Coffee and Code - Keras.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Coffee and Code - 19/10/2017\n",
    "## Artificial Neural Networks\n",
    "### Keras and Tensorflow\n",
    "\n",
    "\n",
    "Here we import Sequential from keras.models, which is a standard feedforward network.\n",
    "Dense and Activation are layers of the network.\n",
    "np_utils is useful later for converting numbers into one-hot vectors.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Activation\n",
    "from keras.utils import np_utils\n",
    "import numpy as np\n",
    "import sklearn.datasets\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Basic Categorisation\n",
    "We are setting up a dataset here, with two crescent shaped groups, categorised differently."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAD8CAYAAAB+UHOxAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nGBgYGFyk/D86z15RULExMAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xb054da78d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "X, y = sklearn.datasets.make_moons(200, noise=0.25, random_state=0)\n",
    "yc = y\n",
    "y = y.reshape(y.size,1).astype(float)\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "ax.scatter(X[:,0], X[:,1], s=40, c=yc, cmap = plt.cm.Spectral, zorder = 1)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We want to be able to predict which group a new data point will belong to. Neural networks can do this very easily. <br>\n",
    "\n",
    "We start by using the previously imported Sequential class. <br>\n",
    "\n",
    "Just by having gone through this many times before, I choose to have an input layer with two neurons (for the two coordinates), 50 neurons in the hidden layer (it probably need not be this high, my neural net can do it with 10), and one in the output (this will be a number between 1 and 0, where 1 is red and 0 is blue. <br>\n",
    "\n",
    "Activation functions in both layers are sigmoids. This was chosen mostly because it is common, and I don't want to change the data beforehand (as sigmoid outputs numbers in the range 0 to 1). <br>\n",
    "\n",
    "Mean squared error is the error function I defined earlier. There are better error functions for other problems, but this will suffice here. Adam refers to adaptive moment estimation, which is essentially an algorithm which speeds up the process of learning."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "model.add(Dense(input_dim = 2, units = 50, activation='sigmoid'))\n",
    "#model.add(Activation('sigmoid'))\n",
    "model.add(Dense(input_dim = 50, units = 1, activation='sigmoid'))\n",
    "#model.add(Activation('sigmoid'))\n",
    "model.compile(optimizer='adam',\n",
    "              loss='mean_squared_error',\n",
    "              metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Okay, after compiling the network, we want to fit the data. This is done easily by calling the fit() method of the network."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2828 - acc: 0.5000     \n",
      "Epoch 2/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2740 - acc: 0.5000     \n",
      "Epoch 3/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2652 - acc: 0.5000     \n",
      "Epoch 4/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2585 - acc: 0.5000     \n",
      "Epoch 5/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2538 - acc: 0.5000     \n",
      "Epoch 6/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2487 - acc: 0.5000     \n",
      "Epoch 7/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2448 - acc: 0.6550     \n",
      "Epoch 8/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2414 - acc: 0.8100     \n",
      "Epoch 9/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2378 - acc: 0.8250     \n",
      "Epoch 10/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2344 - acc: 0.8250     \n",
      "Epoch 11/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2307 - acc: 0.8050     \n",
      "Epoch 12/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2275 - acc: 0.8300     \n",
      "Epoch 13/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2243 - acc: 0.8000     \n",
      "Epoch 14/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2212 - acc: 0.8050     \n",
      "Epoch 15/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2180 - acc: 0.8050     \n",
      "Epoch 16/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2149 - acc: 0.8050     \n",
      "Epoch 17/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2116 - acc: 0.8050     \n",
      "Epoch 18/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.2088 - acc: 0.812 - 0s - loss: 0.2087 - acc: 0.8050     \n",
      "Epoch 19/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2060 - acc: 0.8000     \n",
      "Epoch 20/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2028 - acc: 0.8100     \n",
      "Epoch 21/2500\n",
      "200/200 [==============================] - 0s - loss: 0.2001 - acc: 0.8100     \n",
      "Epoch 22/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1971 - acc: 0.8100     \n",
      "Epoch 23/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1944 - acc: 0.8150     \n",
      "Epoch 24/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1917 - acc: 0.8100     \n",
      "Epoch 25/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1889 - acc: 0.8050     \n",
      "Epoch 26/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1862 - acc: 0.8100     \n",
      "Epoch 27/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1837 - acc: 0.8100     \n",
      "Epoch 28/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1814 - acc: 0.8050     \n",
      "Epoch 29/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1790 - acc: 0.8050     \n",
      "Epoch 30/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1765 - acc: 0.8050     \n",
      "Epoch 31/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1744 - acc: 0.8100     \n",
      "Epoch 32/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1719 - acc: 0.8100     \n",
      "Epoch 33/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1697 - acc: 0.8050     \n",
      "Epoch 34/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1676 - acc: 0.8100     \n",
      "Epoch 35/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1655 - acc: 0.8050     \n",
      "Epoch 36/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1634 - acc: 0.8150     \n",
      "Epoch 37/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1616 - acc: 0.8050     \n",
      "Epoch 38/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1598 - acc: 0.8100     \n",
      "Epoch 39/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1580 - acc: 0.8100     \n",
      "Epoch 40/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1563 - acc: 0.8150     \n",
      "Epoch 41/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1548 - acc: 0.8150     \n",
      "Epoch 42/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1532 - acc: 0.8050     \n",
      "Epoch 43/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1516 - acc: 0.8150     \n",
      "Epoch 44/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1502 - acc: 0.8100     \n",
      "Epoch 45/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1488 - acc: 0.8100     \n",
      "Epoch 46/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1475 - acc: 0.8150     \n",
      "Epoch 47/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1462 - acc: 0.8200     \n",
      "Epoch 48/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1450 - acc: 0.8150     \n",
      "Epoch 49/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1437 - acc: 0.8150     \n",
      "Epoch 50/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1425 - acc: 0.8150     \n",
      "Epoch 51/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1415 - acc: 0.8150     \n",
      "Epoch 52/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1402 - acc: 0.8200     \n",
      "Epoch 53/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1393 - acc: 0.8150     \n",
      "Epoch 54/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1382 - acc: 0.8150     \n",
      "Epoch 55/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1373 - acc: 0.8150     \n",
      "Epoch 56/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1362 - acc: 0.8150     \n",
      "Epoch 57/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1351 - acc: 0.8150     \n",
      "Epoch 58/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1343 - acc: 0.8150     \n",
      "Epoch 59/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1334 - acc: 0.8150     \n",
      "Epoch 60/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1326 - acc: 0.8250     \n",
      "Epoch 61/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1318 - acc: 0.8200     \n",
      "Epoch 62/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1311 - acc: 0.8200     \n",
      "Epoch 63/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1305 - acc: 0.8200     \n",
      "Epoch 64/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1298 - acc: 0.8200     \n",
      "Epoch 65/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1292 - acc: 0.8200     \n",
      "Epoch 66/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1285 - acc: 0.8200     \n",
      "Epoch 67/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1278 - acc: 0.8250     \n",
      "Epoch 68/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1272 - acc: 0.8250     \n",
      "Epoch 69/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1266 - acc: 0.8250     \n",
      "Epoch 70/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1260 - acc: 0.8200     \n",
      "Epoch 71/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1253 - acc: 0.8250     \n",
      "Epoch 72/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1248 - acc: 0.8250     \n",
      "Epoch 73/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1243 - acc: 0.8250     \n",
      "Epoch 74/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1238 - acc: 0.8250     \n",
      "Epoch 75/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1233 - acc: 0.8250     \n",
      "Epoch 76/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1228 - acc: 0.8250     \n",
      "Epoch 77/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1223 - acc: 0.8300     \n",
      "Epoch 78/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1218 - acc: 0.8250     \n",
      "Epoch 79/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1214 - acc: 0.8250     \n",
      "Epoch 80/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1210 - acc: 0.8250     \n",
      "Epoch 81/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1207 - acc: 0.8250     \n",
      "Epoch 82/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1202 - acc: 0.8250     \n",
      "Epoch 83/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1198 - acc: 0.8300     \n",
      "Epoch 84/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1194 - acc: 0.8300     \n",
      "Epoch 85/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1380 - acc: 0.750 - 0s - loss: 0.1191 - acc: 0.8250     \n",
      "Epoch 86/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1188 - acc: 0.8300     \n",
      "Epoch 87/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.1184 - acc: 0.8300     \n",
      "Epoch 88/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1180 - acc: 0.8250     \n",
      "Epoch 89/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1177 - acc: 0.8250     \n",
      "Epoch 90/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1174 - acc: 0.8250     \n",
      "Epoch 91/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1171 - acc: 0.8250     \n",
      "Epoch 92/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1168 - acc: 0.8250     \n",
      "Epoch 93/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1164 - acc: 0.8300     \n",
      "Epoch 94/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1163 - acc: 0.8300     \n",
      "Epoch 95/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1162 - acc: 0.8250     \n",
      "Epoch 96/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1158 - acc: 0.8250     \n",
      "Epoch 97/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1153 - acc: 0.8300     \n",
      "Epoch 98/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1151 - acc: 0.8300     \n",
      "Epoch 99/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1150 - acc: 0.8300     \n",
      "Epoch 100/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1147 - acc: 0.8350     \n",
      "Epoch 101/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1145 - acc: 0.8300     \n",
      "Epoch 102/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1144 - acc: 0.8300     \n",
      "Epoch 103/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1195 - acc: 0.843 - 0s - loss: 0.1138 - acc: 0.8350     \n",
      "Epoch 104/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1137 - acc: 0.8350     \n",
      "Epoch 105/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1049 - acc: 0.843 - 0s - loss: 0.1137 - acc: 0.8350     \n",
      "Epoch 106/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1134 - acc: 0.8300     \n",
      "Epoch 107/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1132 - acc: 0.8300     \n",
      "Epoch 108/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1129 - acc: 0.8350     \n",
      "Epoch 109/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1126 - acc: 0.8350     \n",
      "Epoch 110/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1124 - acc: 0.8350     \n",
      "Epoch 111/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1124 - acc: 0.8300     \n",
      "Epoch 112/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1122 - acc: 0.8300     \n",
      "Epoch 113/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1120 - acc: 0.8300     \n",
      "Epoch 114/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1117 - acc: 0.8300     \n",
      "Epoch 115/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1116 - acc: 0.8300     \n",
      "Epoch 116/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1113 - acc: 0.8300     \n",
      "Epoch 117/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1112 - acc: 0.8350     \n",
      "Epoch 118/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1111 - acc: 0.8350     \n",
      "Epoch 119/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1109 - acc: 0.8350     \n",
      "Epoch 120/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1107 - acc: 0.8350     \n",
      "Epoch 121/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1106 - acc: 0.8300     \n",
      "Epoch 122/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1104 - acc: 0.8300     \n",
      "Epoch 123/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1189 - acc: 0.843 - 0s - loss: 0.1104 - acc: 0.8350     \n",
      "Epoch 124/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1101 - acc: 0.8300     \n",
      "Epoch 125/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0992 - acc: 0.843 - 0s - loss: 0.1100 - acc: 0.8300     \n",
      "Epoch 126/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1098 - acc: 0.8300     \n",
      "Epoch 127/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1098 - acc: 0.8300     \n",
      "Epoch 128/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1097 - acc: 0.8300     \n",
      "Epoch 129/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1227 - acc: 0.812 - 0s - loss: 0.1095 - acc: 0.8300     \n",
      "Epoch 130/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1095 - acc: 0.8300     \n",
      "Epoch 131/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1092 - acc: 0.8300     \n",
      "Epoch 132/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1091 - acc: 0.8300     \n",
      "Epoch 133/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1090 - acc: 0.8300     \n",
      "Epoch 134/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1090 - acc: 0.8300     \n",
      "Epoch 135/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1087 - acc: 0.8300     \n",
      "Epoch 136/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1085 - acc: 0.8300     \n",
      "Epoch 137/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1084 - acc: 0.8300     \n",
      "Epoch 138/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1084 - acc: 0.8300     \n",
      "Epoch 139/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1083 - acc: 0.8300     \n",
      "Epoch 140/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1081 - acc: 0.8300     \n",
      "Epoch 141/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1082 - acc: 0.8300     \n",
      "Epoch 142/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1080 - acc: 0.8300     \n",
      "Epoch 143/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1079 - acc: 0.8300     \n",
      "Epoch 144/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1080 - acc: 0.8300     \n",
      "Epoch 145/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1077 - acc: 0.8350     \n",
      "Epoch 146/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1076 - acc: 0.8300     \n",
      "Epoch 147/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1074 - acc: 0.8300     \n",
      "Epoch 148/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1074 - acc: 0.8300     \n",
      "Epoch 149/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1073 - acc: 0.8300     \n",
      "Epoch 150/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1072 - acc: 0.8300     \n",
      "Epoch 151/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1071 - acc: 0.8300     \n",
      "Epoch 152/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1070 - acc: 0.8300     \n",
      "Epoch 153/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1070 - acc: 0.8300     \n",
      "Epoch 154/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1070 - acc: 0.8300     \n",
      "Epoch 155/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1068 - acc: 0.8300     \n",
      "Epoch 156/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1068 - acc: 0.8300     \n",
      "Epoch 157/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1067 - acc: 0.8300     \n",
      "Epoch 158/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1069 - acc: 0.8300     \n",
      "Epoch 159/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1066 - acc: 0.8300     \n",
      "Epoch 160/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1065 - acc: 0.8300     \n",
      "Epoch 161/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1065 - acc: 0.8300     \n",
      "Epoch 162/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1065 - acc: 0.8300     \n",
      "Epoch 163/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1063 - acc: 0.8300     \n",
      "Epoch 164/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1063 - acc: 0.8300     \n",
      "Epoch 165/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1062 - acc: 0.8300     \n",
      "Epoch 166/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1062 - acc: 0.8300     \n",
      "Epoch 167/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1061 - acc: 0.8300     \n",
      "Epoch 168/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1061 - acc: 0.8300     \n",
      "Epoch 169/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1061 - acc: 0.8300     \n",
      "Epoch 170/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1062 - acc: 0.8300     \n",
      "Epoch 171/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.1060 - acc: 0.8300     \n",
      "Epoch 172/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1058 - acc: 0.8300     \n",
      "Epoch 173/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1058 - acc: 0.8300     \n",
      "Epoch 174/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1058 - acc: 0.8300     \n",
      "Epoch 175/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1057 - acc: 0.8300     \n",
      "Epoch 176/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1056 - acc: 0.8300     \n",
      "Epoch 177/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1056 - acc: 0.8300     \n",
      "Epoch 178/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1056 - acc: 0.8300     \n",
      "Epoch 179/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1057 - acc: 0.8300     \n",
      "Epoch 180/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1056 - acc: 0.8300     \n",
      "Epoch 181/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1054 - acc: 0.8300     \n",
      "Epoch 182/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1054 - acc: 0.8300     \n",
      "Epoch 183/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1054 - acc: 0.8350     \n",
      "Epoch 184/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1053 - acc: 0.8400     \n",
      "Epoch 185/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1053 - acc: 0.8400     \n",
      "Epoch 186/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1054 - acc: 0.8400     \n",
      "Epoch 187/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1054 - acc: 0.8450     \n",
      "Epoch 188/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1053 - acc: 0.8450     \n",
      "Epoch 189/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1053 - acc: 0.8450     \n",
      "Epoch 190/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1052 - acc: 0.8400     \n",
      "Epoch 191/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1051 - acc: 0.8400     \n",
      "Epoch 192/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1051 - acc: 0.8400     \n",
      "Epoch 193/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1051 - acc: 0.8400     \n",
      "Epoch 194/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1051 - acc: 0.8450     \n",
      "Epoch 195/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1053 - acc: 0.8450     \n",
      "Epoch 196/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1051 - acc: 0.8450     \n",
      "Epoch 197/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1051 - acc: 0.8450     \n",
      "Epoch 198/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1052 - acc: 0.8450     \n",
      "Epoch 199/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1051 - acc: 0.8450     \n",
      "Epoch 200/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1049 - acc: 0.8450     \n",
      "Epoch 201/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1048 - acc: 0.8450     \n",
      "Epoch 202/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1047 - acc: 0.8400     \n",
      "Epoch 203/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1049 - acc: 0.8300     \n",
      "Epoch 204/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1047 - acc: 0.8300     \n",
      "Epoch 205/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1048 - acc: 0.8300     \n",
      "Epoch 206/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1049 - acc: 0.8300     \n",
      "Epoch 207/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1048 - acc: 0.8300     \n",
      "Epoch 208/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1048 - acc: 0.8300     \n",
      "Epoch 209/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1047 - acc: 0.8300     \n",
      "Epoch 210/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1046 - acc: 0.8350     \n",
      "Epoch 211/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1045 - acc: 0.8400     \n",
      "Epoch 212/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1047 - acc: 0.8300     \n",
      "Epoch 213/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1046 - acc: 0.8300     \n",
      "Epoch 214/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1052 - acc: 0.8300     \n",
      "Epoch 215/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1050 - acc: 0.8350     \n",
      "Epoch 216/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1050 - acc: 0.8350     \n",
      "Epoch 217/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1049 - acc: 0.8350     \n",
      "Epoch 218/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1047 - acc: 0.8350     \n",
      "Epoch 219/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1046 - acc: 0.8350     \n",
      "Epoch 220/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1044 - acc: 0.8300     \n",
      "Epoch 221/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1043 - acc: 0.8400     \n",
      "Epoch 222/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1043 - acc: 0.8400     \n",
      "Epoch 223/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1043 - acc: 0.8400     \n",
      "Epoch 224/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1045 - acc: 0.8400     \n",
      "Epoch 225/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1043 - acc: 0.8400     \n",
      "Epoch 226/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1042 - acc: 0.8400     \n",
      "Epoch 227/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1042 - acc: 0.8400     \n",
      "Epoch 228/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1043 - acc: 0.8350     \n",
      "Epoch 229/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1043 - acc: 0.8300     \n",
      "Epoch 230/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1043 - acc: 0.8300     \n",
      "Epoch 231/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1043 - acc: 0.8350     \n",
      "Epoch 232/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1042 - acc: 0.8350     \n",
      "Epoch 233/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1042 - acc: 0.8400     \n",
      "Epoch 234/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1042 - acc: 0.8400     \n",
      "Epoch 235/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1041 - acc: 0.8400     \n",
      "Epoch 236/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1041 - acc: 0.8400     \n",
      "Epoch 237/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1041 - acc: 0.8400     \n",
      "Epoch 238/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1041 - acc: 0.8400     \n",
      "Epoch 239/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1041 - acc: 0.8400     \n",
      "Epoch 240/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1041 - acc: 0.8400     \n",
      "Epoch 241/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1040 - acc: 0.8400     \n",
      "Epoch 242/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1041 - acc: 0.8400     \n",
      "Epoch 243/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1041 - acc: 0.8350     \n",
      "Epoch 244/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1041 - acc: 0.8350     \n",
      "Epoch 245/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1041 - acc: 0.8350     \n",
      "Epoch 246/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1041 - acc: 0.8350     \n",
      "Epoch 247/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1042 - acc: 0.8350     \n",
      "Epoch 248/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1040 - acc: 0.8400     \n",
      "Epoch 249/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1041 - acc: 0.8400     \n",
      "Epoch 250/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1040 - acc: 0.8400     \n",
      "Epoch 251/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1041 - acc: 0.8400     \n",
      "Epoch 252/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1040 - acc: 0.8400     \n",
      "Epoch 253/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1040 - acc: 0.8400     \n",
      "Epoch 254/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8400     \n",
      "Epoch 255/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1041 - acc: 0.8400     \n",
      "Epoch 256/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8400     \n",
      "Epoch 257/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.1040 - acc: 0.8400     \n",
      "Epoch 258/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8400     \n",
      "Epoch 259/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8400     \n",
      "Epoch 260/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 261/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1040 - acc: 0.8350     \n",
      "Epoch 262/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8400     \n",
      "Epoch 263/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 264/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8350     \n",
      "Epoch 265/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8350     \n",
      "Epoch 266/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1040 - acc: 0.8400     \n",
      "Epoch 267/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8400     \n",
      "Epoch 268/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8400     \n",
      "Epoch 269/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8400     \n",
      "Epoch 270/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 271/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8400     \n",
      "Epoch 272/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 273/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8350     \n",
      "Epoch 274/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1040 - acc: 0.8400     \n",
      "Epoch 275/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1040 - acc: 0.8400     \n",
      "Epoch 276/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8400     \n",
      "Epoch 277/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1040 - acc: 0.8400     \n",
      "Epoch 278/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8400     \n",
      "Epoch 279/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8400     \n",
      "Epoch 280/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 281/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 282/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 283/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 284/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 285/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 286/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 287/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 288/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 289/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 290/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1040 - acc: 0.8350     \n",
      "Epoch 291/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1041 - acc: 0.8350     \n",
      "Epoch 292/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8400     \n",
      "Epoch 293/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1040 - acc: 0.8350     \n",
      "Epoch 294/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 295/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 296/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 297/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 298/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8400     \n",
      "Epoch 299/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 300/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 301/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 302/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 303/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 304/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 305/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8400     \n",
      "Epoch 306/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 307/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1014 - acc: 0.843 - 0s - loss: 0.1039 - acc: 0.8400     \n",
      "Epoch 308/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 309/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 310/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 311/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 312/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 313/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 314/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 315/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 316/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 317/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 318/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 319/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0988 - acc: 0.843 - 0s - loss: 0.1040 - acc: 0.8400     \n",
      "Epoch 320/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 321/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 322/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 323/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 324/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 325/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 326/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 327/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 328/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 329/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 330/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 331/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 332/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 333/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 334/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 335/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1310 - acc: 0.875 - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 336/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 337/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 338/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 339/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 340/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 341/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 342/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 343/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 344/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 345/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 346/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 347/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 348/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 349/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 350/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 351/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 352/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 353/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 354/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 355/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 356/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 357/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 358/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 359/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 360/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 361/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 362/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 363/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 364/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 365/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 366/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0973 - acc: 0.875 - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 367/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8350     \n",
      "Epoch 368/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8350     \n",
      "Epoch 369/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 370/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1040 - acc: 0.8350     \n",
      "Epoch 371/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 372/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 373/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 374/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 375/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 376/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 377/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 378/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 379/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 380/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 381/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 382/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 383/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 384/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8300     \n",
      "Epoch 385/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8300     \n",
      "Epoch 386/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 387/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 388/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 389/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 390/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 391/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 392/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 393/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8300     \n",
      "Epoch 394/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 395/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8350     \n",
      "Epoch 396/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 397/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 398/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 399/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 400/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 401/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 402/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 403/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 404/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 405/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8300     \n",
      "Epoch 406/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 407/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 408/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 409/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 410/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 411/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 412/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 413/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 414/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 415/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8400     \n",
      "Epoch 416/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8300     \n",
      "Epoch 417/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 418/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 419/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 420/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 421/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 422/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 423/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 424/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 425/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 426/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 427/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8300     \n",
      "Epoch 428/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 429/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8300     \n",
      "Epoch 430/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 431/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 432/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 433/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 434/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 435/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 436/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 437/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 438/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 439/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1093 - acc: 0.812 - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 440/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8400     \n",
      "Epoch 441/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 442/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 443/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 444/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 445/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 446/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 447/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 448/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 449/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8450     \n",
      "Epoch 450/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 451/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 452/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 453/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 454/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 455/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 456/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 457/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8300     \n",
      "Epoch 458/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 459/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 460/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 461/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 462/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 463/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 464/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 465/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 466/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 467/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 468/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 469/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 470/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 471/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 472/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 473/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 474/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 475/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 476/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 477/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 478/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 479/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 480/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 481/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 482/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 483/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 484/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 485/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 486/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 487/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 488/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 489/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 490/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 491/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 492/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 493/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 494/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 495/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 496/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 497/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 498/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 499/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 500/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 501/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 502/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 503/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 504/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 505/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 506/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 507/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 508/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 509/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 510/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 511/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 512/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 513/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 514/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 515/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 516/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 517/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 518/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 519/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 520/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 521/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 522/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 523/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 524/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 525/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 526/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 527/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8300     \n",
      "Epoch 528/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 529/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 530/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 531/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 532/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 533/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 534/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 535/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 536/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0789 - acc: 0.875 - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 537/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 538/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 539/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 540/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 541/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 542/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 543/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 544/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 545/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 546/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1040 - acc: 0.8350     \n",
      "Epoch 547/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 548/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 549/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 550/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 551/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 552/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 553/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 554/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 555/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 556/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 557/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 558/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 559/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 560/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8300     \n",
      "Epoch 561/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 562/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 563/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 564/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 565/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 566/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 567/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 568/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 569/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 570/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 571/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 572/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 573/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 574/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 575/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8300     \n",
      "Epoch 576/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 577/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 578/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 579/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 580/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8300     \n",
      "Epoch 581/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 582/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 583/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 584/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 585/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 586/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1029 - acc: 0.843 - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 587/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 588/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 589/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 590/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 591/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 592/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 593/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 594/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 595/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 596/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 597/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 598/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 599/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 600/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8300     \n",
      "Epoch 601/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 602/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 603/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 604/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 605/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 606/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 607/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 608/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 609/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 610/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 611/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 612/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 613/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 614/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 615/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 616/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 617/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 618/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 619/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 620/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 621/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 622/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 623/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 624/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 625/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 626/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 627/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 628/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 629/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 630/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 631/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 632/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 633/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 634/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 635/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 636/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 637/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 638/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 639/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 640/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 641/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 642/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 643/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 644/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 645/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 646/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 647/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 648/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 649/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 650/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 651/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 652/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 653/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 654/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 655/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 656/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 657/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 658/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 659/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 660/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 661/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 662/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 663/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 664/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 665/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 666/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 667/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 668/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 669/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 670/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 671/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 672/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 673/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 674/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 675/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 676/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 677/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 678/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 679/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 680/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 681/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 682/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 683/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 684/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 685/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 686/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 687/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 688/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 689/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 690/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 691/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 692/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 693/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 694/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 695/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 696/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 697/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 698/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 699/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 700/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1040 - acc: 0.8450     \n",
      "Epoch 701/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8450     \n",
      "Epoch 702/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 703/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 704/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 705/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 706/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 707/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 708/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 709/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 710/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 711/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 712/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 713/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 714/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 715/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 716/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 717/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 718/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 719/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 720/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 721/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 722/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 723/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 724/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 725/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 726/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 727/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 728/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 729/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 730/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 731/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 732/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 733/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8300     \n",
      "Epoch 734/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 735/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 736/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 737/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 738/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 739/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 740/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1002 - acc: 0.843 - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 741/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 742/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 743/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 744/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 745/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 746/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 747/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 748/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 749/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 750/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 751/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 752/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 753/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 754/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 755/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 756/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 757/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 758/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 759/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 760/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 761/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 762/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 763/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 764/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 765/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 766/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 767/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 768/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 769/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 770/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8450     \n",
      "Epoch 771/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 772/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 773/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 774/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 775/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 776/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 777/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 778/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 779/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 780/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 781/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 782/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 783/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8300     \n",
      "Epoch 784/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 785/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 786/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 787/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 788/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 789/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8300     \n",
      "Epoch 790/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 791/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 792/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 793/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 794/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 795/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 796/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 797/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 798/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0824 - acc: 0.875 - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 799/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1336 - acc: 0.812 - 0s - loss: 0.1035 - acc: 0.8400     \n",
      "Epoch 800/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 801/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 802/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 803/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 804/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 805/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 806/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 807/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 808/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 809/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8400     \n",
      "Epoch 810/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 811/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8350     \n",
      "Epoch 812/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 813/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8350     \n",
      "Epoch 814/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8350     \n",
      "Epoch 815/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1042 - acc: 0.8350     \n",
      "Epoch 816/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 817/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 818/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 819/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 820/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 821/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 822/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 823/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 824/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 825/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 826/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 827/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 828/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 829/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 830/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 831/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 832/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 833/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 834/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 835/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 836/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 837/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 838/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 839/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 840/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 841/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 842/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 843/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8300     \n",
      "Epoch 844/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 845/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 846/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 847/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 848/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 849/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 850/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 851/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 852/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 853/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 854/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 855/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 856/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 857/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 858/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 859/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 860/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 861/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 862/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 863/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 864/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 865/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 866/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 867/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 868/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 869/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 870/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 871/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 872/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 873/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 874/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 875/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 876/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 877/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 878/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 879/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 880/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 881/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 882/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 883/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 884/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 885/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 886/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 887/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8450     \n",
      "Epoch 888/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8400     \n",
      "Epoch 889/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8450     \n",
      "Epoch 890/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1041 - acc: 0.8450     \n",
      "Epoch 891/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8450     \n",
      "Epoch 892/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 893/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 894/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 895/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 896/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 897/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1071 - acc: 0.843 - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 898/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 899/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 900/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 901/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8300     \n",
      "Epoch 902/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 903/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 904/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 905/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 906/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 907/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 908/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8300     \n",
      "Epoch 909/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 910/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8300     \n",
      "Epoch 911/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 912/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 913/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 914/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 915/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1236 - acc: 0.781 - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 916/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 917/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 918/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 919/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 920/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 921/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 922/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1054 - acc: 0.843 - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 923/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 924/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 925/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 926/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 927/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 928/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 929/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 930/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 931/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 932/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 933/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 934/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 935/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 936/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 937/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 938/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 939/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 940/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 941/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 942/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 943/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 944/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 945/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 946/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 947/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 948/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 949/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 950/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 951/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 952/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1102 - acc: 0.843 - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 953/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 954/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 955/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 956/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 957/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 958/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 959/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 960/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 961/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8300     \n",
      "Epoch 962/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 963/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 964/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 965/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 966/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 967/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 968/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 969/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 970/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 971/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 972/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8400     \n",
      "Epoch 973/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8400     \n",
      "Epoch 974/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 975/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 976/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 977/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 978/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 979/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 980/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 981/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 982/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 983/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 984/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 985/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 986/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 987/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 988/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0767 - acc: 0.843 - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 989/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 990/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 991/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 992/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 993/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 994/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 995/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 996/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 997/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8300     \n",
      "Epoch 998/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 999/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8400     \n",
      "Epoch 1000/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 1001/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1002/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1003/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1004/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1005/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1006/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1007/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 1008/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 1009/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 1010/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1011/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 1012/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1013/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1014/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8350     \n",
      "Epoch 1015/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 1016/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 1017/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1018/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1019/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1020/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1021/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 1022/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 1023/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 1024/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 1025/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 1026/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 1027/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 1028/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 1029/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1030/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1031/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8300     \n",
      "Epoch 1032/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 1033/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1034/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1035/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1036/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 1037/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1038/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1039/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1040/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1041/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 1042/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 1043/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1044/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 1045/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1046/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1047/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1048/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1049/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1050/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1051/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 1052/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 1053/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 1054/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1055/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1056/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1057/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8400     \n",
      "Epoch 1058/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1059/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1060/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1061/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 1062/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1063/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1039 - acc: 0.8350     \n",
      "Epoch 1064/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 1065/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1066/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1067/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1068/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1069/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 1070/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1071/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1072/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1073/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1074/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1075/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1076/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8400     \n",
      "Epoch 1077/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8400     \n",
      "Epoch 1078/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1079/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8400     \n",
      "Epoch 1080/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1081/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1082/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1083/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1084/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1085/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1086/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1087/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 1088/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 1089/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 1090/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1091/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1092/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 1093/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1094/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1095/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8300     \n",
      "Epoch 1096/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 1097/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1098/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1099/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8300     \n",
      "Epoch 1100/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1101/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1102/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1103/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 1104/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1105/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1106/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1107/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1108/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 1109/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1110/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1111/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1112/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1113/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1114/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8400     \n",
      "Epoch 1115/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1116/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 1117/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1118/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1119/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1120/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1121/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 1122/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8300     \n",
      "Epoch 1123/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1124/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1125/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 1126/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1127/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1128/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1129/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1130/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1131/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1132/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1133/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1134/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1135/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 1136/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8400     \n",
      "Epoch 1137/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1138/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1139/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1140/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1141/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1142/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1143/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1144/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1145/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1146/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1147/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1148/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1149/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8300     \n",
      "Epoch 1150/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1151/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0886 - acc: 0.843 - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1152/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1153/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1154/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1155/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1156/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1157/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1158/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1159/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1160/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1225 - acc: 0.750 - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1161/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8400     \n",
      "Epoch 1162/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8450     \n",
      "Epoch 1163/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8450     \n",
      "Epoch 1164/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1165/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1166/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1167/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8400     \n",
      "Epoch 1168/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1169/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1170/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1171/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8400     \n",
      "Epoch 1172/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 1173/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8450     \n",
      "Epoch 1174/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8400     \n",
      "Epoch 1175/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1176/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 1177/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1178/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 1179/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1180/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8300     \n",
      "Epoch 1181/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1182/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1183/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1184/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1185/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8300     \n",
      "Epoch 1186/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1187/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1188/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1189/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1190/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1191/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1192/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1193/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1194/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1195/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1196/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1197/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1198/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 1199/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8300     \n",
      "Epoch 1200/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 1201/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1202/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8350     \n",
      "Epoch 1203/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1037 - acc: 0.8400     \n",
      "Epoch 1204/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8400     \n",
      "Epoch 1205/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8350     \n",
      "Epoch 1206/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1207/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1038 - acc: 0.8350     \n",
      "Epoch 1208/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1209/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1210/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1211/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1212/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1213/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8300     \n",
      "Epoch 1214/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1215/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1216/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1217/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1218/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1219/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1220/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1221/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1222/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1223/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1224/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1225/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1226/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1227/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1228/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1229/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1230/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1231/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1232/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1233/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1234/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1235/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1236/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1237/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1238/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1239/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1240/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1241/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1242/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1243/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1244/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8400     \n",
      "Epoch 1245/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8450     \n",
      "Epoch 1246/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1247/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1248/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8400     \n",
      "Epoch 1249/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1250/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1251/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1252/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1253/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1254/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1255/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1256/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1257/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1258/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1259/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8400     \n",
      "Epoch 1260/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1261/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1262/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8400     \n",
      "Epoch 1263/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1264/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8400     \n",
      "Epoch 1265/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1266/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1267/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8350     \n",
      "Epoch 1268/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8350     \n",
      "Epoch 1269/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1270/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1271/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1272/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1273/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1274/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1275/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1276/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1277/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8400     \n",
      "Epoch 1278/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1279/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1280/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8450     \n",
      "Epoch 1281/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8400     \n",
      "Epoch 1282/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8350     \n",
      "Epoch 1283/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1284/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1285/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1286/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1287/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8400     \n",
      "Epoch 1288/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1289/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1290/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1291/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1292/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1293/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1294/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1295/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1296/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1297/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1298/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1299/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1300/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1301/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1302/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1303/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8350     \n",
      "Epoch 1304/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1305/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8300     \n",
      "Epoch 1306/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1307/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1287 - acc: 0.843 - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1308/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8300     \n",
      "Epoch 1309/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1310/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1311/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1312/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8350     \n",
      "Epoch 1313/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1314/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8350     \n",
      "Epoch 1315/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1316/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1317/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1318/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8350     \n",
      "Epoch 1319/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8350     \n",
      "Epoch 1320/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1321/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8400     \n",
      "Epoch 1322/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8400     \n",
      "Epoch 1323/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8350     \n",
      "Epoch 1324/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8350     \n",
      "Epoch 1325/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1326/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1327/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8350     \n",
      "Epoch 1328/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1329/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8350     \n",
      "Epoch 1330/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8350     \n",
      "Epoch 1331/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0730 - acc: 0.843 - 0s - loss: 0.1030 - acc: 0.8350     \n",
      "Epoch 1332/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1333/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1334/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1335/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1336/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1337/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1338/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8400     \n",
      "Epoch 1339/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8450     \n",
      "Epoch 1340/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8450     \n",
      "Epoch 1341/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1034 - acc: 0.8450     \n",
      "Epoch 1342/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8400     \n",
      "Epoch 1343/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1035 - acc: 0.8400     \n",
      "Epoch 1344/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1036 - acc: 0.8400     \n",
      "Epoch 1345/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8450     \n",
      "Epoch 1346/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8450     \n",
      "Epoch 1347/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8400     \n",
      "Epoch 1348/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8350     \n",
      "Epoch 1349/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8350     \n",
      "Epoch 1350/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8400     \n",
      "Epoch 1351/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1352/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8400     \n",
      "Epoch 1353/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8400     \n",
      "Epoch 1354/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8400     \n",
      "Epoch 1355/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8400     \n",
      "Epoch 1356/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0905 - acc: 0.875 - 0s - loss: 0.1030 - acc: 0.8350     \n",
      "Epoch 1357/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8350     \n",
      "Epoch 1358/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8350     \n",
      "Epoch 1359/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8350     \n",
      "Epoch 1360/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8400     \n",
      "Epoch 1361/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8400     \n",
      "Epoch 1362/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8350     \n",
      "Epoch 1363/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1364/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8350     \n",
      "Epoch 1365/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1366/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8350     \n",
      "Epoch 1367/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8350     \n",
      "Epoch 1368/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8350     \n",
      "Epoch 1369/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8350     \n",
      "Epoch 1370/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8400     \n",
      "Epoch 1371/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8350     \n",
      "Epoch 1372/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8450     \n",
      "Epoch 1373/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8400     \n",
      "Epoch 1374/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1032 - acc: 0.8450     \n",
      "Epoch 1375/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8450     \n",
      "Epoch 1376/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8400     \n",
      "Epoch 1377/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8350     \n",
      "Epoch 1378/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8350     \n",
      "Epoch 1379/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8350     \n",
      "Epoch 1380/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0632 - acc: 0.906 - 0s - loss: 0.1029 - acc: 0.8400     \n",
      "Epoch 1381/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8350     \n",
      "Epoch 1382/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8350     \n",
      "Epoch 1383/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8350     \n",
      "Epoch 1384/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1031 - acc: 0.8350     \n",
      "Epoch 1385/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8350     \n",
      "Epoch 1386/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8350     \n",
      "Epoch 1387/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8350     \n",
      "Epoch 1388/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8350     \n",
      "Epoch 1389/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8350     \n",
      "Epoch 1390/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8350     \n",
      "Epoch 1391/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8350     \n",
      "Epoch 1392/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1033 - acc: 0.8450     \n",
      "Epoch 1393/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1030 - acc: 0.8400     \n",
      "Epoch 1394/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8450     \n",
      "Epoch 1395/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8400     \n",
      "Epoch 1396/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8350     \n",
      "Epoch 1397/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8350     \n",
      "Epoch 1398/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8400     \n",
      "Epoch 1399/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8400     \n",
      "Epoch 1400/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8350     \n",
      "Epoch 1401/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8350     \n",
      "Epoch 1402/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8350     \n",
      "Epoch 1403/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8350     \n",
      "Epoch 1404/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8350     \n",
      "Epoch 1405/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1029 - acc: 0.8350     \n",
      "Epoch 1406/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8350     \n",
      "Epoch 1407/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8350     \n",
      "Epoch 1408/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8350     \n",
      "Epoch 1409/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8350     \n",
      "Epoch 1410/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8350     \n",
      "Epoch 1411/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8400     \n",
      "Epoch 1412/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8400     \n",
      "Epoch 1413/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8400     \n",
      "Epoch 1414/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8400     \n",
      "Epoch 1415/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8400     \n",
      "Epoch 1416/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8400     \n",
      "Epoch 1417/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8400     \n",
      "Epoch 1418/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8400     \n",
      "Epoch 1419/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8400     \n",
      "Epoch 1420/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8350     \n",
      "Epoch 1421/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8350     \n",
      "Epoch 1422/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8350     \n",
      "Epoch 1423/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8400     \n",
      "Epoch 1424/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8400     \n",
      "Epoch 1425/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8400     \n",
      "Epoch 1426/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8400     \n",
      "Epoch 1427/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8400     \n",
      "Epoch 1428/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8450     \n",
      "Epoch 1429/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8450     \n",
      "Epoch 1430/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8350     \n",
      "Epoch 1431/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8400     \n",
      "Epoch 1432/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8400     \n",
      "Epoch 1433/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8450     \n",
      "Epoch 1434/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1025 - acc: 0.8400     \n",
      "Epoch 1435/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8400     \n",
      "Epoch 1436/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8400     \n",
      "Epoch 1437/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8350     \n",
      "Epoch 1438/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8400     \n",
      "Epoch 1439/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8400     \n",
      "Epoch 1440/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8400     \n",
      "Epoch 1441/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8400     \n",
      "Epoch 1442/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8400     \n",
      "Epoch 1443/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1025 - acc: 0.8450     \n",
      "Epoch 1444/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8400     \n",
      "Epoch 1445/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8450     \n",
      "Epoch 1446/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1028 - acc: 0.8400     \n",
      "Epoch 1447/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8400     \n",
      "Epoch 1448/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8400     \n",
      "Epoch 1449/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8450     \n",
      "Epoch 1450/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8450     \n",
      "Epoch 1451/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8400     \n",
      "Epoch 1452/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1025 - acc: 0.8400     \n",
      "Epoch 1453/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8400     \n",
      "Epoch 1454/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1025 - acc: 0.8400     \n",
      "Epoch 1455/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8400     \n",
      "Epoch 1456/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1025 - acc: 0.8400     \n",
      "Epoch 1457/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1025 - acc: 0.8400     \n",
      "Epoch 1458/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1025 - acc: 0.8400     \n",
      "Epoch 1459/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8350     \n",
      "Epoch 1460/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1025 - acc: 0.8350     \n",
      "Epoch 1461/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8350     \n",
      "Epoch 1462/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8350     \n",
      "Epoch 1463/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8350     \n",
      "Epoch 1464/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1025 - acc: 0.8350     \n",
      "Epoch 1465/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1024 - acc: 0.8350     \n",
      "Epoch 1466/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1024 - acc: 0.8350     \n",
      "Epoch 1467/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1024 - acc: 0.8350     \n",
      "Epoch 1468/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1027 - acc: 0.8350     \n",
      "Epoch 1469/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8400     \n",
      "Epoch 1470/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8400     \n",
      "Epoch 1471/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1024 - acc: 0.8400     \n",
      "Epoch 1472/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1024 - acc: 0.8400     \n",
      "Epoch 1473/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1026 - acc: 0.8350     \n",
      "Epoch 1474/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1025 - acc: 0.8350     \n",
      "Epoch 1475/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1024 - acc: 0.8350     \n",
      "Epoch 1476/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1023 - acc: 0.8350     \n",
      "Epoch 1477/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1023 - acc: 0.8350     \n",
      "Epoch 1478/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1024 - acc: 0.8400     \n",
      "Epoch 1479/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1023 - acc: 0.8400     \n",
      "Epoch 1480/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1023 - acc: 0.8400     \n",
      "Epoch 1481/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1023 - acc: 0.8400     \n",
      "Epoch 1482/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1023 - acc: 0.8400     \n",
      "Epoch 1483/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1023 - acc: 0.8400     \n",
      "Epoch 1484/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1023 - acc: 0.8400     \n",
      "Epoch 1485/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1024 - acc: 0.8400     \n",
      "Epoch 1486/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1024 - acc: 0.8400     \n",
      "Epoch 1487/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1023 - acc: 0.8400     \n",
      "Epoch 1488/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1023 - acc: 0.8400     \n",
      "Epoch 1489/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1023 - acc: 0.8400     \n",
      "Epoch 1490/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1022 - acc: 0.8450     \n",
      "Epoch 1491/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1022 - acc: 0.8400     \n",
      "Epoch 1492/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1022 - acc: 0.8400     \n",
      "Epoch 1493/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1022 - acc: 0.8400     \n",
      "Epoch 1494/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1021 - acc: 0.8400     \n",
      "Epoch 1495/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1022 - acc: 0.8400     \n",
      "Epoch 1496/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1023 - acc: 0.8450     \n",
      "Epoch 1497/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1024 - acc: 0.8400     \n",
      "Epoch 1498/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1023 - acc: 0.8450     \n",
      "Epoch 1499/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1023 - acc: 0.8450     \n",
      "Epoch 1500/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1022 - acc: 0.8400     \n",
      "Epoch 1501/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1021 - acc: 0.8450     \n",
      "Epoch 1502/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1021 - acc: 0.8400     \n",
      "Epoch 1503/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1022 - acc: 0.8400     \n",
      "Epoch 1504/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1021 - acc: 0.8400     \n",
      "Epoch 1505/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1020 - acc: 0.8450     \n",
      "Epoch 1506/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1021 - acc: 0.8400     \n",
      "Epoch 1507/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1022 - acc: 0.8400     \n",
      "Epoch 1508/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1022 - acc: 0.8400     \n",
      "Epoch 1509/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1022 - acc: 0.8450     \n",
      "Epoch 1510/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1020 - acc: 0.8400     \n",
      "Epoch 1511/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1021 - acc: 0.8400     \n",
      "Epoch 1512/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1022 - acc: 0.8350     \n",
      "Epoch 1513/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1023 - acc: 0.8400     \n",
      "Epoch 1514/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1025 - acc: 0.8400     \n",
      "Epoch 1515/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1023 - acc: 0.8400     \n",
      "Epoch 1516/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1021 - acc: 0.8350     \n",
      "Epoch 1517/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1020 - acc: 0.8400     \n",
      "Epoch 1518/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1021 - acc: 0.8400     \n",
      "Epoch 1519/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1020 - acc: 0.8400     \n",
      "Epoch 1520/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1020 - acc: 0.8400     \n",
      "Epoch 1521/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1020 - acc: 0.8400     \n",
      "Epoch 1522/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1021 - acc: 0.8400     \n",
      "Epoch 1523/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1019 - acc: 0.8400     \n",
      "Epoch 1524/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1020 - acc: 0.8400     \n",
      "Epoch 1525/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1019 - acc: 0.8400     \n",
      "Epoch 1526/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1019 - acc: 0.8400     \n",
      "Epoch 1527/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1019 - acc: 0.8400     \n",
      "Epoch 1528/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1019 - acc: 0.8450     \n",
      "Epoch 1529/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1019 - acc: 0.8450     \n",
      "Epoch 1530/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1019 - acc: 0.8450     \n",
      "Epoch 1531/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1019 - acc: 0.8400     \n",
      "Epoch 1532/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1020 - acc: 0.8400     \n",
      "Epoch 1533/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1019 - acc: 0.8400     \n",
      "Epoch 1534/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.1019 - acc: 0.8400     \n",
      "Epoch 1535/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1020 - acc: 0.8400     \n",
      "Epoch 1536/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1018 - acc: 0.8400     \n",
      "Epoch 1537/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1018 - acc: 0.8450     \n",
      "Epoch 1538/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1018 - acc: 0.8450     \n",
      "Epoch 1539/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1017 - acc: 0.8450     \n",
      "Epoch 1540/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1021 - acc: 0.8400     \n",
      "Epoch 1541/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1021 - acc: 0.8450     \n",
      "Epoch 1542/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1020 - acc: 0.8400     \n",
      "Epoch 1543/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1019 - acc: 0.8450     \n",
      "Epoch 1544/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1018 - acc: 0.8450     \n",
      "Epoch 1545/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1018 - acc: 0.8450     \n",
      "Epoch 1546/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1017 - acc: 0.8450     \n",
      "Epoch 1547/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1017 - acc: 0.8450     \n",
      "Epoch 1548/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1017 - acc: 0.8400     \n",
      "Epoch 1549/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1017 - acc: 0.8400     \n",
      "Epoch 1550/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1017 - acc: 0.8400     \n",
      "Epoch 1551/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1017 - acc: 0.8350     \n",
      "Epoch 1552/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1017 - acc: 0.8350     \n",
      "Epoch 1553/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1016 - acc: 0.8400     \n",
      "Epoch 1554/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1016 - acc: 0.8400     \n",
      "Epoch 1555/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1016 - acc: 0.8400     \n",
      "Epoch 1556/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1016 - acc: 0.8450     \n",
      "Epoch 1557/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1016 - acc: 0.8450     \n",
      "Epoch 1558/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1017 - acc: 0.8450     \n",
      "Epoch 1559/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1015 - acc: 0.8450     \n",
      "Epoch 1560/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1015 - acc: 0.8400     \n",
      "Epoch 1561/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1016 - acc: 0.8400     \n",
      "Epoch 1562/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1015 - acc: 0.8450     \n",
      "Epoch 1563/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1017 - acc: 0.8450     \n",
      "Epoch 1564/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1016 - acc: 0.8450     \n",
      "Epoch 1565/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1016 - acc: 0.8450     \n",
      "Epoch 1566/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1015 - acc: 0.8450     \n",
      "Epoch 1567/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1014 - acc: 0.8450     \n",
      "Epoch 1568/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1015 - acc: 0.8450     \n",
      "Epoch 1569/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1015 - acc: 0.8400     \n",
      "Epoch 1570/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1015 - acc: 0.8450     \n",
      "Epoch 1571/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1015 - acc: 0.8450     \n",
      "Epoch 1572/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1014 - acc: 0.8450     \n",
      "Epoch 1573/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1014 - acc: 0.8450     \n",
      "Epoch 1574/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1014 - acc: 0.8450     \n",
      "Epoch 1575/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1014 - acc: 0.8450     \n",
      "Epoch 1576/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1017 - acc: 0.8400     \n",
      "Epoch 1577/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1013 - acc: 0.8450     \n",
      "Epoch 1578/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1015 - acc: 0.8450     \n",
      "Epoch 1579/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1013 - acc: 0.8450     \n",
      "Epoch 1580/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1013 - acc: 0.8450     \n",
      "Epoch 1581/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1013 - acc: 0.8450     \n",
      "Epoch 1582/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1013 - acc: 0.8450     \n",
      "Epoch 1583/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1014 - acc: 0.8450     \n",
      "Epoch 1584/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1013 - acc: 0.8450     \n",
      "Epoch 1585/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1013 - acc: 0.8450     \n",
      "Epoch 1586/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1011 - acc: 0.8450     \n",
      "Epoch 1587/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1013 - acc: 0.8450     \n",
      "Epoch 1588/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1013 - acc: 0.8400     \n",
      "Epoch 1589/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1013 - acc: 0.8400     \n",
      "Epoch 1590/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1012 - acc: 0.8400     \n",
      "Epoch 1591/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1012 - acc: 0.8450     \n",
      "Epoch 1592/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1012 - acc: 0.8450     \n",
      "Epoch 1593/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1012 - acc: 0.8450     \n",
      "Epoch 1594/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1012 - acc: 0.8450     \n",
      "Epoch 1595/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1014 - acc: 0.8500     \n",
      "Epoch 1596/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1013 - acc: 0.8400     \n",
      "Epoch 1597/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1012 - acc: 0.8450     \n",
      "Epoch 1598/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1011 - acc: 0.8450     \n",
      "Epoch 1599/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1011 - acc: 0.8450     \n",
      "Epoch 1600/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1011 - acc: 0.8450     \n",
      "Epoch 1601/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1013 - acc: 0.8450     \n",
      "Epoch 1602/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1010 - acc: 0.8450     \n",
      "Epoch 1603/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1010 - acc: 0.8450     \n",
      "Epoch 1604/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1010 - acc: 0.8450     \n",
      "Epoch 1605/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1010 - acc: 0.8450     \n",
      "Epoch 1606/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1010 - acc: 0.8450     \n",
      "Epoch 1607/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1009 - acc: 0.8450     \n",
      "Epoch 1608/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1009 - acc: 0.8450     \n",
      "Epoch 1609/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1011 - acc: 0.8450     \n",
      "Epoch 1610/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1009 - acc: 0.8450     \n",
      "Epoch 1611/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1009 - acc: 0.8450     \n",
      "Epoch 1612/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1009 - acc: 0.8450     \n",
      "Epoch 1613/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1009 - acc: 0.8450     \n",
      "Epoch 1614/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1009 - acc: 0.8450     \n",
      "Epoch 1615/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1009 - acc: 0.8450     \n",
      "Epoch 1616/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1009 - acc: 0.8450     \n",
      "Epoch 1617/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1008 - acc: 0.8450     \n",
      "Epoch 1618/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1007 - acc: 0.8450     \n",
      "Epoch 1619/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.1009 - acc: 0.8450     \n",
      "Epoch 1620/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1009 - acc: 0.8450     \n",
      "Epoch 1621/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1011 - acc: 0.8450     \n",
      "Epoch 1622/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1011 - acc: 0.8400     \n",
      "Epoch 1623/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1008 - acc: 0.8450     \n",
      "Epoch 1624/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1008 - acc: 0.8450     \n",
      "Epoch 1625/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1007 - acc: 0.8450     \n",
      "Epoch 1626/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1006 - acc: 0.8450     \n",
      "Epoch 1627/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1006 - acc: 0.8450     \n",
      "Epoch 1628/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1006 - acc: 0.8450     \n",
      "Epoch 1629/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1006 - acc: 0.8450     \n",
      "Epoch 1630/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1004 - acc: 0.8450     \n",
      "Epoch 1631/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1005 - acc: 0.8400     \n",
      "Epoch 1632/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1008 - acc: 0.8400     \n",
      "Epoch 1633/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1007 - acc: 0.8400     \n",
      "Epoch 1634/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1006 - acc: 0.8500     \n",
      "Epoch 1635/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1005 - acc: 0.8450     \n",
      "Epoch 1636/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1004 - acc: 0.8450     \n",
      "Epoch 1637/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1004 - acc: 0.8450     \n",
      "Epoch 1638/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1006 - acc: 0.8450     \n",
      "Epoch 1639/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1003 - acc: 0.8450     \n",
      "Epoch 1640/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1004 - acc: 0.8450     \n",
      "Epoch 1641/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1003 - acc: 0.8450     \n",
      "Epoch 1642/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1003 - acc: 0.8450     \n",
      "Epoch 1643/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1004 - acc: 0.8450     \n",
      "Epoch 1644/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1004 - acc: 0.8450     \n",
      "Epoch 1645/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1003 - acc: 0.8450     \n",
      "Epoch 1646/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0479 - acc: 0.937 - 0s - loss: 0.1003 - acc: 0.8450     \n",
      "Epoch 1647/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1002 - acc: 0.8450     \n",
      "Epoch 1648/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1572 - acc: 0.687 - 0s - loss: 0.1002 - acc: 0.8450     \n",
      "Epoch 1649/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1003 - acc: 0.8450     \n",
      "Epoch 1650/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1002 - acc: 0.8450     \n",
      "Epoch 1651/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1001 - acc: 0.8450     \n",
      "Epoch 1652/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1001 - acc: 0.8450     \n",
      "Epoch 1653/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1001 - acc: 0.8450     \n",
      "Epoch 1654/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1001 - acc: 0.8450     \n",
      "Epoch 1655/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1002 - acc: 0.8450     \n",
      "Epoch 1656/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1000 - acc: 0.8450     \n",
      "Epoch 1657/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1002 - acc: 0.8450     \n",
      "Epoch 1658/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1001 - acc: 0.8450     \n",
      "Epoch 1659/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1001 - acc: 0.8450     \n",
      "Epoch 1660/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1002 - acc: 0.8450     \n",
      "Epoch 1661/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1000 - acc: 0.8450     \n",
      "Epoch 1662/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1001 - acc: 0.8450     \n",
      "Epoch 1663/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1000 - acc: 0.8450     \n",
      "Epoch 1664/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1001 - acc: 0.8450     \n",
      "Epoch 1665/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1001 - acc: 0.8450     \n",
      "Epoch 1666/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1000 - acc: 0.8450     \n",
      "Epoch 1667/2500\n",
      "200/200 [==============================] - 0s - loss: 0.1000 - acc: 0.8400     \n",
      "Epoch 1668/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0999 - acc: 0.8450     \n",
      "Epoch 1669/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0999 - acc: 0.8450     \n",
      "Epoch 1670/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0999 - acc: 0.8450     \n",
      "Epoch 1671/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0998 - acc: 0.8450     \n",
      "Epoch 1672/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0997 - acc: 0.8450     \n",
      "Epoch 1673/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0998 - acc: 0.8450     \n",
      "Epoch 1674/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0997 - acc: 0.8450     \n",
      "Epoch 1675/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0997 - acc: 0.8450     \n",
      "Epoch 1676/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0996 - acc: 0.8450     \n",
      "Epoch 1677/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0996 - acc: 0.8450     \n",
      "Epoch 1678/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0997 - acc: 0.8450     \n",
      "Epoch 1679/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0996 - acc: 0.8450     \n",
      "Epoch 1680/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0997 - acc: 0.8450     \n",
      "Epoch 1681/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0997 - acc: 0.8450     \n",
      "Epoch 1682/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0995 - acc: 0.8450     \n",
      "Epoch 1683/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0995 - acc: 0.8450     \n",
      "Epoch 1684/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0995 - acc: 0.8450     \n",
      "Epoch 1685/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0995 - acc: 0.8500     \n",
      "Epoch 1686/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0996 - acc: 0.8500     \n",
      "Epoch 1687/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0995 - acc: 0.8450     \n",
      "Epoch 1688/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0993 - acc: 0.8450     \n",
      "Epoch 1689/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0996 - acc: 0.8450     \n",
      "Epoch 1690/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0993 - acc: 0.8450     \n",
      "Epoch 1691/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0993 - acc: 0.8450     \n",
      "Epoch 1692/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0993 - acc: 0.8450     \n",
      "Epoch 1693/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0996 - acc: 0.8450     \n",
      "Epoch 1694/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0994 - acc: 0.8450     \n",
      "Epoch 1695/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0993 - acc: 0.8450     \n",
      "Epoch 1696/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0992 - acc: 0.8450     \n",
      "Epoch 1697/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0995 - acc: 0.8500     \n",
      "Epoch 1698/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0994 - acc: 0.8500     \n",
      "Epoch 1699/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0991 - acc: 0.8450     \n",
      "Epoch 1700/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0991 - acc: 0.8450     \n",
      "Epoch 1701/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0991 - acc: 0.8450     \n",
      "Epoch 1702/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0990 - acc: 0.8450     \n",
      "Epoch 1703/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.0991 - acc: 0.8450     \n",
      "Epoch 1704/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0990 - acc: 0.8450     \n",
      "Epoch 1705/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0990 - acc: 0.8450     \n",
      "Epoch 1706/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0534 - acc: 0.906 - 0s - loss: 0.0990 - acc: 0.8450     \n",
      "Epoch 1707/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0990 - acc: 0.8450     \n",
      "Epoch 1708/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0989 - acc: 0.8450     \n",
      "Epoch 1709/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0989 - acc: 0.8450     \n",
      "Epoch 1710/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0989 - acc: 0.8450     \n",
      "Epoch 1711/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0989 - acc: 0.8450     \n",
      "Epoch 1712/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0988 - acc: 0.8450     \n",
      "Epoch 1713/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0988 - acc: 0.8450     \n",
      "Epoch 1714/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0988 - acc: 0.8450     \n",
      "Epoch 1715/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0986 - acc: 0.8450     \n",
      "Epoch 1716/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0986 - acc: 0.8450     \n",
      "Epoch 1717/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0988 - acc: 0.8500     \n",
      "Epoch 1718/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0988 - acc: 0.8500     \n",
      "Epoch 1719/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0987 - acc: 0.8500     \n",
      "Epoch 1720/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0987 - acc: 0.8500     \n",
      "Epoch 1721/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0987 - acc: 0.8500     \n",
      "Epoch 1722/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0986 - acc: 0.8500     \n",
      "Epoch 1723/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0985 - acc: 0.8500     \n",
      "Epoch 1724/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0986 - acc: 0.8500     \n",
      "Epoch 1725/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0984 - acc: 0.8500     \n",
      "Epoch 1726/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0984 - acc: 0.8450     \n",
      "Epoch 1727/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0984 - acc: 0.8450     \n",
      "Epoch 1728/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0983 - acc: 0.8450     \n",
      "Epoch 1729/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0983 - acc: 0.8450     \n",
      "Epoch 1730/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0983 - acc: 0.8450     \n",
      "Epoch 1731/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0982 - acc: 0.8450     \n",
      "Epoch 1732/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0983 - acc: 0.8450     \n",
      "Epoch 1733/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0983 - acc: 0.8450     \n",
      "Epoch 1734/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0982 - acc: 0.8450     \n",
      "Epoch 1735/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0983 - acc: 0.8450     \n",
      "Epoch 1736/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0983 - acc: 0.8450     \n",
      "Epoch 1737/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0981 - acc: 0.8450     \n",
      "Epoch 1738/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0981 - acc: 0.8450     \n",
      "Epoch 1739/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0980 - acc: 0.8500     \n",
      "Epoch 1740/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0980 - acc: 0.8500     \n",
      "Epoch 1741/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0980 - acc: 0.8500     \n",
      "Epoch 1742/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0982 - acc: 0.8400     \n",
      "Epoch 1743/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0980 - acc: 0.8450     \n",
      "Epoch 1744/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0979 - acc: 0.8450     \n",
      "Epoch 1745/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0980 - acc: 0.8450     \n",
      "Epoch 1746/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0978 - acc: 0.8400     \n",
      "Epoch 1747/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0977 - acc: 0.8450     \n",
      "Epoch 1748/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0977 - acc: 0.8500     \n",
      "Epoch 1749/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0977 - acc: 0.8450     \n",
      "Epoch 1750/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0978 - acc: 0.8500     \n",
      "Epoch 1751/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0977 - acc: 0.8500     \n",
      "Epoch 1752/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0977 - acc: 0.8450     \n",
      "Epoch 1753/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0976 - acc: 0.8450     \n",
      "Epoch 1754/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0976 - acc: 0.8450     \n",
      "Epoch 1755/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0976 - acc: 0.8450     \n",
      "Epoch 1756/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0977 - acc: 0.8500     \n",
      "Epoch 1757/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0976 - acc: 0.8500     \n",
      "Epoch 1758/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0974 - acc: 0.8500     \n",
      "Epoch 1759/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0974 - acc: 0.8450     \n",
      "Epoch 1760/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0976 - acc: 0.8450     \n",
      "Epoch 1761/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0975 - acc: 0.8500     \n",
      "Epoch 1762/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0974 - acc: 0.8500     \n",
      "Epoch 1763/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0973 - acc: 0.8500     \n",
      "Epoch 1764/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0885 - acc: 0.875 - 0s - loss: 0.0973 - acc: 0.8450     \n",
      "Epoch 1765/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0972 - acc: 0.8450     \n",
      "Epoch 1766/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0972 - acc: 0.8450     \n",
      "Epoch 1767/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0972 - acc: 0.8500     \n",
      "Epoch 1768/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0971 - acc: 0.8450     \n",
      "Epoch 1769/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0971 - acc: 0.8500     \n",
      "Epoch 1770/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0971 - acc: 0.8500     \n",
      "Epoch 1771/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0970 - acc: 0.8450     \n",
      "Epoch 1772/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0970 - acc: 0.8500     \n",
      "Epoch 1773/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0971 - acc: 0.8500     \n",
      "Epoch 1774/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0970 - acc: 0.8500     \n",
      "Epoch 1775/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0971 - acc: 0.8500     \n",
      "Epoch 1776/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0970 - acc: 0.8500     \n",
      "Epoch 1777/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0968 - acc: 0.8500     \n",
      "Epoch 1778/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0967 - acc: 0.8500     \n",
      "Epoch 1779/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0967 - acc: 0.8500     \n",
      "Epoch 1780/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0967 - acc: 0.8500     \n",
      "Epoch 1781/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0967 - acc: 0.8450     \n",
      "Epoch 1782/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0970 - acc: 0.8550     \n",
      "Epoch 1783/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0968 - acc: 0.8550     \n",
      "Epoch 1784/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0968 - acc: 0.8550     \n",
      "Epoch 1785/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0968 - acc: 0.8550     \n",
      "Epoch 1786/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0968 - acc: 0.8550     \n",
      "Epoch 1787/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.0966 - acc: 0.8550     \n",
      "Epoch 1788/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0965 - acc: 0.8450     \n",
      "Epoch 1789/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0964 - acc: 0.8500     \n",
      "Epoch 1790/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0964 - acc: 0.8500     \n",
      "Epoch 1791/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0963 - acc: 0.8500     \n",
      "Epoch 1792/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0963 - acc: 0.8500     \n",
      "Epoch 1793/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0962 - acc: 0.8500     \n",
      "Epoch 1794/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0963 - acc: 0.8500     \n",
      "Epoch 1795/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0963 - acc: 0.8500     \n",
      "Epoch 1796/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0961 - acc: 0.8500     \n",
      "Epoch 1797/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0960 - acc: 0.8500     \n",
      "Epoch 1798/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0961 - acc: 0.8500     \n",
      "Epoch 1799/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0960 - acc: 0.8500     \n",
      "Epoch 1800/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0960 - acc: 0.8550     \n",
      "Epoch 1801/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0960 - acc: 0.8500     \n",
      "Epoch 1802/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0959 - acc: 0.8500     \n",
      "Epoch 1803/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0962 - acc: 0.8500     \n",
      "Epoch 1804/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0960 - acc: 0.8500     \n",
      "Epoch 1805/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0959 - acc: 0.8500     \n",
      "Epoch 1806/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0958 - acc: 0.8500     \n",
      "Epoch 1807/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1279 - acc: 0.750 - 0s - loss: 0.0957 - acc: 0.8500     \n",
      "Epoch 1808/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0957 - acc: 0.8500     \n",
      "Epoch 1809/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0956 - acc: 0.8500     \n",
      "Epoch 1810/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0961 - acc: 0.8550     \n",
      "Epoch 1811/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0958 - acc: 0.8550     \n",
      "Epoch 1812/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0959 - acc: 0.8550     \n",
      "Epoch 1813/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0957 - acc: 0.8550     \n",
      "Epoch 1814/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0956 - acc: 0.8550     \n",
      "Epoch 1815/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0955 - acc: 0.8550     \n",
      "Epoch 1816/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0953 - acc: 0.8450     \n",
      "Epoch 1817/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0954 - acc: 0.8450     \n",
      "Epoch 1818/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0952 - acc: 0.8500     \n",
      "Epoch 1819/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0954 - acc: 0.8500     \n",
      "Epoch 1820/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0952 - acc: 0.8500     \n",
      "Epoch 1821/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0953 - acc: 0.8550     \n",
      "Epoch 1822/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0955 - acc: 0.8550     \n",
      "Epoch 1823/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0953 - acc: 0.8550     \n",
      "Epoch 1824/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0951 - acc: 0.8550     \n",
      "Epoch 1825/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0950 - acc: 0.8500     \n",
      "Epoch 1826/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0949 - acc: 0.8500     \n",
      "Epoch 1827/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0949 - acc: 0.8550     \n",
      "Epoch 1828/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0950 - acc: 0.8550     \n",
      "Epoch 1829/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0950 - acc: 0.8600     \n",
      "Epoch 1830/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0949 - acc: 0.8550     \n",
      "Epoch 1831/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0950 - acc: 0.8550     \n",
      "Epoch 1832/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0950 - acc: 0.8550     \n",
      "Epoch 1833/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0949 - acc: 0.8550     \n",
      "Epoch 1834/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0947 - acc: 0.8550     \n",
      "Epoch 1835/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0947 - acc: 0.8550     \n",
      "Epoch 1836/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0947 - acc: 0.8550     \n",
      "Epoch 1837/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0947 - acc: 0.8550     \n",
      "Epoch 1838/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0946 - acc: 0.8550     \n",
      "Epoch 1839/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0945 - acc: 0.8550     \n",
      "Epoch 1840/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0944 - acc: 0.8600     \n",
      "Epoch 1841/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0943 - acc: 0.8500     \n",
      "Epoch 1842/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0943 - acc: 0.8500     \n",
      "Epoch 1843/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0943 - acc: 0.8500     \n",
      "Epoch 1844/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0944 - acc: 0.8500     \n",
      "Epoch 1845/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0942 - acc: 0.8550     \n",
      "Epoch 1846/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0942 - acc: 0.8550     \n",
      "Epoch 1847/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0941 - acc: 0.8550     \n",
      "Epoch 1848/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0941 - acc: 0.8550     \n",
      "Epoch 1849/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0940 - acc: 0.8550     \n",
      "Epoch 1850/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0940 - acc: 0.8500     \n",
      "Epoch 1851/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0941 - acc: 0.8550     \n",
      "Epoch 1852/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0940 - acc: 0.8550     \n",
      "Epoch 1853/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0939 - acc: 0.8500     \n",
      "Epoch 1854/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0938 - acc: 0.8600     \n",
      "Epoch 1855/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0939 - acc: 0.8600     \n",
      "Epoch 1856/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0940 - acc: 0.8550     \n",
      "Epoch 1857/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0940 - acc: 0.8550     \n",
      "Epoch 1858/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0938 - acc: 0.8550     \n",
      "Epoch 1859/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0940 - acc: 0.8600     \n",
      "Epoch 1860/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0938 - acc: 0.8550     \n",
      "Epoch 1861/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0935 - acc: 0.8500     \n",
      "Epoch 1862/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0935 - acc: 0.8600     \n",
      "Epoch 1863/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0935 - acc: 0.8600     \n",
      "Epoch 1864/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0935 - acc: 0.8600     \n",
      "Epoch 1865/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0934 - acc: 0.8500     \n",
      "Epoch 1866/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0933 - acc: 0.8550     \n",
      "Epoch 1867/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0933 - acc: 0.8550     \n",
      "Epoch 1868/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0933 - acc: 0.8550     \n",
      "Epoch 1869/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0933 - acc: 0.8550     \n",
      "Epoch 1870/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0931 - acc: 0.8550     \n",
      "Epoch 1871/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0931 - acc: 0.8600     \n",
      "Epoch 1872/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.0930 - acc: 0.8550     \n",
      "Epoch 1873/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0929 - acc: 0.8600     \n",
      "Epoch 1874/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0930 - acc: 0.8600     \n",
      "Epoch 1875/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0930 - acc: 0.8600     \n",
      "Epoch 1876/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0930 - acc: 0.8600     \n",
      "Epoch 1877/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0929 - acc: 0.8600     \n",
      "Epoch 1878/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0928 - acc: 0.8600     \n",
      "Epoch 1879/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0927 - acc: 0.8600     \n",
      "Epoch 1880/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0927 - acc: 0.8500     \n",
      "Epoch 1881/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0927 - acc: 0.8550     \n",
      "Epoch 1882/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0928 - acc: 0.8550     \n",
      "Epoch 1883/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0927 - acc: 0.8600     \n",
      "Epoch 1884/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0926 - acc: 0.8600     \n",
      "Epoch 1885/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0924 - acc: 0.8550     \n",
      "Epoch 1886/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0924 - acc: 0.8550     \n",
      "Epoch 1887/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0925 - acc: 0.8600     \n",
      "Epoch 1888/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0924 - acc: 0.8600     \n",
      "Epoch 1889/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0924 - acc: 0.8550     \n",
      "Epoch 1890/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0923 - acc: 0.8550     \n",
      "Epoch 1891/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0922 - acc: 0.8650     \n",
      "Epoch 1892/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0922 - acc: 0.8650     \n",
      "Epoch 1893/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0921 - acc: 0.8550     \n",
      "Epoch 1894/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0921 - acc: 0.8550     \n",
      "Epoch 1895/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0920 - acc: 0.8550     \n",
      "Epoch 1896/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0920 - acc: 0.8550     \n",
      "Epoch 1897/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0920 - acc: 0.8550     \n",
      "Epoch 1898/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0920 - acc: 0.8550     \n",
      "Epoch 1899/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0919 - acc: 0.8550     \n",
      "Epoch 1900/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0919 - acc: 0.8550     \n",
      "Epoch 1901/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0919 - acc: 0.8550     \n",
      "Epoch 1902/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0919 - acc: 0.8600     \n",
      "Epoch 1903/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0917 - acc: 0.8600     \n",
      "Epoch 1904/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0917 - acc: 0.8600     \n",
      "Epoch 1905/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0916 - acc: 0.8600     \n",
      "Epoch 1906/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0916 - acc: 0.8600     \n",
      "Epoch 1907/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0915 - acc: 0.8600     \n",
      "Epoch 1908/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0914 - acc: 0.8600     \n",
      "Epoch 1909/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0914 - acc: 0.8600     \n",
      "Epoch 1910/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0914 - acc: 0.8600     \n",
      "Epoch 1911/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0913 - acc: 0.8650     \n",
      "Epoch 1912/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0913 - acc: 0.8650     \n",
      "Epoch 1913/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0912 - acc: 0.8600     \n",
      "Epoch 1914/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0911 - acc: 0.8600     \n",
      "Epoch 1915/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0912 - acc: 0.8650     \n",
      "Epoch 1916/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0911 - acc: 0.8600     \n",
      "Epoch 1917/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0911 - acc: 0.8600     \n",
      "Epoch 1918/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0910 - acc: 0.8700     \n",
      "Epoch 1919/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0910 - acc: 0.8600     \n",
      "Epoch 1920/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0909 - acc: 0.8600     \n",
      "Epoch 1921/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0909 - acc: 0.8600     \n",
      "Epoch 1922/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0908 - acc: 0.8600     \n",
      "Epoch 1923/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0909 - acc: 0.8600     \n",
      "Epoch 1924/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0910 - acc: 0.8650     \n",
      "Epoch 1925/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0908 - acc: 0.8600     \n",
      "Epoch 1926/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0907 - acc: 0.8600     \n",
      "Epoch 1927/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0906 - acc: 0.8600     \n",
      "Epoch 1928/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0905 - acc: 0.8700     \n",
      "Epoch 1929/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0904 - acc: 0.8700     \n",
      "Epoch 1930/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0904 - acc: 0.8600     \n",
      "Epoch 1931/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0904 - acc: 0.8600     \n",
      "Epoch 1932/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0909 - acc: 0.8650     \n",
      "Epoch 1933/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0903 - acc: 0.8600     \n",
      "Epoch 1934/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0902 - acc: 0.8700     \n",
      "Epoch 1935/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0902 - acc: 0.8650     \n",
      "Epoch 1936/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0901 - acc: 0.8650     \n",
      "Epoch 1937/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0902 - acc: 0.8600     \n",
      "Epoch 1938/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0900 - acc: 0.8600     \n",
      "Epoch 1939/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0899 - acc: 0.8700     \n",
      "Epoch 1940/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0899 - acc: 0.8700     \n",
      "Epoch 1941/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0902 - acc: 0.8650     \n",
      "Epoch 1942/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0899 - acc: 0.8650     \n",
      "Epoch 1943/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0898 - acc: 0.8650     \n",
      "Epoch 1944/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0901 - acc: 0.8650     \n",
      "Epoch 1945/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0898 - acc: 0.8650     \n",
      "Epoch 1946/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0897 - acc: 0.8650     \n",
      "Epoch 1947/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0896 - acc: 0.8650     \n",
      "Epoch 1948/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0895 - acc: 0.8650     \n",
      "Epoch 1949/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0895 - acc: 0.8650     \n",
      "Epoch 1950/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0895 - acc: 0.8650     \n",
      "Epoch 1951/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0896 - acc: 0.8650     \n",
      "Epoch 1952/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0893 - acc: 0.8650     \n",
      "Epoch 1953/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0893 - acc: 0.8650     \n",
      "Epoch 1954/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0891 - acc: 0.8650     \n",
      "Epoch 1955/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0891 - acc: 0.8700     \n",
      "Epoch 1956/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0893 - acc: 0.8700     \n",
      "Epoch 1957/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.0894 - acc: 0.8650     \n",
      "Epoch 1958/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0891 - acc: 0.8650     \n",
      "Epoch 1959/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0891 - acc: 0.8650     \n",
      "Epoch 1960/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0890 - acc: 0.8650     \n",
      "Epoch 1961/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0889 - acc: 0.8650     \n",
      "Epoch 1962/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0889 - acc: 0.8700     \n",
      "Epoch 1963/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0889 - acc: 0.8650     \n",
      "Epoch 1964/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0888 - acc: 0.8700     \n",
      "Epoch 1965/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0887 - acc: 0.8700     \n",
      "Epoch 1966/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0887 - acc: 0.8700     \n",
      "Epoch 1967/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0886 - acc: 0.8700     \n",
      "Epoch 1968/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0887 - acc: 0.8650     \n",
      "Epoch 1969/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0885 - acc: 0.8700     \n",
      "Epoch 1970/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0885 - acc: 0.8700     \n",
      "Epoch 1971/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0884 - acc: 0.8650     \n",
      "Epoch 1972/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0886 - acc: 0.8700     \n",
      "Epoch 1973/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0885 - acc: 0.8700     \n",
      "Epoch 1974/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0884 - acc: 0.8700     \n",
      "Epoch 1975/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0884 - acc: 0.8750     \n",
      "Epoch 1976/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0883 - acc: 0.8750     \n",
      "Epoch 1977/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0882 - acc: 0.8750     \n",
      "Epoch 1978/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0881 - acc: 0.8700     \n",
      "Epoch 1979/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0881 - acc: 0.8700     \n",
      "Epoch 1980/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0880 - acc: 0.8700     \n",
      "Epoch 1981/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0879 - acc: 0.8650     \n",
      "Epoch 1982/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0879 - acc: 0.8650     \n",
      "Epoch 1983/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0879 - acc: 0.8650     \n",
      "Epoch 1984/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0878 - acc: 0.8700     \n",
      "Epoch 1985/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0877 - acc: 0.8700     \n",
      "Epoch 1986/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0878 - acc: 0.8700     \n",
      "Epoch 1987/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0879 - acc: 0.8700     \n",
      "Epoch 1988/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0877 - acc: 0.8700     \n",
      "Epoch 1989/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0878 - acc: 0.8700     \n",
      "Epoch 1990/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0878 - acc: 0.8700     \n",
      "Epoch 1991/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0876 - acc: 0.8700     \n",
      "Epoch 1992/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0877 - acc: 0.8700     \n",
      "Epoch 1993/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0874 - acc: 0.8750     \n",
      "Epoch 1994/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0874 - acc: 0.8750     \n",
      "Epoch 1995/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0874 - acc: 0.8700     \n",
      "Epoch 1996/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0872 - acc: 0.8750     \n",
      "Epoch 1997/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0872 - acc: 0.8750     \n",
      "Epoch 1998/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0872 - acc: 0.8750     \n",
      "Epoch 1999/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0873 - acc: 0.8750     \n",
      "Epoch 2000/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0873 - acc: 0.8750     \n",
      "Epoch 2001/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0870 - acc: 0.8700     \n",
      "Epoch 2002/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0870 - acc: 0.8700     \n",
      "Epoch 2003/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0869 - acc: 0.8700     \n",
      "Epoch 2004/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0868 - acc: 0.8700     \n",
      "Epoch 2005/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0868 - acc: 0.8700     \n",
      "Epoch 2006/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0867 - acc: 0.8700     \n",
      "Epoch 2007/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0867 - acc: 0.8700     \n",
      "Epoch 2008/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0869 - acc: 0.8800     \n",
      "Epoch 2009/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0866 - acc: 0.8750     \n",
      "Epoch 2010/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0865 - acc: 0.8700     \n",
      "Epoch 2011/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0866 - acc: 0.8750     \n",
      "Epoch 2012/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0866 - acc: 0.8750     \n",
      "Epoch 2013/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0865 - acc: 0.8750     \n",
      "Epoch 2014/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0863 - acc: 0.8750     \n",
      "Epoch 2015/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0863 - acc: 0.8700     \n",
      "Epoch 2016/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0863 - acc: 0.8850     \n",
      "Epoch 2017/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0864 - acc: 0.8850     \n",
      "Epoch 2018/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0864 - acc: 0.8850     \n",
      "Epoch 2019/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0864 - acc: 0.8850     \n",
      "Epoch 2020/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0862 - acc: 0.8900     \n",
      "Epoch 2021/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0861 - acc: 0.8850     \n",
      "Epoch 2022/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0860 - acc: 0.8800     \n",
      "Epoch 2023/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0858 - acc: 0.8750     \n",
      "Epoch 2024/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0859 - acc: 0.8750     \n",
      "Epoch 2025/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0858 - acc: 0.8750     \n",
      "Epoch 2026/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0858 - acc: 0.8800     \n",
      "Epoch 2027/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0857 - acc: 0.8800     \n",
      "Epoch 2028/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0857 - acc: 0.8800     \n",
      "Epoch 2029/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0857 - acc: 0.8850     \n",
      "Epoch 2030/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0856 - acc: 0.8850     \n",
      "Epoch 2031/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0855 - acc: 0.8800     \n",
      "Epoch 2032/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0855 - acc: 0.8800     \n",
      "Epoch 2033/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0856 - acc: 0.8800     \n",
      "Epoch 2034/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0856 - acc: 0.8800     \n",
      "Epoch 2035/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0855 - acc: 0.8800     \n",
      "Epoch 2036/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0853 - acc: 0.8800     \n",
      "Epoch 2037/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0852 - acc: 0.8750     \n",
      "Epoch 2038/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0852 - acc: 0.8750     \n",
      "Epoch 2039/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0852 - acc: 0.8750     \n",
      "Epoch 2040/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0851 - acc: 0.8750     \n",
      "Epoch 2041/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0851 - acc: 0.8850     \n",
      "Epoch 2042/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.0851 - acc: 0.8800     \n",
      "Epoch 2043/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0851 - acc: 0.8850     \n",
      "Epoch 2044/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0848 - acc: 0.8850     \n",
      "Epoch 2045/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0926 - acc: 0.906 - 0s - loss: 0.0848 - acc: 0.8850     \n",
      "Epoch 2046/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0847 - acc: 0.8850     \n",
      "Epoch 2047/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0847 - acc: 0.8800     \n",
      "Epoch 2048/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0847 - acc: 0.8850     \n",
      "Epoch 2049/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0847 - acc: 0.8800     \n",
      "Epoch 2050/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0846 - acc: 0.8850     \n",
      "Epoch 2051/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0847 - acc: 0.8800     \n",
      "Epoch 2052/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0846 - acc: 0.8800     \n",
      "Epoch 2053/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0846 - acc: 0.8800     \n",
      "Epoch 2054/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0844 - acc: 0.8850     \n",
      "Epoch 2055/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0844 - acc: 0.8800     \n",
      "Epoch 2056/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0845 - acc: 0.8850     \n",
      "Epoch 2057/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0846 - acc: 0.8850     \n",
      "Epoch 2058/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0845 - acc: 0.8850     \n",
      "Epoch 2059/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0844 - acc: 0.8800     \n",
      "Epoch 2060/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0842 - acc: 0.8850     \n",
      "Epoch 2061/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0840 - acc: 0.8900     \n",
      "Epoch 2062/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0839 - acc: 0.8850     \n",
      "Epoch 2063/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0841 - acc: 0.8850     \n",
      "Epoch 2064/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0839 - acc: 0.8950     \n",
      "Epoch 2065/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0607 - acc: 0.906 - 0s - loss: 0.0838 - acc: 0.8900     \n",
      "Epoch 2066/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0837 - acc: 0.8900     \n",
      "Epoch 2067/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0837 - acc: 0.8850     \n",
      "Epoch 2068/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0837 - acc: 0.8850     \n",
      "Epoch 2069/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0837 - acc: 0.8850     \n",
      "Epoch 2070/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0837 - acc: 0.8850     \n",
      "Epoch 2071/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0838 - acc: 0.8850     \n",
      "Epoch 2072/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0837 - acc: 0.8850     \n",
      "Epoch 2073/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0836 - acc: 0.8850     \n",
      "Epoch 2074/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0834 - acc: 0.8800     \n",
      "Epoch 2075/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0833 - acc: 0.8850     \n",
      "Epoch 2076/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0833 - acc: 0.8850     \n",
      "Epoch 2077/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0832 - acc: 0.8850     \n",
      "Epoch 2078/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0832 - acc: 0.8900     \n",
      "Epoch 2079/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0833 - acc: 0.8900     \n",
      "Epoch 2080/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0831 - acc: 0.8900     \n",
      "Epoch 2081/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0831 - acc: 0.8900     \n",
      "Epoch 2082/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0830 - acc: 0.8850     \n",
      "Epoch 2083/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0830 - acc: 0.8900     \n",
      "Epoch 2084/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0830 - acc: 0.8850     \n",
      "Epoch 2085/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0830 - acc: 0.8850     \n",
      "Epoch 2086/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0830 - acc: 0.8800     \n",
      "Epoch 2087/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0829 - acc: 0.8900     \n",
      "Epoch 2088/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0830 - acc: 0.8850     \n",
      "Epoch 2089/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0827 - acc: 0.8850     \n",
      "Epoch 2090/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0826 - acc: 0.8900     \n",
      "Epoch 2091/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0825 - acc: 0.8950     \n",
      "Epoch 2092/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0828 - acc: 0.8900     \n",
      "Epoch 2093/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0826 - acc: 0.8950     \n",
      "Epoch 2094/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0823 - acc: 0.8950     \n",
      "Epoch 2095/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0822 - acc: 0.8900     \n",
      "Epoch 2096/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0824 - acc: 0.8900     \n",
      "Epoch 2097/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0825 - acc: 0.8900     \n",
      "Epoch 2098/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0822 - acc: 0.8900     \n",
      "Epoch 2099/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0823 - acc: 0.8900     \n",
      "Epoch 2100/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0822 - acc: 0.8950     \n",
      "Epoch 2101/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0823 - acc: 0.8900     \n",
      "Epoch 2102/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0823 - acc: 0.8900     \n",
      "Epoch 2103/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0825 - acc: 0.8900     \n",
      "Epoch 2104/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0820 - acc: 0.8950     \n",
      "Epoch 2105/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0819 - acc: 0.8950     \n",
      "Epoch 2106/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0819 - acc: 0.8950     \n",
      "Epoch 2107/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0818 - acc: 0.8950     \n",
      "Epoch 2108/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0818 - acc: 0.8950     \n",
      "Epoch 2109/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0818 - acc: 0.8900     \n",
      "Epoch 2110/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0818 - acc: 0.8900     \n",
      "Epoch 2111/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0819 - acc: 0.8900     \n",
      "Epoch 2112/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0818 - acc: 0.8900     \n",
      "Epoch 2113/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0818 - acc: 0.8900     \n",
      "Epoch 2114/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.1453 - acc: 0.843 - 0s - loss: 0.0816 - acc: 0.8950     \n",
      "Epoch 2115/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0816 - acc: 0.8900     \n",
      "Epoch 2116/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0815 - acc: 0.8950     \n",
      "Epoch 2117/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0813 - acc: 0.8950     \n",
      "Epoch 2118/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0813 - acc: 0.9000     \n",
      "Epoch 2119/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0812 - acc: 0.9000     \n",
      "Epoch 2120/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0812 - acc: 0.8950     \n",
      "Epoch 2121/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0811 - acc: 0.8950     \n",
      "Epoch 2122/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0810 - acc: 0.8950     \n",
      "Epoch 2123/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0811 - acc: 0.8900     \n",
      "Epoch 2124/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0810 - acc: 0.8900     \n",
      "Epoch 2125/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0809 - acc: 0.8900     \n",
      "Epoch 2126/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.0808 - acc: 0.8950     \n",
      "Epoch 2127/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0811 - acc: 0.8950     \n",
      "Epoch 2128/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0808 - acc: 0.9000     \n",
      "Epoch 2129/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0807 - acc: 0.9000     \n",
      "Epoch 2130/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0807 - acc: 0.9000     \n",
      "Epoch 2131/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0807 - acc: 0.9000     \n",
      "Epoch 2132/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0806 - acc: 0.9000     \n",
      "Epoch 2133/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0806 - acc: 0.9000     \n",
      "Epoch 2134/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0805 - acc: 0.9000     \n",
      "Epoch 2135/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0804 - acc: 0.9000     \n",
      "Epoch 2136/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0804 - acc: 0.9000     \n",
      "Epoch 2137/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0804 - acc: 0.9000     \n",
      "Epoch 2138/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0804 - acc: 0.9000     \n",
      "Epoch 2139/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0802 - acc: 0.9000     \n",
      "Epoch 2140/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0802 - acc: 0.9000     \n",
      "Epoch 2141/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0802 - acc: 0.9000     \n",
      "Epoch 2142/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0802 - acc: 0.9000     \n",
      "Epoch 2143/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0800 - acc: 0.9050     \n",
      "Epoch 2144/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0800 - acc: 0.9000     \n",
      "Epoch 2145/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0799 - acc: 0.9000     \n",
      "Epoch 2146/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0800 - acc: 0.9000     \n",
      "Epoch 2147/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0799 - acc: 0.9000     \n",
      "Epoch 2148/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0798 - acc: 0.9000     \n",
      "Epoch 2149/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0797 - acc: 0.9000     \n",
      "Epoch 2150/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0798 - acc: 0.9050     \n",
      "Epoch 2151/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0798 - acc: 0.9000     \n",
      "Epoch 2152/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0797 - acc: 0.9000     \n",
      "Epoch 2153/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0797 - acc: 0.9000     \n",
      "Epoch 2154/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0796 - acc: 0.9000     \n",
      "Epoch 2155/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0796 - acc: 0.9000     \n",
      "Epoch 2156/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0795 - acc: 0.9050     \n",
      "Epoch 2157/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0794 - acc: 0.9050     \n",
      "Epoch 2158/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0795 - acc: 0.9000     \n",
      "Epoch 2159/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0793 - acc: 0.8950     \n",
      "Epoch 2160/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0793 - acc: 0.9050     \n",
      "Epoch 2161/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0793 - acc: 0.9050     \n",
      "Epoch 2162/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0792 - acc: 0.9000     \n",
      "Epoch 2163/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0792 - acc: 0.9000     \n",
      "Epoch 2164/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0791 - acc: 0.9050     \n",
      "Epoch 2165/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0791 - acc: 0.9050     \n",
      "Epoch 2166/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0791 - acc: 0.9050     \n",
      "Epoch 2167/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0789 - acc: 0.9000     \n",
      "Epoch 2168/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0788 - acc: 0.9050     \n",
      "Epoch 2169/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0789 - acc: 0.9000     \n",
      "Epoch 2170/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0790 - acc: 0.9050     \n",
      "Epoch 2171/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0788 - acc: 0.9050     \n",
      "Epoch 2172/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0787 - acc: 0.9000     \n",
      "Epoch 2173/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0786 - acc: 0.9100     \n",
      "Epoch 2174/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0787 - acc: 0.9050     \n",
      "Epoch 2175/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0787 - acc: 0.9050     \n",
      "Epoch 2176/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0787 - acc: 0.9100     \n",
      "Epoch 2177/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0785 - acc: 0.9050     \n",
      "Epoch 2178/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0784 - acc: 0.9050     \n",
      "Epoch 2179/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0784 - acc: 0.9050     \n",
      "Epoch 2180/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0784 - acc: 0.9000     \n",
      "Epoch 2181/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0783 - acc: 0.9050     \n",
      "Epoch 2182/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0786 - acc: 0.9000     \n",
      "Epoch 2183/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0783 - acc: 0.9050     \n",
      "Epoch 2184/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0783 - acc: 0.9050     \n",
      "Epoch 2185/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0782 - acc: 0.9050     \n",
      "Epoch 2186/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0783 - acc: 0.9050     \n",
      "Epoch 2187/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0781 - acc: 0.9050     \n",
      "Epoch 2188/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0780 - acc: 0.9050     \n",
      "Epoch 2189/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0779 - acc: 0.9050     \n",
      "Epoch 2190/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0779 - acc: 0.9050     \n",
      "Epoch 2191/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0707 - acc: 0.906 - 0s - loss: 0.0779 - acc: 0.9050     \n",
      "Epoch 2192/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0779 - acc: 0.9100     \n",
      "Epoch 2193/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0779 - acc: 0.9100     \n",
      "Epoch 2194/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0778 - acc: 0.9100     \n",
      "Epoch 2195/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0778 - acc: 0.9100     \n",
      "Epoch 2196/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0777 - acc: 0.9100     \n",
      "Epoch 2197/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0777 - acc: 0.9100     \n",
      "Epoch 2198/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0834 - acc: 0.875 - 0s - loss: 0.0775 - acc: 0.9000     \n",
      "Epoch 2199/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0776 - acc: 0.9050     \n",
      "Epoch 2200/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0778 - acc: 0.9100     \n",
      "Epoch 2201/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0778 - acc: 0.9050     \n",
      "Epoch 2202/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0774 - acc: 0.9050     \n",
      "Epoch 2203/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0773 - acc: 0.9100     \n",
      "Epoch 2204/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0383 - acc: 0.968 - 0s - loss: 0.0774 - acc: 0.9100     \n",
      "Epoch 2205/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0774 - acc: 0.9100     \n",
      "Epoch 2206/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0772 - acc: 0.9100     \n",
      "Epoch 2207/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0773 - acc: 0.9100     \n",
      "Epoch 2208/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0774 - acc: 0.9050     \n",
      "Epoch 2209/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0772 - acc: 0.9050     \n",
      "Epoch 2210/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.0770 - acc: 0.9050     \n",
      "Epoch 2211/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0770 - acc: 0.9050     \n",
      "Epoch 2212/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0771 - acc: 0.9050     \n",
      "Epoch 2213/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0769 - acc: 0.9100     \n",
      "Epoch 2214/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0769 - acc: 0.9100     \n",
      "Epoch 2215/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0769 - acc: 0.9100     \n",
      "Epoch 2216/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0767 - acc: 0.9100     \n",
      "Epoch 2217/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0768 - acc: 0.9100     \n",
      "Epoch 2218/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0767 - acc: 0.9100     \n",
      "Epoch 2219/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0766 - acc: 0.9100     \n",
      "Epoch 2220/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0766 - acc: 0.9100     \n",
      "Epoch 2221/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0765 - acc: 0.9100     \n",
      "Epoch 2222/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0877 - acc: 0.875 - 0s - loss: 0.0765 - acc: 0.9100     \n",
      "Epoch 2223/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0764 - acc: 0.9100     \n",
      "Epoch 2224/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0765 - acc: 0.9150     \n",
      "Epoch 2225/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0764 - acc: 0.9100     \n",
      "Epoch 2226/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0763 - acc: 0.9150     \n",
      "Epoch 2227/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0765 - acc: 0.9100     \n",
      "Epoch 2228/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0764 - acc: 0.9100     \n",
      "Epoch 2229/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0764 - acc: 0.9150     \n",
      "Epoch 2230/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0762 - acc: 0.9100     \n",
      "Epoch 2231/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0762 - acc: 0.9100     \n",
      "Epoch 2232/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0764 - acc: 0.9100     \n",
      "Epoch 2233/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0761 - acc: 0.9150     \n",
      "Epoch 2234/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0761 - acc: 0.9150     \n",
      "Epoch 2235/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0760 - acc: 0.9150     \n",
      "Epoch 2236/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0760 - acc: 0.9100     \n",
      "Epoch 2237/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0759 - acc: 0.9100     \n",
      "Epoch 2238/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0759 - acc: 0.9150     \n",
      "Epoch 2239/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0760 - acc: 0.9100     \n",
      "Epoch 2240/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0758 - acc: 0.9100     \n",
      "Epoch 2241/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0757 - acc: 0.9150     \n",
      "Epoch 2242/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0757 - acc: 0.9100     \n",
      "Epoch 2243/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0757 - acc: 0.9100     \n",
      "Epoch 2244/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0757 - acc: 0.9100     \n",
      "Epoch 2245/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0756 - acc: 0.9100     \n",
      "Epoch 2246/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0755 - acc: 0.9100     \n",
      "Epoch 2247/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0755 - acc: 0.9150     \n",
      "Epoch 2248/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0755 - acc: 0.9150     \n",
      "Epoch 2249/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0754 - acc: 0.9150     \n",
      "Epoch 2250/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0754 - acc: 0.9150     \n",
      "Epoch 2251/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0753 - acc: 0.9150     \n",
      "Epoch 2252/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0753 - acc: 0.9100     \n",
      "Epoch 2253/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0753 - acc: 0.9150     \n",
      "Epoch 2254/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0752 - acc: 0.9150     \n",
      "Epoch 2255/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0752 - acc: 0.9200     \n",
      "Epoch 2256/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0752 - acc: 0.9100     \n",
      "Epoch 2257/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0751 - acc: 0.9150     \n",
      "Epoch 2258/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0751 - acc: 0.9150     \n",
      "Epoch 2259/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0750 - acc: 0.9150     \n",
      "Epoch 2260/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0751 - acc: 0.9150     \n",
      "Epoch 2261/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0750 - acc: 0.9150     \n",
      "Epoch 2262/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0750 - acc: 0.9150     \n",
      "Epoch 2263/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0749 - acc: 0.9200     \n",
      "Epoch 2264/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0748 - acc: 0.9150     \n",
      "Epoch 2265/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0748 - acc: 0.9200     \n",
      "Epoch 2266/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0747 - acc: 0.9100     \n",
      "Epoch 2267/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0747 - acc: 0.9150     \n",
      "Epoch 2268/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0748 - acc: 0.9150     \n",
      "Epoch 2269/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0746 - acc: 0.9150     \n",
      "Epoch 2270/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0746 - acc: 0.9150     \n",
      "Epoch 2271/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0746 - acc: 0.9150     \n",
      "Epoch 2272/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0746 - acc: 0.9150     \n",
      "Epoch 2273/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0745 - acc: 0.9150     \n",
      "Epoch 2274/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0744 - acc: 0.9150     \n",
      "Epoch 2275/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0743 - acc: 0.9150     \n",
      "Epoch 2276/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0743 - acc: 0.9150     \n",
      "Epoch 2277/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0745 - acc: 0.9150     \n",
      "Epoch 2278/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0744 - acc: 0.9150     \n",
      "Epoch 2279/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0744 - acc: 0.9150     \n",
      "Epoch 2280/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0741 - acc: 0.9200     \n",
      "Epoch 2281/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0742 - acc: 0.9150     \n",
      "Epoch 2282/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0741 - acc: 0.9150     \n",
      "Epoch 2283/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0741 - acc: 0.9150     \n",
      "Epoch 2284/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0741 - acc: 0.9150     \n",
      "Epoch 2285/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0741 - acc: 0.9150     \n",
      "Epoch 2286/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0739 - acc: 0.9150     \n",
      "Epoch 2287/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0739 - acc: 0.9150     \n",
      "Epoch 2288/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0739 - acc: 0.9150     \n",
      "Epoch 2289/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0739 - acc: 0.9150     \n",
      "Epoch 2290/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0738 - acc: 0.9150     \n",
      "Epoch 2291/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0738 - acc: 0.9150     \n",
      "Epoch 2292/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0737 - acc: 0.9150     \n",
      "Epoch 2293/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0737 - acc: 0.9200     \n",
      "Epoch 2294/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0736 - acc: 0.9200     \n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 2295/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0736 - acc: 0.9200     \n",
      "Epoch 2296/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0735 - acc: 0.9200     \n",
      "Epoch 2297/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0735 - acc: 0.9150     \n",
      "Epoch 2298/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0735 - acc: 0.9150     \n",
      "Epoch 2299/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0734 - acc: 0.9200     \n",
      "Epoch 2300/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0736 - acc: 0.9150     \n",
      "Epoch 2301/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0735 - acc: 0.9150     \n",
      "Epoch 2302/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0733 - acc: 0.9150     \n",
      "Epoch 2303/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0733 - acc: 0.9150     \n",
      "Epoch 2304/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0732 - acc: 0.9200     \n",
      "Epoch 2305/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0732 - acc: 0.9200     \n",
      "Epoch 2306/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0731 - acc: 0.9200     \n",
      "Epoch 2307/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0732 - acc: 0.9200     \n",
      "Epoch 2308/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0731 - acc: 0.9150     \n",
      "Epoch 2309/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0731 - acc: 0.9150     \n",
      "Epoch 2310/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0731 - acc: 0.9150     \n",
      "Epoch 2311/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0731 - acc: 0.9150     \n",
      "Epoch 2312/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0731 - acc: 0.9150     \n",
      "Epoch 2313/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0729 - acc: 0.9150     \n",
      "Epoch 2314/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0729 - acc: 0.9150     \n",
      "Epoch 2315/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0728 - acc: 0.9150     \n",
      "Epoch 2316/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0730 - acc: 0.9200     \n",
      "Epoch 2317/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0729 - acc: 0.9200     \n",
      "Epoch 2318/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0732 - acc: 0.9200     \n",
      "Epoch 2319/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0729 - acc: 0.9200     \n",
      "Epoch 2320/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0727 - acc: 0.9200     \n",
      "Epoch 2321/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0726 - acc: 0.9150     \n",
      "Epoch 2322/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0725 - acc: 0.9150     \n",
      "Epoch 2323/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0725 - acc: 0.9200     \n",
      "Epoch 2324/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0726 - acc: 0.9150     \n",
      "Epoch 2325/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0726 - acc: 0.9150     \n",
      "Epoch 2326/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0724 - acc: 0.9200     \n",
      "Epoch 2327/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0724 - acc: 0.9200     \n",
      "Epoch 2328/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0724 - acc: 0.9150     \n",
      "Epoch 2329/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0723 - acc: 0.9150     \n",
      "Epoch 2330/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0724 - acc: 0.9150     \n",
      "Epoch 2331/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0722 - acc: 0.9150     \n",
      "Epoch 2332/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0722 - acc: 0.9150     \n",
      "Epoch 2333/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0721 - acc: 0.9150     \n",
      "Epoch 2334/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0721 - acc: 0.9200     \n",
      "Epoch 2335/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0721 - acc: 0.9200     \n",
      "Epoch 2336/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0964 - acc: 0.906 - 0s - loss: 0.0720 - acc: 0.9200     \n",
      "Epoch 2337/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0723 - acc: 0.9200     \n",
      "Epoch 2338/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0720 - acc: 0.9200     \n",
      "Epoch 2339/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0720 - acc: 0.9200     \n",
      "Epoch 2340/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0718 - acc: 0.9200     \n",
      "Epoch 2341/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0720 - acc: 0.9200     \n",
      "Epoch 2342/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0720 - acc: 0.9150     \n",
      "Epoch 2343/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0719 - acc: 0.9150     \n",
      "Epoch 2344/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0718 - acc: 0.9150     \n",
      "Epoch 2345/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0718 - acc: 0.9150     \n",
      "Epoch 2346/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0717 - acc: 0.9150     \n",
      "Epoch 2347/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0717 - acc: 0.9200     \n",
      "Epoch 2348/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0716 - acc: 0.9200     \n",
      "Epoch 2349/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0717 - acc: 0.9250     \n",
      "Epoch 2350/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0717 - acc: 0.9200     \n",
      "Epoch 2351/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0717 - acc: 0.9200     \n",
      "Epoch 2352/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0716 - acc: 0.9200     \n",
      "Epoch 2353/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0715 - acc: 0.9250     \n",
      "Epoch 2354/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0718 - acc: 0.9200     \n",
      "Epoch 2355/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0717 - acc: 0.9250     \n",
      "Epoch 2356/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0713 - acc: 0.9150     \n",
      "Epoch 2357/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0713 - acc: 0.9200     \n",
      "Epoch 2358/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0713 - acc: 0.9200     \n",
      "Epoch 2359/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0713 - acc: 0.9200     \n",
      "Epoch 2360/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0712 - acc: 0.9200     \n",
      "Epoch 2361/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0713 - acc: 0.9200     \n",
      "Epoch 2362/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0712 - acc: 0.9200     \n",
      "Epoch 2363/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0711 - acc: 0.9200     \n",
      "Epoch 2364/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0714 - acc: 0.9250     \n",
      "Epoch 2365/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0710 - acc: 0.9250     \n",
      "Epoch 2366/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0710 - acc: 0.9200     \n",
      "Epoch 2367/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0709 - acc: 0.9250     \n",
      "Epoch 2368/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0709 - acc: 0.9200     \n",
      "Epoch 2369/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0710 - acc: 0.9200     \n",
      "Epoch 2370/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0711 - acc: 0.9250     \n",
      "Epoch 2371/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0708 - acc: 0.9250     \n",
      "Epoch 2372/2500\n",
      "200/200 [==============================] - ETA: 0s - loss: 0.0467 - acc: 0.937 - 0s - loss: 0.0707 - acc: 0.9200     \n",
      "Epoch 2373/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0707 - acc: 0.9200     \n",
      "Epoch 2374/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0708 - acc: 0.9200     \n",
      "Epoch 2375/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0707 - acc: 0.9300     \n",
      "Epoch 2376/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0707 - acc: 0.9300     \n",
      "Epoch 2377/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0707 - acc: 0.9300     \n",
      "Epoch 2378/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0705 - acc: 0.9300     \n",
      "Epoch 2379/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.0705 - acc: 0.9250     \n",
      "Epoch 2380/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0705 - acc: 0.9250     \n",
      "Epoch 2381/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0705 - acc: 0.9300     \n",
      "Epoch 2382/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0705 - acc: 0.9300     \n",
      "Epoch 2383/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0705 - acc: 0.9300     \n",
      "Epoch 2384/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0705 - acc: 0.9250     \n",
      "Epoch 2385/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0704 - acc: 0.9250     \n",
      "Epoch 2386/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0702 - acc: 0.9250     \n",
      "Epoch 2387/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0702 - acc: 0.9250     \n",
      "Epoch 2388/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0703 - acc: 0.9300     \n",
      "Epoch 2389/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0703 - acc: 0.9250     \n",
      "Epoch 2390/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0704 - acc: 0.9250     \n",
      "Epoch 2391/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0703 - acc: 0.9250     \n",
      "Epoch 2392/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0702 - acc: 0.9300     \n",
      "Epoch 2393/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0701 - acc: 0.9300     \n",
      "Epoch 2394/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0701 - acc: 0.9250     \n",
      "Epoch 2395/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0700 - acc: 0.9250     \n",
      "Epoch 2396/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0700 - acc: 0.9300     \n",
      "Epoch 2397/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0699 - acc: 0.9300     \n",
      "Epoch 2398/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0699 - acc: 0.9300     \n",
      "Epoch 2399/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0699 - acc: 0.9300     \n",
      "Epoch 2400/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0698 - acc: 0.9300     \n",
      "Epoch 2401/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0698 - acc: 0.9350     \n",
      "Epoch 2402/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0698 - acc: 0.9300     \n",
      "Epoch 2403/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0697 - acc: 0.9250     \n",
      "Epoch 2404/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0697 - acc: 0.9250     \n",
      "Epoch 2405/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0697 - acc: 0.9300     \n",
      "Epoch 2406/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0696 - acc: 0.9300     \n",
      "Epoch 2407/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0698 - acc: 0.9300     \n",
      "Epoch 2408/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0697 - acc: 0.9300     \n",
      "Epoch 2409/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0696 - acc: 0.9350     \n",
      "Epoch 2410/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0695 - acc: 0.9250     \n",
      "Epoch 2411/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0699 - acc: 0.9300     \n",
      "Epoch 2412/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0695 - acc: 0.9300     \n",
      "Epoch 2413/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0694 - acc: 0.9300     \n",
      "Epoch 2414/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0694 - acc: 0.9300     \n",
      "Epoch 2415/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0693 - acc: 0.9300     \n",
      "Epoch 2416/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0693 - acc: 0.9250     \n",
      "Epoch 2417/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0693 - acc: 0.9250     \n",
      "Epoch 2418/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0693 - acc: 0.9250     \n",
      "Epoch 2419/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0693 - acc: 0.9250     \n",
      "Epoch 2420/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0692 - acc: 0.9300     \n",
      "Epoch 2421/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0692 - acc: 0.9300     \n",
      "Epoch 2422/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0691 - acc: 0.9300     \n",
      "Epoch 2423/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0691 - acc: 0.9300     \n",
      "Epoch 2424/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0692 - acc: 0.9300     \n",
      "Epoch 2425/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0690 - acc: 0.9300     \n",
      "Epoch 2426/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0690 - acc: 0.9300     \n",
      "Epoch 2427/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0690 - acc: 0.9300     \n",
      "Epoch 2428/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0689 - acc: 0.9300     \n",
      "Epoch 2429/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0690 - acc: 0.9300     \n",
      "Epoch 2430/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0689 - acc: 0.9250     \n",
      "Epoch 2431/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0688 - acc: 0.9300     \n",
      "Epoch 2432/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0688 - acc: 0.9300     \n",
      "Epoch 2433/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0686 - acc: 0.9350     \n",
      "Epoch 2434/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0687 - acc: 0.9400     \n",
      "Epoch 2435/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0692 - acc: 0.9400     \n",
      "Epoch 2436/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0691 - acc: 0.9400     \n",
      "Epoch 2437/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0690 - acc: 0.9400     \n",
      "Epoch 2438/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0687 - acc: 0.9350     \n",
      "Epoch 2439/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0687 - acc: 0.9400     \n",
      "Epoch 2440/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0685 - acc: 0.9400     \n",
      "Epoch 2441/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0686 - acc: 0.9300     \n",
      "Epoch 2442/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0685 - acc: 0.9250     \n",
      "Epoch 2443/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0687 - acc: 0.9400     \n",
      "Epoch 2444/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0686 - acc: 0.9350     \n",
      "Epoch 2445/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0684 - acc: 0.9400     \n",
      "Epoch 2446/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0683 - acc: 0.9400     \n",
      "Epoch 2447/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0683 - acc: 0.9250     \n",
      "Epoch 2448/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0683 - acc: 0.9250     \n",
      "Epoch 2449/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0683 - acc: 0.9250     \n",
      "Epoch 2450/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0682 - acc: 0.9300     \n",
      "Epoch 2451/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0682 - acc: 0.9300     \n",
      "Epoch 2452/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0681 - acc: 0.9300     \n",
      "Epoch 2453/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0681 - acc: 0.9300     \n",
      "Epoch 2454/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0681 - acc: 0.9300     \n",
      "Epoch 2455/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0681 - acc: 0.9250     \n",
      "Epoch 2456/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0681 - acc: 0.9300     \n",
      "Epoch 2457/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0680 - acc: 0.9300     \n",
      "Epoch 2458/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0682 - acc: 0.9400     \n",
      "Epoch 2459/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0681 - acc: 0.9400     \n",
      "Epoch 2460/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0680 - acc: 0.9400     \n",
      "Epoch 2461/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0680 - acc: 0.9400     \n",
      "Epoch 2462/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0679 - acc: 0.9400     \n",
      "Epoch 2463/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0678 - acc: 0.9400     \n",
      "Epoch 2464/2500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200/200 [==============================] - 0s - loss: 0.0678 - acc: 0.9400     \n",
      "Epoch 2465/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0677 - acc: 0.9300     \n",
      "Epoch 2466/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0678 - acc: 0.9300     \n",
      "Epoch 2467/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0677 - acc: 0.9300     \n",
      "Epoch 2468/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0677 - acc: 0.9300     \n",
      "Epoch 2469/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0677 - acc: 0.9300     \n",
      "Epoch 2470/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0676 - acc: 0.9350     \n",
      "Epoch 2471/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0678 - acc: 0.9400     \n",
      "Epoch 2472/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0676 - acc: 0.9400     \n",
      "Epoch 2473/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0676 - acc: 0.9400     \n",
      "Epoch 2474/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0675 - acc: 0.9300     \n",
      "Epoch 2475/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0675 - acc: 0.9300     \n",
      "Epoch 2476/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0675 - acc: 0.9300     \n",
      "Epoch 2477/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0675 - acc: 0.9300     \n",
      "Epoch 2478/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0676 - acc: 0.9300     \n",
      "Epoch 2479/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0677 - acc: 0.9300     \n",
      "Epoch 2480/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0675 - acc: 0.9300     \n",
      "Epoch 2481/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0675 - acc: 0.9300     \n",
      "Epoch 2482/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0674 - acc: 0.9300     \n",
      "Epoch 2483/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0674 - acc: 0.9300     \n",
      "Epoch 2484/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0673 - acc: 0.9300     \n",
      "Epoch 2485/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0671 - acc: 0.9300     \n",
      "Epoch 2486/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0672 - acc: 0.9400     \n",
      "Epoch 2487/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0672 - acc: 0.9400     \n",
      "Epoch 2488/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0672 - acc: 0.9400     \n",
      "Epoch 2489/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0671 - acc: 0.9400     \n",
      "Epoch 2490/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0671 - acc: 0.9400     \n",
      "Epoch 2491/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0671 - acc: 0.9400     \n",
      "Epoch 2492/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0671 - acc: 0.9400     \n",
      "Epoch 2493/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0670 - acc: 0.9400     \n",
      "Epoch 2494/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0670 - acc: 0.9400     \n",
      "Epoch 2495/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0669 - acc: 0.9400     \n",
      "Epoch 2496/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0669 - acc: 0.9400     \n",
      "Epoch 2497/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0668 - acc: 0.9400     \n",
      "Epoch 2498/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0669 - acc: 0.9400     \n",
      "Epoch 2499/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0669 - acc: 0.9400     \n",
      "Epoch 2500/2500\n",
      "200/200 [==============================] - 0s - loss: 0.0669 - acc: 0.9400     \n"
     ]
    }
   ],
   "source": [
    "hist = model.fit(X,y,epochs=2500,batch_size=32,verbose=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here I have just retained the history object output by fitting the network. It just contains information on the loss and accuracy of the learning process."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZEAAAD8CAYAAAC2PJlnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAIABJREFUeJzt3XmYVNW19/HvgqaZBGVyYlBQoqJR\nIYjmapQMRtRETdREEq+Y5Iqzcbpv8GrUmMGBRE3iEIkhagbnaIhKcIIYjSgYBQEFASdAQZkDzdjr\n/WNVWdXd1d3VRXdXne7f53nqqapz9qmzTlV3rdp7n7O3uTsiIiKFaFPsAEREJLmUREREpGBKIiIi\nUjAlERERKZiSiIiIFExJRERECqYkIiIiBVMSERGRgimJiIhIwcqKteM2bdp4x44di7V7EZFEWr9+\nvbt7yVQAipZEOnbsyLp164q1exGRRDKzimLHkK1kspmIiCSPkoiIiBRMSURERAqmJCIiIgVTEhER\nkYIpiYiISMGUREREpGBFu06kYM8/D5MmwZVXQrt2xY5GROQT7nDppXDjjXD33fDRR7B0Key+O+yz\nD0yeHOW++lU46KCihtporFhzrHfu3NkLutjw5z+H//1fWLsWttuu8QMTESnQ/PkwcGDudTvuCMuW\ngRncdhucdVZh+zCz9e7eufAoG1fyaiJlqZC3bCluHCIiwIIFUQPZYw9Ytar2csuXw0knwYMPNl9s\nzSF5fSLpJqzNm4sbh4i0eq+/DnvuGbWP3/wG1qypvezWrdC+ffPF1lyUREREGqiyMm4LFmSWTZ0K\nK1bUvV1LTCLJa85SEhGRInrlFRg6tObye+6JW106dWqamIopeUlEfSIiUkTz5zes/I9+BIsXQ79+\n8M1vNk1MxZS8JKKaiIg0s2nTYNiwwra98srGjaXUKImIiNTj4INrX7fffjBrVnSwf+Mb0LUr3HUX\nvPkm/N//NVuIRZO8JKLmLBFpBI8/DvffDyNGQM+e0L073H57nHE1ciQ8/XQkgq9/PU7hre6jj2K7\nXH7wg6aNvZQk72LDJ56AY4+Fl14qvH4pIq2e2bZtv2lTcQbN0MWG20rNWSJSgA8/hEcfhfJy6Nu3\n8Ncp0u/ukpW8JJJuzlISEZEG2GWXYkfQMiUviaRrIuoTEZF6bN4czVb1NV1NnAgdO8Zv1LVrYfvt\noU2bODV3//1jWZ8+0LZt88SdJMlNIqqJiEgO6YTxs5/lf3bUiBFNF09Ll7wkouYsEcnDjBlVnw8c\nCG+9FY+PPDJG1f32t6FzyXRRJ1NeScTMRgC/BNoCd7r7ddXW7waMB3oBK4BT3X1RI8caVBMRkSyT\nJsF3vgMffFB1+f33V31+1VWRNKRx1TsAo5m1BW4FjgYGASPNbFC1Yj8H7nH3/YFrgGsbO9BPqE9E\nRLKMGFEzgVT3la/A4Yc3TzytTT41kWHAfHdfCGBm9wHHA3OyygwCLko9ngw82phBVqGaiEirN2NG\ndIbne6nZ3/7WtPG0Zvkkkd7A+1nPFwHVBwGYAZxINHl9DehiZj3cfXl2ITMbDYwGKC8vLzBi9YmI\ntHYnnADvvFN3mZ49YbfdYPjw5oio9conieQ6Oa765TaXAreY2enAc8BioEZ7k7uPA8ZBXLHeoEjT\n1Jwl0mrMmRNzk5vBq69G4qisrJpANmyIeTrSZ2XpYsDmlU8SWQRkX9/ZB1iSXcDdlwBfBzCz7YAT\n3X11YwVZhZqzRFqFf/4z+jFuvRXOOQeGDMldLj3RU3ogRGle+cxsOA0YaGb9zawcOAWYkF3AzHqa\nWfq1LiPO1Goaas4SaRUWLoz7F1+Mcarq8/LLsHJl08YkNdWbRNx9C3AeMAl4A3jA3Web2TVmdlyq\n2HBgrpnNA3YCftpE8aomItICXXJJzavKr7gi7v/4x9qnld1hh8zjjh2rPpfmkbxRfDdsiL+Wa6+F\nMWMaPzARaXbpBFJZmXlc21Alp54KFRXR1PW5z8Hgwc0TY6nQKL7bSs1ZIi1KZWXm8ec/D//4R+1l\nzzor5vyQ0pFPn0hpSY+ApiQi0iL85z+Zx3UlEICLLqp7vTS/5CURs+gXURIRaRHuvjv/snvu2XRx\nSGGSl0Qgetk2bix2FCKyjTZuhAsuqH39McfEBYP77RcTSbVJ5jdWszKzEWY218zmm1mNjmMzu9jM\n5pjZTDN7JjX2YXrdVjN7LXWbUH3bXJLXJwLRsb5hQ7GjEJEC/PvfsOuuMUf5U0/lLnPbbXD22c0b\nV0uQNdbhkcQ1ftPMbIK7Zw9T9Sow1N3Xm9nZwA3AN1PrKtz9wIbsM5lJpEOHOD1DRBKlogI+8xk4\n8EB47bXay/Xq1XwxtTD1jnXo7pOzyk8FTt2WHSazcqiaiEjJq6yErVsz3ZebN2dG282VQA4/PP6t\n33gDTjyx+eJMoDIzm551G521LtdYh73reK3vAROznndIveZUMzshr2DyDruUqCYiUtKqX+MxZgxc\nd13usmmHHhrdnXvv3XRxtRBb3H1oLevyGeswCpqdCgwFjsha3M/dl5jZAOBZM3vd3RfUFUwyk4hq\nIiIlK/u6j7TaEsiFF0bnes+emSvUZZvUO9YhgJl9CbgcOMLdPzlLKTUOIu6+0MymAIOBFphEVBMR\nKSmLF8Npp8Ett+Q/r/no0XDTTU0bVyv0yViHxGjqpwDfyi5gZoOBO4AR7r4sa3k3YL27bzSznsCh\nRKd7ndQnIiLbbMIEePbZuJr80RxT0mXP6fGZz0TCufDCZguv1chzrMOxwHbAg9VO5d0HmG5mM4jJ\nBa+rdlZXTskbOwvga1+DBQtg5szGDUpEavXkk7B2bVxh3qsXHHts9GN89BHMmxdlOnWC9etrbqs5\nPhqPxs5qDKqJiDS7o46queyFF6o+z5VApGVLZnOW+kREmszq1fDXv8Ly5TGb4D//CY88kv/2V10V\nNY/sm7RcqomISBVjx8JPt2FGoD59Gi8WKX3JTCIdOiiJiDTApk1x7UbbtjH+1KpV8VusrCwuCHSP\nf6ny8syMgg318svQubOu82htktmc1bGjmrOkVXvssUgKb72VWXbmmbkncnrssfiXKS+HkSPhz3+G\nbt1iYMOjj44L/Dp0iFkBO3WCe++tfb9HHFH7uoMOgkGDNEhia5PMmkjHjpnxFNLT5Yq0IvfcE/ev\nvAIDB8bjcePi3r1qMnnsscwFgA88kKkpLF1a+wCIaSeeCA8/HI/33z8eP/IITJ8OhxwSNZnhw+HD\nDxvlsCSJ3L0ot06dOnnBbrwx+utWriz8NXL44Q/jZQ880H369EZ96bycf777n//c/PudN8/98MPd\nFy6svcwVV7j/6lcNf+2bb4739IUXCo9vW23Z4v6Vr7gPH+5+zTV1lx01yv3RRzPPb7vNfcyY3GWf\neiqO7Vvfiuf33x/Pp0+P+5NPzpS9446qXc1PPJFZt2GDe69emXVXXeU+ebJ7587Vu6dz3/bdt+rz\nQYNqLsu+9eiR3+tCxHfGGfH4L39p4BsvTQJY50X63s51S2YSSf9HLlpU+GvkkP3Pc/PNjfrSDdp/\ncxszJvY7dmztZQqNLb3dqFEFh7fNli6t+cWYS2VlzTJ1bfPFL1Zdn3582GG1v076NmBAZt2LL9Zc\nf+yx+X/RH3NM1ecnnhi39PPjjov7rl3dBw6MdeedF8kv1+tdcon7iBHuEydGfAsWuJ95pvvatYW9\n/9K4lEQaI4n88Y8R+rx5hb9GNZs3V/1HGjzY/ZZb3EePjn/okSPdTzklaing3q6d+znnuHfv7n7S\nSfH4nHPchw1z79jR/de/jnL9+0cNZ+vWmvv87W/d33svHt9yS+4vrAcfdJ8xo+a2W7a4H320+6WX\nuj//fHwpHHmk+yOP1Cw7bZr72We79+zpfu658SvXPX5Zpt9KcP/FL2p/f7JjGzs2fjl///vub79d\ntdzEie7/+lfN7QYMcL/ySvc773S/7z73Z591//nPM+VWrIhkln4/brrJ/ZVXcscyYYL7yy+7z5oV\nx7R6dc33a9Mm9+uvd1+/3v2tt3InkbFjq76369dnyrRvn0mu4D51aqbclClxHIcfnlm/1165v5Af\neMD9tNNyrxs/Pv62vvrV/BNG9duoUe4ff5z7+NLPKytr/1z33z/3352ULiWRxkgijzwSob/6auGv\nUc2rrxb+jwzuO+wQX2i1rZ87t+r+Vq6M5fvsU/UXcPV/5tr+wf/859r3Vd2oUbnLVF92++21vz/p\nMps2Vd3m8svrjresLHeM6eVr1kS53/8+nl90UebLPPvXeq59pG9HH11z3bhxcX/11ZGMqh/7Rx/F\n4/33z2z74Yd1f8ZpbdrE8+OP37a/mUJvw4a5d+kSj197LX5QDBwYz888MxPn7be7Dx1a+2fq7v73\nv8d2115bdzkpHaWWRJLZsd45dcV/HsOmbNoEv/51nHXSoUNMs3nQQTFMQ69ecZYKwJo1cf/MM3DC\nCTG8Q30+/3mYPDkev/xydHDmOjsG4C9/iX1VVMDHH2eGiXjjjZrn5F91FfTvHxd7pU2YEF8hn/oU\njB9f94gvl10Wx7v99tClS+45rM86q+aym26CFSviPUt/ZZWVxSx0aeeeW3WbmTPh2muhR4+YsS7t\niiti2733hlmzau5ry5a4HzMm5pH4zW8yMbRtG48XLoRrronXWLIk9tG+fc3XmjgR/t//i+NN+9Wv\n4v7BB+H996uWv/hi2G67TPzjxsWZSUuX1nztbL/9LUydmumkXry47vK1qayE7t3jNNs097gfMADe\nfrvqsvTf1NattZ/5lP57ynbWWbk/52xHHZXZj0hBipW9tqkm8vzz8R03aVK9Rb/3vdy/KMF9zz0z\n5f72t1g2bZp7t275/SL8zW8yj1etyryubrrVdXOv2ox0wAGZv8Orr45lw4dnlp1wQnSyi7i7o5pI\nI8izJrJqVZzSWF36F+f8+ZnH6V+rXbvG/AbVffRR3HfqBMuWxS/Jrl2jNtK1a+ZX8KpV8O678Wt6\ny5ao/ZSVxWvuu28DjzPlxz+GH/6wsG3rc9FFcMYZcX5/XV5/HX7/e7jxxnj+9NNRW1q1KkZlrc1n\nPwsvvlhz+V57wdy5+cX48MOZme5mz44zu088Mcbg/PWv46v4ggtqbjd9OgxNTd3z5pt1XwQ3e3bc\nb90an9umTVHrqayMmkBZWZziCnD++TGMeZs2sPvuMVvfpk1x6949/lbSf0dmUWb16vhbKC+P13ju\nuajJdOgAO+6YieOHP4zrPXr2zCy7//68Kt0ixVGs7LUtNZFlL8xzB5/8vT/kXP/664X/Sly2zH3n\nnXP/etxWDYljt93ivk2bTBdQIbe6+mkgczprPr+eH34483zDhlhWvT+n+u1HP8r9fl50Uf7HsGRJ\n3A8blnkvzzwzls2aFZ30tcWc/biuDuyGfH733rvNfwoiBUM1kW23cGlnegHTpqxjeI71//xnzWXn\nnBO1g8rKqBHMmgW9e2f6RCDa/nv1ivkQ7r8fdtoJDjggrm1sDK+8AtOmwUsvRZ/L9tvHqPZTpkQb\n/XnnxaQ+O+4Yv7SXLIlfpD17Ri1g+fKI/8kn4ec/jzb6du1g0aIYr6iiIi78mjwZ9twThgyJY334\nYfjGN+KYpkyJIbw/+CD6cI49NmKbOjV+hU+cGPsbMCBqaW++CaefHmW+9rW42rlPn0zfhBn861/R\nX/HUUzB4cNTWnn46ahqXXBK/2h95JF5vw4aI8Ygj4mK1zZuj5rZ0aWy7fj289168N+3aRQ1gl11g\n0qSqNYmbb4aTTorj22cf+NnPYn3//jF/9w47RLn33ot+HoCHHoLjjourtEeOhHfeyVypnY93343j\n/+Y3t+WvQKRlSeR8Ij/7wWr+74YduLH3L7h40cWfLF+0CPr2zb1NkQ5TRKRRldp8Iokc5ebKG+L9\n27SyahK65Zbc5Z95pqkjEhFpnRLZnLWVMjZSTsfKTBK5/364/vpMmV12ieYgERFpOomsiQCsozMd\nKjPTqJ1yStX16cHoRESk6SSyJgKwnk6UbVrH4sXwj39UXbdgQXQMi4hI00psEllHZzqzLucsat27\nN388IiKtUaKbszpTtWO9rCxOFc33lE0REdk2LSqJnHZa1at/RUSkaeWVRMxshJnNNbP5ZjYmx/p+\nZjbZzF41s5lmdkzjh1pVriRy5ZVNvVcREclWbxIxs7bArcDRwCBgpJlVH2npCuABdx8MnALc1tiB\nZisry51EunZtyr2KiEh1+dREhgHz3X2hu28C7gOOr1bGgfRX+PZAk16hccEFNZPIuedWHcJERESa\nXj5nZ/UGsmdkWAQcXK3M1cCTZnY+0Bn4UqNEV4sOHTJJZOxYuPTSptybiIjUJp+aSK5plqqPRDUS\nuMvd+wDHAH8wsxqvbWajzWy6mU3fkp6VqABmcZ1IZ9bVuMhQRESaTz5JZBGQPaxhH2o2V30PeADA\n3V8EOgA9q5XB3ce5+1B3H1pWVvglKu5QYZ3pzHr69NbIiiIixZJPEpkGDDSz/mZWTnScT6hW5j3g\niwBmtg+RRD5qzECrW2epQSwrKppyNyIiUod6k4i7bwHOAyYBbxBnYc02s2vM7LhUsUuAM8xsBnAv\ncLo34Rjz7rCe/OdZFxGRppFXm5K7PwE8UW3ZlVmP5wCHNm5odcWTVRNZty5mkhIRkWaX2CvWVRMR\nESm+RCYRd1hvSiIiIsWW2CRSYZ3iiZKIiEjRJDKJgGoiIiKlIJFJxD2uWAeUREREsuQxYO7FZjYn\nNVjuM2a2W9a6UWb2Vuo2Kp/9JTaJqCYiIlJVngPmvgoMdff9gYeAG1LbdgeuIoa1GgZcZWb1jkiY\nyCQCUNEmlUTWr6+7oIhI61HvgLnuPtnd01+cU4lRSACOAp5y9xXuvhJ4ChhR3w4TmUTUnCUiklOu\nAXN711H+e8DEArcFEjzH+gbrGCMxKomISOtSZmbTs56Pc/dxqcf5DJgbBc1OBYYCRzR02yrB1Feg\nFLkTCaRTJyUREWlttrj70FrW5TNgLmb2JeBy4Ah335i17fBq206pL5jENmeZAZ07K4mIiGTUO2Cu\nmQ0G7gCOc/dlWasmAV82s26pDvUvp5bVKZE1EUglkS5dYO3aYociIlIS3H2LmaUHzG0LjE8PmAtM\nd/cJwFhgO+BBMwN4z92Pc/cVZvZjIhEBXOPuK+rbZyKTyCfjA3fpAmvWFDUWEZFSkseAubXOPOvu\n44HxDdlfspuzunZVTUREpIgSmUQgK4moJiIiUjSJTCJqzhIRKQ2JTSJqzhIRKb5EJhFQc5aISClI\nZBL5pDmra1fYsAE2by5qPCIirVUikwhkXScCatISESmSRCaRKjURUJOWiEiRJDaJfNInAkoiIiJF\nksgkAmrOEhEpBYlMImrOEhEpDYlNImrOEhEpvkQmEVBzlohIKUhkElFzlohIaUhsEjEDttsuFiiJ\niIgURSKTCKSSSNu2MbuhmrNERIoikUnEs6eO1/hZIiJFk8gkAqmaCGgkXxGRIkpkEqlSE9GcIiIi\nRZPYJFKlJqIkIiJSFIlMIqDmLBGRUpDIJKLmLBGR0pDYJFKlJrJ6dVHjERFprfJKImY2wszmmtl8\nMxuTY/1NZvZa6jbPzFY1fqjV95l60K1bJJHKyqbepYiIVFNWXwEzawvcChwJLAKmmdkEd5+TLuPu\nF2WVPx8Y3ASxfqJKc1b37pFA1qyBHXZoyt2KiEg1+dREhgHz3X2hu28C7gOOr6P8SODexgiuNlWa\ns3r0iPvly5tylyIikkM+SaQ38H7W80WpZTWY2W5Af+DZbQ+tbp8kke7d437FiqbepYiIVFNvcxZg\nOZZ5jmUApwAPufvWnC9kNhoYDVBeXp5XgLnUaM4CJRERkSLIpyayCOib9bwPsKSWsqdQR1OWu49z\n96HuPrSsLJ/8VdvrqDlLRKQU5JNEpgEDzay/mZUTiWJC9UJmthfQDXixcUPMTc1ZIiLFV28Scfct\nwHnAJOAN4AF3n21m15jZcVlFRwL3uXttTV2NpsoeunWLeyUREZFml1ebkrs/ATxRbdmV1Z5f3Xhh\n1e+TmkhZWVxwqOYsEZFml9gr1qvo0UM1ERGRIkhsErHsc8a6d1cSEREpgkQmEciRRNScJSLS7BKZ\nRNScJSJSGhKbRKrURHr0gI8/Llo8IiKtVSKTCFRLIjvtBCtXwsaNRYtHRKQU5DHq+uFm9m8z22Jm\nJ1VbtzVrRPYa1wPmUvhl40VUozlr553jftky6Nu3RnkRkdYgn1HXgfeA04FLc7xEhbsf2JB9JrIm\nUqM5K51EPvywKPGIiJSIekddd/d33H0m0CiTMCUyiYCSiIhIDnmPul6LDmY23cymmtkJ+WzQMpqz\ndtop7pcubfZYRESaWZmZTc96Ps7dx6UeN2TU9Vz6ufsSMxsAPGtmr7v7gjqDacCLl5QaHeugmoiI\ntAZb3H1oLesaMup6De6+JHW/0MymELPU1plEEtmcVaMm0r59DMSoJCIirVteo67nYmbdzKx96nFP\n4FBgTt1bJTiJWPVK2667wuLFRYlHRKQU5DPqupkdZGaLgJOBO8xsdmrzfYDpZjYDmAxcV+2srpxa\nRnMWQL9+8N57RYlFRKRU1DfqurtPI5q5qm/3L+DTDd1fYmsiNey2G7z7brPHIiLSmiU2ieSsiSxf\nDuvWFSUmEZHWKJFJBHIkkd12i3s1aYmINJtEJpFam7NATVoiIs0osUkkZ3MWKImIiDSjRCYRqOUU\n37IyJRERkWaUyCSSszmrbVvo00d9IiIizSiRSQRy1EQgmrRUExERaTaJTCI5ayIA/fvDgjqHeRER\nkUaU2CSSsyay337wwQeaKldEpJkkMolALUnkgAPi/vXXmzUWEZHWKpFJpNbmrP33j/uZM5stFhGR\n1iyxSSRnTWSnnWDHHWHGjGaPSUSkNUpkEoFakghEbUQ1ERGRZpHIJFJrcxZEv8js2bBlS7PFIyLS\nWiU2idRaExkyBDZsgFmzmjUmEZHWKJFJBOpIIp/7XNw/91yzxSIi0lolMonU2ZzVty/svruSiIhI\nM0hkEoE6aiIQtZHnnqsn24iIyLZKZBKpNzccfjh89BHMndss8YiItFaJTSJ11kS+8IW4f/zxZolH\nRKS1SmQSgXqSyIABMHQo3Hdfs8UjItIa5ZVEzGyEmc01s/lmNqaWMt8wszlmNtvM/ty4YVaVV1fH\nKafA9Okwf35ThiIi0qrVm0TMrC1wK3A0MAgYaWaDqpUZCFwGHOru+wIXNkGsn6i3OQvgG9+I+z/9\nqSlDERFp1fKpiQwD5rv7QnffBNwHHF+tzBnAre6+EsDdlzVumDXVm0T69oURI+D22+PiQxERaXT5\nJJHewPtZzxellmX7FPApM3vBzKaa2YjGCjCXvM/cvfRSWLoU7ryzKcMREWm18kkiuX7zV/8aLwMG\nAsOBkcCdZrZDjRcyG21m081s+pZtGNsqr+YsiLO0DjoIrr0W1q8veH8iIpJbPklkEdA363kfYEmO\nMn91983u/jYwl0gqVbj7OHcf6u5Dy8rKCo0ZyDOJmMFPfwpLlsAZZ2zT/kREpKZ8ksg0YKCZ9Tez\ncuAUYEK1Mo8Cnwcws55E89bCxgw0W4MuRD/ySNhnH3jgAVjYZCGJiLRK9SYRd98CnAdMAt4AHnD3\n2WZ2jZkdlyo2CVhuZnOAycD/uvvypgoa8qyJpD3+OHTuDP/93xoiXkSkEZkXaXypzp07+7p16wra\ndvjwqI384x8N2Ojee+Fb34LTToO77mpgFhIRKQ1mtt7dOxc7jrRt65gokrw71rONHAlz5sBPfgLt\n2sG4cdAmsRfsi4iUhEQmESiwIvGjH8HGjTB2LEydChMnxvUkIiJSkET+FC+4Ba5NG7j+evjFL2IK\n3WHD4IorIrGIiEiDJTaJFNylYQYXXxzzjXToEKcA9+wJJ58cnSwVFY0aq4hIS5bIJAKN0C/+uc/F\n4Izjx0OvXvDQQ9Fj36lT1FjM4vaHP8DmzZG5NMmViEgViTw767DDoH17eOaZRgxo9mx47LHofL/n\nnprr27fPNHvts0/c77BDlN9770g03bpFuVdfhQMOiKvlp0+PCbLmzYPvfhduvjnKffe7kcRWr4Yp\nU+B//ieSVNeusGgRfPrTcZX9Y49FjemAA6Bfv3itt96C8nL4299i2Ptdd43Xb9cOvv3teL3//AcO\nOSTi2m8/mDUrEuLFF8PWrXEse+wR+7jxxojjsMMiuS5eHMfRt2+sX7MmTpF+++1Itr17w9FHZ4aT\nOflkOPhgWL4c1q6NcvPnw1e+EjEuXRqvD/DlL8OBB8LMmXE/YACsWwdPPQU77hgxV1TARRfBJZdE\nuRdfjKbHqVNh553h2GNhwYLY56BB8K9/wbPPwn/9V7zemjVRu6yoiM9g06Z4T2fMgFtuideeNg2e\nfz62v/VWePNN6NED3nsvjnW//aKWesMNsf2ee8b7W1YGH3wQn1N5ebwf++4LgwfHZ/POO3EM77wD\nffpEeZFGVGpnZyUyiRx6KHTsCE8/3chBZVuzBlasiE74HXeMxzNnxhf0fvvFl3BlZXw57bFHJBOR\nbO3aRTLr1w922y1+PGzeHIlq992j+fTjjyMZd+sWF8P26hVlhgyJMqtXQ5cu0LZtsY9GSkR9SSQ1\nduEvgbbAne5+XbX1hwM3A/sDp7j7Q1nrRgFXpJ7+xN3vrjeepCaRTp3ix2tJym76SjeLVVZGM1l6\nuXt8WZSVxbrKyviiqKiIX7ibN0eNoV27qtutXx/Lu3SJRNahQyzbuBG23z4upty8OdMkV14er925\nc6xbty7zuhUV0L171Dx69IjMvGpVptbVrl2UbdMmXmfTpohjzZpIrFu3xv4rKmIfEDWR8vLYfuvW\neL7zzrBsWdTcttsulq9cmdnWPeJq0ybidIcPP4w4dtghajjbbRfLN26EXXaJY968OWLq2DHKbNkS\n+161KrZL92+l39+KCvj3v+MLu1u3KLtkSdQqPv444lm7Nl6rT5+o4XzpS/Dww1GDgtjHfffFZ/XF\nL8aX/l/+EjWg99+PGl+/fnH/6U9HEujaNWJdvjz201C9esH++8d7vmABvPxyXDi7885RGxs+POLu\n21fJphWoK4mkpu6YBxxJDEcw6BdWAAALfElEQVQ1DRjp7nOyyuwOdAUuBSakk4iZdQemA0OJ8RFf\nAT6THp291niSmET+67/iu6Zkk4hIbdJnhVRWRiJcuTIS7MqVkWReeCESaFlZJPF77oH+/SORfPhh\n1GLqstNOcPrpMf3B2rUxknW6+VVahHqSyGeBq939qNTzywDc/docZe8CHstKIiOB4e5+Zur5HcAU\nd7+3rngS22CrC84lkdJ/uG3aRO1qu+2qXqv0zW9WLX93jtaERYuiz2X+/Ognmj8/am3TpkX/0/XX\nZ8qOH19125/8JPqU9toranDS0uSauuPgbdi2+rQfNSQyiegkKWnV+vSJ22GHRa2jupUrYfLk6MMr\nK4tElJ4m+oor4gbRFNa3bzRpnnpqnITQoUOzHYYUrMzMpmc9H+fu41KP85m6ozYFbZvYJKKaiEgt\nunWDr389bpBJGu5xhtorr8RZg6+8EmeXQQwDBDG+XHl5NKGNGROPpdRscfehtazLZ+qO2iwi5oTK\n3nZKfRslsk/kkEOi3/Tvf2/koERam40bYcKE+Gd68MFYtnZt3PftG0lk/foYwPSII4oXp3yinj6R\nMqJj/YvAYqJj/VvuPjtH2buo2ifSnehMH5Iq8m+iY31FXfEk8mJDNWeJNJL27eM6n9/9Ls66W706\naimXXBLXAy1YENfFDB8ez3/wg0zTmJScfKbuMLODzGwRcDJwh5nNTm27AvgxkXimAdfUl0AgoTWR\nYcPijNSJExs5KBGpasOGuEj1yiszpylDXMOyeHGcTXbQQUUNsbUptYsNE1kTAfWJiDSLDh1iaukP\nPohrXN59F44/Ps4Q27w5ftGZRZlZs4odrRRBIpOImrNEiqRfP3j00cyV91deGRe53nlnXFw5eHAM\nF7NgQbEjlWaS2CSimohIkfXtG3P0rFoVTVvXXRdDSVxxRYw1ZhanIr/0kn75tWCJTCKgJCJSUnbd\nNTrdX3gB5s6Fyy+P5YsXx+mUbdrA5z8fnffSoiQyiehHjUgJ+9Sn4sp49zjTK23KlGj6OvDAGHV5\nw4aihSiNJ5FJBFQTEUmEz3wmM2jnXXfFsC4zZsTglR07RvPXn/6kX4YJlsgkor83kYTp0gVGjYoR\nkNevj6vhIUZWPvXUGGzyjDOif0USJbFJRDURkYTq2BGuvTb+kd99N+Z22HHHzLwqRxwR16YsX65f\njAmQyCQCSiIiLUK/fjHD5OLF8NxzcMEFcXrwaafFhF7dusXkXen5aqTkJDKJ6MeJSAtjFlMz//KX\nkUR+97uYunj16hhyZfDgqKFMmhSzjErJSGwSUU1EpIVq3x6++12YPTuujP/Vr2LZc8/BiBEx5tEP\nfpAZgkWKKpFJBJRERFqF3r3h/PNjSuDHH88sv+GGaO7ad98YhXjLluLF2MolMomoOUukFTrmmPjn\nd4dXX4WhQ2HOnBjLa5dd4OyzYzIuJZRmldgkopqISCt24IExHfB//gO33BIjCd99N3zhC3G6cP/+\n8J3v6BdnM0hkEgElEREBOneGc8+FJ56ADz+MDvmuXWMO+rvuiuFWtt8+rqDfurXY0bZIiUwi+nEh\nIjV07Rod8qtXRzPX6NHw2c/G1fI//GHMN28GN94ImzYVO9oWI5GTUu23H+y9d2Z6aBGRWi1dCnfc\nAQ8/DDNnxrLtt4djj4WddoKrrornCVFqk1IlMonsuy/ss4+SiIg0UEVFDL3y9NPwyCPxHKIWc+GF\nMfRK794l3V6uJJKyLUlk0KBIJA8+2MhBiUjrsXkzPPMMPPkk3HRT1XUHHxzNYI8+GqMSl5BSSyKJ\n7BOBkv6hICJJ0K5dXLx4443R0TpvXqbv5KWX4I03YK+94svmiCPgttvi4kepIpFJRB3rItLoBg6E\na66JGsqqVdHsdfzxse655+IssL59Y/j6738fFi7UmF7kmUTMbISZzTWz+WY2Jsf6083sIzN7LXX7\nn8YPNUPXiYhIk9p++5j75NFHI1GsXg1jx8YMjhUVMRTLHnvEKcYnnAAdOsDvf98qL3Sst0/EzNoC\n84AjgUXANGCku8/JKnM6MNTdz8t3x9vSJ7L33nGt0X33FbS5iEjhtm6NibWmTo0ZGp96Kve0vy++\nCEOGQHl5o+6+1PpEyvIoMwyY7+4LAczsPuB4YE6dWzUhNWeJSNG0bRvJYcgQOOecqH289FI0ef3u\ndzEKMcQ1KgADBsCnPx1nBI0eDbvt1qKaUvJpzuoNvJ/1fFFqWXUnmtlMM3vIzPo2SnQ5jB8f/V8t\n6DMQkSQrK4uJtS67DObPj1+5s2dHR/zQoTHa8F//GhNx9e8f43z96U/FjrrR5FMTyfV1Xb0u8Dfg\nXnffaGZnAXcDX6jxQmajgdEA5QVW8Xr0gJNOgtNPL2hzEZGmN2hQ3M4+O55XVsbgkK+8Emd97bpr\nceNrRPn0iXwWuNrdj0o9vwzA3a+tpXxbYIW713kJ6Lb0iYiItFal1ieST3PWNGCgmfU3s3LgFGBC\ndgEz2yXr6XHAG40XooiIlKp6m7PcfYuZnQdMAtoC4919tpldA0x39wnABWZ2HLAFWAGc3oQxi4hI\niUjksCciIq1VEpuzREREclISERGRgimJiIhIwZRERESkYEoiIiJSsKKdnWVmlUBFgZuXEacTtyY6\n5tZBx9w6bMsxd3T3kqkAFC2JbAszm+7uQ4sdR3PSMbcOOubWoSUdc8lkMxERSR4lERERKVhSk8i4\nYgdQBDrm1kHH3Dq0mGNOZJ+IiIiUhqTWREREpAQkLomY2Qgzm2tm881sTLHjaUxm9o6ZvW5mr5nZ\n9NSy7mb2lJm9lbrvllpuZvar1Psw08yGFDf6/JjZeDNbZmazspY1+BjNbFSq/FtmNqoYx5KPWo73\najNbnPqcXzOzY7LWXZY63rlmdlTW8sT83ZtZXzObbGZvmNlsM/t+anlL/pxrO+YW/VkD4O6JuRFD\n0S8ABgDlwAxgULHjasTjewfoWW3ZDcCY1OMxwPWpx8cAE4mZJw8BXip2/Hke4+HAEGBWoccIdAcW\npu67pR53K/axNeB4rwYuzVF2UOpvuj3QP/W33jZpf/fALsCQ1OMuwLzUsbXkz7m2Y27Rn7W7J64m\nMgyY7+4L3X0TcB9wfJFjamrHE9MNk7o/IWv5PR6mAjtUmxysJLn7c8ScM9kaeoxHAU+5+wp3Xwk8\nBYxo+ugbrpbjrc3xwH3uvtHd3wbmE3/zifq7d/cP3P3fqcdriUnqetOyP+fajrk2LeKzhuQ1Z/UG\n3s96voi6P6ikceBJM3slNR89wE7u/gHEHyqwY2p5S3ovGnqMLeHYz0s13YxPN+vQAo/XzHYHBgMv\n0Uo+52rHDC38s05aErEcy1rS6WWHuvsQ4GjgXDM7vI6yLf29gNqPMenHfjuwB3Ag8AHwi9TyFnW8\nZrYd8DBwobuvqatojmWJPO4cx9ziP+ukJZFFQN+s532AJUWKpdG5+5LU/TLgEaJquzTdTJW6X5Yq\n3pLei4YeY6KP3d2XuvtWd68Efkt8ztCCjtfM2hFfpn9y97+kFrfozznXMbeGzzppSWQaMNDM+ptZ\nOXAKMKHIMTUKM+tsZl3Sj4EvA7OI40uflTIK+Gvq8QTgtNSZLYcAq9NNBQnU0GOcBHzZzLqlmge+\nnFqWCNX6rr5GfM4Qx3uKmbU3s/7AQOBlEvZ3b2YG/A54w91vzFrVYj/n2o65pX/WQLLOzvLMmRzz\niDMYLi92PI14XAOIMzFmALPTxwb0AJ4B3krdd08tN+DW1PvwOjC02MeQ53HeS1TrNxO/ur5XyDEC\n3yU6I+cD3yn2cTXweP+QOp6ZxBfELlnlL08d71zg6Kzlifm7Bw4jmmBmAq+lbse08M+5tmNu0Z+1\nu+uKdRERKVzSmrNERKSEKImIiEjBlERERKRgSiIiIlIwJRERESmYkoiIiBRMSURERAqmJCIiIgX7\n/73mS7Mk5n8FAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xb054a239b0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "ax1 = ax.twinx()\n",
    "ax.plot(hist.history['acc'], 'b')\n",
    "ax1.plot(hist.history['loss'], 'r')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The red region in the plot below is the region the network believes objects should fall into the red category, and the same for the blue. We can see that with these very noisy moons, the network does a pretty good job, but cannot entirely separate them. If we'd had more input parameters, perhaps we could have certainly separated the groups."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAD8CAYAAAB+UHOxAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ntAPQaDSaXYp2ABqNRrNL+f/ygfLrA9sOCwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xb054f284e0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "ax.scatter(X[:,0], X[:,1], s=40, c=yc, cmap = plt.cm.Spectral, zorder = 1)\n",
    "xmin, xmax = ax.get_xlim()\n",
    "ymin, ymax = ax.get_ylim()\n",
    "\n",
    "XX, YY = np.mgrid[xmin:xmax:200j, ymin:ymax:200j]\n",
    "X_in = np.c_[XX.ravel(), YY.ravel()]\n",
    "\n",
    "Z = model.predict(X_in)\n",
    "#Z = destandardise(Z,meany,stdy)\n",
    "Z = Z.reshape(XX.shape)\n",
    "\n",
    "Z = np.where(Z > 0.5, 1., 0.)\n",
    "\n",
    "ax.pcolormesh(XX, YY, Z, cmap=plt.cm.RdBu, zorder = 0)\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### More difficult categoristion - MNIST Handwritten Digit Dataset\n",
    "\n",
    "This is a standard example for people new to neural networks. <br>\n",
    "\n",
    "The dataset consists of 60,000 28 x 28 pixel images of handwritten digits, with the correct labels, which are generally used for training a network. There are also 10,000 test examples, which are used to check the network is performing as expected. While not necessarily a useful real world case by itself, it can be used to demonstrate understanding of how neural networks function, as well as benchmarking speed tests of networks, testing accuracy, etc. <br>\n",
    "\n",
    "Luckily, keras comes with a routine to download the data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.datasets import mnist\n",
    "(X_train, y_train), (X_test, y_test) = mnist.load_data()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First 5 images in the dataset. Clearly a 7, 2, 1, 0 and 4. I'll use these numbers to quickly test the network we set up."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWQAAABFCAYAAAB9nJwHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xb057f72860>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "f, axarr = plt.subplots(1,6)\n",
    "for i in range(6):\n",
    "    axarr[i].imshow(X_test[i],cmap=plt.cm.Greys)\n",
    "    axarr[i].xaxis.set_major_locator(plt.NullLocator())\n",
    "    axarr[i].yaxis.set_major_locator(plt.NullLocator())\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I flatten the data out to be a list of 784 pixel values for each image. This is how I have generally seen it done, but keras also allows us to have 2D input layers. You can read about this on your own."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "X_train_flat = X_train.reshape(X_train.shape[0], 784)\n",
    "X_test_flat = X_test.reshape(X_test.shape[0], 784)\n",
    "\n",
    "Y_train = np_utils.to_categorical(y_train, 10)\n",
    "Y_test = np_utils.to_categorical(y_test, 10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Similar to before. How many neurons should we have in the input layer? How many in the output layer?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "num_model = Sequential()\n",
    "num_model.add(Dense(input_dim = 784, units = 30, activation='sigmoid'))\n",
    "#model.add(Activation('sigmoid'))\n",
    "num_model.add(Dense(input_dim = 30, units = 10, activation='softmax'))\n",
    "#model.add(Activation('sigmoid'))\n",
    "num_model.compile(optimizer='rmsprop',\n",
    "              loss='categorical_crossentropy',\n",
    "              metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[5 2 5 5 2]\n"
     ]
    }
   ],
   "source": [
    "print(np.argmax(num_model.predict(X_test_flat[0:5]),axis=1))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Clearly the model does a terrible job at predicting the values of the images.\n",
    "\n",
    "No better than random chance at the moment (this is literally true, as the weights of the neurons are assigned randomly at first).\n",
    "\n",
    "Perhaps training will give a better result. As the dataset contains 60,000 images, we can run through it 30 times to get a reasonable result."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.7431 - acc: 0.8002     \n",
      "Epoch 2/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.3764 - acc: 0.8877     \n",
      "Epoch 3/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.3417 - acc: 0.8970     \n",
      "Epoch 4/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.3200 - acc: 0.9040     \n",
      "Epoch 5/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.3033 - acc: 0.9084     \n",
      "Epoch 6/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2944 - acc: 0.9118     \n",
      "Epoch 7/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2870 - acc: 0.9141     \n",
      "Epoch 8/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2860 - acc: 0.9146     \n",
      "Epoch 9/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2771 - acc: 0.9182     \n",
      "Epoch 10/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2742 - acc: 0.9180     \n",
      "Epoch 11/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2728 - acc: 0.9187     - ETA: 1s - lo\n",
      "Epoch 12/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2688 - acc: 0.9199     \n",
      "Epoch 13/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2661 - acc: 0.9220     - ETA: 0s - loss: 0.2639 - acc: 0.\n",
      "Epoch 14/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2644 - acc: 0.9205     \n",
      "Epoch 15/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2625 - acc: 0.9230     \n",
      "Epoch 16/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2628 - acc: 0.9234     \n",
      "Epoch 17/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2636 - acc: 0.9219     - ETA: 0s - loss: 0.2630 - ac\n",
      "Epoch 18/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2612 - acc: 0.9226     \n",
      "Epoch 19/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2579 - acc: 0.9243     - ETA: 1s - l\n",
      "Epoch 20/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2606 - acc: 0.9235     \n",
      "Epoch 21/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2599 - acc: 0.9228     - ETA: 0s - loss: 0.2590 - acc: 0.\n",
      "Epoch 22/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2615 - acc: 0.9240     \n",
      "Epoch 23/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2576 - acc: 0.9248     \n",
      "Epoch 24/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2596 - acc: 0.9246     - ETA: 1s - loss: 0. - ETA: 0s - loss: 0.2611 - ac\n",
      "Epoch 25/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2546 - acc: 0.9263     \n",
      "Epoch 26/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2536 - acc: 0.9256     \n",
      "Epoch 27/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2496 - acc: 0.9271     \n",
      "Epoch 28/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2513 - acc: 0.9273     \n",
      "Epoch 29/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2504 - acc: 0.9270     \n",
      "Epoch 30/30\n",
      "60000/60000 [==============================] - 3s - loss: 0.2488 - acc: 0.9275     \n"
     ]
    }
   ],
   "source": [
    "hist = num_model.fit(X_train_flat, Y_train, epochs=30, batch_size=32)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZEAAAD8CAYAAAC2PJlnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xb057ff20b8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "ax1 = ax.twinx()\n",
    "ax.plot(hist.history['acc'], 'b')\n",
    "ax1.plot(hist.history['loss'], 'r')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[7 2 1 0 4 1]\n"
     ]
    }
   ],
   "source": [
    "print(np.argmax(num_model.predict(X_test_flat[0:6]),axis=1))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Clearly the code is doing much better now. According to the accuracy plot, it is categorising about 92% of images correctly. Not bad for a few minutes of work!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " 9568/10000 [===========================>..] - ETA: 0s"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.25998682493120434, 0.92210000000000003]"
      ]
     },
     "execution_count": 79,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "num_model.evaluate(X_test_flat, Y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:py36]",
   "language": "python",
   "name": "conda-env-py36-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}