Preceptron.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import keras\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras.optimizers import Adam\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[16.52810469 12.76546486]\n",
      " [13.80031442 11.93151544]\n",
      " [14.95747597 14.19269369]\n",
      " ...\n",
      " [ 7.50641288  6.31686769]\n",
      " [ 5.84131367  3.71619716]\n",
      " [ 7.7715489   3.37805926]]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x2502a996d30>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX4AAAD8CAYAAABw1c+bAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJztvXt0XNWV5//dVSrZJUNLFnYIlm1M8mPBj4cfQQF3m2QAT5s3ESaxgaQ76Tw8TJLJAGmD6c4ygmYGg1fHTNLJJOaxkm4IbROMMK9gFvBrAj0msfEDPIEhPIItMWCwJcAurJJ0fn/ce6pu3Trn3HMf9VLtz1paku7z1K2qfc7ZZ+/vJiEEGIZhmOYhVesGMAzDMNWFDT/DMEyTwYafYRimyWDDzzAM02Sw4WcYhmky2PAzDMM0GWz4GYZhmgw2/AzDME0GG36GYZgmo6XWDVAxZcoUMWvWrFo3g2EYpmHYunXre0KIqTbH1qXhnzVrFrZs2VLrZjAMwzQMRPQn22PZ1cMwDNNksOFnGIZpMtjwMwzDNBls+BmGYZoMNvwMwzBNRmBUDxHdBeACAO8KIU5yt60DcJx7SAeAQSHEXMW5bwL4EMAogBEhRHdC7WYYpoL0bevH6sdfwcBgDtM6slh+9nHomddV62YxCWETzvkLAP8E4J/lBiHEUvk3Ef0jgCHD+WcKId6L2kCGYapL37Z+XLfhReTyowCA/sEcrtvwIgCw8R8nBLp6hBDPANin2kdEBGAJgHsTbhfDMDVi9eOvFIy+JJcfxerHX6lRi5ikievj/xyAd4QQr2r2CwCbiGgrES2LeS+GYarAwGAu1Ham8YibuXsZzKP9BUKIASL6BIAniOhldwZRhtsxLAOAmTNnxmwWwzBRmdaRRb/CyE/ryNagNeHh9YlgIo/4iagFwGIA63THCCEG3N/vAngAwKmGY9cKIbqFEN1Tp1rJTTAMUwGWn30cspl0ybZsJo3lZx+nOaN+kOsT/YM5CBTXJ/q29de6aXVFHFfPfwTwshBij2onEU0iosPl3wAWAXgpxv0YhqkCPfO6cPPik9HVkQUB6OrI4ubFJzfEqJnXJ+ywCee8F8AZAKYQ0R4A1wsh7gRwKXxuHiKaBuAOIcR5AI4E8ICz/osWAL8SQvwm2eYzTH0xXtwMPfO6GrLdvD5hR6DhF0Jcptn+NcW2AQDnuX+/DmBOzPYxTMPAYZC1p9HXJ6oFZ+4yTEKwm6H2qNYnAODAoRH283uoSz1+hmlEGtnN4HdRnXn8VDz98t6Gc1nJNt7w0C7sP5gvbB/M5Xn25YFH/AyTEDp3Qr27GVSRMHdvfqthI2N65nWhrbV8TMuzryJs+BkmIRo1DFLlovLTaEazkWdf1YBdPQyTENKF0GhRPbbGsJGMZrUXeRstmosNP8MkSCOGQeqMpOq4RmH52ceVRFgBlZt9NWI0F7t6GMalb1s/Fqx6CseseAQLVj0Vyqcd59xao4uE8dIILisv1UxCa8RoLhJC1LoNZXR3d4stW7bUuhlME+EftQGOsbMxFqpzM2nCpNYWDOXyDTH1Hy9RPbXgmBWPQGVFCcAbq86vWjuIaKttzRN29TAMzKO2IIOnOjc/KjCYc8IJG2Hq34guqnqhEZPG2PAzDOJFgdgcE9SJ8Ii7+iS1IFvN9YSkYMPPMIg3arNdHNV1EKrFwbs3v1XY3wgzhkYjyQXZRozm4sVdhkG8GHybxVFA34mMxzj6eifpBdmeeV14bsVZeGPV+XhuxVl1bfQBHvEzDIB4ozb/uR1tGXz08QjyY8UlP1MnMh7j6OudZk/wYsPPMC5xFjj954bxHycVR1+NJKJGS1TS0YgLsknChp9hKkCYTkS1OOgnyO1UjSSiRkxU0tGIC7JJwoafYWqEd/Tcns1gYiaFwYP5SFE9QeGoSYzU44S81huNuCCbJGz4GaYG+EfPg7k8spk01iydG8n4mHzWNiN1m45hvPnFmzl3gQ0/w9SApEfPHW2ZEv15STaT0t6rd+MurH78FfQP5kBAIftU58IZ737x8bJ+YQOHczJMDUh69KxTXjmYH9MuHA/m8oV9/tNVoY2NKjttg6omQSPVIAgLG36GqQG6UXKKKNDYqAThhnLlo31JmihSG/2dUDWFz6pNIwqtxSHQ1UNEdwG4AMC7QoiT3G29AL4FYK972N8JIR5VnHsOgP8BIA3gDiHEqoTazTANjS6SZ1QIY6RM37Z+LP/1DuRHnTF6/2AOy3+9A+3ZTEEbyM+oEMhm0oFJYn5UndN49YuPt/WLIGxG/L8AcI5i+xohxFz3R2X00wB+AuBcACcAuIyITojTWIYZL8jRs2o0bhpp3vDQroLRl+RHBfKjY9CN67s6srjklK7CvdJEmNQ6vmSY46KbgQmg4WS2bQg0/EKIZwDsi3DtUwH8UQjxuhBiGMC/AvhChOswzLikZ14XxjTOed1IU7WACwAHhkfx5fkzy4x/NpPGmcdPxf1b+zHq3mtUCAyPjCGTLj1a/ud14TRynYEwmGQ3xqO/P05Uz3eJ6K8BbAHwfSHEft/+LgC7Pf/vAXBajPsxTN0QJQJEdU6SkTI39ZyM7qM7y1Q+731+d8HoS/JjAh3ZDCZNaNG+hqTCQBsBb1y/6v1o1HwFHVaFWIhoFoCHPT7+IwG8B2cm9A8AjhJCfN13zpcAnC2E+Kb7/18BOFUI8V8091gGYBkAzJw585Q//elPEV8Sw1QWm6ItKpnl+7f2l51zySldZdu9dGQz6L3oxMJ1596wSevLv82XA6Bqp58ug7FesOoppRHs6sjiuRVn4Qd9L+KezW+VRATZFq+pFEl0RPVSWCUsYQqxRIrqEUK8I4QYFUKMAbgdjlvHzx4AMzz/TwcwYLjmWiFEtxCie+rUqVGaxTQp1S6ZGBQBogoNvHvzW8pzHtn5don/3c9gLo/l9+0otKv3ohON7Qpqpx+TGyMoKcxv9IHaRsIkFZKpm22Nl3wFIKLhJ6KjPP9eDOAlxWG/B3AsER1DRK0ALgWwMcr9GEZHnC971HODDOL31++wjqDZfzCPdb8vd8V4yY+JgjE1jV797bKNSNEZa5MB/PsHXlSOigHnOc5a8QhmrXgEc2/YVDXfeFIhmeM5X0ESaPiJ6F4A/wvAcUS0h4i+AeBWInqRiHYCOBPAVe6x04joUQAQQowA+C6AxwH8AcB6IcSuCr0OpkmJ82WPeq7OILZnM7huw4tGI67CH6WjwmvEuyxHpGFGqKpO4szjp5YtFhOAttYUDgzbdWz+GUslSSokczznK0gCF3eFEJcpNt+pOXYAwHme/x8FUBbqyTBJUYmSiUHn6pQdiRA6Vt4WrxG3VZZUHeeVZtBdH3BmQ/dv7S87VgB49d0DodouZyyVNpxJLpSP13wFCWfuMg1NHH9s1Nht3YhwUBNqGZdMikqMuu2IVHXcl+fPtHJj2KwPhKF/MFfxcNBmcNEkhVVUT7Xp7u4WW7ZsqXUzmAbAJsImzLleVNcxRY3oomAkmTQBAiWVuYLwR/UkgU1h96vWbdf68ONQ6aif8RJeGoUwUT1s+JmGJ86XXZ5rMtgy5BFAWUeRSRMmtbZgKJdHezaDA8MjJT576VrxXuPKddsD72XTfpvXHXSMruOcmElpk8V0LPh0J3735v7ANQsZDhrl9SRxzniFDT/DhEQXuy2xNYaZFOGwiS2FgioqQ6SLxZcG0dagmzqhaR1ZzDoii39/bZ8xzl43S9GtBUiO/cQkvL73IEaFQJoIl502Azf1nIy+bf244aFdxuekioePMnOLM9tLmnrogMIYftbjZxgE173N5UetfN75MYH9B/NIE6F/MFcWhtm3rR8HhkfKzpN+/L5t/Vh+346CO6h/MIfl9+0ouQag9sHnR0WhQ+kfzFlloOoWsoOGgweHx/DazeeVbZeLoia3l2ptJUp9gnqpCNaIJSnZ8DMM7OrehkGGdEojcN+Wt7D59f3aUM/DJragZ14X5t6wqWwNID8m0LvRiYQOGk3bIPMNVj/+SmQ/flDkk2m/arE1SoRVvShq1ksHFAY2/AyDYK0WwFloPTQyFrpzyOVH8dxrZp1DGRGkk2MYzOVL5JjjMDGTit3JBUVNmWSiddcLG4pZqYpgYd029dIBhYHDOZmGopJqkT3zuvDcirNw29K5yrDA3otOLAmPnNyWQSYVrciJHxtjlYTRBxCp8/JCUI/aJX3b+vHBx3qjr0qQixKKqTvnzOOnxpLwCJvN3YgSDzziZxqGavlSvaN/1ahPFd5pWh8IwmvgJmtq5yZJiGhSJQLOs7lq3fayZyPfI9M9VCPhoGeuQnWOXwwv7GckitvGNqGunmDDzzQM9ehLlfft3bgrlGtD4g/fvP7CExNx6aSJMKGFcDA/ptwXVlbCCwGFjk4a1i1/2oenX95r1QF2tGWU26Nky/rPWbDqKeNnJMiNE8VtE6XTqjVs+JmGIeyXMmqIXZiZhY30sY6vzJ+Jm3pOLtlms9YQRCZNWP3FOQDK8w5spKCDUClyqpQ6tedbHBj1vQsS0At6X6OuGzSaxAP7+JmGoG9bP1Ia6WLVlzKOamcY8TZbaYNsJlVS+lBl9CU987riuQlE8ToqaYebek4ubPcSZ7kizPzBVBgeiPfemfztNu9rs8g+8IifqXukIVC5J3RfyjhuoTAzC5vIjTBJRUmsGXhF0XQjUbnNOwKO6/u3JWj0HOe9M/nbr9JkTHvfw0Z020SBDT9T9+hG1WkirUGNE2IXZrqvOzZNhDEhjIbDpkqXCVN2rc3r1D3XoKzdONiMnnWdnk1naDLcug7V/742mtsmCmz4mbpHZ8TGhNB+QcMYb9syiSqDpRthBo3wVf7mMH5yKe+gy5DtaMtgwaqnjKNWU9Zul/v84nYCk9syaGvV1/VVoVt81lUp8+M3/tKV04jRN5WCDT9T90RZcLP9kqsM8P1b+3HJKV1lipUml4nONaBbpFSNtm0NrPd1nHn81LIOI5MmfPTxSCEsVLc4rXuuXhE1WVEsShRQNpPG9ReGUxbt29avvZdtG3SLuDcvdtY2xrsbxwY2/EzdE2WkZuur1fmTn355r1JBElAbc53apC6KJGpWpzf8s29bP9b9bndZh5FJlYdx5vKj6N24q6Tds45QG/4zjy/WvO6Z14Utf9qHuze/FbmdtsjnZbqmDaY1gudWnNWUht4PG36m6oQN1Yu64Gbjq40SImob6qkzQL0bdwWKwqm4bencknv0btyl1PZXxe4DjuyDV8RN9xrvfX437tn8VonbKwx+2WVbtVHTzEJm5M67cVNhJqOrVRBnjaBZYMPPVJWo2beVWnAL60YKE3GiM6yDuTwumHNU6FF078ZdhXv8oO/FSAljXnSOE6/AXJh1B6B0Jta3rR9/t2FnSUfUP5jDleu248p128vqHJhcOZec0oV1v99dktgm6/kCpZ+dMGsE9SCnXAvY8DNVpd6yb3V1ab3uDi9hRpOmUf3TL+8NLc8wmMvjmBWPoK01bV3sPC42Rr8tk0IuP1ZiOP3y0ipkpz8xkzJGMnV1ZPH0y3uV2cyqer62awSNKKecFGz4mapSb0qG0oftHdkKAPdv7Uf30Z0ASl1MRPrM075t/SUGY/nZx2mrbQ0M5rBm6dzQWb8CqJrRt6W1JY3//Q/nlmxb/fgrViUmg+ocBMXgA+WfnS5Nh0twZkpy0T6lmBnUWgKkWgRm7hLRXUT0LhG95Nm2moheJqKdRPQAEXVozn2TiF4kou1ExCW1mLpUMnz65b1KGYLejbvKMkhNgSX+7NKeeV2YrNGlmdaRLcmsbWQGc/kyFcwkOnJvnkaQPLOX5WcfB1XgpwBw9+a3Cu+nbmZQi0FIJVVnVdhINvwCwDm+bU8AOEkIMRvA/wFwneH8M4UQc21LgjHjm3pMiTf54sOMxlWyDiccdXjZcd7XK6Wg68X4+w1mNpNGxsJK+GUV2rPqDs+WFAF/lm3BVeu2Y8Gqp3Dm8VOdYvUKDg6PlHW4cXIP/B2JjVGOY7jjSFREJdDVI4R4hohm+bZt8vy7GcAXk20WM16px5T4KBE2OrydyA/6XlQWYPnMzPYyaWfT/eOqaYbBu34go2ZslUe9HZ+qvGQYxgRK8hDu39qPpZ+dgUd2vl22LrL/YL7MN69z9wThH4To1gGkGunAYA4dbRl89PFISbnMSktBxyUJH//XAazT7BMANhGRAPBzIcRa3UWIaBmAZQAwc+bMBJrF1Cv1lhKvyxPQFVef3JbBB7kRpTH2jhbvfX638n7//vq+QlZtezZjNJIy8iXJspAmvOsHh0acaJwgUTUvA4M5J8w0oaIxEplbsW3lImW2st9QhnlmJnkNnVH2rgmpPiNhDHct1r1iGX4i+nsAIwDu0RyyQAgxQESfAPAEEb0shHhGdaDbKawFgO7u7irJRTGMfhYCqGWNr7/wRO0+72hRN0oXohgFZBpJy+vJ9kXNoI2KNF5hZkQdFSwkIw2hjaH0v6eqhVzAcW3945I5yuI6A67rRYXNu2BruCtVQtJEZMNPRF8FcAGAhUKoP41CiAH397tE9ACAUwEoDT/THMSJm7Y9N8o95CxEniurSwVJN5juE9dF49X7iZpBGxcZfaSLTvKSzaSNi99xn4c0hLaG0juzVNVNIABfnj+zzOgnNbuyNdy10BCKZPiJ6BwA1wL4D0KIg5pjJgFICSE+dP9eBODGyC1lGhKvEZZuDekGCOMLtY25jhObrdPt0QmuBbmsLjttRmRD3eVG/Xh5+uW9ka4VBxl9dNW67YGj3JsXn2wMu7zstBl4eMfbkRPP+gdz+PR1j2L+pyZj34Hhqkl4+LERrgtjuGux7hVo+InoXgBnAJhCRHsAXA8nimcCHPcNAGwWQlxBRNMA3CGEOA/AkQAecPe3APiVEOI3FXkVTF3iN6SqL7ytL9R2ASzOQpntubYzCllo5d7nd4ce6fYP5nDiyt/gv11c7HRqEWa4/Ozj0Let3zpKpj2b0Rp2VRbwpNY0cvnRsloACz7diTffz5WN7EeFwHOv7cOCT3fif7/9YcGtNKElOPQojoQH4Bh8nXprJk2Y1NqCoVw+kuGu9rqXTVTPZYrNd2qOHQBwnvv36wDmxGod09DYVqeyMWi2C2BxFspszjXNKIDyUdtNPSej++hOqxGznwPDo/i+R5LA1teepJ5+z7wuLFj1lNWxV67bDpNysqpNumS0N9/P4bkVZ+GYFY8oz/v31/ZhoicseDBXHtkTBZNi6fKzj8Pqx1/BPZvfQkdbBhNaUoGGvl4lIbj0IlMxwixuRT3Gvz1OgpjuGAEUYrNNwmu6WOzVj78S2RCPupIEgDP61sWy+9ubJGFmGkmtPQ8M5vCDvheNi6u25THDoMszOfP4qSXv7/6DeRwaGcOapXO1ip+1iM+3hQ0/UzFsjK2tL9Q28StOgpjqXIn80upG3KpkL2mI4rpo5Pk987owqbV6KisdbhJWLbKqZZZtWGyftS7hSlen+OmX94buaMLUbq42bPiZiqEypJkUYXJbpuRLZTP11X0h/efaHqdjoiFNNZcfta4CJZGL2nFIERUMU5iYelsyiirrKQC9Fzlhq8vPPk55TD1i00mpRuJXrtuOuTdsKugtPbfiLLyx6vzCaD6KC7HedKm8sEgbUzGksfVmfh42saWkKpMcedn4QG0XwMIslHmLm9v4xkeFKDvOlOxlk6B15vFT8cAL/Vp/96gQBf91klnG3vv7JY8pRejduKsQ1nrqMZOVWchRMYndRcV2ZqdbezKtE5hCSHV+/FrE59vCI36m4sgMUKCYXt+3rb/mPlDv/QF737hAUdNGziiuv/DEMv97Jk0ggjKLdXJbBm+6I8qbek7GrhvPwVfmz1SKiwFFF4Gtn9+GyW0ZPLfiLKXk8eiYwGAuX3hfkjL6X5k/E7ctnYuWkDOnINJEuOQUuw7fNOLWuWJsff/ez3A96lJJ2PAzFcXk56y1D9Q26kiFLEhesrDnt+9Cnc4PONv9HdxNPSfjjVXna41/wWAlMFJOEQoZyNVyPUxuy+CmnpOtJZvDMCoE7t/abzVoCBpxq55HFN9/XLdjJSFN0m1N6e7uFlu2sIpzTdi5HnjyRmBoD9A+HVi4Epi9JPLldOF40rjp9r2x6vzI97RF1zYvpmxTbztV+jE2TG7LlBUkP3Hlb5Run8ltGQzl8mUx71GZ3JbB+bOPipRnEJWo4mm2pIlw2WkzjNnWNtm5tjWDTZ/vanyGS+5JtNVWBZlH/EyRneuBh74HDO0GIJzfD33P2R4RU3hlrbX5g+6TzaTxj0vmaCWTvedHHTV7XV+Ao+ip8/UPHgxv9DNpKkTnqO599+a3qmb0CeHr3mYzKW1NAxWjQpRo7vcP5nDVuu34QV8x10KOxE3XtXU71vozHBU2/EyRx64F8r4vZj7nzAAiYvJz1toHqrq/33ffM68rsJ192/qRiuGz9rq37jGEMNqY5xShJGpq9RfnVCQSKAj/04iaVPZxfgzbVi7CbUvnGpPDTAg4z9Wv2S+vq+vYZX6GiVp/hqPCUT2Mw871QE6zgDe029kfweVjo0NSq8xGW40UKZAmXSLehUTpNog7ah4IucCsY0wAH+RGsGbp3JLXV0n3igq5BiKfa9T7y5GzfC1RBdQEoJTtkBFgOpfNYC5fVlLTfz5QX/UlbGDDzzgEjeof+p7zO6LxjxuiWSmC7t+3rR83PLSrZJFWLiR2H90Za4HYS5KuAW/4p5yxBBU+TxoZMSSJsgbiHznH7chM7jhT5xSk81Trz3AU2NXDOAztMe+P6fJpRORo3lRoI4mIGAJw5vFTATiiZUlQFh1V5fyrjz4uLYdoyopW4a236yVOqUpT52pyzfQP5upCZiFJ2PAzDu3Tg48J6hzGGUGjebmAGBcBFEIR/9vFJyOdUJZs/2AOC1Y9hSvXbU+8IlYQeY/GEFBcULXNfB4TwjiK1nUkuqsTzMa9Z16XcbG3XjR2koINP+OwcCWQCRhF2XQONSZO0Ws/SY3mbfCO0P9sYjIe2ChRNDpkLHqYPql/MFfyPvTM68KY5VpIkOtLFSN/29K52uMFglU7r7/wRO2spF40dpKCffyMg/TdP3mjG87pi8PIZJ3OoY6JU4RFRRLyCGHG2VIzxot8F6JExSQ5xpf++r5t/Vj+6x3WMwhvNitgX5rRJipG5VvX+f9tXEPyWrpqY/WgsZMUPOJnisxeAlz1ErD4diA7ubg92wlc+KNYiVzVIOlM4LB+6UogzWtHW0Ybj19psh7hup55XVj9xTmhxepkaORHH+t1iySZVHRN/bjhlT3zuqzyNhodNvxMKTKJyxvaORJypLNzPbDmJKC3w/kdIwFMhc6dE1sN0dfunvRzBXeCiYtSz+LZ1u/h9QmX49nW7+Gi1LOhXo8NUv/9tqVzC7HnhOQWg00cGhkri4G3ddl4GczlrSKL8q600w/6XsSnr3sUs1Y8gk9f92hJEpaOJGQSGjU2Pwws2cCUsuYk19Xjo32G4+oJknOQHYc3ESyTLc4YYkpCqNLts5k0bl58snGa7w0t9IdodmQzuH3eG/jsi9cr2903ukA7/b8o9SxWZe5AGw0Xth0UrViR/yY2jp1u/bpsmdyWQVtrSyFm/MChEesattlMOpHQ00ooa/r5yvyZSj3+r8yfWShpWUnqtXKWiTCSDWz4mVJ6O6D1DmeyeoMuCeo4TJ2CBbp4cKmtousUvDLQKh/1s63fw/TUe2XXHcAUfG74xxjVjFR15+0Zm4LTh3/kvMQ0Yelni/oxKYP+T6UwaQ5Vg2wmjUMj5bV1VUxuy+CD3IiyvWkivHbzeRVoYeMTxvDz4i5TSvt0teGmtF7OwWu0dSGfQ3ucY22uYcDkzrHNElYtTE6jcuMNAJ8U72uNfjaTRlfqfeW+aeRsV4mwzVrxiPKcSlJpo9/VkcX+A4dwMD+m3D+hJWU128ikCddfeKJ2hlXNzivKqL9RZgpWPn4iuouI3iWilzzbOonoCSJ61f09WXPuV91jXiWirybVcMZAHB+7KqwzkwWE5kvrN/S6kM/26eZOwZIgUSxV9SQv2o5DTFFuJwit3/7mxScjl/2k8rxUx3S8uep8XH/hiVj9+Csl6xFhF0YbgedWnIXFp+jDfQdz+cDQVgKw+otz0DOvy/iMqhFPH6VWRK3rS4TBdnH3FwDO8W1bAeBJIcSxAJ50/y+BiDoBXA/gNACnArhe10EwCRFXYXP2Esf10j4DADm/C/8r8Bv6YxehLHpdhoKaOgVL4i686TqOW0eW4KBoLdtOBExPvYdVmTuUxn/lgUvKzhtJTwQWrtQagkqMWmvZlRAco/f0y3uNxwW9am+s/WWnaT5vQFXi6aNEiNW6vkQYrAy/EOIZAH4Fry8A+KX79y8B9ChOPRvAE0KIfUKI/QCeQHkHwiSJyZ1iiwzr7B10fs9eop8JeGP7d64HdvwKZV/xfA7Y8C3gg37HZWS6RgBxozZ0Faw2jp2OFflvYs/YFOXCZRsN45qWYufZkc1g9eOv4NfDf1E4b0wQ9oxNwU10BTB7idYQVGLEL0XRokgZJHHvq9dvT1QIzrSAW414+vFWY9dPHB//kUKItwFACPE2EX1CcUwXAK/DeI+7jakUCbhTlJQkeGkiclSdjhfh+n8zk4D8wciFXuKIYsnz/MJrgGP8Nw6fjtcnXK4cQUu/fSZF6L3oRFzl+qHleRIaBnqh/8KPChEpwiYoiWtgMIc1S+cGKljK68h6u/dv7Q88vj2bMUYPJaH/Nqk1XVJ/ebIm2asa8fRR6uXWc41dP5WO41d9f5QfESJaRkRbiGjL3r3mKSNjIAF3ipZCgtda5/8Ny0rXEGw7l5GPS2cTKiqYCyC12N9cdT7eXHV+SVx8V0cWQ61HKs8bEEc4GvdfmlMopq1Cbtft78hmSvIDbMf/3lq/KmRNAP+M6CvzZ5b8v2bp3JJ6v0FFSQSA3otOxJurzq/ojGJ4ZKzELfbRxyNls7NqxdNHcSk2Uvx/HMP/DhEdBQDu73cVx+wB4HXWTQcwoLqYEGKtEKJbCNE9derUGM0aZ4Q1gDYumbjt0a0h2HYuuoVim3tUAP+C8OQLbyp7hjmr1uZDAAAgAElEQVRMwOqR0k4q6Iu+/OzjkFGI2xwYdrJXn1txFt5cdT7W+DoeE6aBtVeO2ft6buo5Gc+tOAtrXC2bq9ZtL9Mxams1T/6lnzquEdO5uQgoS+7KjwlMam2pSc3aKC7Feq6x68c6jp+IZgF4WAhxkvv/agDvCyFWEdEKAJ1CiGt853QC2ArgM+6mFwCcIoTQVPxw4Dh+l6BkKNN5fpcM4FTYkhm52U7g3FvCyzDo4vTlNYc/AkaH1fsllAauN3wETLkAV71Uvr0SuM9QDO3BgDgCt+SXFBKyCMCX3USioPC9eTduUror/EllXvy5ChelnsU1Lesxjd7DgJiCW0eW4JkJZ+LDj9Wx7jKnQbaroy2DQ/nRslDLbCaNS07pCnT1yNcsa8jOvWGTddKYF5nP4L+fye1Vqdq1lQ67rEVYZ+Jx/ER0L4AzAEwhoj1wInVWAVhPRN8A8BaAL7nHdgO4QgjxTSHEPiL6BwC/dy91Y5DRZzxEjXufvaR0/871QN+3gTHPlzW3D3jwO8XjbTG5c3L7gFTG6QBy+4HWNmD4QPlxp3wt2j102xMuEA+g8AxPVySMyVJ+3Ud3Bq43DGoEyUwLfsvPPg5XrdsOgfLM4OnkRBhddwjYKNSZwTJ6SBpTnShaLj+qzI5V4XVb9V50YuhKWN58BlnAxmsUdVnXlfCP24j5xTHcSYsFVgIrwy+EuEyza6Hi2C0Avun5/y4Ad0VqXbOT1ELtkzeWGn3J6HCo5CkA+gQvyVgeaJ0EXPuG8//DVwNbf+G4dyjtGP0LfhjtHllFJLB/VjS0G2MbvgXa8C2QzBaO0QnoDLSulJ8XWYv3fPpt2Yh965/9ZfnrcDuvnvbpeCp1ITaOnY5rWtaXyEEAToTRda33YcvEv9RG0iQhzSDRVcL6/vodgaGp/sxpeb7XwPZu3KWcQVTKP24Ku/SW04xquIOuXw9w5m49ozOAYRdqTR1F2E5EJbtguuYFPww29Kp7+GcogONG8tf+VcyKCgtXcm0AiGz8TdLMplG7NB7n02/LRuy3ZO7ASyfMAuC6ehSd16rMHUBen1F8JN5TSlQkRZoIY0IY6xAD5TVwMynCYRNbMHgwj2lu1NDqx1/BVeu2o6MtAyGAoVxx37rf7VYKt3VkM+i96MRYhlI3ag8Ku4xruBshrJMNfz2j07YJu1BrGqWH7UTKdPsTuKbqHt71CIlqhmKafQChJSH8eN0ufvxuCK+hkXo817SWj9izNIzPvvZjAP/J2aDovGTewICYgukK40/t00ONvMOg0jfyhllKAxokkeEfOXtdTv2DOaObadKElthGXzdqDwq7jGu4GyGsk2WZ6xldFm1YI7ZwpeN795NujRbt49Xtr1QEUW6/evvQ7mJ00871sAqGVM1qLKOleuZ14cvzZ5bdxe+G8GfpSkOsG7GXtEkz6+qi9/Dk2NzyjGLPM+6Z15WI0U8TKSNRgmQITBIZcQrRxx0dm0btQdFYQWG6QTRCWCeP+Osd/0Jt1GsAyUT1+BdS51wOvLop2YVVwDxLkS6cliys6kz5ZyAK14rJJXRTz8klC5JfPex3uCazDm0P/l/g/3Ne8+rHpyiNnG7EXtImzWslAr6Ufgb3jX4eC1PbMY3ex4A4AtMvvLnQzr5t/VbVucg9qD2bwYHhkRKhOpUfXhLH7RHHeMcdHccR89OpvIYp5mK6fj3Ahr9ZSKIDURnMHb+qTHWuoLWEfM68ziBRzUAiREsVXBs71wMP/RzIlXYa3Qf+Bv0oj7K5dWRJmV5/SZt2rldHPrm00TAWprYXJJ4JwJrRuQV9lNWPv2JXYlEUwyJ/0Pci7n1+N0aFQJoIl5yij0yK4/aIWroynaLYo+Mgd4spGisJwx0nu7wasOFn7AnSAbINqbQJv5T/P3BFcMKXD2kIidLOjMR/7TjRUo9dq3wG17Xehwc/Ljf8j4jPgfLAda334Ui8B/K+XlWehgIpFQGURxPZjqpTRDhmxSOFEb90D40Kgfu39hdCU8vuHcFfLdc6+gdzkWoF+/3PUUIro47a/fdas3RuXRvwqHAhlmpQiTjzWhCmSAsI6P66E9Hjff3ZycChD30RO655oLRj5L1hmKZ72hC2WIybIKY0NunnHLE5BQKEE0b/1VgEpoxbjilfwFbgLeoClCY16QrTRKFLYVRNFc9Ur0t1vDT+bZlUWRKZrmOQCW5h7+9vS5gOI8696gEuxFJPhPQnV7QdXuMLOAuoYTqi7GSNoSLFqFUAW9z0jR2/Ku5Xnu9+9eXI3vuMgvIGgvC4cKQh6P7gQqxqvRNZHCoe53G/eA3ARalncc3B9ZjW9z7GUqSNhqD26bj5jJPtDc3O9VZG/6Boxa0+qYgOj65OkiGdqnj1sG4P1ZqAFISThtx7LV2nJbfHWWMI625phPj7pGDDX2kSqDpljW5m4e98vAYnkY5INyIXxeStsMhnZJM34GmFUhVwaA8e9BjzfpwOMQxcm1mPafR+qfsFwPZH1uIJuhtdE96DAFCQ2zHNjheuRM/sEIbGIJM9IlJIQeBtHIHVo0uxcWxByf6PPh5B37Z+ZUhlezYDIhTi6MPOBlSGLowB1bme+gdzWLDqKSw/+7iSDkCH1PivZkx8I8TfJwUb/kpTKZlkP6aZRZBcsm1HpAuxNBHF6EuG9tjlDbjoAjvfwZSy0dzGsdOx8dDpzkj0Kk8i1WPX4vr8PlDKfM0Ssp3hO03N+y8EcHX+ihJdID/5MVFinE2GOYorKG40jmkUf92GF7HlT/sC9YHkWkY1Y+J10tP1FH+fFBzHX2kqKZPsxTSzsOlkbI6J0mZ/4ZUwyPvJvIFsZ+hLHBSt2DQyB+sOfguvT7i8rIxi/2AOx6x4BL03XY+RB/8LkNuHUHVSMlknLBbFRCdvmcXA1+bDP6fQzTFsjbMqpjyTIqMMcxxDp7qfl1x+FPc+v9vKNTUwmKtaTHzftv6CaqqXTAIRRvUIG/5KU2mZZIlpZmFjsFU6OH4C26wouXjK18pfvw0xnpEQKFTDum/08/hSy28xPfUeUpoyigLAN4fvRsvox3Y3oDT8CXWh660uXAnVeD5FKKn0pZMxlsY5qLNRSQWv/tIcbFu5CLctnVtmVAlFt0yUWrHe++mwTTib1pGtmtTx6sdfKcltkBw2MV4Gcb3Crp5KY1O5KglMuj42fvJDHzqCat5krGMXlSdnmVi8Vv06Z84vumpk5I4JSpeGicpnZelq6hfFKJjnJnyvdBEXRTkEb9UsbYatH40sduiFwdlLtBFCMnxTJ5ssR7y2YmI6V5B3fcAfehlHUVLeT+dmSrtyFia8o/pqxMTrZlA6ddVGh0f81UBVwzZpTDMLKf1gYizvROF4i59subO8GEpmkvr8zCT96/TW7DUZ/Wxn6TFDux3jeMsx1oVeRtITcUfrVwqjQ28MvBf/9gExJfDaoLQ2WS3SwqCmgP2AOAJpIty8+ORChSzViDeJ4t5ScqGrI1vmVvJeK5Qby0XnprnstBnGtZNaFDCJK9PQaPCIf7xgmlnIaJ9AAqbg+ZxjnPM5AKXx2BjLlytneglaYAbcEb2iDbl9Tqcz5/LS0FDA0SCacHghNLVl4Ur0zl6CXrl/jXom9C5NKSlwsl8chmHRglYq9/MWEGPa16dbhOww+NKxcCUO3v+dkqxeGb45JkTg4m2SUSima0WVKTaFguoE2gjQFqipJHFlGhoNNvz1TpjkL5Usw8NXu/H0CSXq5far4/mDtP2topgMbcznHLfThT8K5zbTKJwemLYQt7x5B7Ku0T2CPsKwSONQpgMT8oPqaxlmHMvPPg7/dv9P8P3UuhLd/cc+/lwh9LKM2Utw68Zd+Obw3QUdnltHnEpfNrVtk4x4MV0r6Vh6k75QrUbYjaCvkyRs+KtNGEMeN/lr5/oQRt8yuZ5S+sQjU4UsSsUL7ZTXD6s5pJkJffrJGwGfXHIrjQJthwMLV4eWw+5JP4ezW25HFqWVspAHVj8+UWtA5p6/DH+54c+RGw4/0kxylHrm8VOVo/Azj5+KezSj86hhnzp9IYK+pm81ShnWu75OkrCPv5qELSIepI0TxJM3wmjMvXLP3V+HVdS6yXhTqvy1yNcc1+gDZh+/SWa5ICO91vl/wzKD8ueecjnsbKejBLphmV7C+ckbC0ZfUtDUNxjInvRz2HrYlXh94pfxbOv38LXDfmft3zZFvIT1yT/98l7t9qT936aqZiYZCOuIKSYQHvFXk7BZvDbJX6YZhMm9oitcvuVO9fE20Thi1DGOG77lXP/YRdEzd1UMH1CvI6hmRn3fdmWoXdfUyCEgr1fBLODNHVBlPetmXZpnPY3e1xtI99pt7rWnp95DL/0cSJ8IwG5WU6Ia+uS1wIN7cHDTJ/HsgUvQP/wXAOx88iYf/5qlcxP1f+vcSjr3VjNJKVQLHvFXk7BZvEHJX0EzCO0ImdRui5nzywu2pDJOwRUxVn68EneGIaOC4hj9Vl8EkVzkffjq0tH9Q1eWd6hjedclJZzfNkY/rISzF82zfhtHlBtIOTvZ8K14Mzrv9Tyfg7bc27iR1pbkKgRF+5hG9UnH0odNymomKYVqwYa/moTN4tWFaB67yM5wqM6XqpmqGYaqKPtY3hk5J51pHES2Exg+WL49nysPO7Ux6kYM1c1sO2vFs85hAgZOuabUQJYYaQ1h5TwMpRu9mAxlkDE2VdoKS9iOpNlCLatBZFcPER0HYJ1n06cArBRC3OY55gwADwJ4w920QQgRcjgzjghbQ1e1MHnsovKQRj9Du53RcNgKWTqDk9sHnHhx8H0TxyD+lhQql5fXfaZblPZ3hIr3KrtwJT7rf9Y2Ya1hO1mDm6nkf4OhlEa3d+Ougl7NxEyy48KoC7TNFmpZDSIbfiHEKwDmAgARpQH0A3hAcehvhRAXRL3PuCJKFq8/imXNSZbGV4SvkGWSQC4JpYwhk2xDtjOaIFxYUhln3UB2krID9nbOOlfVsYvKt9lEHAWN5qNIVWjetwFxROFvW0N5aKTo0tt/MB85e9dP1FwA7/5mCbWsBokUYiGiRQCuF0Is8G0/A8DfhjX8464QS5JEKUyiW8j1s3O9VkYAICcjV2JZRCQ0UhKh0h1MqtUJYhr1lURsydq9Lttn6kdXBEZeM4qch6KS10h6Im6iK/DLj061NpQ6iQWppR+HSl47CtUID602tSjEcimAezX7/pyIdgAYgNMJ7FIdRETLACwDgJkzZybUrHFIlMIktj7j2UtKC7L77+sl6oi8fUaw4QM0dWgJOObzwJ7fxXc5iVFgzDeat63jC0SX1da5++LULVbMJMsymA14SyWqSGIRtRYLtDrjHmf2MV6I7cQjolYAFwG4T7H7BQBHCyHmAPgxgD7ddYQQa4UQ3UKI7qlTp8Zt1vhFt+C7+Hat9ouVz1hGmuT2Qamy6Xc/RFnslSGeKrq/URxBP/S98s4n2+nE4X91o2MkvRLNmUnlEUBBxA0xtVEzVeHPEdAtKke5rtRJWrjS6QRUOQ0+vDHyOpJYRK32Aq0p9j8JjaNGJ4kR/7kAXhBCvOPfIYT4wPP3o0T0UyKaIoSwlEJkSpCLjvmcuj4tEDrjtHDdkvNkLSuhdz8sXAk8+J1SV4kRcoz+LtUyEIAX/iU45v/JG508gexk4NBQcXvsqB4P2U5gxGLkP3LIvN9E2OzjMITM9lYZQS9JLaJWe4HWZNw5PDSZcM7LoHHzENEniRxBcSI61b2fWi6RMeMPAxSjpeqbQPTRpDLSRBT92KrzZy8pj/k3IpyFZp3/fGzYbPRz+4ohnLl95W6aJJBFVS78UXABmSQ7myQJme1tMnZJqmRWS1dfYjLuHB4ac8RPRG0A/hLAf/JsuwIAhBA/A/BFAP+ZiEYA5ABcKpJYTY5KGJ2cesM26zfKaFIbq74b6G13RsHn3lJ63YevDm/8qhoKGoHppwIPXOF0QJSCM06xTVyrE0ImCZqyaJNedK2mFo5JdI7DQ2MafiHEQQBH+Lb9zPP3PwH4pzj3SIy4gme1Jm7tXm+nJ/3TrpQxMm1mI57b57h1gKKMgU7aoZF549+Kf1tnKvuIOLhILMrEVJBHQTWNYDUjaUyvi8NDm0mrJ6xOTr2h+0JLYTTTa/B3el53i22EkFd2OaykwHiEFF7SiIMLf5TJKR88gc/2fQviwfdBYWemIZMEq2UEqx1JE/S6mkmJU0XzGP64I+ZasnO9JrwRjltCZVxssk/DIp9VXT6zFJDtqExugQrVjCDi4MK7EHlR6lmsytxRLM4SdmYaIUmwGkawFkJrzW7cTTSPVk9YnZx6QY4iTQbNv3i3c72jTikXQ5NSx4RwEreihjKmDAummUnBC6p+vIvYi38OXPsG0DtUGupZhoX0tPW9XWQorFbqebcxrNK7EHlNy/qSilwAwgu3VaPUZ0g4kqa+aB7Db6pJWw1MevGm41VCbCq8o/DHri0XW0uK3D4gN2g24jpMkTj5g8D1+xzD3TsEawO9eG25cTv3FoU4HYq5ALp8B1u8nxsb0TXAWHvBG02iLfpel7MseziSpr5oHldPFJ2csOxcX5r5KqNhALXv963NagE1RQp+IN6ZSxLujlSrE2KpZAwYS2jkLPHPvGwylI1uEM+YhlLAKX8DXPDD4rYNyxBJ7M2b17BzfTEKyAaN28e7EDkgpmC6yvjX+8xUhcfd+ET2k1jZegl+7dYIAJovkqaeSESrJ2kaUqtn53p1QlOhGLjKGPvKHUbVqfGm/Bv1dlykKyS336A+OcMdZVbh80FpYGJ7McpIJZYWdP7FPyu+ft370PPTotHtbddeTgiAVP2azGvwd/Ch8GkeuciIl+4PnsCq1juRhSdBLK6kQy1ISD+IsSeMVg8b/qQw+XjDEMXgHvMfHCkDm5mC34hoRd8omi5QEsg2AqUzNFNbbDpNr7DaDZ3GkfqYAFJe4+9tk1WHpKlhbCPu1sj5JhLd9yGquB0TSC1E2pikfLA2Rs7PG//mJFS9uinYIPndDdnJhuLpMYx++wyg81OlsfG25HOOC+Xin5Uaif8+TR/dlM8Fu2+8ryfAPUMA9oxNwbTUe0hRuvjchg/Yy2L7SWVK1wZ0xt2fhFdYPA7ZEdSyA2nkKLomoHkWdyuN0Qer8hvofOTCMS5+OQTVYqWXQlUqC+SXb+d6YPgjfTuisvh2R5fnjWeiX0OGqcrF0H86TW/0iyeZd3tj7y0WeG8dWYKPRWuxkxjaHW/9ZMLhjuF9+GpPwXe3bsKGZY77yb/wH1ReU0fU85KiUaPomoTxZ/jDRs8kxcKVQLpVs9NnkLKdTvlDnTHP7XOczNlOlGjuGI2VsA+HlF++x64NIbIWgr5vu5m9Md2I+ZyzXnHjVOC9l+O3S4wVPxPKspSl9Gb+uTy0Mg65/W7W810ofzaeWsVeAx1Se6dA1POSotZRdIyR8WX4aznKmb0E+MJPfDHkmlF96yQnwmTO5fpjRoed47yx2AtX6o8H3JFpQLSN/PLtXF+5ZKekQ0m10UUR8EYC+eWdPRABk6GbDUWEUu7Ce0CH6DXQUV0mtXa1xJCf7tvWjwWrnsIxKx7BglVPoW9bf+Xb22SMLx9/rWUZ/L7Z3g71cfLL9+omBPqkvXIMs5c4IaBGnRyfpPKxi9yQ0d3OjED6z5NI6qJUdE2bWpLPObMdwNj5KSN74hDmmcvPSEjtnZL9Uc5LkgiCgVwkpTqMrxF/rUc5foL8nDbt8s9YvLHoWjySyhf8sDjtloYnqUzeRjT6kty+6OGYobYH7dMgPyNRXSYN6mrhIinVYXwZ/npbUAr68tm0S85YvGsXNr58KRPQ2+GM8MMkg6UySEzaYNzgrrm0tpXvymSdNRv/gnwq41QWC7vW4f2MRHWZVKrSV4VhaYfqML5cPSGVCStOULawqr0qpF9aHmc1YqfSoi226LKNmxlKARf/XP08vLUKZs4vf6+tF1Nd95w3dBQoukuiGOxKVvqqECYdfSY5xl8CV6Mlv9hkgcoyi5VGJoJ52xaUBdxoZDuBjwfDuakyk4C2zmgJSdoEORWaTO56/vwmjN/HDzjSDpWs1jVeCJPANb5cPUBdKhMaCWqf1zdfad54RhEBNV5cPuS4XUZy4dcm8gejrx+FcjP6Oohqhl/WCdUu0disjC9Xjw31NiMICqs0yRCoCoNnskBLVn3NwJmDKI2AevJGVEWrpxoc8/ngYu46CovxEaJklO48jZyDiibMdGUd/coz/kb8JmqdzajCNKJrn1GM31ctEsvC4P4FPJUscSYLnPK14Axgr6GphU5Ppdjzu2Cjn1Ik4GWyTkisKmvYmxOhShqUg4x8rrgg3z5Dk7ynmVlxpitTAZrL8Nc6m1GFaUQXNbJDd7yNlo/X0IQtjFLP2CxSj+UduQnvc5tzObDjV+UzqGxnqWibfzDx8NWlOv1itNhRXPDD8vdH1Rk0QPgl05jEdvUQ0ZsAPgQwCmDEv7hARATgfwA4D8BBAF8TQrwQ976RqLc4f0CfaJPtLDXsqgiNoBqv/uM3LAtuz7GLin/HXVtQuaLqGanl412sXXOSuv2tk5znq9qfz6ndSt5kQtX7o4oKqrYbst5coUxFSMrHf6YQQlM6COcCONb9OQ3A/3R/V59aZjOqvlCA3oUgQypNhM1UtlH9fHWT5/gZ0d093tcQWbu+yqjqFwcNFnT7dZ2maZBRrfBLnXGPWCyeaTyq4er5AoB/Fg6bAXQQ0VFVuG85tcpmVK0t9H3bKRiicyHYfNHCzmAshMlKzrU5XoXXFTV7iVMLt1Hwu/6CkgK1g4aIPvtKiwya1rnq0RXKVIQkDL8AsImIthKRypfQBcA7bNzjbqs+tcpmVH2hxvJqZUzpQrAhyCj5jQhgp/K55iTHR+1fmMx2lmenekm3Oj5yVRit6Z7G4ugJMeV4+2ODOj/vYEG1P5UBUqqvVsqZ4emMejWCD0zGvR5doUxFSMLwLxBCfAaOS+c7RPR5337V0Kcslo2IlhHRFiLasnfv3gSapaEWcf5hvjhhjtWNyA/uc7Td/Zrvctp+1UuOgdaN5od2O0Jw3oXJVMZx3fT8VL/o23qYuoBIb4dj8Pyy1Zms045Kzwi6vwEceNf+eG+HGjRYUO2fcLimsPyYO8PTGPVqjLhNxr3eJE+YihHb8AshBtzf7wJ4AMCpvkP2APAO96YDGFBcZ60QolsI0T116tS4zaovwnxxwhw7e4la2jkv1w0MCUElBsuCsbzjq5+9RJ8Aldtf/Ns/es3tc4rZtk4qHtOSLR4bhUy29Hqq/YtvdxZNrdcYqNz1FzRY8O/3PgcTfqNejRG37vNFKWdhnyOLmoJYhp+IJhHR4fJvAIsA+PPXNwL4a3KYD2BICPF2nPs2HDp3gGoEHPZLFiTt7MdrRKTBss3OlcbTxsWkEoYby5cuZuf2OZ2DlEgOA6WB6acCIxqtfu9aSZjrd3+9uNAZ1dcepvP2vh/VGHHrZoli1AlbnXN5wwm7MeGJO+I/EsCzRLQDwO8APCKE+A0RXUFEV7jHPArgdQB/BHA7gG/HvGfjoXIH9PzUKdwS90sWdjSoMiJhDIu2ehU5I0Y50rcNBc3nokX8iFFHYkJX9EUKpwH210+3OrODuL523fNR4X321Qg+kJ9Flbsun3MGEo0kecJEYvyJtDULhZC8EOGWUvQLKA3nO3aRM9qzibeXomQPX11eQtAkF1FtpDxFWIE76fqyEWQzxbz796mesUqEzSaOPolYe614HDlGn2k4woi0NZ9Wz3jAH29tg1du+cHvFCOKhnYD2/4FmPdXxUpdJkzVw/IRk7WynU7R9yTr/0YtOmOaQXn32STPRUnQCorlTyrWvh4qdDE1o7kkG8YLquiPIEbc41UF1keHgV0PmOWFJWGqh9kgE7389YqznU40TjVCPb20T7fztUeJwEkioiypyJ8GrdDFJAOP+BuRKEZXGgedG8bGPeOvHhYlqzfd6oR95vaXj3q9bpLHrg2oLVwBvK8vqKBPUAROpaQPkor8CSoSxIxr2PA3IlGNrs05qVZgTOVyoVJ/tG31MC/ealU6orixkkDVNmkUs5Od/zcsc7YtXGl2lcR1x5g6jbAuGtO1GrBCF5MM7OppRKJKKZjIdjpGQmn04Rg/v2icbR4ApYuJWkGGJoobKwn8GdPSLbN4reMm8ydemWLeg9wxplDRoIiiMC6aepQhZ+oCNvwqKq2XEhXZrg3LnOiZbCcKRcBjSSinnNGuKd5d5QqSxjHIDy/G7EeWtZIH0N1XZ8Rf3aTP6NXNrIZ2q43xhmVOlJTpfsrEu4AwYNbeYTSwq8dPvSoU+tuV2+dmpq4tJhz5wwdtfeTZDuCtzQF+fnLuoXoG595irs0bNnO5kgVgKKXOPNa1UetTd9uoWhDXhZBSWjOjEU5o7Mz5dj58WxcNa+8wGnjE76deR0k2I0FvxMgFP7SXY8jtc2PyTQj9M5i9xDDqV0ggmKiEG8vLxA69q0Q10zN1Wjq3iS6EVIwajK77fJPM3mXtHUYDG34/9TpKitIuWyNKaVjJPpjupSr3CCpKINgSVkOofUY4N1duv9pVAqj94ccuKpfWkOgGBLq2t88wG92hPcmGWXLIJqOBXT1+6jWxJUq7pMF94ApDIhPZJznZ3MsUHmgb4ihdGdrsUk/bF6503FS2bq326WpXia6S1qub3PBTjRtM1RmqIp68BnfDMihfl2wboC7as+akcKGXHLLJaGDD7yfoS1srorZr9pKAkosCjo5MwIjf9l46oxJl7STQ3y9K7ynLHVIamHW6U2Dd9nlFnempOsMgg/vWZrXchbfGsr/DjLruFCVkk8svjntYq0dFvX7wo7ZrzUnRFkxbJwHDB+M9AxtNIbkY2j6jXO/GKy/hx6+do723xfPSPSOTdg+ouLhui/d56F53mLbZZFuHQZVHodIUYuoO1uqJS70mtvzx93QAAAv6SURBVMh2SePhTSgytdc62cod+QcZIh02wmQqpKtJVyz+4SvLaxMnPQsLmlGVPb8I6xd+oypGi/cwXaea605h6zgzDQkv7jYaUZJyrBdMRXEUGcXoP/id0nZtuTN8MpZ/wXT2EuDvBpwEMNWCbJhEKG+8vB9TfLxq3+K1TuRUGKJGjFUzOqdegxuYRGFXT6MRd9ofKIkQUZb3lmMSlGO2aEOQS8Lk3ur+RnijnQRRpZCr6X6ppluJSZQwrh4e8TcacUdkpkIcQPRRZJIa/N426LKoTaPnnevNawpb7qpNNnbUkXuYbN24cAhoU8A+/kYjiXBTaTCiRi+pFk2TIpVx/Pm9HY4+0KEPi1W2vGsApoxaeYwWURufdZyIsWqtO3EIaFPArp5GI8lpvynCBChfqC0UavGFf2ayAFKeIu8eMpPU21XYFmQxRdqEqbjVO2R3XJLUa8QY0/CEcfWw4W9EkjQeqo4k3QoIoa9nqyLbWTo6B5zR+2f+uhhfr4PSjn4OpeyN9uLb1R2g7WIypYHrDe4pWRNAurBsJKUZpoaw4a83ajnKC7p31Bj/MtzwxoKRJ6cDGT2UwLX9t3KNtuq1halDrBvxP3y1OhOY0sDEdnURmTDwqJ+pABzHX0/UUu3T5t5JhellJzsx+4URuwg2+mELoUvEaFEpVPUMdZIIXnShrTvX6wXrxGhxBuB/lraJWfWq/so0FZGjeohoBhE9TUR/IKJdRPRfFcecQURDRLTd/Wm+0IBaqn3a3DuJWHAZBWIds0/OaFslj2yLP3fBW6vAVn5CFTH05I3B50u8UUSFnAGUJ6R521mv6q9MUxEnnHMEwPeFEP8vgPkAvkNEJyiO+60QYq7703yf7lomxNjcWxW+l251/PNGyPklQwtz++3bJTsbXadDaRQKzOja4a9o5U3WMjY75Zy74VvOjz8RLqzba2iPuWqY36hzghRTB0Q2/EKIt4UQL7h/fwjgDwC6kmrYuKGWmug291bFiH/hJ0DPT0u3dX+jPHO1d6iY5Wv7erzhi7qY8Yt/5iQ0XfuG0w4dQ7ud0foDVwTPNtKtxUVkHflc+Epm7dODjbZ3v6zfq7oOw1SJRHz8RDQLwDwAzyt2/zkR7QAwAOBvhRC7NNdYBmAZAMycOTOJZtUHtVT7tL23zlcexue8cKVZUA0oj4yxiRmfvSRgwVYErBOQc93hA3ZJZlI/p6wjSQNQ3KckzFWDNOo71zuRT37SrZwgxVSV2Jm7RHQYgPsBXCmE+MC3+wUARwsh5gD4MYA+3XWEEGuFEN1CiO6pU6fGbVb9UM2sy1ree/YSR7fehKrYur9ymKptUatytc8oXtfWFSWfUcns5nagfZr6+Fc3mdvn7WifvFEdItt6GC/sMlUl1oifiDJwjP49QogN/v3ejkAI8SgR/ZSIpggh3otz34ajlmqf1by3ybjaVtRS4Z8Z2Cy++mc2NrV8vUqZ/memq2kwtMfXPkNUj84lFGZ9hGESILLhJyICcCeAPwghlIpXRPRJAO8IIQQRnQpnhvF+1HsydY7WuIasu6vCa4x1uQfSh69yGQW5ooKkqIOkMmw62Hqt7sY0HXFG/AsA/BWAF4lou7vt7wDMBAAhxM8AfBHAfyaiEQA5AJeKeswYY5JBqfsfQbc+yn2CZCvk9qjZuEms1dRrdTem6eDM3Uaj3rM+q9W+WjyHJO5Z7+8f07CwZMN4Raelzzoy4WDjy4xDWLJhvKJLFMrta860/ygGnCUTGIYLsTQUpkShZkv7j1KCEmDJBIYBG/7GIij6o5nS/qMacJZMYBg2/A1FUCJTM4UFRjXgtZTQYJg6gQ1/IyEzcbOd5fuaLSwwqgHnmrIMw4a/4Zi9xJE+WHx7bWQg6oWoBryWEhoMUydwOCfTuJiiejhkk2kyOJyTaQ50MgkcsskwRtjVw4w/OGSTYYyw4W9WVGUHxwscsskwRtjwNyNRk58aBQ7ZZBgjbPibkfHuCuGQTYYxwoa/GRnvrhAO2WQYIxzV04w0Q0GQWlY9Y5g6h0f8zQi7QhimqWHD34ywK4Rhmhp29TQr7AphmKaFR/wMwzBNBht+hmGYJiOW4Seic4joFSL6IxGtUOyfQETr3P3PE9GsOPdjGIZh4hPZ8BNRGsBPAJwL4AQAlxHRCb7DvgFgvxDi/wGwBsAtUe/HMAzDJEOcEf+pAP4ohHhdCDEM4F8BfMF3zBcA/NL9+9cAFhIRxbgnwzAME5M4hr8LgDcLaI+7TXmMEGIEwBCAI2Lck2EYholJHMOvGrn7q7rYHOMcSLSMiLYQ0Za9e/fGaBbDMAxjIk4c/x4AMzz/TwcwoDlmDxG1AGgHsE91MSHEWgBrAYCI9hLRn2K0zZYpAN6rwn2ShttdXbjd1aVR2w3Utu1H2x4Yx/D/HsCxRHQMgH4AlwK43HfMRgBfBfC/AHwRwFPCotajEGJqjHZZQ0RbbEuV1RPc7urC7a4ujdpuoHHaHtnwCyFGiOi7AB4HkAZwlxBiFxHdCGCLEGIjgDsB/AsR/RHOSP/SJBrNMAzDRCeWZIMQ4lEAj/q2rfT8/TGAL8W5B8MwDJMszZ65u7bWDYgIt7u6cLurS6O2G2iQtpOFy51hGIYZRzT7iJ9hGKbpGPeGn4hmENHTRPQHItpFRP9VccwZRDRERNvdn7qoSEJEbxLRi26btij2ExH9yNVC2klEn6lFO31tOs7zHLcT0QdEdKXvmLp43kR0FxG9S0QvebZ1EtETRPSq+3uy5tyvuse8SkRfrV6rte1eTUQvu5+DB4ioQ3Ou8TNVSTTt7iWifs9n4TzNuUZdsEqiafc6T5vfJKLtmnNr9ryNCCHG9Q+AowB8xv37cAD/B8AJvmPOAPBwrduqaPubAKYY9p8H4DE4iXLzATxf6zb72pcG8H8BHF2PzxvA5wF8BsBLnm23Aljh/r0CwC2K8zoBvO7+nuz+PbnG7V4EoMX9+xZVu20+UzVody+Av7X4HL0G4FMAWgHs8H+Hq91u3/5/BLCy3p636Wfcj/iFEG8LIV5w//4QwB9QLi3RqHwBwD8Lh80AOojoqFo3ysNCAK8JIaqRjBcaIcQzKE8o9OpL/RJAj+LUswE8IYTYJ4TYD+AJAOdUrKE+VO0WQmwSjiwKAGyGk1BZV2ietw02umAVw9RuV3tsCYB7q9WeJBj3ht+LKws9D8Dzit1/TkQ7iOgxIjqxqg3TIwBsIqKtRLRMsd9GL6mWXAr9F6IenzcAHCmEeBtwBg0APqE4pt6f+9fhzARVBH2masF3XRfVXRrXWj0/788BeEcI8apmfz0+7+Yx/ER0GID7AVwphPjAt/sFOO6IOQB+DKCv2u3TsEAI8Rk40tffIaLP+/ZbayFVGyJqBXARgPsUu+v1edtSz8/97wGMALhHc0jQZ6ra/E8AnwYwF8DbcNwmfur2eQO4DObRfr09bwBNYviJKAPH6N8jhNjg3y+E+EAI8ZH796MAMkQ0pcrNLEMIMeD+fhfAA3CmvF5s9JJqxbkAXhBCvOPfUa/P2+Ud6S5zf7+rOKYun7u7yHwBgC8L18Hsx+IzVVWEEO8IIUaFEGMAbte0p16fdwuAxQDW6Y6pt+ctGfeG3/XB3QngD0KIH2qO+aR7HIjoVDjP5f3qtVLZpklEdLj8G87i3Uu+wzYC+Gs3umc+gCHppqgDtCOhenzeHqS+FNzfDyqOeRzAIiKa7LomFrnbagYRnQPgWgAXCSEOao6x+UxVFd+a1MVQt6egC+bOJC+F8z7Vmv8I4GUhxB7Vznp83gVqvbpc6R8Ap8OZFu4EsN39OQ/AFQCucI/5LoBdcKIFNgP4izpo96fc9uxw2/b37nZvuwlOFbTXALwIoLvW7Xbb1QbHkLd7ttXd84bTMb0NIA9nVPkNOPUingTwqvu70z22G8AdnnO/DuCP7s/f1EG7/wjHDy4/4z9zj50G4FHTZ6rG7f4X97O7E44xP8rfbvf/8+BE5L1WD+12t/9CfqY9x9bN8zb9cOYuwzBMkzHuXT0MwzBMKWz4GYZhmgw2/AzDME0GG36GYZgmgw0/wzBMk8GGn2EYpslgw88wDNNksOFnGIZpMv5/LgC8Abj76DUAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "n_pts=500\n",
    "np.random.seed(0)\n",
    "Xa=np.array([np.random.normal(13,2,n_pts),\n",
    "            np.random.normal(12,2,n_pts)]).T\n",
    "#print(Xa)\n",
    "Xb=np.array([np.random.normal(8,2,n_pts),\n",
    "            np.random.normal(6,2,n_pts)]).T\n",
    "#print(Xb)\n",
    "X=np.vstack((Xa,Xb))\n",
    "y=np.matrix(np.append(np.zeros(n_pts),np.ones(n_pts))).T\n",
    "print(X)\n",
    "plt.scatter(X[:n_pts,0],X[:n_pts,1])\n",
    "plt.scatter(X[n_pts:,0],X[n_pts:,1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/500\n",
      "1000/1000 [==============================] - 0s 159us/step - loss: 1.4820 - acc: 0.5360\n",
      "Epoch 2/500\n",
      "1000/1000 [==============================] - 0s 20us/step - loss: 0.6442 - acc: 0.5990\n",
      "Epoch 3/500\n",
      "1000/1000 [==============================] - 0s 20us/step - loss: 0.4453 - acc: 0.8460\n",
      "Epoch 4/500\n",
      "1000/1000 [==============================] - 0s 20us/step - loss: 0.3872 - acc: 0.8960\n",
      "Epoch 5/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.3437 - acc: 0.9060\n",
      "Epoch 6/500\n",
      "1000/1000 [==============================] - 0s 19us/step - loss: 0.3108 - acc: 0.9250\n",
      "Epoch 7/500\n",
      "1000/1000 [==============================] - 0s 20us/step - loss: 0.2915 - acc: 0.9310\n",
      "Epoch 8/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.2637 - acc: 0.9390\n",
      "Epoch 9/500\n",
      "1000/1000 [==============================] - 0s 20us/step - loss: 0.2371 - acc: 0.9580\n",
      "Epoch 10/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.2213 - acc: 0.9630\n",
      "Epoch 11/500\n",
      "1000/1000 [==============================] - 0s 19us/step - loss: 0.2060 - acc: 0.9580\n",
      "Epoch 12/500\n",
      "1000/1000 [==============================] - 0s 20us/step - loss: 0.1946 - acc: 0.9600\n",
      "Epoch 13/500\n",
      "1000/1000 [==============================] - 0s 19us/step - loss: 0.1825 - acc: 0.9670\n",
      "Epoch 14/500\n",
      "1000/1000 [==============================] - 0s 20us/step - loss: 0.1731 - acc: 0.9670\n",
      "Epoch 15/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.1674 - acc: 0.9650\n",
      "Epoch 16/500\n",
      "1000/1000 [==============================] - 0s 20us/step - loss: 0.1592 - acc: 0.9680\n",
      "Epoch 17/500\n",
      "1000/1000 [==============================] - 0s 21us/step - loss: 0.1511 - acc: 0.9700\n",
      "Epoch 18/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.1460 - acc: 0.9670\n",
      "Epoch 19/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.1401 - acc: 0.9760\n",
      "Epoch 20/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.1346 - acc: 0.9740\n",
      "Epoch 21/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.1323 - acc: 0.9740\n",
      "Epoch 22/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.1274 - acc: 0.9740\n",
      "Epoch 23/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.1286 - acc: 0.9720\n",
      "Epoch 24/500\n",
      "1000/1000 [==============================] - 0s 19us/step - loss: 0.1252 - acc: 0.9740\n",
      "Epoch 25/500\n",
      "1000/1000 [==============================] - 0s 19us/step - loss: 0.1173 - acc: 0.9720\n",
      "Epoch 26/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.1127 - acc: 0.9790\n",
      "Epoch 27/500\n",
      "1000/1000 [==============================] - 0s 19us/step - loss: 0.1110 - acc: 0.9750\n",
      "Epoch 28/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.1101 - acc: 0.9780\n",
      "Epoch 29/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.1085 - acc: 0.9750\n",
      "Epoch 30/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.1037 - acc: 0.9780\n",
      "Epoch 31/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.1028 - acc: 0.9770\n",
      "Epoch 32/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.1023 - acc: 0.9770\n",
      "Epoch 33/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.1005 - acc: 0.9750\n",
      "Epoch 34/500\n",
      "1000/1000 [==============================] - 0s 19us/step - loss: 0.0971 - acc: 0.9780\n",
      "Epoch 35/500\n",
      "1000/1000 [==============================] - 0s 19us/step - loss: 0.0951 - acc: 0.9770\n",
      "Epoch 36/500\n",
      "1000/1000 [==============================] - 0s 19us/step - loss: 0.0949 - acc: 0.9780\n",
      "Epoch 37/500\n",
      "1000/1000 [==============================] - 0s 19us/step - loss: 0.0934 - acc: 0.9770\n",
      "Epoch 38/500\n",
      "1000/1000 [==============================] - 0s 20us/step - loss: 0.0924 - acc: 0.9780\n",
      "Epoch 39/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0946 - acc: 0.9760\n",
      "Epoch 40/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0929 - acc: 0.9750\n",
      "Epoch 41/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.1025 - acc: 0.9670\n",
      "Epoch 42/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0903 - acc: 0.9760\n",
      "Epoch 43/500\n",
      "1000/1000 [==============================] - 0s 19us/step - loss: 0.0875 - acc: 0.9770\n",
      "Epoch 44/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0923 - acc: 0.9730\n",
      "Epoch 45/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0938 - acc: 0.9690\n",
      "Epoch 46/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0986 - acc: 0.9670\n",
      "Epoch 47/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0850 - acc: 0.9740\n",
      "Epoch 48/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0829 - acc: 0.9790\n",
      "Epoch 49/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0817 - acc: 0.9750\n",
      "Epoch 50/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0874 - acc: 0.9750\n",
      "Epoch 51/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0850 - acc: 0.9770\n",
      "Epoch 52/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0799 - acc: 0.9760\n",
      "Epoch 53/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0787 - acc: 0.9770\n",
      "Epoch 54/500\n",
      "1000/1000 [==============================] - 0s 19us/step - loss: 0.0796 - acc: 0.9780\n",
      "Epoch 55/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0825 - acc: 0.9760\n",
      "Epoch 56/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0848 - acc: 0.9710\n",
      "Epoch 57/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0777 - acc: 0.9750\n",
      "Epoch 58/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0764 - acc: 0.9790\n",
      "Epoch 59/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0767 - acc: 0.9740\n",
      "Epoch 60/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0814 - acc: 0.9730\n",
      "Epoch 61/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0748 - acc: 0.9760\n",
      "Epoch 62/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0751 - acc: 0.9780\n",
      "Epoch 63/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0717 - acc: 0.9780\n",
      "Epoch 64/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0747 - acc: 0.9730\n",
      "Epoch 65/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0777 - acc: 0.9760\n",
      "Epoch 66/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0744 - acc: 0.9820\n",
      "Epoch 67/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0735 - acc: 0.9760\n",
      "Epoch 68/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0733 - acc: 0.9810\n",
      "Epoch 69/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0768 - acc: 0.9770\n",
      "Epoch 70/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0743 - acc: 0.9740\n",
      "Epoch 71/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0764 - acc: 0.9720\n",
      "Epoch 72/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0730 - acc: 0.9760\n",
      "Epoch 73/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0710 - acc: 0.9780\n",
      "Epoch 74/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0739 - acc: 0.9750\n",
      "Epoch 75/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0745 - acc: 0.9730\n",
      "Epoch 76/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0745 - acc: 0.9700\n",
      "Epoch 77/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0708 - acc: 0.9740\n",
      "Epoch 78/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0726 - acc: 0.9730\n",
      "Epoch 79/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0686 - acc: 0.9770\n",
      "Epoch 80/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0708 - acc: 0.9750\n",
      "Epoch 81/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0684 - acc: 0.9790\n",
      "Epoch 82/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0693 - acc: 0.9760\n",
      "Epoch 83/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0683 - acc: 0.9740\n",
      "Epoch 84/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0708 - acc: 0.9740\n",
      "Epoch 85/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0772 - acc: 0.9730\n",
      "Epoch 86/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0754 - acc: 0.9740\n",
      "Epoch 87/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0675 - acc: 0.9780\n",
      "Epoch 88/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0700 - acc: 0.9750\n",
      "Epoch 89/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0694 - acc: 0.9760\n",
      "Epoch 90/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0713 - acc: 0.9780\n",
      "Epoch 91/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0854 - acc: 0.9680\n",
      "Epoch 92/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0737 - acc: 0.9720\n",
      "Epoch 93/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0681 - acc: 0.9770\n",
      "Epoch 94/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0679 - acc: 0.9760\n",
      "Epoch 95/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0728 - acc: 0.9740\n",
      "Epoch 96/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0790 - acc: 0.9700\n",
      "Epoch 97/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0788 - acc: 0.9690\n",
      "Epoch 98/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0663 - acc: 0.9760\n",
      "Epoch 99/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0680 - acc: 0.9710\n",
      "Epoch 100/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0708 - acc: 0.9740\n",
      "Epoch 101/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0727 - acc: 0.9730\n",
      "Epoch 102/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0639 - acc: 0.9770\n",
      "Epoch 103/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0677 - acc: 0.9770\n",
      "Epoch 104/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0650 - acc: 0.9790\n",
      "Epoch 105/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0724 - acc: 0.9750\n",
      "Epoch 106/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0701 - acc: 0.9750\n",
      "Epoch 107/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0892 - acc: 0.9680\n",
      "Epoch 108/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0743 - acc: 0.9720\n",
      "Epoch 109/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0733 - acc: 0.9760\n",
      "Epoch 110/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0753 - acc: 0.9680\n",
      "Epoch 111/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0677 - acc: 0.9770\n",
      "Epoch 112/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0720 - acc: 0.9770\n",
      "Epoch 113/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0650 - acc: 0.9780\n",
      "Epoch 114/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0645 - acc: 0.9810\n",
      "Epoch 115/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0682 - acc: 0.9770\n",
      "Epoch 116/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0738 - acc: 0.9690\n",
      "Epoch 117/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0706 - acc: 0.9700\n",
      "Epoch 118/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0652 - acc: 0.9770\n",
      "Epoch 119/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0651 - acc: 0.9760\n",
      "Epoch 120/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0677 - acc: 0.9730\n",
      "Epoch 121/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0636 - acc: 0.9780\n",
      "Epoch 122/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0650 - acc: 0.9750\n",
      "Epoch 123/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0735 - acc: 0.9760\n",
      "Epoch 124/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0749 - acc: 0.9700\n",
      "Epoch 125/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0686 - acc: 0.9770\n",
      "Epoch 126/500\n",
      "1000/1000 [==============================] - 0s 21us/step - loss: 0.0728 - acc: 0.9730\n",
      "Epoch 127/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0632 - acc: 0.9770\n",
      "Epoch 128/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0689 - acc: 0.9740\n",
      "Epoch 129/500\n",
      "1000/1000 [==============================] - 0s 20us/step - loss: 0.0756 - acc: 0.9740\n",
      "Epoch 130/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0670 - acc: 0.9790\n",
      "Epoch 131/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0626 - acc: 0.9770\n",
      "Epoch 132/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0648 - acc: 0.9740\n",
      "Epoch 133/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0685 - acc: 0.9720\n",
      "Epoch 134/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0659 - acc: 0.9780\n",
      "Epoch 135/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0691 - acc: 0.9740\n",
      "Epoch 136/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0640 - acc: 0.9770\n",
      "Epoch 137/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0734 - acc: 0.9730\n",
      "Epoch 138/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0731 - acc: 0.9760\n",
      "Epoch 139/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0628 - acc: 0.9780\n",
      "Epoch 140/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0664 - acc: 0.9780\n",
      "Epoch 141/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0632 - acc: 0.9750\n",
      "Epoch 142/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0708 - acc: 0.9740\n",
      "Epoch 143/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0652 - acc: 0.9790\n",
      "Epoch 144/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0816 - acc: 0.9720\n",
      "Epoch 145/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0726 - acc: 0.9710\n",
      "Epoch 146/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0707 - acc: 0.9750\n",
      "Epoch 147/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0805 - acc: 0.9680\n",
      "Epoch 148/500\n",
      "1000/1000 [==============================] - 0s 21us/step - loss: 0.0608 - acc: 0.9780\n",
      "Epoch 149/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0635 - acc: 0.9770\n",
      "Epoch 150/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0647 - acc: 0.9740\n",
      "Epoch 151/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0662 - acc: 0.9710\n",
      "Epoch 152/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0647 - acc: 0.9760\n",
      "Epoch 153/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0678 - acc: 0.9730\n",
      "Epoch 154/500\n",
      "1000/1000 [==============================] - 0s 20us/step - loss: 0.0643 - acc: 0.9770\n",
      "Epoch 155/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0683 - acc: 0.9760\n",
      "Epoch 156/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0687 - acc: 0.9720\n",
      "Epoch 157/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0640 - acc: 0.9750\n",
      "Epoch 158/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0616 - acc: 0.9770\n",
      "Epoch 159/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0625 - acc: 0.9770\n",
      "Epoch 160/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0717 - acc: 0.9720\n",
      "Epoch 161/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0647 - acc: 0.9740\n",
      "Epoch 162/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0617 - acc: 0.9740\n",
      "Epoch 163/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0673 - acc: 0.9790\n",
      "Epoch 164/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0730 - acc: 0.9710\n",
      "Epoch 165/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0660 - acc: 0.9790\n",
      "Epoch 166/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0687 - acc: 0.9740\n",
      "Epoch 167/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0704 - acc: 0.9740\n",
      "Epoch 168/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0656 - acc: 0.9780\n",
      "Epoch 169/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0679 - acc: 0.9780\n",
      "Epoch 170/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0776 - acc: 0.9640\n",
      "Epoch 171/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0719 - acc: 0.9710\n",
      "Epoch 172/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0641 - acc: 0.9710\n",
      "Epoch 173/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0724 - acc: 0.9710\n",
      "Epoch 174/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0654 - acc: 0.9770\n",
      "Epoch 175/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0748 - acc: 0.9730\n",
      "Epoch 176/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0672 - acc: 0.9730\n",
      "Epoch 177/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0680 - acc: 0.9760\n",
      "Epoch 178/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0803 - acc: 0.9670\n",
      "Epoch 179/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0642 - acc: 0.9760\n",
      "Epoch 180/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0708 - acc: 0.9710\n",
      "Epoch 181/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0663 - acc: 0.9760\n",
      "Epoch 182/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0673 - acc: 0.9760\n",
      "Epoch 183/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0682 - acc: 0.9740\n",
      "Epoch 184/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0608 - acc: 0.9780\n",
      "Epoch 185/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0625 - acc: 0.9750\n",
      "Epoch 186/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0700 - acc: 0.9720\n",
      "Epoch 187/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0645 - acc: 0.9760\n",
      "Epoch 188/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0639 - acc: 0.9760\n",
      "Epoch 189/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0699 - acc: 0.9700\n",
      "Epoch 190/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0771 - acc: 0.9710\n",
      "Epoch 191/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0727 - acc: 0.9720\n",
      "Epoch 192/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0761 - acc: 0.9710\n",
      "Epoch 193/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0663 - acc: 0.9770\n",
      "Epoch 194/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0648 - acc: 0.9740\n",
      "Epoch 195/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0658 - acc: 0.9750\n",
      "Epoch 196/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0711 - acc: 0.9700\n",
      "Epoch 197/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0647 - acc: 0.9760\n",
      "Epoch 198/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0642 - acc: 0.9800\n",
      "Epoch 199/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0613 - acc: 0.9760\n",
      "Epoch 200/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0711 - acc: 0.9720\n",
      "Epoch 201/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0900 - acc: 0.9580\n",
      "Epoch 202/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0916 - acc: 0.9680\n",
      "Epoch 203/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0672 - acc: 0.9750\n",
      "Epoch 204/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0809 - acc: 0.9670\n",
      "Epoch 205/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0639 - acc: 0.9730\n",
      "Epoch 206/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0676 - acc: 0.9750\n",
      "Epoch 207/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0681 - acc: 0.9730\n",
      "Epoch 208/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0626 - acc: 0.9790\n",
      "Epoch 209/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0649 - acc: 0.9750\n",
      "Epoch 210/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0674 - acc: 0.9720\n",
      "Epoch 211/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0681 - acc: 0.9750\n",
      "Epoch 212/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0694 - acc: 0.9680\n",
      "Epoch 213/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0612 - acc: 0.9760\n",
      "Epoch 214/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0669 - acc: 0.9780\n",
      "Epoch 215/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0624 - acc: 0.9780\n",
      "Epoch 216/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0651 - acc: 0.9790\n",
      "Epoch 217/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0644 - acc: 0.9760\n",
      "Epoch 218/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0742 - acc: 0.9710\n",
      "Epoch 219/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0640 - acc: 0.9730\n",
      "Epoch 220/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0628 - acc: 0.9770\n",
      "Epoch 221/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0676 - acc: 0.9730\n",
      "Epoch 222/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0808 - acc: 0.9640\n",
      "Epoch 223/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0604 - acc: 0.9770\n",
      "Epoch 224/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0633 - acc: 0.9780\n",
      "Epoch 225/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0669 - acc: 0.9760\n",
      "Epoch 226/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0658 - acc: 0.9680\n",
      "Epoch 227/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0648 - acc: 0.9770\n",
      "Epoch 228/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0848 - acc: 0.9630\n",
      "Epoch 229/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0767 - acc: 0.9690\n",
      "Epoch 230/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0680 - acc: 0.9720\n",
      "Epoch 231/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0681 - acc: 0.9760\n",
      "Epoch 232/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0688 - acc: 0.9740\n",
      "Epoch 233/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0606 - acc: 0.9770\n",
      "Epoch 234/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0611 - acc: 0.9750\n",
      "Epoch 235/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0670 - acc: 0.9720\n",
      "Epoch 236/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0763 - acc: 0.9670\n",
      "Epoch 237/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0701 - acc: 0.9760\n",
      "Epoch 238/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0718 - acc: 0.9730\n",
      "Epoch 239/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0609 - acc: 0.9790\n",
      "Epoch 240/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0617 - acc: 0.9760\n",
      "Epoch 241/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0640 - acc: 0.9770\n",
      "Epoch 242/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0635 - acc: 0.9780\n",
      "Epoch 243/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0849 - acc: 0.9630\n",
      "Epoch 244/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0734 - acc: 0.9710\n",
      "Epoch 245/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0652 - acc: 0.9760\n",
      "Epoch 246/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0651 - acc: 0.9750\n",
      "Epoch 247/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0660 - acc: 0.9740\n",
      "Epoch 248/500\n",
      "1000/1000 [==============================] - 0s 20us/step - loss: 0.0705 - acc: 0.9720\n",
      "Epoch 249/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0616 - acc: 0.9750\n",
      "Epoch 250/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0655 - acc: 0.9760\n",
      "Epoch 251/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0618 - acc: 0.9790\n",
      "Epoch 252/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0640 - acc: 0.9760\n",
      "Epoch 253/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0679 - acc: 0.9750\n",
      "Epoch 254/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0614 - acc: 0.9770\n",
      "Epoch 255/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0729 - acc: 0.9750\n",
      "Epoch 256/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0758 - acc: 0.9720\n",
      "Epoch 257/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0656 - acc: 0.9760\n",
      "Epoch 258/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0658 - acc: 0.9750\n",
      "Epoch 259/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0608 - acc: 0.9790\n",
      "Epoch 260/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0666 - acc: 0.9770\n",
      "Epoch 261/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0620 - acc: 0.9760\n",
      "Epoch 262/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0585 - acc: 0.9810\n",
      "Epoch 263/500\n",
      "1000/1000 [==============================] - 0s 19us/step - loss: 0.0662 - acc: 0.9760\n",
      "Epoch 264/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0616 - acc: 0.9760\n",
      "Epoch 265/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0608 - acc: 0.9780\n",
      "Epoch 266/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0607 - acc: 0.9780\n",
      "Epoch 267/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0657 - acc: 0.9750\n",
      "Epoch 268/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0646 - acc: 0.9750\n",
      "Epoch 269/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0701 - acc: 0.9760\n",
      "Epoch 270/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0712 - acc: 0.9710\n",
      "Epoch 271/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0708 - acc: 0.9750\n",
      "Epoch 272/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0717 - acc: 0.9690\n",
      "Epoch 273/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0869 - acc: 0.9640\n",
      "Epoch 274/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0793 - acc: 0.9670\n",
      "Epoch 275/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0712 - acc: 0.9770\n",
      "Epoch 276/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0603 - acc: 0.9750\n",
      "Epoch 277/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0638 - acc: 0.9760\n",
      "Epoch 278/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0629 - acc: 0.9760\n",
      "Epoch 279/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0709 - acc: 0.9720\n",
      "Epoch 280/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0691 - acc: 0.9760\n",
      "Epoch 281/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0670 - acc: 0.9690\n",
      "Epoch 282/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0628 - acc: 0.9780\n",
      "Epoch 283/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0612 - acc: 0.9780\n",
      "Epoch 284/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0643 - acc: 0.9760\n",
      "Epoch 285/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0722 - acc: 0.9690\n",
      "Epoch 286/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0700 - acc: 0.9740\n",
      "Epoch 287/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0794 - acc: 0.9680\n",
      "Epoch 288/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0623 - acc: 0.9740\n",
      "Epoch 289/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0635 - acc: 0.9750\n",
      "Epoch 290/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0638 - acc: 0.9770\n",
      "Epoch 291/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0668 - acc: 0.9760\n",
      "Epoch 292/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0701 - acc: 0.9720\n",
      "Epoch 293/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0686 - acc: 0.9730\n",
      "Epoch 294/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0645 - acc: 0.9770\n",
      "Epoch 295/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0817 - acc: 0.9660\n",
      "Epoch 296/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0594 - acc: 0.9780\n",
      "Epoch 297/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0761 - acc: 0.9690\n",
      "Epoch 298/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0622 - acc: 0.9760\n",
      "Epoch 299/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0612 - acc: 0.9780\n",
      "Epoch 300/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0725 - acc: 0.9710\n",
      "Epoch 301/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0614 - acc: 0.9760\n",
      "Epoch 302/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0621 - acc: 0.9780\n",
      "Epoch 303/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0784 - acc: 0.9730\n",
      "Epoch 304/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0816 - acc: 0.9680\n",
      "Epoch 305/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0738 - acc: 0.9700\n",
      "Epoch 306/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0737 - acc: 0.9730\n",
      "Epoch 307/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0654 - acc: 0.9760\n",
      "Epoch 308/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0643 - acc: 0.9770\n",
      "Epoch 309/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0644 - acc: 0.9750\n",
      "Epoch 310/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0701 - acc: 0.9730\n",
      "Epoch 311/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0706 - acc: 0.9730\n",
      "Epoch 312/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0673 - acc: 0.9770\n",
      "Epoch 313/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0618 - acc: 0.9760\n",
      "Epoch 314/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0652 - acc: 0.9740\n",
      "Epoch 315/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0659 - acc: 0.9740\n",
      "Epoch 316/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0608 - acc: 0.9750\n",
      "Epoch 317/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0619 - acc: 0.9750\n",
      "Epoch 318/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0678 - acc: 0.9780\n",
      "Epoch 319/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0792 - acc: 0.9670\n",
      "Epoch 320/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0659 - acc: 0.9730\n",
      "Epoch 321/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0647 - acc: 0.9710\n",
      "Epoch 322/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0619 - acc: 0.9760\n",
      "Epoch 323/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0657 - acc: 0.9750\n",
      "Epoch 324/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0615 - acc: 0.9780\n",
      "Epoch 325/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0616 - acc: 0.9780\n",
      "Epoch 326/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0658 - acc: 0.9730\n",
      "Epoch 327/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0641 - acc: 0.9770\n",
      "Epoch 328/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0604 - acc: 0.9770\n",
      "Epoch 329/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0621 - acc: 0.9760\n",
      "Epoch 330/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0625 - acc: 0.9770\n",
      "Epoch 331/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0788 - acc: 0.9700\n",
      "Epoch 332/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0631 - acc: 0.9770\n",
      "Epoch 333/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0641 - acc: 0.9750\n",
      "Epoch 334/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0648 - acc: 0.9760\n",
      "Epoch 335/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0830 - acc: 0.9680\n",
      "Epoch 336/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0707 - acc: 0.9720\n",
      "Epoch 337/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0761 - acc: 0.9700\n",
      "Epoch 338/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0655 - acc: 0.9780\n",
      "Epoch 339/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0613 - acc: 0.9770\n",
      "Epoch 340/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0665 - acc: 0.9720\n",
      "Epoch 341/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0631 - acc: 0.9760\n",
      "Epoch 342/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0674 - acc: 0.9730\n",
      "Epoch 343/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0651 - acc: 0.9760\n",
      "Epoch 344/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0723 - acc: 0.9760\n",
      "Epoch 345/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0648 - acc: 0.9750\n",
      "Epoch 346/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0666 - acc: 0.9750\n",
      "Epoch 347/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0662 - acc: 0.9710\n",
      "Epoch 348/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0652 - acc: 0.9780\n",
      "Epoch 349/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0711 - acc: 0.9750\n",
      "Epoch 350/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0740 - acc: 0.9770\n",
      "Epoch 351/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0641 - acc: 0.9760\n",
      "Epoch 352/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0684 - acc: 0.9800\n",
      "Epoch 353/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0726 - acc: 0.9730\n",
      "Epoch 354/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0641 - acc: 0.9760\n",
      "Epoch 355/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0680 - acc: 0.9750\n",
      "Epoch 356/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0679 - acc: 0.9740\n",
      "Epoch 357/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0742 - acc: 0.9640\n",
      "Epoch 358/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0632 - acc: 0.9740\n",
      "Epoch 359/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0717 - acc: 0.9730\n",
      "Epoch 360/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0674 - acc: 0.9770\n",
      "Epoch 361/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0711 - acc: 0.9700\n",
      "Epoch 362/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0608 - acc: 0.9780\n",
      "Epoch 363/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0656 - acc: 0.9760\n",
      "Epoch 364/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0702 - acc: 0.9740\n",
      "Epoch 365/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0725 - acc: 0.9770\n",
      "Epoch 366/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0649 - acc: 0.9750\n",
      "Epoch 367/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0650 - acc: 0.9710\n",
      "Epoch 368/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0642 - acc: 0.9770\n",
      "Epoch 369/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0623 - acc: 0.9790\n",
      "Epoch 370/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0631 - acc: 0.9770\n",
      "Epoch 371/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0647 - acc: 0.9760\n",
      "Epoch 372/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0637 - acc: 0.9720\n",
      "Epoch 373/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0630 - acc: 0.9770\n",
      "Epoch 374/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0620 - acc: 0.9760\n",
      "Epoch 375/500\n",
      "1000/1000 [==============================] - 0s 18us/step - loss: 0.0643 - acc: 0.9780\n",
      "Epoch 376/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0634 - acc: 0.9790\n",
      "Epoch 377/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0806 - acc: 0.9650\n",
      "Epoch 378/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0650 - acc: 0.9720\n",
      "Epoch 379/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0612 - acc: 0.9760\n",
      "Epoch 380/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0718 - acc: 0.9740\n",
      "Epoch 381/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0582 - acc: 0.9800\n",
      "Epoch 382/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0741 - acc: 0.9720\n",
      "Epoch 383/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0693 - acc: 0.9720\n",
      "Epoch 384/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0671 - acc: 0.9750\n",
      "Epoch 385/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0642 - acc: 0.9780\n",
      "Epoch 386/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0636 - acc: 0.9770\n",
      "Epoch 387/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0798 - acc: 0.9660\n",
      "Epoch 388/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0703 - acc: 0.9760\n",
      "Epoch 389/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0684 - acc: 0.9700\n",
      "Epoch 390/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0626 - acc: 0.9770\n",
      "Epoch 391/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0648 - acc: 0.9780\n",
      "Epoch 392/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0645 - acc: 0.9740\n",
      "Epoch 393/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0647 - acc: 0.9770\n",
      "Epoch 394/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0723 - acc: 0.9750\n",
      "Epoch 395/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0715 - acc: 0.9750\n",
      "Epoch 396/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0627 - acc: 0.9750\n",
      "Epoch 397/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0655 - acc: 0.9740\n",
      "Epoch 398/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0623 - acc: 0.9750\n",
      "Epoch 399/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0720 - acc: 0.9730\n",
      "Epoch 400/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0620 - acc: 0.9780\n",
      "Epoch 401/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0630 - acc: 0.9740\n",
      "Epoch 402/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0646 - acc: 0.9760\n",
      "Epoch 403/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0698 - acc: 0.9720\n",
      "Epoch 404/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0733 - acc: 0.9740\n",
      "Epoch 405/500\n",
      "1000/1000 [==============================] - 0s 22us/step - loss: 0.0761 - acc: 0.9720\n",
      "Epoch 406/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0683 - acc: 0.9700\n",
      "Epoch 407/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0643 - acc: 0.9800\n",
      "Epoch 408/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0669 - acc: 0.9770\n",
      "Epoch 409/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0640 - acc: 0.9760\n",
      "Epoch 410/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0611 - acc: 0.9760\n",
      "Epoch 411/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0625 - acc: 0.9780\n",
      "Epoch 412/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0631 - acc: 0.9760\n",
      "Epoch 413/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0613 - acc: 0.9790\n",
      "Epoch 414/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0610 - acc: 0.9760\n",
      "Epoch 415/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0636 - acc: 0.9740\n",
      "Epoch 416/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0733 - acc: 0.9720\n",
      "Epoch 417/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0669 - acc: 0.9730\n",
      "Epoch 418/500\n",
      "1000/1000 [==============================] - 0s 19us/step - loss: 0.0653 - acc: 0.9750\n",
      "Epoch 419/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0638 - acc: 0.9750\n",
      "Epoch 420/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0687 - acc: 0.9720\n",
      "Epoch 421/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0654 - acc: 0.9720\n",
      "Epoch 422/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0673 - acc: 0.9730\n",
      "Epoch 423/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0720 - acc: 0.9720\n",
      "Epoch 424/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0649 - acc: 0.9780\n",
      "Epoch 425/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0627 - acc: 0.9770\n",
      "Epoch 426/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0645 - acc: 0.9750\n",
      "Epoch 427/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0638 - acc: 0.9750\n",
      "Epoch 428/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0690 - acc: 0.9750\n",
      "Epoch 429/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0648 - acc: 0.9750\n",
      "Epoch 430/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0675 - acc: 0.9740\n",
      "Epoch 431/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0715 - acc: 0.9710\n",
      "Epoch 432/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0626 - acc: 0.9760\n",
      "Epoch 433/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0662 - acc: 0.9780\n",
      "Epoch 434/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0655 - acc: 0.9750\n",
      "Epoch 435/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0628 - acc: 0.9810\n",
      "Epoch 436/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0716 - acc: 0.9700\n",
      "Epoch 437/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0610 - acc: 0.9740\n",
      "Epoch 438/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0615 - acc: 0.9770\n",
      "Epoch 439/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0646 - acc: 0.9740\n",
      "Epoch 440/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0642 - acc: 0.9770\n",
      "Epoch 441/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0613 - acc: 0.9750\n",
      "Epoch 442/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0624 - acc: 0.9770\n",
      "Epoch 443/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0626 - acc: 0.9780\n",
      "Epoch 444/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0613 - acc: 0.9780\n",
      "Epoch 445/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0712 - acc: 0.9720\n",
      "Epoch 446/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0612 - acc: 0.9750\n",
      "Epoch 447/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0640 - acc: 0.9720\n",
      "Epoch 448/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0700 - acc: 0.9740\n",
      "Epoch 449/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0683 - acc: 0.9750\n",
      "Epoch 450/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0670 - acc: 0.9770\n",
      "Epoch 451/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0789 - acc: 0.9710\n",
      "Epoch 452/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0670 - acc: 0.9720\n",
      "Epoch 453/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0629 - acc: 0.9770\n",
      "Epoch 454/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0716 - acc: 0.9770\n",
      "Epoch 455/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0652 - acc: 0.9750\n",
      "Epoch 456/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0781 - acc: 0.9690\n",
      "Epoch 457/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0728 - acc: 0.9690\n",
      "Epoch 458/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0673 - acc: 0.9750\n",
      "Epoch 459/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0646 - acc: 0.9780\n",
      "Epoch 460/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0628 - acc: 0.9730\n",
      "Epoch 461/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0666 - acc: 0.9770\n",
      "Epoch 462/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.1010 - acc: 0.9590\n",
      "Epoch 463/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0762 - acc: 0.9720\n",
      "Epoch 464/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0744 - acc: 0.9720\n",
      "Epoch 465/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0734 - acc: 0.9690\n",
      "Epoch 466/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0647 - acc: 0.9730\n",
      "Epoch 467/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0677 - acc: 0.9760\n",
      "Epoch 468/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0650 - acc: 0.9760\n",
      "Epoch 469/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0681 - acc: 0.9740\n",
      "Epoch 470/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0640 - acc: 0.9730\n",
      "Epoch 471/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0663 - acc: 0.9740\n",
      "Epoch 472/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0703 - acc: 0.9720\n",
      "Epoch 473/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0765 - acc: 0.9650\n",
      "Epoch 474/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0763 - acc: 0.9690\n",
      "Epoch 475/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0653 - acc: 0.9770\n",
      "Epoch 476/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0622 - acc: 0.9760\n",
      "Epoch 477/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0766 - acc: 0.9720\n",
      "Epoch 478/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0698 - acc: 0.9730\n",
      "Epoch 479/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0618 - acc: 0.9770\n",
      "Epoch 480/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0645 - acc: 0.9750\n",
      "Epoch 481/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0696 - acc: 0.9730\n",
      "Epoch 482/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0651 - acc: 0.9760\n",
      "Epoch 483/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0651 - acc: 0.9770\n",
      "Epoch 484/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0637 - acc: 0.9750\n",
      "Epoch 485/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0909 - acc: 0.9710\n",
      "Epoch 486/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0686 - acc: 0.9780\n",
      "Epoch 487/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0722 - acc: 0.9730\n",
      "Epoch 488/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0699 - acc: 0.9730\n",
      "Epoch 489/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0668 - acc: 0.9750\n",
      "Epoch 490/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0629 - acc: 0.9780\n",
      "Epoch 491/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0649 - acc: 0.9770\n",
      "Epoch 492/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0648 - acc: 0.9780\n",
      "Epoch 493/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0649 - acc: 0.9750\n",
      "Epoch 494/500\n",
      "1000/1000 [==============================] - 0s 13us/step - loss: 0.0770 - acc: 0.9700\n",
      "Epoch 495/500\n",
      "1000/1000 [==============================] - 0s 17us/step - loss: 0.0628 - acc: 0.9750\n",
      "Epoch 496/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0674 - acc: 0.9740\n",
      "Epoch 497/500\n",
      "1000/1000 [==============================] - 0s 16us/step - loss: 0.0746 - acc: 0.9740\n",
      "Epoch 498/500\n",
      "1000/1000 [==============================] - 0s 14us/step - loss: 0.0657 - acc: 0.9750\n",
      "Epoch 499/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0698 - acc: 0.9710\n",
      "Epoch 500/500\n",
      "1000/1000 [==============================] - 0s 15us/step - loss: 0.0614 - acc: 0.9760\n"
     ]
    }
   ],
   "source": [
    "model=Sequential()\n",
    "model.add(Dense(units=1,input_shape=(2,),activation='sigmoid'))\n",
    "adam=Adam(lr=0.1)\n",
    "model.compile(adam,loss='binary_crossentropy',metrics=['accuracy'])\n",
    "h=model.fit(x=X,y=y,verbose=1,batch_size=50,epochs = 500,shuffle='true')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x2502b089320>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(h.history['acc'])\n",
    "plt.title('accuracy')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x2502b0df390>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(h.history['loss'])\n",
    "plt.title('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['loss'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_decision_boundary(X,y,model):\n",
    "    x_span=np.linspace(min(X[:,0])-1,max(X[:,0])+1)\n",
    "    y_span=np.linspace(min(X[:,1])-1,max(X[:,1])+1)\n",
    "    xx,yy=np.meshgrid(x_span,y_span)\n",
    "    xx_,yy_=xx.ravel(),yy.ravel()\n",
    "    grid=np.c_[xx_,yy_]\n",
    "    pred_func=model.predict(grid)\n",
    "    z=pred_func.reshape(xx.shape)\n",
    "    plt.contourf(xx,yy,z)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x2502c2f74a8>"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_decision_boundary(X,y,model)\n",
    "plt.scatter(X[:n_pts,0],X[:n_pts,1])\n",
    "plt.scatter(X[n_pts:,0],X[n_pts:,1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x2502c297320>"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}