKzPMzdNNUkonAdNiHbsluR8pFd/AEHY0SbDbWYhiCprE9t/wx84Qe/jNc//wW8/Y2v49HDB3j/0bsY44QXDufo2wa0DbcPt/HSyy/gjTfewK1b50CcEbZH4DSpLGcCxzGloOj+AKjAJ+qumR+nKH4gMCQoPicwhic+RtF8G09KZU9Q60x9OMFmlj2sWPj93rQPsopcF7tc3JO+uW5A9xpBJuQYmPJpWmNr8aZYjef1sstOSvj5fUfyp+9XCITCL5dyZwkZvTHMAzRYicSnllU4VB2ASpQE59sOHkFL3aoGdr9UtQ8ptbVHqe6WFYFyW1h3Qjmaevawoao4lH7BehHB4CClUq4N5ZrNTEIj+T78dWOUQL3C8JXplE2+jXPIgAHcSgHVihupBoNPsRMIw8JMD3IwmUg/ou/a9206yA3OFacCknx/ydRKYwUIHcncil0wKVgd1xK3b93CF9/8IrMCV48xxgm3esfh/AxjAGfRcPv2OaJv1PYDyGAHXsO9LKQBrQefQ54uhZWPS/cIK4gEhKeD6Vlt5ZetGplVhBKQeyQOBXwNQVU7R3YzAh58IRZnE4LwdXwoZSOR1x6Fw0Tk9i2ffwrOK+AKu31YbkD6OlJKVoJSMmUA6l4E4QWx1295gGj5QCWPWc/qNSEyKHnx86v/4pLp2PkiC5ml77J+2wHNXdGTkCtK0VIGAlQYzfsDopm2OcVrN0tobt70luOAtCyhSw8z2RopwLbCEWip0ZcFa7nxU+PFw77UiFr33uQ7J4D6rt4rmEnrviAeB5pOpbUiDLMGilTTdsdBKKWDB6Z8Vf12y2BFWQdw1nE4bLh9924RS5zbhbIKTCf14pbPAKAafTSgq99Uq8NNi2/BniA0HpNNODM1QWgG+QQ6OC26Ck2iKuiqtXYSNhvuttalFPwv3tMEkGpqYRdipIKuSfpwoTRAMZxcGgQKJNoNM6CxYvNhi1BrRFlMKamJXNkIGQp2nvIY+0REx8ilQGfK0qYgdDAlWX+ZqEpWv93Fslqxg4UYjGwGWCrcyWlXdsn8B6zMCDiWfkybKWVGmk0e4AxE1zqedvTiCLpgSGB2KbNOklZqk3pAtc25FvpDXjdGCRg0efMZyWW76qla7T2DEEA1GmHgYz1keB8lqM1+Z/KgjXRfAi5Qs4UUL52DH+gfVprMApe1RbQ+PTTdx8o+hF4k1KE5PG11k+UZVGByAiugCB3QicjOYhohFEf4O6hManhrmmJjuJ9aJzYLGYmavSCGaSGp/ep7Hdn0w2vcyhpP2ALy0Jmb7gPbdjRiW6bakaL07sy+YydZuy7NE4zFGEnY2fHf6Sbs+cwcTP40NZjxTWAhFMcXAaXOwN4Eo7DP/p54vzv2up5bKFTPNr3Dav3tTAta070RTXn4iUup6fZOQzYqRyOXmWvNZrLt+lS34OQELN6UXQlWITpVDilSqJmNZTSxzsKHvW6OEqhIsDIFXT3XFSjsYbFaWnVpf8A1/A3Ayb6e8r2Gf2geENPBRBA/N3a/RbcDRRPV3TEFI4koOaYk1aEyFHdMY++HV6ArgQAt/MAuhywyiS0A05vL7YgItGnVkVDWGaHgWuu9AknVfVmuSxPiSR3QqzwhwVnNodM9kWTCIdW4wzGZFVSi/SLaOBOsFnhYQpaAMyZF+5UbMaprT1vR/d0aOyIeOiBpBZ2tshWhmMWcg59odF/cbyLTJKGO8AkAkCOLcWro7FScAj68XhOc13NHIzoBgDlCBoZKMGMpNxshtTrFJjeTfJKVgaAoklpteD9zwBkCZAJj1QFkmP8hYppWKyCkNWfJnV0S+N7hTA3Qrw/ovPa6MUrAxST8s4RLAuHIaYBWuvgBhkYa/IHpqHET/HTgRX6YfCjFW8j4SqeEXHEHlMGKqIU30JA9Miau9xBrD9fNahps2nLov30/MBrZQeWddbSxSLCbrDe8UIQi0Ou6/MMM/w6vmjrsbsJUHfrDvrkvnnpOWdy267/nv6/n07Nb8LAURsgVYCOS2K1diL3ofXSW4npNgX/fHlwLB2zXc9NeaHDnTjGHcv12bdzOG7HDkIb5+qOHd3poSE3w3VlPKzoHp2N/Taw/z4UTfTWsfgJSIFhZFiapvP8K4roWIPghlW9V6s/qpm4MldPBKhwTVwarO9LzXjdGCVAVr4DOzKaNpnYmHciEgKVds2lrcy0wLZt/04toKVBLKdchrBgzuJhrE1sKL8wV9KsONiB9eYqzznoC7ip9TcceBoBNFE5+lWXPDmxBlkiciKnGE7JCZYUnhcCbjJwY08E55+L5vGbHZSy4joaClYbYKdenajMAKVhZ3cnZhWTM7eBxcwFNKNdiofZvpdp8abWSlYBnQghXCVboRYODHhGBLcG6jZbkiMym+Qmz/O+GqPl/RDksOHZAlxp8BUMjoBgC9ywB5NhZxTD/wHr9etPadJsvKT2mUVvxGkIB2xSiaAk1QzF7byoOoMCclMGYq1Euf1gZC+2x5yZmBA5CXCfJzEGIgr0a6RbUPmrfSRNOXfuaqv+e141RAlEjmDR8ku8y0JYwZwQ67TrrjrTyAxRCa1BSb7M56MLvrg43svBCE/YxF0UT6u1uiL4TDAkCx4sJYfj+9duL/GKX2Pfoay44XFHcAPnihoa7qHWEMyQoP9eYwHbEEWcLf6bHmMkit4UubIa8CpWVAepQNoQonK2CY8M5bSOQPQoBYXlr9phXrnvmxIYd+cXCDsBuoC0zm5VscgUMy7Fy89qJ8P0lXQbCBq1VzPLlqdjmsvDNnIgVSE3dse8LgSJGhZucGiHlUjIzsIyM7993mGs/jN6a9nJ4jxFCsKHWdBIGCU9hDN9n26dXARs6S2cRncKIVfvj0tcPed0IJWCZTD90AocAQhF/a7KhmvUREw2NAzMVcMumABgMITtZd2DaBOm5hr2IGmmhhK+ddWijSZBabS8gwukakS6lYaEOuw6sIWBFV7ueTtShCeQiOuWEJyO1PdFDi+PBo/6um16ayOSo+FA9u9tMG125rj9nrBiK8u0+YJimEzcMeHjFII8idaAhmDmB1n1YSEsdUkXkDWwKgLFzbpdiGwiiAz1jX4UFyEx2cE4gwYamrbFQjK1FupRJVu8AgDGedEzDSjt0/0XTS8UU9uvPLXXnJaOuhuWe2KWIOKBmRgUJSiE6cYdBvPthUQLnqsNG5mBj0GgsCUZwPuPOf2SJshR9M/V6Sk8QuZGqDZz62D1XoMVBll8GIQF3FQo7G38YJhAN+IFt9ltVlAEpi9/ks4VRX/ECqjyNZ1VbpvFOrrACKm1UBzhdRGoUMJehNO2z/NpdYEn3Q+s7V91LafGFLBzwK3c33B8vdNuCzUGBCbBsOhGVZ5epKjc1dQ+ZplqjEAjrHLgmE2SlAUv4ucyCirplxwvWv7MUSR0arB8wg42kIEL/RJQfb4vq/9MqwT4Q4wPax1iB3tQq57ohycWKmywfd1lAojxab26rGEDi+ucCV9phCK1ldVwqlLS7jpFaan9sddee2tW0q4oqsTaNHTDHpRX9eVHfl8wY9THX5IwL17iJO8wYmdZdQcc51rWJtCoatmIEu7jLB183QglweQSaspUGHwlcTqXC0NAwxfDjbg4FlSpKDZTVJVSX6NryKiqVYLQ05U+ac1DWtiRBrSdjj/RzJyhZ3XZCcYB1zDR9Zk7pjlgbL3ZZSzHaIutZmmYApLilTZrOVNPeWqWvgBAKlkDKgri2fsolCsHfdUKpNDysY2YCk4U8rnWfY5AgdJqMQltwg2s55QeEXKMNqweDT1yTLx5t0ndusQZlqvx1Aujq71igqnYw4O7EIyXUk2hDsX80sMdhF3xuzYrxAB8pFSJWKe/uIrt9RAUcHdzbGo3JVOKqyFjgSLk99KdfnqIW7w+dGoymsilBLkGiYUsrCsZwhlzbqAEprapfM2c92wFEEfO0skqtt4r5sK8j1J5cSvB7H7pez1cP/4xfEe79b0HdN4ekMLn/4MSyGlXybV8tE2ioKP/eghr2Z6ai3oo3zCxk4OBQFdqkUYj8OmmaiF1yK3RdKYCV6gqbMRi6VTOLtLKxC3T9N7xnfMYmPj31Wi8Y6IAgkcXMvIaU9m2rUGDJ7/GgkmOxyEjXREUKY1mpKP/fiKRlgVBvRH3Xv52AehH4H+wweS4r6/vxG+EUquF2wHES3gWff1Okfu8mzLA95e+uZqC2uuEHhKsuUwo39WdjorlDMUROalhTiC91N0sH+1m8547rWB5CyMBoEckuyFMZlfRSJmDW59Qe6MlXsZ0/uCPPrWiR7wHPfd0IJABAfozdAlkY97AvRUC/3vUFB6AknBHsVc+Pxlzt3g90M9IAxNKzIKHMfc3DaypITkd3TyLmmaC0LMcQDBXdnhodE3Oqrl7MjVUrTlVt648E23NlImdyxoF8zqZS0pGovDQQNRWXKU9u+TCcljtgqA7Qug35hYfY7PTYtURr7vaj0lX52X1rpbQQmhqkOAQCmH2xIwOL1IQ56qCMQeEnc9CfrG2XKwBrlLVHS8uKaxM6EEQH5YrEajdf6cApKpDuwSxTKFDY6kAmordykRCxs75GWOJo9I0EslDpsbWyUVzb1N6MI82NsjjfkALuArOQO5GQzCd7HgQCMweJcDgxnjU5QXtIMzQhW4O7vqktHij/QwqLMa5Qt+nnH70bowTC2tEzn3DdWkGHrogvZQ50wpFwjnioEshBu2IOttgJrBWLAV4q2LPQgu8sFaNILWpBcVtyIRc2dfDX5M/BcJ29CkzEscVITG5+UpNbY7tgZrXdSCkcC8/OvYAEu63PVm7dMBlRwccPvlqjIpxa++jUbFPoC1AgUK6UA7DkWYjOPbkio36zrVjGtYva/QIw3BYryo+LgOjY7FZkFLNvosmSQFlIGm/tjaLnU+ncxGphnolNqVMoVWtzK8zHt+RKAalDyRdHn40Vv4CUQ6w1zsnD3MwK81oA6j2pzkmWnyjbw2uUs4GFMDJt4srlsUK2EkkvXDoTce1HEB8B+G+EEkgozZJAT07U6crNzgEypiQna5FteZSXhoQwU9NYgC3XBu81YQmPF0pWz40uIgIxBzx+y1CLv6YJehFQArPKmBODkHoYInvGng6tNH7NWXRaEYotuNgGKZadIF2SMTlnlc/Qek2iBUXEGGicbEM2swwrUqWxh3bGrIWVK+hD0x/gjWV1rWE32woQyuHvKd48sNZW6TZTp1egT2iru0RDvy1rn22JZ8i6I4Acoyw6wkSrIes9pEAZ7uxSynUQ9D/V9ykN210XsgzKnJMxIgeftdZEMM4eUd3xjBFNnAkRtQjyOxI4NbqvXS7VifPMsWFW/cceye3j2I5pTRGHMEPxAcoeU7ts22YXA1AmJqm8WzBOwDNE5FNp87wev/jg68bEBIiuREyxjktvaP0nPPmlrD9A37HJ0km7Vu27vwMUBZk+++TfaHGo/DlweprbbWohP0HYHHYVrJB2D2BNbj8t6mul46NZKQWQzvFKhSVdFluoa2gDNIBNipCFNr4/W9IVnOKw1n10GJqh56vJjaja+bUHfqPuOWxAU/dkhgPEGeBqrmwA7z+DZKvypaFYSrV6433S8NtvdesyReXT3u2y1gEo+JW1xvD108Yyy1DwNx389bNm/bviK7LS8rixchtGVWt9LGN1kgMVO2Bwzq6LERXgCkcuj4LUuqHE2lcA3Mdwloj36rhPIQfFXJz9yN3fV3AxvS/Pf90IJKDnAEDyxeKrK0DiHQhGmr3wPndsEIqy7t1BmdaKR1+QUu9bAwN7qqfht4ZqJOMKDcuQOEINtEWNlX+eRQaK6m1ni0tf2iiNm+vCj6lahzkV/Z8rPeQFaiK/tGDjFTKa5H94AAAgAElEQVQTJVCZVSzWiqVHCJvoQjENprU4DpChYoqEYhAOnkpZpafzCAE5mFbvA3UxMH+TzVmeqZy71gocQkr0FkvReFHk4FYADbSGbi5bnAIsbkGVawvMJCECPEBkIFUWQH6J3SVnOiJClY6qaWjr/ZTSoH5S1qVQ11ZNUw3pW049l7YsLFPU2pkdp8z1HDmWIknua5c8nCKr3buWT2CP67K5IGvqsPcmDs2eCalgaeea/aEIDO7nt9NqUeA8UHJI4fboDIDNxBby8xHF0AMAd/8pSrC0PIwQpnxvf2enKGDhwELJM+0fLis3ZiLGIEqp4FCUAvNBCqxgo8kh3QFPkXvQaEG6u/E0WonGU8/fU6rRBzNlOlKlqGMEeiOs7qEDBAfNomQYmax8FOpoFv6IMnfuwuMFddaC64xra7YCbITCJ6GsGU0j5LkfJPNYcfh/dx2l9bbjK4XwrLhlSStroHubrqeQi5KqtKuOelrTQO6uI+WpCtWIZRBYmap+ANFI8knR1jW1iZSAJoJ0Vqq1N3ByVTjFR4Pljk2Mb8nNcxl6PX9U23IGGRu5Dp0uLpWaXoOEougDEye0BDa30B9UQN6Tlgpe72tAPvC6Me6AgzGUqVwbZwEF5WaEI9ASAkGAgrGZu58UA0zvywAW7N0HHH0LJt5kDhiIccz00vIfFKb6sxdevQiyqVtNaNNyl1oC5JBoDkImPOw0c9DXnxMxZ0X/eRiiDgog1FEHQmtk+ImwQ13GVhO3sMpZUa4NapWz1mimC5J8rfX3tSD+ZvrvLPgmWu0JMtzX5oOs56m18UGJWczCcnlq765DXOlT2IWym7H4BR/I6Vvpw9z6tZ+moceeZivZrL2L0KzLddiQTskBq/FJgJTrhrXrek2jjaxnyboOMSnjmOT8tQoVaz2FctYodu2Q4yo+B3ZJCk5/7+uGIIGoDqrAZA7fbaqUyoOt+YQIPkDl7cHpr+UfhC0QFu2XqAzZqL8JTwX/EmVdI0WAjY6IhhkDMyeOuWFmYpMwYHJsuSj8xSbjHfB/T4PlNIdo6ILnAMpFCT3QpqISyJ+HlB8kqKWsoOyDmH4tOaHGVpvC1HSATGZ1uo1uECc7ZQWrFqVlgDz8ttY6GoYYg2caPOpBgO7NwDvhsFNat6ZW4KnPOCCa2BqK44FgncGEU1iNAo1EtF6xC0JpE3DkjtShNo8+1DOC7xGNKY0IB1gbPOGasx0DyI6+U8hQhqMCcjADkdd1dyl3tyZyUXaodboPDYg54QpVwL69bj1IADpZX2WK1CiXTs9H6nZHNQmB1RvAJrwpMhzp3oPtkemiZqp3ZqBtMm4fUNj71w1RArYrO16T4F1FfGUFocBagodwl4mpX7HQA/bf7QtybBasQLAQxQpo2ebZKkAR/omdx1KlvQVtd0YBsmjdgpwSxXJXdhazNDzKMviDRECzGpcglZWgv7SzbvvUkL/PX0fx6Vduej9LYVkNFUHpNet3l4X0eviZIlANRRe5Blq/HSiTX+QUpR+R5CbFgbA6NBsvdEA8CWCRs7R2CX3jOpi9TtrSXkh+/Pepwqa1da0ssVupM77OdWIEXoHFQoRas50E8f8ld96LD8Q6/NsBu1OsSOFg1qJ3wRyUlX3xroQ1gbcP5V6U4kTFSFwPsTb8e183RAksPnf5ZogiekzVAjRRc0+wIK36f0CCBaDm1U9viVMytCpAw2mellJxwCaYegJYwOT3Z+TuAEdBZZ61WQUfmVYmIaXAdmNFuf2AMhZwxQaN28ysUWE+CHMy3806A+elGXgD2o6rL4EJMgopO7yXCpR6nVqXxaPvyzHwTX3oJMQi6AQUU5DiCBAlbI5rKBVmS1f3L2ufAHKAtF2oq3BjQrGr8ccqiV2Kge7bkubZ1OEJKyHs3v1IMCDZWIyFNAXXRsBR9FhukgTG/RIYNuXfDSsOPZcHsja0UoiqSeN3giPneCWVwLuWQPq2wk+UZAJdNc4ZOVfdRCzdwbqW1ZyU7fV2ygG7bEPJDP/ZegemSrabMcSHv26MEuBrB+kQFahipi5XJ5g5azOWLcidXY/dwsvHTJMueECau1vwF6smHQoMAXD9lX49sEpZ9b1ocJ247VTo9xEsF3Xac//yxk1ZlGn1jd1W2VrLIjBb4joKShTdietZhLI4SPEDTBTa3RtsSHy1VlbRP2Q3p8v1MIzvsf8e7zMF5+nOLGG3FXZWAkm6Nn3rXRmz8zOuqS/DNQn3w4HfQAbfszU22plItOlYwJKiPUoyMqlAafgA8/1NSsWxGpNzFIukEbJC0V7nTOx7MpRE+THkk3u2IpdMcQ5XKaKpi/VSGH4xp7JEovZTXJABoKW7CpmfofNSnG1835jARwYGI+JzEfG/R8Q/iYh/HBH/kd7/zyLirYj4Ff3z07vv/KcR8bWI+I2I+Nc/6hrQYtE6YVkSWddaHUDVWwq0qLEED5EPdSLHrJx4Kq3DMz+wBeffAYHcKPwedtGCPeA3+YQTVh68+IAOdoRSPQSENbOuyU9WpC0VfVe0kyQe4eQJ+n0neACHcvyNFtppMsO6jEA2dwQ2AlpKCkjxFxqOAxi57o8NNnctrAhAGVsIgNM1B8+xrMYm8pEpjsVliJBlkTvSmsgSgUDffZ57OOdk+/LWkDPQgz31MRuQHS55rSAZGK/owRLs3nj4MjTEFeyx4zqH3lJZEfMSshAcOR1R8Q/SsBOHFoX0EqszdE4yJjsCPQG3oHNAdMbECL43w3IAHAopEnV1obENvF63YhyS755iVLZKhSsNxbb3k41yzZdonTLa0YUumZIcaMo8BeY8AhjFWkyIjtwaWnTkXrN84PVxkMAJwH+cmb8cES8A+IcR8bf1d/9lZv7n1w9z/DCAfwfAVwF8GsD/GhFfzmqt8rzXCnABKB63tQCVA3vX+yE9P29d3HYFsM+6fpEvB31r2GhiZQEkQM6lR+zeU033yVbFv5xSMLomtT+rBzdSAAvm0rhH+cY09OnHB2Afbuk9cwmsGKsVVxYpGCZG+b+LhadrQHrIhVE+bIH131wL3mMHau1sSRuc6YCLFAFokEfE6vevL3quYDqoNgdz2onV/yMIsUft8XKdVufjnTxwgeoZI6LYod0xJOzSdi0W+SqB6AsJAiFiThT9Afv1uLaiDgjqpzJIQU5oNkAwWIcVVATEUbGkpFCLrqWIw3JZUrgmE70TkRBliLlpdyxMN1+odI6JbesoBCI44AgDPd1rJ+Xa6+NMIHobGiuemY8i4tfBUePPe/0MgL+emZcAfjsivgbgxwH8n9/vOhMshdx0JL2Wra386aiZATxMIcExJHYUua2VFrzL8qtNmV0BIPC7hQYkxI1C7JyrFfaW4HQbpbhMoXUBCxAYI+u9mBO9dTRZf9gaN2BLj7Hk0I05eXgGzBZ3enMx1sw5aFj57xBSABQrUMFS07OWkggXxxgqaz2UimiG0iZIwBCa+9GFzFpCY75JyJmC5ABpyGF3QvUbbKjhaxGDZwBHuRJ9109haJ3dSCVdA1LBSz6nsdpJEdczXYOWOupA8fmxMDZDKSrY0RrMRVCyzHltEhCym1BvbyKO6X5Cwt8a/Z1aI6ZWl5HawOeYKqxyC3UkC5IcwwKYKmYRVd9bNaCdkNlqcvYGK89OslBpe95OS6PMQMvnzyL8p+IJRMQXAPwogL+rt/5cRPxqRPy3EfGy3vsMgK/vvvYNfH+lAT6ifL50znxpRvu5tNBhdG1disyp9uAO7nDxyQ2P8n8ToaCUTJmsfiRTkXOfu+UTlza3bXDprnkBtFyrDHrvviwLK/gd16Pr+yKoiI6mXgFV9466aB2ERnzM69kkKp0q30ZuhRmSU1TWMnb6qIKFKBUKh+5SdOoEdtkI/bcHkIRclOlIvbIFdkMzMULDPdPKxA/kXHjKJfGiiDMQquUQxCfpqYnnoFQdfRGhCaOHWeXUzu3DxB87y1p3W2H48MNKfKGAmYmT+iw6PchAroLIche1naIBY7eudA8q5r9TNukeiY0Bb7sT31P2m5YjyH5EIagW4ls0y7eyWU6LBxUKW8I9/+x9bCUQEffA8eR/PjMfAvivAPwAgB8BkcJ/8XF/S7/38xHxDyLiHzx+713Btt0h4dOL8qnctgtOTCwClraOhiE46d75PRp6b/RzlfqzkuitEUK1rGGj3oj6t6o8CAGXpV9BSdGLuyPW3KSewCHZ/qwlvfFjTE0x1pzj7CozTUQ25KBVdNEP73LW2LOme56nQaVT0HRB6DkGiuQ0OeZ8C6DPYIvrOYEcSF239MwyIBJMPa/WdQr28syoa67GXkUDDlugO37iTu5CEAhU6y6fNyvBHoFz0L9lpoHIhuPE3AFqIagPknp4fxNdvAiE/Hkp5pGjni26VLkOpYeHINdvZ4D8iiptoFKNaEwJ58TMtcYRidGhArcQf2PqGRmf2RDYgFIEAdLCT+m5m/6s2rvRz6lsBoCK04BzilGZKfl7cw6ug9PBJJNQJt0Q5vtogY+lBCLiACqAv5aZv8Q9zG9n5kiG0v9rEPIDwFsAPrf7+mf13rVXZv5iZv5YZv7YvZdfqxspiyET5chw6ee2AoiwPxf2l+RNR+7ed1dXWtIhzThlFWrsdGNQzotf2rduBlhDI3eoZZbHXiimzGsoKOkghqG80m9t5/M2CafZc/zJvHZ9xh+UOfE9YBe/CAuIhAO07em/axxhbrorraqgaa4iFF1MYMkxkViPZcsPdslxCTIk0O6q06TyOpzH1r17IcJK2YFyw3O+WwG7gtW2cEZGzposeTED0e85Ky9hsnZCRYFiIQOjxz1ocUq2JfsCbjpURoixS+GF1n/YXdThI6KC5FLBYSMtpR09n2DPfK2CN6NV7XvI0JWeDc9lQCkjIoDcrfnzXx8ZEwjuzH8D4Ncz8xd2739K8QIA+FkA/0h//psA/vuI+AUwMPhDAP7eR12n74YtLnIKBbesVJ7KMhkpeDz3SUUgAFQ4tCv+GFyITfGDGTyflcYKHYrMEqpNDLmifNZM+qjNbkr+tuRhas0o4cimnsHMwbY1OCblE+uDtQ9CEa4H/e0GIoQccoGkJotFmwUPM4O+eAQ4jZgfHTvh5FoBmU0598bnmYnsau5myC/Ya9fBh5TBPg0YlwLbclOANgsluCR7+VDT2yXjSmU9ppuU0tolx/jgrPU6pJHi7uu3aiRXAtkSTX76EBuQt82b9gFj3IBoo0GFRfBhCmRbwd+loFJFW7H4Jkofh6wy4wFOPI+S48o+uT9D/WzDac9dSeNMVKzGo9zIW5BLxIeRvBpVhJDzKhizqeQ+0gA1+Yyrs9L3vj5OduAnAfwcgF+LiF/Re38RwL8bET+ix/sdAP8hKED/OCL+BoB/AmYW/uxHZwZs9JZ/ZYuaZvAl4XHC5amidFZzQZS1SqneKTdiZYolvHNqNkBWEMjdXtlGKgBz93fNQdDk11GXAIUA+OeaYiRhD+XO1c6W1tAINPy8OjyOLoMKwAeyN04WSpc11rOiIKuRkY1IIRdTHGVu2T2IkTGvlWSprDB9cgNwzzrkuu1rNpy98W2kCFrpkXFaVwZ5o9ovuumrFTsfQf8h/35viiOoBMpPLCDBD13vVrTWyEE71l0tJJly1/QhmE7tNG/hoLx+OV5S8w+m4kv1hHXD11xKv1E2TOvWNTo+wwy/2KG+9VmiuWVsCn0m4MIyZAJW6nDYI6sKNRRw/n5g4ONkB/7Oc37ib32f7/wlAH/po357//LhTVnTzhWr6qdAQ0aXb6mKKYiPn4RlB52Mgo4qK/UmDgxsGWi942oI61nhIIFg95kUM5FwdMgPU2PSILdg1hDRqPtfnYtUyiYlM6oCZwnaUlYAYqgpJwoSEoE0ccBl9RyLSHEahDQSam9euHOlFqHv0jCIx5Buf5Uc90ZeLIBWDEpEZ+DOJBMv19RB3hIDogJK6dGnFhyfs+ofpMeZIozV5swHoLIYCojMPPHHhBjmsJLIQjQsuwWiqVZU06fCz5IU7tyUlhyJ2aIYfgE3N3W8YnEGpqjmrOp092k+xMjEoXW4j4KV0xBtrVmeUrEsGNWIsNWS6A6AA98xNJZNhiGHirZa6OBD+X/HHpQp64HTnMiR6DGFIuQOKAK1mEfP1wL/VNmB/z9f1rrQAbShNNPKAtMlNG7HxaV0rl+LrRSUAZIj4z1XZ2LYomPxCQDAkL+0/1LCKyCoHHY6EJm09jlFLtFhQLqDcSwFkLvMguG6lMdCELYvPEEphbIi8QAqiSjhCUWVZD2KQy93ARLAFsz3l8XTPc3wIA2IIqNZCJPKKNaFYV5rZKt7i/QZVaDSDB1Z5rKsEVTc8oNrbcu3z5KF+k3D4QQymL3wzSdiZ8V3NjmgOYVTLdmaF6oyGOVHFxKahaoYc1C8yHEdkOxEV0nxmUJlHT5OEY6MRsmDQVzPYOFZcgqRS70LDKTcu4RIRlHyG0ZqZeYUH2taF6Nno9OdXO8iHd/zuhm0Yd84Ei27hJI3PUAt2FpHn8G+AsEOs4ldbb0h8ZzVnMHYMuowyDFg0TUPFny43fRiBRJHZNXrl7/o9KBfO3hWVnyBZphDX+w+cVBTffCQ0PjzhlBzCE7A1iiKYPbC58l58jFnkYgZh06E0El38wyCcXEWAhjM6kff4PQlMDGmaKuNQy36rp8ikHA2IlJVajR9cLswNMZZWD+w4h0t3NSFQu52Y9D6lvIID2uRUnf6LBrGWM1hTQ46gh55C3LjtRG16sW2Qqi+HxgdVNQgPdjrWI1LEUhs7BMBgKzALk7IKmuO4CgUpPtV7fy71ACTaGoJFqrHCKR4HXz4joZZWRPeG8VkRCK2qCY7MRIRtPo5FJeaJr0x63PKgWzsr+ng8gmpdG5fsymf87oZSkCHGOVTZvncy2Lx/wZQ5Z+toM6yfEbE9ZP2CDPrUj4iLkP1EnVZCdqZKT9Wn3bwq253Lasj2LSEgn3yYwMpuGkfDjDuyZJZswAtkK2QTTo4EF0ukJpd6mcKTaS/TwiqnwP1kwuP1lplcvw5B5/MEnzCZBcmPc/zXWtARWedyui3ZT2vfdpuXGo3vQ7rU73WWu/LncppJMWDZutcMC3XdUz8cUOXEMfefP3Y72XUt6S8HD1oJRO74ItQ0ygjsXodMN6CbHIbVzVg3Yj+s9ZtkvPh1fBvxv662lOIMLe5UE3uZZsLprr7URlTBZJTUNGdij7sdTOUAEBLCEPiXGuPZGprTvQeOAOY1oLYZfo80Bg8o8GDx1+PzDWNClmdclmroBZhYF0BVUaC6YMO9FmHCwHUcUlRMgOE6E3XExVt7oN84aCkfl2BwrlTCswRU3P3Rh7B3A2wDGUMyCVv6NGwYcemk/BvbQXmECvi7BxhNPY/ODhYWuvXxU/YgAROmGLxiW0ZTQH+Xoc4mtZ38lCVYp5LzxUKklWeKr02gWYm/fQ9xRXpfoVAt/Vz7AarZgKQy2LjKvXZdbCclXUlYKSzAsBUs1KOMWyI2EC+ARWs3b2RswadLOXN18iFUFzgEfJhIkhS4/WJuhijmbXuEYEtCPtP06iMbccLvBdzUag1GCYxG3Gp6YammoGJWJ2W52pH/n2AwM1RAoHlxxCeL1+K2p3QMKH+7DEpQAXFpAzkyyWwItEKSLFXv0guYHmq4Zr3IWLTpBcqgylYF4olnFRQwrkIYDUXb3LB0V3PAtBAoAVwNdlXrkuBWEmQuMKxWVGQsosyPcptQTQWNykjsqx7rMv2g1AKhcwKMDNxzBOj4+3ArEnOshgzVnPzlpq2DLoJPgPFeJMBojKEEI8UjucjBq0vLabcArl4c7LQp5uGiCb8wzXL4D8j+XuepFwmXwhOzg0CZkcKdQFIYSm7fQEWRU2x5yhfB33WgtekhLLozAr3rApW3cBqRh4wHJnTnYqUDQJg3oabq3omYlHY4UjC9dhUYGImOR0hvos3wlWYFUicXDcrz2BBTNUgRAS2edPdAVhQbdUXzDWvnI8gXz8ApEuIlw9qIaky1V3BDBc5qvQzEPLdRNPd54hhaLYQhKG9/yYicNqhBx8GoAyvDq5QiuURLh02O1A9BJqGmPjxHGvYqfAAXY5pdyknmmctSMA4UgxFHYaU6z5YynmMNJUzPEZN/ATZf+zWzSknR7OhbHutVMF4px+9ZMudoyUMONrN79E9cEPQrsEu5jdUYU/W/1SQNcAD5+68+1WqaydpvwDbz2Mu94a+vXfUhsRlPQth1TXTRK3lFpmwZBHZy+1uw+u2Qs/eYtXD0Hq7IavdRlOoel0noPhEyWOiiFkta1xHdJM9Aim2rS6O571ujBIArADWkZ+IHQ00VU5pWCkgFFGwsAn3zcENJWNPP67FR+tLkB20yiCES9C2yLKxtVNzpSzq7gIM4G1SQCmykDbaDMHoehbBYB6EpjFiKDejJiMJ7TRZWx+WRSySWDZX7C1+L3+OdnE2IGMVW9kguuRk6uBtnQGukSe0yc5CobZV7sDrBD8PsJXpXMoTC7X5cszNN8w2ZZkFh9UQ17UIp4ptMKZy0loyW9JwUrFNr7StkHeQHNYGKgjsoGOPUCFSFhsRkDg4wAhC9GnDoPtzELMLpVEo/ZxcyF0Oau1P0IhEQo1KlzGhgiONfUr7R6MSn3OSXJSMDK5ISV9KD1p2qaxQfABCd1KXkrVZmz2VWmYZcSL7HwIl4IUz624F5PZaOdeuAjDJmwcjWJSRblO1tw5AQTbY99aIMIXeY/fDlWpxMZEOYdTv8D49XJPW4trTSJCXVeA9Nk7q2ek6RnT5YvOP6ymlaTRg6OfrSZFNAcGIBVtbBeiUDYkaWVIIKhBVVn0CD5Y7985r1t1ZDx/0axvAZ0yoxBCImOqUu1iCKYXHzsYrgFUIN1dajvfleoGlJEj/3ilCRCEuKzn+fdTB9m6F90Qw3NyKsUcTMXff0P15x8P6WvvAnvDS0im52rVz169cU0K5u5MEAA+b8SPE+n46FczPe6q03YeFlkssKNl276azSKhOT9fOzQdeN0YJ+BRNEXzo9UDwkAy/BlWeOd0GTi6OpgqzqZLh5oMiL9NWNOV/daGBCaCx2YZ9XYgw4s10g1Bb2U2KqEUrqvLiKFx/lrAfaPbWdA7dhxtK60H5AKavLBmOYGfKIDd3QFqHy3E3uigM1Hk4SiFR8faBOlZYWqgzPhAgrNRvOa4yGO0qAlWkah/Am3A6L/TcESIuhenA9ngNj+VGNAazEokjoFgIhbihYZuJoRjPaaoZa8wii3Hy0ixlVZ61LORmBdT4/PM0kFtX0ZfmCUKVeM3X3lDHN5QiVXzHNSzTGl+xqAnHN5xZ4P24g7TR4UinpbtQG5UH46JL2dhUrVJ27V0yLc7+BX1trs5DTLq3mUGfcjr75dTr8183RgksQyoxlCXLsurrkBE672IBtpY8+UsLymc3+SgRorPa10MJkf0+psZ4ulZWAdc2HiAkm2HCUd02/5jX73WBNt8HR61ZlYdMfuq5i367+9k058FlxDtmIBVBW6ahrCXrLLqXqH4RcF0DZOkrOeXvy3xYgfpGc6KeQlclCpt5rfow/ewpFiYC2VIBSe2bu+UKKaAwCl/s3iQkomKavqKhcKqPgTatlMUAgYI6WJwM/z0cQEw3gW3YdSIQelsN5gqRTKKmwo7TKEbXEkd3NBsyBZLL9ZnFSgijs108KyWnyERvm+SA++qjPLHqDIyMminqsnameBcf5Dosvva6OUqgot5cFOd72VJbARMNwZk7C0vLoKmujemUGjvutW0hxlvDHDs/6gMLo5grA0pgzzryBgglWTG2GIEsz821kVlxZglhYI2SWcKNOckIkx856tCKC1fuiGGgcv3ZilCzSkqbuO+8TpdPXznsWM00HfuknFg42bhk3+l4yQzJKjXyO1qhmCFSlp+beswFWPoduwDayzF0sKxMxA48qNJtBtRyjYfKbTBaUE3ROvJesnX62KJPu7DrJCLYaKuKMTPZmk2uItN0VODhoi8pmplEUqyNIHS02TjOExJB5qmfESjmKSB0hqwGs0OOzVk46Ml6lKYmq9eC1kIxbtXOw4xSfMuiL4tTCDasFGMVCzl7Zuj4nNfNUQKwwOtmtTi5e0AL9VCtPANCLjmVoFtlFjAg7GcCwJz1iWsWDr6kbUiWRXbWcZHQvEE+uFJMgTpU5ni7E+4yPwXc4cCAC1iAWb6x78XBUFNYmWWg/0noql8TrDH91w1R3dRiP8YbgJSig2R7NLXu0HITUgoU1Fb74PbgjG5bzhIZZNBFNpF/lI0R2tlTwGeK344P+N9wZiPWQckopIRM+VM2FlNuEoOlbqJSy2yLqHt05ge+1g7llVxkfbnW2R2UbMGZgWlwJqf4LXA2BKomNWZSWDih2g2iCugZ1+/E2tdcCqKCjj4i2hNgBWBl/xTc1H6ABKbnvW6MEsggfAvBP7RlAdlQZKvCk611uUlzNRqRkAAomMnDnpiV/ptwntYHwrnvqhdIdsQ5zoFNVsK+Xk/5yJnU8NF2FWqSuglMnIAkqcfNNFjG2ouJR8orrXXTPN+Qv1cHwlZIrag4KqyjYQNEl/UB5cFRrj1rJi/djGv1CxNtbgVBM0mPZcxFa50TpwZkuFNyKw5/HSbEEtgkWnLDLfux+y5QhLcruw4TZ6BnDKY7u1AK6xjUGAUdJ7ssRYAydC+1jZyKxkegYRSDsymNBlHAZwxZce7P5TwBYOv3DhUFl5ZgDKGUhSTJswjRStzsA3FddfgZK1iurZUEeRlyU06iWsVc8xJNAd96IQAaxMn9TV97IHuin7gvrbla1bEmrqNJcR/2ujlKwD6aBM3AZ9ecqSCnCTblt+40vT9HQ7WLtgLQTDNIR8r1WO5B0ZUTxcHuLCSD44h1RcNg65MyGhLMPdKwr+xKkKZPlR8rS7fT+olc7pEOp12GCCvIxhiRLJdUwTWLL4Z8X6EAACAASURBVPUleXbaMOsjrkcP3SfSOWy7JLGgcYaCcgZlq5iqev+F3OK001JQrs6Id5QcEClnW/XkfYevt1Mb1rNGCoa95n/QTYAOXxj+iVS2gsSYy5IPrO68kYFsgZwegy5UOcFApix6WXFkWfhCDzYqoffKZfKBlPtVD2OpjyXXAF2v6gepIS36rzXXURkOp1CNYDPgtnA5h2T/+Wfv5iiB6aQQX577ZhLOzCF23kQ3tNHgCxavMCVAwRwMVG0q6bWWxppA2xTFcm9+APz9oIvRGyPJbU5Z+xDkpvXuSAxp7UCgCZaWYHjRY3U2rpD9zm9GrnwzomOOoTQar2cOAWw1g1FtKN8+pisuA2N63Rb05VoCwCrKyphwhDmZsEZguR4zOO8+oEGWPsck5hLJKI5Q5J3gWk35tDPY2qwJMawATBYNuDmoJpg7chaPozudl64ajPonkBVnyLAbshCHfX5APQkHadCmPmwyAidQT5zJXk8rV2c0FKwdYDq1ajyg9m45GNnX9zJEeHKreRukOSt+xRUgI5TuzMDQUaZyjeXCINDmoAyGc2PMUGWy2GhEYpuWDpAa02y4RpUXV53Vh7xujBIw7PaB6FoMbrC1XCqna9i/aJ08mKhDFnXwl+hEBKZHfksbz/3nC9qjYHxAAiur2sUqozKgwDL/HSiskM0GXGWgfC4SclCWQVk3KrG27sP5hH0sJxSBN8qB7tHmNZCcOmOlEn5/j67MseTY8BqXUdZ5SjcZlchm1++1VRdhCFRrLIWjQ5GFcvw8KxpebEjFdXzPYd5D2KhyN9z+NEwStnKDeViBiLGUaTIQCECxPX5fbxVPwIgheq/9cEcqIx+E94SGqmo5FJDchBK5pYESoEB9LmKlDKvHgtY+YZCi3FFRlM0eDWRnHMEp8ww+eOuo9XDnqZ1I8D3HpW86Ekgop1xEFpQ1GH0nNEX2kL8jwWzeQLP9Gss9i1xR19kFBAOI1ByDsiJaUAkNCz75ajqAC6B1kJpn68pFLzisZ/DusZe/C0iYzslMRN9FcyU9mVDNPRSzAFrOCi2NXIE0XgeSwinITqGYcyiOEKyZUIquZdehlAvUOWrdjlGDDzMDS5GmPU/67OVmMcYS4WYd5GuwlCpL8LJWjWviANeKiHNfekS1O0u4qjBwmkNDW3hvbPahZp1QwHKKsrxLoaavh5WhmHMiurosBlOopNzOcicA1B4FxEQVagQYezAc14dZNj1RAUumUhmncXcrZKBNWnSeXZmnFB2o6mX4LBMrgDunyEXqMZmNma5mpaB6l9Z5FjxKDWB7utIsH/J6/t/8M3zZ2mRwDlzlQ+V3AZAvugJPAQkrZOmC/1RnWqyBnTQQ6z8qhrCL5pb11Eev+aL+jE3CHtrqnPudZZn4jskie3VUo9f1GdNXKxKf2E0PXn6xn6yCY9C1hFKK3VoPKjJLsCa+JdYsu9zJsD4/M2sQBuDU7G4tw2ZOMRNhrLTL4NBg5jXLE47D5LquD6j3tg6EqgrtoqxHUavt6GD8pDaq7kmeibdFf5W1ZwYKlpwsH2Ll4peYaD2A7xlvXpwSZDViKYVfYuO4iuNHqLiX7ytVGNey1aTtArnBKsoqidZzpgPCvkpgvxNwCzRLS9jF28PKD7xuBBJA8EYmyB93NyHaKhF+9Mr0IeOY7Tkmts50nDW7DxIXLyp/3cOMxFaoYAzCv6YMgmXQte3rINJ1mK5Lj5VnrqOhfGLuNgFwnwK5AOFI/iqtzQwc50mwvIl0Y+hrpESB77FLHSGwz/cnoCIZXrpgtJWacasX3f87V486c+wdLQscxHNnnCR358456pD1nzvF1hxtAzCn0pwxyv/nT/gzcW1vU625clBhbZrHwAIfftDu4kqfbVKsqx0Y35db4IrMUgL8PWeiAiSXNZuXnMUDqTl/pQhccZmVjXDNg5eURiurJ4VnVcwcNV/CnY84yj1wkEmb6CoPmYUYWmgqYQPjRhGIDaqC3UR4WjGrOa1wGBNoz589cjOQgJ0YHka+YRJFfST9wVl/7+k3LTU92J9Tempv7WyaVhVaXFPcpflLW2cdeP/9/n5KOwcDbgO7wRc7FFJcA/XUL/dBv1IHONYa1AHz94Fi2g39ozfrc85g1HupIKEdbBjFLGi+Fr8V2tj/Hvvjz4XMpHhiv3BYOW5JbB3uCmi22mS5BKbsXLdONLA8dK21a0htIbNQwQ4PpVPI139KGSQFGUsEds9teVrFOU3dpsHAqdPDpXT3hijlvrAPg/0/LXutFWAWZdb3Fx8gi+CVTVyOaIzbRIiktn7XsjVcJq+tbIB6UcjBY3073ZnJvQ+kGJUf/roZSAC23g3d0Ew+l2PdhEX0wUeQsbWB/r/DTs3z8OB0zKrx9xqEpvXkYGzBTSgAENYqxZRjqnebg1KrHJhfaJVHb2x4LcVCP25OsLKrhMGw2QchLfXspivo3WK19mr+rHL4I7Oq6JwZyEqVqbGHSTVaP84KBEydbqIS2wBL3jFzVc8FSDoCVvq1FXNucl2TJczRHa/hIV+BqYV0uO66t8wa5EFLLaUQwFSLdR7sQKLV4RlaRwfgxmSzVA+odSCzHMJJCXDWaJUHUUnvKc6kAgNbO0hWqKwnRpmdQmEBRHRG2xuoaKA2Z0aN0bBqe5uSIQmoTiGaGuka5SlYOGEiGBXdFO+hCVUMAGhxLQBOhDARk41fBlABUCIA0Ym/T/3AjVECUz6686zFnpKk8uCAi1XVbhJaaUq39EI48jr15x3cRLC/PZavtuoGclmqCM2cTyBWgHAPUBjtZaCvw9YJYgovjMEDjjoQAuyyrIT/kYb09j915HR/E5MKLLnpuX5ah705uL9iFzvLXfB1JtDEr6vgnCyO06YOfIFJKXcyjsYJxxNAqOuOiSyOXDMQpjUywqmAK+9hX5lplqTbqNVnpqGsFSZj/D2s9PbruYK7tcepIqBJuXKKF9iZ61qmFBZYVh7ArsQ7losEwJWVElfYVXC4tjgAcvUoGrlc2XT7dr5fdgVLFiPoyoxkP8mmArKYa8+zkUFAMCe3CKvZCMVAQ2V22O+DrxuhBBJqaWVhDgl/i5V1078i3MQ5C8g2c7kLdirtpu7CCxDSwpZfKlJFLn2vj2njNBVmJllakShLzZdJRxBmsxBY6YhCK+Ygv7t/yS5NwP0M3K/f0/gmgJyhdFACGAU7ozX0ahzR4c49q2ceFZMLqiBEcZqMGHdT3RTDYJGOkZMsL1KNT11v3ySgqINtdRHBOoguIDcUTTe5xgFRR9m1BGAL7hXwRWvXavinhDmwApcBB0SXK5Dp0CTQN65ncSNUgNTkLhXKyF3wc4pY2IRuKp0HKVwH+9wiXAd9uitkQ1OQdIry2rc1ZWnfrHYmin/Q5VK6sGiC99F1jxlA+//aO79QW7urvP/GfNc+iTRi1AQJSahJFcSLNoYgEcSLiK3mJhZSyFVzERBsC+1FL1KEotAbC61QkIpFIZXSxKYthkKhtgZ61aSxJjFRUr9YQT9SU61Jm0L99nrn6MV4njHf/eWc831pztl7w1kTztlrr73W+853/hnjGc/4MzeFzeWhT5Ek5ZE5R3mPruQdKSHvAKnaX49q90IIBAfCq6ETvSBLsWal2aKqOIDDWWtKK914y6EMvxp0LzwEeZu7kQvpnJVKPI6RiUYPiUgqBYqMbEh+GhCV+VPQdLiyLlwpAKmUoAklnZKEoK7RQe1fmp6ZFmJTa9Q4PIhxYuY15GTnxMgzV/igzslZg3mFUMZYz2R+ZEQu0lOa2KZN9Vkq5zjGETobUKoo5Bos2FHXUvlt5i6TZejkJyvQgrhOJIIyP9wPOb+ZBCMdwSCCeE7GcBVoMVzalP7+ShrLcpe1q86IcnSIb+41Jz5q3fB6FxUfQdPRIzwmmkePJ0JlCljLWB6avVSyLNqlg31wLdR6OaVMpKhNexVKVXY5PM1EubuLPKziK9ozck3sQ16GgJjO8VSQkK5xsiJ4SHtJYjAiXhkRH4+IT0XEZyPip/T+myLiYxHxXER8KCIe6P1X6Pfn9Pdvf6l7AO3+yyYzbgy3dLVP0Fktod00pBKKxqjoN0z8lOUehmdtuwoGGkdWT7CpYPhmz0H3p02VesdRfB00tNSjFlLAcA04PU+UNmOgoqPuQvVFiE8aJPqXgshm8u11WJFs676x/PQel9CiykMHPYapCFmrPdZ3j6nRjXpEvNo0sZassfB4pjRt9TMUbL9cZjduVfMSqgrkDhE6mMWAvQbvhqCeHhubI9ImlMAbZS/2GOa6dK8zPez6YcgNyxw6fLY8TisMG0WzhkjNkPQLDPtDf6trbJGKpcgDXDfnUSjV+suILLxeoZ+xQgBWz4xSjVQc8/G4YKGX4x34U+AdmfkXqBOIfzgi3g78NPAzmfkdwJ8A79Pn3wf8id7/GX3uJVvN0WiW3D7qEs46pca51dJsZvP3XRo2gtjkdkGTnXCekxfOe+2RDElvD+jyLoirLy2RXt6Gx2iPRQePsIV+bESUK/BBRAXYOOgkgvMcXO9A1gIZ1j6Gx5kiIINrqoBqRsHLev4VWDMziHmqXs2KnD4Lxm9xUgJOjdEMVEd/sHW56kI3ZaPuzHmGqFNtayPbplww2IZIyoU2cpN2VKEMdK19In6sFtcAhuofSvtBpca2LRzSWoTOb9UMb5BbAmcixcLh/Hm5ztxLzTtBQWAnECjYK0cQD0ZpwwjiVKZQ5CTzXHUnBk3AhZDeKUYVUBlXxHYzWtL+97qdzU5t8NYnU0JyeZ4qniVIlXtnRMdcJNAH5N7QdfW9a2ZXQD6NjbGFKgtrs0eJiZKJK7tx9zQ+or2kEMhqX9GvV/qXwDuAD+v9DwA/qtfv0u/o7z8YZloee5/62QEzLA3kB9R2JNOwSZMqW/uosUwe1pKZ2O/dn7GPGGkn55VbgUWVnPY58qHsrFB/mrwM9LmDTctUpF091y4m3Nor+nm8jA9aWch46jQjlePAULfGwr+jPVA7wgfflM0ejQBSm99KvDGMN7i+U1mIa9ybvPOHJXnM7K+nWRr0Bh9CFqwN+lMJdaRWGIkd7qf5sAY7ulQ1OgcTpm9YgjTcPyG5LMUxKBTQZ1am4juoTd8pvzbPdAkvr/63el+fTfv91waL43M0Yk3W4bglOo0IQ88gccqZYD94RNB3UB/2OYt/yAOymcVlbLrWCjhTx5yV+Jit/rI4gYjYgF8HvgP4WeDzwJcy86yP/AHwer1+PfD71ac8R8SXgW8F/uiRN0gVTpA0rUQPQd7c12slURiyhmDa2Goy5/TYl80qtM2DFGkmO9HJHWhjQgUB2e51yK0r7JYbqlxSrgtXzHGd11eltxTaqhkbTWgmncc0JYRUXWbbxGmYkMtUVmAt3iaCZA50GXZUC38M5nlN/PGwTy/UzunPLK2FEFfCqkC8V+qrtVEEg012qeCsipIa5laEsnkcPSCJjoslYvDCTrk3IzgDp72QXQoZEBv2LaiXEmjOZCgEUIJNZB11cs9ZQmk4aCu9Qf0MqsSbU/sguM4y22JaaJcJM2QWOER7nwXzzzrHgki2GYy4whWVaeEvobsXyeehuJKE9njuHM6jUPCVddCYKU+AAOYoVePtlVkl167G1gKvEObewWpVI0Lmou+VrAN6boaT3mgvK1goM/fMfAvwBuB7ge96Od97XIuIH4uIT0TEJ77ypT8yemtNtrS6Nc9yt1gTI+0LjgSMjnhL2ZUzNcjTNtcyJ3wfw6YcJn9SdQrKBZbSNhoN7EIsTUvb6LWMrQ3oc+jDtp60g0Oc95wrv0GaIVpjS6NJqCRB5tISHUTUaED/cKCUnmEodkHHajtc2i60KdTStf0FpSt7TxpHz+wMwdSmJA6hvcg/Ljt1ZRhqNKIw2UIXQlIsBDG04bNDX8Oy2jpYMjZvhIijcajl4UiACuDaMUlsO76IvC28lgID8nQ0JwY+s23q9rBIWLia8Ui6eKxp2AhWDgErYrXyKg5mTSoaMZbyr9OyFzYMmztEmcO5AsYiTApnVcxS/oeDqezWvMHpPKR9TRGDmfkl4KPA9wGvjggjiTcAz+v188AbazDiBHwT8McPudbPZ+bbMvNtr/rm1wALpoVtuphdgku7CU/9gb3BjGvVFJwiuUIQq2yw/n4e/LX1LhM6rbTSidfa7UVoO30vTQWjglvGxp6y1VkFQQwjdwmROntQ5bn03fM0NJRW3cqbsaWsF+fIz6qfn5HkKEDZhJ2Ei8NTF+4emFtuZBC2lWnCc5N9usVJImyZUTnPEggo9iC02KJ3XSpFN+cidc1PDgqmjnC5LV15bpAb5vvxeDsHgvJADPW9BaJ5m6gQ4Do7EnIk1wlzFpbZAnQiB3NWGu2eruJfa2g6KIneqpjEQ0ZYmexZyWOT4pIATrUwNgmTEaUHcltKhSFhGBWYtI3gFBtjBnGm54FZRUU3r/3YKjEqy1PCtoRlcRFIwUmhJLQruZXQUkakdNRjSo6/HO/AayPi1Xr9DcAPAb9NCYN362PvBX5Frz+i39Hffy1fShRhmLwk/bI3i/FpppN1GMkNvas3XQSyNGgeAoiwnMamZqZz7Fv6kMqJtwAwXK/np7mDumAe7EVfU+MmFReBjrLS/XKWvRobPrrMNlzb6L4+8nwEIthqqo9lrJYm7K1rtXQQFNaPCyU54tAFQOrLjlX35LOurfDTQltV5RlkggnWctRA1LrbPFc+CksIYXgDZcjlu8OYgrryILSHPTUXFgQWMKUoss0ZfdoatjWztnbo6K+lZntewWunPme0GYfnKa+T0pL768cxdm+rnkKi9ZeFGFQhv7mIrpdwRKNakzNW6nOHf2tll7b3ekJjif/S4d1WAM0PPKK9HE7gdcAHxAsM4Jcz899ExG8BH4yIvwf8BvAL+vwvAL8UEc8B/xN4z8u4B2C3XZV6qsi4UXUBAxVWqI238qiWKAgTJhFca2OextZkTfENZYeShtESFJFcCeZiviDM8qvC7cyGswCTkwRE9dVwPw+jbV/9KhAkeIkW6oAhib+ClmoyKXMWl9hGk12Md9qXt+Ljde89A0eKmmjzY7lg5yLXQqm3uwTv6PGY8lOXV8SuK1hFxAxxo0N/ySC2jcw6nWnEYDPkHxAxO5S60Vh47rOLto6ErrFu6a1/dZxBKQfLndGbQ5sqEMu+cRKr73qG6cvZJGIVG8l0SbJs12eF7W5EVghyZqWY7y38VRVp9RZXcj5YEMR1ZQWeNwWQMbQ2dpySbfOk3N2OlUg8Ea7xOBzxOS1oyuXYyq23R8pzNInTo5HASwqBzPw08D0Pef93KX7gxe//X+CvvNR1v+p7x1JasSxNTBJGduXWIrYUYacFnQkPxoYLLqozdQm919rSEj+1RSwMWHCrOxA6GxCZDodr1KEvxiLWAeBgm44/6Mvp72FJ7QMjripKPbyU0bNlE51bbE1wOZehtVCtRbkVExuYqf5bUdvM9oI1dDkdIH+DgkT2cy3QPkQU8xsSMOl+b/VTPuk5tnbpdl+tpXL0d48emQbs6eAcab8D7+M5B7M3GrNBa0BjiHoOjXXEqmOQSYmOo/lDX7sDvSiiscnRfl/bM0SDRnaKdhzv2WMlI3GfxCmbn7DSsZC2aVL3rUlKSb7Y7IYUInL/CqSt8GVd7IZKicHcH11a6F5EDALrgAxH2wk6NvNljTCL9OlRA8z8TIoUGTm7+AIBc18FqkZrHw2wNkJwpQ3kRaMIrJC8N4wfmlSxr5CgVOYCKoZ5tg1s1/rX0rbos0RtHh1RXxt91gqIhJNMpEEHjWOTgBjs5yIpp8pkn0YlkvhYWmtthzZEUryCirHUMywtV2aUfObebHno07D2q8V+amQlzZoyAGZp34woO7jyXmVuiKdARF2qGMh5b64mZQ4M/eySBcNwv/p6zrNIPnuQ6pqD8uFPxXaYYB0gz31yzlUOHJCJBuTORjbK2qfJzoL0OezNquKoxQEILQwrHtXHF791vqrkqG2s9VjjVx87i6MYHUU4VOtxMk/RB9cg3BEgIhtiViRGkYS1Jq+0trsYi5KjHtbujRCoDWJNCrWyshdFQ8bMPim3a9yP2kC7o7D8fam1Xlh4E93UlAdVBKx89/6TuxN2uWgaNMDkDcRKYrNAATrMJcxSdeVZhN1Nn0It+kI/rgiAbD/3cLm3+lAN5ALUOG6sU3VX5NIUQx0Omzg8t8Ypl925wnEPc6K/m4jSjqU9H4dPDxOlPQ92/VkkZz+yzYM1L/rvOAmCSDEKptPml+6heR7KPdiluW3KzI5PqOts1sJCQcvz4g2vZ7VJozkfPe9yqpZE1Hxnoyd3us5+lNKYpWB2Sk4n0Zmc0Y+pk4bSv3lX2KT1devdGcFJ9qN1zBG9WHg9qt0LIdBjk5OxqXLMVBSbTmJzuiwjO4EIoomX0hIGQaPXFxShVodZHESMbO1AcfOzQXPbqb1JMxnbJs1V95u6acE8LXCtjjJtyp0XgnQFDKIJIe/siJWZBw6MLbs/rZH7ASn4OJOMreGjg5imteAwpXqoh0iIRpgM5yNImy4ZuISfBezO1FkGJVQqBdbpravHzUorGCECTtKs55FNtkamClxYXDjWwOnLszdzJlyL5GsnrJKy6gTjqPVhKGxhOutY7pNeRwZnjAZLkI2MugZV6KPPtVQpNo9NCbHJJjvcQsNenT2AOeWRCIVJRpc7S3S4TJrQLcG6RfUx56APj/VERHk9xoTYok1fk41bDgWzueDNiX3f28tQrk4LfI3xfhTPN9u9EAJQi27YdicUikmvlQ659EN6o7n1wrFPvLFuL7fa5hIeGvTWw55E7HqxP9xfXtfKdE25+p5hWh4rA8uWtElQn1hYoZRcLJMCw8ilmUpO1CZxcQrHQhSJVkLEbPVmN5c2+Q4HQSYSsTVodaVIsaVJE/NQpYWnNHwhC7HYSlu1qUETjWu0l7gN6a91PFwJ6HUdhAIs2EdmZ3CCwl7DUZbiW+Ymb2QeQR/L2SPU4tiFDIYPe3VglDX/Ng4nKhXKSHKF42YoUMwVrJAHcnEE6lajjRWERiuLofEsNSXhwsZZAnAs5Y8hR6TGzsUDM0RK1vjXd/bFoRqpKS08tDecYfqwdi+EQEE4DZILQlISeWQd92jXtE9gjhhc42wsUVZeiOmgIE+R0EKkJimJEMLQBqjDIjytaJYH2yhtszvuNmen0dakonvtuFZeL4K9YKEz3IaQwr470aQE2mBA7rIpZVOnFp+EjHPbc5abNENFK8TCl/Yu1FBE1AD2QgYbguw12haAlVlXwmqXIhqZtVla0y6hNFMmhzZ0Ba5AcQQSyGLHut8pj4igvLmcjMp5qAChxAEyrmcYHfEeWvjBnGdOUQLKB9SSNhlFuOmo4zFrY1cdBkTi1UOe8FHns+tQWEA5axTlsUzK/j/Gl9SizSZBUYRpeW02tjDhucxZu5LN/oTg/3nusNVp1ZlUeHqWGRgiel30dcya/11IymQtaLynFcWKFKwzFIaibR/e7oUQAESiC9ZO+9+nYrZa3pb2iSI8uiiWNHrbcnq9y00zYBE7Jn8Ah6KmQYUWamuxaJB7MBFqQmai48lZwicW0ljMd/ZZfmbUx4i+L8SqRoPtwqxFWLuy+qRdZftQAcseEVzEZ6LFKdPA763xo4FHC8fNI7wQiWM1TtmAsjCS3F9pJBWhAzu02dsJb9/3wQMAizRlbQ4n48ScbXBnrgg8o6+QwWuElmHYrfENrw+Pbwm3cegPVJ+DKBJPfIbjEiqGQV6gKY+I8wCsZTO7ZJoRxbR3IEpB5WHt1GOUkjqxMXc9o3if0Z+p55jh4uqOW6i/1VitTW/UU3REShjV+05mXX2450jArVw88oFak+cSAMzoBZ1M+W+1gbXwLQDIOroM5nLtxCbbNNbASGvHwQyozY6gsQcwCtSm70Xb2qvEt8OJPVFLIPTPuAmLfZ6Aa8yZMHTdPLPopFxcXW+gKgyRK+JrmpwCLcrw+rNo6ufyZgHY9mn0WWM9VJ127lxRG/QFKt4+GxE1AG/PgX7zo9W1/IYRlzb49I7JiuokJyex7ruSY8YYFWSlzWDTKjtYKpsE7BN6R6Gm0zgp9wMswk6hfkzFDEggzy6prrU2TTAXHEclwVzRpGrd1PhWlaqpIjYQ45qKUCjTa5/ZGzIiYJxLCeyarxgku1yGtfGrjFkQu0TlKM7fyU5hJty6Z3jE9/JUke1NqfAe1pp7SLs3QmALB6rQm6BY0qobt8+dESfLXuX4L6m/tG8hiGPZgQwFXswdx4lPC4rNElVG3dG/ULCiN54rw7gSzbix8WvHOoS5cTRrg3iH7HbdWftABR01bAFnKJ7n2gxTseNjDJy+1sxxUAFNi0CowBrZms0lSBvOKMJ0CFERdp3Vycb22PZZDp4PrDF1U5Uy10iX2aCYD2ck9lFn8lPOUXkdW89FynTIlZ0cPn3JVCld3qzQzd41FRvZoeq/w8pCx7Ln7GPkA2l5w+U8e0WV8A7dTWPmhC5UwSnkur3OOjXoJMFRKeJRrtGoOkOp52gTYgzybK8Ohr+8kPX5k0J75zzXfUR0Z+rwkU3ej32uMPRGEW7nMiMZ5K7PZIezPLR9TbkDT7NFIhtraSuZQrVQdcJI29tRkjxzQS7Hw08mHRoRZnhRkcdajHadjGlXjTWvUMgBaiU0wVYuxCG43gC7icgSCDt5wydTASpJcDNko3G5jklMMiY7Oz4lCC3OA8/ZbdmxSChF28kLckf59seBBPX70qazS1tL6+YyvSrPwfnq0snpvs9V0CMNEFaQisODtatxVOM8kLylbREpzIq5Z5kDZSpoJUdV0nX/HJ5t8rHNDP/TmA1BHaOj9KLDpkYJ+2hlkJ1ZmqncDcV7Ok8EwKdjuUDInqHvBG2xS6hPBU/lvtZGUKdaGfMg0yQO66PnyutRJljtgdScGWkJde2+JQAADEJJREFU/fTalPlp2PCQdm+QgFdSZBJ760eReQVr2vZu3zgdusqsAJIZwZxVCXfmrgCfTQtCmXN5ZsTglFmlwKNiDDogxf0ZEiwjYFdhi5x0QVDbyxGAUpAzmUOoU5tjL+5H+iakGRNn6lWR0EoMCraFQFK+bOFcK9/p9yliboSWZEwGQwhAYziGKvNSgousCrRGLNLWcx4ChmI5YeWgLDfhnB16bKY2okJop1xUp8O87PKzb8OC2Qs6iBn4vMMNe2JqTIS38A6ac6pCsdl4PwvEaRDTCTQLGVqZWMiSZubNPFH+vdgo3L2XKZGTbatnP08J17EV6TqT5NzI7RQid284qYaE/TqZOloAVlfGafRnIatUHSlhPziryvHQOlN5GXJWOHd6zWUVtDlnVoJTQpzMRVSl4U64e8y5A/dGCLQNnoc3wsUt5eEV5HaNPqHBm5pj2pYrkJc99NJsdo9JrBpxOPjHpbdsHx9djWDZqkWK0YFgnDRw+W9TfvJCH1U2zhC+Pdp90UK95W2YNm41DMdmQWjN58zA9GCwvBP7PLdmjlQNAj3Xka+wOdAK3kgMDjEUFVE3dUhKeoMZGRw1LM6PsMtRtrQ0t49YqwCam7ER4XkAnN00ItZZfllmyzYG+9y7InRTpoa+PZca15mVt9B9W/Pmh54ZRbjalrQQ6jRrCRSRh+YQANNGFYJNcK0aFMtWKdMrR2CPURiJxKRDi4WQHCnpzT5zjZ89UY5D2PaUKRk4HboS4VjHst1/JLDsetG+moipYplb+bNF0lTeNAKr0WaDpzjCbp6aCJdcjzEqhBXYx2T3GQQWLkj2MCssmOIOar0UQTa8hQ/w3y5At5LsdTjFzMmMnSA5jSt8iIjFEtTpQ1UDZVOoaJQMmFTobwh25sG+PNi4mFi0posSgNuoDVMOJ6hAoaQD9sQkR1bgygxIFQ9JEYFVE7QCrWbuN8K7g0JVp1jVm2v8aZ4kIzjLVfkgHqBdXAkvaOKalC1UcUrAPv1DvO7yjdTczghBeBO5CVMJRHsJ1LGVEnFwEbp2mY01JmQV9HByk/f2JpxyzgrE2nocghzuX2pj1kY+U/XVtnElpbMrzqPmek50JkiNzzaC6ywj8SQBdDXkF0goN6bMQUXUNrKNM2XODVVOGnX+AEluVSF78QOu//PV7Z4IgSWnZK22pkUFRWBryXlMS0FSr16FNsE4SHp9SsSWg3BqkWfb/Buj4/bNYHvNLD2zPl+LRSRhIk0hSaxJTZVbOAm6Y2EUshn1+XrG09LaIc+goB8O0gGuYmvXZ29IosnP6I0Vsm9rBIoMm0INzns4XKNJzRrZqfvrQZaU9URl9t/LdbjGJAWvUu9tUdDXdvnMa0Jo63hunpVB1XMwAVo3KAswZDUmcvxgt6kFDqHELmlEZhGFuzkSQTm7LmFw3a5JeaisENhaAA0qm9DnGMxc4x3FVLYArrGYCrhaHgrjx/BqNaczWdWaxfyH0Gl5gNz3usZMS2+5Ou2NanAjwXhjXl4EKQ/tXgiBWjO1QwxNi8zTw5PMsbNPndxiiNPanoZNXkhhVdeD5wmu72xsHcxiG3356MFB2D6B11J46nMhO64GeitBkYg6uoKgCz9ACJaBw1Eh29W19eIazZJHFGG06/y8oZOSciqQ2sE/YbVegDpmkLFxFoO/ZS29E3SBihsehaDhvmfD0NQnFFa5q/KsIC0/XakpKwhnF2n4gIWunLUHp4p8M+mqkZtNXlWpNmQ+uCbEzFoPI7SZY5Jj1DN2YpB6bb98yJVmUIlLoiY2SbY84Q2Zs7IoZ0LutOBJIFVN+MTy+ljYDLneptZaEacSNgk5UoFgHg8JniEfi9bkPou3ilmoq+SAvDAygze5QQ31e53LC3MaFV4d1JqJUIizeJAVu/Hwdm+8Aw4sqdcgXdCDtxJU/Dm/p+9gQWA7tH5fMdnRGsDfsFUVnAt+HT5KepEW4ddKEQsEay4jB0n1DM6h6DT1IXW9wA7O5Qq1tdbnE2T5qc2gR9/V2qs+u54NmlCd9mHLpLBc1LjOTM7MdSAHVqZrLM21IK3rGKom9A7yo004WgXhgihN0FkzRotMfAVTWJ7L7M/tJUD7++HLF0pyhihGdUYBN1bUEnIagyGSbkwOCTXS5qmK0+nfW7xonUAqO9V5HdMauP8TkWpBoDkzgiRVafnYzYhKE46u39yfnUaY/ayVeh5hj1e5qKcSh+z5GiK0G1ikcjce0e4FEgC4npWSumshboI7JslGBju7/labrUI+TL6I5c+dKYb9lOVarAgtmjEaYzD3sxZ0eB/VZopo5nuHMjWSJtVmQ91QjLnrwIuSB857QcbNCCNskpQT3KW193JHF2suIXcaV0UuzuQ66LDaeS6351lCZZ73tr2t7cuDUEdXxRiMvfgAM+J5BmJwHXV/C4Wr2DqSrxbgbGSTIw+HhGRn1Fko1mm/FoAbL0iIlOzdGzXkoFK8IwthZSUmDeS9yRWCvUs87BN2x3ikpnn6+7QV0cQuFk6aC3MGEYzNpebhvEu4SJtu2yBjK796KADIAkloqza0szLXutn2UiRzK4i+zQBxQbWeS8BU9evB+dxpTLVmsal3kHLogJYoxDPSptHUXpkyITb2WejjRG32FyTah1KYZyiOhZZ6X9UiHyMhbqtFxP8A/g+Pq0j89Ntr7vj+96EPd33/Sx+e7v3/bGa+9sVv3gshABARn8jMtz2r978Pfbjr+1/6cDf3vzecwKVd2qXdTbsIgUu7tGe83Sch8PPP+P3h7vtw1/eHSx9u/f73hhO4tEu7tLtp9wkJXNqlXdodtDsXAhHxwxHxuYh4LiLef4v3/b2I+M2I+GREfELvfUtE/GpE/I5+fvMTvucvRsQXI+Izh/cees+o9o80Lp+OiLc+pfv/ZEQ8r3H4ZES88/C3v6P7fy4i/tITuP8bI+KjEfFbEfHZiPibev82x+BRfbjNcXhlRHw8Ij6lPvyU3n9TRHxM9/pQRDzQ+6/Q78/p79/+9fbhRnOk3l38oxIcPw+8GXgAfAr47lu69+8Br3nRe38feL9evx/46Sd8zx8A3gp85qXuCbwT+LdUaMrbgY89pfv/JPC3H/LZ79Z8vAJ4k+Zp+zrv/zrgrXr9jcB/1X1ucwwe1YfbHIcAXqXXV8DH9Hy/DLxH7/8c8ON6/deAn9Pr9wAfepLr8q6RwPcCz2Xm72bmC8AHgXfdYX/eBXxArz8A/OiTvHhm/kfqaLaXc893Af80q/0n6gDY1z2F+z+qvQv4YGb+aWb+N+A5HnLi1Nd4/y9k5n/R6/9NnWn5em53DB7Vh0e1pzEOmZlf0a9X+pfAO4AP6/0Xj4PH58PAD8bjMoK+xnbXQuD1wO8ffv8DHj8hT7Il8O8i4tcj4sf03rdl5hf0+r8D33YL/XjUPW9zbP6G4PYvHkygp3p/QdrvobTgnYzBi/oAtzgOEbFFxCeBLwK/SiGML2Wmc36P9+k+6O9fBr716+2D210Lgbts35+ZbwV+BPjrEfEDxz9mYa9bdZ3cxT2Bfwz8OeAtwBeAf/C0bxgRrwL+JfC3MvN/Hf92W2PwkD7c6jhk5p6ZbwHeQCGL73qa93tcu2sh8DzwxsPvb9B7T71l5vP6+UXgX1MT8YeGm/r5xVvoyqPueStjk5l/qAU5gX/CgrpP5f4RcUVtvn+Wmf9Kb9/qGDysD7c9Dm6Z+SXgo8D3UeaOk/qO9+k+6O/fBPzxk+rDXQuB/wx8p1jRBxTp8ZGnfdOI+DMR8Y1+DfxF4DO693v1sfcCv/K0+/KYe34E+KtiyN8OfPkAmZ9Ye5GN/ZepcfD93yNm+k3AdwIf/zrvFdTR9b+dmf/w8KdbG4NH9eGWx+G1EfFqvf4G4IcobuKjwLv1sRePg8fn3cCvCTE9mfYkWcb/T6b0nRRD+3ngJ27pnm+mGN9PAZ/1fSk76z8AvwP8e+BbnvB9/zkFNa8pm+99j7onxSD/rMblN4G3PaX7/5Ku/2ktttcdPv8Tuv/ngB95Avf/fgrqfxr4pP6985bH4FF9uM1x+PPAb+henwH+7mFdfpwiH/8F8Aq9/0r9/pz+/uYnuS4vEYOXdmnPeLtrc+DSLu3S7rhdhMClXdoz3i5C4NIu7RlvFyFwaZf2jLeLELi0S3vG20UIXNqlPePtIgQu7dKe8XYRApd2ac94+38LI+c1opghagAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 建立讀取影像的函數\n",
    "def read_img(file):\n",
    "    image_string = tf.io.read_file(file)  # 讀取檔案\n",
    "    # 將讀入檔案以影像格式來解碼\n",
    "    image_decode = tf.image.decode_image(image_string)\n",
    "    # 將影像增加一個維度(number,height,width,channel)方便之後存入紀錄檔中\n",
    "    # image_decode = tf.expand_dims(image_decode, axis=0)\n",
    "    return image_decode\n",
    "\n",
    "img = read_img('image/airplane.png')  # 讀入影像資訊\n",
    "plt.imshow(img)  # 顯示讀入的影像資訊"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "image_string = tf.io.read_file('image/airplane.png')  # 讀取檔案\n",
    "image_decode = tf.image.decode_image(image_string)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "with summary_writer.as_default():  # summary_writer作為預設寫入的紀錄檔\n",
    "    tf.summary.image(\"Airplane\", [image_decode], step=0)  # 存入影像資訊"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "2. 一次儲存五張影像到紀錄檔中並顯示"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "airplane.png\t    airplane_flip.png  airplane_zoom.png\r\n",
      "airplane_color.png  airplane_rot.png   elephant.jpg\r\n"
     ]
    }
   ],
   "source": [
    "!ls image/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "img_files = ['airplane_zoom.png', 'airplane_flip.png', 'airplane_color.png', 'airplane_rot.png',\n",
    "             'airplane.png']  # 創建一個陣列用來儲存讀入的影像 \n",
    "\n",
    "imgs = [] \n",
    "\n",
    "for file in img_files: \n",
    "    imgs.append(read_img('image/'+file))  # 讀取影像並存入陣列中 \n",
    "\n",
    "with summary_writer.as_default():  # summary_writer作為預設寫入的紀錄檔 \n",
    "    # 一次存入五張影像(注意:如果max_outputs沒設定為5，就只會儲存3張影像) \n",
    "    tf.summary.image(\"Airplane Augmentation\", imgs, max_outputs=5, step=0) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "3. 將五張影像以不同Step(時間)儲存"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "with summary_writer.as_default():  # summary_writer作為預設寫入的紀錄檔\n",
    "    # 每次儲存一張影像，並儲存在不同Step中\n",
    "    for i, img in enumerate(imgs):\n",
    "        tf.summary.image(\"Save image each step\", [img], step=i)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Text"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 建立一個陣列，裡面包含了對話記錄\n",
    "texts = [\"小明：Cubee小助理最近好想學深度學習的技術哦!\", \n",
    "         \"Cubee：這是當然的阿，這可現今最火的技術呢!\", \n",
    "         \"小明：那我該如何入門呢?\", \n",
    "         \"Cubee：推薦你一本書「輕鬆學會Google TensorFlow2.0深度學習」。\", \n",
    "         \"小明：這本書沒有深度學習經驗的人也能學會嗎?\", \n",
    "         \"Cubee：這是當然的，你只需要基礎Python能力就可以學會了!\", \n",
    "         \"小明：太好了那我要趕快去買了!\"]\n",
    "\n",
    "with summary_writer.as_default():  # summary_writer作為預設寫入的紀錄檔\n",
    "    # 將每一段字串資訊以不同Step存入到記錄檔中\n",
    "    for i, text in enumerate(texts):\n",
    "        tf.summary.text(\"Chat record\", text, step=i)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Audio"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 建立讀取音訊的函數\n",
    "def read_audio(file):\n",
    "    audio_string = tf.io.read_file(file)  # 讀取檔案\n",
    "    # 將讀入檔案以音訊格式來解碼\n",
    "    audio, fs = tf.audio.decode_wav(audio_string)\n",
    "    # 因為tf.summary.audio要求輸入格式為[k(clips), t(frames), c(channels)]\n",
    "    # 而解碼後的音訊只有[t(frames), c(channels)]，所以需要增加一個維度給音訊\n",
    "    audio = tf.expand_dims(audio, axis=0)\n",
    "    return audio, fs\n",
    "\n",
    "audio, fs = read_audio('./audio/cat.wav')  # 讀取音訊檔\n",
    "\n",
    "with summary_writer.as_default():  # summary_writer作為預設寫入的紀錄檔\n",
    "    tf.summary.audio('cat', audio, fs, step=0)  # 存入音訊資訊"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Histogram\n",
    "\n",
    "目前 TensorFlow-gpu 2.0-alpha有問題，等待下一版修復。\n",
    "\n",
    "https://github.com/tensorflow/tensorboard/issues/1993"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "建立一個常態分佈"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = tf.random.normal([64, 100], dtype=tf.float64)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "儲存常態分佈分佈"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "with summary_writer.as_default():\n",
    "    tf.summary.histogram('Normal distribution', data, step=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "儲存多個常態分佈，並且各個之間平均值都相差0.01"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "with summary_writer.as_default():\n",
    "    for i, offset in enumerate(tf.range(0, 10, delta=0.1, dtype=tf.float64)):\n",
    "        tf.summary.histogram('Normal distribution 2', data+offset, step=i)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 實驗一：使用tf.summary.image紀錄訓練結果"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Import 必要套件"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import io\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "import tensorflow_datasets as tfds\n",
    "import matplotlib.pyplot as plt\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras import layers\n",
    "from preprocessing import parse_aug_fn, parse_fn"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Confusion matrix"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Confusion matrix函數：透過tf.math.confusion_matrix來計算Confusion matrix。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[2 0 0]\n",
      " [0 2 1]\n",
      " [1 0 2]]\n"
     ]
    }
   ],
   "source": [
    "y_true = [2, 1, 0, 2, 2, 0, 1, 1]\n",
    "y_pred = [0, 1, 0, 2, 2, 0, 2, 1]\n",
    "cm = tf.math.confusion_matrix(y_true, y_pred, num_classes=3).numpy()\n",
    "print(cm)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "建立plot_confusion_matrix函式：將剛剛上方計算的Confusion matrix陣列以Matplotlib圖片來表示，而Confusion matrix中的數字改成百分比型式。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def plot_confusion_matrix(cm, class_names):\n",
    "    \"\"\"\n",
    "    產生一張圖表示的Confusion matrix\n",
    "    \n",
    "    Args:\n",
    "    cm (shape = [n, n]): 傳入Confusion matrix\n",
    "    class_names (shape = [n]): 傳入類別\n",
    "    \"\"\"\n",
    "    # 標準化confusion matrix.\n",
    "    cm = np.around(cm.astype('float') / cm.sum(axis=1)[:, np.newaxis], decimals=2)\n",
    "    \n",
    "    figure = plt.figure(figsize=(8, 8))\n",
    "    plt.imshow(cm, interpolation='nearest', cmap=plt.cm.Blues)\n",
    "    plt.title(\"Confusion matrix\")\n",
    "    tick_index = np.arange(len(class_names))\n",
    "    # matplotlib 3.1.1 bug，如果不設定ylim在[-0.5~2.5]，圖片y軸範圍會被縮小成[0~2]\n",
    "    plt.ylim([-0.5, 2.5])\n",
    "    # Y軸顯示類別名稱\n",
    "    plt.yticks(tick_index, class_names)\n",
    "    # X軸顯示類別名稱，並將類別名稱旋轉45度(避免文字重疊)\n",
    "    plt.xticks(tick_index, class_names, rotation=45)\n",
    "    # 再圖片右邊產生一條顏色刻度條\n",
    "    plt.colorbar()\n",
    "\n",
    "    # 在每一格Confusion matrix輸入預測百分比\n",
    "    threshold = cm.max() / 2.\n",
    "    for i in range(cm.shape[0]):\n",
    "        for j in range(cm.shape[1]):\n",
    "            # 如果格內背景顏色太深使用白色文字顯示，反之使用黑色文字\n",
    "            color = \"white\" if cm[i, j] > threshold else \"black\"\n",
    "            plt.text(j, i, cm[i, j], horizontalalignment=\"center\", color=color)\n",
    "            \n",
    "    plt.ylabel('True label')\n",
    "    plt.xlabel('Predicted label')\n",
    "    # 將圖片的位置進行調整，避免x或y軸的文字被遮擋\n",
    "    plt.tight_layout()\n",
    "    return figure\n",
    "\n",
    "# Example\n",
    "img = plot_confusion_matrix(cm, [0, 1, 2])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "建立plot_to_image函數：將Matplotlib圖片轉TensorFlow型式的圖片，這樣才能夠透過tf.summary.image紀錄影像到TensorBoard。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_to_image(figure):\n",
    "    \"\"\"將Matplotlib plot的圖片轉TensorFlow的張量格式\"\"\"\n",
    "    # 將Matplotlib plot的圖片以PNG的格式儲存到記憶體中\n",
    "    buf = io.BytesIO()\n",
    "    plt.savefig(buf, format='png')\n",
    "    # 關閉plt圖片，防止圖片直接顯示在Jupyter notebook介面中\n",
    "    plt.close(figure)\n",
    "    buf.seek(0)\n",
    "    # 將記憶體中的資料轉成TensorFlow格式\n",
    "    image = tf.image.decode_png(buf.getvalue(), channels=4)\n",
    "    image = tf.expand_dims(image, 0)\n",
    "    return image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Example\n",
    "y_true = [2, 0, 2, 2, 0, 1]\n",
    "y_pred = [0, 0, 2, 2, 0, 2]\n",
    "cm = tf.math.confusion_matrix(y_true, y_pred, num_classes=3).numpy()\n",
    "img = plot_confusion_matrix(cm, [0, 1, 2])\n",
    "img_show = plot_to_image(img)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 創建Callback函數\n",
    "創建回調函數：訓練過程中每個epoch結束，會產生一張Confusion matrix的圖片，並將圖片紀錄在Tensorboard上。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "class ConfusionMatrix(tf.keras.callbacks.Callback):\n",
    "    def __init__(self, log_dir, test_data, class_name):\n",
    "        super(ConfusionMatrix, self).__init__()\n",
    "        self.log_dir = log_dir\n",
    "        self.test_data = test_data\n",
    "        self.class_names = class_name\n",
    "        self.num_classes = len(class_name)\n",
    "\n",
    "    def on_train_begin(self, logs=None):\n",
    "        path = os.path.join(self.log_dir, 'confusion_matrix')\n",
    "        # 創建TensorBoard紀錄檔\n",
    "        self.writer = tf.summary.create_file_writer(path)\n",
    "\n",
    "    def on_epoch_end(self, epoch, logs=None):\n",
    "        # 計算Confusion matrix\n",
    "        total_cm = np.zeros([10, 10])\n",
    "        for x, y_true in self.test_data:\n",
    "            y_pred = self.model.predict(x)\n",
    "            y_pred = np.argmax(y_pred, axis=1)\n",
    "            y_true = np.argmax(y_true, axis=1)\n",
    "            cm = tf.math.confusion_matrix(y_true, y_pred, num_classes=self.num_classes).numpy()\n",
    "            total_cm += cm\n",
    "        \n",
    "        # 將Confusion matrix轉成Matplotlib圖片\n",
    "        figure = plot_confusion_matrix(total_cm, class_names=self.class_names)\n",
    "        # 將Matplotlib圖片轉成TensorFlow型式的圖片\n",
    "        cm_image = plot_to_image(figure)\n",
    "\n",
    "        # 將圖片紀錄在TensorBoard log中\n",
    "        with self.writer.as_default():\n",
    "            tf.summary.image(\"Confusion Matrix\", cm_image, step=epoch)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 訓練網路模型"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "載入CIFAR-10數據集："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 將train Data重新分成1:9等分，分別分給valid data, train data\n",
    "valid_split, train_split = tfds.Split.TRAIN.subsplit([10, 90])\n",
    "# 取得訓練數據，並順便讀取data的資訊\n",
    "train_data, info = tfds.load(\"cifar10\", split=train_split, with_info=True)\n",
    "# 取得驗證數據\n",
    "valid_data = tfds.load(\"cifar10\", split=valid_split)\n",
    "# 取得測試數據\n",
    "test_data = tfds.load(\"cifar10\", split=tfds.Split.TEST)\n",
    "# 取得CIFAR-10數據集的類別\n",
    "class_name = info.features['label'].names"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Dataset 設定；"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "AUTOTUNE = tf.data.experimental.AUTOTUNE  # 自動調整模式\n",
    "batch_size = 64  # 批次大小\n",
    "train_num = int(info.splits['train'].num_examples / 10) * 9  # 訓練資料數量\n",
    "\n",
    "train_data = train_data.shuffle(train_num)  # 打散資料集\n",
    "# 載入預處理「 parse_aug_fn」function，cpu數量為自動調整模式\n",
    "train_data = train_data.map(map_func=parse_aug_fn, num_parallel_calls=AUTOTUNE)\n",
    "# 設定批次大小並將prefetch模式開啟(暫存空間為自動調整模式)\n",
    "train_data = train_data.batch(batch_size).prefetch(buffer_size=AUTOTUNE)\n",
    "\n",
    "# 載入預處理「 parse_fn」function，cpu數量為自動調整模式\n",
    "valid_data = valid_data.map(map_func=parse_fn, num_parallel_calls=AUTOTUNE)\n",
    "# 設定批次大小並將prefetch模式開啟(暫存空間為自動調整模式)\n",
    "valid_data = valid_data.batch(batch_size).prefetch(buffer_size=AUTOTUNE)\n",
    "\n",
    "# 載入預處理「 parse_fn」function，cpu數量為自動調整模式\n",
    "test_data = test_data.map(map_func=parse_fn, num_parallel_calls=AUTOTUNE)\n",
    "# 設定批次大小並將prefetch模式開啟(暫存空間為自動調整模式)\n",
    "test_data = test_data.batch(batch_size).prefetch(buffer_size=AUTOTUNE)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "建立網路模型："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model 1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         [(None, 32, 32, 3)]       0         \n",
      "_________________________________________________________________\n",
      "conv2d (Conv2D)              (None, 30, 30, 64)        1792      \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 15, 15, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 13, 13, 128)       73856     \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 11, 11, 256)       295168    \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 9, 9, 128)         295040    \n",
      "_________________________________________________________________\n",
      "conv2d_4 (Conv2D)            (None, 7, 7, 64)          73792     \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 3136)              0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 64)                200768    \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 10)                650       \n",
      "=================================================================\n",
      "Total params: 941,066\n",
      "Trainable params: 941,066\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "inputs = keras.Input(shape=(32, 32, 3))\n",
    "x = layers.Conv2D(64, 3, activation='relu', kernel_initializer='glorot_uniform')(inputs)\n",
    "x = layers.MaxPool2D()(x)\n",
    "x = layers.Conv2D(128, 3, activation='relu', kernel_initializer='glorot_uniform')(x)\n",
    "x = layers.Conv2D(256, 3, activation='relu', kernel_initializer='glorot_uniform')(x)\n",
    "x = layers.Conv2D(128, 3, activation='relu', kernel_initializer='glorot_uniform')(x)\n",
    "x = layers.Conv2D(64, 3, activation='relu', kernel_initializer='glorot_uniform')(x)\n",
    "x = layers.Flatten()(x)\n",
    "x = layers.Dense(64, activation='relu')(x)\n",
    "x = layers.Dropout(0.5)(x)\n",
    "outputs = layers.Dense(10)(x)\n",
    "# 建立網路模型(將輸入到輸出所有經過的網路層連接起來)\n",
    "model_1 = keras.Model(inputs, outputs, name='model-1')\n",
    "model_1.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "建立Callback function："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 儲存訓練記錄檔\n",
    "logs_dirs = 'lab7-logs-images'\n",
    "model_cbk = keras.callbacks.TensorBoard(logs_dirs)\n",
    "# 儲存Confusion matrix圖片\n",
    "save_cm = ConfusionMatrix(logs_dirs, test_data, class_name)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "設定訓練使用的優化器、客自化損失函數和客自化指標函數："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_1.compile(keras.optimizers.Adam(), \n",
    "                loss=keras.losses.CategoricalCrossentropy(from_logits=True), \n",
    "                metrics=[keras.metrics.CategoricalAccuracy()])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "訓練網路模型："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "      1/Unknown - 4s 4s/step - loss: 2.3080 - categorical_accuracy: 0.1094"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING: Logging before flag parsing goes to stderr.\n",
      "W0629 20:55:55.693831 139815137187648 callbacks.py:241] Method (on_train_batch_end) is slow compared to the batch update (0.128922). Check your callbacks.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 16s 22ms/step - loss: 2.0773 - categorical_accuracy: 0.2196 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.8722 - categorical_accuracy: 0.3069 - val_loss: 1.5641 - val_categorical_accuracy: 0.4276\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.7459 - categorical_accuracy: 0.3602 - val_loss: 1.4246 - val_categorical_accuracy: 0.4756\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.6568 - categorical_accuracy: 0.3955 - val_loss: 1.3370 - val_categorical_accuracy: 0.5268\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.5911 - categorical_accuracy: 0.4272 - val_loss: 1.2556 - val_categorical_accuracy: 0.5528\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.5312 - categorical_accuracy: 0.4474 - val_loss: 1.1968 - val_categorical_accuracy: 0.5718\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.4757 - categorical_accuracy: 0.4735 - val_loss: 1.1474 - val_categorical_accuracy: 0.5916\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.4323 - categorical_accuracy: 0.4931 - val_loss: 1.0755 - val_categorical_accuracy: 0.6198\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.4027 - categorical_accuracy: 0.5083 - val_loss: 1.0345 - val_categorical_accuracy: 0.6330\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.3665 - categorical_accuracy: 0.5201 - val_loss: 1.0419 - val_categorical_accuracy: 0.6278\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.3257 - categorical_accuracy: 0.5379 - val_loss: 1.0120 - val_categorical_accuracy: 0.6378\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.3041 - categorical_accuracy: 0.5440 - val_loss: 0.9637 - val_categorical_accuracy: 0.6552\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.2738 - categorical_accuracy: 0.5560 - val_loss: 0.9440 - val_categorical_accuracy: 0.6630\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.2508 - categorical_accuracy: 0.5664 - val_loss: 0.9294 - val_categorical_accuracy: 0.6730\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.2361 - categorical_accuracy: 0.5700 - val_loss: 0.9401 - val_categorical_accuracy: 0.6660\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.2085 - categorical_accuracy: 0.5823 - val_loss: 0.9062 - val_categorical_accuracy: 0.6832\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.1907 - categorical_accuracy: 0.5896 - val_loss: 0.8691 - val_categorical_accuracy: 0.6872\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.1738 - categorical_accuracy: 0.5967 - val_loss: 0.8436 - val_categorical_accuracy: 0.7036\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.1539 - categorical_accuracy: 0.6036 - val_loss: 0.8786 - val_categorical_accuracy: 0.6874\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.1531 - categorical_accuracy: 0.6014 - val_loss: 0.8849 - val_categorical_accuracy: 0.6942\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.1375 - categorical_accuracy: 0.6084 - val_loss: 0.8444 - val_categorical_accuracy: 0.6994\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.1252 - categorical_accuracy: 0.6145 - val_loss: 0.8000 - val_categorical_accuracy: 0.7138\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.1134 - categorical_accuracy: 0.6179 - val_loss: 0.8485 - val_categorical_accuracy: 0.7026\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.1002 - categorical_accuracy: 0.6219 - val_loss: 0.8660 - val_categorical_accuracy: 0.6976\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.0940 - categorical_accuracy: 0.6268 - val_loss: 0.8149 - val_categorical_accuracy: 0.7134\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.0799 - categorical_accuracy: 0.6298 - val_loss: 0.8194 - val_categorical_accuracy: 0.7154\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.0667 - categorical_accuracy: 0.6379 - val_loss: 0.8084 - val_categorical_accuracy: 0.7146\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.0582 - categorical_accuracy: 0.6410 - val_loss: 0.8281 - val_categorical_accuracy: 0.7092\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.0377 - categorical_accuracy: 0.6450 - val_loss: 0.7642 - val_categorical_accuracy: 0.7272\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.0338 - categorical_accuracy: 0.6482 - val_loss: 0.7615 - val_categorical_accuracy: 0.7342\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.0368 - categorical_accuracy: 0.6473 - val_loss: 0.7374 - val_categorical_accuracy: 0.7374\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.0186 - categorical_accuracy: 0.6523 - val_loss: 0.7699 - val_categorical_accuracy: 0.7312\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.0145 - categorical_accuracy: 0.6547 - val_loss: 0.7697 - val_categorical_accuracy: 0.7382\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.0063 - categorical_accuracy: 0.6584 - val_loss: 0.7422 - val_categorical_accuracy: 0.7440\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.0084 - categorical_accuracy: 0.6580 - val_loss: 0.7333 - val_categorical_accuracy: 0.7506\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.9929 - categorical_accuracy: 0.6629 - val_loss: 0.7339 - val_categorical_accuracy: 0.7520\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.9795 - categorical_accuracy: 0.6680 - val_loss: 0.7138 - val_categorical_accuracy: 0.7570\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.9781 - categorical_accuracy: 0.6678 - val_loss: 0.7284 - val_categorical_accuracy: 0.7454\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.9742 - categorical_accuracy: 0.6712 - val_loss: 0.7655 - val_categorical_accuracy: 0.7434\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.9719 - categorical_accuracy: 0.6720 - val_loss: 0.7323 - val_categorical_accuracy: 0.7526\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.9592 - categorical_accuracy: 0.6760 - val_loss: 0.7198 - val_categorical_accuracy: 0.7600\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.9558 - categorical_accuracy: 0.6794 - val_loss: 0.7087 - val_categorical_accuracy: 0.7572\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.9525 - categorical_accuracy: 0.6800 - val_loss: 0.7323 - val_categorical_accuracy: 0.7506\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.9456 - categorical_accuracy: 0.6818 - val_loss: 0.6928 - val_categorical_accuracy: 0.7608\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.9408 - categorical_accuracy: 0.6815 - val_loss: 0.7070 - val_categorical_accuracy: 0.7596\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.9282 - categorical_accuracy: 0.6863 - val_loss: 0.6910 - val_categorical_accuracy: 0.7612\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.9267 - categorical_accuracy: 0.6887 - val_loss: 0.6681 - val_categorical_accuracy: 0.7806\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.9267 - categorical_accuracy: 0.6887 - val_loss: 0.6976 - val_categorical_accuracy: 0.7642\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.9226 - categorical_accuracy: 0.6904 - val_loss: 0.6883 - val_categorical_accuracy: 0.7704\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.9202 - categorical_accuracy: 0.6938 - val_loss: 0.7116 - val_categorical_accuracy: 0.7630\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.9067 - categorical_accuracy: 0.6952 - val_loss: 0.6943 - val_categorical_accuracy: 0.7680\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.9027 - categorical_accuracy: 0.6996 - val_loss: 0.6813 - val_categorical_accuracy: 0.7680\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8965 - categorical_accuracy: 0.6980 - val_loss: 0.6967 - val_categorical_accuracy: 0.7672\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8970 - categorical_accuracy: 0.7010 - val_loss: 0.6991 - val_categorical_accuracy: 0.7686\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8867 - categorical_accuracy: 0.7015 - val_loss: 0.6921 - val_categorical_accuracy: 0.7716\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8958 - categorical_accuracy: 0.6963 - val_loss: 0.6816 - val_categorical_accuracy: 0.7692\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8858 - categorical_accuracy: 0.7011 - val_loss: 0.7108 - val_categorical_accuracy: 0.7602\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8806 - categorical_accuracy: 0.7067 - val_loss: 0.7294 - val_categorical_accuracy: 0.7580\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8752 - categorical_accuracy: 0.7059 - val_loss: 0.6783 - val_categorical_accuracy: 0.7772\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8759 - categorical_accuracy: 0.7076 - val_loss: 0.7205 - val_categorical_accuracy: 0.7628\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8617 - categorical_accuracy: 0.7107 - val_loss: 0.6663 - val_categorical_accuracy: 0.7784\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8668 - categorical_accuracy: 0.7078 - val_loss: 0.6598 - val_categorical_accuracy: 0.7774\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8628 - categorical_accuracy: 0.7105 - val_loss: 0.6858 - val_categorical_accuracy: 0.7744\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8647 - categorical_accuracy: 0.7096 - val_loss: 0.6970 - val_categorical_accuracy: 0.7680\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8684 - categorical_accuracy: 0.7080 - val_loss: 0.6555 - val_categorical_accuracy: 0.7784\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8536 - categorical_accuracy: 0.7159 - val_loss: 0.6238 - val_categorical_accuracy: 0.7956\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8533 - categorical_accuracy: 0.7147 - val_loss: 0.6414 - val_categorical_accuracy: 0.7828\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8486 - categorical_accuracy: 0.7167 - val_loss: 0.6618 - val_categorical_accuracy: 0.7820\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8383 - categorical_accuracy: 0.7192 - val_loss: 0.6918 - val_categorical_accuracy: 0.7724\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8446 - categorical_accuracy: 0.7164 - val_loss: 0.6716 - val_categorical_accuracy: 0.7752\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8498 - categorical_accuracy: 0.7176 - val_loss: 0.6765 - val_categorical_accuracy: 0.7768\n",
      "Epoch 72/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8296 - categorical_accuracy: 0.7238 - val_loss: 0.6774 - val_categorical_accuracy: 0.7800\n",
      "Epoch 73/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8234 - categorical_accuracy: 0.7236 - val_loss: 0.6773 - val_categorical_accuracy: 0.7786\n",
      "Epoch 74/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8303 - categorical_accuracy: 0.7236 - val_loss: 0.6638 - val_categorical_accuracy: 0.7846\n",
      "Epoch 75/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8321 - categorical_accuracy: 0.7209 - val_loss: 0.6638 - val_categorical_accuracy: 0.7836\n",
      "Epoch 76/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8347 - categorical_accuracy: 0.7223 - val_loss: 0.6816 - val_categorical_accuracy: 0.7780\n",
      "Epoch 77/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8256 - categorical_accuracy: 0.7217 - val_loss: 0.6773 - val_categorical_accuracy: 0.7794\n",
      "Epoch 78/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8192 - categorical_accuracy: 0.7267 - val_loss: 0.6864 - val_categorical_accuracy: 0.7796\n",
      "Epoch 79/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8209 - categorical_accuracy: 0.7256 - val_loss: 0.6556 - val_categorical_accuracy: 0.7866\n",
      "Epoch 80/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8162 - categorical_accuracy: 0.7273 - val_loss: 0.6653 - val_categorical_accuracy: 0.7852\n",
      "Epoch 81/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8160 - categorical_accuracy: 0.7270 - val_loss: 0.6853 - val_categorical_accuracy: 0.7782\n",
      "Epoch 82/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8149 - categorical_accuracy: 0.7290 - val_loss: 0.7170 - val_categorical_accuracy: 0.7656\n",
      "Epoch 83/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8137 - categorical_accuracy: 0.7282 - val_loss: 0.6941 - val_categorical_accuracy: 0.7746\n",
      "Epoch 84/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8166 - categorical_accuracy: 0.7279 - val_loss: 0.6457 - val_categorical_accuracy: 0.7954\n",
      "Epoch 85/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8115 - categorical_accuracy: 0.7284 - val_loss: 0.6801 - val_categorical_accuracy: 0.7830\n",
      "Epoch 86/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8067 - categorical_accuracy: 0.7295 - val_loss: 0.6796 - val_categorical_accuracy: 0.7898\n",
      "Epoch 87/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8046 - categorical_accuracy: 0.7312 - val_loss: 0.6675 - val_categorical_accuracy: 0.7834\n",
      "Epoch 88/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8048 - categorical_accuracy: 0.7311 - val_loss: 0.6443 - val_categorical_accuracy: 0.7944\n",
      "Epoch 89/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8092 - categorical_accuracy: 0.7295 - val_loss: 0.6601 - val_categorical_accuracy: 0.7930\n",
      "Epoch 90/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.7936 - categorical_accuracy: 0.7363 - val_loss: 0.6373 - val_categorical_accuracy: 0.7982\n",
      "Epoch 91/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8026 - categorical_accuracy: 0.7344 - val_loss: 0.6554 - val_categorical_accuracy: 0.7890\n",
      "Epoch 92/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.7920 - categorical_accuracy: 0.7382 - val_loss: 0.6913 - val_categorical_accuracy: 0.7792\n",
      "Epoch 93/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.7962 - categorical_accuracy: 0.7347 - val_loss: 0.6235 - val_categorical_accuracy: 0.7936\n",
      "Epoch 94/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7870 - categorical_accuracy: 0.7377 - val_loss: 0.6503 - val_categorical_accuracy: 0.7860\n",
      "Epoch 95/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7942 - categorical_accuracy: 0.7359 - val_loss: 0.6484 - val_categorical_accuracy: 0.7928\n",
      "Epoch 96/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.7863 - categorical_accuracy: 0.7395 - val_loss: 0.6702 - val_categorical_accuracy: 0.7900\n",
      "Epoch 97/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 13s 18ms/step - loss: 0.7890 - categorical_accuracy: 0.7370 - val_loss: 0.6618 - val_categorical_accuracy: 0.7960\n",
      "Epoch 98/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7761 - categorical_accuracy: 0.7423 - val_loss: 0.6465 - val_categorical_accuracy: 0.7918\n",
      "Epoch 99/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.7992 - categorical_accuracy: 0.7354 - val_loss: 0.6544 - val_categorical_accuracy: 0.7866\n",
      "Epoch 100/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.7820 - categorical_accuracy: 0.7400 - val_loss: 0.6489 - val_categorical_accuracy: 0.7966\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x7f28b0183da0>"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_1.fit(train_data,\n",
    "            epochs=100, \n",
    "            validation_data=valid_data,\n",
    "            callbacks=[model_cbk, save_cm])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 實驗二：使用TensorBoard超參數調校工具來訓練多個網路模型"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Import必要套件："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import tensorflow as tf\n",
    "import tensorflow_datasets as tfds\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras import layers\n",
    "# 從資料夾中的preprocessing.py檔案中Import parse_aug_fn和parse_fn函數\n",
    "from preprocessing import parse_aug_fn, parse_fn"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Import TensorBoard超參數工具所需要的套件"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorboard.plugins.hparams import api as hp"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 設定TensorBoard超參數調校"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "hparam_ia = hp.HParam('Imgae_Augmentation', hp.Discrete([False, True]))\n",
    "hparam_bn = hp.HParam('Batch_Normalization', hp.Discrete([False, True]))\n",
    "hparam_init = hp.HParam('Weight_Initialization', hp.Discrete(['RandomNormal_0.01std', 'glorot_normal', 'he_normal']))\n",
    "hparam_lr = hp.HParam('Learning_Rate', hp.Discrete([0.001, 0.01, 0.03]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 將實驗摘要寫入紀錄檔"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "metric = 'Accuracy'\n",
    "log_dirs = \"lab7-logs-hparams/hparam_tuning\"\n",
    "with tf.summary.create_file_writer(log_dirs).as_default():\n",
    "    hp.hparams_config(\n",
    "        hparams=[hparam_ia, hparam_bn, hparam_init, hparam_lr],\n",
    "        metrics=[hp.Metric(metric, display_name='Accuracy')],\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 訓練網路模型"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "準備訓練資料，分別為「沒有影像增強」和「有影像增強」的訓練資料"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "AUTOTUNE = tf.data.experimental.AUTOTUNE  # 自動調整模式\n",
    "batch_size = 64  # 批次大小\n",
    "# 將train Data重新分成1:9等分，分別分給valid data, train data\n",
    "valid_split, train_split = tfds.Split.TRAIN.subsplit([10, 90])\n",
    "\n",
    "# 取得訓練數據\n",
    "train_data_noaug, info = tfds.load(\"cifar10\", split=train_split, with_info=True)\n",
    "train_data_aug = tfds.load(\"cifar10\", split=train_split)\n",
    "# 取得驗證數據\n",
    "valid_data = tfds.load(\"cifar10\", split=valid_split)\n",
    "\n",
    "train_num = int(info.splits['train'].num_examples / 5) * 4  # 訓練資料數量\n",
    "\n",
    "train_data_noaug = train_data_noaug.shuffle(train_num)  # 打散資料集\n",
    "# 載入預處理「 parse_aug_fn」function，cpu數量為自動調整模式\n",
    "train_data_noaug = train_data_noaug.map(map_func=parse_fn, num_parallel_calls=AUTOTUNE)\n",
    "# 設定批次大小並將prefetch模式開啟(暫存空間為自動調整模式)\n",
    "train_data_noaug = train_data_noaug.batch(batch_size).prefetch(buffer_size=AUTOTUNE)\n",
    "\n",
    "train_data_aug = train_data_aug.shuffle(train_num)  # 打散資料集\n",
    "# 載入預處理「 parse_fn」function，cpu數量為自動調整模式\n",
    "train_data_aug = train_data_aug.map(map_func=parse_aug_fn, num_parallel_calls=AUTOTUNE)\n",
    "# 設定批次大小並將prefetch模式開啟(暫存空間為自動調整模式)\n",
    "train_data_aug = train_data_aug.batch(batch_size).prefetch(buffer_size=AUTOTUNE)\n",
    "\n",
    "# 載入預處理「 parse_fn」function，cpu數量為自動調整模式\n",
    "valid_data = valid_data.map(map_func=parse_fn, num_parallel_calls=AUTOTUNE)\n",
    "# 設定批次大小並將prefetch模式開啟(暫存空間為自動調整模式)\n",
    "valid_data = valid_data.batch(batch_size).prefetch(buffer_size=AUTOTUNE)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "客製化超參數回調函式：用於紀錄每一次訓練模型的超參數數值和最終訓練結果的指標值。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "class HyperparameterCallback(tf.keras.callbacks.Callback):\n",
    "    # 類別創建時調用\n",
    "    def __init__(self, log_dir, hparams):\n",
    "        super(HyperparameterCallback, self).__init__()\n",
    "        self.log_dir = log_dir\n",
    "        self.hparams = hparams\n",
    "        self.best_accuracy = 0\n",
    "        self.writer = None\n",
    "        \n",
    "    # 訓練開始前調用\n",
    "    def on_train_begin(self, logs=None):\n",
    "        self.writer = tf.summary.create_file_writer(self.log_dir)\n",
    "\n",
    "    # 每一個Epcoh結束後調用\n",
    "    def on_epoch_end(self, epoch, logs=None):\n",
    "        current_accuracy = logs.get('val_categorical_accuracy')\n",
    "        if current_accuracy > self.best_accuracy:\n",
    "            self.best_accuracy = current_accuracy\n",
    "            \n",
    "    # 訓練結束時調用\n",
    "    def on_train_end(self, logs=None):\n",
    "        with self.writer.as_default():\n",
    "            hp.hparams(self.hparams)  # record the values used in this trial\n",
    "            tf.summary.scalar(metric, self.best_accuracy, step=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "建立一個函式負責創建、編譯和訓練網路模型，網路層配置如下：\n",
    "- keras.Input：輸入層(輸入影像大小為32x32x3)。\n",
    "- layers.Conv2D：卷積層(使用ReLU激活函數，以及3x3大小的kernel)。\n",
    "- layers.MaxPool2D：池化層(對特徵圖下採樣)。\n",
    "- layers.Flatten：扁平層(特徵圖轉成一維Tensor)。\n",
    "- layers.Dropout：Dropout層(每次訓練隨機丟棄50%網路)。\n",
    "- layers.Dense：全連接層(隱藏層使用ReLU激活函數，輸出層使用Softmax激活函數)。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train_test_model(logs_dir, hparams):\n",
    "    \"\"\"\n",
    "    logs_dir:傳入目前執行的任務log檔的位置\n",
    "    hparams:傳入超參數\n",
    "    \"\"\"\n",
    "    # 指派網路模型初始化的方法\n",
    "    if hparams[hparam_init] == \"glorot_normal\":\n",
    "        init = keras.initializers.glorot_normal()\n",
    "    elif hparams[hparam_init] == \"he_normal\":\n",
    "        init = keras.initializers.he_normal()\n",
    "    else:\n",
    "        init = keras.initializers.RandomNormal(0, 0.01)\n",
    "\n",
    "    inputs = keras.Input(shape=(32, 32, 3))\n",
    "    x = layers.Conv2D(64, (3, 3), kernel_initializer=init)(inputs)\n",
    "    if hparams[hparam_bn]: x = layers.BatchNormalization()(x)\n",
    "    x = layers.ReLU()(x)\n",
    "    x = layers.MaxPool2D()(x)\n",
    "    x = layers.Conv2D(128, (3, 3), kernel_initializer=init)(x)\n",
    "    if hparams[hparam_bn]: x = layers.BatchNormalization()(x)\n",
    "    x = layers.ReLU()(x)\n",
    "    x = layers.Conv2D(256, (3, 3), kernel_initializer=init)(x)\n",
    "    if hparams[hparam_bn]: x = layers.BatchNormalization()(x)\n",
    "    x = layers.ReLU()(x)\n",
    "    x = layers.Conv2D(128, (3, 3), kernel_initializer=init)(x)\n",
    "    if hparams[hparam_bn]: x = layers.BatchNormalization()(x)\n",
    "    x = layers.ReLU()(x)\n",
    "    x = layers.Conv2D(64, (3, 3), kernel_initializer=init)(x)\n",
    "    if hparams[hparam_bn]: x = layers.BatchNormalization()(x)\n",
    "    x = layers.ReLU()(x)\n",
    "    x = layers.Flatten()(x)\n",
    "    x = layers.Dense(64, kernel_initializer=init)(x)\n",
    "    if hparams[hparam_bn]: x = layers.BatchNormalization()(x)\n",
    "    x = layers.ReLU()(x)\n",
    "    x = layers.Dropout(0.5)(x)\n",
    "    outputs = layers.Dense(10, activation='softmax')(x)\n",
    "    # 建立網路模型(將輸入到輸出所有經過的網路層連接起來)\n",
    "    model = keras.Model(inputs, outputs, name='model')\n",
    "\n",
    "    # 儲存訓練記錄檔\n",
    "    model_tb = keras.callbacks.TensorBoard(log_dir=logs_dir, write_graph=False)\n",
    "\n",
    "    # 儲存最好的網路模型權重\n",
    "    model_mckp = keras.callbacks.ModelCheckpoint(logs_dir +'/best-model.hdf5', \n",
    "                                                 monitor='val_categorical_accuracy', \n",
    "                                                 save_best_only=True, \n",
    "                                                 mode='max')\n",
    "    \n",
    "    # 設定停止訓練的條件(當Accuracy超過30迭代沒有上升的話訓練會終止)\n",
    "    model_els = keras.callbacks.EarlyStopping(monitor='val_categorical_accuracy' , \n",
    "                                              min_delta=0, \n",
    "                                              patience=30, \n",
    "                                              mode='max')\n",
    "    # 客自化超參數回調函式，紀錄訓練模型的超參數和指標(準確率)\n",
    "    model_hparam = HyperparameterCallback(logs_dir + 'hparam_tuning', hparams)\n",
    "\n",
    "\n",
    "    # 設定訓練使用的優化器、損失函數和指標函數\n",
    "    # 優化器學習率為超參數：0.001、0.01或0.03\n",
    "    model.compile(keras.optimizers.Adam(hparams[hparam_lr]), \n",
    "                  loss=keras.losses.CategoricalCrossentropy(), \n",
    "                  metrics=[keras.metrics.CategoricalAccuracy()])\n",
    "    \n",
    "    # 超參數：使用「經過影像增強的數據」或「不經過影像增強的數據」訓練網路\n",
    "    if hparams[hparam_ia]:\n",
    "        history = model.fit(train_data_aug,\n",
    "                            epochs=2, \n",
    "                            validation_data=valid_data,\n",
    "                            callbacks=[model_tb, model_mckp, model_els, model_hparam])\n",
    "    else:\n",
    "        history = model.fit(train_data_noaug,\n",
    "                            epochs=2, \n",
    "                            validation_data=valid_data,\n",
    "                            callbacks=[model_tb, model_mckp, model_els, model_hparam])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "訓練36種不同參數的網路模型："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--- Running training session 1\n",
      "Epoch 1/100\n",
      "      1/Unknown - 12s 12s/step - loss: 2.3026 - categorical_accuracy: 0.0625"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING: Logging before flag parsing goes to stderr.\n",
      "W0423 14:11:27.514188 140303049860864 callbacks.py:236] Method (on_train_batch_end) is slow compared to the batch update (0.192300). Check your callbacks.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 23s 33ms/step - loss: 2.1177 - categorical_accuracy: 0.1981 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 1.9106 - categorical_accuracy: 0.2809 - val_loss: 1.7897 - val_categorical_accuracy: 0.3300\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.8074 - categorical_accuracy: 0.3233 - val_loss: 1.6816 - val_categorical_accuracy: 0.3842\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 1.7261 - categorical_accuracy: 0.3523 - val_loss: 1.5660 - val_categorical_accuracy: 0.4282\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 1.6565 - categorical_accuracy: 0.3870 - val_loss: 1.4883 - val_categorical_accuracy: 0.4610\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 22s 31ms/step - loss: 1.6009 - categorical_accuracy: 0.4085 - val_loss: 1.4367 - val_categorical_accuracy: 0.4802\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 1.5570 - categorical_accuracy: 0.4233 - val_loss: 1.3943 - val_categorical_accuracy: 0.4888\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 1.5122 - categorical_accuracy: 0.4409 - val_loss: 1.3324 - val_categorical_accuracy: 0.5122\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 19s 28ms/step - loss: 1.4624 - categorical_accuracy: 0.4601 - val_loss: 1.3385 - val_categorical_accuracy: 0.5064\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.4340 - categorical_accuracy: 0.4744 - val_loss: 1.2818 - val_categorical_accuracy: 0.5290\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 1.4058 - categorical_accuracy: 0.4860 - val_loss: 1.2663 - val_categorical_accuracy: 0.5398\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 1.3830 - categorical_accuracy: 0.4939 - val_loss: 1.2476 - val_categorical_accuracy: 0.5414\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.3597 - categorical_accuracy: 0.4983 - val_loss: 1.2359 - val_categorical_accuracy: 0.5460\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 20s 29ms/step - loss: 1.3332 - categorical_accuracy: 0.5111 - val_loss: 1.2015 - val_categorical_accuracy: 0.5586\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.3124 - categorical_accuracy: 0.5199 - val_loss: 1.1861 - val_categorical_accuracy: 0.5630\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 1.2896 - categorical_accuracy: 0.5270 - val_loss: 1.1873 - val_categorical_accuracy: 0.5540\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.2656 - categorical_accuracy: 0.5394 - val_loss: 1.1643 - val_categorical_accuracy: 0.5722\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 1.2528 - categorical_accuracy: 0.5436 - val_loss: 1.1877 - val_categorical_accuracy: 0.5642\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 1.2349 - categorical_accuracy: 0.5483 - val_loss: 1.1629 - val_categorical_accuracy: 0.5714\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.2250 - categorical_accuracy: 0.5518 - val_loss: 1.2051 - val_categorical_accuracy: 0.5614\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 1.2077 - categorical_accuracy: 0.5562 - val_loss: 1.1801 - val_categorical_accuracy: 0.5658\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 1.1905 - categorical_accuracy: 0.5622 - val_loss: 1.2027 - val_categorical_accuracy: 0.5614\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 1.1721 - categorical_accuracy: 0.5669 - val_loss: 1.1526 - val_categorical_accuracy: 0.5816\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 15s 22ms/step - loss: 1.1604 - categorical_accuracy: 0.5744 - val_loss: 1.1729 - val_categorical_accuracy: 0.5854\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 15s 22ms/step - loss: 1.1495 - categorical_accuracy: 0.5781 - val_loss: 1.1692 - val_categorical_accuracy: 0.5802\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 15s 22ms/step - loss: 1.1333 - categorical_accuracy: 0.5834 - val_loss: 1.1729 - val_categorical_accuracy: 0.5922\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 15s 22ms/step - loss: 1.1192 - categorical_accuracy: 0.5923 - val_loss: 1.1430 - val_categorical_accuracy: 0.5856\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 1.0953 - categorical_accuracy: 0.5961 - val_loss: 1.1511 - val_categorical_accuracy: 0.5880\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 14s 21ms/step - loss: 1.0830 - categorical_accuracy: 0.6014 - val_loss: 1.1587 - val_categorical_accuracy: 0.5880\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 1.0734 - categorical_accuracy: 0.6045 - val_loss: 1.1851 - val_categorical_accuracy: 0.5916\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 1.0553 - categorical_accuracy: 0.6098 - val_loss: 1.1519 - val_categorical_accuracy: 0.5982\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.0489 - categorical_accuracy: 0.6137 - val_loss: 1.1662 - val_categorical_accuracy: 0.5950\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 1.0404 - categorical_accuracy: 0.6152 - val_loss: 1.1417 - val_categorical_accuracy: 0.6014\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 1.0238 - categorical_accuracy: 0.6189 - val_loss: 1.1624 - val_categorical_accuracy: 0.5938\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 1.0146 - categorical_accuracy: 0.6234 - val_loss: 1.1582 - val_categorical_accuracy: 0.6026\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 1.0112 - categorical_accuracy: 0.6269 - val_loss: 1.1753 - val_categorical_accuracy: 0.5986\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.9968 - categorical_accuracy: 0.6308 - val_loss: 1.1596 - val_categorical_accuracy: 0.6030\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.9758 - categorical_accuracy: 0.6387 - val_loss: 1.2236 - val_categorical_accuracy: 0.6020\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.9652 - categorical_accuracy: 0.6408 - val_loss: 1.1747 - val_categorical_accuracy: 0.6060\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.9552 - categorical_accuracy: 0.6421 - val_loss: 1.2063 - val_categorical_accuracy: 0.6094\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 0.9500 - categorical_accuracy: 0.6441 - val_loss: 1.1998 - val_categorical_accuracy: 0.6046\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.9403 - categorical_accuracy: 0.6510 - val_loss: 1.2170 - val_categorical_accuracy: 0.5984\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 15s 22ms/step - loss: 0.9309 - categorical_accuracy: 0.6505 - val_loss: 1.2687 - val_categorical_accuracy: 0.5966\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.9199 - categorical_accuracy: 0.6544 - val_loss: 1.2151 - val_categorical_accuracy: 0.6076\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.9092 - categorical_accuracy: 0.6566 - val_loss: 1.2890 - val_categorical_accuracy: 0.5960\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8991 - categorical_accuracy: 0.6628 - val_loss: 1.2881 - val_categorical_accuracy: 0.5958\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.8882 - categorical_accuracy: 0.6648 - val_loss: 1.2834 - val_categorical_accuracy: 0.5980\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 15s 22ms/step - loss: 0.8809 - categorical_accuracy: 0.6665 - val_loss: 1.2783 - val_categorical_accuracy: 0.6016\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.8725 - categorical_accuracy: 0.6701 - val_loss: 1.2755 - val_categorical_accuracy: 0.6070\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.8663 - categorical_accuracy: 0.6737 - val_loss: 1.2980 - val_categorical_accuracy: 0.6046\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.8500 - categorical_accuracy: 0.6755 - val_loss: 1.3561 - val_categorical_accuracy: 0.6104\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.8472 - categorical_accuracy: 0.6781 - val_loss: 1.3784 - val_categorical_accuracy: 0.6008\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.8343 - categorical_accuracy: 0.6832 - val_loss: 1.3258 - val_categorical_accuracy: 0.6074\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.8273 - categorical_accuracy: 0.6864 - val_loss: 1.3682 - val_categorical_accuracy: 0.6060\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.8228 - categorical_accuracy: 0.6878 - val_loss: 1.3839 - val_categorical_accuracy: 0.6084\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.8067 - categorical_accuracy: 0.6894 - val_loss: 1.3828 - val_categorical_accuracy: 0.6072\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 0.7951 - categorical_accuracy: 0.6970 - val_loss: 1.4641 - val_categorical_accuracy: 0.5986\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 0.7900 - categorical_accuracy: 0.6980 - val_loss: 1.3971 - val_categorical_accuracy: 0.6096\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.7784 - categorical_accuracy: 0.7010 - val_loss: 1.3891 - val_categorical_accuracy: 0.5968\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 0.7856 - categorical_accuracy: 0.6959 - val_loss: 1.4677 - val_categorical_accuracy: 0.6046\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 0.7648 - categorical_accuracy: 0.7044 - val_loss: 1.4282 - val_categorical_accuracy: 0.6088\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7593 - categorical_accuracy: 0.7074 - val_loss: 1.5282 - val_categorical_accuracy: 0.5992\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.7543 - categorical_accuracy: 0.7096 - val_loss: 1.5402 - val_categorical_accuracy: 0.5980\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.7450 - categorical_accuracy: 0.7123 - val_loss: 1.5296 - val_categorical_accuracy: 0.5948\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.7423 - categorical_accuracy: 0.7121 - val_loss: 1.4886 - val_categorical_accuracy: 0.6082\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.7348 - categorical_accuracy: 0.7156 - val_loss: 1.4906 - val_categorical_accuracy: 0.6046\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7343 - categorical_accuracy: 0.7161 - val_loss: 1.5149 - val_categorical_accuracy: 0.6120\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.7261 - categorical_accuracy: 0.7186 - val_loss: 1.5244 - val_categorical_accuracy: 0.6000\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.7095 - categorical_accuracy: 0.7251 - val_loss: 1.5344 - val_categorical_accuracy: 0.6080\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 20s 29ms/step - loss: 0.7216 - categorical_accuracy: 0.7216 - val_loss: 1.6314 - val_categorical_accuracy: 0.5984\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6987 - categorical_accuracy: 0.7274 - val_loss: 1.5954 - val_categorical_accuracy: 0.6008\n",
      "Epoch 72/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.6972 - categorical_accuracy: 0.7280 - val_loss: 1.6154 - val_categorical_accuracy: 0.5992\n",
      "Epoch 73/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.7084 - categorical_accuracy: 0.7261 - val_loss: 1.5602 - val_categorical_accuracy: 0.5968\n",
      "Epoch 74/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6926 - categorical_accuracy: 0.7298 - val_loss: 1.6330 - val_categorical_accuracy: 0.5940\n",
      "Epoch 75/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.6852 - categorical_accuracy: 0.7314 - val_loss: 1.6874 - val_categorical_accuracy: 0.6042\n",
      "Epoch 76/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6946 - categorical_accuracy: 0.7300 - val_loss: 1.6421 - val_categorical_accuracy: 0.6048\n",
      "Epoch 77/100\n",
      "704/704 [==============================] - 17s 23ms/step - loss: 0.6765 - categorical_accuracy: 0.7381 - val_loss: 1.5851 - val_categorical_accuracy: 0.6020\n",
      "Epoch 78/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6777 - categorical_accuracy: 0.7361 - val_loss: 1.6169 - val_categorical_accuracy: 0.6058\n",
      "Epoch 79/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6702 - categorical_accuracy: 0.7365 - val_loss: 1.7175 - val_categorical_accuracy: 0.5970\n",
      "Epoch 80/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.6623 - categorical_accuracy: 0.7395 - val_loss: 1.6811 - val_categorical_accuracy: 0.6014\n",
      "Epoch 81/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6629 - categorical_accuracy: 0.7381 - val_loss: 1.6508 - val_categorical_accuracy: 0.6030\n",
      "Epoch 82/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.6592 - categorical_accuracy: 0.7423 - val_loss: 1.7176 - val_categorical_accuracy: 0.6000\n",
      "Epoch 83/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.6548 - categorical_accuracy: 0.7443 - val_loss: 1.6907 - val_categorical_accuracy: 0.6028\n",
      "Epoch 84/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6583 - categorical_accuracy: 0.7440 - val_loss: 1.6871 - val_categorical_accuracy: 0.5996\n",
      "Epoch 85/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.6542 - categorical_accuracy: 0.7445 - val_loss: 1.8370 - val_categorical_accuracy: 0.5930\n",
      "Epoch 86/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.6420 - categorical_accuracy: 0.7504 - val_loss: 1.7410 - val_categorical_accuracy: 0.6028\n",
      "Epoch 87/100\n",
      "704/704 [==============================] - 15s 22ms/step - loss: 0.6373 - categorical_accuracy: 0.7496 - val_loss: 1.7843 - val_categorical_accuracy: 0.5996\n",
      "Epoch 88/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.6353 - categorical_accuracy: 0.7508 - val_loss: 1.8351 - val_categorical_accuracy: 0.5992\n",
      "Epoch 89/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.6366 - categorical_accuracy: 0.7492 - val_loss: 1.7823 - val_categorical_accuracy: 0.6004\n",
      "Epoch 90/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.6307 - categorical_accuracy: 0.7534 - val_loss: 1.8504 - val_categorical_accuracy: 0.5954\n",
      "Epoch 91/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 0.6263 - categorical_accuracy: 0.7544 - val_loss: 1.8443 - val_categorical_accuracy: 0.6038\n",
      "Epoch 92/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.6223 - categorical_accuracy: 0.7555 - val_loss: 1.8228 - val_categorical_accuracy: 0.5984\n",
      "Epoch 93/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.6118 - categorical_accuracy: 0.7609 - val_loss: 1.8869 - val_categorical_accuracy: 0.5976\n",
      "Epoch 94/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 0.6050 - categorical_accuracy: 0.7636 - val_loss: 2.0015 - val_categorical_accuracy: 0.5922\n",
      "Epoch 95/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.6082 - categorical_accuracy: 0.7625 - val_loss: 1.9687 - val_categorical_accuracy: 0.5894\n",
      "Epoch 96/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.5985 - categorical_accuracy: 0.7651 - val_loss: 1.9772 - val_categorical_accuracy: 0.5880\n",
      "Epoch 97/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 20s 29ms/step - loss: 0.6056 - categorical_accuracy: 0.7634 - val_loss: 1.9225 - val_categorical_accuracy: 0.5884\n",
      "--- Running training session 2\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.3087 - categorical_accuracy: 0.1011 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3035 - categorical_accuracy: 0.1014 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 2.3035 - categorical_accuracy: 0.1015 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 2.3035 - categorical_accuracy: 0.1015 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 19s 28ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 14s 21ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "--- Running training session 3\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 3.3538 - categorical_accuracy: 0.1038 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.3048 - categorical_accuracy: 0.1039 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3048 - categorical_accuracy: 0.1038 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3048 - categorical_accuracy: 0.1038 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1038 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3048 - categorical_accuracy: 0.1038 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 17/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 19s 28ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "--- Running training session 4\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 1.9055 - categorical_accuracy: 0.2754 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 1.5714 - categorical_accuracy: 0.4204 - val_loss: 1.3370 - val_categorical_accuracy: 0.5078\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 1.4065 - categorical_accuracy: 0.4924 - val_loss: 1.2398 - val_categorical_accuracy: 0.5498\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 1.2897 - categorical_accuracy: 0.5390 - val_loss: 1.1276 - val_categorical_accuracy: 0.5870\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.1865 - categorical_accuracy: 0.5776 - val_loss: 1.0690 - val_categorical_accuracy: 0.6124\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.1055 - categorical_accuracy: 0.6103 - val_loss: 1.0275 - val_categorical_accuracy: 0.6310\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.0295 - categorical_accuracy: 0.6381 - val_loss: 0.9734 - val_categorical_accuracy: 0.6612\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.9606 - categorical_accuracy: 0.6610 - val_loss: 0.9278 - val_categorical_accuracy: 0.6772\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.9111 - categorical_accuracy: 0.6792 - val_loss: 1.0047 - val_categorical_accuracy: 0.6546\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.8616 - categorical_accuracy: 0.6935 - val_loss: 0.9572 - val_categorical_accuracy: 0.6752\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.8106 - categorical_accuracy: 0.7135 - val_loss: 0.9513 - val_categorical_accuracy: 0.6872\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.7705 - categorical_accuracy: 0.7264 - val_loss: 0.9849 - val_categorical_accuracy: 0.6922\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.7310 - categorical_accuracy: 0.7390 - val_loss: 0.9947 - val_categorical_accuracy: 0.6886\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.6878 - categorical_accuracy: 0.7546 - val_loss: 0.9443 - val_categorical_accuracy: 0.7076\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.6494 - categorical_accuracy: 0.7695 - val_loss: 1.0326 - val_categorical_accuracy: 0.7008\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.6206 - categorical_accuracy: 0.7790 - val_loss: 1.0007 - val_categorical_accuracy: 0.7106\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5866 - categorical_accuracy: 0.7918 - val_loss: 1.0610 - val_categorical_accuracy: 0.7084\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.5687 - categorical_accuracy: 0.7952 - val_loss: 1.0866 - val_categorical_accuracy: 0.7118\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.5339 - categorical_accuracy: 0.8109 - val_loss: 1.0881 - val_categorical_accuracy: 0.7100\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.5064 - categorical_accuracy: 0.8179 - val_loss: 1.2089 - val_categorical_accuracy: 0.6988\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.4808 - categorical_accuracy: 0.8289 - val_loss: 1.1790 - val_categorical_accuracy: 0.7142\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.4551 - categorical_accuracy: 0.8382 - val_loss: 1.2702 - val_categorical_accuracy: 0.7004\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.4299 - categorical_accuracy: 0.8458 - val_loss: 1.3953 - val_categorical_accuracy: 0.7030\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.4262 - categorical_accuracy: 0.8502 - val_loss: 1.3318 - val_categorical_accuracy: 0.7090\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.4023 - categorical_accuracy: 0.8579 - val_loss: 1.4344 - val_categorical_accuracy: 0.7120\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.4042 - categorical_accuracy: 0.8581 - val_loss: 1.3323 - val_categorical_accuracy: 0.7190\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.3646 - categorical_accuracy: 0.8720 - val_loss: 1.3342 - val_categorical_accuracy: 0.7176\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 0.3510 - categorical_accuracy: 0.8769 - val_loss: 1.3574 - val_categorical_accuracy: 0.7196\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 0.3264 - categorical_accuracy: 0.8846 - val_loss: 1.3685 - val_categorical_accuracy: 0.7198\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.3273 - categorical_accuracy: 0.8873 - val_loss: 1.4423 - val_categorical_accuracy: 0.7108\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.3008 - categorical_accuracy: 0.8947 - val_loss: 1.4657 - val_categorical_accuracy: 0.7242\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.2919 - categorical_accuracy: 0.9005 - val_loss: 1.4442 - val_categorical_accuracy: 0.7230\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.2755 - categorical_accuracy: 0.9048 - val_loss: 1.6222 - val_categorical_accuracy: 0.7100\n",
      "Epoch 34/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 21s 29ms/step - loss: 0.2612 - categorical_accuracy: 0.9087 - val_loss: 1.6809 - val_categorical_accuracy: 0.7122\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.2609 - categorical_accuracy: 0.9134 - val_loss: 1.4529 - val_categorical_accuracy: 0.7202\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.2402 - categorical_accuracy: 0.9164 - val_loss: 1.8888 - val_categorical_accuracy: 0.7026\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.2361 - categorical_accuracy: 0.9202 - val_loss: 1.6141 - val_categorical_accuracy: 0.7196\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.2294 - categorical_accuracy: 0.9240 - val_loss: 1.6579 - val_categorical_accuracy: 0.7262\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 0.2239 - categorical_accuracy: 0.9252 - val_loss: 1.8094 - val_categorical_accuracy: 0.7142\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.2256 - categorical_accuracy: 0.9260 - val_loss: 1.6316 - val_categorical_accuracy: 0.7216\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.2096 - categorical_accuracy: 0.9307 - val_loss: 1.7704 - val_categorical_accuracy: 0.7138\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.1942 - categorical_accuracy: 0.9351 - val_loss: 1.9234 - val_categorical_accuracy: 0.7150\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.1955 - categorical_accuracy: 0.9352 - val_loss: 1.7876 - val_categorical_accuracy: 0.7184\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.1936 - categorical_accuracy: 0.9374 - val_loss: 1.8239 - val_categorical_accuracy: 0.7186\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.1847 - categorical_accuracy: 0.9402 - val_loss: 1.8491 - val_categorical_accuracy: 0.7132\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1778 - categorical_accuracy: 0.9427 - val_loss: 1.9974 - val_categorical_accuracy: 0.7204\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.1728 - categorical_accuracy: 0.9438 - val_loss: 1.8972 - val_categorical_accuracy: 0.7170\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.1720 - categorical_accuracy: 0.9459 - val_loss: 1.9343 - val_categorical_accuracy: 0.7140\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.1726 - categorical_accuracy: 0.9448 - val_loss: 1.9265 - val_categorical_accuracy: 0.7178\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.1548 - categorical_accuracy: 0.9487 - val_loss: 1.8954 - val_categorical_accuracy: 0.7196\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 22s 31ms/step - loss: 0.1671 - categorical_accuracy: 0.9476 - val_loss: 1.9679 - val_categorical_accuracy: 0.7072\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.1584 - categorical_accuracy: 0.9503 - val_loss: 1.9570 - val_categorical_accuracy: 0.7134\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.1547 - categorical_accuracy: 0.9494 - val_loss: 2.1521 - val_categorical_accuracy: 0.7222\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.1426 - categorical_accuracy: 0.9547 - val_loss: 2.2010 - val_categorical_accuracy: 0.7148\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.1457 - categorical_accuracy: 0.9542 - val_loss: 2.2607 - val_categorical_accuracy: 0.7146\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 0.1440 - categorical_accuracy: 0.9551 - val_loss: 2.1199 - val_categorical_accuracy: 0.7208\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1425 - categorical_accuracy: 0.9544 - val_loss: 2.1831 - val_categorical_accuracy: 0.7188\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.1422 - categorical_accuracy: 0.9557 - val_loss: 2.2095 - val_categorical_accuracy: 0.7072\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.1389 - categorical_accuracy: 0.9567 - val_loss: 2.1922 - val_categorical_accuracy: 0.7256\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.1366 - categorical_accuracy: 0.9568 - val_loss: 2.3858 - val_categorical_accuracy: 0.7100\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1413 - categorical_accuracy: 0.9568 - val_loss: 2.3578 - val_categorical_accuracy: 0.7150\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.1249 - categorical_accuracy: 0.9616 - val_loss: 2.4344 - val_categorical_accuracy: 0.7238\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 15s 22ms/step - loss: 0.1222 - categorical_accuracy: 0.9611 - val_loss: 2.4289 - val_categorical_accuracy: 0.7188\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1418 - categorical_accuracy: 0.9569 - val_loss: 2.3315 - val_categorical_accuracy: 0.7218\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 15s 22ms/step - loss: 0.1384 - categorical_accuracy: 0.9574 - val_loss: 2.2657 - val_categorical_accuracy: 0.7148\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 15s 22ms/step - loss: 0.1191 - categorical_accuracy: 0.9631 - val_loss: 2.4449 - val_categorical_accuracy: 0.7126\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1226 - categorical_accuracy: 0.9612 - val_loss: 2.4675 - val_categorical_accuracy: 0.7136\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.1156 - categorical_accuracy: 0.9645 - val_loss: 2.3150 - val_categorical_accuracy: 0.7216\n",
      "--- Running training session 5\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 3.0691 - categorical_accuracy: 0.1017 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.3035 - categorical_accuracy: 0.1017 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3035 - categorical_accuracy: 0.1017 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 15s 22ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 20s 29ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 20s 29ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 20s 29ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "--- Running training session 6\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 23.2937 - categorical_accuracy: 0.1033 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 2.3048 - categorical_accuracy: 0.1039 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 2.3048 - categorical_accuracy: 0.1039 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 15s 22ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 31/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 21s 29ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "--- Running training session 7\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.1123 - categorical_accuracy: 0.1919 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 1.6973 - categorical_accuracy: 0.3564 - val_loss: 1.4489 - val_categorical_accuracy: 0.4832\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 14s 21ms/step - loss: 1.4819 - categorical_accuracy: 0.4522 - val_loss: 1.3307 - val_categorical_accuracy: 0.5122\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 1.2986 - categorical_accuracy: 0.5327 - val_loss: 1.1253 - val_categorical_accuracy: 0.5968\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 1.1678 - categorical_accuracy: 0.5841 - val_loss: 1.1156 - val_categorical_accuracy: 0.6168\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 15s 22ms/step - loss: 1.0515 - categorical_accuracy: 0.6298 - val_loss: 1.0532 - val_categorical_accuracy: 0.6444\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.9764 - categorical_accuracy: 0.6531 - val_loss: 0.9656 - val_categorical_accuracy: 0.6650\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.8927 - categorical_accuracy: 0.6883 - val_loss: 0.9747 - val_categorical_accuracy: 0.6706\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8087 - categorical_accuracy: 0.7152 - val_loss: 1.0053 - val_categorical_accuracy: 0.6696\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.7447 - categorical_accuracy: 0.7363 - val_loss: 0.9990 - val_categorical_accuracy: 0.6766\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.6975 - categorical_accuracy: 0.7546 - val_loss: 1.0464 - val_categorical_accuracy: 0.6744\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.6404 - categorical_accuracy: 0.7747 - val_loss: 1.0503 - val_categorical_accuracy: 0.6876\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.5926 - categorical_accuracy: 0.7927 - val_loss: 1.0730 - val_categorical_accuracy: 0.6868\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 22s 31ms/step - loss: 0.5593 - categorical_accuracy: 0.8014 - val_loss: 1.0653 - val_categorical_accuracy: 0.6930\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.5214 - categorical_accuracy: 0.8166 - val_loss: 1.1817 - val_categorical_accuracy: 0.7040\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.4916 - categorical_accuracy: 0.8249 - val_loss: 1.2611 - val_categorical_accuracy: 0.7094\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.4456 - categorical_accuracy: 0.8425 - val_loss: 1.2495 - val_categorical_accuracy: 0.7136\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.4160 - categorical_accuracy: 0.8531 - val_loss: 1.3888 - val_categorical_accuracy: 0.7008\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.3882 - categorical_accuracy: 0.8606 - val_loss: 1.3086 - val_categorical_accuracy: 0.7134\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.3623 - categorical_accuracy: 0.8741 - val_loss: 1.3360 - val_categorical_accuracy: 0.7038\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.3446 - categorical_accuracy: 0.8790 - val_loss: 1.4172 - val_categorical_accuracy: 0.7094\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.3270 - categorical_accuracy: 0.8878 - val_loss: 1.4403 - val_categorical_accuracy: 0.7106\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.3123 - categorical_accuracy: 0.8921 - val_loss: 1.4329 - val_categorical_accuracy: 0.7136\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.2927 - categorical_accuracy: 0.8996 - val_loss: 1.4473 - val_categorical_accuracy: 0.7092\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.2640 - categorical_accuracy: 0.9075 - val_loss: 1.4426 - val_categorical_accuracy: 0.7204\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 19s 28ms/step - loss: 0.2690 - categorical_accuracy: 0.9078 - val_loss: 1.5962 - val_categorical_accuracy: 0.7258\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 17s 23ms/step - loss: 0.2640 - categorical_accuracy: 0.9106 - val_loss: 1.6457 - val_categorical_accuracy: 0.7144\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.2488 - categorical_accuracy: 0.9165 - val_loss: 1.7037 - val_categorical_accuracy: 0.7036\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.2264 - categorical_accuracy: 0.9238 - val_loss: 1.6265 - val_categorical_accuracy: 0.7090\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.2295 - categorical_accuracy: 0.9234 - val_loss: 1.5562 - val_categorical_accuracy: 0.7178\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.2101 - categorical_accuracy: 0.9292 - val_loss: 1.8484 - val_categorical_accuracy: 0.7140\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.2124 - categorical_accuracy: 0.9311 - val_loss: 1.7112 - val_categorical_accuracy: 0.7148\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.2034 - categorical_accuracy: 0.9333 - val_loss: 1.8558 - val_categorical_accuracy: 0.7150\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 0.2061 - categorical_accuracy: 0.9325 - val_loss: 2.0082 - val_categorical_accuracy: 0.7008\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1874 - categorical_accuracy: 0.9394 - val_loss: 1.9070 - val_categorical_accuracy: 0.7110\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1916 - categorical_accuracy: 0.9378 - val_loss: 1.7988 - val_categorical_accuracy: 0.7070\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 15s 22ms/step - loss: 0.1607 - categorical_accuracy: 0.9472 - val_loss: 1.9039 - val_categorical_accuracy: 0.7042\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.1761 - categorical_accuracy: 0.9432 - val_loss: 1.9749 - val_categorical_accuracy: 0.6990\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.1688 - categorical_accuracy: 0.9448 - val_loss: 1.9098 - val_categorical_accuracy: 0.7074\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1685 - categorical_accuracy: 0.9450 - val_loss: 2.0480 - val_categorical_accuracy: 0.7144\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.1577 - categorical_accuracy: 0.9498 - val_loss: 2.0549 - val_categorical_accuracy: 0.7034\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.1582 - categorical_accuracy: 0.9482 - val_loss: 2.0362 - val_categorical_accuracy: 0.7164\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.1481 - categorical_accuracy: 0.9531 - val_loss: 1.9664 - val_categorical_accuracy: 0.7060\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 14s 21ms/step - loss: 0.1586 - categorical_accuracy: 0.9504 - val_loss: 1.9549 - val_categorical_accuracy: 0.7114\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.1401 - categorical_accuracy: 0.9563 - val_loss: 1.9358 - val_categorical_accuracy: 0.7180\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 15s 22ms/step - loss: 0.1435 - categorical_accuracy: 0.9540 - val_loss: 2.2915 - val_categorical_accuracy: 0.6916\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.1443 - categorical_accuracy: 0.9546 - val_loss: 1.9621 - val_categorical_accuracy: 0.7126\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.1381 - categorical_accuracy: 0.9562 - val_loss: 2.1066 - val_categorical_accuracy: 0.7064\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 20s 29ms/step - loss: 0.1356 - categorical_accuracy: 0.9571 - val_loss: 2.1267 - val_categorical_accuracy: 0.7172\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.1317 - categorical_accuracy: 0.9601 - val_loss: 2.0423 - val_categorical_accuracy: 0.7090\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 0.1341 - categorical_accuracy: 0.9600 - val_loss: 2.3750 - val_categorical_accuracy: 0.7050\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1339 - categorical_accuracy: 0.9584 - val_loss: 2.1688 - val_categorical_accuracy: 0.7188\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.1256 - categorical_accuracy: 0.9604 - val_loss: 2.1422 - val_categorical_accuracy: 0.7214\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 0.1360 - categorical_accuracy: 0.9581 - val_loss: 2.0137 - val_categorical_accuracy: 0.7186\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.1261 - categorical_accuracy: 0.9608 - val_loss: 2.2963 - val_categorical_accuracy: 0.7160\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.1187 - categorical_accuracy: 0.9633 - val_loss: 2.0781 - val_categorical_accuracy: 0.7172\n",
      "--- Running training session 8\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.7674 - categorical_accuracy: 0.1023 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3035 - categorical_accuracy: 0.1017 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3035 - categorical_accuracy: 0.1014 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3035 - categorical_accuracy: 0.1015 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 15s 22ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 14s 21ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 15s 22ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3035 - categorical_accuracy: 0.1016 - val_loss: 2.3024 - val_categorical_accuracy: 0.1038\n",
      "--- Running training session 9\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 81.2991 - categorical_accuracy: 0.1026 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1039 - val_loss: 2.3036 - val_categorical_accuracy: 0.1038\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.3048 - categorical_accuracy: 0.1044 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1038 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1038 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 9/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 19s 28ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3048 - categorical_accuracy: 0.1037 - val_loss: 2.3037 - val_categorical_accuracy: 0.1038\n",
      "--- Running training session 10\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 19s 28ms/step - loss: 1.6066 - categorical_accuracy: 0.4276 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.1760 - categorical_accuracy: 0.5869 - val_loss: 1.2121 - val_categorical_accuracy: 0.5692\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.0052 - categorical_accuracy: 0.6553 - val_loss: 1.0346 - val_categorical_accuracy: 0.6202\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8801 - categorical_accuracy: 0.6952 - val_loss: 1.2469 - val_categorical_accuracy: 0.5704\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7839 - categorical_accuracy: 0.7337 - val_loss: 1.2577 - val_categorical_accuracy: 0.5906\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7071 - categorical_accuracy: 0.7590 - val_loss: 0.9975 - val_categorical_accuracy: 0.6590\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6280 - categorical_accuracy: 0.7862 - val_loss: 0.9465 - val_categorical_accuracy: 0.6804\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.5651 - categorical_accuracy: 0.8104 - val_loss: 0.8541 - val_categorical_accuracy: 0.7098\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.5108 - categorical_accuracy: 0.8258 - val_loss: 0.9690 - val_categorical_accuracy: 0.6998\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.4541 - categorical_accuracy: 0.8444 - val_loss: 0.8214 - val_categorical_accuracy: 0.7374\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4057 - categorical_accuracy: 0.8598 - val_loss: 0.7409 - val_categorical_accuracy: 0.7736\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.3542 - categorical_accuracy: 0.8789 - val_loss: 1.2034 - val_categorical_accuracy: 0.6998\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.3226 - categorical_accuracy: 0.8882 - val_loss: 0.6974 - val_categorical_accuracy: 0.7854\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.2861 - categorical_accuracy: 0.8999 - val_loss: 0.7966 - val_categorical_accuracy: 0.7642\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.2535 - categorical_accuracy: 0.9123 - val_loss: 0.9874 - val_categorical_accuracy: 0.7354\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.2291 - categorical_accuracy: 0.9202 - val_loss: 1.5671 - val_categorical_accuracy: 0.6410\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1992 - categorical_accuracy: 0.9313 - val_loss: 1.0739 - val_categorical_accuracy: 0.7496\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1799 - categorical_accuracy: 0.9386 - val_loss: 1.4612 - val_categorical_accuracy: 0.6928\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1595 - categorical_accuracy: 0.9455 - val_loss: 1.0226 - val_categorical_accuracy: 0.7606\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1454 - categorical_accuracy: 0.9495 - val_loss: 0.9770 - val_categorical_accuracy: 0.7716\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1513 - categorical_accuracy: 0.9480 - val_loss: 0.9862 - val_categorical_accuracy: 0.7868\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1349 - categorical_accuracy: 0.9532 - val_loss: 0.9141 - val_categorical_accuracy: 0.7984\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.1130 - categorical_accuracy: 0.9613 - val_loss: 0.9237 - val_categorical_accuracy: 0.7976\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1061 - categorical_accuracy: 0.9644 - val_loss: 1.2467 - val_categorical_accuracy: 0.7582\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1009 - categorical_accuracy: 0.9660 - val_loss: 1.0988 - val_categorical_accuracy: 0.7710\n",
      "Epoch 26/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0949 - categorical_accuracy: 0.9680 - val_loss: 1.4090 - val_categorical_accuracy: 0.7334\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0953 - categorical_accuracy: 0.9675 - val_loss: 1.0023 - val_categorical_accuracy: 0.7960\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.0873 - categorical_accuracy: 0.9704 - val_loss: 0.9816 - val_categorical_accuracy: 0.7954\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0796 - categorical_accuracy: 0.9729 - val_loss: 1.2687 - val_categorical_accuracy: 0.7780\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 20s 29ms/step - loss: 0.0791 - categorical_accuracy: 0.9739 - val_loss: 1.1028 - val_categorical_accuracy: 0.7892\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0748 - categorical_accuracy: 0.9755 - val_loss: 1.0798 - val_categorical_accuracy: 0.7976\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.0712 - categorical_accuracy: 0.9754 - val_loss: 1.0596 - val_categorical_accuracy: 0.7890\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0658 - categorical_accuracy: 0.9781 - val_loss: 1.3293 - val_categorical_accuracy: 0.7706\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.0673 - categorical_accuracy: 0.9774 - val_loss: 1.4750 - val_categorical_accuracy: 0.7536\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 19s 28ms/step - loss: 0.0634 - categorical_accuracy: 0.9791 - val_loss: 1.2811 - val_categorical_accuracy: 0.7766\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 24s 35ms/step - loss: 0.0604 - categorical_accuracy: 0.9796 - val_loss: 1.0593 - val_categorical_accuracy: 0.8062\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.0628 - categorical_accuracy: 0.9796 - val_loss: 1.0987 - val_categorical_accuracy: 0.8152\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.0585 - categorical_accuracy: 0.9809 - val_loss: 1.3306 - val_categorical_accuracy: 0.7890\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0575 - categorical_accuracy: 0.9812 - val_loss: 1.3471 - val_categorical_accuracy: 0.7872\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0549 - categorical_accuracy: 0.9816 - val_loss: 1.1842 - val_categorical_accuracy: 0.7948\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0632 - categorical_accuracy: 0.9792 - val_loss: 1.8539 - val_categorical_accuracy: 0.7164\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.0543 - categorical_accuracy: 0.9828 - val_loss: 1.0845 - val_categorical_accuracy: 0.8098\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0484 - categorical_accuracy: 0.9839 - val_loss: 1.5708 - val_categorical_accuracy: 0.7664\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0447 - categorical_accuracy: 0.9853 - val_loss: 1.3213 - val_categorical_accuracy: 0.7920\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0419 - categorical_accuracy: 0.9864 - val_loss: 1.2894 - val_categorical_accuracy: 0.8042\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.0509 - categorical_accuracy: 0.9830 - val_loss: 1.5896 - val_categorical_accuracy: 0.7734\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0431 - categorical_accuracy: 0.9853 - val_loss: 1.1621 - val_categorical_accuracy: 0.8180\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0373 - categorical_accuracy: 0.9876 - val_loss: 1.1996 - val_categorical_accuracy: 0.8148\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0503 - categorical_accuracy: 0.9840 - val_loss: 1.7014 - val_categorical_accuracy: 0.7410\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0455 - categorical_accuracy: 0.9851 - val_loss: 1.2359 - val_categorical_accuracy: 0.7958\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0374 - categorical_accuracy: 0.9876 - val_loss: 1.3637 - val_categorical_accuracy: 0.7906\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0413 - categorical_accuracy: 0.9865 - val_loss: 1.2036 - val_categorical_accuracy: 0.8152\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0376 - categorical_accuracy: 0.9884 - val_loss: 1.5947 - val_categorical_accuracy: 0.7700\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0447 - categorical_accuracy: 0.9854 - val_loss: 1.3333 - val_categorical_accuracy: 0.7960\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0403 - categorical_accuracy: 0.9867 - val_loss: 1.3076 - val_categorical_accuracy: 0.7976\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0368 - categorical_accuracy: 0.9880 - val_loss: 1.4769 - val_categorical_accuracy: 0.7734\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0304 - categorical_accuracy: 0.9898 - val_loss: 1.3919 - val_categorical_accuracy: 0.7980\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0378 - categorical_accuracy: 0.9874 - val_loss: 1.5450 - val_categorical_accuracy: 0.7758\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0405 - categorical_accuracy: 0.9870 - val_loss: 1.5085 - val_categorical_accuracy: 0.7854\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0348 - categorical_accuracy: 0.9882 - val_loss: 1.4772 - val_categorical_accuracy: 0.7860\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0289 - categorical_accuracy: 0.9908 - val_loss: 1.4698 - val_categorical_accuracy: 0.7822\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0389 - categorical_accuracy: 0.9872 - val_loss: 1.5778 - val_categorical_accuracy: 0.7740\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0333 - categorical_accuracy: 0.9892 - val_loss: 1.3087 - val_categorical_accuracy: 0.8116\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0278 - categorical_accuracy: 0.9909 - val_loss: 1.4379 - val_categorical_accuracy: 0.8086\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0295 - categorical_accuracy: 0.9901 - val_loss: 1.2286 - val_categorical_accuracy: 0.8202\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0299 - categorical_accuracy: 0.9904 - val_loss: 1.6363 - val_categorical_accuracy: 0.7800\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0372 - categorical_accuracy: 0.9885 - val_loss: 1.3341 - val_categorical_accuracy: 0.8106\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0293 - categorical_accuracy: 0.9898 - val_loss: 1.2840 - val_categorical_accuracy: 0.8168\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0262 - categorical_accuracy: 0.9910 - val_loss: 1.4032 - val_categorical_accuracy: 0.8084\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.0270 - categorical_accuracy: 0.9911 - val_loss: 1.4309 - val_categorical_accuracy: 0.8070\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.0365 - categorical_accuracy: 0.9883 - val_loss: 1.5142 - val_categorical_accuracy: 0.7876\n",
      "Epoch 72/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0261 - categorical_accuracy: 0.9916 - val_loss: 1.3798 - val_categorical_accuracy: 0.7988\n",
      "Epoch 73/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0308 - categorical_accuracy: 0.9907 - val_loss: 1.3457 - val_categorical_accuracy: 0.8104\n",
      "Epoch 74/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0232 - categorical_accuracy: 0.9927 - val_loss: 1.3788 - val_categorical_accuracy: 0.8120\n",
      "Epoch 75/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0273 - categorical_accuracy: 0.9914 - val_loss: 1.6070 - val_categorical_accuracy: 0.7872\n",
      "Epoch 76/100\n",
      "704/704 [==============================] - 22s 31ms/step - loss: 0.0294 - categorical_accuracy: 0.9903 - val_loss: 1.5151 - val_categorical_accuracy: 0.7972\n",
      "Epoch 77/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0228 - categorical_accuracy: 0.9924 - val_loss: 1.3463 - val_categorical_accuracy: 0.8192\n",
      "Epoch 78/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.0235 - categorical_accuracy: 0.9925 - val_loss: 1.5871 - val_categorical_accuracy: 0.8044\n",
      "Epoch 79/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.0313 - categorical_accuracy: 0.9899 - val_loss: 1.3217 - val_categorical_accuracy: 0.8184\n",
      "Epoch 80/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0231 - categorical_accuracy: 0.9923 - val_loss: 1.6465 - val_categorical_accuracy: 0.7870\n",
      "Epoch 81/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.0266 - categorical_accuracy: 0.9910 - val_loss: 1.2609 - val_categorical_accuracy: 0.8224\n",
      "Epoch 82/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0265 - categorical_accuracy: 0.9918 - val_loss: 1.7230 - val_categorical_accuracy: 0.7854\n",
      "Epoch 83/100\n",
      "704/704 [==============================] - 24s 34ms/step - loss: 0.0265 - categorical_accuracy: 0.9913 - val_loss: 1.4862 - val_categorical_accuracy: 0.7998\n",
      "Epoch 84/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0221 - categorical_accuracy: 0.9931 - val_loss: 1.2349 - val_categorical_accuracy: 0.8296\n",
      "Epoch 85/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0198 - categorical_accuracy: 0.9937 - val_loss: 1.6004 - val_categorical_accuracy: 0.8006\n",
      "Epoch 86/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0249 - categorical_accuracy: 0.9920 - val_loss: 2.0992 - val_categorical_accuracy: 0.7280\n",
      "Epoch 87/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0270 - categorical_accuracy: 0.9911 - val_loss: 1.3135 - val_categorical_accuracy: 0.8174\n",
      "Epoch 88/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.0226 - categorical_accuracy: 0.9927 - val_loss: 1.5292 - val_categorical_accuracy: 0.7940\n",
      "Epoch 89/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0185 - categorical_accuracy: 0.9940 - val_loss: 1.3443 - val_categorical_accuracy: 0.8186\n",
      "Epoch 90/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0201 - categorical_accuracy: 0.9935 - val_loss: 1.3977 - val_categorical_accuracy: 0.8280\n",
      "Epoch 91/100\n",
      "704/704 [==============================] - 23s 33ms/step - loss: 0.0249 - categorical_accuracy: 0.9919 - val_loss: 1.5740 - val_categorical_accuracy: 0.7910\n",
      "Epoch 92/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 0.0243 - categorical_accuracy: 0.9924 - val_loss: 1.4723 - val_categorical_accuracy: 0.8038\n",
      "Epoch 93/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0244 - categorical_accuracy: 0.9924 - val_loss: 1.7261 - val_categorical_accuracy: 0.7882\n",
      "Epoch 94/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0196 - categorical_accuracy: 0.9935 - val_loss: 1.4074 - val_categorical_accuracy: 0.8156\n",
      "Epoch 95/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0182 - categorical_accuracy: 0.9940 - val_loss: 1.7197 - val_categorical_accuracy: 0.7894\n",
      "Epoch 96/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0238 - categorical_accuracy: 0.9925 - val_loss: 2.1900 - val_categorical_accuracy: 0.7572\n",
      "Epoch 97/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0282 - categorical_accuracy: 0.9911 - val_loss: 1.3306 - val_categorical_accuracy: 0.8198\n",
      "Epoch 98/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0152 - categorical_accuracy: 0.9952 - val_loss: 1.5185 - val_categorical_accuracy: 0.8094\n",
      "Epoch 99/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0255 - categorical_accuracy: 0.9921 - val_loss: 1.4721 - val_categorical_accuracy: 0.8064\n",
      "Epoch 100/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0167 - categorical_accuracy: 0.9947 - val_loss: 1.3076 - val_categorical_accuracy: 0.8210\n",
      "--- Running training session 11\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 1.6661 - categorical_accuracy: 0.3884 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 1.2921 - categorical_accuracy: 0.5417 - val_loss: 1.7630 - val_categorical_accuracy: 0.4200\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 1.0951 - categorical_accuracy: 0.6140 - val_loss: 1.2792 - val_categorical_accuracy: 0.5716\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.9751 - categorical_accuracy: 0.6627 - val_loss: 1.8947 - val_categorical_accuracy: 0.4724\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.8804 - categorical_accuracy: 0.6944 - val_loss: 1.1129 - val_categorical_accuracy: 0.6180\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.7924 - categorical_accuracy: 0.7296 - val_loss: 0.9667 - val_categorical_accuracy: 0.6622\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7308 - categorical_accuracy: 0.7530 - val_loss: 1.6719 - val_categorical_accuracy: 0.5464\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.6572 - categorical_accuracy: 0.7771 - val_loss: 1.6458 - val_categorical_accuracy: 0.5456\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.6065 - categorical_accuracy: 0.7966 - val_loss: 1.2629 - val_categorical_accuracy: 0.6250\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.5400 - categorical_accuracy: 0.8166 - val_loss: 1.9122 - val_categorical_accuracy: 0.5542\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.4936 - categorical_accuracy: 0.8328 - val_loss: 1.7998 - val_categorical_accuracy: 0.5774\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.4572 - categorical_accuracy: 0.8430 - val_loss: 1.1384 - val_categorical_accuracy: 0.7032\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4084 - categorical_accuracy: 0.8604 - val_loss: 1.0596 - val_categorical_accuracy: 0.7122\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.3668 - categorical_accuracy: 0.8751 - val_loss: 1.1024 - val_categorical_accuracy: 0.7092\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.3302 - categorical_accuracy: 0.8865 - val_loss: 0.8881 - val_categorical_accuracy: 0.7660\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.3000 - categorical_accuracy: 0.8962 - val_loss: 0.9993 - val_categorical_accuracy: 0.7440\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.2794 - categorical_accuracy: 0.9037 - val_loss: 1.4746 - val_categorical_accuracy: 0.6666\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.2564 - categorical_accuracy: 0.9115 - val_loss: 1.0063 - val_categorical_accuracy: 0.7462\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.2313 - categorical_accuracy: 0.9210 - val_loss: 1.1107 - val_categorical_accuracy: 0.7418\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.2135 - categorical_accuracy: 0.9266 - val_loss: 1.3390 - val_categorical_accuracy: 0.7102\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.1951 - categorical_accuracy: 0.9330 - val_loss: 1.2860 - val_categorical_accuracy: 0.7394\n",
      "Epoch 22/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1918 - categorical_accuracy: 0.9359 - val_loss: 2.1154 - val_categorical_accuracy: 0.6206\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 24s 34ms/step - loss: 0.1731 - categorical_accuracy: 0.9412 - val_loss: 1.0919 - val_categorical_accuracy: 0.7708\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.1642 - categorical_accuracy: 0.9447 - val_loss: 1.0850 - val_categorical_accuracy: 0.7764\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 22s 32ms/step - loss: 0.1549 - categorical_accuracy: 0.9466 - val_loss: 1.2351 - val_categorical_accuracy: 0.7536\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.1427 - categorical_accuracy: 0.9524 - val_loss: 1.2497 - val_categorical_accuracy: 0.7572\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.1355 - categorical_accuracy: 0.9549 - val_loss: 1.5109 - val_categorical_accuracy: 0.7332\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1260 - categorical_accuracy: 0.9586 - val_loss: 1.1185 - val_categorical_accuracy: 0.7926\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 26s 37ms/step - loss: 0.1219 - categorical_accuracy: 0.9586 - val_loss: 1.2017 - val_categorical_accuracy: 0.7688\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 26s 37ms/step - loss: 0.1158 - categorical_accuracy: 0.9617 - val_loss: 1.4619 - val_categorical_accuracy: 0.7590\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 26s 37ms/step - loss: 0.1139 - categorical_accuracy: 0.9623 - val_loss: 1.3665 - val_categorical_accuracy: 0.7468\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1097 - categorical_accuracy: 0.9636 - val_loss: 1.2515 - val_categorical_accuracy: 0.7748\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.1014 - categorical_accuracy: 0.9659 - val_loss: 1.4300 - val_categorical_accuracy: 0.7634\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1004 - categorical_accuracy: 0.9667 - val_loss: 1.5082 - val_categorical_accuracy: 0.7618\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0989 - categorical_accuracy: 0.9682 - val_loss: 1.2074 - val_categorical_accuracy: 0.7842\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0900 - categorical_accuracy: 0.9705 - val_loss: 1.2704 - val_categorical_accuracy: 0.7696\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 26s 37ms/step - loss: 0.0858 - categorical_accuracy: 0.9720 - val_loss: 2.2242 - val_categorical_accuracy: 0.6944\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.0892 - categorical_accuracy: 0.9708 - val_loss: 1.2719 - val_categorical_accuracy: 0.7772\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0859 - categorical_accuracy: 0.9728 - val_loss: 1.8227 - val_categorical_accuracy: 0.7222\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1006 - categorical_accuracy: 0.9685 - val_loss: 1.3560 - val_categorical_accuracy: 0.7720\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0786 - categorical_accuracy: 0.9740 - val_loss: 1.1995 - val_categorical_accuracy: 0.8012\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0738 - categorical_accuracy: 0.9760 - val_loss: 1.3046 - val_categorical_accuracy: 0.7786\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0747 - categorical_accuracy: 0.9756 - val_loss: 1.4263 - val_categorical_accuracy: 0.7692\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0816 - categorical_accuracy: 0.9730 - val_loss: 1.1663 - val_categorical_accuracy: 0.7932\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0715 - categorical_accuracy: 0.9767 - val_loss: 1.3519 - val_categorical_accuracy: 0.7758\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0714 - categorical_accuracy: 0.9767 - val_loss: 1.3339 - val_categorical_accuracy: 0.7938\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0675 - categorical_accuracy: 0.9782 - val_loss: 2.1252 - val_categorical_accuracy: 0.7318\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0715 - categorical_accuracy: 0.9775 - val_loss: 1.3993 - val_categorical_accuracy: 0.7814\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0659 - categorical_accuracy: 0.9781 - val_loss: 1.4122 - val_categorical_accuracy: 0.7854\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0679 - categorical_accuracy: 0.9782 - val_loss: 1.3720 - val_categorical_accuracy: 0.7836\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0603 - categorical_accuracy: 0.9811 - val_loss: 1.5833 - val_categorical_accuracy: 0.7704\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.0665 - categorical_accuracy: 0.9787 - val_loss: 1.2840 - val_categorical_accuracy: 0.7914\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 20s 29ms/step - loss: 0.0608 - categorical_accuracy: 0.9809 - val_loss: 1.6883 - val_categorical_accuracy: 0.7742\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0573 - categorical_accuracy: 0.9823 - val_loss: 1.4406 - val_categorical_accuracy: 0.7956\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0602 - categorical_accuracy: 0.9813 - val_loss: 1.6369 - val_categorical_accuracy: 0.7724\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0564 - categorical_accuracy: 0.9828 - val_loss: 1.4514 - val_categorical_accuracy: 0.7848\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0613 - categorical_accuracy: 0.9800 - val_loss: 1.9562 - val_categorical_accuracy: 0.7366\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0514 - categorical_accuracy: 0.9831 - val_loss: 1.6908 - val_categorical_accuracy: 0.7868\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0552 - categorical_accuracy: 0.9824 - val_loss: 1.7831 - val_categorical_accuracy: 0.7624\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0481 - categorical_accuracy: 0.9849 - val_loss: 1.5007 - val_categorical_accuracy: 0.7880\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0575 - categorical_accuracy: 0.9817 - val_loss: 1.8789 - val_categorical_accuracy: 0.7602\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0515 - categorical_accuracy: 0.9837 - val_loss: 1.4409 - val_categorical_accuracy: 0.7902\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0534 - categorical_accuracy: 0.9827 - val_loss: 1.3440 - val_categorical_accuracy: 0.7946\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0498 - categorical_accuracy: 0.9845 - val_loss: 1.4153 - val_categorical_accuracy: 0.7968\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0531 - categorical_accuracy: 0.9835 - val_loss: 1.6453 - val_categorical_accuracy: 0.7730\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0435 - categorical_accuracy: 0.9858 - val_loss: 1.4463 - val_categorical_accuracy: 0.7956\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0468 - categorical_accuracy: 0.9854 - val_loss: 1.6273 - val_categorical_accuracy: 0.7886\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0444 - categorical_accuracy: 0.9859 - val_loss: 1.5875 - val_categorical_accuracy: 0.7984\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.0506 - categorical_accuracy: 0.9838 - val_loss: 1.6106 - val_categorical_accuracy: 0.7824\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0476 - categorical_accuracy: 0.9845 - val_loss: 1.4350 - val_categorical_accuracy: 0.8006\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0440 - categorical_accuracy: 0.9863 - val_loss: 1.5795 - val_categorical_accuracy: 0.7924\n",
      "--- Running training session 12\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 31s 45ms/step - loss: 1.7116 - categorical_accuracy: 0.3682 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.4126 - categorical_accuracy: 0.4918 - val_loss: 1.3437 - val_categorical_accuracy: 0.5232\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.2507 - categorical_accuracy: 0.5604 - val_loss: 1.2281 - val_categorical_accuracy: 0.5714\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 26s 38ms/step - loss: 1.1369 - categorical_accuracy: 0.6005 - val_loss: 1.1472 - val_categorical_accuracy: 0.5896\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 1.0474 - categorical_accuracy: 0.6362 - val_loss: 1.6993 - val_categorical_accuracy: 0.4968\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.9652 - categorical_accuracy: 0.6666 - val_loss: 1.1939 - val_categorical_accuracy: 0.6144\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 22s 31ms/step - loss: 0.9014 - categorical_accuracy: 0.6914 - val_loss: 1.3200 - val_categorical_accuracy: 0.5938\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.8451 - categorical_accuracy: 0.7098 - val_loss: 1.2352 - val_categorical_accuracy: 0.5968\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 22s 31ms/step - loss: 0.8025 - categorical_accuracy: 0.7239 - val_loss: 1.4873 - val_categorical_accuracy: 0.5672\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.7500 - categorical_accuracy: 0.7451 - val_loss: 1.0635 - val_categorical_accuracy: 0.6676\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6800 - categorical_accuracy: 0.7686 - val_loss: 1.1376 - val_categorical_accuracy: 0.6330\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 22s 32ms/step - loss: 0.6486 - categorical_accuracy: 0.7817 - val_loss: 1.4249 - val_categorical_accuracy: 0.5764\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.5985 - categorical_accuracy: 0.7952 - val_loss: 0.9506 - val_categorical_accuracy: 0.7084\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5661 - categorical_accuracy: 0.8079 - val_loss: 1.0295 - val_categorical_accuracy: 0.7150\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 22s 32ms/step - loss: 0.5213 - categorical_accuracy: 0.8219 - val_loss: 1.0507 - val_categorical_accuracy: 0.6906\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4892 - categorical_accuracy: 0.8323 - val_loss: 1.0346 - val_categorical_accuracy: 0.7226\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4538 - categorical_accuracy: 0.8443 - val_loss: 1.0579 - val_categorical_accuracy: 0.7312\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.4246 - categorical_accuracy: 0.8557 - val_loss: 2.0132 - val_categorical_accuracy: 0.5896\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.3992 - categorical_accuracy: 0.8635 - val_loss: 1.5083 - val_categorical_accuracy: 0.6528\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.3774 - categorical_accuracy: 0.8708 - val_loss: 1.8697 - val_categorical_accuracy: 0.5522\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.3620 - categorical_accuracy: 0.8768 - val_loss: 1.3765 - val_categorical_accuracy: 0.6716\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.3252 - categorical_accuracy: 0.8908 - val_loss: 1.1994 - val_categorical_accuracy: 0.7148\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.3079 - categorical_accuracy: 0.8957 - val_loss: 1.3021 - val_categorical_accuracy: 0.7278\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.3019 - categorical_accuracy: 0.8985 - val_loss: 1.1010 - val_categorical_accuracy: 0.7580\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.2835 - categorical_accuracy: 0.9044 - val_loss: 1.2763 - val_categorical_accuracy: 0.7422\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.2677 - categorical_accuracy: 0.9099 - val_loss: 1.1422 - val_categorical_accuracy: 0.7766\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.2425 - categorical_accuracy: 0.9177 - val_loss: 1.0741 - val_categorical_accuracy: 0.7750\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.2374 - categorical_accuracy: 0.9214 - val_loss: 1.3082 - val_categorical_accuracy: 0.7344\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.2367 - categorical_accuracy: 0.9214 - val_loss: 1.1029 - val_categorical_accuracy: 0.7756\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.2204 - categorical_accuracy: 0.9264 - val_loss: 1.6888 - val_categorical_accuracy: 0.7352\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.2080 - categorical_accuracy: 0.9315 - val_loss: 1.3594 - val_categorical_accuracy: 0.7460\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.2011 - categorical_accuracy: 0.9343 - val_loss: 1.2322 - val_categorical_accuracy: 0.7562\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1926 - categorical_accuracy: 0.9365 - val_loss: 1.3595 - val_categorical_accuracy: 0.7492\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1917 - categorical_accuracy: 0.9391 - val_loss: 1.6878 - val_categorical_accuracy: 0.7260\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1925 - categorical_accuracy: 0.9359 - val_loss: 1.4030 - val_categorical_accuracy: 0.7654\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1684 - categorical_accuracy: 0.9447 - val_loss: 1.8508 - val_categorical_accuracy: 0.7244\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1649 - categorical_accuracy: 0.9469 - val_loss: 1.4724 - val_categorical_accuracy: 0.7608\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1610 - categorical_accuracy: 0.9478 - val_loss: 1.3395 - val_categorical_accuracy: 0.7806\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1569 - categorical_accuracy: 0.9502 - val_loss: 1.3796 - val_categorical_accuracy: 0.7768\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1602 - categorical_accuracy: 0.9488 - val_loss: 1.5149 - val_categorical_accuracy: 0.7822\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1531 - categorical_accuracy: 0.9518 - val_loss: 1.5073 - val_categorical_accuracy: 0.7308\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1409 - categorical_accuracy: 0.9554 - val_loss: 1.6301 - val_categorical_accuracy: 0.7654\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1401 - categorical_accuracy: 0.9556 - val_loss: 1.5769 - val_categorical_accuracy: 0.7550\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.1372 - categorical_accuracy: 0.9576 - val_loss: 1.7574 - val_categorical_accuracy: 0.7604\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1266 - categorical_accuracy: 0.9586 - val_loss: 1.4767 - val_categorical_accuracy: 0.7798\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1290 - categorical_accuracy: 0.9593 - val_loss: 1.6650 - val_categorical_accuracy: 0.7806\n",
      "Epoch 47/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1259 - categorical_accuracy: 0.9600 - val_loss: 1.4795 - val_categorical_accuracy: 0.7616\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.1273 - categorical_accuracy: 0.9602 - val_loss: 1.7071 - val_categorical_accuracy: 0.7630\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 20s 29ms/step - loss: 0.1143 - categorical_accuracy: 0.9639 - val_loss: 1.5670 - val_categorical_accuracy: 0.7880\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.1295 - categorical_accuracy: 0.9601 - val_loss: 1.5932 - val_categorical_accuracy: 0.7742\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.1155 - categorical_accuracy: 0.9645 - val_loss: 1.5153 - val_categorical_accuracy: 0.7830\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.1106 - categorical_accuracy: 0.9667 - val_loss: 1.5194 - val_categorical_accuracy: 0.7780\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 24s 34ms/step - loss: 0.1139 - categorical_accuracy: 0.9638 - val_loss: 1.4154 - val_categorical_accuracy: 0.7898\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.1078 - categorical_accuracy: 0.9676 - val_loss: 1.4592 - val_categorical_accuracy: 0.7814\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1137 - categorical_accuracy: 0.9663 - val_loss: 2.3386 - val_categorical_accuracy: 0.7334\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 24s 35ms/step - loss: 0.1057 - categorical_accuracy: 0.9678 - val_loss: 1.6232 - val_categorical_accuracy: 0.7864\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 24s 34ms/step - loss: 0.1078 - categorical_accuracy: 0.9671 - val_loss: 1.7131 - val_categorical_accuracy: 0.7800\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1029 - categorical_accuracy: 0.9689 - val_loss: 1.5486 - val_categorical_accuracy: 0.7850\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 25s 35ms/step - loss: 0.1043 - categorical_accuracy: 0.9690 - val_loss: 2.1581 - val_categorical_accuracy: 0.7600\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1024 - categorical_accuracy: 0.9688 - val_loss: 1.8373 - val_categorical_accuracy: 0.7780\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 26s 37ms/step - loss: 0.0939 - categorical_accuracy: 0.9718 - val_loss: 2.1310 - val_categorical_accuracy: 0.7444\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.0964 - categorical_accuracy: 0.9708 - val_loss: 1.4503 - val_categorical_accuracy: 0.7796\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1017 - categorical_accuracy: 0.9693 - val_loss: 1.5740 - val_categorical_accuracy: 0.7910\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0924 - categorical_accuracy: 0.9733 - val_loss: 1.7338 - val_categorical_accuracy: 0.7758\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0988 - categorical_accuracy: 0.9701 - val_loss: 1.6309 - val_categorical_accuracy: 0.7878\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.0883 - categorical_accuracy: 0.9747 - val_loss: 2.7875 - val_categorical_accuracy: 0.7372\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 22s 31ms/step - loss: 0.0979 - categorical_accuracy: 0.9708 - val_loss: 1.6445 - val_categorical_accuracy: 0.7862\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0888 - categorical_accuracy: 0.9729 - val_loss: 1.6148 - val_categorical_accuracy: 0.7832\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0996 - categorical_accuracy: 0.9709 - val_loss: 1.7831 - val_categorical_accuracy: 0.7698\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0771 - categorical_accuracy: 0.9770 - val_loss: 1.9530 - val_categorical_accuracy: 0.7886\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.0890 - categorical_accuracy: 0.9730 - val_loss: 1.8878 - val_categorical_accuracy: 0.7792\n",
      "Epoch 72/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0960 - categorical_accuracy: 0.9722 - val_loss: 1.8092 - val_categorical_accuracy: 0.7744\n",
      "Epoch 73/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0931 - categorical_accuracy: 0.9740 - val_loss: 1.5983 - val_categorical_accuracy: 0.8004\n",
      "Epoch 74/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0738 - categorical_accuracy: 0.9776 - val_loss: 1.8108 - val_categorical_accuracy: 0.7960\n",
      "Epoch 75/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0816 - categorical_accuracy: 0.9756 - val_loss: 1.7494 - val_categorical_accuracy: 0.7996\n",
      "Epoch 76/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0859 - categorical_accuracy: 0.9752 - val_loss: 1.8421 - val_categorical_accuracy: 0.7978\n",
      "Epoch 77/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0799 - categorical_accuracy: 0.9770 - val_loss: 2.0705 - val_categorical_accuracy: 0.7930\n",
      "Epoch 78/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0973 - categorical_accuracy: 0.9725 - val_loss: 2.1082 - val_categorical_accuracy: 0.7950\n",
      "Epoch 79/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0727 - categorical_accuracy: 0.9786 - val_loss: 1.8862 - val_categorical_accuracy: 0.7942\n",
      "Epoch 80/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0759 - categorical_accuracy: 0.9779 - val_loss: 2.2186 - val_categorical_accuracy: 0.7806\n",
      "Epoch 81/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0840 - categorical_accuracy: 0.9761 - val_loss: 1.8374 - val_categorical_accuracy: 0.7884\n",
      "Epoch 82/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0809 - categorical_accuracy: 0.9768 - val_loss: 1.7988 - val_categorical_accuracy: 0.7832\n",
      "Epoch 83/100\n",
      "704/704 [==============================] - 17s 23ms/step - loss: 0.0706 - categorical_accuracy: 0.9791 - val_loss: 1.8282 - val_categorical_accuracy: 0.7932\n",
      "Epoch 84/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0827 - categorical_accuracy: 0.9762 - val_loss: 2.0641 - val_categorical_accuracy: 0.7934\n",
      "Epoch 85/100\n",
      "704/704 [==============================] - 22s 32ms/step - loss: 0.0689 - categorical_accuracy: 0.9803 - val_loss: 2.0699 - val_categorical_accuracy: 0.7912\n",
      "Epoch 86/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0804 - categorical_accuracy: 0.9777 - val_loss: 1.8741 - val_categorical_accuracy: 0.7898\n",
      "Epoch 87/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.0827 - categorical_accuracy: 0.9776 - val_loss: 2.2175 - val_categorical_accuracy: 0.7822\n",
      "Epoch 88/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0672 - categorical_accuracy: 0.9806 - val_loss: 2.2574 - val_categorical_accuracy: 0.7964\n",
      "Epoch 89/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.0728 - categorical_accuracy: 0.9802 - val_loss: 2.2946 - val_categorical_accuracy: 0.7802\n",
      "Epoch 90/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0763 - categorical_accuracy: 0.9781 - val_loss: 2.6880 - val_categorical_accuracy: 0.7512\n",
      "Epoch 91/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0738 - categorical_accuracy: 0.9790 - val_loss: 2.1004 - val_categorical_accuracy: 0.7778\n",
      "Epoch 92/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0651 - categorical_accuracy: 0.9806 - val_loss: 2.4472 - val_categorical_accuracy: 0.7604\n",
      "Epoch 93/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0698 - categorical_accuracy: 0.9802 - val_loss: 1.8581 - val_categorical_accuracy: 0.7966\n",
      "Epoch 94/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0770 - categorical_accuracy: 0.9781 - val_loss: 1.5698 - val_categorical_accuracy: 0.7842\n",
      "Epoch 95/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.0630 - categorical_accuracy: 0.9810 - val_loss: 2.1364 - val_categorical_accuracy: 0.7880\n",
      "Epoch 96/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.0713 - categorical_accuracy: 0.9796 - val_loss: 2.3016 - val_categorical_accuracy: 0.7580\n",
      "Epoch 97/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0737 - categorical_accuracy: 0.9791 - val_loss: 2.1373 - val_categorical_accuracy: 0.7754\n",
      "Epoch 98/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0651 - categorical_accuracy: 0.9823 - val_loss: 1.8690 - val_categorical_accuracy: 0.7932\n",
      "Epoch 99/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 0.0601 - categorical_accuracy: 0.9825 - val_loss: 2.1465 - val_categorical_accuracy: 0.7928\n",
      "Epoch 100/100\n",
      "704/704 [==============================] - 22s 31ms/step - loss: 0.0662 - categorical_accuracy: 0.9815 - val_loss: 1.8043 - val_categorical_accuracy: 0.7840\n",
      "--- Running training session 13\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 26s 38ms/step - loss: 1.5610 - categorical_accuracy: 0.4480 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.1143 - categorical_accuracy: 0.6142 - val_loss: 1.4928 - val_categorical_accuracy: 0.5172\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.9301 - categorical_accuracy: 0.6814 - val_loss: 1.1470 - val_categorical_accuracy: 0.6074\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.8158 - categorical_accuracy: 0.7223 - val_loss: 1.1204 - val_categorical_accuracy: 0.6222\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 24s 34ms/step - loss: 0.7178 - categorical_accuracy: 0.7567 - val_loss: 1.3403 - val_categorical_accuracy: 0.5722\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 26s 37ms/step - loss: 0.6362 - categorical_accuracy: 0.7842 - val_loss: 1.2021 - val_categorical_accuracy: 0.6146\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 23s 33ms/step - loss: 0.5755 - categorical_accuracy: 0.8045 - val_loss: 1.7920 - val_categorical_accuracy: 0.4902\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 20s 29ms/step - loss: 0.5089 - categorical_accuracy: 0.8273 - val_loss: 1.0995 - val_categorical_accuracy: 0.6598\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.4515 - categorical_accuracy: 0.8454 - val_loss: 1.4654 - val_categorical_accuracy: 0.6074\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4096 - categorical_accuracy: 0.8603 - val_loss: 1.0263 - val_categorical_accuracy: 0.6964\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.3539 - categorical_accuracy: 0.8791 - val_loss: 1.1895 - val_categorical_accuracy: 0.6740\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 17s 23ms/step - loss: 0.3108 - categorical_accuracy: 0.8938 - val_loss: 1.2569 - val_categorical_accuracy: 0.6442\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.2793 - categorical_accuracy: 0.9033 - val_loss: 0.9859 - val_categorical_accuracy: 0.7284\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 25s 35ms/step - loss: 0.2571 - categorical_accuracy: 0.9112 - val_loss: 1.0603 - val_categorical_accuracy: 0.7274\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.2193 - categorical_accuracy: 0.9247 - val_loss: 0.8657 - val_categorical_accuracy: 0.7702\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.2002 - categorical_accuracy: 0.9304 - val_loss: 0.9545 - val_categorical_accuracy: 0.7618\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 25s 35ms/step - loss: 0.1727 - categorical_accuracy: 0.9402 - val_loss: 1.0357 - val_categorical_accuracy: 0.7484\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.1606 - categorical_accuracy: 0.9449 - val_loss: 0.8891 - val_categorical_accuracy: 0.7854\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.1457 - categorical_accuracy: 0.9491 - val_loss: 0.8899 - val_categorical_accuracy: 0.7952\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1269 - categorical_accuracy: 0.9568 - val_loss: 0.9386 - val_categorical_accuracy: 0.7856\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1244 - categorical_accuracy: 0.9563 - val_loss: 1.1136 - val_categorical_accuracy: 0.7650\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.1133 - categorical_accuracy: 0.9629 - val_loss: 0.9869 - val_categorical_accuracy: 0.7840\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1035 - categorical_accuracy: 0.9652 - val_loss: 0.9163 - val_categorical_accuracy: 0.7996\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1007 - categorical_accuracy: 0.9658 - val_loss: 0.9574 - val_categorical_accuracy: 0.7906\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1001 - categorical_accuracy: 0.9663 - val_loss: 0.9429 - val_categorical_accuracy: 0.8012\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0836 - categorical_accuracy: 0.9723 - val_loss: 1.0149 - val_categorical_accuracy: 0.7778\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0855 - categorical_accuracy: 0.9718 - val_loss: 1.3367 - val_categorical_accuracy: 0.7574\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0791 - categorical_accuracy: 0.9734 - val_loss: 0.9676 - val_categorical_accuracy: 0.8030\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0743 - categorical_accuracy: 0.9750 - val_loss: 1.0450 - val_categorical_accuracy: 0.7988\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0776 - categorical_accuracy: 0.9738 - val_loss: 1.9334 - val_categorical_accuracy: 0.6930\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0715 - categorical_accuracy: 0.9758 - val_loss: 0.9995 - val_categorical_accuracy: 0.8058\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0687 - categorical_accuracy: 0.9773 - val_loss: 1.1343 - val_categorical_accuracy: 0.7972\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0618 - categorical_accuracy: 0.9793 - val_loss: 1.2146 - val_categorical_accuracy: 0.7856\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.0661 - categorical_accuracy: 0.9776 - val_loss: 1.2285 - val_categorical_accuracy: 0.7738\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.0691 - categorical_accuracy: 0.9771 - val_loss: 1.2068 - val_categorical_accuracy: 0.7872\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0506 - categorical_accuracy: 0.9829 - val_loss: 1.1122 - val_categorical_accuracy: 0.8002\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0554 - categorical_accuracy: 0.9814 - val_loss: 1.0956 - val_categorical_accuracy: 0.8024\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0582 - categorical_accuracy: 0.9808 - val_loss: 1.1059 - val_categorical_accuracy: 0.8026\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0494 - categorical_accuracy: 0.9836 - val_loss: 1.0428 - val_categorical_accuracy: 0.8174\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0465 - categorical_accuracy: 0.9844 - val_loss: 1.2271 - val_categorical_accuracy: 0.7914\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 17s 23ms/step - loss: 0.0571 - categorical_accuracy: 0.9813 - val_loss: 1.3538 - val_categorical_accuracy: 0.7846\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0509 - categorical_accuracy: 0.9832 - val_loss: 1.2775 - val_categorical_accuracy: 0.7950\n",
      "Epoch 43/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0425 - categorical_accuracy: 0.9861 - val_loss: 1.0224 - val_categorical_accuracy: 0.8256\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0465 - categorical_accuracy: 0.9848 - val_loss: 1.5167 - val_categorical_accuracy: 0.7670\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.0476 - categorical_accuracy: 0.9843 - val_loss: 1.1325 - val_categorical_accuracy: 0.8054\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 22s 31ms/step - loss: 0.0406 - categorical_accuracy: 0.9864 - val_loss: 1.1762 - val_categorical_accuracy: 0.8020\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0440 - categorical_accuracy: 0.9858 - val_loss: 1.1776 - val_categorical_accuracy: 0.8082\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0393 - categorical_accuracy: 0.9873 - val_loss: 1.2360 - val_categorical_accuracy: 0.8004\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 20s 29ms/step - loss: 0.0446 - categorical_accuracy: 0.9854 - val_loss: 1.3665 - val_categorical_accuracy: 0.7878\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 20s 29ms/step - loss: 0.0429 - categorical_accuracy: 0.9862 - val_loss: 1.2736 - val_categorical_accuracy: 0.7946\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 28s 40ms/step - loss: 0.0391 - categorical_accuracy: 0.9871 - val_loss: 1.3861 - val_categorical_accuracy: 0.7926\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0406 - categorical_accuracy: 0.9864 - val_loss: 1.2735 - val_categorical_accuracy: 0.7980\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0372 - categorical_accuracy: 0.9880 - val_loss: 1.2382 - val_categorical_accuracy: 0.8030\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0371 - categorical_accuracy: 0.9883 - val_loss: 1.2708 - val_categorical_accuracy: 0.7888\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0412 - categorical_accuracy: 0.9866 - val_loss: 1.1802 - val_categorical_accuracy: 0.8074\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.0326 - categorical_accuracy: 0.9890 - val_loss: 1.3499 - val_categorical_accuracy: 0.7854\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0358 - categorical_accuracy: 0.9886 - val_loss: 1.3988 - val_categorical_accuracy: 0.7946\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0341 - categorical_accuracy: 0.9890 - val_loss: 1.3222 - val_categorical_accuracy: 0.7940\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0354 - categorical_accuracy: 0.9887 - val_loss: 1.7912 - val_categorical_accuracy: 0.7470\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0346 - categorical_accuracy: 0.9880 - val_loss: 1.3657 - val_categorical_accuracy: 0.7912\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0376 - categorical_accuracy: 0.9876 - val_loss: 1.2856 - val_categorical_accuracy: 0.8040\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0269 - categorical_accuracy: 0.9908 - val_loss: 1.3160 - val_categorical_accuracy: 0.7970\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0274 - categorical_accuracy: 0.9907 - val_loss: 1.3406 - val_categorical_accuracy: 0.8056\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0411 - categorical_accuracy: 0.9863 - val_loss: 1.2478 - val_categorical_accuracy: 0.8138\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0339 - categorical_accuracy: 0.9889 - val_loss: 1.3806 - val_categorical_accuracy: 0.8012\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0272 - categorical_accuracy: 0.9913 - val_loss: 1.2328 - val_categorical_accuracy: 0.8142\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0285 - categorical_accuracy: 0.9903 - val_loss: 1.4078 - val_categorical_accuracy: 0.7952\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0275 - categorical_accuracy: 0.9909 - val_loss: 1.2642 - val_categorical_accuracy: 0.8228\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0323 - categorical_accuracy: 0.9894 - val_loss: 1.3301 - val_categorical_accuracy: 0.8088\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0268 - categorical_accuracy: 0.9914 - val_loss: 1.3267 - val_categorical_accuracy: 0.8150\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0277 - categorical_accuracy: 0.9911 - val_loss: 1.2696 - val_categorical_accuracy: 0.8142\n",
      "Epoch 72/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0257 - categorical_accuracy: 0.9916 - val_loss: 1.5198 - val_categorical_accuracy: 0.7766\n",
      "Epoch 73/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0296 - categorical_accuracy: 0.9904 - val_loss: 1.5610 - val_categorical_accuracy: 0.7878\n",
      "--- Running training session 14\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 1.5856 - categorical_accuracy: 0.4192 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 1.2309 - categorical_accuracy: 0.5623 - val_loss: 1.6006 - val_categorical_accuracy: 0.4506\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 1.0551 - categorical_accuracy: 0.6306 - val_loss: 1.6203 - val_categorical_accuracy: 0.4654\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.9280 - categorical_accuracy: 0.6781 - val_loss: 1.1210 - val_categorical_accuracy: 0.6238\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.8275 - categorical_accuracy: 0.7125 - val_loss: 1.1697 - val_categorical_accuracy: 0.5990\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.7429 - categorical_accuracy: 0.7455 - val_loss: 0.8747 - val_categorical_accuracy: 0.7042\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.6731 - categorical_accuracy: 0.7717 - val_loss: 1.3552 - val_categorical_accuracy: 0.5814\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.6075 - categorical_accuracy: 0.7928 - val_loss: 1.3637 - val_categorical_accuracy: 0.6154\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.5478 - categorical_accuracy: 0.8121 - val_loss: 1.1687 - val_categorical_accuracy: 0.6696\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.4892 - categorical_accuracy: 0.8313 - val_loss: 0.9658 - val_categorical_accuracy: 0.7270\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.4359 - categorical_accuracy: 0.8490 - val_loss: 1.6921 - val_categorical_accuracy: 0.5964\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.3942 - categorical_accuracy: 0.8644 - val_loss: 1.0438 - val_categorical_accuracy: 0.7260\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.3698 - categorical_accuracy: 0.8725 - val_loss: 0.9908 - val_categorical_accuracy: 0.7438\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.3228 - categorical_accuracy: 0.8870 - val_loss: 0.9709 - val_categorical_accuracy: 0.7452\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.2924 - categorical_accuracy: 0.8974 - val_loss: 1.0402 - val_categorical_accuracy: 0.7462\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.2710 - categorical_accuracy: 0.9060 - val_loss: 1.2931 - val_categorical_accuracy: 0.7026\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.2512 - categorical_accuracy: 0.9127 - val_loss: 1.3416 - val_categorical_accuracy: 0.7172\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.2202 - categorical_accuracy: 0.9239 - val_loss: 1.1117 - val_categorical_accuracy: 0.7358\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.2102 - categorical_accuracy: 0.9272 - val_loss: 0.9343 - val_categorical_accuracy: 0.7830\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.1896 - categorical_accuracy: 0.9358 - val_loss: 1.2783 - val_categorical_accuracy: 0.7308\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.1683 - categorical_accuracy: 0.9418 - val_loss: 1.4805 - val_categorical_accuracy: 0.7138\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 22s 31ms/step - loss: 0.1668 - categorical_accuracy: 0.9442 - val_loss: 1.4077 - val_categorical_accuracy: 0.7036\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.1532 - categorical_accuracy: 0.9485 - val_loss: 1.1093 - val_categorical_accuracy: 0.7650\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.1474 - categorical_accuracy: 0.9503 - val_loss: 1.3192 - val_categorical_accuracy: 0.7476\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.1372 - categorical_accuracy: 0.9535 - val_loss: 1.2026 - val_categorical_accuracy: 0.7640\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.1326 - categorical_accuracy: 0.9564 - val_loss: 1.2515 - val_categorical_accuracy: 0.7752\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.1235 - categorical_accuracy: 0.9585 - val_loss: 1.0967 - val_categorical_accuracy: 0.7896\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 23s 32ms/step - loss: 0.1253 - categorical_accuracy: 0.9569 - val_loss: 1.3001 - val_categorical_accuracy: 0.7780\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.1200 - categorical_accuracy: 0.9605 - val_loss: 1.2123 - val_categorical_accuracy: 0.7774\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.1085 - categorical_accuracy: 0.9639 - val_loss: 1.3964 - val_categorical_accuracy: 0.7632\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1032 - categorical_accuracy: 0.9662 - val_loss: 1.2224 - val_categorical_accuracy: 0.7964\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 24s 33ms/step - loss: 0.1047 - categorical_accuracy: 0.9642 - val_loss: 1.2957 - val_categorical_accuracy: 0.7762\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0969 - categorical_accuracy: 0.9685 - val_loss: 1.4263 - val_categorical_accuracy: 0.7422\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0938 - categorical_accuracy: 0.9698 - val_loss: 1.3333 - val_categorical_accuracy: 0.7788\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0891 - categorical_accuracy: 0.9703 - val_loss: 1.2793 - val_categorical_accuracy: 0.7848\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0863 - categorical_accuracy: 0.9717 - val_loss: 1.2535 - val_categorical_accuracy: 0.7954\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0874 - categorical_accuracy: 0.9718 - val_loss: 1.3553 - val_categorical_accuracy: 0.7782\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0840 - categorical_accuracy: 0.9725 - val_loss: 1.3978 - val_categorical_accuracy: 0.7922\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0777 - categorical_accuracy: 0.9743 - val_loss: 1.5806 - val_categorical_accuracy: 0.7576\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0766 - categorical_accuracy: 0.9753 - val_loss: 1.4125 - val_categorical_accuracy: 0.7746\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0739 - categorical_accuracy: 0.9766 - val_loss: 1.6971 - val_categorical_accuracy: 0.7614\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0768 - categorical_accuracy: 0.9751 - val_loss: 1.4168 - val_categorical_accuracy: 0.7828\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0703 - categorical_accuracy: 0.9767 - val_loss: 1.3810 - val_categorical_accuracy: 0.7926\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0724 - categorical_accuracy: 0.9762 - val_loss: 1.2960 - val_categorical_accuracy: 0.7906\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0707 - categorical_accuracy: 0.9775 - val_loss: 1.6882 - val_categorical_accuracy: 0.7644\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0713 - categorical_accuracy: 0.9771 - val_loss: 1.9976 - val_categorical_accuracy: 0.7360\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0685 - categorical_accuracy: 0.9775 - val_loss: 1.4257 - val_categorical_accuracy: 0.7682\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0629 - categorical_accuracy: 0.9793 - val_loss: 1.4671 - val_categorical_accuracy: 0.7764\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0694 - categorical_accuracy: 0.9783 - val_loss: 1.3074 - val_categorical_accuracy: 0.7914\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0588 - categorical_accuracy: 0.9820 - val_loss: 1.2705 - val_categorical_accuracy: 0.7998\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 17s 23ms/step - loss: 0.0628 - categorical_accuracy: 0.9796 - val_loss: 1.5038 - val_categorical_accuracy: 0.7824\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0659 - categorical_accuracy: 0.9790 - val_loss: 1.6695 - val_categorical_accuracy: 0.7630\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0563 - categorical_accuracy: 0.9810 - val_loss: 1.2571 - val_categorical_accuracy: 0.8194\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0583 - categorical_accuracy: 0.9818 - val_loss: 1.6122 - val_categorical_accuracy: 0.7958\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0523 - categorical_accuracy: 0.9832 - val_loss: 1.5335 - val_categorical_accuracy: 0.7876\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0505 - categorical_accuracy: 0.9834 - val_loss: 1.7996 - val_categorical_accuracy: 0.7688\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0533 - categorical_accuracy: 0.9834 - val_loss: 1.5988 - val_categorical_accuracy: 0.7796\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0520 - categorical_accuracy: 0.9840 - val_loss: 1.6921 - val_categorical_accuracy: 0.7778\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0524 - categorical_accuracy: 0.9836 - val_loss: 1.3492 - val_categorical_accuracy: 0.8082\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0570 - categorical_accuracy: 0.9824 - val_loss: 1.5229 - val_categorical_accuracy: 0.7808\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0485 - categorical_accuracy: 0.9846 - val_loss: 1.8165 - val_categorical_accuracy: 0.7742\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0486 - categorical_accuracy: 0.9839 - val_loss: 1.4994 - val_categorical_accuracy: 0.7816\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.0470 - categorical_accuracy: 0.9846 - val_loss: 1.4689 - val_categorical_accuracy: 0.7902\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.0458 - categorical_accuracy: 0.9858 - val_loss: 1.5755 - val_categorical_accuracy: 0.7882\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.0508 - categorical_accuracy: 0.9834 - val_loss: 1.3944 - val_categorical_accuracy: 0.8124\n",
      "Epoch 66/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0435 - categorical_accuracy: 0.9854 - val_loss: 1.5011 - val_categorical_accuracy: 0.8034\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0482 - categorical_accuracy: 0.9848 - val_loss: 1.5432 - val_categorical_accuracy: 0.7944\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0415 - categorical_accuracy: 0.9875 - val_loss: 1.4410 - val_categorical_accuracy: 0.8130\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0504 - categorical_accuracy: 0.9846 - val_loss: 1.5901 - val_categorical_accuracy: 0.7910\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 23s 32ms/step - loss: 0.0433 - categorical_accuracy: 0.9867 - val_loss: 1.4710 - val_categorical_accuracy: 0.8038\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.0455 - categorical_accuracy: 0.9860 - val_loss: 1.6648 - val_categorical_accuracy: 0.7918\n",
      "Epoch 72/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.0425 - categorical_accuracy: 0.9863 - val_loss: 1.4004 - val_categorical_accuracy: 0.8004\n",
      "Epoch 73/100\n",
      "704/704 [==============================] - 23s 33ms/step - loss: 0.0463 - categorical_accuracy: 0.9859 - val_loss: 1.4976 - val_categorical_accuracy: 0.8096\n",
      "Epoch 74/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0398 - categorical_accuracy: 0.9873 - val_loss: 1.4907 - val_categorical_accuracy: 0.8076\n",
      "Epoch 75/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0402 - categorical_accuracy: 0.9871 - val_loss: 1.4619 - val_categorical_accuracy: 0.8034\n",
      "Epoch 76/100\n",
      "704/704 [==============================] - 26s 37ms/step - loss: 0.0388 - categorical_accuracy: 0.9872 - val_loss: 1.6334 - val_categorical_accuracy: 0.7918\n",
      "Epoch 77/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0406 - categorical_accuracy: 0.9872 - val_loss: 1.6385 - val_categorical_accuracy: 0.7942\n",
      "Epoch 78/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.0433 - categorical_accuracy: 0.9868 - val_loss: 1.5660 - val_categorical_accuracy: 0.8048\n",
      "Epoch 79/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.0355 - categorical_accuracy: 0.9891 - val_loss: 1.5966 - val_categorical_accuracy: 0.8056\n",
      "Epoch 80/100\n",
      "704/704 [==============================] - 26s 37ms/step - loss: 0.0429 - categorical_accuracy: 0.9865 - val_loss: 1.6540 - val_categorical_accuracy: 0.8028\n",
      "Epoch 81/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.0365 - categorical_accuracy: 0.9886 - val_loss: 1.6852 - val_categorical_accuracy: 0.7988\n",
      "Epoch 82/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0406 - categorical_accuracy: 0.9871 - val_loss: 1.5661 - val_categorical_accuracy: 0.8046\n",
      "Epoch 83/100\n",
      "704/704 [==============================] - 22s 31ms/step - loss: 0.0344 - categorical_accuracy: 0.9890 - val_loss: 1.8196 - val_categorical_accuracy: 0.7930\n",
      "--- Running training session 15\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 22s 32ms/step - loss: 1.7358 - categorical_accuracy: 0.3566 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.4150 - categorical_accuracy: 0.4922 - val_loss: 1.8043 - val_categorical_accuracy: 0.4092\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.2396 - categorical_accuracy: 0.5614 - val_loss: 2.3061 - val_categorical_accuracy: 0.3818\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.1317 - categorical_accuracy: 0.6024 - val_loss: 1.1830 - val_categorical_accuracy: 0.5860\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.0326 - categorical_accuracy: 0.6416 - val_loss: 1.3125 - val_categorical_accuracy: 0.5660\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.9516 - categorical_accuracy: 0.6680 - val_loss: 1.1679 - val_categorical_accuracy: 0.6002\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8849 - categorical_accuracy: 0.6920 - val_loss: 1.2104 - val_categorical_accuracy: 0.6068\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.8275 - categorical_accuracy: 0.7156 - val_loss: 1.4392 - val_categorical_accuracy: 0.5844\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7755 - categorical_accuracy: 0.7350 - val_loss: 1.3779 - val_categorical_accuracy: 0.6242\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7181 - categorical_accuracy: 0.7555 - val_loss: 1.3369 - val_categorical_accuracy: 0.6158\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6724 - categorical_accuracy: 0.7676 - val_loss: 1.2825 - val_categorical_accuracy: 0.6542\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 17s 23ms/step - loss: 0.6209 - categorical_accuracy: 0.7893 - val_loss: 1.1236 - val_categorical_accuracy: 0.7032\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5849 - categorical_accuracy: 0.7998 - val_loss: 1.3573 - val_categorical_accuracy: 0.6610\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.5397 - categorical_accuracy: 0.8171 - val_loss: 1.0975 - val_categorical_accuracy: 0.7018\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.4963 - categorical_accuracy: 0.8303 - val_loss: 1.2020 - val_categorical_accuracy: 0.6942\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4789 - categorical_accuracy: 0.8358 - val_loss: 0.9105 - val_categorical_accuracy: 0.7454\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.4490 - categorical_accuracy: 0.8453 - val_loss: 1.2381 - val_categorical_accuracy: 0.6894\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.4106 - categorical_accuracy: 0.8582 - val_loss: 1.4048 - val_categorical_accuracy: 0.7064\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.3785 - categorical_accuracy: 0.8690 - val_loss: 1.6786 - val_categorical_accuracy: 0.6968\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.3553 - categorical_accuracy: 0.8777 - val_loss: 1.2244 - val_categorical_accuracy: 0.7266\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.3423 - categorical_accuracy: 0.8829 - val_loss: 1.4240 - val_categorical_accuracy: 0.6920\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.3043 - categorical_accuracy: 0.8975 - val_loss: 1.7720 - val_categorical_accuracy: 0.6862\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.2969 - categorical_accuracy: 0.8999 - val_loss: 1.9601 - val_categorical_accuracy: 0.6672\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.2842 - categorical_accuracy: 0.9037 - val_loss: 1.8357 - val_categorical_accuracy: 0.6966\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.2637 - categorical_accuracy: 0.9120 - val_loss: 1.5110 - val_categorical_accuracy: 0.7098\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.2615 - categorical_accuracy: 0.9133 - val_loss: 1.3909 - val_categorical_accuracy: 0.7330\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.2378 - categorical_accuracy: 0.9209 - val_loss: 1.3667 - val_categorical_accuracy: 0.7662\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.2277 - categorical_accuracy: 0.9243 - val_loss: 1.4445 - val_categorical_accuracy: 0.7176\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.2152 - categorical_accuracy: 0.9295 - val_loss: 1.1685 - val_categorical_accuracy: 0.7768\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.2123 - categorical_accuracy: 0.9299 - val_loss: 1.1771 - val_categorical_accuracy: 0.7720\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.1979 - categorical_accuracy: 0.9355 - val_loss: 1.1616 - val_categorical_accuracy: 0.7980\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.2047 - categorical_accuracy: 0.9337 - val_loss: 1.5701 - val_categorical_accuracy: 0.7342\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1865 - categorical_accuracy: 0.9382 - val_loss: 1.6799 - val_categorical_accuracy: 0.7478\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.1769 - categorical_accuracy: 0.9421 - val_loss: 1.6546 - val_categorical_accuracy: 0.7488\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.1755 - categorical_accuracy: 0.9443 - val_loss: 1.8653 - val_categorical_accuracy: 0.7170\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1599 - categorical_accuracy: 0.9480 - val_loss: 1.8411 - val_categorical_accuracy: 0.7548\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 25s 35ms/step - loss: 0.1656 - categorical_accuracy: 0.9480 - val_loss: 1.7302 - val_categorical_accuracy: 0.7498\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 20s 29ms/step - loss: 0.1659 - categorical_accuracy: 0.9477 - val_loss: 1.4334 - val_categorical_accuracy: 0.7790\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.1521 - categorical_accuracy: 0.9516 - val_loss: 1.9722 - val_categorical_accuracy: 0.7190\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 28s 40ms/step - loss: 0.1475 - categorical_accuracy: 0.9522 - val_loss: 1.8745 - val_categorical_accuracy: 0.7518\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1482 - categorical_accuracy: 0.9522 - val_loss: 1.3181 - val_categorical_accuracy: 0.7824\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 19s 28ms/step - loss: 0.1375 - categorical_accuracy: 0.9568 - val_loss: 1.6263 - val_categorical_accuracy: 0.7788\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.1439 - categorical_accuracy: 0.9554 - val_loss: 1.7620 - val_categorical_accuracy: 0.7660\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.1304 - categorical_accuracy: 0.9594 - val_loss: 1.4191 - val_categorical_accuracy: 0.7896\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.1293 - categorical_accuracy: 0.9600 - val_loss: 1.5171 - val_categorical_accuracy: 0.7692\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.1354 - categorical_accuracy: 0.9582 - val_loss: 1.6484 - val_categorical_accuracy: 0.7684\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 22s 31ms/step - loss: 0.1279 - categorical_accuracy: 0.9604 - val_loss: 1.8135 - val_categorical_accuracy: 0.7072\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1337 - categorical_accuracy: 0.9591 - val_loss: 1.4879 - val_categorical_accuracy: 0.7674\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 27s 39ms/step - loss: 0.1187 - categorical_accuracy: 0.9629 - val_loss: 1.6965 - val_categorical_accuracy: 0.7824\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1297 - categorical_accuracy: 0.9601 - val_loss: 1.5676 - val_categorical_accuracy: 0.7958\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1307 - categorical_accuracy: 0.9607 - val_loss: 1.4902 - val_categorical_accuracy: 0.7938\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1132 - categorical_accuracy: 0.9653 - val_loss: 1.4674 - val_categorical_accuracy: 0.7894\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0972 - categorical_accuracy: 0.9703 - val_loss: 1.9223 - val_categorical_accuracy: 0.7598\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.1181 - categorical_accuracy: 0.9640 - val_loss: 1.6021 - val_categorical_accuracy: 0.7692\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0999 - categorical_accuracy: 0.9688 - val_loss: 1.6910 - val_categorical_accuracy: 0.7722\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1168 - categorical_accuracy: 0.9651 - val_loss: 1.8720 - val_categorical_accuracy: 0.7766\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1005 - categorical_accuracy: 0.9690 - val_loss: 1.8975 - val_categorical_accuracy: 0.7608\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 17s 23ms/step - loss: 0.1053 - categorical_accuracy: 0.9682 - val_loss: 1.7627 - val_categorical_accuracy: 0.7908\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.1115 - categorical_accuracy: 0.9673 - val_loss: 2.1460 - val_categorical_accuracy: 0.7582\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1020 - categorical_accuracy: 0.9698 - val_loss: 1.7726 - val_categorical_accuracy: 0.7804\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0880 - categorical_accuracy: 0.9740 - val_loss: 1.7444 - val_categorical_accuracy: 0.7908\n",
      "--- Running training session 16\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 22s 31ms/step - loss: 1.5088 - categorical_accuracy: 0.4667 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 1.0672 - categorical_accuracy: 0.6313 - val_loss: 1.1867 - val_categorical_accuracy: 0.5808\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.8910 - categorical_accuracy: 0.6974 - val_loss: 1.2811 - val_categorical_accuracy: 0.5764\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.7730 - categorical_accuracy: 0.7399 - val_loss: 0.9878 - val_categorical_accuracy: 0.6718\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6773 - categorical_accuracy: 0.7722 - val_loss: 0.8908 - val_categorical_accuracy: 0.6936\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6029 - categorical_accuracy: 0.7970 - val_loss: 1.2505 - val_categorical_accuracy: 0.6146\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5300 - categorical_accuracy: 0.8192 - val_loss: 1.2802 - val_categorical_accuracy: 0.6110\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4628 - categorical_accuracy: 0.8439 - val_loss: 0.7506 - val_categorical_accuracy: 0.7608\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.4111 - categorical_accuracy: 0.8606 - val_loss: 0.9975 - val_categorical_accuracy: 0.6882\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.3618 - categorical_accuracy: 0.8779 - val_loss: 0.9138 - val_categorical_accuracy: 0.7110\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.3178 - categorical_accuracy: 0.8900 - val_loss: 1.5118 - val_categorical_accuracy: 0.6334\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.2743 - categorical_accuracy: 0.9065 - val_loss: 1.2272 - val_categorical_accuracy: 0.6944\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.2481 - categorical_accuracy: 0.9154 - val_loss: 0.9776 - val_categorical_accuracy: 0.7356\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.2165 - categorical_accuracy: 0.9259 - val_loss: 0.9263 - val_categorical_accuracy: 0.7594\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1866 - categorical_accuracy: 0.9365 - val_loss: 0.8050 - val_categorical_accuracy: 0.7890\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.1798 - categorical_accuracy: 0.9391 - val_loss: 1.0112 - val_categorical_accuracy: 0.7562\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1611 - categorical_accuracy: 0.9440 - val_loss: 0.9671 - val_categorical_accuracy: 0.7738\n",
      "Epoch 18/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 18s 25ms/step - loss: 0.1381 - categorical_accuracy: 0.9536 - val_loss: 0.9545 - val_categorical_accuracy: 0.7700\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1324 - categorical_accuracy: 0.9545 - val_loss: 1.0457 - val_categorical_accuracy: 0.7600\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.1240 - categorical_accuracy: 0.9579 - val_loss: 0.9037 - val_categorical_accuracy: 0.7902\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1104 - categorical_accuracy: 0.9615 - val_loss: 1.1092 - val_categorical_accuracy: 0.7726\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1021 - categorical_accuracy: 0.9664 - val_loss: 0.9397 - val_categorical_accuracy: 0.7964\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 27s 38ms/step - loss: 0.1027 - categorical_accuracy: 0.9651 - val_loss: 0.9497 - val_categorical_accuracy: 0.7982\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0920 - categorical_accuracy: 0.9698 - val_loss: 0.9973 - val_categorical_accuracy: 0.7954\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 22s 31ms/step - loss: 0.0792 - categorical_accuracy: 0.9733 - val_loss: 1.0186 - val_categorical_accuracy: 0.7948\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0831 - categorical_accuracy: 0.9726 - val_loss: 1.5852 - val_categorical_accuracy: 0.7292\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0846 - categorical_accuracy: 0.9722 - val_loss: 1.0744 - val_categorical_accuracy: 0.7832\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0721 - categorical_accuracy: 0.9762 - val_loss: 1.0360 - val_categorical_accuracy: 0.7976\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0760 - categorical_accuracy: 0.9746 - val_loss: 1.1935 - val_categorical_accuracy: 0.7804\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0676 - categorical_accuracy: 0.9777 - val_loss: 1.2166 - val_categorical_accuracy: 0.7800\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0704 - categorical_accuracy: 0.9763 - val_loss: 1.2173 - val_categorical_accuracy: 0.7872\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0626 - categorical_accuracy: 0.9794 - val_loss: 1.0954 - val_categorical_accuracy: 0.7882\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 24s 35ms/step - loss: 0.0639 - categorical_accuracy: 0.9789 - val_loss: 1.2637 - val_categorical_accuracy: 0.7794\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 23s 32ms/step - loss: 0.0561 - categorical_accuracy: 0.9812 - val_loss: 1.3332 - val_categorical_accuracy: 0.7754\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 26s 37ms/step - loss: 0.0605 - categorical_accuracy: 0.9806 - val_loss: 1.0573 - val_categorical_accuracy: 0.8056\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0619 - categorical_accuracy: 0.9794 - val_loss: 1.2645 - val_categorical_accuracy: 0.7846\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0515 - categorical_accuracy: 0.9826 - val_loss: 1.0303 - val_categorical_accuracy: 0.8130\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.0496 - categorical_accuracy: 0.9838 - val_loss: 1.2602 - val_categorical_accuracy: 0.7902\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 17s 23ms/step - loss: 0.0536 - categorical_accuracy: 0.9834 - val_loss: 1.2960 - val_categorical_accuracy: 0.7826\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0466 - categorical_accuracy: 0.9843 - val_loss: 1.1754 - val_categorical_accuracy: 0.7982\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0488 - categorical_accuracy: 0.9839 - val_loss: 1.2174 - val_categorical_accuracy: 0.7892\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0506 - categorical_accuracy: 0.9837 - val_loss: 1.1230 - val_categorical_accuracy: 0.8078\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0417 - categorical_accuracy: 0.9860 - val_loss: 1.1833 - val_categorical_accuracy: 0.7936\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0474 - categorical_accuracy: 0.9847 - val_loss: 1.2722 - val_categorical_accuracy: 0.8006\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0470 - categorical_accuracy: 0.9847 - val_loss: 1.6460 - val_categorical_accuracy: 0.7486\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0409 - categorical_accuracy: 0.9868 - val_loss: 1.1988 - val_categorical_accuracy: 0.7996\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0434 - categorical_accuracy: 0.9862 - val_loss: 1.3615 - val_categorical_accuracy: 0.7952\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0372 - categorical_accuracy: 0.9880 - val_loss: 1.1635 - val_categorical_accuracy: 0.8064\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 17s 23ms/step - loss: 0.0435 - categorical_accuracy: 0.9855 - val_loss: 1.4655 - val_categorical_accuracy: 0.7784\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0421 - categorical_accuracy: 0.9866 - val_loss: 1.2439 - val_categorical_accuracy: 0.7978\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0405 - categorical_accuracy: 0.9870 - val_loss: 1.5896 - val_categorical_accuracy: 0.7666\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0344 - categorical_accuracy: 0.9884 - val_loss: 1.1300 - val_categorical_accuracy: 0.8186\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0358 - categorical_accuracy: 0.9879 - val_loss: 1.2868 - val_categorical_accuracy: 0.8064\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0413 - categorical_accuracy: 0.9872 - val_loss: 1.3290 - val_categorical_accuracy: 0.8146\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0339 - categorical_accuracy: 0.9889 - val_loss: 1.3014 - val_categorical_accuracy: 0.8022\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 17s 23ms/step - loss: 0.0332 - categorical_accuracy: 0.9893 - val_loss: 1.4546 - val_categorical_accuracy: 0.7974\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0361 - categorical_accuracy: 0.9885 - val_loss: 1.2684 - val_categorical_accuracy: 0.8006\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0314 - categorical_accuracy: 0.9897 - val_loss: 1.3728 - val_categorical_accuracy: 0.8064\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0343 - categorical_accuracy: 0.9883 - val_loss: 1.2307 - val_categorical_accuracy: 0.8134\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0291 - categorical_accuracy: 0.9904 - val_loss: 2.2406 - val_categorical_accuracy: 0.7300\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0397 - categorical_accuracy: 0.9871 - val_loss: 1.1718 - val_categorical_accuracy: 0.8198\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0257 - categorical_accuracy: 0.9917 - val_loss: 1.1815 - val_categorical_accuracy: 0.8216\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0344 - categorical_accuracy: 0.9884 - val_loss: 1.4103 - val_categorical_accuracy: 0.8048\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0357 - categorical_accuracy: 0.9890 - val_loss: 1.2269 - val_categorical_accuracy: 0.8208\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0236 - categorical_accuracy: 0.9927 - val_loss: 1.2589 - val_categorical_accuracy: 0.8242\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0327 - categorical_accuracy: 0.9898 - val_loss: 1.2744 - val_categorical_accuracy: 0.8118\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0310 - categorical_accuracy: 0.9898 - val_loss: 1.5119 - val_categorical_accuracy: 0.7808\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0299 - categorical_accuracy: 0.9901 - val_loss: 1.4214 - val_categorical_accuracy: 0.7992\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 23s 33ms/step - loss: 0.0265 - categorical_accuracy: 0.9914 - val_loss: 1.4274 - val_categorical_accuracy: 0.7946\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0288 - categorical_accuracy: 0.9902 - val_loss: 1.4571 - val_categorical_accuracy: 0.7966\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0283 - categorical_accuracy: 0.9912 - val_loss: 1.2976 - val_categorical_accuracy: 0.8148\n",
      "Epoch 72/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0297 - categorical_accuracy: 0.9902 - val_loss: 1.9785 - val_categorical_accuracy: 0.7480\n",
      "Epoch 73/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0278 - categorical_accuracy: 0.9910 - val_loss: 1.6700 - val_categorical_accuracy: 0.7536\n",
      "Epoch 74/100\n",
      "704/704 [==============================] - 24s 35ms/step - loss: 0.0285 - categorical_accuracy: 0.9904 - val_loss: 1.2389 - val_categorical_accuracy: 0.8238\n",
      "Epoch 75/100\n",
      "704/704 [==============================] - 27s 38ms/step - loss: 0.0246 - categorical_accuracy: 0.9922 - val_loss: 1.2968 - val_categorical_accuracy: 0.8112\n",
      "Epoch 76/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.0195 - categorical_accuracy: 0.9939 - val_loss: 1.6827 - val_categorical_accuracy: 0.7898\n",
      "Epoch 77/100\n",
      "704/704 [==============================] - 17s 23ms/step - loss: 0.0251 - categorical_accuracy: 0.9920 - val_loss: 1.4518 - val_categorical_accuracy: 0.7868\n",
      "Epoch 78/100\n",
      "704/704 [==============================] - 22s 31ms/step - loss: 0.0253 - categorical_accuracy: 0.9919 - val_loss: 1.3609 - val_categorical_accuracy: 0.8052\n",
      "Epoch 79/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0275 - categorical_accuracy: 0.9911 - val_loss: 1.4703 - val_categorical_accuracy: 0.8016\n",
      "Epoch 80/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0194 - categorical_accuracy: 0.9940 - val_loss: 1.5042 - val_categorical_accuracy: 0.8116\n",
      "Epoch 81/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0340 - categorical_accuracy: 0.9897 - val_loss: 1.2406 - val_categorical_accuracy: 0.8154\n",
      "Epoch 82/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0210 - categorical_accuracy: 0.9933 - val_loss: 1.4606 - val_categorical_accuracy: 0.8012\n",
      "Epoch 83/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0196 - categorical_accuracy: 0.9937 - val_loss: 1.3859 - val_categorical_accuracy: 0.8090\n",
      "Epoch 84/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0255 - categorical_accuracy: 0.9918 - val_loss: 1.3597 - val_categorical_accuracy: 0.8042\n",
      "Epoch 85/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0234 - categorical_accuracy: 0.9928 - val_loss: 1.4151 - val_categorical_accuracy: 0.8012\n",
      "Epoch 86/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0226 - categorical_accuracy: 0.9927 - val_loss: 1.5334 - val_categorical_accuracy: 0.8012\n",
      "Epoch 87/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0210 - categorical_accuracy: 0.9931 - val_loss: 1.5522 - val_categorical_accuracy: 0.8050\n",
      "Epoch 88/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.0246 - categorical_accuracy: 0.9926 - val_loss: 1.4939 - val_categorical_accuracy: 0.7928\n",
      "Epoch 89/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0261 - categorical_accuracy: 0.9920 - val_loss: 1.3217 - val_categorical_accuracy: 0.8116\n",
      "Epoch 90/100\n",
      "704/704 [==============================] - 23s 33ms/step - loss: 0.0202 - categorical_accuracy: 0.9934 - val_loss: 1.2842 - val_categorical_accuracy: 0.8196\n",
      "Epoch 91/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0233 - categorical_accuracy: 0.9927 - val_loss: 3.0298 - val_categorical_accuracy: 0.6772\n",
      "Epoch 92/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0242 - categorical_accuracy: 0.9921 - val_loss: 1.2134 - val_categorical_accuracy: 0.8252\n",
      "Epoch 93/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0194 - categorical_accuracy: 0.9937 - val_loss: 1.5527 - val_categorical_accuracy: 0.7994\n",
      "Epoch 94/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0203 - categorical_accuracy: 0.9936 - val_loss: 1.5673 - val_categorical_accuracy: 0.7944\n",
      "Epoch 95/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0275 - categorical_accuracy: 0.9911 - val_loss: 1.4195 - val_categorical_accuracy: 0.8096\n",
      "Epoch 96/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0188 - categorical_accuracy: 0.9938 - val_loss: 1.4311 - val_categorical_accuracy: 0.8208\n",
      "Epoch 97/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0161 - categorical_accuracy: 0.9951 - val_loss: 1.3236 - val_categorical_accuracy: 0.8162\n",
      "Epoch 98/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0169 - categorical_accuracy: 0.9944 - val_loss: 1.4616 - val_categorical_accuracy: 0.8036\n",
      "Epoch 99/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0294 - categorical_accuracy: 0.9906 - val_loss: 1.6023 - val_categorical_accuracy: 0.7858\n",
      "Epoch 100/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0189 - categorical_accuracy: 0.9940 - val_loss: 1.4282 - val_categorical_accuracy: 0.8194\n",
      "--- Running training session 17\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 22s 32ms/step - loss: 1.5579 - categorical_accuracy: 0.4298 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 1.2045 - categorical_accuracy: 0.5742 - val_loss: 1.3323 - val_categorical_accuracy: 0.5482\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 1.0250 - categorical_accuracy: 0.6432 - val_loss: 1.0828 - val_categorical_accuracy: 0.6162\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.9080 - categorical_accuracy: 0.6857 - val_loss: 1.2200 - val_categorical_accuracy: 0.6000\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.8106 - categorical_accuracy: 0.7230 - val_loss: 1.1539 - val_categorical_accuracy: 0.6158\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.7317 - categorical_accuracy: 0.7520 - val_loss: 1.2949 - val_categorical_accuracy: 0.5846\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.6603 - categorical_accuracy: 0.7787 - val_loss: 0.9983 - val_categorical_accuracy: 0.6706\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 17s 23ms/step - loss: 0.5907 - categorical_accuracy: 0.8011 - val_loss: 1.2076 - val_categorical_accuracy: 0.6482\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.5291 - categorical_accuracy: 0.8192 - val_loss: 1.5284 - val_categorical_accuracy: 0.6112\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.4695 - categorical_accuracy: 0.8408 - val_loss: 2.3182 - val_categorical_accuracy: 0.5362\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.4247 - categorical_accuracy: 0.8552 - val_loss: 1.3154 - val_categorical_accuracy: 0.6460\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.3790 - categorical_accuracy: 0.8706 - val_loss: 1.4862 - val_categorical_accuracy: 0.6622\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.3377 - categorical_accuracy: 0.8844 - val_loss: 1.3976 - val_categorical_accuracy: 0.6592\n",
      "Epoch 14/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 18s 26ms/step - loss: 0.3047 - categorical_accuracy: 0.8946 - val_loss: 0.9425 - val_categorical_accuracy: 0.7566\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.2757 - categorical_accuracy: 0.9043 - val_loss: 1.0653 - val_categorical_accuracy: 0.7356\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.2525 - categorical_accuracy: 0.9135 - val_loss: 1.1557 - val_categorical_accuracy: 0.7354\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 20s 29ms/step - loss: 0.2300 - categorical_accuracy: 0.9209 - val_loss: 1.0596 - val_categorical_accuracy: 0.7510\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.2060 - categorical_accuracy: 0.9297 - val_loss: 1.2250 - val_categorical_accuracy: 0.7452\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.1933 - categorical_accuracy: 0.9342 - val_loss: 1.1578 - val_categorical_accuracy: 0.7474\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.1774 - categorical_accuracy: 0.9394 - val_loss: 2.8227 - val_categorical_accuracy: 0.6238\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1692 - categorical_accuracy: 0.9431 - val_loss: 2.1247 - val_categorical_accuracy: 0.6530\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 22s 32ms/step - loss: 0.1578 - categorical_accuracy: 0.9478 - val_loss: 1.9993 - val_categorical_accuracy: 0.6746\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 26s 37ms/step - loss: 0.1475 - categorical_accuracy: 0.9502 - val_loss: 1.5382 - val_categorical_accuracy: 0.7010\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.1398 - categorical_accuracy: 0.9528 - val_loss: 1.4004 - val_categorical_accuracy: 0.7538\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 20s 29ms/step - loss: 0.1346 - categorical_accuracy: 0.9547 - val_loss: 1.2802 - val_categorical_accuracy: 0.7678\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 22s 31ms/step - loss: 0.1210 - categorical_accuracy: 0.9596 - val_loss: 2.0700 - val_categorical_accuracy: 0.6878\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.1181 - categorical_accuracy: 0.9610 - val_loss: 1.0853 - val_categorical_accuracy: 0.7852\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.1148 - categorical_accuracy: 0.9624 - val_loss: 1.3755 - val_categorical_accuracy: 0.7488\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.1099 - categorical_accuracy: 0.9632 - val_loss: 1.1616 - val_categorical_accuracy: 0.7814\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 0.1015 - categorical_accuracy: 0.9669 - val_loss: 2.1665 - val_categorical_accuracy: 0.7114\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.0963 - categorical_accuracy: 0.9691 - val_loss: 1.0779 - val_categorical_accuracy: 0.8040\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0944 - categorical_accuracy: 0.9691 - val_loss: 1.1300 - val_categorical_accuracy: 0.7890\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 22s 31ms/step - loss: 0.0955 - categorical_accuracy: 0.9694 - val_loss: 2.3621 - val_categorical_accuracy: 0.6802\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 23s 33ms/step - loss: 0.0897 - categorical_accuracy: 0.9711 - val_loss: 1.6775 - val_categorical_accuracy: 0.7646\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.0872 - categorical_accuracy: 0.9716 - val_loss: 1.4673 - val_categorical_accuracy: 0.7548\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0869 - categorical_accuracy: 0.9713 - val_loss: 1.2869 - val_categorical_accuracy: 0.7600\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0792 - categorical_accuracy: 0.9735 - val_loss: 1.2990 - val_categorical_accuracy: 0.7856\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0850 - categorical_accuracy: 0.9720 - val_loss: 1.4249 - val_categorical_accuracy: 0.7826\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0789 - categorical_accuracy: 0.9749 - val_loss: 1.1149 - val_categorical_accuracy: 0.8042\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0647 - categorical_accuracy: 0.9796 - val_loss: 1.2350 - val_categorical_accuracy: 0.7974\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0710 - categorical_accuracy: 0.9771 - val_loss: 1.7861 - val_categorical_accuracy: 0.7446\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0734 - categorical_accuracy: 0.9760 - val_loss: 1.4889 - val_categorical_accuracy: 0.7794\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0712 - categorical_accuracy: 0.9778 - val_loss: 1.6235 - val_categorical_accuracy: 0.7552\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0664 - categorical_accuracy: 0.9791 - val_loss: 1.2857 - val_categorical_accuracy: 0.7988\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0587 - categorical_accuracy: 0.9809 - val_loss: 1.4502 - val_categorical_accuracy: 0.7676\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0694 - categorical_accuracy: 0.9781 - val_loss: 1.4882 - val_categorical_accuracy: 0.7700\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0631 - categorical_accuracy: 0.9800 - val_loss: 1.5362 - val_categorical_accuracy: 0.7886\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0596 - categorical_accuracy: 0.9808 - val_loss: 1.5039 - val_categorical_accuracy: 0.7856\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0637 - categorical_accuracy: 0.9795 - val_loss: 1.7244 - val_categorical_accuracy: 0.7884\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0644 - categorical_accuracy: 0.9801 - val_loss: 1.2293 - val_categorical_accuracy: 0.8036\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0558 - categorical_accuracy: 0.9824 - val_loss: 1.7165 - val_categorical_accuracy: 0.7378\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0588 - categorical_accuracy: 0.9814 - val_loss: 1.3757 - val_categorical_accuracy: 0.7898\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0554 - categorical_accuracy: 0.9825 - val_loss: 1.8976 - val_categorical_accuracy: 0.7644\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0590 - categorical_accuracy: 0.9812 - val_loss: 1.1934 - val_categorical_accuracy: 0.8066\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0489 - categorical_accuracy: 0.9841 - val_loss: 1.6865 - val_categorical_accuracy: 0.7842\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0516 - categorical_accuracy: 0.9836 - val_loss: 1.6544 - val_categorical_accuracy: 0.7806\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.0515 - categorical_accuracy: 0.9832 - val_loss: 1.6791 - val_categorical_accuracy: 0.7878\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0508 - categorical_accuracy: 0.9834 - val_loss: 1.2684 - val_categorical_accuracy: 0.8098\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0475 - categorical_accuracy: 0.9853 - val_loss: 1.3045 - val_categorical_accuracy: 0.8046\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0498 - categorical_accuracy: 0.9841 - val_loss: 1.5721 - val_categorical_accuracy: 0.7976\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0480 - categorical_accuracy: 0.9850 - val_loss: 1.4104 - val_categorical_accuracy: 0.8094\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0499 - categorical_accuracy: 0.9841 - val_loss: 1.4201 - val_categorical_accuracy: 0.8068\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0491 - categorical_accuracy: 0.9844 - val_loss: 1.5436 - val_categorical_accuracy: 0.7952\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.0470 - categorical_accuracy: 0.9849 - val_loss: 1.6794 - val_categorical_accuracy: 0.7826\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0435 - categorical_accuracy: 0.9859 - val_loss: 1.6885 - val_categorical_accuracy: 0.7810\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.0480 - categorical_accuracy: 0.9850 - val_loss: 2.4017 - val_categorical_accuracy: 0.7516\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.0428 - categorical_accuracy: 0.9864 - val_loss: 2.3326 - val_categorical_accuracy: 0.7446\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0450 - categorical_accuracy: 0.9853 - val_loss: 1.5443 - val_categorical_accuracy: 0.7928\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 23s 32ms/step - loss: 0.0398 - categorical_accuracy: 0.9871 - val_loss: 1.6620 - val_categorical_accuracy: 0.7886\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0475 - categorical_accuracy: 0.9848 - val_loss: 1.3772 - val_categorical_accuracy: 0.8142\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 26s 36ms/step - loss: 0.0426 - categorical_accuracy: 0.9870 - val_loss: 1.3630 - val_categorical_accuracy: 0.7986\n",
      "Epoch 72/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0436 - categorical_accuracy: 0.9862 - val_loss: 1.4786 - val_categorical_accuracy: 0.7880\n",
      "Epoch 73/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0379 - categorical_accuracy: 0.9880 - val_loss: 1.4648 - val_categorical_accuracy: 0.8064\n",
      "Epoch 74/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0383 - categorical_accuracy: 0.9878 - val_loss: 1.8370 - val_categorical_accuracy: 0.7422\n",
      "Epoch 75/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0403 - categorical_accuracy: 0.9873 - val_loss: 1.5206 - val_categorical_accuracy: 0.7982\n",
      "Epoch 76/100\n",
      "704/704 [==============================] - 23s 32ms/step - loss: 0.0391 - categorical_accuracy: 0.9882 - val_loss: 1.8887 - val_categorical_accuracy: 0.7868\n",
      "Epoch 77/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0349 - categorical_accuracy: 0.9894 - val_loss: 1.6190 - val_categorical_accuracy: 0.8074\n",
      "Epoch 78/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0398 - categorical_accuracy: 0.9871 - val_loss: 1.8769 - val_categorical_accuracy: 0.7990\n",
      "Epoch 79/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.0401 - categorical_accuracy: 0.9875 - val_loss: 1.7890 - val_categorical_accuracy: 0.7992\n",
      "Epoch 80/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0375 - categorical_accuracy: 0.9883 - val_loss: 1.6914 - val_categorical_accuracy: 0.7950\n",
      "Epoch 81/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.0375 - categorical_accuracy: 0.9884 - val_loss: 1.6799 - val_categorical_accuracy: 0.8090\n",
      "Epoch 82/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0400 - categorical_accuracy: 0.9877 - val_loss: 1.5717 - val_categorical_accuracy: 0.8090\n",
      "Epoch 83/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 0.0353 - categorical_accuracy: 0.9892 - val_loss: 1.6703 - val_categorical_accuracy: 0.8026\n",
      "Epoch 84/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0336 - categorical_accuracy: 0.9894 - val_loss: 1.5560 - val_categorical_accuracy: 0.8030\n",
      "Epoch 85/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0353 - categorical_accuracy: 0.9893 - val_loss: 1.7449 - val_categorical_accuracy: 0.7790\n",
      "Epoch 86/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0337 - categorical_accuracy: 0.9896 - val_loss: 1.7972 - val_categorical_accuracy: 0.7822\n",
      "Epoch 87/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0324 - categorical_accuracy: 0.9892 - val_loss: 1.7502 - val_categorical_accuracy: 0.8092\n",
      "Epoch 88/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0401 - categorical_accuracy: 0.9872 - val_loss: 1.6336 - val_categorical_accuracy: 0.7994\n",
      "Epoch 89/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0340 - categorical_accuracy: 0.9897 - val_loss: 2.6567 - val_categorical_accuracy: 0.7308\n",
      "Epoch 90/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0339 - categorical_accuracy: 0.9901 - val_loss: 1.5481 - val_categorical_accuracy: 0.8098\n",
      "Epoch 91/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0311 - categorical_accuracy: 0.9908 - val_loss: 1.7736 - val_categorical_accuracy: 0.7834\n",
      "Epoch 92/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0319 - categorical_accuracy: 0.9902 - val_loss: 1.5771 - val_categorical_accuracy: 0.8104\n",
      "Epoch 93/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0381 - categorical_accuracy: 0.9888 - val_loss: 2.1834 - val_categorical_accuracy: 0.7614\n",
      "Epoch 94/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0342 - categorical_accuracy: 0.9891 - val_loss: 1.4798 - val_categorical_accuracy: 0.8272\n",
      "Epoch 95/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0304 - categorical_accuracy: 0.9909 - val_loss: 1.9511 - val_categorical_accuracy: 0.7658\n",
      "Epoch 96/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0374 - categorical_accuracy: 0.9887 - val_loss: 1.4955 - val_categorical_accuracy: 0.8158\n",
      "Epoch 97/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0302 - categorical_accuracy: 0.9909 - val_loss: 1.7404 - val_categorical_accuracy: 0.8034\n",
      "Epoch 98/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0364 - categorical_accuracy: 0.9891 - val_loss: 1.9291 - val_categorical_accuracy: 0.7908\n",
      "Epoch 99/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0283 - categorical_accuracy: 0.9909 - val_loss: 1.9551 - val_categorical_accuracy: 0.8000\n",
      "Epoch 100/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0328 - categorical_accuracy: 0.9897 - val_loss: 1.8811 - val_categorical_accuracy: 0.7894\n",
      "--- Running training session 18\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 22s 32ms/step - loss: 1.6869 - categorical_accuracy: 0.3712 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 1.3825 - categorical_accuracy: 0.5033 - val_loss: 6.7336 - val_categorical_accuracy: 0.1816\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 1.2204 - categorical_accuracy: 0.5691 - val_loss: 1.1268 - val_categorical_accuracy: 0.6028\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 1.0940 - categorical_accuracy: 0.6205 - val_loss: 1.0195 - val_categorical_accuracy: 0.6534\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.9925 - categorical_accuracy: 0.6551 - val_loss: 1.0549 - val_categorical_accuracy: 0.6298\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.9238 - categorical_accuracy: 0.6838 - val_loss: 1.4165 - val_categorical_accuracy: 0.5634\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.8530 - categorical_accuracy: 0.7081 - val_loss: 1.0208 - val_categorical_accuracy: 0.6754\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.7953 - categorical_accuracy: 0.7297 - val_loss: 1.1231 - val_categorical_accuracy: 0.6616\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.7264 - categorical_accuracy: 0.7557 - val_loss: 0.9303 - val_categorical_accuracy: 0.6930\n",
      "Epoch 10/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6798 - categorical_accuracy: 0.7690 - val_loss: 0.8735 - val_categorical_accuracy: 0.7216\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 21s 29ms/step - loss: 0.6287 - categorical_accuracy: 0.7873 - val_loss: 0.9363 - val_categorical_accuracy: 0.7214\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5820 - categorical_accuracy: 0.8012 - val_loss: 1.0093 - val_categorical_accuracy: 0.7258\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.5331 - categorical_accuracy: 0.8198 - val_loss: 0.8984 - val_categorical_accuracy: 0.7526\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5028 - categorical_accuracy: 0.8303 - val_loss: 1.2438 - val_categorical_accuracy: 0.6902\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4648 - categorical_accuracy: 0.8430 - val_loss: 1.3208 - val_categorical_accuracy: 0.6878\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 28s 40ms/step - loss: 0.4320 - categorical_accuracy: 0.8530 - val_loss: 1.0820 - val_categorical_accuracy: 0.7300\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 28s 40ms/step - loss: 0.3993 - categorical_accuracy: 0.8640 - val_loss: 1.0256 - val_categorical_accuracy: 0.7430\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 23s 33ms/step - loss: 0.3715 - categorical_accuracy: 0.8725 - val_loss: 1.0558 - val_categorical_accuracy: 0.7676\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.3542 - categorical_accuracy: 0.8795 - val_loss: 1.8392 - val_categorical_accuracy: 0.6586\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.3287 - categorical_accuracy: 0.8901 - val_loss: 1.5922 - val_categorical_accuracy: 0.6786\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 27s 39ms/step - loss: 0.3082 - categorical_accuracy: 0.8980 - val_loss: 1.7363 - val_categorical_accuracy: 0.6788\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.2937 - categorical_accuracy: 0.9041 - val_loss: 1.2997 - val_categorical_accuracy: 0.7280\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.2785 - categorical_accuracy: 0.9090 - val_loss: 1.6074 - val_categorical_accuracy: 0.7206\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 28s 40ms/step - loss: 0.2565 - categorical_accuracy: 0.9131 - val_loss: 1.1485 - val_categorical_accuracy: 0.7714\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.2509 - categorical_accuracy: 0.9167 - val_loss: 1.5612 - val_categorical_accuracy: 0.7212\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.2362 - categorical_accuracy: 0.9220 - val_loss: 1.3357 - val_categorical_accuracy: 0.7534\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.2267 - categorical_accuracy: 0.9250 - val_loss: 1.3109 - val_categorical_accuracy: 0.7754\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 22s 31ms/step - loss: 0.2165 - categorical_accuracy: 0.9296 - val_loss: 1.1982 - val_categorical_accuracy: 0.7352\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 26s 37ms/step - loss: 0.1997 - categorical_accuracy: 0.9360 - val_loss: 1.7429 - val_categorical_accuracy: 0.6962\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.2011 - categorical_accuracy: 0.9334 - val_loss: 2.2484 - val_categorical_accuracy: 0.6352\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.1865 - categorical_accuracy: 0.9395 - val_loss: 1.2880 - val_categorical_accuracy: 0.7614\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1771 - categorical_accuracy: 0.9431 - val_loss: 1.6473 - val_categorical_accuracy: 0.7518\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 22s 31ms/step - loss: 0.1726 - categorical_accuracy: 0.9459 - val_loss: 1.4481 - val_categorical_accuracy: 0.7576\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1696 - categorical_accuracy: 0.9461 - val_loss: 1.5378 - val_categorical_accuracy: 0.7534\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1535 - categorical_accuracy: 0.9518 - val_loss: 1.5997 - val_categorical_accuracy: 0.7568\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1528 - categorical_accuracy: 0.9528 - val_loss: 1.6909 - val_categorical_accuracy: 0.7542\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1609 - categorical_accuracy: 0.9491 - val_loss: 1.6203 - val_categorical_accuracy: 0.7602\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1494 - categorical_accuracy: 0.9528 - val_loss: 1.7563 - val_categorical_accuracy: 0.7350\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1492 - categorical_accuracy: 0.9531 - val_loss: 1.4428 - val_categorical_accuracy: 0.7842\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1358 - categorical_accuracy: 0.9573 - val_loss: 2.2742 - val_categorical_accuracy: 0.7060\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1426 - categorical_accuracy: 0.9552 - val_loss: 1.7256 - val_categorical_accuracy: 0.7662\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1346 - categorical_accuracy: 0.9591 - val_loss: 1.2162 - val_categorical_accuracy: 0.7864\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1264 - categorical_accuracy: 0.9596 - val_loss: 2.7577 - val_categorical_accuracy: 0.6934\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1296 - categorical_accuracy: 0.9591 - val_loss: 1.7643 - val_categorical_accuracy: 0.7812\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1155 - categorical_accuracy: 0.9642 - val_loss: 1.7463 - val_categorical_accuracy: 0.7750\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1283 - categorical_accuracy: 0.9601 - val_loss: 1.6387 - val_categorical_accuracy: 0.7750\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1189 - categorical_accuracy: 0.9639 - val_loss: 1.7069 - val_categorical_accuracy: 0.7806\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1165 - categorical_accuracy: 0.9649 - val_loss: 1.7859 - val_categorical_accuracy: 0.7472\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1167 - categorical_accuracy: 0.9643 - val_loss: 2.0032 - val_categorical_accuracy: 0.7310\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1083 - categorical_accuracy: 0.9667 - val_loss: 1.9144 - val_categorical_accuracy: 0.7360\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1246 - categorical_accuracy: 0.9616 - val_loss: 1.7320 - val_categorical_accuracy: 0.7692\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1033 - categorical_accuracy: 0.9679 - val_loss: 1.6304 - val_categorical_accuracy: 0.7868\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.1049 - categorical_accuracy: 0.9688 - val_loss: 2.1533 - val_categorical_accuracy: 0.7536\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1040 - categorical_accuracy: 0.9689 - val_loss: 1.9703 - val_categorical_accuracy: 0.7530\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.1067 - categorical_accuracy: 0.9676 - val_loss: 1.8191 - val_categorical_accuracy: 0.7612\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0909 - categorical_accuracy: 0.9728 - val_loss: 1.7751 - val_categorical_accuracy: 0.7798\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 16s 22ms/step - loss: 0.1061 - categorical_accuracy: 0.9682 - val_loss: 2.4854 - val_categorical_accuracy: 0.7084\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 19s 28ms/step - loss: 0.1045 - categorical_accuracy: 0.9687 - val_loss: 1.6583 - val_categorical_accuracy: 0.7794\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0917 - categorical_accuracy: 0.9733 - val_loss: 3.5869 - val_categorical_accuracy: 0.6978\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0932 - categorical_accuracy: 0.9717 - val_loss: 2.0237 - val_categorical_accuracy: 0.7710\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0933 - categorical_accuracy: 0.9724 - val_loss: 1.7916 - val_categorical_accuracy: 0.7994\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.0949 - categorical_accuracy: 0.9727 - val_loss: 1.9684 - val_categorical_accuracy: 0.7896\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 22s 32ms/step - loss: 0.0865 - categorical_accuracy: 0.9737 - val_loss: 1.6635 - val_categorical_accuracy: 0.8002\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 28s 40ms/step - loss: 0.0913 - categorical_accuracy: 0.9733 - val_loss: 1.7572 - val_categorical_accuracy: 0.7894\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.0809 - categorical_accuracy: 0.9761 - val_loss: 1.5798 - val_categorical_accuracy: 0.7938\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0957 - categorical_accuracy: 0.9728 - val_loss: 1.8161 - val_categorical_accuracy: 0.7886\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0801 - categorical_accuracy: 0.9760 - val_loss: 1.6661 - val_categorical_accuracy: 0.7950\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0865 - categorical_accuracy: 0.9745 - val_loss: 1.8710 - val_categorical_accuracy: 0.7928\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0843 - categorical_accuracy: 0.9749 - val_loss: 2.5496 - val_categorical_accuracy: 0.7430\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0901 - categorical_accuracy: 0.9750 - val_loss: 1.7811 - val_categorical_accuracy: 0.7748\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.0777 - categorical_accuracy: 0.9774 - val_loss: 1.7182 - val_categorical_accuracy: 0.7942\n",
      "Epoch 72/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.0731 - categorical_accuracy: 0.9779 - val_loss: 1.7925 - val_categorical_accuracy: 0.8028\n",
      "Epoch 73/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0854 - categorical_accuracy: 0.9744 - val_loss: 2.5485 - val_categorical_accuracy: 0.7686\n",
      "Epoch 74/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0778 - categorical_accuracy: 0.9770 - val_loss: 1.9646 - val_categorical_accuracy: 0.7974\n",
      "Epoch 75/100\n",
      "704/704 [==============================] - 23s 33ms/step - loss: 0.0805 - categorical_accuracy: 0.9764 - val_loss: 1.8096 - val_categorical_accuracy: 0.7776\n",
      "Epoch 76/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.0751 - categorical_accuracy: 0.9779 - val_loss: 2.0598 - val_categorical_accuracy: 0.7868\n",
      "Epoch 77/100\n",
      "704/704 [==============================] - 24s 34ms/step - loss: 0.0823 - categorical_accuracy: 0.9769 - val_loss: 2.1314 - val_categorical_accuracy: 0.7768\n",
      "Epoch 78/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0803 - categorical_accuracy: 0.9775 - val_loss: 1.9231 - val_categorical_accuracy: 0.8024\n",
      "Epoch 79/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.0673 - categorical_accuracy: 0.9798 - val_loss: 2.0441 - val_categorical_accuracy: 0.7952\n",
      "Epoch 80/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.0728 - categorical_accuracy: 0.9795 - val_loss: 1.9787 - val_categorical_accuracy: 0.7866\n",
      "Epoch 81/100\n",
      "704/704 [==============================] - 17s 23ms/step - loss: 0.0747 - categorical_accuracy: 0.9790 - val_loss: 2.3939 - val_categorical_accuracy: 0.7906\n",
      "Epoch 82/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0671 - categorical_accuracy: 0.9807 - val_loss: 2.0022 - val_categorical_accuracy: 0.8030\n",
      "Epoch 83/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0739 - categorical_accuracy: 0.9784 - val_loss: 1.9796 - val_categorical_accuracy: 0.7934\n",
      "Epoch 84/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0753 - categorical_accuracy: 0.9789 - val_loss: 2.7495 - val_categorical_accuracy: 0.7566\n",
      "Epoch 85/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0701 - categorical_accuracy: 0.9797 - val_loss: 2.8367 - val_categorical_accuracy: 0.7602\n",
      "Epoch 86/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0762 - categorical_accuracy: 0.9785 - val_loss: 2.2255 - val_categorical_accuracy: 0.7782\n",
      "Epoch 87/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0656 - categorical_accuracy: 0.9810 - val_loss: 2.0636 - val_categorical_accuracy: 0.7854\n",
      "Epoch 88/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0667 - categorical_accuracy: 0.9815 - val_loss: 1.7659 - val_categorical_accuracy: 0.7920\n",
      "Epoch 89/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0767 - categorical_accuracy: 0.9791 - val_loss: 1.9928 - val_categorical_accuracy: 0.8016\n",
      "Epoch 90/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0695 - categorical_accuracy: 0.9802 - val_loss: 2.6134 - val_categorical_accuracy: 0.7602\n",
      "Epoch 91/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0667 - categorical_accuracy: 0.9808 - val_loss: 1.9382 - val_categorical_accuracy: 0.8124\n",
      "Epoch 92/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0665 - categorical_accuracy: 0.9814 - val_loss: 2.1851 - val_categorical_accuracy: 0.7926\n",
      "Epoch 93/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0584 - categorical_accuracy: 0.9836 - val_loss: 2.0978 - val_categorical_accuracy: 0.8038\n",
      "Epoch 94/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.0761 - categorical_accuracy: 0.9793 - val_loss: 2.4065 - val_categorical_accuracy: 0.7866\n",
      "Epoch 95/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0687 - categorical_accuracy: 0.9817 - val_loss: 2.2649 - val_categorical_accuracy: 0.7916\n",
      "Epoch 96/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0675 - categorical_accuracy: 0.9822 - val_loss: 2.2128 - val_categorical_accuracy: 0.8034\n",
      "Epoch 97/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0672 - categorical_accuracy: 0.9819 - val_loss: 2.0204 - val_categorical_accuracy: 0.8018\n",
      "Epoch 98/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0676 - categorical_accuracy: 0.9819 - val_loss: 1.8489 - val_categorical_accuracy: 0.8050\n",
      "Epoch 99/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0605 - categorical_accuracy: 0.9839 - val_loss: 2.1110 - val_categorical_accuracy: 0.7842\n",
      "Epoch 100/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 0.0627 - categorical_accuracy: 0.9824 - val_loss: 2.1787 - val_categorical_accuracy: 0.7988\n",
      "--- Running training session 19\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 17s 23ms/step - loss: 2.3028 - categorical_accuracy: 0.0983 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3028 - categorical_accuracy: 0.0993 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3028 - categorical_accuracy: 0.0986 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 6/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3028 - categorical_accuracy: 0.0988 - val_loss: 2.3027 - val_categorical_accuracy: 0.0970\n",
      "--- Running training session 20\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.3093 - categorical_accuracy: 0.0980 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0980 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3039 - categorical_accuracy: 0.0980 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3039 - categorical_accuracy: 0.0980 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "--- Running training session 21\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 3.5887 - categorical_accuracy: 0.0974 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3061 - categorical_accuracy: 0.0975 - val_loss: 2.3039 - val_categorical_accuracy: 0.1054\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1054\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0979 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3062 - categorical_accuracy: 0.0978 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0977 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 14s 21ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "--- Running training session 22\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.0949 - categorical_accuracy: 0.2036 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 1.9211 - categorical_accuracy: 0.2894 - val_loss: 1.6718 - val_categorical_accuracy: 0.3830\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 1.8124 - categorical_accuracy: 0.3316 - val_loss: 1.5671 - val_categorical_accuracy: 0.4274\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 1.7355 - categorical_accuracy: 0.3655 - val_loss: 1.4829 - val_categorical_accuracy: 0.4620\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.6723 - categorical_accuracy: 0.3873 - val_loss: 1.3396 - val_categorical_accuracy: 0.5074\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 1.6186 - categorical_accuracy: 0.4118 - val_loss: 1.3105 - val_categorical_accuracy: 0.5160\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 1.5837 - categorical_accuracy: 0.4270 - val_loss: 1.2511 - val_categorical_accuracy: 0.5398\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 1.5426 - categorical_accuracy: 0.4447 - val_loss: 1.1958 - val_categorical_accuracy: 0.5678\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 9/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.5120 - categorical_accuracy: 0.4534 - val_loss: 1.1751 - val_categorical_accuracy: 0.5810\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.4742 - categorical_accuracy: 0.4758 - val_loss: 1.1341 - val_categorical_accuracy: 0.6034\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.4376 - categorical_accuracy: 0.4906 - val_loss: 1.0453 - val_categorical_accuracy: 0.6250\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.4074 - categorical_accuracy: 0.5068 - val_loss: 1.0830 - val_categorical_accuracy: 0.6262\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.3747 - categorical_accuracy: 0.5145 - val_loss: 1.0857 - val_categorical_accuracy: 0.6142\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 1.3396 - categorical_accuracy: 0.5343 - val_loss: 0.9845 - val_categorical_accuracy: 0.6556\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 1.3194 - categorical_accuracy: 0.5398 - val_loss: 1.0434 - val_categorical_accuracy: 0.6382\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.2989 - categorical_accuracy: 0.5478 - val_loss: 0.9893 - val_categorical_accuracy: 0.6532\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 1.2691 - categorical_accuracy: 0.5564 - val_loss: 0.9846 - val_categorical_accuracy: 0.6554\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.2586 - categorical_accuracy: 0.5630 - val_loss: 0.9064 - val_categorical_accuracy: 0.6844\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.2379 - categorical_accuracy: 0.5710 - val_loss: 0.9221 - val_categorical_accuracy: 0.6846\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.2228 - categorical_accuracy: 0.5774 - val_loss: 0.9100 - val_categorical_accuracy: 0.6846\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.2027 - categorical_accuracy: 0.5861 - val_loss: 0.8918 - val_categorical_accuracy: 0.6856\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 1.1875 - categorical_accuracy: 0.5878 - val_loss: 0.8893 - val_categorical_accuracy: 0.6886\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.1707 - categorical_accuracy: 0.5962 - val_loss: 0.8719 - val_categorical_accuracy: 0.6978\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.1547 - categorical_accuracy: 0.6025 - val_loss: 0.8518 - val_categorical_accuracy: 0.7020\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.1423 - categorical_accuracy: 0.6048 - val_loss: 0.8414 - val_categorical_accuracy: 0.7036\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.1276 - categorical_accuracy: 0.6126 - val_loss: 0.8161 - val_categorical_accuracy: 0.7196\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.1197 - categorical_accuracy: 0.6148 - val_loss: 0.8053 - val_categorical_accuracy: 0.7188\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.1175 - categorical_accuracy: 0.6134 - val_loss: 0.8048 - val_categorical_accuracy: 0.7210\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.1055 - categorical_accuracy: 0.6216 - val_loss: 0.8371 - val_categorical_accuracy: 0.7058\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.0974 - categorical_accuracy: 0.6262 - val_loss: 0.8033 - val_categorical_accuracy: 0.7198\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 1.0750 - categorical_accuracy: 0.6306 - val_loss: 0.7877 - val_categorical_accuracy: 0.7216\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.0680 - categorical_accuracy: 0.6358 - val_loss: 0.7761 - val_categorical_accuracy: 0.7300\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.0653 - categorical_accuracy: 0.6355 - val_loss: 0.7965 - val_categorical_accuracy: 0.7274\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.0517 - categorical_accuracy: 0.6403 - val_loss: 0.8091 - val_categorical_accuracy: 0.7234\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.0415 - categorical_accuracy: 0.6467 - val_loss: 0.7744 - val_categorical_accuracy: 0.7336\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.0317 - categorical_accuracy: 0.6478 - val_loss: 0.7450 - val_categorical_accuracy: 0.7402\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.0285 - categorical_accuracy: 0.6500 - val_loss: 0.7863 - val_categorical_accuracy: 0.7314\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.0256 - categorical_accuracy: 0.6514 - val_loss: 0.7286 - val_categorical_accuracy: 0.7476\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.0034 - categorical_accuracy: 0.6611 - val_loss: 0.7481 - val_categorical_accuracy: 0.7444\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.0035 - categorical_accuracy: 0.6597 - val_loss: 0.7344 - val_categorical_accuracy: 0.7534\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.0041 - categorical_accuracy: 0.6578 - val_loss: 0.7399 - val_categorical_accuracy: 0.7434\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.9946 - categorical_accuracy: 0.6647 - val_loss: 0.7513 - val_categorical_accuracy: 0.7410\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.9733 - categorical_accuracy: 0.6698 - val_loss: 0.7187 - val_categorical_accuracy: 0.7540\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.9718 - categorical_accuracy: 0.6685 - val_loss: 0.7050 - val_categorical_accuracy: 0.7554\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 0.9646 - categorical_accuracy: 0.6706 - val_loss: 0.7230 - val_categorical_accuracy: 0.7586\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.9693 - categorical_accuracy: 0.6734 - val_loss: 0.7080 - val_categorical_accuracy: 0.7590\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 14s 21ms/step - loss: 0.9612 - categorical_accuracy: 0.6734 - val_loss: 0.7139 - val_categorical_accuracy: 0.7586\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.9552 - categorical_accuracy: 0.6782 - val_loss: 0.6918 - val_categorical_accuracy: 0.7626\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 0.9436 - categorical_accuracy: 0.6831 - val_loss: 0.6803 - val_categorical_accuracy: 0.7700\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.9456 - categorical_accuracy: 0.6822 - val_loss: 0.6944 - val_categorical_accuracy: 0.7638\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.9354 - categorical_accuracy: 0.6864 - val_loss: 0.6659 - val_categorical_accuracy: 0.7744\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 0.9291 - categorical_accuracy: 0.6873 - val_loss: 0.6959 - val_categorical_accuracy: 0.7670\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.9280 - categorical_accuracy: 0.6883 - val_loss: 0.7022 - val_categorical_accuracy: 0.7618\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.9290 - categorical_accuracy: 0.6867 - val_loss: 0.6695 - val_categorical_accuracy: 0.7726\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.9161 - categorical_accuracy: 0.6941 - val_loss: 0.6748 - val_categorical_accuracy: 0.7762\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.9134 - categorical_accuracy: 0.6916 - val_loss: 0.6937 - val_categorical_accuracy: 0.7754\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.9083 - categorical_accuracy: 0.6948 - val_loss: 0.6724 - val_categorical_accuracy: 0.7806\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.9046 - categorical_accuracy: 0.6981 - val_loss: 0.6830 - val_categorical_accuracy: 0.7692\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8931 - categorical_accuracy: 0.7023 - val_loss: 0.6467 - val_categorical_accuracy: 0.7908\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8911 - categorical_accuracy: 0.7012 - val_loss: 0.7038 - val_categorical_accuracy: 0.7700\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.8912 - categorical_accuracy: 0.7028 - val_loss: 0.6989 - val_categorical_accuracy: 0.7704\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8900 - categorical_accuracy: 0.7005 - val_loss: 0.6549 - val_categorical_accuracy: 0.7848\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8819 - categorical_accuracy: 0.7048 - val_loss: 0.6810 - val_categorical_accuracy: 0.7772\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8766 - categorical_accuracy: 0.7084 - val_loss: 0.6853 - val_categorical_accuracy: 0.7816\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.8723 - categorical_accuracy: 0.7088 - val_loss: 0.6537 - val_categorical_accuracy: 0.7818\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8738 - categorical_accuracy: 0.7073 - val_loss: 0.6805 - val_categorical_accuracy: 0.7784\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.8641 - categorical_accuracy: 0.7089 - val_loss: 0.6680 - val_categorical_accuracy: 0.7800\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8643 - categorical_accuracy: 0.7109 - val_loss: 0.6336 - val_categorical_accuracy: 0.7942\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8586 - categorical_accuracy: 0.7148 - val_loss: 0.6564 - val_categorical_accuracy: 0.7820\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8532 - categorical_accuracy: 0.7132 - val_loss: 0.6320 - val_categorical_accuracy: 0.7948\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8554 - categorical_accuracy: 0.7118 - val_loss: 0.6505 - val_categorical_accuracy: 0.7886\n",
      "Epoch 72/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8600 - categorical_accuracy: 0.7123 - val_loss: 0.6282 - val_categorical_accuracy: 0.7900\n",
      "Epoch 73/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.8491 - categorical_accuracy: 0.7146 - val_loss: 0.6471 - val_categorical_accuracy: 0.7908\n",
      "Epoch 74/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.8472 - categorical_accuracy: 0.7160 - val_loss: 0.6688 - val_categorical_accuracy: 0.7802\n",
      "Epoch 75/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8498 - categorical_accuracy: 0.7160 - val_loss: 0.6734 - val_categorical_accuracy: 0.7864\n",
      "Epoch 76/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8335 - categorical_accuracy: 0.7198 - val_loss: 0.6306 - val_categorical_accuracy: 0.7916\n",
      "Epoch 77/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8333 - categorical_accuracy: 0.7229 - val_loss: 0.6182 - val_categorical_accuracy: 0.8032\n",
      "Epoch 78/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8465 - categorical_accuracy: 0.7192 - val_loss: 0.6591 - val_categorical_accuracy: 0.7892\n",
      "Epoch 79/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8407 - categorical_accuracy: 0.7221 - val_loss: 0.6547 - val_categorical_accuracy: 0.7898\n",
      "Epoch 80/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8140 - categorical_accuracy: 0.7276 - val_loss: 0.6182 - val_categorical_accuracy: 0.7964\n",
      "Epoch 81/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8308 - categorical_accuracy: 0.7243 - val_loss: 0.6590 - val_categorical_accuracy: 0.7872\n",
      "Epoch 82/100\n",
      "704/704 [==============================] - 14s 21ms/step - loss: 0.8325 - categorical_accuracy: 0.7229 - val_loss: 0.6363 - val_categorical_accuracy: 0.7934\n",
      "Epoch 83/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8246 - categorical_accuracy: 0.7236 - val_loss: 0.6280 - val_categorical_accuracy: 0.7932\n",
      "Epoch 84/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8250 - categorical_accuracy: 0.7252 - val_loss: 0.6630 - val_categorical_accuracy: 0.7848\n",
      "Epoch 85/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8112 - categorical_accuracy: 0.7328 - val_loss: 0.6136 - val_categorical_accuracy: 0.7950\n",
      "Epoch 86/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8139 - categorical_accuracy: 0.7288 - val_loss: 0.6305 - val_categorical_accuracy: 0.7962\n",
      "Epoch 87/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8150 - categorical_accuracy: 0.7297 - val_loss: 0.6490 - val_categorical_accuracy: 0.7912\n",
      "Epoch 88/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8046 - categorical_accuracy: 0.7327 - val_loss: 0.6198 - val_categorical_accuracy: 0.7978\n",
      "Epoch 89/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8011 - categorical_accuracy: 0.7318 - val_loss: 0.6389 - val_categorical_accuracy: 0.7948\n",
      "Epoch 90/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8035 - categorical_accuracy: 0.7344 - val_loss: 0.6527 - val_categorical_accuracy: 0.7910\n",
      "Epoch 91/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8034 - categorical_accuracy: 0.7334 - val_loss: 0.6385 - val_categorical_accuracy: 0.7942\n",
      "Epoch 92/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.8010 - categorical_accuracy: 0.7368 - val_loss: 0.6081 - val_categorical_accuracy: 0.8086\n",
      "Epoch 93/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7955 - categorical_accuracy: 0.7347 - val_loss: 0.6184 - val_categorical_accuracy: 0.8046\n",
      "Epoch 94/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7922 - categorical_accuracy: 0.7364 - val_loss: 0.6510 - val_categorical_accuracy: 0.8004\n",
      "Epoch 95/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.7931 - categorical_accuracy: 0.7372 - val_loss: 0.6406 - val_categorical_accuracy: 0.7892\n",
      "Epoch 96/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8023 - categorical_accuracy: 0.7320 - val_loss: 0.6237 - val_categorical_accuracy: 0.8036\n",
      "Epoch 97/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.7887 - categorical_accuracy: 0.7370 - val_loss: 0.6215 - val_categorical_accuracy: 0.7996\n",
      "Epoch 98/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7851 - categorical_accuracy: 0.7397 - val_loss: 0.6043 - val_categorical_accuracy: 0.8110\n",
      "Epoch 99/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 0.7832 - categorical_accuracy: 0.7399 - val_loss: 0.6295 - val_categorical_accuracy: 0.7996\n",
      "Epoch 100/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7834 - categorical_accuracy: 0.7400 - val_loss: 0.6365 - val_categorical_accuracy: 0.7956\n",
      "--- Running training session 23\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 3.1754 - categorical_accuracy: 0.0972 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3039 - categorical_accuracy: 0.0977 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 5/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 14s 21ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "--- Running training session 24\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 23.9706 - categorical_accuracy: 0.0973 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3061 - categorical_accuracy: 0.0972 - val_loss: 2.3039 - val_categorical_accuracy: 0.1038\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0975 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 14s 21ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 14s 21ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "--- Running training session 25\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 16s 23ms/step - loss: 2.0639 - categorical_accuracy: 0.2242 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.8671 - categorical_accuracy: 0.3069 - val_loss: 1.6354 - val_categorical_accuracy: 0.4116\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 1.7298 - categorical_accuracy: 0.3670 - val_loss: 1.4478 - val_categorical_accuracy: 0.4816\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.6304 - categorical_accuracy: 0.4102 - val_loss: 1.3366 - val_categorical_accuracy: 0.5290\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.5448 - categorical_accuracy: 0.4443 - val_loss: 1.2113 - val_categorical_accuracy: 0.5732\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.4804 - categorical_accuracy: 0.4750 - val_loss: 1.1096 - val_categorical_accuracy: 0.6030\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.4156 - categorical_accuracy: 0.5016 - val_loss: 1.1137 - val_categorical_accuracy: 0.6104\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.3740 - categorical_accuracy: 0.5178 - val_loss: 1.0174 - val_categorical_accuracy: 0.6330\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.3358 - categorical_accuracy: 0.5329 - val_loss: 1.0510 - val_categorical_accuracy: 0.6354\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.3070 - categorical_accuracy: 0.5423 - val_loss: 1.0059 - val_categorical_accuracy: 0.6508\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.2739 - categorical_accuracy: 0.5539 - val_loss: 0.9299 - val_categorical_accuracy: 0.6684\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.2475 - categorical_accuracy: 0.5680 - val_loss: 0.9769 - val_categorical_accuracy: 0.6530\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.2170 - categorical_accuracy: 0.5771 - val_loss: 0.9206 - val_categorical_accuracy: 0.6788\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.1944 - categorical_accuracy: 0.5887 - val_loss: 0.8990 - val_categorical_accuracy: 0.6862\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.1689 - categorical_accuracy: 0.5968 - val_loss: 0.8931 - val_categorical_accuracy: 0.6900\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.1516 - categorical_accuracy: 0.6073 - val_loss: 0.8797 - val_categorical_accuracy: 0.6970\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.1301 - categorical_accuracy: 0.6110 - val_loss: 0.8807 - val_categorical_accuracy: 0.6932\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.1151 - categorical_accuracy: 0.6183 - val_loss: 0.8459 - val_categorical_accuracy: 0.7068\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.1070 - categorical_accuracy: 0.6229 - val_loss: 0.8186 - val_categorical_accuracy: 0.7258\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.0882 - categorical_accuracy: 0.6260 - val_loss: 0.8222 - val_categorical_accuracy: 0.7130\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.0775 - categorical_accuracy: 0.6329 - val_loss: 0.8038 - val_categorical_accuracy: 0.7158\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.0564 - categorical_accuracy: 0.6403 - val_loss: 0.7715 - val_categorical_accuracy: 0.7332\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 1.0473 - categorical_accuracy: 0.6441 - val_loss: 0.7930 - val_categorical_accuracy: 0.7274\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.0383 - categorical_accuracy: 0.6473 - val_loss: 0.7851 - val_categorical_accuracy: 0.7334\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 1.0262 - categorical_accuracy: 0.6514 - val_loss: 0.7427 - val_categorical_accuracy: 0.7416\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 1.0132 - categorical_accuracy: 0.6573 - val_loss: 0.7662 - val_categorical_accuracy: 0.7424\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 1.0102 - categorical_accuracy: 0.6556 - val_loss: 0.7862 - val_categorical_accuracy: 0.7288\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 0.9994 - categorical_accuracy: 0.6597 - val_loss: 0.7372 - val_categorical_accuracy: 0.7496\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.9828 - categorical_accuracy: 0.6662 - val_loss: 0.7469 - val_categorical_accuracy: 0.7480\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.9782 - categorical_accuracy: 0.6664 - val_loss: 0.7150 - val_categorical_accuracy: 0.7554\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.9684 - categorical_accuracy: 0.6694 - val_loss: 0.7377 - val_categorical_accuracy: 0.7500\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.9659 - categorical_accuracy: 0.6737 - val_loss: 0.7091 - val_categorical_accuracy: 0.7592\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.9604 - categorical_accuracy: 0.6772 - val_loss: 0.7033 - val_categorical_accuracy: 0.7650\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.9453 - categorical_accuracy: 0.6802 - val_loss: 0.7216 - val_categorical_accuracy: 0.7546\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.9529 - categorical_accuracy: 0.6795 - val_loss: 0.7067 - val_categorical_accuracy: 0.7620\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.9343 - categorical_accuracy: 0.6859 - val_loss: 0.7088 - val_categorical_accuracy: 0.7666\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.9268 - categorical_accuracy: 0.6873 - val_loss: 0.7245 - val_categorical_accuracy: 0.7562\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.9295 - categorical_accuracy: 0.6877 - val_loss: 0.6964 - val_categorical_accuracy: 0.7654\n",
      "Epoch 39/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 14s 20ms/step - loss: 0.9195 - categorical_accuracy: 0.6893 - val_loss: 0.6974 - val_categorical_accuracy: 0.7686\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.9037 - categorical_accuracy: 0.6935 - val_loss: 0.6944 - val_categorical_accuracy: 0.7678\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.9145 - categorical_accuracy: 0.6918 - val_loss: 0.6602 - val_categorical_accuracy: 0.7836\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8998 - categorical_accuracy: 0.6958 - val_loss: 0.7384 - val_categorical_accuracy: 0.7580\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8997 - categorical_accuracy: 0.6978 - val_loss: 0.6799 - val_categorical_accuracy: 0.7772\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.8933 - categorical_accuracy: 0.6978 - val_loss: 0.6596 - val_categorical_accuracy: 0.7800\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.8806 - categorical_accuracy: 0.7036 - val_loss: 0.6516 - val_categorical_accuracy: 0.7816\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8948 - categorical_accuracy: 0.7025 - val_loss: 0.6679 - val_categorical_accuracy: 0.7788\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.8788 - categorical_accuracy: 0.7039 - val_loss: 0.6509 - val_categorical_accuracy: 0.7806\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8710 - categorical_accuracy: 0.7084 - val_loss: 0.6618 - val_categorical_accuracy: 0.7878\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8765 - categorical_accuracy: 0.7047 - val_loss: 0.6666 - val_categorical_accuracy: 0.7778\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8650 - categorical_accuracy: 0.7094 - val_loss: 0.6508 - val_categorical_accuracy: 0.7800\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8706 - categorical_accuracy: 0.7106 - val_loss: 0.6570 - val_categorical_accuracy: 0.7864\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8536 - categorical_accuracy: 0.7150 - val_loss: 0.6544 - val_categorical_accuracy: 0.7792\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.8599 - categorical_accuracy: 0.7122 - val_loss: 0.6615 - val_categorical_accuracy: 0.7848\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8505 - categorical_accuracy: 0.7166 - val_loss: 0.6448 - val_categorical_accuracy: 0.7910\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8522 - categorical_accuracy: 0.7154 - val_loss: 0.6368 - val_categorical_accuracy: 0.7924\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8451 - categorical_accuracy: 0.7162 - val_loss: 0.6770 - val_categorical_accuracy: 0.7848\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8386 - categorical_accuracy: 0.7203 - val_loss: 0.6351 - val_categorical_accuracy: 0.7936\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8375 - categorical_accuracy: 0.7204 - val_loss: 0.6717 - val_categorical_accuracy: 0.7824\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.8387 - categorical_accuracy: 0.7201 - val_loss: 0.6613 - val_categorical_accuracy: 0.7890\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8369 - categorical_accuracy: 0.7240 - val_loss: 0.6174 - val_categorical_accuracy: 0.8020\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8239 - categorical_accuracy: 0.7262 - val_loss: 0.6319 - val_categorical_accuracy: 0.7954\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8237 - categorical_accuracy: 0.7250 - val_loss: 0.6681 - val_categorical_accuracy: 0.7938\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8206 - categorical_accuracy: 0.7250 - val_loss: 0.6157 - val_categorical_accuracy: 0.7972\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.8090 - categorical_accuracy: 0.7312 - val_loss: 0.6570 - val_categorical_accuracy: 0.7898\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8189 - categorical_accuracy: 0.7282 - val_loss: 0.6775 - val_categorical_accuracy: 0.7808\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8114 - categorical_accuracy: 0.7331 - val_loss: 0.6265 - val_categorical_accuracy: 0.7984\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.8102 - categorical_accuracy: 0.7289 - val_loss: 0.6931 - val_categorical_accuracy: 0.7824\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.8052 - categorical_accuracy: 0.7326 - val_loss: 0.6235 - val_categorical_accuracy: 0.7942\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.8006 - categorical_accuracy: 0.7311 - val_loss: 0.6091 - val_categorical_accuracy: 0.7950\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.7952 - categorical_accuracy: 0.7349 - val_loss: 0.6409 - val_categorical_accuracy: 0.7914\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7861 - categorical_accuracy: 0.7372 - val_loss: 0.6503 - val_categorical_accuracy: 0.7872\n",
      "Epoch 72/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7818 - categorical_accuracy: 0.7408 - val_loss: 0.6187 - val_categorical_accuracy: 0.8004\n",
      "Epoch 73/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7873 - categorical_accuracy: 0.7400 - val_loss: 0.6354 - val_categorical_accuracy: 0.7930\n",
      "Epoch 74/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7946 - categorical_accuracy: 0.7359 - val_loss: 0.6323 - val_categorical_accuracy: 0.7950\n",
      "Epoch 75/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7930 - categorical_accuracy: 0.7371 - val_loss: 0.6146 - val_categorical_accuracy: 0.8000\n",
      "Epoch 76/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7871 - categorical_accuracy: 0.7390 - val_loss: 0.6421 - val_categorical_accuracy: 0.7950\n",
      "Epoch 77/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.7931 - categorical_accuracy: 0.7345 - val_loss: 0.6148 - val_categorical_accuracy: 0.8046\n",
      "Epoch 78/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7895 - categorical_accuracy: 0.7402 - val_loss: 0.6252 - val_categorical_accuracy: 0.7986\n",
      "Epoch 79/100\n",
      "704/704 [==============================] - 14s 21ms/step - loss: 0.7782 - categorical_accuracy: 0.7428 - val_loss: 0.5955 - val_categorical_accuracy: 0.8134\n",
      "Epoch 80/100\n",
      "704/704 [==============================] - 14s 21ms/step - loss: 0.7821 - categorical_accuracy: 0.7413 - val_loss: 0.6561 - val_categorical_accuracy: 0.7880\n",
      "Epoch 81/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 0.7964 - categorical_accuracy: 0.7351 - val_loss: 0.6520 - val_categorical_accuracy: 0.7906\n",
      "Epoch 82/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 0.7852 - categorical_accuracy: 0.7404 - val_loss: 0.6285 - val_categorical_accuracy: 0.7910\n",
      "Epoch 83/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7824 - categorical_accuracy: 0.7398 - val_loss: 0.6327 - val_categorical_accuracy: 0.8000\n",
      "Epoch 84/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7760 - categorical_accuracy: 0.7453 - val_loss: 0.5935 - val_categorical_accuracy: 0.8078\n",
      "Epoch 85/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7651 - categorical_accuracy: 0.7474 - val_loss: 0.6679 - val_categorical_accuracy: 0.7864\n",
      "Epoch 86/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.7700 - categorical_accuracy: 0.7438 - val_loss: 0.6535 - val_categorical_accuracy: 0.7906\n",
      "Epoch 87/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.7626 - categorical_accuracy: 0.7466 - val_loss: 0.6011 - val_categorical_accuracy: 0.8076\n",
      "Epoch 88/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 0.7696 - categorical_accuracy: 0.7461 - val_loss: 0.6360 - val_categorical_accuracy: 0.7940\n",
      "Epoch 89/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7645 - categorical_accuracy: 0.7484 - val_loss: 0.6421 - val_categorical_accuracy: 0.7974\n",
      "Epoch 90/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.7618 - categorical_accuracy: 0.7500 - val_loss: 0.6480 - val_categorical_accuracy: 0.7940\n",
      "Epoch 91/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7641 - categorical_accuracy: 0.7489 - val_loss: 0.5885 - val_categorical_accuracy: 0.8104\n",
      "Epoch 92/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7545 - categorical_accuracy: 0.7520 - val_loss: 0.6173 - val_categorical_accuracy: 0.7992\n",
      "Epoch 93/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7580 - categorical_accuracy: 0.7492 - val_loss: 0.6509 - val_categorical_accuracy: 0.7988\n",
      "Epoch 94/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7532 - categorical_accuracy: 0.7518 - val_loss: 0.6460 - val_categorical_accuracy: 0.7958\n",
      "Epoch 95/100\n",
      "704/704 [==============================] - 14s 21ms/step - loss: 0.7591 - categorical_accuracy: 0.7495 - val_loss: 0.6096 - val_categorical_accuracy: 0.8016\n",
      "Epoch 96/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7620 - categorical_accuracy: 0.7492 - val_loss: 0.6380 - val_categorical_accuracy: 0.7970\n",
      "Epoch 97/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.7503 - categorical_accuracy: 0.7523 - val_loss: 0.6461 - val_categorical_accuracy: 0.7974\n",
      "Epoch 98/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7595 - categorical_accuracy: 0.7503 - val_loss: 0.6828 - val_categorical_accuracy: 0.7820\n",
      "Epoch 99/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 0.7445 - categorical_accuracy: 0.7541 - val_loss: 0.6317 - val_categorical_accuracy: 0.7962\n",
      "Epoch 100/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 0.7462 - categorical_accuracy: 0.7546 - val_loss: 0.5914 - val_categorical_accuracy: 0.8070\n",
      "--- Running training session 26\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 2.8302 - categorical_accuracy: 0.0981 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3039 - categorical_accuracy: 0.0978 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0980 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0980 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3039 - categorical_accuracy: 0.0979 - val_loss: 2.3027 - val_categorical_accuracy: 0.1054\n",
      "--- Running training session 27\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 206.5490 - categorical_accuracy: 0.0974 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 14s 19ms/step - loss: 2.3748 - categorical_accuracy: 0.0978 - val_loss: 2.3038 - val_categorical_accuracy: 0.1038\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3061 - categorical_accuracy: 0.0975 - val_loss: 2.3039 - val_categorical_accuracy: 0.1038\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 4/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3062 - categorical_accuracy: 0.0975 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0978 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0978 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3062 - categorical_accuracy: 0.0978 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0977 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 13s 18ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 14s 21ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 14s 20ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 15s 21ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 13s 19ms/step - loss: 2.3062 - categorical_accuracy: 0.0976 - val_loss: 2.3040 - val_categorical_accuracy: 0.1038\n",
      "--- Running training session 28\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 26s 37ms/step - loss: 1.9507 - categorical_accuracy: 0.2964 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.6471 - categorical_accuracy: 0.4084 - val_loss: 1.7332 - val_categorical_accuracy: 0.4056\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.4779 - categorical_accuracy: 0.4744 - val_loss: 1.2967 - val_categorical_accuracy: 0.5322\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.3763 - categorical_accuracy: 0.5153 - val_loss: 1.2142 - val_categorical_accuracy: 0.5596\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.2950 - categorical_accuracy: 0.5491 - val_loss: 1.8665 - val_categorical_accuracy: 0.4502\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.2212 - categorical_accuracy: 0.5752 - val_loss: 1.0669 - val_categorical_accuracy: 0.6126\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.1620 - categorical_accuracy: 0.5972 - val_loss: 1.0646 - val_categorical_accuracy: 0.6162\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.1144 - categorical_accuracy: 0.6156 - val_loss: 0.9922 - val_categorical_accuracy: 0.6452\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.0686 - categorical_accuracy: 0.6337 - val_loss: 1.4188 - val_categorical_accuracy: 0.5570\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.0275 - categorical_accuracy: 0.6508 - val_loss: 0.9875 - val_categorical_accuracy: 0.6540\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.9959 - categorical_accuracy: 0.6602 - val_loss: 0.7448 - val_categorical_accuracy: 0.7358\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.9610 - categorical_accuracy: 0.6741 - val_loss: 0.7536 - val_categorical_accuracy: 0.7326\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.9352 - categorical_accuracy: 0.6836 - val_loss: 0.6773 - val_categorical_accuracy: 0.7632\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.9038 - categorical_accuracy: 0.6964 - val_loss: 0.6991 - val_categorical_accuracy: 0.7596\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.8881 - categorical_accuracy: 0.6987 - val_loss: 0.7457 - val_categorical_accuracy: 0.7486\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8551 - categorical_accuracy: 0.7125 - val_loss: 0.6183 - val_categorical_accuracy: 0.7876\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8373 - categorical_accuracy: 0.7185 - val_loss: 0.7980 - val_categorical_accuracy: 0.7310\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8245 - categorical_accuracy: 0.7258 - val_loss: 0.6318 - val_categorical_accuracy: 0.7862\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8009 - categorical_accuracy: 0.7339 - val_loss: 0.6974 - val_categorical_accuracy: 0.7634\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7862 - categorical_accuracy: 0.7374 - val_loss: 0.6141 - val_categorical_accuracy: 0.7928\n",
      "Epoch 21/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7713 - categorical_accuracy: 0.7446 - val_loss: 0.8302 - val_categorical_accuracy: 0.7200\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7598 - categorical_accuracy: 0.7459 - val_loss: 0.5496 - val_categorical_accuracy: 0.8144\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7434 - categorical_accuracy: 0.7519 - val_loss: 0.5569 - val_categorical_accuracy: 0.8132\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7273 - categorical_accuracy: 0.7583 - val_loss: 0.5116 - val_categorical_accuracy: 0.8240\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7224 - categorical_accuracy: 0.7610 - val_loss: 0.5759 - val_categorical_accuracy: 0.8000\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7066 - categorical_accuracy: 0.7645 - val_loss: 0.5139 - val_categorical_accuracy: 0.8250\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7037 - categorical_accuracy: 0.7677 - val_loss: 0.5490 - val_categorical_accuracy: 0.8120\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6853 - categorical_accuracy: 0.7740 - val_loss: 0.5485 - val_categorical_accuracy: 0.8170\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6853 - categorical_accuracy: 0.7712 - val_loss: 0.5645 - val_categorical_accuracy: 0.8056\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6670 - categorical_accuracy: 0.7802 - val_loss: 0.5408 - val_categorical_accuracy: 0.8208\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6631 - categorical_accuracy: 0.7800 - val_loss: 0.5092 - val_categorical_accuracy: 0.8296\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6465 - categorical_accuracy: 0.7844 - val_loss: 0.5103 - val_categorical_accuracy: 0.8266\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6408 - categorical_accuracy: 0.7891 - val_loss: 0.5896 - val_categorical_accuracy: 0.8014\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6401 - categorical_accuracy: 0.7865 - val_loss: 0.4842 - val_categorical_accuracy: 0.8408\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6291 - categorical_accuracy: 0.7928 - val_loss: 0.5476 - val_categorical_accuracy: 0.8184\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6253 - categorical_accuracy: 0.7950 - val_loss: 0.5329 - val_categorical_accuracy: 0.8264\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6181 - categorical_accuracy: 0.7969 - val_loss: 0.4692 - val_categorical_accuracy: 0.8396\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6114 - categorical_accuracy: 0.7988 - val_loss: 0.4841 - val_categorical_accuracy: 0.8346\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6095 - categorical_accuracy: 0.7997 - val_loss: 0.4951 - val_categorical_accuracy: 0.8316\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5993 - categorical_accuracy: 0.8034 - val_loss: 0.4946 - val_categorical_accuracy: 0.8368\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5933 - categorical_accuracy: 0.8046 - val_loss: 0.5685 - val_categorical_accuracy: 0.8148\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5929 - categorical_accuracy: 0.8055 - val_loss: 0.5097 - val_categorical_accuracy: 0.8256\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5799 - categorical_accuracy: 0.8095 - val_loss: 0.4788 - val_categorical_accuracy: 0.8438\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5759 - categorical_accuracy: 0.8104 - val_loss: 0.5254 - val_categorical_accuracy: 0.8222\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5762 - categorical_accuracy: 0.8113 - val_loss: 0.5011 - val_categorical_accuracy: 0.8286\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5714 - categorical_accuracy: 0.8132 - val_loss: 0.5491 - val_categorical_accuracy: 0.8196\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5689 - categorical_accuracy: 0.8133 - val_loss: 0.4538 - val_categorical_accuracy: 0.8410\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5588 - categorical_accuracy: 0.8175 - val_loss: 0.5182 - val_categorical_accuracy: 0.8236\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5573 - categorical_accuracy: 0.8165 - val_loss: 0.5043 - val_categorical_accuracy: 0.8304\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5422 - categorical_accuracy: 0.8224 - val_loss: 0.5976 - val_categorical_accuracy: 0.8062\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5416 - categorical_accuracy: 0.8209 - val_loss: 0.4528 - val_categorical_accuracy: 0.8484\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5324 - categorical_accuracy: 0.8265 - val_loss: 0.4609 - val_categorical_accuracy: 0.8418\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5383 - categorical_accuracy: 0.8234 - val_loss: 0.5703 - val_categorical_accuracy: 0.8172\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5285 - categorical_accuracy: 0.8262 - val_loss: 0.4475 - val_categorical_accuracy: 0.8530\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5290 - categorical_accuracy: 0.8258 - val_loss: 0.5073 - val_categorical_accuracy: 0.8312\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.5257 - categorical_accuracy: 0.8285 - val_loss: 0.6089 - val_categorical_accuracy: 0.8094\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5163 - categorical_accuracy: 0.8323 - val_loss: 0.5199 - val_categorical_accuracy: 0.8336\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5205 - categorical_accuracy: 0.8299 - val_loss: 0.4812 - val_categorical_accuracy: 0.8410\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5127 - categorical_accuracy: 0.8335 - val_loss: 0.4501 - val_categorical_accuracy: 0.8490\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5128 - categorical_accuracy: 0.8333 - val_loss: 0.5689 - val_categorical_accuracy: 0.8180\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5038 - categorical_accuracy: 0.8340 - val_loss: 0.4384 - val_categorical_accuracy: 0.8572\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5096 - categorical_accuracy: 0.8321 - val_loss: 0.4560 - val_categorical_accuracy: 0.8474\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4998 - categorical_accuracy: 0.8360 - val_loss: 0.4750 - val_categorical_accuracy: 0.8502\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4984 - categorical_accuracy: 0.8370 - val_loss: 0.4358 - val_categorical_accuracy: 0.8600\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4890 - categorical_accuracy: 0.8400 - val_loss: 0.4489 - val_categorical_accuracy: 0.8532\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4952 - categorical_accuracy: 0.8393 - val_loss: 0.4320 - val_categorical_accuracy: 0.8574\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4850 - categorical_accuracy: 0.8416 - val_loss: 0.4316 - val_categorical_accuracy: 0.8558\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4800 - categorical_accuracy: 0.8428 - val_loss: 0.4758 - val_categorical_accuracy: 0.8498\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4818 - categorical_accuracy: 0.8422 - val_loss: 0.4840 - val_categorical_accuracy: 0.8398\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4800 - categorical_accuracy: 0.8434 - val_loss: 0.5474 - val_categorical_accuracy: 0.8248\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4778 - categorical_accuracy: 0.8445 - val_loss: 0.4814 - val_categorical_accuracy: 0.8418\n",
      "Epoch 72/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4759 - categorical_accuracy: 0.8443 - val_loss: 0.5716 - val_categorical_accuracy: 0.8222\n",
      "Epoch 73/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4667 - categorical_accuracy: 0.8488 - val_loss: 0.5143 - val_categorical_accuracy: 0.8318\n",
      "Epoch 74/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4710 - categorical_accuracy: 0.8471 - val_loss: 0.5428 - val_categorical_accuracy: 0.8296\n",
      "Epoch 75/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4742 - categorical_accuracy: 0.8454 - val_loss: 0.4481 - val_categorical_accuracy: 0.8544\n",
      "Epoch 76/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4642 - categorical_accuracy: 0.8496 - val_loss: 0.4755 - val_categorical_accuracy: 0.8426\n",
      "Epoch 77/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.4622 - categorical_accuracy: 0.8499 - val_loss: 0.4082 - val_categorical_accuracy: 0.8650\n",
      "Epoch 78/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4592 - categorical_accuracy: 0.8503 - val_loss: 0.4394 - val_categorical_accuracy: 0.8578\n",
      "Epoch 79/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4654 - categorical_accuracy: 0.8480 - val_loss: 0.4536 - val_categorical_accuracy: 0.8534\n",
      "Epoch 80/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4499 - categorical_accuracy: 0.8540 - val_loss: 0.4415 - val_categorical_accuracy: 0.8604\n",
      "Epoch 81/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4552 - categorical_accuracy: 0.8508 - val_loss: 0.4380 - val_categorical_accuracy: 0.8608\n",
      "Epoch 82/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4500 - categorical_accuracy: 0.8546 - val_loss: 0.4726 - val_categorical_accuracy: 0.8536\n",
      "Epoch 83/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4540 - categorical_accuracy: 0.8540 - val_loss: 0.4616 - val_categorical_accuracy: 0.8504\n",
      "Epoch 84/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4431 - categorical_accuracy: 0.8565 - val_loss: 0.4309 - val_categorical_accuracy: 0.8598\n",
      "Epoch 85/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4440 - categorical_accuracy: 0.8570 - val_loss: 0.4842 - val_categorical_accuracy: 0.8432\n",
      "Epoch 86/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4423 - categorical_accuracy: 0.8553 - val_loss: 0.4483 - val_categorical_accuracy: 0.8538\n",
      "Epoch 87/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4413 - categorical_accuracy: 0.8557 - val_loss: 0.3874 - val_categorical_accuracy: 0.8770\n",
      "Epoch 88/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4390 - categorical_accuracy: 0.8603 - val_loss: 0.5084 - val_categorical_accuracy: 0.8378\n",
      "Epoch 89/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4385 - categorical_accuracy: 0.8582 - val_loss: 0.4920 - val_categorical_accuracy: 0.8356\n",
      "Epoch 90/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4407 - categorical_accuracy: 0.8580 - val_loss: 0.4172 - val_categorical_accuracy: 0.8656\n",
      "Epoch 91/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4385 - categorical_accuracy: 0.8581 - val_loss: 0.4681 - val_categorical_accuracy: 0.8532\n",
      "Epoch 92/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4291 - categorical_accuracy: 0.8605 - val_loss: 0.4778 - val_categorical_accuracy: 0.8490\n",
      "Epoch 93/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4300 - categorical_accuracy: 0.8610 - val_loss: 0.4420 - val_categorical_accuracy: 0.8566\n",
      "Epoch 94/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4206 - categorical_accuracy: 0.8620 - val_loss: 0.5135 - val_categorical_accuracy: 0.8382\n",
      "Epoch 95/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4318 - categorical_accuracy: 0.8620 - val_loss: 0.4507 - val_categorical_accuracy: 0.8550\n",
      "Epoch 96/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4309 - categorical_accuracy: 0.8583 - val_loss: 0.4631 - val_categorical_accuracy: 0.8502\n",
      "Epoch 97/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4208 - categorical_accuracy: 0.8616 - val_loss: 0.4149 - val_categorical_accuracy: 0.8666\n",
      "Epoch 98/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4194 - categorical_accuracy: 0.8646 - val_loss: 0.5644 - val_categorical_accuracy: 0.8250\n",
      "Epoch 99/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4161 - categorical_accuracy: 0.8657 - val_loss: 0.5898 - val_categorical_accuracy: 0.8258\n",
      "Epoch 100/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4185 - categorical_accuracy: 0.8641 - val_loss: 0.4399 - val_categorical_accuracy: 0.8582\n",
      "--- Running training session 29\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 26s 37ms/step - loss: 1.9202 - categorical_accuracy: 0.2977 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 1.6941 - categorical_accuracy: 0.3848 - val_loss: 2.0061 - val_categorical_accuracy: 0.3614\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.5593 - categorical_accuracy: 0.4437 - val_loss: 1.3518 - val_categorical_accuracy: 0.4990\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.4609 - categorical_accuracy: 0.4792 - val_loss: 1.2906 - val_categorical_accuracy: 0.5280\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 1.3906 - categorical_accuracy: 0.5128 - val_loss: 1.3204 - val_categorical_accuracy: 0.5314\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.3139 - categorical_accuracy: 0.5396 - val_loss: 1.0708 - val_categorical_accuracy: 0.6078\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.2655 - categorical_accuracy: 0.5564 - val_loss: 1.0838 - val_categorical_accuracy: 0.6206\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.2249 - categorical_accuracy: 0.5770 - val_loss: 0.8892 - val_categorical_accuracy: 0.6820\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.1729 - categorical_accuracy: 0.5937 - val_loss: 0.9854 - val_categorical_accuracy: 0.6592\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.1405 - categorical_accuracy: 0.6082 - val_loss: 0.8724 - val_categorical_accuracy: 0.6978\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.0934 - categorical_accuracy: 0.6268 - val_loss: 0.9478 - val_categorical_accuracy: 0.6670\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.0651 - categorical_accuracy: 0.6364 - val_loss: 0.8019 - val_categorical_accuracy: 0.7252\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.0287 - categorical_accuracy: 0.6520 - val_loss: 0.8003 - val_categorical_accuracy: 0.7170\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 1.0082 - categorical_accuracy: 0.6590 - val_loss: 0.7258 - val_categorical_accuracy: 0.7462\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.9760 - categorical_accuracy: 0.6716 - val_loss: 0.7608 - val_categorical_accuracy: 0.7454\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.9478 - categorical_accuracy: 0.6824 - val_loss: 0.6718 - val_categorical_accuracy: 0.7668\n",
      "Epoch 17/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 18s 25ms/step - loss: 0.9390 - categorical_accuracy: 0.6867 - val_loss: 0.7076 - val_categorical_accuracy: 0.7610\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.9202 - categorical_accuracy: 0.6928 - val_loss: 0.7208 - val_categorical_accuracy: 0.7554\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8950 - categorical_accuracy: 0.7017 - val_loss: 0.6907 - val_categorical_accuracy: 0.7674\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8715 - categorical_accuracy: 0.7080 - val_loss: 0.8861 - val_categorical_accuracy: 0.7150\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8579 - categorical_accuracy: 0.7128 - val_loss: 0.6794 - val_categorical_accuracy: 0.7702\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8445 - categorical_accuracy: 0.7172 - val_loss: 0.6157 - val_categorical_accuracy: 0.8056\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8348 - categorical_accuracy: 0.7225 - val_loss: 0.5763 - val_categorical_accuracy: 0.8102\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8181 - categorical_accuracy: 0.7285 - val_loss: 0.6187 - val_categorical_accuracy: 0.7952\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8021 - categorical_accuracy: 0.7315 - val_loss: 0.6919 - val_categorical_accuracy: 0.7722\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7921 - categorical_accuracy: 0.7372 - val_loss: 0.9565 - val_categorical_accuracy: 0.7076\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7801 - categorical_accuracy: 0.7388 - val_loss: 0.6309 - val_categorical_accuracy: 0.7924\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7607 - categorical_accuracy: 0.7455 - val_loss: 0.6538 - val_categorical_accuracy: 0.7846\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7608 - categorical_accuracy: 0.7485 - val_loss: 0.6472 - val_categorical_accuracy: 0.7814\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7414 - categorical_accuracy: 0.7530 - val_loss: 0.5178 - val_categorical_accuracy: 0.8252\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7422 - categorical_accuracy: 0.7556 - val_loss: 0.5040 - val_categorical_accuracy: 0.8384\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7290 - categorical_accuracy: 0.7582 - val_loss: 0.6464 - val_categorical_accuracy: 0.7806\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7152 - categorical_accuracy: 0.7618 - val_loss: 0.6014 - val_categorical_accuracy: 0.8012\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7149 - categorical_accuracy: 0.7630 - val_loss: 0.5654 - val_categorical_accuracy: 0.8054\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7118 - categorical_accuracy: 0.7652 - val_loss: 0.5380 - val_categorical_accuracy: 0.8220\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6953 - categorical_accuracy: 0.7704 - val_loss: 0.5371 - val_categorical_accuracy: 0.8224\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6922 - categorical_accuracy: 0.7709 - val_loss: 0.6151 - val_categorical_accuracy: 0.7904\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6873 - categorical_accuracy: 0.7742 - val_loss: 0.6763 - val_categorical_accuracy: 0.7820\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6811 - categorical_accuracy: 0.7739 - val_loss: 0.5260 - val_categorical_accuracy: 0.8234\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6710 - categorical_accuracy: 0.7796 - val_loss: 0.5404 - val_categorical_accuracy: 0.8216\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6630 - categorical_accuracy: 0.7822 - val_loss: 0.5396 - val_categorical_accuracy: 0.8146\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6575 - categorical_accuracy: 0.7814 - val_loss: 0.6128 - val_categorical_accuracy: 0.7978\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6572 - categorical_accuracy: 0.7841 - val_loss: 0.5010 - val_categorical_accuracy: 0.8294\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6419 - categorical_accuracy: 0.7872 - val_loss: 0.5815 - val_categorical_accuracy: 0.8120\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.6395 - categorical_accuracy: 0.7896 - val_loss: 0.5665 - val_categorical_accuracy: 0.8124\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6346 - categorical_accuracy: 0.7926 - val_loss: 0.5157 - val_categorical_accuracy: 0.8334\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.6338 - categorical_accuracy: 0.7896 - val_loss: 0.4809 - val_categorical_accuracy: 0.8368\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6238 - categorical_accuracy: 0.7915 - val_loss: 0.4950 - val_categorical_accuracy: 0.8374\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6182 - categorical_accuracy: 0.7964 - val_loss: 0.5434 - val_categorical_accuracy: 0.8280\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6215 - categorical_accuracy: 0.7945 - val_loss: 0.5911 - val_categorical_accuracy: 0.8008\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6118 - categorical_accuracy: 0.7984 - val_loss: 0.4995 - val_categorical_accuracy: 0.8310\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6055 - categorical_accuracy: 0.8006 - val_loss: 0.5913 - val_categorical_accuracy: 0.8106\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5956 - categorical_accuracy: 0.8053 - val_loss: 0.5828 - val_categorical_accuracy: 0.8024\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6042 - categorical_accuracy: 0.8015 - val_loss: 0.5479 - val_categorical_accuracy: 0.8230\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.5976 - categorical_accuracy: 0.8042 - val_loss: 0.5596 - val_categorical_accuracy: 0.8140\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5858 - categorical_accuracy: 0.8081 - val_loss: 0.5740 - val_categorical_accuracy: 0.8148\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5914 - categorical_accuracy: 0.8046 - val_loss: 0.5301 - val_categorical_accuracy: 0.8290\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5796 - categorical_accuracy: 0.8096 - val_loss: 0.5414 - val_categorical_accuracy: 0.8284\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5806 - categorical_accuracy: 0.8090 - val_loss: 0.6232 - val_categorical_accuracy: 0.8040\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5755 - categorical_accuracy: 0.8106 - val_loss: 0.4996 - val_categorical_accuracy: 0.8356\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5750 - categorical_accuracy: 0.8102 - val_loss: 0.5159 - val_categorical_accuracy: 0.8308\n",
      "--- Running training session 30\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 27s 38ms/step - loss: 1.9703 - categorical_accuracy: 0.2697 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.8054 - categorical_accuracy: 0.3331 - val_loss: 1.9491 - val_categorical_accuracy: 0.3486\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.7076 - categorical_accuracy: 0.3791 - val_loss: 1.5540 - val_categorical_accuracy: 0.4400\n",
      "Epoch 4/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 17s 24ms/step - loss: 1.6121 - categorical_accuracy: 0.4160 - val_loss: 1.5518 - val_categorical_accuracy: 0.4672\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.5437 - categorical_accuracy: 0.4456 - val_loss: 1.3808 - val_categorical_accuracy: 0.5056\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.4814 - categorical_accuracy: 0.4726 - val_loss: 1.4322 - val_categorical_accuracy: 0.5024\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.4321 - categorical_accuracy: 0.4928 - val_loss: 1.1885 - val_categorical_accuracy: 0.5836\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.3973 - categorical_accuracy: 0.5070 - val_loss: 1.1424 - val_categorical_accuracy: 0.5950\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.3447 - categorical_accuracy: 0.5279 - val_loss: 1.0285 - val_categorical_accuracy: 0.6230\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.3137 - categorical_accuracy: 0.5412 - val_loss: 1.3725 - val_categorical_accuracy: 0.5402\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.2877 - categorical_accuracy: 0.5493 - val_loss: 1.1222 - val_categorical_accuracy: 0.6150\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 1.2590 - categorical_accuracy: 0.5667 - val_loss: 1.0227 - val_categorical_accuracy: 0.6368\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.2316 - categorical_accuracy: 0.5751 - val_loss: 0.9717 - val_categorical_accuracy: 0.6600\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.1993 - categorical_accuracy: 0.5860 - val_loss: 0.9776 - val_categorical_accuracy: 0.6590\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.1725 - categorical_accuracy: 0.5966 - val_loss: 0.8775 - val_categorical_accuracy: 0.6862\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.1418 - categorical_accuracy: 0.6092 - val_loss: 0.8401 - val_categorical_accuracy: 0.7100\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.1226 - categorical_accuracy: 0.6192 - val_loss: 0.9288 - val_categorical_accuracy: 0.6828\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.1027 - categorical_accuracy: 0.6278 - val_loss: 0.7721 - val_categorical_accuracy: 0.7354\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.0789 - categorical_accuracy: 0.6366 - val_loss: 0.7846 - val_categorical_accuracy: 0.7272\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 1.0602 - categorical_accuracy: 0.6439 - val_loss: 0.9823 - val_categorical_accuracy: 0.6858\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.0397 - categorical_accuracy: 0.6493 - val_loss: 0.7088 - val_categorical_accuracy: 0.7606\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.0275 - categorical_accuracy: 0.6536 - val_loss: 0.7719 - val_categorical_accuracy: 0.7382\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.0104 - categorical_accuracy: 0.6602 - val_loss: 0.7298 - val_categorical_accuracy: 0.7526\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.9950 - categorical_accuracy: 0.6672 - val_loss: 0.6928 - val_categorical_accuracy: 0.7620\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.9699 - categorical_accuracy: 0.6746 - val_loss: 0.6572 - val_categorical_accuracy: 0.7750\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.9649 - categorical_accuracy: 0.6777 - val_loss: 0.7120 - val_categorical_accuracy: 0.7702\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.9531 - categorical_accuracy: 0.6802 - val_loss: 0.6479 - val_categorical_accuracy: 0.7814\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.9408 - categorical_accuracy: 0.6850 - val_loss: 0.6883 - val_categorical_accuracy: 0.7684\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.9338 - categorical_accuracy: 0.6843 - val_loss: 0.6431 - val_categorical_accuracy: 0.7912\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.9108 - categorical_accuracy: 0.6949 - val_loss: 0.6699 - val_categorical_accuracy: 0.7798\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.9019 - categorical_accuracy: 0.7001 - val_loss: 0.6879 - val_categorical_accuracy: 0.7710\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8917 - categorical_accuracy: 0.7038 - val_loss: 0.6315 - val_categorical_accuracy: 0.7894\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8807 - categorical_accuracy: 0.7068 - val_loss: 0.7739 - val_categorical_accuracy: 0.7510\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8828 - categorical_accuracy: 0.7058 - val_loss: 0.6972 - val_categorical_accuracy: 0.7712\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8687 - categorical_accuracy: 0.7121 - val_loss: 0.6930 - val_categorical_accuracy: 0.7682\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8541 - categorical_accuracy: 0.7169 - val_loss: 0.6596 - val_categorical_accuracy: 0.7854\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.8555 - categorical_accuracy: 0.7175 - val_loss: 0.5924 - val_categorical_accuracy: 0.8000\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8474 - categorical_accuracy: 0.7206 - val_loss: 0.6887 - val_categorical_accuracy: 0.7726\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8422 - categorical_accuracy: 0.7194 - val_loss: 0.6052 - val_categorical_accuracy: 0.7976\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8368 - categorical_accuracy: 0.7216 - val_loss: 0.5677 - val_categorical_accuracy: 0.8078\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.8231 - categorical_accuracy: 0.7257 - val_loss: 0.5841 - val_categorical_accuracy: 0.7998\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8241 - categorical_accuracy: 0.7279 - val_loss: 0.6297 - val_categorical_accuracy: 0.7928\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8130 - categorical_accuracy: 0.7290 - val_loss: 0.6760 - val_categorical_accuracy: 0.7764\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.8024 - categorical_accuracy: 0.7334 - val_loss: 0.5927 - val_categorical_accuracy: 0.8086\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8006 - categorical_accuracy: 0.7334 - val_loss: 0.5331 - val_categorical_accuracy: 0.8276\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7987 - categorical_accuracy: 0.7346 - val_loss: 0.6398 - val_categorical_accuracy: 0.7980\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7901 - categorical_accuracy: 0.7385 - val_loss: 0.5878 - val_categorical_accuracy: 0.8038\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7888 - categorical_accuracy: 0.7396 - val_loss: 0.5495 - val_categorical_accuracy: 0.8186\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.7801 - categorical_accuracy: 0.7426 - val_loss: 0.5501 - val_categorical_accuracy: 0.8200\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7682 - categorical_accuracy: 0.7464 - val_loss: 0.4876 - val_categorical_accuracy: 0.8420\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7730 - categorical_accuracy: 0.7440 - val_loss: 0.5858 - val_categorical_accuracy: 0.8160\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7620 - categorical_accuracy: 0.7475 - val_loss: 0.5555 - val_categorical_accuracy: 0.8214\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7578 - categorical_accuracy: 0.7500 - val_loss: 0.5236 - val_categorical_accuracy: 0.8276\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7585 - categorical_accuracy: 0.7487 - val_loss: 0.5836 - val_categorical_accuracy: 0.8096\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7485 - categorical_accuracy: 0.7514 - val_loss: 0.5389 - val_categorical_accuracy: 0.8184\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7524 - categorical_accuracy: 0.7526 - val_loss: 0.5618 - val_categorical_accuracy: 0.8162\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7406 - categorical_accuracy: 0.7586 - val_loss: 0.5341 - val_categorical_accuracy: 0.8284\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7407 - categorical_accuracy: 0.7550 - val_loss: 0.5411 - val_categorical_accuracy: 0.8196\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7352 - categorical_accuracy: 0.7560 - val_loss: 0.5381 - val_categorical_accuracy: 0.8232\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7296 - categorical_accuracy: 0.7581 - val_loss: 0.5670 - val_categorical_accuracy: 0.8160\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7213 - categorical_accuracy: 0.7631 - val_loss: 0.5707 - val_categorical_accuracy: 0.8160\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7182 - categorical_accuracy: 0.7648 - val_loss: 0.5462 - val_categorical_accuracy: 0.8298\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7229 - categorical_accuracy: 0.7628 - val_loss: 0.5813 - val_categorical_accuracy: 0.8148\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7155 - categorical_accuracy: 0.7637 - val_loss: 0.7229 - val_categorical_accuracy: 0.7742\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.7057 - categorical_accuracy: 0.7694 - val_loss: 0.5190 - val_categorical_accuracy: 0.8342\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7082 - categorical_accuracy: 0.7702 - val_loss: 0.4951 - val_categorical_accuracy: 0.8362\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6992 - categorical_accuracy: 0.7695 - val_loss: 0.5481 - val_categorical_accuracy: 0.8296\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6996 - categorical_accuracy: 0.7715 - val_loss: 0.5306 - val_categorical_accuracy: 0.8314\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7036 - categorical_accuracy: 0.7675 - val_loss: 0.5162 - val_categorical_accuracy: 0.8284\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7011 - categorical_accuracy: 0.7695 - val_loss: 0.5124 - val_categorical_accuracy: 0.8390\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6924 - categorical_accuracy: 0.7745 - val_loss: 0.5680 - val_categorical_accuracy: 0.8068\n",
      "Epoch 72/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6896 - categorical_accuracy: 0.7742 - val_loss: 0.5712 - val_categorical_accuracy: 0.8220\n",
      "Epoch 73/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6898 - categorical_accuracy: 0.7726 - val_loss: 0.5378 - val_categorical_accuracy: 0.8300\n",
      "Epoch 74/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6873 - categorical_accuracy: 0.7749 - val_loss: 0.5333 - val_categorical_accuracy: 0.8278\n",
      "Epoch 75/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6807 - categorical_accuracy: 0.7782 - val_loss: 0.5342 - val_categorical_accuracy: 0.8342\n",
      "Epoch 76/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6784 - categorical_accuracy: 0.7797 - val_loss: 0.5590 - val_categorical_accuracy: 0.8144\n",
      "Epoch 77/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6762 - categorical_accuracy: 0.7759 - val_loss: 0.4962 - val_categorical_accuracy: 0.8436\n",
      "Epoch 78/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6642 - categorical_accuracy: 0.7811 - val_loss: 0.5262 - val_categorical_accuracy: 0.8312\n",
      "Epoch 79/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6617 - categorical_accuracy: 0.7831 - val_loss: 0.5701 - val_categorical_accuracy: 0.8258\n",
      "Epoch 80/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6638 - categorical_accuracy: 0.7804 - val_loss: 0.5081 - val_categorical_accuracy: 0.8354\n",
      "Epoch 81/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6648 - categorical_accuracy: 0.7818 - val_loss: 0.5040 - val_categorical_accuracy: 0.8400\n",
      "Epoch 82/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6733 - categorical_accuracy: 0.7791 - val_loss: 0.5124 - val_categorical_accuracy: 0.8342\n",
      "Epoch 83/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6624 - categorical_accuracy: 0.7847 - val_loss: 0.5305 - val_categorical_accuracy: 0.8284\n",
      "Epoch 84/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6579 - categorical_accuracy: 0.7845 - val_loss: 0.5227 - val_categorical_accuracy: 0.8302\n",
      "Epoch 85/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6559 - categorical_accuracy: 0.7859 - val_loss: 0.4788 - val_categorical_accuracy: 0.8434\n",
      "Epoch 86/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.6635 - categorical_accuracy: 0.7844 - val_loss: 0.5519 - val_categorical_accuracy: 0.8160\n",
      "Epoch 87/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6552 - categorical_accuracy: 0.7861 - val_loss: 0.4847 - val_categorical_accuracy: 0.8376\n",
      "Epoch 88/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6469 - categorical_accuracy: 0.7885 - val_loss: 0.5016 - val_categorical_accuracy: 0.8358\n",
      "Epoch 89/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.6545 - categorical_accuracy: 0.7839 - val_loss: 0.4914 - val_categorical_accuracy: 0.8414\n",
      "Epoch 90/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6462 - categorical_accuracy: 0.7898 - val_loss: 0.5422 - val_categorical_accuracy: 0.8274\n",
      "Epoch 91/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6385 - categorical_accuracy: 0.7893 - val_loss: 0.5475 - val_categorical_accuracy: 0.8200\n",
      "Epoch 92/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6470 - categorical_accuracy: 0.7882 - val_loss: 0.4834 - val_categorical_accuracy: 0.8398\n",
      "Epoch 93/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.6394 - categorical_accuracy: 0.7900 - val_loss: 0.5513 - val_categorical_accuracy: 0.8222\n",
      "Epoch 94/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6403 - categorical_accuracy: 0.7905 - val_loss: 0.4972 - val_categorical_accuracy: 0.8400\n",
      "Epoch 95/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6379 - categorical_accuracy: 0.7910 - val_loss: 0.5124 - val_categorical_accuracy: 0.8392\n",
      "Epoch 96/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.6435 - categorical_accuracy: 0.7901 - val_loss: 0.5073 - val_categorical_accuracy: 0.8354\n",
      "Epoch 97/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6259 - categorical_accuracy: 0.7948 - val_loss: 0.5468 - val_categorical_accuracy: 0.8298\n",
      "Epoch 98/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6309 - categorical_accuracy: 0.7940 - val_loss: 0.5405 - val_categorical_accuracy: 0.8316\n",
      "Epoch 99/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.6239 - categorical_accuracy: 0.7947 - val_loss: 0.5507 - val_categorical_accuracy: 0.8290\n",
      "Epoch 100/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6236 - categorical_accuracy: 0.7976 - val_loss: 0.5060 - val_categorical_accuracy: 0.8374\n",
      "--- Running training session 31\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 26s 37ms/step - loss: 1.8609 - categorical_accuracy: 0.3320 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.5426 - categorical_accuracy: 0.4538 - val_loss: 1.4505 - val_categorical_accuracy: 0.4668\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.3771 - categorical_accuracy: 0.5193 - val_loss: 1.3280 - val_categorical_accuracy: 0.5640\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.2709 - categorical_accuracy: 0.5600 - val_loss: 1.0894 - val_categorical_accuracy: 0.6232\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.1953 - categorical_accuracy: 0.5892 - val_loss: 1.3354 - val_categorical_accuracy: 0.5570\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.1453 - categorical_accuracy: 0.6046 - val_loss: 0.8888 - val_categorical_accuracy: 0.6852\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.0904 - categorical_accuracy: 0.6263 - val_loss: 0.8786 - val_categorical_accuracy: 0.6942\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.0453 - categorical_accuracy: 0.6450 - val_loss: 1.0995 - val_categorical_accuracy: 0.6388\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.0134 - categorical_accuracy: 0.6586 - val_loss: 0.8702 - val_categorical_accuracy: 0.6962\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.9665 - categorical_accuracy: 0.6727 - val_loss: 0.7650 - val_categorical_accuracy: 0.7392\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.9457 - categorical_accuracy: 0.6793 - val_loss: 0.7955 - val_categorical_accuracy: 0.7204\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.9082 - categorical_accuracy: 0.6956 - val_loss: 0.7834 - val_categorical_accuracy: 0.7278\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8861 - categorical_accuracy: 0.7031 - val_loss: 0.8479 - val_categorical_accuracy: 0.7104\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8728 - categorical_accuracy: 0.7065 - val_loss: 0.6240 - val_categorical_accuracy: 0.7816\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8365 - categorical_accuracy: 0.7207 - val_loss: 0.7630 - val_categorical_accuracy: 0.7374\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8312 - categorical_accuracy: 0.7193 - val_loss: 0.5927 - val_categorical_accuracy: 0.7972\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.7997 - categorical_accuracy: 0.7336 - val_loss: 0.6225 - val_categorical_accuracy: 0.7920\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7848 - categorical_accuracy: 0.7380 - val_loss: 0.6284 - val_categorical_accuracy: 0.7946\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.7707 - categorical_accuracy: 0.7399 - val_loss: 0.5665 - val_categorical_accuracy: 0.8064\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.7587 - categorical_accuracy: 0.7465 - val_loss: 0.5690 - val_categorical_accuracy: 0.8034\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7494 - categorical_accuracy: 0.7518 - val_loss: 0.6732 - val_categorical_accuracy: 0.7738\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7374 - categorical_accuracy: 0.7545 - val_loss: 0.6165 - val_categorical_accuracy: 0.7908\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.7208 - categorical_accuracy: 0.7618 - val_loss: 0.5851 - val_categorical_accuracy: 0.8044\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7133 - categorical_accuracy: 0.7641 - val_loss: 0.5877 - val_categorical_accuracy: 0.7978\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6964 - categorical_accuracy: 0.7691 - val_loss: 0.6646 - val_categorical_accuracy: 0.7828\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.6960 - categorical_accuracy: 0.7699 - val_loss: 0.6944 - val_categorical_accuracy: 0.7734\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6839 - categorical_accuracy: 0.7746 - val_loss: 0.7360 - val_categorical_accuracy: 0.7520\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6655 - categorical_accuracy: 0.7792 - val_loss: 0.5900 - val_categorical_accuracy: 0.8060\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6579 - categorical_accuracy: 0.7825 - val_loss: 0.6095 - val_categorical_accuracy: 0.7970\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6510 - categorical_accuracy: 0.7851 - val_loss: 0.5022 - val_categorical_accuracy: 0.8268\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6481 - categorical_accuracy: 0.7873 - val_loss: 0.5771 - val_categorical_accuracy: 0.7994\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6413 - categorical_accuracy: 0.7897 - val_loss: 0.5680 - val_categorical_accuracy: 0.8040\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6287 - categorical_accuracy: 0.7924 - val_loss: 0.5333 - val_categorical_accuracy: 0.8204\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6252 - categorical_accuracy: 0.7947 - val_loss: 0.6322 - val_categorical_accuracy: 0.7872\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6202 - categorical_accuracy: 0.7963 - val_loss: 0.5046 - val_categorical_accuracy: 0.8308\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.6083 - categorical_accuracy: 0.8003 - val_loss: 0.4779 - val_categorical_accuracy: 0.8382\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6060 - categorical_accuracy: 0.8018 - val_loss: 0.4958 - val_categorical_accuracy: 0.8368\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5965 - categorical_accuracy: 0.8031 - val_loss: 0.5299 - val_categorical_accuracy: 0.8272\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.5905 - categorical_accuracy: 0.8074 - val_loss: 0.4377 - val_categorical_accuracy: 0.8516\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5874 - categorical_accuracy: 0.8089 - val_loss: 0.5887 - val_categorical_accuracy: 0.8076\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.5850 - categorical_accuracy: 0.8102 - val_loss: 0.4402 - val_categorical_accuracy: 0.8436\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5753 - categorical_accuracy: 0.8123 - val_loss: 0.5996 - val_categorical_accuracy: 0.7990\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5677 - categorical_accuracy: 0.8130 - val_loss: 0.4814 - val_categorical_accuracy: 0.8454\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5585 - categorical_accuracy: 0.8140 - val_loss: 0.4876 - val_categorical_accuracy: 0.8330\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.5584 - categorical_accuracy: 0.8165 - val_loss: 0.4539 - val_categorical_accuracy: 0.8538\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5501 - categorical_accuracy: 0.8192 - val_loss: 0.5248 - val_categorical_accuracy: 0.8286\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5528 - categorical_accuracy: 0.8199 - val_loss: 0.4673 - val_categorical_accuracy: 0.8462\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5450 - categorical_accuracy: 0.8207 - val_loss: 0.5167 - val_categorical_accuracy: 0.8288\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5423 - categorical_accuracy: 0.8231 - val_loss: 0.6229 - val_categorical_accuracy: 0.8118\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5418 - categorical_accuracy: 0.8227 - val_loss: 0.4465 - val_categorical_accuracy: 0.8492\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5332 - categorical_accuracy: 0.8265 - val_loss: 0.4340 - val_categorical_accuracy: 0.8536\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5314 - categorical_accuracy: 0.8280 - val_loss: 0.5078 - val_categorical_accuracy: 0.8340\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5216 - categorical_accuracy: 0.8295 - val_loss: 0.5222 - val_categorical_accuracy: 0.8320\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5183 - categorical_accuracy: 0.8300 - val_loss: 0.4641 - val_categorical_accuracy: 0.8434\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5204 - categorical_accuracy: 0.8310 - val_loss: 0.5285 - val_categorical_accuracy: 0.8258\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5137 - categorical_accuracy: 0.8312 - val_loss: 0.4830 - val_categorical_accuracy: 0.8394\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5174 - categorical_accuracy: 0.8310 - val_loss: 0.4921 - val_categorical_accuracy: 0.8360\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5068 - categorical_accuracy: 0.8351 - val_loss: 0.5425 - val_categorical_accuracy: 0.8218\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4999 - categorical_accuracy: 0.8361 - val_loss: 0.5171 - val_categorical_accuracy: 0.8338\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5016 - categorical_accuracy: 0.8357 - val_loss: 0.5348 - val_categorical_accuracy: 0.8326\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4960 - categorical_accuracy: 0.8378 - val_loss: 0.6452 - val_categorical_accuracy: 0.8066\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4916 - categorical_accuracy: 0.8397 - val_loss: 0.5873 - val_categorical_accuracy: 0.8070\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4979 - categorical_accuracy: 0.8378 - val_loss: 0.4631 - val_categorical_accuracy: 0.8470\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4895 - categorical_accuracy: 0.8413 - val_loss: 0.5279 - val_categorical_accuracy: 0.8222\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4874 - categorical_accuracy: 0.8407 - val_loss: 0.4728 - val_categorical_accuracy: 0.8424\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4731 - categorical_accuracy: 0.8465 - val_loss: 0.5550 - val_categorical_accuracy: 0.8266\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4783 - categorical_accuracy: 0.8447 - val_loss: 0.4944 - val_categorical_accuracy: 0.8380\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4791 - categorical_accuracy: 0.8430 - val_loss: 0.5241 - val_categorical_accuracy: 0.8282\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4781 - categorical_accuracy: 0.8440 - val_loss: 0.5122 - val_categorical_accuracy: 0.8350\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4710 - categorical_accuracy: 0.8461 - val_loss: 0.5324 - val_categorical_accuracy: 0.8280\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4716 - categorical_accuracy: 0.8473 - val_loss: 0.4164 - val_categorical_accuracy: 0.8624\n",
      "Epoch 72/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4666 - categorical_accuracy: 0.8488 - val_loss: 0.5479 - val_categorical_accuracy: 0.8328\n",
      "Epoch 73/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4583 - categorical_accuracy: 0.8508 - val_loss: 0.5068 - val_categorical_accuracy: 0.8340\n",
      "Epoch 74/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4591 - categorical_accuracy: 0.8505 - val_loss: 0.4244 - val_categorical_accuracy: 0.8616\n",
      "Epoch 75/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4623 - categorical_accuracy: 0.8499 - val_loss: 0.6041 - val_categorical_accuracy: 0.8192\n",
      "Epoch 76/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4589 - categorical_accuracy: 0.8503 - val_loss: 0.4364 - val_categorical_accuracy: 0.8582\n",
      "Epoch 77/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4564 - categorical_accuracy: 0.8519 - val_loss: 0.5317 - val_categorical_accuracy: 0.8352\n",
      "Epoch 78/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4581 - categorical_accuracy: 0.8519 - val_loss: 0.7502 - val_categorical_accuracy: 0.7740\n",
      "Epoch 79/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4565 - categorical_accuracy: 0.8519 - val_loss: 0.5483 - val_categorical_accuracy: 0.8232\n",
      "Epoch 80/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4501 - categorical_accuracy: 0.8560 - val_loss: 0.5710 - val_categorical_accuracy: 0.8232\n",
      "Epoch 81/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.4425 - categorical_accuracy: 0.8577 - val_loss: 0.4896 - val_categorical_accuracy: 0.8504\n",
      "Epoch 82/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4479 - categorical_accuracy: 0.8562 - val_loss: 0.4479 - val_categorical_accuracy: 0.8554\n",
      "Epoch 83/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4358 - categorical_accuracy: 0.8590 - val_loss: 0.6914 - val_categorical_accuracy: 0.8074\n",
      "Epoch 84/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4351 - categorical_accuracy: 0.8597 - val_loss: 0.4319 - val_categorical_accuracy: 0.8622\n",
      "Epoch 85/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4394 - categorical_accuracy: 0.8582 - val_loss: 0.5197 - val_categorical_accuracy: 0.8422\n",
      "Epoch 86/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4363 - categorical_accuracy: 0.8563 - val_loss: 0.5228 - val_categorical_accuracy: 0.8346\n",
      "Epoch 87/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4364 - categorical_accuracy: 0.8571 - val_loss: 0.6631 - val_categorical_accuracy: 0.8038\n",
      "Epoch 88/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.4287 - categorical_accuracy: 0.8605 - val_loss: 0.4959 - val_categorical_accuracy: 0.8500\n",
      "Epoch 89/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4307 - categorical_accuracy: 0.8610 - val_loss: 0.5433 - val_categorical_accuracy: 0.8378\n",
      "Epoch 90/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4276 - categorical_accuracy: 0.8628 - val_loss: 0.4896 - val_categorical_accuracy: 0.8484\n",
      "Epoch 91/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4318 - categorical_accuracy: 0.8618 - val_loss: 0.4539 - val_categorical_accuracy: 0.8514\n",
      "Epoch 92/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4284 - categorical_accuracy: 0.8612 - val_loss: 0.8613 - val_categorical_accuracy: 0.7764\n",
      "Epoch 93/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4227 - categorical_accuracy: 0.8637 - val_loss: 0.4453 - val_categorical_accuracy: 0.8618\n",
      "Epoch 94/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4178 - categorical_accuracy: 0.8646 - val_loss: 0.4978 - val_categorical_accuracy: 0.8426\n",
      "Epoch 95/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4154 - categorical_accuracy: 0.8665 - val_loss: 0.5681 - val_categorical_accuracy: 0.8228\n",
      "Epoch 96/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4204 - categorical_accuracy: 0.8658 - val_loss: 0.5412 - val_categorical_accuracy: 0.8410\n",
      "Epoch 97/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4113 - categorical_accuracy: 0.8681 - val_loss: 0.6170 - val_categorical_accuracy: 0.8172\n",
      "Epoch 98/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4086 - categorical_accuracy: 0.8677 - val_loss: 0.5267 - val_categorical_accuracy: 0.8344\n",
      "Epoch 99/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4087 - categorical_accuracy: 0.8666 - val_loss: 0.4814 - val_categorical_accuracy: 0.8554\n",
      "Epoch 100/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4147 - categorical_accuracy: 0.8647 - val_loss: 0.4810 - val_categorical_accuracy: 0.8510\n",
      "--- Running training session 32\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 27s 39ms/step - loss: 1.8971 - categorical_accuracy: 0.3045 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.6765 - categorical_accuracy: 0.3920 - val_loss: 1.4867 - val_categorical_accuracy: 0.4490\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.5359 - categorical_accuracy: 0.4528 - val_loss: 1.3533 - val_categorical_accuracy: 0.5048\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.4314 - categorical_accuracy: 0.4935 - val_loss: 1.3316 - val_categorical_accuracy: 0.5394\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.3620 - categorical_accuracy: 0.5238 - val_loss: 1.1542 - val_categorical_accuracy: 0.5834\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.2829 - categorical_accuracy: 0.5534 - val_loss: 1.0534 - val_categorical_accuracy: 0.6182\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.2269 - categorical_accuracy: 0.5734 - val_loss: 1.0752 - val_categorical_accuracy: 0.6088\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.1814 - categorical_accuracy: 0.5931 - val_loss: 0.9826 - val_categorical_accuracy: 0.6604\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.1293 - categorical_accuracy: 0.6103 - val_loss: 1.0049 - val_categorical_accuracy: 0.6576\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.0939 - categorical_accuracy: 0.6268 - val_loss: 0.8962 - val_categorical_accuracy: 0.6904\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.0676 - categorical_accuracy: 0.6383 - val_loss: 0.7811 - val_categorical_accuracy: 0.7216\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.0266 - categorical_accuracy: 0.6518 - val_loss: 0.8065 - val_categorical_accuracy: 0.7194\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.9876 - categorical_accuracy: 0.6646 - val_loss: 1.0023 - val_categorical_accuracy: 0.6738\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.9579 - categorical_accuracy: 0.6767 - val_loss: 0.7833 - val_categorical_accuracy: 0.7338\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.9344 - categorical_accuracy: 0.6845 - val_loss: 0.7691 - val_categorical_accuracy: 0.7446\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.9136 - categorical_accuracy: 0.6914 - val_loss: 0.6174 - val_categorical_accuracy: 0.7860\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8943 - categorical_accuracy: 0.6983 - val_loss: 0.6510 - val_categorical_accuracy: 0.7764\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.8811 - categorical_accuracy: 0.7030 - val_loss: 0.7564 - val_categorical_accuracy: 0.7424\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8542 - categorical_accuracy: 0.7160 - val_loss: 0.8445 - val_categorical_accuracy: 0.7144\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.8452 - categorical_accuracy: 0.7189 - val_loss: 0.5755 - val_categorical_accuracy: 0.8016\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.8234 - categorical_accuracy: 0.7258 - val_loss: 0.6908 - val_categorical_accuracy: 0.7652\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.8148 - categorical_accuracy: 0.7292 - val_loss: 0.6073 - val_categorical_accuracy: 0.7930\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8001 - categorical_accuracy: 0.7335 - val_loss: 0.6368 - val_categorical_accuracy: 0.7782\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7887 - categorical_accuracy: 0.7361 - val_loss: 0.7234 - val_categorical_accuracy: 0.7564\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.7757 - categorical_accuracy: 0.7439 - val_loss: 0.6279 - val_categorical_accuracy: 0.7868\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.7645 - categorical_accuracy: 0.7459 - val_loss: 0.6659 - val_categorical_accuracy: 0.7762\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7507 - categorical_accuracy: 0.7492 - val_loss: 0.5285 - val_categorical_accuracy: 0.8260\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7346 - categorical_accuracy: 0.7558 - val_loss: 0.5756 - val_categorical_accuracy: 0.8040\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.7387 - categorical_accuracy: 0.7555 - val_loss: 0.5968 - val_categorical_accuracy: 0.7992\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7280 - categorical_accuracy: 0.7595 - val_loss: 0.5713 - val_categorical_accuracy: 0.8090\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7249 - categorical_accuracy: 0.7589 - val_loss: 0.5447 - val_categorical_accuracy: 0.8146\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7118 - categorical_accuracy: 0.7662 - val_loss: 0.6005 - val_categorical_accuracy: 0.8064\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.7008 - categorical_accuracy: 0.7688 - val_loss: 0.5459 - val_categorical_accuracy: 0.8186\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6939 - categorical_accuracy: 0.7679 - val_loss: 0.5334 - val_categorical_accuracy: 0.8172\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6865 - categorical_accuracy: 0.7708 - val_loss: 0.6755 - val_categorical_accuracy: 0.7808\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6815 - categorical_accuracy: 0.7762 - val_loss: 0.5124 - val_categorical_accuracy: 0.8296\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.6743 - categorical_accuracy: 0.7782 - val_loss: 0.5739 - val_categorical_accuracy: 0.8114\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6681 - categorical_accuracy: 0.7776 - val_loss: 0.6712 - val_categorical_accuracy: 0.7898\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6602 - categorical_accuracy: 0.7803 - val_loss: 0.5400 - val_categorical_accuracy: 0.8210\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.6514 - categorical_accuracy: 0.7831 - val_loss: 0.5142 - val_categorical_accuracy: 0.8372\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6423 - categorical_accuracy: 0.7879 - val_loss: 0.4942 - val_categorical_accuracy: 0.8338\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6454 - categorical_accuracy: 0.7850 - val_loss: 0.5396 - val_categorical_accuracy: 0.8178\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6333 - categorical_accuracy: 0.7901 - val_loss: 0.5692 - val_categorical_accuracy: 0.8158\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6366 - categorical_accuracy: 0.7906 - val_loss: 0.5075 - val_categorical_accuracy: 0.8350\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6265 - categorical_accuracy: 0.7940 - val_loss: 0.6967 - val_categorical_accuracy: 0.7758\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6256 - categorical_accuracy: 0.7948 - val_loss: 0.5432 - val_categorical_accuracy: 0.8164\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6140 - categorical_accuracy: 0.7980 - val_loss: 0.5983 - val_categorical_accuracy: 0.8116\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6002 - categorical_accuracy: 0.8021 - val_loss: 0.5530 - val_categorical_accuracy: 0.8208\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6053 - categorical_accuracy: 0.7997 - val_loss: 0.6104 - val_categorical_accuracy: 0.8128\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6036 - categorical_accuracy: 0.7994 - val_loss: 0.5861 - val_categorical_accuracy: 0.8096\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5946 - categorical_accuracy: 0.8040 - val_loss: 0.4525 - val_categorical_accuracy: 0.8466\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5958 - categorical_accuracy: 0.8040 - val_loss: 0.6202 - val_categorical_accuracy: 0.8058\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5872 - categorical_accuracy: 0.8070 - val_loss: 0.5231 - val_categorical_accuracy: 0.8312\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5852 - categorical_accuracy: 0.8085 - val_loss: 0.4606 - val_categorical_accuracy: 0.8440\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5755 - categorical_accuracy: 0.8117 - val_loss: 0.4636 - val_categorical_accuracy: 0.8460\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5761 - categorical_accuracy: 0.8122 - val_loss: 0.4881 - val_categorical_accuracy: 0.8416\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.5713 - categorical_accuracy: 0.8134 - val_loss: 0.4905 - val_categorical_accuracy: 0.8370\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5696 - categorical_accuracy: 0.8134 - val_loss: 0.7668 - val_categorical_accuracy: 0.7588\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5706 - categorical_accuracy: 0.8141 - val_loss: 0.6634 - val_categorical_accuracy: 0.7936\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5601 - categorical_accuracy: 0.8167 - val_loss: 0.5236 - val_categorical_accuracy: 0.8286\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5686 - categorical_accuracy: 0.8153 - val_loss: 0.4574 - val_categorical_accuracy: 0.8484\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5561 - categorical_accuracy: 0.8159 - val_loss: 0.5688 - val_categorical_accuracy: 0.8192\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5572 - categorical_accuracy: 0.8175 - val_loss: 0.4694 - val_categorical_accuracy: 0.8390\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5524 - categorical_accuracy: 0.8176 - val_loss: 0.5007 - val_categorical_accuracy: 0.8290\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5563 - categorical_accuracy: 0.8181 - val_loss: 0.5152 - val_categorical_accuracy: 0.8382\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5444 - categorical_accuracy: 0.8210 - val_loss: 0.5094 - val_categorical_accuracy: 0.8362\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5374 - categorical_accuracy: 0.8253 - val_loss: 0.5283 - val_categorical_accuracy: 0.8352\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5412 - categorical_accuracy: 0.8234 - val_loss: 0.4808 - val_categorical_accuracy: 0.8426\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5341 - categorical_accuracy: 0.8250 - val_loss: 0.4701 - val_categorical_accuracy: 0.8436\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5332 - categorical_accuracy: 0.8270 - val_loss: 0.5459 - val_categorical_accuracy: 0.8256\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5256 - categorical_accuracy: 0.8275 - val_loss: 0.4829 - val_categorical_accuracy: 0.8460\n",
      "Epoch 72/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.5267 - categorical_accuracy: 0.8256 - val_loss: 0.4455 - val_categorical_accuracy: 0.8540\n",
      "Epoch 73/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5226 - categorical_accuracy: 0.8304 - val_loss: 0.5855 - val_categorical_accuracy: 0.8176\n",
      "Epoch 74/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5203 - categorical_accuracy: 0.8307 - val_loss: 0.7197 - val_categorical_accuracy: 0.7942\n",
      "Epoch 75/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5183 - categorical_accuracy: 0.8311 - val_loss: 0.6227 - val_categorical_accuracy: 0.8104\n",
      "Epoch 76/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5200 - categorical_accuracy: 0.8285 - val_loss: 0.5169 - val_categorical_accuracy: 0.8390\n",
      "Epoch 77/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5147 - categorical_accuracy: 0.8314 - val_loss: 0.4882 - val_categorical_accuracy: 0.8396\n",
      "Epoch 78/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5085 - categorical_accuracy: 0.8342 - val_loss: 0.5299 - val_categorical_accuracy: 0.8392\n",
      "Epoch 79/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5117 - categorical_accuracy: 0.8312 - val_loss: 0.4550 - val_categorical_accuracy: 0.8528\n",
      "Epoch 80/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5018 - categorical_accuracy: 0.8365 - val_loss: 0.5284 - val_categorical_accuracy: 0.8292\n",
      "Epoch 81/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5024 - categorical_accuracy: 0.8366 - val_loss: 0.5479 - val_categorical_accuracy: 0.8294\n",
      "Epoch 82/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5049 - categorical_accuracy: 0.8359 - val_loss: 0.5286 - val_categorical_accuracy: 0.8328\n",
      "Epoch 83/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5007 - categorical_accuracy: 0.8354 - val_loss: 0.5597 - val_categorical_accuracy: 0.8294\n",
      "Epoch 84/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4992 - categorical_accuracy: 0.8361 - val_loss: 0.4640 - val_categorical_accuracy: 0.8528\n",
      "Epoch 85/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4956 - categorical_accuracy: 0.8364 - val_loss: 0.6539 - val_categorical_accuracy: 0.8114\n",
      "Epoch 86/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4936 - categorical_accuracy: 0.8386 - val_loss: 0.5436 - val_categorical_accuracy: 0.8324\n",
      "Epoch 87/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4939 - categorical_accuracy: 0.8386 - val_loss: 0.4502 - val_categorical_accuracy: 0.8576\n",
      "Epoch 88/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4844 - categorical_accuracy: 0.8419 - val_loss: 0.4628 - val_categorical_accuracy: 0.8544\n",
      "Epoch 89/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4847 - categorical_accuracy: 0.8418 - val_loss: 0.4404 - val_categorical_accuracy: 0.8568\n",
      "Epoch 90/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4835 - categorical_accuracy: 0.8433 - val_loss: 0.5123 - val_categorical_accuracy: 0.8358\n",
      "Epoch 91/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4807 - categorical_accuracy: 0.8444 - val_loss: 0.5475 - val_categorical_accuracy: 0.8376\n",
      "Epoch 92/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4748 - categorical_accuracy: 0.8448 - val_loss: 0.5409 - val_categorical_accuracy: 0.8356\n",
      "Epoch 93/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4802 - categorical_accuracy: 0.8443 - val_loss: 0.4833 - val_categorical_accuracy: 0.8496\n",
      "Epoch 94/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4743 - categorical_accuracy: 0.8456 - val_loss: 0.5038 - val_categorical_accuracy: 0.8378\n",
      "Epoch 95/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4707 - categorical_accuracy: 0.8474 - val_loss: 0.6418 - val_categorical_accuracy: 0.8078\n",
      "Epoch 96/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4647 - categorical_accuracy: 0.8476 - val_loss: 0.5383 - val_categorical_accuracy: 0.8338\n",
      "Epoch 97/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4694 - categorical_accuracy: 0.8471 - val_loss: 0.4645 - val_categorical_accuracy: 0.8518\n",
      "Epoch 98/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4662 - categorical_accuracy: 0.8506 - val_loss: 0.4864 - val_categorical_accuracy: 0.8482\n",
      "Epoch 99/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4625 - categorical_accuracy: 0.8488 - val_loss: 0.5702 - val_categorical_accuracy: 0.8354\n",
      "Epoch 100/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4585 - categorical_accuracy: 0.8511 - val_loss: 0.4650 - val_categorical_accuracy: 0.8572\n",
      "--- Running training session 33\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 27s 38ms/step - loss: 1.9988 - categorical_accuracy: 0.2548 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.8236 - categorical_accuracy: 0.3230 - val_loss: 1.7432 - val_categorical_accuracy: 0.3714\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.7089 - categorical_accuracy: 0.3710 - val_loss: 1.4775 - val_categorical_accuracy: 0.4650\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.6229 - categorical_accuracy: 0.4066 - val_loss: 1.3463 - val_categorical_accuracy: 0.5078\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.5538 - categorical_accuracy: 0.4425 - val_loss: 1.4144 - val_categorical_accuracy: 0.4782\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.4855 - categorical_accuracy: 0.4696 - val_loss: 1.2689 - val_categorical_accuracy: 0.5488\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 1.4214 - categorical_accuracy: 0.4979 - val_loss: 1.2730 - val_categorical_accuracy: 0.5356\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.3772 - categorical_accuracy: 0.5116 - val_loss: 1.1372 - val_categorical_accuracy: 0.5822\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.3398 - categorical_accuracy: 0.5291 - val_loss: 1.1978 - val_categorical_accuracy: 0.5744\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.3018 - categorical_accuracy: 0.5429 - val_loss: 1.1813 - val_categorical_accuracy: 0.6032\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.2633 - categorical_accuracy: 0.5576 - val_loss: 0.9415 - val_categorical_accuracy: 0.6578\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.2325 - categorical_accuracy: 0.5711 - val_loss: 0.9062 - val_categorical_accuracy: 0.6754\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.2065 - categorical_accuracy: 0.5804 - val_loss: 0.9280 - val_categorical_accuracy: 0.6796\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.1850 - categorical_accuracy: 0.5896 - val_loss: 1.0603 - val_categorical_accuracy: 0.6252\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.1590 - categorical_accuracy: 0.6018 - val_loss: 0.8187 - val_categorical_accuracy: 0.7194\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.1337 - categorical_accuracy: 0.6137 - val_loss: 0.7947 - val_categorical_accuracy: 0.7160\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 1.1137 - categorical_accuracy: 0.6208 - val_loss: 0.8340 - val_categorical_accuracy: 0.7188\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.0983 - categorical_accuracy: 0.6245 - val_loss: 0.8336 - val_categorical_accuracy: 0.7170\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 1.0871 - categorical_accuracy: 0.6307 - val_loss: 0.6971 - val_categorical_accuracy: 0.7566\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.0621 - categorical_accuracy: 0.6394 - val_loss: 0.7951 - val_categorical_accuracy: 0.7274\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.0483 - categorical_accuracy: 0.6444 - val_loss: 0.8066 - val_categorical_accuracy: 0.7284\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.0305 - categorical_accuracy: 0.6505 - val_loss: 0.8726 - val_categorical_accuracy: 0.7070\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.0216 - categorical_accuracy: 0.6534 - val_loss: 0.7902 - val_categorical_accuracy: 0.7362\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.0029 - categorical_accuracy: 0.6596 - val_loss: 0.7412 - val_categorical_accuracy: 0.7454\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.9866 - categorical_accuracy: 0.6670 - val_loss: 0.6993 - val_categorical_accuracy: 0.7642\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 19s 28ms/step - loss: 0.9708 - categorical_accuracy: 0.6714 - val_loss: 0.8013 - val_categorical_accuracy: 0.7268\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.9575 - categorical_accuracy: 0.6781 - val_loss: 0.7279 - val_categorical_accuracy: 0.7560\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.9533 - categorical_accuracy: 0.6807 - val_loss: 0.6947 - val_categorical_accuracy: 0.7666\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.9518 - categorical_accuracy: 0.6790 - val_loss: 0.6316 - val_categorical_accuracy: 0.7942\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.9387 - categorical_accuracy: 0.6834 - val_loss: 0.6468 - val_categorical_accuracy: 0.7874\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.9210 - categorical_accuracy: 0.6919 - val_loss: 0.7456 - val_categorical_accuracy: 0.7510\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.9194 - categorical_accuracy: 0.6952 - val_loss: 0.6255 - val_categorical_accuracy: 0.7914\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.9126 - categorical_accuracy: 0.6948 - val_loss: 0.6687 - val_categorical_accuracy: 0.7832\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.9039 - categorical_accuracy: 0.6979 - val_loss: 0.6185 - val_categorical_accuracy: 0.7882\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8970 - categorical_accuracy: 0.7044 - val_loss: 0.6047 - val_categorical_accuracy: 0.8000\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8802 - categorical_accuracy: 0.7058 - val_loss: 0.6993 - val_categorical_accuracy: 0.7730\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8773 - categorical_accuracy: 0.7079 - val_loss: 0.6717 - val_categorical_accuracy: 0.7814\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8653 - categorical_accuracy: 0.7097 - val_loss: 0.5953 - val_categorical_accuracy: 0.8056\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8590 - categorical_accuracy: 0.7131 - val_loss: 0.6549 - val_categorical_accuracy: 0.7794\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8563 - categorical_accuracy: 0.7130 - val_loss: 0.6351 - val_categorical_accuracy: 0.7874\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8547 - categorical_accuracy: 0.7152 - val_loss: 0.6521 - val_categorical_accuracy: 0.7828\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8412 - categorical_accuracy: 0.7195 - val_loss: 0.6204 - val_categorical_accuracy: 0.8042\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8471 - categorical_accuracy: 0.7168 - val_loss: 0.6938 - val_categorical_accuracy: 0.7738\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8317 - categorical_accuracy: 0.7231 - val_loss: 0.6946 - val_categorical_accuracy: 0.7670\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8333 - categorical_accuracy: 0.7217 - val_loss: 0.6522 - val_categorical_accuracy: 0.7828\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8168 - categorical_accuracy: 0.7293 - val_loss: 0.6588 - val_categorical_accuracy: 0.7824\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8161 - categorical_accuracy: 0.7303 - val_loss: 0.6121 - val_categorical_accuracy: 0.7964\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8185 - categorical_accuracy: 0.7269 - val_loss: 0.5797 - val_categorical_accuracy: 0.8146\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8040 - categorical_accuracy: 0.7306 - val_loss: 0.6061 - val_categorical_accuracy: 0.8068\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8060 - categorical_accuracy: 0.7361 - val_loss: 0.6542 - val_categorical_accuracy: 0.7872\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7868 - categorical_accuracy: 0.7396 - val_loss: 0.6195 - val_categorical_accuracy: 0.7970\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8079 - categorical_accuracy: 0.7326 - val_loss: 0.6044 - val_categorical_accuracy: 0.8022\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7876 - categorical_accuracy: 0.7378 - val_loss: 0.5719 - val_categorical_accuracy: 0.8106\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7692 - categorical_accuracy: 0.7437 - val_loss: 0.5232 - val_categorical_accuracy: 0.8244\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.7828 - categorical_accuracy: 0.7411 - val_loss: 0.5705 - val_categorical_accuracy: 0.8176\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7717 - categorical_accuracy: 0.7471 - val_loss: 0.6065 - val_categorical_accuracy: 0.8062\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7736 - categorical_accuracy: 0.7422 - val_loss: 0.5802 - val_categorical_accuracy: 0.8168\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.7725 - categorical_accuracy: 0.7435 - val_loss: 0.6117 - val_categorical_accuracy: 0.8072\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7629 - categorical_accuracy: 0.7460 - val_loss: 0.5062 - val_categorical_accuracy: 0.8340\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7527 - categorical_accuracy: 0.7516 - val_loss: 0.6153 - val_categorical_accuracy: 0.7952\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.7405 - categorical_accuracy: 0.7542 - val_loss: 0.5983 - val_categorical_accuracy: 0.8078\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7519 - categorical_accuracy: 0.7527 - val_loss: 0.5137 - val_categorical_accuracy: 0.8316\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7557 - categorical_accuracy: 0.7500 - val_loss: 0.5921 - val_categorical_accuracy: 0.8092\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.7419 - categorical_accuracy: 0.7549 - val_loss: 0.5762 - val_categorical_accuracy: 0.8176\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.7390 - categorical_accuracy: 0.7537 - val_loss: 0.5421 - val_categorical_accuracy: 0.8204\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.7333 - categorical_accuracy: 0.7595 - val_loss: 0.5290 - val_categorical_accuracy: 0.8306\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.7310 - categorical_accuracy: 0.7580 - val_loss: 0.5063 - val_categorical_accuracy: 0.8312\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7164 - categorical_accuracy: 0.7648 - val_loss: 0.5273 - val_categorical_accuracy: 0.8266\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7251 - categorical_accuracy: 0.7608 - val_loss: 0.5539 - val_categorical_accuracy: 0.8228\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7288 - categorical_accuracy: 0.7610 - val_loss: 0.6419 - val_categorical_accuracy: 0.7898\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7200 - categorical_accuracy: 0.7599 - val_loss: 0.5093 - val_categorical_accuracy: 0.8442\n",
      "Epoch 72/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.7241 - categorical_accuracy: 0.7618 - val_loss: 0.5653 - val_categorical_accuracy: 0.8112\n",
      "Epoch 73/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7106 - categorical_accuracy: 0.7676 - val_loss: 0.6083 - val_categorical_accuracy: 0.8120\n",
      "Epoch 74/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7079 - categorical_accuracy: 0.7668 - val_loss: 0.4706 - val_categorical_accuracy: 0.8454\n",
      "Epoch 75/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7199 - categorical_accuracy: 0.7632 - val_loss: 0.5577 - val_categorical_accuracy: 0.8170\n",
      "Epoch 76/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6992 - categorical_accuracy: 0.7669 - val_loss: 0.5791 - val_categorical_accuracy: 0.8250\n",
      "Epoch 77/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7016 - categorical_accuracy: 0.7708 - val_loss: 0.6720 - val_categorical_accuracy: 0.7812\n",
      "Epoch 78/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7066 - categorical_accuracy: 0.7676 - val_loss: 0.5592 - val_categorical_accuracy: 0.8174\n",
      "Epoch 79/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7048 - categorical_accuracy: 0.7683 - val_loss: 0.5425 - val_categorical_accuracy: 0.8414\n",
      "Epoch 80/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.7014 - categorical_accuracy: 0.7716 - val_loss: 0.5664 - val_categorical_accuracy: 0.8180\n",
      "Epoch 81/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6926 - categorical_accuracy: 0.7721 - val_loss: 0.5109 - val_categorical_accuracy: 0.8346\n",
      "Epoch 82/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6930 - categorical_accuracy: 0.7714 - val_loss: 0.5561 - val_categorical_accuracy: 0.8228\n",
      "Epoch 83/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6844 - categorical_accuracy: 0.7750 - val_loss: 0.5221 - val_categorical_accuracy: 0.8376\n",
      "Epoch 84/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.6923 - categorical_accuracy: 0.7738 - val_loss: 0.5539 - val_categorical_accuracy: 0.8200\n",
      "Epoch 85/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6919 - categorical_accuracy: 0.7718 - val_loss: 0.5270 - val_categorical_accuracy: 0.8344\n",
      "Epoch 86/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6782 - categorical_accuracy: 0.7770 - val_loss: 0.6119 - val_categorical_accuracy: 0.8118\n",
      "Epoch 87/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.6806 - categorical_accuracy: 0.7749 - val_loss: 0.6692 - val_categorical_accuracy: 0.7878\n",
      "Epoch 88/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6799 - categorical_accuracy: 0.7779 - val_loss: 0.5265 - val_categorical_accuracy: 0.8318\n",
      "Epoch 89/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6803 - categorical_accuracy: 0.7784 - val_loss: 0.5532 - val_categorical_accuracy: 0.8208\n",
      "Epoch 90/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6739 - categorical_accuracy: 0.7781 - val_loss: 0.4654 - val_categorical_accuracy: 0.8534\n",
      "Epoch 91/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6781 - categorical_accuracy: 0.7802 - val_loss: 0.5587 - val_categorical_accuracy: 0.8200\n",
      "Epoch 92/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6683 - categorical_accuracy: 0.7810 - val_loss: 0.5210 - val_categorical_accuracy: 0.8370\n",
      "Epoch 93/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6653 - categorical_accuracy: 0.7808 - val_loss: 0.4986 - val_categorical_accuracy: 0.8442\n",
      "Epoch 94/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6681 - categorical_accuracy: 0.7816 - val_loss: 0.5498 - val_categorical_accuracy: 0.8284\n",
      "Epoch 95/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6535 - categorical_accuracy: 0.7873 - val_loss: 0.5406 - val_categorical_accuracy: 0.8342\n",
      "Epoch 96/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6531 - categorical_accuracy: 0.7862 - val_loss: 0.6923 - val_categorical_accuracy: 0.7820\n",
      "Epoch 97/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6577 - categorical_accuracy: 0.7865 - val_loss: 0.5254 - val_categorical_accuracy: 0.8342\n",
      "Epoch 98/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6510 - categorical_accuracy: 0.7841 - val_loss: 0.5393 - val_categorical_accuracy: 0.8348\n",
      "Epoch 99/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.6628 - categorical_accuracy: 0.7826 - val_loss: 0.5645 - val_categorical_accuracy: 0.8212\n",
      "Epoch 100/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6573 - categorical_accuracy: 0.7852 - val_loss: 0.5470 - val_categorical_accuracy: 0.8258\n",
      "--- Running training session 34\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 28s 39ms/step - loss: 1.8398 - categorical_accuracy: 0.3398 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.4818 - categorical_accuracy: 0.4785 - val_loss: 1.1987 - val_categorical_accuracy: 0.5704\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.3349 - categorical_accuracy: 0.5334 - val_loss: 1.0669 - val_categorical_accuracy: 0.6204\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.2480 - categorical_accuracy: 0.5689 - val_loss: 1.0838 - val_categorical_accuracy: 0.6236\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.1759 - categorical_accuracy: 0.5941 - val_loss: 0.8929 - val_categorical_accuracy: 0.6818\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.1244 - categorical_accuracy: 0.6126 - val_loss: 0.9529 - val_categorical_accuracy: 0.6672\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 1.0749 - categorical_accuracy: 0.6332 - val_loss: 0.8882 - val_categorical_accuracy: 0.6826\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.0311 - categorical_accuracy: 0.6478 - val_loss: 0.7459 - val_categorical_accuracy: 0.7326\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.9972 - categorical_accuracy: 0.6614 - val_loss: 0.7489 - val_categorical_accuracy: 0.7346\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.9581 - categorical_accuracy: 0.6736 - val_loss: 0.7126 - val_categorical_accuracy: 0.7556\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.9378 - categorical_accuracy: 0.6809 - val_loss: 0.6399 - val_categorical_accuracy: 0.7772\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.9013 - categorical_accuracy: 0.6950 - val_loss: 0.6486 - val_categorical_accuracy: 0.7816\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8813 - categorical_accuracy: 0.7034 - val_loss: 0.7248 - val_categorical_accuracy: 0.7530\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8633 - categorical_accuracy: 0.7091 - val_loss: 0.7004 - val_categorical_accuracy: 0.7592\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8354 - categorical_accuracy: 0.7189 - val_loss: 0.6223 - val_categorical_accuracy: 0.7904\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8181 - categorical_accuracy: 0.7254 - val_loss: 0.5908 - val_categorical_accuracy: 0.7968\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8038 - categorical_accuracy: 0.7312 - val_loss: 0.6011 - val_categorical_accuracy: 0.7908\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.7871 - categorical_accuracy: 0.7357 - val_loss: 0.5530 - val_categorical_accuracy: 0.8088\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7767 - categorical_accuracy: 0.7414 - val_loss: 0.5330 - val_categorical_accuracy: 0.8206\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.7540 - categorical_accuracy: 0.7499 - val_loss: 0.5244 - val_categorical_accuracy: 0.8256\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7411 - categorical_accuracy: 0.7560 - val_loss: 0.6459 - val_categorical_accuracy: 0.7878\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7337 - categorical_accuracy: 0.7561 - val_loss: 0.5458 - val_categorical_accuracy: 0.8140\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7234 - categorical_accuracy: 0.7600 - val_loss: 0.5655 - val_categorical_accuracy: 0.8002\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7072 - categorical_accuracy: 0.7658 - val_loss: 0.5418 - val_categorical_accuracy: 0.8166\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7006 - categorical_accuracy: 0.7694 - val_loss: 0.4984 - val_categorical_accuracy: 0.8354\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6813 - categorical_accuracy: 0.7731 - val_loss: 0.5664 - val_categorical_accuracy: 0.8048\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6742 - categorical_accuracy: 0.7768 - val_loss: 0.6289 - val_categorical_accuracy: 0.7926\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6662 - categorical_accuracy: 0.7768 - val_loss: 0.5408 - val_categorical_accuracy: 0.8178\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.6597 - categorical_accuracy: 0.7801 - val_loss: 0.5804 - val_categorical_accuracy: 0.8074\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6588 - categorical_accuracy: 0.7820 - val_loss: 0.5246 - val_categorical_accuracy: 0.8238\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6481 - categorical_accuracy: 0.7882 - val_loss: 0.5209 - val_categorical_accuracy: 0.8254\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6376 - categorical_accuracy: 0.7889 - val_loss: 0.4853 - val_categorical_accuracy: 0.8348\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6269 - categorical_accuracy: 0.7921 - val_loss: 0.4741 - val_categorical_accuracy: 0.8438\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6240 - categorical_accuracy: 0.7940 - val_loss: 0.4750 - val_categorical_accuracy: 0.8420\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6149 - categorical_accuracy: 0.7980 - val_loss: 0.7266 - val_categorical_accuracy: 0.7682\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6107 - categorical_accuracy: 0.7972 - val_loss: 0.5842 - val_categorical_accuracy: 0.8036\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6030 - categorical_accuracy: 0.8008 - val_loss: 0.5008 - val_categorical_accuracy: 0.8308\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5960 - categorical_accuracy: 0.8028 - val_loss: 0.4660 - val_categorical_accuracy: 0.8468\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5903 - categorical_accuracy: 0.8062 - val_loss: 0.5850 - val_categorical_accuracy: 0.8082\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5864 - categorical_accuracy: 0.8072 - val_loss: 0.5330 - val_categorical_accuracy: 0.8240\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5834 - categorical_accuracy: 0.8109 - val_loss: 0.5112 - val_categorical_accuracy: 0.8322\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5801 - categorical_accuracy: 0.8092 - val_loss: 0.4847 - val_categorical_accuracy: 0.8430\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5754 - categorical_accuracy: 0.8119 - val_loss: 0.4917 - val_categorical_accuracy: 0.8370\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5715 - categorical_accuracy: 0.8126 - val_loss: 0.5502 - val_categorical_accuracy: 0.8188\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5630 - categorical_accuracy: 0.8127 - val_loss: 0.5193 - val_categorical_accuracy: 0.8320\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5556 - categorical_accuracy: 0.8155 - val_loss: 0.5346 - val_categorical_accuracy: 0.8288\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5460 - categorical_accuracy: 0.8196 - val_loss: 0.5102 - val_categorical_accuracy: 0.8380\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5542 - categorical_accuracy: 0.8156 - val_loss: 0.4983 - val_categorical_accuracy: 0.8362\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5436 - categorical_accuracy: 0.8226 - val_loss: 0.5919 - val_categorical_accuracy: 0.8074\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5413 - categorical_accuracy: 0.8229 - val_loss: 0.4333 - val_categorical_accuracy: 0.8548\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5382 - categorical_accuracy: 0.8209 - val_loss: 0.4223 - val_categorical_accuracy: 0.8616\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5377 - categorical_accuracy: 0.8258 - val_loss: 0.5181 - val_categorical_accuracy: 0.8298\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.5360 - categorical_accuracy: 0.8233 - val_loss: 0.4181 - val_categorical_accuracy: 0.8648\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.5178 - categorical_accuracy: 0.8309 - val_loss: 0.6035 - val_categorical_accuracy: 0.8112\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5237 - categorical_accuracy: 0.8285 - val_loss: 0.4645 - val_categorical_accuracy: 0.8492\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5217 - categorical_accuracy: 0.8294 - val_loss: 0.5097 - val_categorical_accuracy: 0.8378\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5145 - categorical_accuracy: 0.8304 - val_loss: 0.5834 - val_categorical_accuracy: 0.8238\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5087 - categorical_accuracy: 0.8324 - val_loss: 0.5185 - val_categorical_accuracy: 0.8334\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5024 - categorical_accuracy: 0.8365 - val_loss: 0.4831 - val_categorical_accuracy: 0.8444\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5005 - categorical_accuracy: 0.8353 - val_loss: 0.5627 - val_categorical_accuracy: 0.8230\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4982 - categorical_accuracy: 0.8377 - val_loss: 0.5311 - val_categorical_accuracy: 0.8308\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4953 - categorical_accuracy: 0.8378 - val_loss: 0.4464 - val_categorical_accuracy: 0.8518\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5046 - categorical_accuracy: 0.8349 - val_loss: 0.4580 - val_categorical_accuracy: 0.8502\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4940 - categorical_accuracy: 0.8374 - val_loss: 0.4495 - val_categorical_accuracy: 0.8498\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4880 - categorical_accuracy: 0.8403 - val_loss: 0.4754 - val_categorical_accuracy: 0.8472\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4873 - categorical_accuracy: 0.8412 - val_loss: 0.4969 - val_categorical_accuracy: 0.8406\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4791 - categorical_accuracy: 0.8436 - val_loss: 0.4666 - val_categorical_accuracy: 0.8506\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4811 - categorical_accuracy: 0.8426 - val_loss: 0.4955 - val_categorical_accuracy: 0.8418\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4827 - categorical_accuracy: 0.8432 - val_loss: 0.4090 - val_categorical_accuracy: 0.8688\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4742 - categorical_accuracy: 0.8446 - val_loss: 0.4464 - val_categorical_accuracy: 0.8564\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4699 - categorical_accuracy: 0.8452 - val_loss: 0.5875 - val_categorical_accuracy: 0.8194\n",
      "Epoch 72/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4674 - categorical_accuracy: 0.8475 - val_loss: 0.4669 - val_categorical_accuracy: 0.8548\n",
      "Epoch 73/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4642 - categorical_accuracy: 0.8486 - val_loss: 0.4695 - val_categorical_accuracy: 0.8520\n",
      "Epoch 74/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4648 - categorical_accuracy: 0.8490 - val_loss: 0.4352 - val_categorical_accuracy: 0.8652\n",
      "Epoch 75/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.4626 - categorical_accuracy: 0.8491 - val_loss: 0.5655 - val_categorical_accuracy: 0.8232\n",
      "Epoch 76/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4614 - categorical_accuracy: 0.8483 - val_loss: 0.4991 - val_categorical_accuracy: 0.8424\n",
      "Epoch 77/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4602 - categorical_accuracy: 0.8506 - val_loss: 0.4559 - val_categorical_accuracy: 0.8550\n",
      "Epoch 78/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4597 - categorical_accuracy: 0.8498 - val_loss: 0.4728 - val_categorical_accuracy: 0.8558\n",
      "Epoch 79/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4529 - categorical_accuracy: 0.8512 - val_loss: 0.4817 - val_categorical_accuracy: 0.8532\n",
      "Epoch 80/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4487 - categorical_accuracy: 0.8522 - val_loss: 0.4918 - val_categorical_accuracy: 0.8516\n",
      "Epoch 81/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4481 - categorical_accuracy: 0.8522 - val_loss: 0.5193 - val_categorical_accuracy: 0.8468\n",
      "Epoch 82/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4503 - categorical_accuracy: 0.8525 - val_loss: 0.4637 - val_categorical_accuracy: 0.8592\n",
      "Epoch 83/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4469 - categorical_accuracy: 0.8552 - val_loss: 0.5133 - val_categorical_accuracy: 0.8440\n",
      "Epoch 84/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4385 - categorical_accuracy: 0.8563 - val_loss: 0.5685 - val_categorical_accuracy: 0.8318\n",
      "Epoch 85/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4495 - categorical_accuracy: 0.8522 - val_loss: 0.4765 - val_categorical_accuracy: 0.8498\n",
      "Epoch 86/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4376 - categorical_accuracy: 0.8586 - val_loss: 0.5505 - val_categorical_accuracy: 0.8306\n",
      "Epoch 87/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4416 - categorical_accuracy: 0.8572 - val_loss: 0.4758 - val_categorical_accuracy: 0.8562\n",
      "Epoch 88/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4354 - categorical_accuracy: 0.8597 - val_loss: 0.4648 - val_categorical_accuracy: 0.8578\n",
      "Epoch 89/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4304 - categorical_accuracy: 0.8608 - val_loss: 0.5183 - val_categorical_accuracy: 0.8422\n",
      "Epoch 90/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4395 - categorical_accuracy: 0.8570 - val_loss: 0.4633 - val_categorical_accuracy: 0.8584\n",
      "Epoch 91/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4296 - categorical_accuracy: 0.8608 - val_loss: 0.5012 - val_categorical_accuracy: 0.8450\n",
      "Epoch 92/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4295 - categorical_accuracy: 0.8608 - val_loss: 0.5011 - val_categorical_accuracy: 0.8496\n",
      "Epoch 93/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4271 - categorical_accuracy: 0.8619 - val_loss: 0.5574 - val_categorical_accuracy: 0.8318\n",
      "Epoch 94/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4199 - categorical_accuracy: 0.8620 - val_loss: 0.4290 - val_categorical_accuracy: 0.8616\n",
      "Epoch 95/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4147 - categorical_accuracy: 0.8670 - val_loss: 0.4419 - val_categorical_accuracy: 0.8570\n",
      "Epoch 96/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.4280 - categorical_accuracy: 0.8621 - val_loss: 0.4506 - val_categorical_accuracy: 0.8568\n",
      "Epoch 97/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4205 - categorical_accuracy: 0.8648 - val_loss: 0.4776 - val_categorical_accuracy: 0.8506\n",
      "Epoch 98/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4204 - categorical_accuracy: 0.8652 - val_loss: 0.4563 - val_categorical_accuracy: 0.8554\n",
      "Epoch 99/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4118 - categorical_accuracy: 0.8664 - val_loss: 0.4177 - val_categorical_accuracy: 0.8684\n",
      "--- Running training session 35\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 29s 41ms/step - loss: 1.8900 - categorical_accuracy: 0.3047 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.6322 - categorical_accuracy: 0.4071 - val_loss: 1.3295 - val_categorical_accuracy: 0.5038\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.4914 - categorical_accuracy: 0.4718 - val_loss: 1.9505 - val_categorical_accuracy: 0.3892\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.3981 - categorical_accuracy: 0.5061 - val_loss: 1.0958 - val_categorical_accuracy: 0.6112\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 1.3125 - categorical_accuracy: 0.5410 - val_loss: 1.0598 - val_categorical_accuracy: 0.6256\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.2488 - categorical_accuracy: 0.5666 - val_loss: 1.1274 - val_categorical_accuracy: 0.6058\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.1898 - categorical_accuracy: 0.5873 - val_loss: 1.0025 - val_categorical_accuracy: 0.6458\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.1520 - categorical_accuracy: 0.6044 - val_loss: 0.8119 - val_categorical_accuracy: 0.7096\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.1032 - categorical_accuracy: 0.6226 - val_loss: 0.8779 - val_categorical_accuracy: 0.6986\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.0692 - categorical_accuracy: 0.6371 - val_loss: 0.7696 - val_categorical_accuracy: 0.7294\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.0367 - categorical_accuracy: 0.6467 - val_loss: 0.8713 - val_categorical_accuracy: 0.7054\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.0036 - categorical_accuracy: 0.6607 - val_loss: 0.9734 - val_categorical_accuracy: 0.6682\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.9796 - categorical_accuracy: 0.6696 - val_loss: 0.8756 - val_categorical_accuracy: 0.7102\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.9587 - categorical_accuracy: 0.6766 - val_loss: 0.9334 - val_categorical_accuracy: 0.6938\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.9326 - categorical_accuracy: 0.6858 - val_loss: 0.6531 - val_categorical_accuracy: 0.7790\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.9076 - categorical_accuracy: 0.6930 - val_loss: 0.7154 - val_categorical_accuracy: 0.7542\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.8900 - categorical_accuracy: 0.7013 - val_loss: 0.6261 - val_categorical_accuracy: 0.7840\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.8658 - categorical_accuracy: 0.7102 - val_loss: 0.7000 - val_categorical_accuracy: 0.7660\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.8492 - categorical_accuracy: 0.7132 - val_loss: 0.8426 - val_categorical_accuracy: 0.7172\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.8391 - categorical_accuracy: 0.7176 - val_loss: 0.6974 - val_categorical_accuracy: 0.7632\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8260 - categorical_accuracy: 0.7220 - val_loss: 0.6097 - val_categorical_accuracy: 0.7960\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8055 - categorical_accuracy: 0.7298 - val_loss: 0.5967 - val_categorical_accuracy: 0.7992\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8007 - categorical_accuracy: 0.7336 - val_loss: 0.6225 - val_categorical_accuracy: 0.7904\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 0.7886 - categorical_accuracy: 0.7386 - val_loss: 0.7239 - val_categorical_accuracy: 0.7626\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.7720 - categorical_accuracy: 0.7431 - val_loss: 0.6634 - val_categorical_accuracy: 0.7800\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7629 - categorical_accuracy: 0.7471 - val_loss: 0.6564 - val_categorical_accuracy: 0.7756\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7546 - categorical_accuracy: 0.7494 - val_loss: 0.5610 - val_categorical_accuracy: 0.8118\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.7389 - categorical_accuracy: 0.7546 - val_loss: 0.7198 - val_categorical_accuracy: 0.7614\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7400 - categorical_accuracy: 0.7534 - val_loss: 0.5896 - val_categorical_accuracy: 0.8028\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7267 - categorical_accuracy: 0.7610 - val_loss: 0.7513 - val_categorical_accuracy: 0.7448\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7143 - categorical_accuracy: 0.7617 - val_loss: 0.5393 - val_categorical_accuracy: 0.8232\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.7039 - categorical_accuracy: 0.7663 - val_loss: 0.5864 - val_categorical_accuracy: 0.8082\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6972 - categorical_accuracy: 0.7710 - val_loss: 0.6915 - val_categorical_accuracy: 0.7680\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6912 - categorical_accuracy: 0.7710 - val_loss: 0.8408 - val_categorical_accuracy: 0.7490\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6941 - categorical_accuracy: 0.7720 - val_loss: 0.6197 - val_categorical_accuracy: 0.7996\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6755 - categorical_accuracy: 0.7766 - val_loss: 0.5521 - val_categorical_accuracy: 0.8146\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6698 - categorical_accuracy: 0.7773 - val_loss: 0.5320 - val_categorical_accuracy: 0.8182\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6601 - categorical_accuracy: 0.7790 - val_loss: 0.4988 - val_categorical_accuracy: 0.8326\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6679 - categorical_accuracy: 0.7817 - val_loss: 0.7529 - val_categorical_accuracy: 0.7562\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6538 - categorical_accuracy: 0.7823 - val_loss: 0.4549 - val_categorical_accuracy: 0.8468\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6453 - categorical_accuracy: 0.7861 - val_loss: 0.5600 - val_categorical_accuracy: 0.8098\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6518 - categorical_accuracy: 0.7848 - val_loss: 0.6802 - val_categorical_accuracy: 0.7856\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6364 - categorical_accuracy: 0.7888 - val_loss: 0.5337 - val_categorical_accuracy: 0.8258\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6345 - categorical_accuracy: 0.7928 - val_loss: 0.5444 - val_categorical_accuracy: 0.8216\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6304 - categorical_accuracy: 0.7908 - val_loss: 0.5104 - val_categorical_accuracy: 0.8242\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6193 - categorical_accuracy: 0.7975 - val_loss: 0.4832 - val_categorical_accuracy: 0.8408\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6159 - categorical_accuracy: 0.7946 - val_loss: 0.6060 - val_categorical_accuracy: 0.8034\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6088 - categorical_accuracy: 0.8000 - val_loss: 0.5707 - val_categorical_accuracy: 0.8126\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6133 - categorical_accuracy: 0.7987 - val_loss: 0.5901 - val_categorical_accuracy: 0.8074\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6031 - categorical_accuracy: 0.8002 - val_loss: 0.6043 - val_categorical_accuracy: 0.8080\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.5980 - categorical_accuracy: 0.8034 - val_loss: 0.6078 - val_categorical_accuracy: 0.8104\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5873 - categorical_accuracy: 0.8056 - val_loss: 0.6415 - val_categorical_accuracy: 0.8020\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5839 - categorical_accuracy: 0.8088 - val_loss: 0.5519 - val_categorical_accuracy: 0.8252\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5839 - categorical_accuracy: 0.8068 - val_loss: 0.4874 - val_categorical_accuracy: 0.8412\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5844 - categorical_accuracy: 0.8067 - val_loss: 0.5010 - val_categorical_accuracy: 0.8378\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5768 - categorical_accuracy: 0.8096 - val_loss: 0.4734 - val_categorical_accuracy: 0.8472\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5689 - categorical_accuracy: 0.8119 - val_loss: 0.5645 - val_categorical_accuracy: 0.8246\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.5716 - categorical_accuracy: 0.8127 - val_loss: 0.5470 - val_categorical_accuracy: 0.8180\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5592 - categorical_accuracy: 0.8144 - val_loss: 0.4689 - val_categorical_accuracy: 0.8510\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.5575 - categorical_accuracy: 0.8162 - val_loss: 0.4909 - val_categorical_accuracy: 0.8394\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5613 - categorical_accuracy: 0.8157 - val_loss: 0.5583 - val_categorical_accuracy: 0.8208\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5591 - categorical_accuracy: 0.8161 - val_loss: 0.5340 - val_categorical_accuracy: 0.8344\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5500 - categorical_accuracy: 0.8174 - val_loss: 0.5100 - val_categorical_accuracy: 0.8382\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.5469 - categorical_accuracy: 0.8211 - val_loss: 0.4939 - val_categorical_accuracy: 0.8442\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5484 - categorical_accuracy: 0.8189 - val_loss: 0.4888 - val_categorical_accuracy: 0.8358\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5456 - categorical_accuracy: 0.8175 - val_loss: 0.4967 - val_categorical_accuracy: 0.8396\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5423 - categorical_accuracy: 0.8217 - val_loss: 0.5186 - val_categorical_accuracy: 0.8312\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.5314 - categorical_accuracy: 0.8240 - val_loss: 0.6074 - val_categorical_accuracy: 0.8134\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5260 - categorical_accuracy: 0.8275 - val_loss: 0.4990 - val_categorical_accuracy: 0.8380\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5302 - categorical_accuracy: 0.8258 - val_loss: 0.6287 - val_categorical_accuracy: 0.8142\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5293 - categorical_accuracy: 0.8264 - val_loss: 0.4598 - val_categorical_accuracy: 0.8506\n",
      "Epoch 72/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5224 - categorical_accuracy: 0.8275 - val_loss: 0.5115 - val_categorical_accuracy: 0.8374\n",
      "Epoch 73/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5238 - categorical_accuracy: 0.8287 - val_loss: 0.5001 - val_categorical_accuracy: 0.8416\n",
      "Epoch 74/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.5242 - categorical_accuracy: 0.8290 - val_loss: 0.7174 - val_categorical_accuracy: 0.7922\n",
      "Epoch 75/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.5131 - categorical_accuracy: 0.8314 - val_loss: 0.5699 - val_categorical_accuracy: 0.8198\n",
      "Epoch 76/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.5108 - categorical_accuracy: 0.8336 - val_loss: 0.6408 - val_categorical_accuracy: 0.8084\n",
      "Epoch 77/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.5113 - categorical_accuracy: 0.8324 - val_loss: 0.4858 - val_categorical_accuracy: 0.8492\n",
      "Epoch 78/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5113 - categorical_accuracy: 0.8322 - val_loss: 0.6518 - val_categorical_accuracy: 0.7988\n",
      "Epoch 79/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5121 - categorical_accuracy: 0.8339 - val_loss: 0.4701 - val_categorical_accuracy: 0.8488\n",
      "Epoch 80/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5071 - categorical_accuracy: 0.8330 - val_loss: 0.4654 - val_categorical_accuracy: 0.8496\n",
      "Epoch 81/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5006 - categorical_accuracy: 0.8355 - val_loss: 0.4753 - val_categorical_accuracy: 0.8518\n",
      "Epoch 82/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4975 - categorical_accuracy: 0.8366 - val_loss: 0.5002 - val_categorical_accuracy: 0.8352\n",
      "Epoch 83/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4994 - categorical_accuracy: 0.8374 - val_loss: 0.5155 - val_categorical_accuracy: 0.8414\n",
      "Epoch 84/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4926 - categorical_accuracy: 0.8381 - val_loss: 0.4637 - val_categorical_accuracy: 0.8536\n",
      "Epoch 85/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.5019 - categorical_accuracy: 0.8364 - val_loss: 0.6239 - val_categorical_accuracy: 0.8110\n",
      "Epoch 86/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4985 - categorical_accuracy: 0.8384 - val_loss: 0.5131 - val_categorical_accuracy: 0.8370\n",
      "Epoch 87/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4895 - categorical_accuracy: 0.8406 - val_loss: 0.4645 - val_categorical_accuracy: 0.8496\n",
      "Epoch 88/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4885 - categorical_accuracy: 0.8412 - val_loss: 0.4779 - val_categorical_accuracy: 0.8446\n",
      "Epoch 89/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4894 - categorical_accuracy: 0.8406 - val_loss: 0.4745 - val_categorical_accuracy: 0.8536\n",
      "Epoch 90/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4808 - categorical_accuracy: 0.8429 - val_loss: 0.4793 - val_categorical_accuracy: 0.8468\n",
      "Epoch 91/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4852 - categorical_accuracy: 0.8402 - val_loss: 0.5033 - val_categorical_accuracy: 0.8424\n",
      "Epoch 92/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.4732 - categorical_accuracy: 0.8462 - val_loss: 0.5156 - val_categorical_accuracy: 0.8398\n",
      "Epoch 93/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4837 - categorical_accuracy: 0.8427 - val_loss: 0.4746 - val_categorical_accuracy: 0.8442\n",
      "Epoch 94/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4741 - categorical_accuracy: 0.8478 - val_loss: 0.4638 - val_categorical_accuracy: 0.8492\n",
      "Epoch 95/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4736 - categorical_accuracy: 0.8471 - val_loss: 0.4294 - val_categorical_accuracy: 0.8598\n",
      "Epoch 96/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.4677 - categorical_accuracy: 0.8497 - val_loss: 0.6077 - val_categorical_accuracy: 0.8200\n",
      "Epoch 97/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4656 - categorical_accuracy: 0.8499 - val_loss: 0.5513 - val_categorical_accuracy: 0.8444\n",
      "Epoch 98/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.4767 - categorical_accuracy: 0.8468 - val_loss: 0.4880 - val_categorical_accuracy: 0.8424\n",
      "Epoch 99/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.4651 - categorical_accuracy: 0.8493 - val_loss: 0.5533 - val_categorical_accuracy: 0.8354\n",
      "Epoch 100/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.4663 - categorical_accuracy: 0.8488 - val_loss: 0.4634 - val_categorical_accuracy: 0.8570\n",
      "--- Running training session 36\n",
      "Epoch 1/100\n",
      "704/704 [==============================] - 30s 43ms/step - loss: 1.9677 - categorical_accuracy: 0.2684 - val_loss: 0.0000e+00 - val_categorical_accuracy: 0.0000e+00\n",
      "Epoch 2/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 1.7912 - categorical_accuracy: 0.3379 - val_loss: 1.5339 - val_categorical_accuracy: 0.4228\n",
      "Epoch 3/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.6756 - categorical_accuracy: 0.3892 - val_loss: 1.7057 - val_categorical_accuracy: 0.3974\n",
      "Epoch 4/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.6017 - categorical_accuracy: 0.4205 - val_loss: 1.3701 - val_categorical_accuracy: 0.5034\n",
      "Epoch 5/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.5236 - categorical_accuracy: 0.4517 - val_loss: 1.2317 - val_categorical_accuracy: 0.5502\n",
      "Epoch 6/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.4636 - categorical_accuracy: 0.4790 - val_loss: 1.3991 - val_categorical_accuracy: 0.5122\n",
      "Epoch 7/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.4191 - categorical_accuracy: 0.4937 - val_loss: 1.1598 - val_categorical_accuracy: 0.5930\n",
      "Epoch 8/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.3733 - categorical_accuracy: 0.5181 - val_loss: 1.0365 - val_categorical_accuracy: 0.6124\n",
      "Epoch 9/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 1.3231 - categorical_accuracy: 0.5381 - val_loss: 1.3194 - val_categorical_accuracy: 0.5732\n",
      "Epoch 10/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 1.2806 - categorical_accuracy: 0.5556 - val_loss: 0.8927 - val_categorical_accuracy: 0.6862\n",
      "Epoch 11/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.2601 - categorical_accuracy: 0.5653 - val_loss: 1.3346 - val_categorical_accuracy: 0.5674\n",
      "Epoch 12/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 1.2253 - categorical_accuracy: 0.5775 - val_loss: 0.9324 - val_categorical_accuracy: 0.6730\n",
      "Epoch 13/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 1.1986 - categorical_accuracy: 0.5876 - val_loss: 0.9931 - val_categorical_accuracy: 0.6520\n",
      "Epoch 14/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.1673 - categorical_accuracy: 0.5964 - val_loss: 0.8943 - val_categorical_accuracy: 0.6968\n",
      "Epoch 15/100\n",
      "704/704 [==============================] - 21s 30ms/step - loss: 1.1418 - categorical_accuracy: 0.6097 - val_loss: 0.8642 - val_categorical_accuracy: 0.6946\n",
      "Epoch 16/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.1263 - categorical_accuracy: 0.6142 - val_loss: 0.8741 - val_categorical_accuracy: 0.7038\n",
      "Epoch 17/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.1014 - categorical_accuracy: 0.6246 - val_loss: 0.9863 - val_categorical_accuracy: 0.6574\n",
      "Epoch 18/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 1.0872 - categorical_accuracy: 0.6272 - val_loss: 0.9020 - val_categorical_accuracy: 0.6884\n",
      "Epoch 19/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.0718 - categorical_accuracy: 0.6364 - val_loss: 0.7357 - val_categorical_accuracy: 0.7548\n",
      "Epoch 20/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 1.0465 - categorical_accuracy: 0.6437 - val_loss: 0.9036 - val_categorical_accuracy: 0.6994\n",
      "Epoch 21/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 1.0358 - categorical_accuracy: 0.6500 - val_loss: 0.7658 - val_categorical_accuracy: 0.7462\n",
      "Epoch 22/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 1.0122 - categorical_accuracy: 0.6577 - val_loss: 0.7525 - val_categorical_accuracy: 0.7436\n",
      "Epoch 23/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.9906 - categorical_accuracy: 0.6658 - val_loss: 0.7406 - val_categorical_accuracy: 0.7520\n",
      "Epoch 24/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.9823 - categorical_accuracy: 0.6707 - val_loss: 0.8087 - val_categorical_accuracy: 0.7356\n",
      "Epoch 25/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.9751 - categorical_accuracy: 0.6726 - val_loss: 0.7187 - val_categorical_accuracy: 0.7590\n",
      "Epoch 26/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.9621 - categorical_accuracy: 0.6771 - val_loss: 0.7716 - val_categorical_accuracy: 0.7544\n",
      "Epoch 27/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.9411 - categorical_accuracy: 0.6818 - val_loss: 0.7242 - val_categorical_accuracy: 0.7678\n",
      "Epoch 28/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.9414 - categorical_accuracy: 0.6866 - val_loss: 0.6119 - val_categorical_accuracy: 0.8004\n",
      "Epoch 29/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.9308 - categorical_accuracy: 0.6887 - val_loss: 0.6426 - val_categorical_accuracy: 0.7810\n",
      "Epoch 30/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.9178 - categorical_accuracy: 0.6919 - val_loss: 0.6864 - val_categorical_accuracy: 0.7742\n",
      "Epoch 31/100\n",
      "704/704 [==============================] - 20s 28ms/step - loss: 0.9076 - categorical_accuracy: 0.6990 - val_loss: 0.6346 - val_categorical_accuracy: 0.7944\n",
      "Epoch 32/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8965 - categorical_accuracy: 0.7028 - val_loss: 0.6681 - val_categorical_accuracy: 0.7768\n",
      "Epoch 33/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8962 - categorical_accuracy: 0.7027 - val_loss: 0.6085 - val_categorical_accuracy: 0.8010\n",
      "Epoch 34/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8740 - categorical_accuracy: 0.7094 - val_loss: 0.7370 - val_categorical_accuracy: 0.7462\n",
      "Epoch 35/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8767 - categorical_accuracy: 0.7069 - val_loss: 0.6200 - val_categorical_accuracy: 0.7890\n",
      "Epoch 36/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8640 - categorical_accuracy: 0.7115 - val_loss: 0.5279 - val_categorical_accuracy: 0.8270\n",
      "Epoch 37/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8508 - categorical_accuracy: 0.7183 - val_loss: 0.6466 - val_categorical_accuracy: 0.7852\n",
      "Epoch 38/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.8412 - categorical_accuracy: 0.7174 - val_loss: 0.5654 - val_categorical_accuracy: 0.8108\n",
      "Epoch 39/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8429 - categorical_accuracy: 0.7201 - val_loss: 0.5652 - val_categorical_accuracy: 0.8104\n",
      "Epoch 40/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8419 - categorical_accuracy: 0.7200 - val_loss: 0.6136 - val_categorical_accuracy: 0.7974\n",
      "Epoch 41/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.8322 - categorical_accuracy: 0.7243 - val_loss: 0.5846 - val_categorical_accuracy: 0.8066\n",
      "Epoch 42/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.8257 - categorical_accuracy: 0.7246 - val_loss: 0.5434 - val_categorical_accuracy: 0.8180\n",
      "Epoch 43/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.8109 - categorical_accuracy: 0.7318 - val_loss: 0.5644 - val_categorical_accuracy: 0.8128\n",
      "Epoch 44/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.8122 - categorical_accuracy: 0.7311 - val_loss: 0.5131 - val_categorical_accuracy: 0.8316\n",
      "Epoch 45/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.8077 - categorical_accuracy: 0.7305 - val_loss: 0.5122 - val_categorical_accuracy: 0.8334\n",
      "Epoch 46/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8008 - categorical_accuracy: 0.7328 - val_loss: 0.6089 - val_categorical_accuracy: 0.7998\n",
      "Epoch 47/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.8030 - categorical_accuracy: 0.7354 - val_loss: 0.5035 - val_categorical_accuracy: 0.8278\n",
      "Epoch 48/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7900 - categorical_accuracy: 0.7397 - val_loss: 0.6408 - val_categorical_accuracy: 0.7792\n",
      "Epoch 49/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.7845 - categorical_accuracy: 0.7402 - val_loss: 0.5368 - val_categorical_accuracy: 0.8224\n",
      "Epoch 50/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7814 - categorical_accuracy: 0.7432 - val_loss: 0.8477 - val_categorical_accuracy: 0.7526\n",
      "Epoch 51/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7785 - categorical_accuracy: 0.7456 - val_loss: 0.5075 - val_categorical_accuracy: 0.8364\n",
      "Epoch 52/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.7788 - categorical_accuracy: 0.7434 - val_loss: 0.5435 - val_categorical_accuracy: 0.8170\n",
      "Epoch 53/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.7660 - categorical_accuracy: 0.7461 - val_loss: 0.5612 - val_categorical_accuracy: 0.8092\n",
      "Epoch 54/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.7634 - categorical_accuracy: 0.7504 - val_loss: 0.5239 - val_categorical_accuracy: 0.8270\n",
      "Epoch 55/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7646 - categorical_accuracy: 0.7496 - val_loss: 0.6785 - val_categorical_accuracy: 0.7800\n",
      "Epoch 56/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.7518 - categorical_accuracy: 0.7520 - val_loss: 0.5186 - val_categorical_accuracy: 0.8270\n",
      "Epoch 57/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.7509 - categorical_accuracy: 0.7506 - val_loss: 0.5200 - val_categorical_accuracy: 0.8330\n",
      "Epoch 58/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7442 - categorical_accuracy: 0.7542 - val_loss: 0.5037 - val_categorical_accuracy: 0.8320\n",
      "Epoch 59/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7407 - categorical_accuracy: 0.7535 - val_loss: 0.4966 - val_categorical_accuracy: 0.8370\n",
      "Epoch 60/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7371 - categorical_accuracy: 0.7583 - val_loss: 0.5381 - val_categorical_accuracy: 0.8296\n",
      "Epoch 61/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7367 - categorical_accuracy: 0.7535 - val_loss: 0.5562 - val_categorical_accuracy: 0.8258\n",
      "Epoch 62/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.7284 - categorical_accuracy: 0.7610 - val_loss: 0.5633 - val_categorical_accuracy: 0.8150\n",
      "Epoch 63/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7358 - categorical_accuracy: 0.7563 - val_loss: 0.5091 - val_categorical_accuracy: 0.8412\n",
      "Epoch 64/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.7204 - categorical_accuracy: 0.7640 - val_loss: 0.5639 - val_categorical_accuracy: 0.8216\n",
      "Epoch 65/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7236 - categorical_accuracy: 0.7632 - val_loss: 0.4953 - val_categorical_accuracy: 0.8390\n",
      "Epoch 66/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7154 - categorical_accuracy: 0.7663 - val_loss: 0.7082 - val_categorical_accuracy: 0.7710\n",
      "Epoch 67/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7137 - categorical_accuracy: 0.7636 - val_loss: 0.5423 - val_categorical_accuracy: 0.8130\n",
      "Epoch 68/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.7189 - categorical_accuracy: 0.7628 - val_loss: 0.5922 - val_categorical_accuracy: 0.8120\n",
      "Epoch 69/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.7099 - categorical_accuracy: 0.7654 - val_loss: 0.6022 - val_categorical_accuracy: 0.8056\n",
      "Epoch 70/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7056 - categorical_accuracy: 0.7673 - val_loss: 0.4626 - val_categorical_accuracy: 0.8516\n",
      "Epoch 71/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.7031 - categorical_accuracy: 0.7712 - val_loss: 0.5618 - val_categorical_accuracy: 0.8236\n",
      "Epoch 72/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6929 - categorical_accuracy: 0.7688 - val_loss: 0.5973 - val_categorical_accuracy: 0.8158\n",
      "Epoch 73/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6872 - categorical_accuracy: 0.7727 - val_loss: 0.4818 - val_categorical_accuracy: 0.8408\n",
      "Epoch 74/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6960 - categorical_accuracy: 0.7710 - val_loss: 0.5383 - val_categorical_accuracy: 0.8246\n",
      "Epoch 75/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6911 - categorical_accuracy: 0.7734 - val_loss: 0.5188 - val_categorical_accuracy: 0.8322\n",
      "Epoch 76/100\n",
      "704/704 [==============================] - 19s 26ms/step - loss: 0.6816 - categorical_accuracy: 0.7762 - val_loss: 0.5654 - val_categorical_accuracy: 0.8216\n",
      "Epoch 77/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6884 - categorical_accuracy: 0.7744 - val_loss: 0.4884 - val_categorical_accuracy: 0.8382\n",
      "Epoch 78/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6818 - categorical_accuracy: 0.7772 - val_loss: 0.4479 - val_categorical_accuracy: 0.8574\n",
      "Epoch 79/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6750 - categorical_accuracy: 0.7790 - val_loss: 0.5280 - val_categorical_accuracy: 0.8352\n",
      "Epoch 80/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6682 - categorical_accuracy: 0.7839 - val_loss: 0.4894 - val_categorical_accuracy: 0.8410\n",
      "Epoch 81/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6747 - categorical_accuracy: 0.7764 - val_loss: 0.5295 - val_categorical_accuracy: 0.8332\n",
      "Epoch 82/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6760 - categorical_accuracy: 0.7790 - val_loss: 0.4580 - val_categorical_accuracy: 0.8538\n",
      "Epoch 83/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6681 - categorical_accuracy: 0.7804 - val_loss: 0.5649 - val_categorical_accuracy: 0.8132\n",
      "Epoch 84/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6708 - categorical_accuracy: 0.7802 - val_loss: 0.4819 - val_categorical_accuracy: 0.8390\n",
      "Epoch 85/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.6620 - categorical_accuracy: 0.7817 - val_loss: 0.4839 - val_categorical_accuracy: 0.8462\n",
      "Epoch 86/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6597 - categorical_accuracy: 0.7809 - val_loss: 0.4802 - val_categorical_accuracy: 0.8454\n",
      "Epoch 87/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6632 - categorical_accuracy: 0.7836 - val_loss: 0.7145 - val_categorical_accuracy: 0.7814\n",
      "Epoch 88/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6697 - categorical_accuracy: 0.7794 - val_loss: 0.5051 - val_categorical_accuracy: 0.8374\n",
      "Epoch 89/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6656 - categorical_accuracy: 0.7818 - val_loss: 0.4634 - val_categorical_accuracy: 0.8534\n",
      "Epoch 90/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6608 - categorical_accuracy: 0.7839 - val_loss: 0.4716 - val_categorical_accuracy: 0.8446\n",
      "Epoch 91/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6539 - categorical_accuracy: 0.7878 - val_loss: 0.5252 - val_categorical_accuracy: 0.8308\n",
      "Epoch 92/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6493 - categorical_accuracy: 0.7889 - val_loss: 0.4953 - val_categorical_accuracy: 0.8418\n",
      "Epoch 93/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6421 - categorical_accuracy: 0.7921 - val_loss: 0.4576 - val_categorical_accuracy: 0.8610\n",
      "Epoch 94/100\n",
      "704/704 [==============================] - 17s 25ms/step - loss: 0.6553 - categorical_accuracy: 0.7870 - val_loss: 0.4716 - val_categorical_accuracy: 0.8484\n",
      "Epoch 95/100\n",
      "704/704 [==============================] - 17s 24ms/step - loss: 0.6520 - categorical_accuracy: 0.7867 - val_loss: 0.5368 - val_categorical_accuracy: 0.8360\n",
      "Epoch 96/100\n",
      "704/704 [==============================] - 18s 25ms/step - loss: 0.6514 - categorical_accuracy: 0.7856 - val_loss: 0.4546 - val_categorical_accuracy: 0.8544\n",
      "Epoch 97/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.6500 - categorical_accuracy: 0.7865 - val_loss: 0.4578 - val_categorical_accuracy: 0.8564\n",
      "Epoch 98/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6407 - categorical_accuracy: 0.7886 - val_loss: 0.5303 - val_categorical_accuracy: 0.8290\n",
      "Epoch 99/100\n",
      "704/704 [==============================] - 18s 26ms/step - loss: 0.6454 - categorical_accuracy: 0.7890 - val_loss: 0.5672 - val_categorical_accuracy: 0.8142\n",
      "Epoch 100/100\n",
      "704/704 [==============================] - 19s 27ms/step - loss: 0.6410 - categorical_accuracy: 0.7896 - val_loss: 0.5317 - val_categorical_accuracy: 0.8364\n"
     ]
    }
   ],
   "source": [
    "session_id = 1  # 訓練任務的id\n",
    "for ia in hparam_ia.domain.values:\n",
    "    for bn in hparam_bn.domain.values:\n",
    "        for init in hparam_init.domain.values:\n",
    "            for lr in hparam_lr.domain.values:\n",
    "                # 顯示目前訓練任務id\n",
    "                print('--- Running training session {}'.format(session_id))\n",
    "                # 設定本次訓練的超參數\n",
    "                hparams = {hparam_ia: ia, hparam_bn: bn, hparam_init: init, hparam_lr: lr}\n",
    "                # 儲放紀錄檔的位置\n",
    "                logs_dir = os.path.join(\"lab7-logs-hparams\", \"run-{}\".format(session_id))\n",
    "                # 建立、編譯及訓練網路模型\n",
    "                train_test_model(logs_dir, hparams)\n",
    "                session_id += 1  # id+1"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  },
  "pycharm": {
   "stem_cell": {
    "cell_type": "raw",
    "metadata": {
     "collapsed": false
    },
    "source": []
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}