lstm_multivariate_time_series.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Libraries needed"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "from keras.layers import Dense, Dropout\n",
    "from keras.layers import LSTM\n",
    "from keras.models import Sequential\n",
    "from keras.optimizers import Adam\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from urad_Preprocessing import get_data_preprocessing\n",
    "row = 14\n",
    "time_prediction = row*3 #in days, it is equal to one week"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Function List (\"Data splitting (training and testing)\", \"ANN\", \"LSTM\", \"convert_dataset\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Function List (\"LSTM\" and \"dataset converter\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#Make the data into classification problem (up and down)\n",
    "def subtract(x1, x0):\n",
    "    return (x1 - x0) > 0\n",
    "\n",
    "def regression_to_class(inp):\n",
    "    len_input = len(inp)\n",
    "    output = np.zeros_like(inp)\n",
    "    for i in range(1,len_input): output[i] = subtract(inp[i], inp[i-1])\n",
    "    #print(inp.shape, output.shape)\n",
    "    return output "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#build LSTM model with return sequences\n",
    "batch_size = 100\n",
    "def build_model_LSTM(learning_rate, time_prediction):\n",
    "    lstm_model = Sequential()\n",
    "    lstm_model.add( LSTM(time_prediction*2, input_shape = ( 1, time_prediction-1), return_sequences = True) ) \n",
    "    lstm_model.add(Dropout(0.5))\n",
    "    lstm_model.add( LSTM(time_prediction*1, return_sequences = True) )\n",
    "    lstm_model.add(Dropout(0.5))\n",
    "    lstm_model.add( LSTM(int(time_prediction/2)) )\n",
    "    lstm_model.add(Dropout(0.5))\n",
    "    lstm_model.add(Dense(1))\n",
    "    adam = Adam(decay = learning_rate*-0.1, lr=learning_rate)\n",
    "    lstm_model.compile(loss='mean_squared_error', optimizer = adam)\n",
    "    return lstm_model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# convert an array of values into a dataset matrix\n",
    "def convert_dataset(dataset, time_prediction, len_data):\n",
    "    # look back default value is in day, it is equal to one week\n",
    "    dataX, dataY = [], []\n",
    "    for i in range(0, len_data-time_prediction+1, row):\n",
    "        #if i > 7000: print(i)\n",
    "        a = dataset[i:i+time_prediction-1]\n",
    "        dataX.append(a)\n",
    "        dataY.append(dataset[i + time_prediction-1])\n",
    "    dataX = np.array(dataX).reshape(-1, 1, time_prediction-1)\n",
    "    return dataX, np.array(dataY)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Preprocessing (to make it into time-series)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Prepare the dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Files from the folder-fb ads/ that contains-22 files are succesfully loaded...\n",
      "(510, 14)\n",
      "Preprocessing is successful!\n",
      "(7140, 1) 510\n"
     ]
    }
   ],
   "source": [
    "np.random.seed(5) #set pseudo checkpoint\n",
    "get_data_preprocessing() #preprocess the raw data and get the important features\n",
    "dataset = pd.read_csv(\"fb_ads_preprocess.csv\", header=None)\n",
    "len_data = len(dataset)\n",
    "dataset = np.array(dataset).flatten().reshape(-1,1)\n",
    "print(dataset.shape, len_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Data Splitting (for training and testing)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(508, 1, 41)\n",
      "(408, 1, 41) (100, 1, 41)\n"
     ]
    }
   ],
   "source": [
    "#Divide the data into training, and test data\n",
    "datasetX, datasetY= convert_dataset(dataset, time_prediction, len_data*row)\n",
    "print(datasetX.shape)\n",
    "\n",
    "#prepare the dataset, and divide them into training and testing dataset\n",
    "train_size = int(len_data * 0.8)\n",
    "test_size = len_data - train_size\n",
    "trainX, trainY, testX, testY =  datasetX[:train_size,:], datasetY[:train_size], \\\n",
    "                                datasetX[train_size:,:], datasetY[train_size:]\n",
    "print(trainX.shape, testX.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Validation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Do Validation and See its Performance "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 326 samples, validate on 82 samples\n",
      "Epoch 1/1000\n",
      "326/326 [==============================] - 3s - loss: 0.0135 - val_loss: 0.0270\n",
      "Epoch 2/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0133 - val_loss: 0.0265\n",
      "Epoch 3/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0129 - val_loss: 0.0261\n",
      "Epoch 4/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0127 - val_loss: 0.0257\n",
      "Epoch 5/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0122 - val_loss: 0.0252\n",
      "Epoch 6/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0122 - val_loss: 0.0248\n",
      "Epoch 7/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0117 - val_loss: 0.0244\n",
      "Epoch 8/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0114 - val_loss: 0.0240\n",
      "Epoch 9/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0112 - val_loss: 0.0235\n",
      "Epoch 10/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0111 - val_loss: 0.0231\n",
      "Epoch 11/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0107 - val_loss: 0.0227\n",
      "Epoch 12/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0107 - val_loss: 0.0223\n",
      "Epoch 13/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0101 - val_loss: 0.0219\n",
      "Epoch 14/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0100 - val_loss: 0.0214\n",
      "Epoch 15/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0095 - val_loss: 0.0210\n",
      "Epoch 16/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0094 - val_loss: 0.0206\n",
      "Epoch 17/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0092 - val_loss: 0.0202\n",
      "Epoch 18/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0089 - val_loss: 0.0198\n",
      "Epoch 19/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0087 - val_loss: 0.0193\n",
      "Epoch 20/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0085 - val_loss: 0.0189\n",
      "Epoch 21/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0081 - val_loss: 0.0185\n",
      "Epoch 22/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0080 - val_loss: 0.0181\n",
      "Epoch 23/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0077 - val_loss: 0.0177\n",
      "Epoch 24/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0073 - val_loss: 0.0173\n",
      "Epoch 25/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0072 - val_loss: 0.0168\n",
      "Epoch 26/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0071 - val_loss: 0.0164\n",
      "Epoch 27/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0069 - val_loss: 0.0160\n",
      "Epoch 28/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0068 - val_loss: 0.0156\n",
      "Epoch 29/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0062 - val_loss: 0.0152\n",
      "Epoch 30/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0063 - val_loss: 0.0148\n",
      "Epoch 31/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0063 - val_loss: 0.0145\n",
      "Epoch 32/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0061 - val_loss: 0.0141\n",
      "Epoch 33/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0056 - val_loss: 0.0137\n",
      "Epoch 34/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0053 - val_loss: 0.0134\n",
      "Epoch 35/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0057 - val_loss: 0.0130\n",
      "Epoch 36/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0050 - val_loss: 0.0126\n",
      "Epoch 37/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0051 - val_loss: 0.0123\n",
      "Epoch 38/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0051 - val_loss: 0.0119\n",
      "Epoch 39/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0050 - val_loss: 0.0116\n",
      "Epoch 40/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0048 - val_loss: 0.0113\n",
      "Epoch 41/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0048 - val_loss: 0.0110\n",
      "Epoch 42/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0047 - val_loss: 0.0107\n",
      "Epoch 43/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0044 - val_loss: 0.0104\n",
      "Epoch 44/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0043 - val_loss: 0.0102\n",
      "Epoch 45/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0047 - val_loss: 0.0100\n",
      "Epoch 46/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0044 - val_loss: 0.0098\n",
      "Epoch 47/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0042 - val_loss: 0.0096\n",
      "Epoch 48/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0046 - val_loss: 0.0094\n",
      "Epoch 49/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0040 - val_loss: 0.0093\n",
      "Epoch 50/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0042 - val_loss: 0.0091\n",
      "Epoch 51/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0042 - val_loss: 0.0090\n",
      "Epoch 52/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0041 - val_loss: 0.0089\n",
      "Epoch 53/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0042 - val_loss: 0.0088\n",
      "Epoch 54/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0041 - val_loss: 0.0087\n",
      "Epoch 55/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0043 - val_loss: 0.0087\n",
      "Epoch 56/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0040 - val_loss: 0.0086\n",
      "Epoch 57/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0040 - val_loss: 0.0086\n",
      "Epoch 58/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0038 - val_loss: 0.0085\n",
      "Epoch 59/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0039 - val_loss: 0.0085\n",
      "Epoch 60/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0038 - val_loss: 0.0084\n",
      "Epoch 61/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0043 - val_loss: 0.0084\n",
      "Epoch 62/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0043 - val_loss: 0.0083\n",
      "Epoch 63/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0037 - val_loss: 0.0083\n",
      "Epoch 64/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0039 - val_loss: 0.0082\n",
      "Epoch 65/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0042 - val_loss: 0.0082\n",
      "Epoch 66/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0044 - val_loss: 0.0082\n",
      "Epoch 67/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0040 - val_loss: 0.0082\n",
      "Epoch 68/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0040 - val_loss: 0.0082\n",
      "Epoch 69/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0042 - val_loss: 0.0082\n",
      "Epoch 70/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0039 - val_loss: 0.0081\n",
      "Epoch 71/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0036 - val_loss: 0.0081\n",
      "Epoch 72/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0041 - val_loss: 0.0080\n",
      "Epoch 73/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0040 - val_loss: 0.0080\n",
      "Epoch 74/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0038 - val_loss: 0.0080\n",
      "Epoch 75/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0041 - val_loss: 0.0079\n",
      "Epoch 76/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0039 - val_loss: 0.0079\n",
      "Epoch 77/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0042 - val_loss: 0.0079\n",
      "Epoch 78/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0039 - val_loss: 0.0079\n",
      "Epoch 79/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0038 - val_loss: 0.0078\n",
      "Epoch 80/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0038 - val_loss: 0.0078\n",
      "Epoch 81/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0037 - val_loss: 0.0078\n",
      "Epoch 82/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0037 - val_loss: 0.0078\n",
      "Epoch 83/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0041 - val_loss: 0.0077\n",
      "Epoch 84/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0038 - val_loss: 0.0077\n",
      "Epoch 85/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0036 - val_loss: 0.0077\n",
      "Epoch 86/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0039 - val_loss: 0.0076\n",
      "Epoch 87/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "326/326 [==============================] - 0s - loss: 0.0038 - val_loss: 0.0076\n",
      "Epoch 88/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0038 - val_loss: 0.0075\n",
      "Epoch 89/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0039 - val_loss: 0.0075\n",
      "Epoch 90/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0039 - val_loss: 0.0075\n",
      "Epoch 91/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0039 - val_loss: 0.0075\n",
      "Epoch 92/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0039 - val_loss: 0.0076\n",
      "Epoch 93/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0038 - val_loss: 0.0075\n",
      "Epoch 94/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0036 - val_loss: 0.0075\n",
      "Epoch 95/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0039 - val_loss: 0.0075\n",
      "Epoch 96/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0075\n",
      "Epoch 97/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0037 - val_loss: 0.0075\n",
      "Epoch 98/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0036 - val_loss: 0.0075\n",
      "Epoch 99/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0039 - val_loss: 0.0075\n",
      "Epoch 100/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0036 - val_loss: 0.0075\n",
      "Epoch 101/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0036 - val_loss: 0.0075\n",
      "Epoch 102/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0037 - val_loss: 0.0075\n",
      "Epoch 103/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0038 - val_loss: 0.0075\n",
      "Epoch 104/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0037 - val_loss: 0.0075\n",
      "Epoch 105/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0037 - val_loss: 0.0075\n",
      "Epoch 106/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0037 - val_loss: 0.0075\n",
      "Epoch 107/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0037 - val_loss: 0.0075\n",
      "Epoch 108/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0038 - val_loss: 0.0075\n",
      "Epoch 109/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0040 - val_loss: 0.0075\n",
      "Epoch 110/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0038 - val_loss: 0.0075\n",
      "Epoch 111/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0038 - val_loss: 0.0075\n",
      "Epoch 112/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0036 - val_loss: 0.0075\n",
      "Epoch 113/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0038 - val_loss: 0.0075\n",
      "Epoch 114/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0039 - val_loss: 0.0075\n",
      "Epoch 115/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0075\n",
      "Epoch 116/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0038 - val_loss: 0.0075\n",
      "Epoch 117/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0037 - val_loss: 0.0075\n",
      "Epoch 118/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0037 - val_loss: 0.0074\n",
      "Epoch 119/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0038 - val_loss: 0.0074\n",
      "Epoch 120/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0038 - val_loss: 0.0074\n",
      "Epoch 121/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0036 - val_loss: 0.0073\n",
      "Epoch 122/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0073\n",
      "Epoch 123/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0036 - val_loss: 0.0073\n",
      "Epoch 124/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0072\n",
      "Epoch 125/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0072\n",
      "Epoch 126/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0072\n",
      "Epoch 127/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0039 - val_loss: 0.0072\n",
      "Epoch 128/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0072\n",
      "Epoch 129/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0038 - val_loss: 0.0073\n",
      "Epoch 130/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0073\n",
      "Epoch 131/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0036 - val_loss: 0.0073\n",
      "Epoch 132/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0038 - val_loss: 0.0073\n",
      "Epoch 133/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0037 - val_loss: 0.0073\n",
      "Epoch 134/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0072\n",
      "Epoch 135/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0072\n",
      "Epoch 136/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0072\n",
      "Epoch 137/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0072\n",
      "Epoch 138/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0072\n",
      "Epoch 139/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0072\n",
      "Epoch 140/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0072\n",
      "Epoch 141/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0072\n",
      "Epoch 142/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0072\n",
      "Epoch 143/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0072\n",
      "Epoch 144/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0072\n",
      "Epoch 145/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0037 - val_loss: 0.0072\n",
      "Epoch 146/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0072\n",
      "Epoch 147/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0036 - val_loss: 0.0071\n",
      "Epoch 148/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0071\n",
      "Epoch 149/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0071\n",
      "Epoch 150/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0071\n",
      "Epoch 151/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0071\n",
      "Epoch 152/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0071\n",
      "Epoch 153/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0036 - val_loss: 0.0071\n",
      "Epoch 154/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0036 - val_loss: 0.0071\n",
      "Epoch 155/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0070\n",
      "Epoch 156/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0070\n",
      "Epoch 157/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0036 - val_loss: 0.0069\n",
      "Epoch 158/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0069\n",
      "Epoch 159/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0037 - val_loss: 0.0069\n",
      "Epoch 160/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0069\n",
      "Epoch 161/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0069\n",
      "Epoch 162/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0069\n",
      "Epoch 163/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0069\n",
      "Epoch 164/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0037 - val_loss: 0.0069\n",
      "Epoch 165/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0069\n",
      "Epoch 166/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0068\n",
      "Epoch 167/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0068\n",
      "Epoch 168/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0068\n",
      "Epoch 169/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0068\n",
      "Epoch 170/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0068\n",
      "Epoch 171/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0068\n",
      "Epoch 172/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0067\n",
      "Epoch 173/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0067\n",
      "Epoch 174/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0067\n",
      "Epoch 175/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0067\n",
      "Epoch 176/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0067\n",
      "Epoch 177/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0067\n",
      "Epoch 178/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0067\n",
      "Epoch 179/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0067\n",
      "Epoch 180/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0067\n",
      "Epoch 181/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0067\n",
      "Epoch 182/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0067\n",
      "Epoch 183/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0067\n",
      "Epoch 184/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0067\n",
      "Epoch 185/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0067\n",
      "Epoch 186/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0067\n",
      "Epoch 187/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0066\n",
      "Epoch 188/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0066\n",
      "Epoch 189/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0066\n",
      "Epoch 190/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0066\n",
      "Epoch 191/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0066\n",
      "Epoch 192/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0066\n",
      "Epoch 193/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0066\n",
      "Epoch 194/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0036 - val_loss: 0.0066\n",
      "Epoch 195/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0066\n",
      "Epoch 196/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0066\n",
      "Epoch 197/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0066\n",
      "Epoch 198/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0066\n",
      "Epoch 199/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0066\n",
      "Epoch 200/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0066\n",
      "Epoch 201/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0066\n",
      "Epoch 202/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0066\n",
      "Epoch 203/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0067\n",
      "Epoch 204/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0067\n",
      "Epoch 205/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0067\n",
      "Epoch 206/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0067\n",
      "Epoch 207/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0067\n",
      "Epoch 208/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0066\n",
      "Epoch 209/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0066\n",
      "Epoch 210/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0065\n",
      "Epoch 211/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0065\n",
      "Epoch 212/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0065\n",
      "Epoch 213/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0064\n",
      "Epoch 214/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0064\n",
      "Epoch 215/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0064\n",
      "Epoch 216/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0064\n",
      "Epoch 217/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0064\n",
      "Epoch 218/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0064\n",
      "Epoch 219/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0064\n",
      "Epoch 220/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0064\n",
      "Epoch 221/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0064\n",
      "Epoch 222/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0064\n",
      "Epoch 223/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0064\n",
      "Epoch 224/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0064\n",
      "Epoch 225/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0064\n",
      "Epoch 226/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0064\n",
      "Epoch 227/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0064\n",
      "Epoch 228/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0064\n",
      "Epoch 229/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0063\n",
      "Epoch 230/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0063\n",
      "Epoch 231/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0063\n",
      "Epoch 232/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0063\n",
      "Epoch 233/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0063\n",
      "Epoch 234/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0063\n",
      "Epoch 235/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0063\n",
      "Epoch 236/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0063\n",
      "Epoch 237/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0062\n",
      "Epoch 238/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0062\n",
      "Epoch 239/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0063\n",
      "Epoch 240/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0062\n",
      "Epoch 241/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0062\n",
      "Epoch 242/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0035 - val_loss: 0.0063\n",
      "Epoch 243/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0062\n",
      "Epoch 244/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0062\n",
      "Epoch 245/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0062\n",
      "Epoch 246/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0062\n",
      "Epoch 247/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0062\n",
      "Epoch 248/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0062\n",
      "Epoch 249/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0062\n",
      "Epoch 250/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0062\n",
      "Epoch 251/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0062\n",
      "Epoch 252/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0062\n",
      "Epoch 253/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0062\n",
      "Epoch 254/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0062\n",
      "Epoch 255/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0062\n",
      "Epoch 256/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0062\n",
      "Epoch 257/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0062\n",
      "Epoch 258/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0062\n",
      "Epoch 259/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0062\n",
      "Epoch 260/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0062\n",
      "Epoch 261/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0062\n",
      "Epoch 262/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0062\n",
      "Epoch 263/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0061\n",
      "Epoch 264/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0060\n",
      "Epoch 265/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0060\n",
      "Epoch 266/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0060\n",
      "Epoch 267/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0060\n",
      "Epoch 268/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0034 - val_loss: 0.0060\n",
      "Epoch 269/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0060\n",
      "Epoch 270/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0060\n",
      "Epoch 271/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0060\n",
      "Epoch 272/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0060\n",
      "Epoch 273/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0060\n",
      "Epoch 274/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0060\n",
      "Epoch 275/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0060\n",
      "Epoch 276/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0060\n",
      "Epoch 277/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0060\n",
      "Epoch 278/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0060\n",
      "Epoch 279/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0059\n",
      "Epoch 280/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0059\n",
      "Epoch 281/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0059\n",
      "Epoch 282/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0059\n",
      "Epoch 283/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0059\n",
      "Epoch 284/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0058\n",
      "Epoch 285/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0058\n",
      "Epoch 286/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0057\n",
      "Epoch 287/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0057\n",
      "Epoch 288/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0056\n",
      "Epoch 289/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0056\n",
      "Epoch 290/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0056\n",
      "Epoch 291/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0057\n",
      "Epoch 292/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0057\n",
      "Epoch 293/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0057\n",
      "Epoch 294/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0057\n",
      "Epoch 295/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0057\n",
      "Epoch 296/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0057\n",
      "Epoch 297/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0056\n",
      "Epoch 298/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0056\n",
      "Epoch 299/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0056\n",
      "Epoch 300/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0056\n",
      "Epoch 301/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0056\n",
      "Epoch 302/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0056\n",
      "Epoch 303/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0055\n",
      "Epoch 304/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0055\n",
      "Epoch 305/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0055\n",
      "Epoch 306/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0055\n",
      "Epoch 307/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0055\n",
      "Epoch 308/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0055\n",
      "Epoch 309/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0055\n",
      "Epoch 310/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0055\n",
      "Epoch 311/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0055\n",
      "Epoch 312/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0055\n",
      "Epoch 313/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0055\n",
      "Epoch 314/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0055\n",
      "Epoch 315/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0055\n",
      "Epoch 316/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0055\n",
      "Epoch 317/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0056\n",
      "Epoch 318/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0056\n",
      "Epoch 319/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0056\n",
      "Epoch 320/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0056\n",
      "Epoch 321/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0056\n",
      "Epoch 322/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0056\n",
      "Epoch 323/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0056\n",
      "Epoch 324/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0056\n",
      "Epoch 325/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0056\n",
      "Epoch 326/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0056\n",
      "Epoch 327/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0056\n",
      "Epoch 328/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0056\n",
      "Epoch 329/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0056\n",
      "Epoch 330/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0056\n",
      "Epoch 331/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0056\n",
      "Epoch 332/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0056\n",
      "Epoch 333/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0056\n",
      "Epoch 334/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0056\n",
      "Epoch 335/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0055\n",
      "Epoch 336/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0055\n",
      "Epoch 337/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0055\n",
      "Epoch 338/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0055\n",
      "Epoch 339/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0032 - val_loss: 0.0055\n",
      "Epoch 340/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0055\n",
      "Epoch 341/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0055\n",
      "Epoch 342/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0055\n",
      "Epoch 343/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0054\n",
      "Epoch 344/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0054\n",
      "Epoch 345/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0054\n",
      "Epoch 346/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0054\n",
      "Epoch 347/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0054\n",
      "Epoch 348/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0033 - val_loss: 0.0054\n",
      "Epoch 349/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0054\n",
      "Epoch 350/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0054\n",
      "Epoch 351/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0054\n",
      "Epoch 352/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0054\n",
      "Epoch 353/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0054\n",
      "Epoch 354/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0054\n",
      "Epoch 355/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0054\n",
      "Epoch 356/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0054\n",
      "Epoch 357/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0054\n",
      "Epoch 358/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0054\n",
      "Epoch 359/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0054\n",
      "Epoch 360/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0054\n",
      "Epoch 361/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0053\n",
      "Epoch 362/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0053\n",
      "Epoch 363/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0053\n",
      "Epoch 364/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0052\n",
      "Epoch 365/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0052\n",
      "Epoch 366/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0052\n",
      "Epoch 367/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0052\n",
      "Epoch 368/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0052\n",
      "Epoch 369/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0052\n",
      "Epoch 370/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0052\n",
      "Epoch 371/1000\n",
      "326/326 [==============================] - ETA: 0s - loss: 0.001 - 0s - loss: 0.0028 - val_loss: 0.0052\n",
      "Epoch 372/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0052\n",
      "Epoch 373/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0052\n",
      "Epoch 374/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0052\n",
      "Epoch 375/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0051\n",
      "Epoch 376/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0051\n",
      "Epoch 377/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0050\n",
      "Epoch 378/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0050\n",
      "Epoch 379/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0050\n",
      "Epoch 380/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0050\n",
      "Epoch 381/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0050\n",
      "Epoch 382/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0050\n",
      "Epoch 383/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0050\n",
      "Epoch 384/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0050\n",
      "Epoch 385/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0050\n",
      "Epoch 386/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0050\n",
      "Epoch 387/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0050\n",
      "Epoch 388/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0050\n",
      "Epoch 389/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0050\n",
      "Epoch 390/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0050\n",
      "Epoch 391/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0050\n",
      "Epoch 392/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0050\n",
      "Epoch 393/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0050\n",
      "Epoch 394/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0050\n",
      "Epoch 395/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0050\n",
      "Epoch 396/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0050\n",
      "Epoch 397/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0050\n",
      "Epoch 398/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0050\n",
      "Epoch 399/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0050\n",
      "Epoch 400/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0050\n",
      "Epoch 401/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0050\n",
      "Epoch 402/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0049\n",
      "Epoch 403/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0049\n",
      "Epoch 404/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0049\n",
      "Epoch 405/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0049\n",
      "Epoch 406/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0048\n",
      "Epoch 407/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0048\n",
      "Epoch 408/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0048\n",
      "Epoch 409/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0048\n",
      "Epoch 410/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0048\n",
      "Epoch 411/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0048\n",
      "Epoch 412/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0048\n",
      "Epoch 413/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0048\n",
      "Epoch 414/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0047\n",
      "Epoch 415/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0047\n",
      "Epoch 416/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0047\n",
      "Epoch 417/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0047\n",
      "Epoch 418/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0047\n",
      "Epoch 419/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0047\n",
      "Epoch 420/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0046\n",
      "Epoch 421/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0046\n",
      "Epoch 422/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0046\n",
      "Epoch 423/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0045\n",
      "Epoch 424/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0045\n",
      "Epoch 425/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0045\n",
      "Epoch 426/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0045\n",
      "Epoch 427/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0045\n",
      "Epoch 428/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0045\n",
      "Epoch 429/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0031 - val_loss: 0.0045\n",
      "Epoch 430/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0046\n",
      "Epoch 431/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0046\n",
      "Epoch 432/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0046\n",
      "Epoch 433/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0046\n",
      "Epoch 434/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0046\n",
      "Epoch 435/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0046\n",
      "Epoch 436/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0046\n",
      "Epoch 437/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0046\n",
      "Epoch 438/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0046\n",
      "Epoch 439/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0046\n",
      "Epoch 440/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0046\n",
      "Epoch 441/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0046\n",
      "Epoch 442/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0046\n",
      "Epoch 443/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0046\n",
      "Epoch 444/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0046\n",
      "Epoch 445/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0046\n",
      "Epoch 446/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0046\n",
      "Epoch 447/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0045\n",
      "Epoch 448/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0045\n",
      "Epoch 449/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0045\n",
      "Epoch 450/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0045\n",
      "Epoch 451/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0045\n",
      "Epoch 452/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0044\n",
      "Epoch 453/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0044\n",
      "Epoch 454/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0044\n",
      "Epoch 455/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0044\n",
      "Epoch 456/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0044\n",
      "Epoch 457/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0044\n",
      "Epoch 458/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0044\n",
      "Epoch 459/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0043\n",
      "Epoch 460/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0043\n",
      "Epoch 461/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0043\n",
      "Epoch 462/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0043\n",
      "Epoch 463/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0043\n",
      "Epoch 464/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0043\n",
      "Epoch 465/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0043\n",
      "Epoch 466/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0044\n",
      "Epoch 467/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0044\n",
      "Epoch 468/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0044\n",
      "Epoch 469/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0044\n",
      "Epoch 470/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0044\n",
      "Epoch 471/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0044\n",
      "Epoch 472/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0044\n",
      "Epoch 473/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0044\n",
      "Epoch 474/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0044\n",
      "Epoch 475/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0044\n",
      "Epoch 476/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0044\n",
      "Epoch 477/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0044\n",
      "Epoch 478/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0044\n",
      "Epoch 479/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0044\n",
      "Epoch 480/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0044\n",
      "Epoch 481/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0044\n",
      "Epoch 482/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0044\n",
      "Epoch 483/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0044\n",
      "Epoch 484/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0044\n",
      "Epoch 485/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0044\n",
      "Epoch 486/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0044\n",
      "Epoch 487/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0043\n",
      "Epoch 488/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0042\n",
      "Epoch 489/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0042\n",
      "Epoch 490/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0041\n",
      "Epoch 491/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0040\n",
      "Epoch 492/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0039\n",
      "Epoch 493/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0039\n",
      "Epoch 494/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0039\n",
      "Epoch 495/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0039\n",
      "Epoch 496/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0039\n",
      "Epoch 497/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0040\n",
      "Epoch 498/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0040\n",
      "Epoch 499/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0040\n",
      "Epoch 500/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0040\n",
      "Epoch 501/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0040\n",
      "Epoch 502/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0040\n",
      "Epoch 503/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0039\n",
      "Epoch 504/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0039\n",
      "Epoch 505/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0039\n",
      "Epoch 506/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0039\n",
      "Epoch 507/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0039\n",
      "Epoch 508/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0039\n",
      "Epoch 509/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0039\n",
      "Epoch 510/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0039\n",
      "Epoch 511/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0039\n",
      "Epoch 512/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0039\n",
      "Epoch 513/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0039\n",
      "Epoch 514/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0038\n",
      "Epoch 515/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0038\n",
      "Epoch 516/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0030 - val_loss: 0.0039\n",
      "Epoch 517/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0039\n",
      "Epoch 518/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0038\n",
      "Epoch 519/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0038\n",
      "Epoch 520/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0038\n",
      "Epoch 521/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0037\n",
      "Epoch 522/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0038\n",
      "Epoch 523/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0037\n",
      "Epoch 524/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0037\n",
      "Epoch 525/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0038\n",
      "Epoch 526/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0038\n",
      "Epoch 527/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0038\n",
      "Epoch 528/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0038\n",
      "Epoch 529/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0038\n",
      "Epoch 530/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0037\n",
      "Epoch 531/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0037\n",
      "Epoch 532/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0037\n",
      "Epoch 533/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0037\n",
      "Epoch 534/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0036\n",
      "Epoch 535/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0036\n",
      "Epoch 536/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0036\n",
      "Epoch 537/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0036\n",
      "Epoch 538/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0036\n",
      "Epoch 539/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0036\n",
      "Epoch 540/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0036\n",
      "Epoch 541/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0036\n",
      "Epoch 542/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0036\n",
      "Epoch 543/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0036\n",
      "Epoch 544/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0035\n",
      "Epoch 545/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0035\n",
      "Epoch 546/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0035\n",
      "Epoch 547/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0035\n",
      "Epoch 548/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0034\n",
      "Epoch 549/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0034\n",
      "Epoch 550/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0034\n",
      "Epoch 551/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0034\n",
      "Epoch 552/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0034\n",
      "Epoch 553/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0034\n",
      "Epoch 554/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0033\n",
      "Epoch 555/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0033\n",
      "Epoch 556/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0029 - val_loss: 0.0033\n",
      "Epoch 557/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0034\n",
      "Epoch 558/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0034\n",
      "Epoch 559/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0034\n",
      "Epoch 560/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0034\n",
      "Epoch 561/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0034\n",
      "Epoch 562/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0034\n",
      "Epoch 563/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0034\n",
      "Epoch 564/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0033\n",
      "Epoch 565/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0033\n",
      "Epoch 566/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0032\n",
      "Epoch 567/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0032\n",
      "Epoch 568/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0032\n",
      "Epoch 569/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0032\n",
      "Epoch 570/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0032\n",
      "Epoch 571/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0032\n",
      "Epoch 572/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0032\n",
      "Epoch 573/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0032\n",
      "Epoch 574/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0032\n",
      "Epoch 575/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0031\n",
      "Epoch 576/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0031\n",
      "Epoch 577/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0030\n",
      "Epoch 578/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0030\n",
      "Epoch 579/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0029\n",
      "Epoch 580/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0029\n",
      "Epoch 581/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0029\n",
      "Epoch 582/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0029\n",
      "Epoch 583/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0029\n",
      "Epoch 584/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0030\n",
      "Epoch 585/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0030\n",
      "Epoch 586/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0030\n",
      "Epoch 587/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0030\n",
      "Epoch 588/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0030\n",
      "Epoch 589/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0030\n",
      "Epoch 590/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0030\n",
      "Epoch 591/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0030\n",
      "Epoch 592/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0030\n",
      "Epoch 593/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0030\n",
      "Epoch 594/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0030\n",
      "Epoch 595/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0029\n",
      "Epoch 596/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0029\n",
      "Epoch 597/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0029\n",
      "Epoch 598/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0029\n",
      "Epoch 599/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0028\n",
      "Epoch 600/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0028\n",
      "Epoch 601/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0028\n",
      "Epoch 602/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0028\n",
      "Epoch 603/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0027\n",
      "Epoch 604/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0027\n",
      "Epoch 605/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0027\n",
      "Epoch 606/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0027\n",
      "Epoch 607/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0027\n",
      "Epoch 608/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0027\n",
      "Epoch 609/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0027\n",
      "Epoch 610/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0027\n",
      "Epoch 611/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0027\n",
      "Epoch 612/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0027\n",
      "Epoch 613/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0027\n",
      "Epoch 614/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0027\n",
      "Epoch 615/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0027\n",
      "Epoch 616/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0027\n",
      "Epoch 617/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0027\n",
      "Epoch 618/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0027\n",
      "Epoch 619/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0027\n",
      "Epoch 620/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0027\n",
      "Epoch 621/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0028 - val_loss: 0.0026\n",
      "Epoch 622/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0026\n",
      "Epoch 623/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0026\n",
      "Epoch 624/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0026\n",
      "Epoch 625/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0026\n",
      "Epoch 626/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0027\n",
      "Epoch 627/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0027\n",
      "Epoch 628/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0027\n",
      "Epoch 629/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0027\n",
      "Epoch 630/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0027\n",
      "Epoch 631/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0026\n",
      "Epoch 632/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0026\n",
      "Epoch 633/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0026\n",
      "Epoch 634/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0025\n",
      "Epoch 635/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0025\n",
      "Epoch 636/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0024\n",
      "Epoch 637/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0024\n",
      "Epoch 638/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0024\n",
      "Epoch 639/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0024\n",
      "Epoch 640/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0024\n",
      "Epoch 641/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0024\n",
      "Epoch 642/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0024\n",
      "Epoch 643/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0024\n",
      "Epoch 644/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0025\n",
      "Epoch 645/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0025\n",
      "Epoch 646/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0025\n",
      "Epoch 647/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0025\n",
      "Epoch 648/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0025\n",
      "Epoch 649/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0025\n",
      "Epoch 650/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0025\n",
      "Epoch 651/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0024\n",
      "Epoch 652/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0024\n",
      "Epoch 653/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0024\n",
      "Epoch 654/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0024\n",
      "Epoch 655/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0023\n",
      "Epoch 656/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0023\n",
      "Epoch 657/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0023\n",
      "Epoch 658/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0027 - val_loss: 0.0023\n",
      "Epoch 659/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0023\n",
      "Epoch 660/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0023\n",
      "Epoch 661/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0023\n",
      "Epoch 662/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0023\n",
      "Epoch 663/1000\n",
      "326/326 [==============================] - ETA: 0s - loss: 0.001 - 0s - loss: 0.0025 - val_loss: 0.0023\n",
      "Epoch 664/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0023\n",
      "Epoch 665/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0023\n",
      "Epoch 666/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0022\n",
      "Epoch 667/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0022\n",
      "Epoch 668/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0022\n",
      "Epoch 669/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0022\n",
      "Epoch 670/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0022\n",
      "Epoch 671/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0022\n",
      "Epoch 672/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0022\n",
      "Epoch 673/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0022\n",
      "Epoch 674/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0022\n",
      "Epoch 675/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0022\n",
      "Epoch 676/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0022\n",
      "Epoch 677/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0022\n",
      "Epoch 678/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0022\n",
      "Epoch 679/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0022\n",
      "Epoch 680/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0022\n",
      "Epoch 681/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0022\n",
      "Epoch 682/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0022\n",
      "Epoch 683/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0022\n",
      "Epoch 684/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0022\n",
      "Epoch 685/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0021\n",
      "Epoch 686/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0021\n",
      "Epoch 687/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0021\n",
      "Epoch 688/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0021\n",
      "Epoch 689/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0021\n",
      "Epoch 690/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0021\n",
      "Epoch 691/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0021\n",
      "Epoch 692/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0021\n",
      "Epoch 693/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0021\n",
      "Epoch 694/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0021\n",
      "Epoch 695/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0021\n",
      "Epoch 696/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0021\n",
      "Epoch 697/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0021\n",
      "Epoch 698/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0020\n",
      "Epoch 699/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0020\n",
      "Epoch 700/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0020\n",
      "Epoch 701/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0020\n",
      "Epoch 702/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0020\n",
      "Epoch 703/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0020\n",
      "Epoch 704/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0020\n",
      "Epoch 705/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0020\n",
      "Epoch 706/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0020\n",
      "Epoch 707/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0020\n",
      "Epoch 708/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0020\n",
      "Epoch 709/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0020\n",
      "Epoch 710/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0020\n",
      "Epoch 711/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0020\n",
      "Epoch 712/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0019\n",
      "Epoch 713/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0019\n",
      "Epoch 714/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0019\n",
      "Epoch 715/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0019\n",
      "Epoch 716/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0019\n",
      "Epoch 717/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0019\n",
      "Epoch 718/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0018\n",
      "Epoch 719/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0018\n",
      "Epoch 720/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0018\n",
      "Epoch 721/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0017\n",
      "Epoch 722/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0017\n",
      "Epoch 723/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0017\n",
      "Epoch 724/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0017\n",
      "Epoch 725/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0017\n",
      "Epoch 726/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0018\n",
      "Epoch 727/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0018\n",
      "Epoch 728/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0018\n",
      "Epoch 729/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0018\n",
      "Epoch 730/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0018\n",
      "Epoch 731/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0018\n",
      "Epoch 732/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0018\n",
      "Epoch 733/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0018\n",
      "Epoch 734/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0018\n",
      "Epoch 735/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0018\n",
      "Epoch 736/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0018\n",
      "Epoch 737/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0018\n",
      "Epoch 738/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0018\n",
      "Epoch 739/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0018\n",
      "Epoch 740/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0018\n",
      "Epoch 741/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0018\n",
      "Epoch 742/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0018\n",
      "Epoch 743/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0018\n",
      "Epoch 744/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0018\n",
      "Epoch 745/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0018\n",
      "Epoch 746/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0018\n",
      "Epoch 747/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0018\n",
      "Epoch 748/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0018\n",
      "Epoch 749/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0018\n",
      "Epoch 750/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0018\n",
      "Epoch 751/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0018\n",
      "Epoch 752/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0017\n",
      "Epoch 753/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0026 - val_loss: 0.0017\n",
      "Epoch 754/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0017\n",
      "Epoch 755/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0017\n",
      "Epoch 756/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0017\n",
      "Epoch 757/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0017\n",
      "Epoch 758/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0017\n",
      "Epoch 759/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0016\n",
      "Epoch 760/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0016\n",
      "Epoch 761/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0016\n",
      "Epoch 762/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0016\n",
      "Epoch 763/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0016\n",
      "Epoch 764/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0016\n",
      "Epoch 765/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0016\n",
      "Epoch 766/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0016\n",
      "Epoch 767/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0016\n",
      "Epoch 768/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0016\n",
      "Epoch 769/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0016\n",
      "Epoch 770/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0016\n",
      "Epoch 771/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0016\n",
      "Epoch 772/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0016\n",
      "Epoch 773/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0016\n",
      "Epoch 774/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0016\n",
      "Epoch 775/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0016\n",
      "Epoch 776/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0016\n",
      "Epoch 777/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0016\n",
      "Epoch 778/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0016\n",
      "Epoch 779/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0016\n",
      "Epoch 780/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0016\n",
      "Epoch 781/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0016\n",
      "Epoch 782/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0016\n",
      "Epoch 783/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0016\n",
      "Epoch 784/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0016\n",
      "Epoch 785/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0016\n",
      "Epoch 786/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0016\n",
      "Epoch 787/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0016\n",
      "Epoch 788/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0016\n",
      "Epoch 789/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0016\n",
      "Epoch 790/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0016\n",
      "Epoch 791/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0016\n",
      "Epoch 792/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0016\n",
      "Epoch 793/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0016\n",
      "Epoch 794/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0016\n",
      "Epoch 795/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0016\n",
      "Epoch 796/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0016\n",
      "Epoch 797/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0016\n",
      "Epoch 798/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0015\n",
      "Epoch 799/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 800/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0015\n",
      "Epoch 801/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0015\n",
      "Epoch 802/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0015\n",
      "Epoch 803/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 804/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0015\n",
      "Epoch 805/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0015\n",
      "Epoch 806/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 807/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0015\n",
      "Epoch 808/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 809/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 810/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0015\n",
      "Epoch 811/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0015\n",
      "Epoch 812/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 813/1000\n",
      "326/326 [==============================] - ETA: 0s - loss: 0.005 - 0s - loss: 0.0021 - val_loss: 0.0015\n",
      "Epoch 814/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 815/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0015\n",
      "Epoch 816/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 817/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0015\n",
      "Epoch 818/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0015\n",
      "Epoch 819/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0015\n",
      "Epoch 820/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0015\n",
      "Epoch 821/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0015\n",
      "Epoch 822/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0015\n",
      "Epoch 823/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 824/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0015\n",
      "Epoch 825/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 826/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0014\n",
      "Epoch 827/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0014\n",
      "Epoch 828/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0014\n",
      "Epoch 829/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 830/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0014\n",
      "Epoch 831/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 832/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0014\n",
      "Epoch 833/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0014\n",
      "Epoch 834/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 835/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0015\n",
      "Epoch 836/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0015\n",
      "Epoch 837/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 838/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0015\n",
      "Epoch 839/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 840/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0014\n",
      "Epoch 841/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0014\n",
      "Epoch 842/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 843/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0014\n",
      "Epoch 844/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 845/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0014\n",
      "Epoch 846/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 847/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0014\n",
      "Epoch 848/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 849/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 850/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0015\n",
      "Epoch 851/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0015\n",
      "Epoch 852/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0015\n",
      "Epoch 853/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 854/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0015\n",
      "Epoch 855/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0015\n",
      "Epoch 856/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0014\n",
      "Epoch 857/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0014\n",
      "Epoch 858/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0014\n",
      "Epoch 859/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0014\n",
      "Epoch 860/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 861/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0014\n",
      "Epoch 862/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0014\n",
      "Epoch 863/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0014\n",
      "Epoch 864/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 865/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 866/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0014\n",
      "Epoch 867/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 868/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0014\n",
      "Epoch 869/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 870/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 871/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 872/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0014\n",
      "Epoch 873/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 874/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 875/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0014\n",
      "Epoch 876/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 877/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 878/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0014\n",
      "Epoch 879/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 880/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 881/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0014\n",
      "Epoch 882/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 883/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 884/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0014\n",
      "Epoch 885/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 886/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 887/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 888/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0014\n",
      "Epoch 889/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 890/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 891/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0018 - val_loss: 0.0015\n",
      "Epoch 892/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0015\n",
      "Epoch 893/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0015\n",
      "Epoch 894/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0019 - val_loss: 0.0015\n",
      "Epoch 895/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0015\n",
      "Epoch 896/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0015\n",
      "Epoch 897/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 898/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 899/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0015\n",
      "Epoch 900/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0015\n",
      "Epoch 901/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0015\n",
      "Epoch 902/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 903/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0015\n",
      "Epoch 904/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 905/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 906/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 907/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 908/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 909/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 910/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 911/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 912/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 913/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0014\n",
      "Epoch 914/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0014\n",
      "Epoch 915/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0014\n",
      "Epoch 916/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 917/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 918/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0014\n",
      "Epoch 919/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0015\n",
      "Epoch 920/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0015\n",
      "Epoch 921/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0015\n",
      "Epoch 922/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 923/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 924/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0015\n",
      "Epoch 925/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0015\n",
      "Epoch 926/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0015\n",
      "Epoch 927/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0015\n",
      "Epoch 928/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 929/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 930/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 931/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 932/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0014\n",
      "Epoch 933/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0014\n",
      "Epoch 934/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 935/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0014\n",
      "Epoch 936/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 937/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0019 - val_loss: 0.0014\n",
      "Epoch 938/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0019 - val_loss: 0.0014\n",
      "Epoch 939/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0014\n",
      "Epoch 940/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0018 - val_loss: 0.0014\n",
      "Epoch 941/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 942/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 943/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0014\n",
      "Epoch 944/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0014\n",
      "Epoch 945/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 946/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0015\n",
      "Epoch 947/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 948/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0015\n",
      "Epoch 949/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0015\n",
      "Epoch 950/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0015\n",
      "Epoch 951/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 952/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0015\n",
      "Epoch 953/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 954/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0015\n",
      "Epoch 955/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0015\n",
      "Epoch 956/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0015\n",
      "Epoch 957/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0016\n",
      "Epoch 958/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0019 - val_loss: 0.0016\n",
      "Epoch 959/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0015\n",
      "Epoch 960/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0015\n",
      "Epoch 961/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0019 - val_loss: 0.0015\n",
      "Epoch 962/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0015\n",
      "Epoch 963/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0018 - val_loss: 0.0015\n",
      "Epoch 964/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0016\n",
      "Epoch 965/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0016\n",
      "Epoch 966/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0016\n",
      "Epoch 967/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0016\n",
      "Epoch 968/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0016\n",
      "Epoch 969/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0025 - val_loss: 0.0016\n",
      "Epoch 970/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0016\n",
      "Epoch 971/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0016\n",
      "Epoch 972/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0016\n",
      "Epoch 973/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0018 - val_loss: 0.0016\n",
      "Epoch 974/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0016\n",
      "Epoch 975/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0016\n",
      "Epoch 976/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0016\n",
      "Epoch 977/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0016\n",
      "Epoch 978/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0016\n",
      "Epoch 979/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0022 - val_loss: 0.0016\n",
      "Epoch 980/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0016\n",
      "Epoch 981/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0019 - val_loss: 0.0016\n",
      "Epoch 982/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0016\n",
      "Epoch 983/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0019 - val_loss: 0.0016\n",
      "Epoch 984/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0016\n",
      "Epoch 985/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0016\n",
      "Epoch 986/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0024 - val_loss: 0.0016\n",
      "Epoch 987/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0019 - val_loss: 0.0016\n",
      "Epoch 988/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0016\n",
      "Epoch 989/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0016\n",
      "Epoch 990/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0016\n",
      "Epoch 991/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0017\n",
      "Epoch 992/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0019 - val_loss: 0.0017\n",
      "Epoch 993/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0017\n",
      "Epoch 994/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0020 - val_loss: 0.0017\n",
      "Epoch 995/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0019 - val_loss: 0.0017\n",
      "Epoch 996/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0017\n",
      "Epoch 997/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0017\n",
      "Epoch 998/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0023 - val_loss: 0.0017\n",
      "Epoch 999/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0017\n",
      "Epoch 1000/1000\n",
      "326/326 [==============================] - 0s - loss: 0.0021 - val_loss: 0.0017\n"
     ]
    }
   ],
   "source": [
    "#using the training data that will be divided again into 80% train, and 20% validation\n",
    "lstm_model = build_model_LSTM(0.0001, time_prediction)\n",
    "history = lstm_model.fit(trainX, trainY, epochs=1000, batch_size=batch_size, validation_split=0.2, shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['loss', 'val_loss'])\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYwAAAEWCAYAAAB1xKBvAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nkVj+P5U4i5gt38ZaAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1bbebd5ab00>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plot training and validation loss using MSE \n",
    "print(history.history.keys())\n",
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.title('LSTM-features: model missing accuracy')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.ylim((.0,0.1))\n",
    "plt.legend(['train', 'validation'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Prediction"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Do The Prediction and Re-Evaluate The Model Performance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/1000\n",
      "408/408 [==============================] - 3s - loss: 0.0161     \n",
      "Epoch 2/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0158     \n",
      "Epoch 3/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0153     \n",
      "Epoch 4/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0149     \n",
      "Epoch 5/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0145     \n",
      "Epoch 6/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0141     \n",
      "Epoch 7/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0138     \n",
      "Epoch 8/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0133     \n",
      "Epoch 9/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0130     \n",
      "Epoch 10/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0124     \n",
      "Epoch 11/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0122     \n",
      "Epoch 12/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0118     \n",
      "Epoch 13/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0115     \n",
      "Epoch 14/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0111     \n",
      "Epoch 15/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0109     \n",
      "Epoch 16/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0106     \n",
      "Epoch 17/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0101     \n",
      "Epoch 18/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0099     \n",
      "Epoch 19/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0095     \n",
      "Epoch 20/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0095     \n",
      "Epoch 21/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0091     \n",
      "Epoch 22/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0086     \n",
      "Epoch 23/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0084     \n",
      "Epoch 24/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0082     \n",
      "Epoch 25/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0077     \n",
      "Epoch 26/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0077     \n",
      "Epoch 27/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0076     \n",
      "Epoch 28/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0070     \n",
      "Epoch 29/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0067     \n",
      "Epoch 30/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0069     \n",
      "Epoch 31/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0065     \n",
      "Epoch 32/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0062     \n",
      "Epoch 33/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0059     \n",
      "Epoch 34/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0057     \n",
      "Epoch 35/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0055     \n",
      "Epoch 36/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0053     \n",
      "Epoch 37/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0054     \n",
      "Epoch 38/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0051     \n",
      "Epoch 39/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0049     \n",
      "Epoch 40/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0049     \n",
      "Epoch 41/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0049     \n",
      "Epoch 42/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0043     \n",
      "Epoch 43/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0048     \n",
      "Epoch 44/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0046     \n",
      "Epoch 45/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0043     \n",
      "Epoch 46/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0043     \n",
      "Epoch 47/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0045     \n",
      "Epoch 48/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0041     \n",
      "Epoch 49/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0040     \n",
      "Epoch 50/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0042     \n",
      "Epoch 51/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0041     \n",
      "Epoch 52/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0040     \n",
      "Epoch 53/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0041     \n",
      "Epoch 54/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0039     \n",
      "Epoch 55/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0041     \n",
      "Epoch 56/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0042     \n",
      "Epoch 57/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0041     \n",
      "Epoch 58/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0041     \n",
      "Epoch 59/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0043     \n",
      "Epoch 60/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0046     \n",
      "Epoch 61/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0043     \n",
      "Epoch 62/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0042     \n",
      "Epoch 63/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0038     \n",
      "Epoch 64/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0041     \n",
      "Epoch 65/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0045     \n",
      "Epoch 66/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0038     \n",
      "Epoch 67/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0041     \n",
      "Epoch 68/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0040     \n",
      "Epoch 69/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0039     \n",
      "Epoch 70/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0036     \n",
      "Epoch 71/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0041     \n",
      "Epoch 72/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0044     \n",
      "Epoch 73/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0040     \n",
      "Epoch 74/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0043     \n",
      "Epoch 75/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0042     \n",
      "Epoch 76/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0043     \n",
      "Epoch 77/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0039     \n",
      "Epoch 78/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0043     \n",
      "Epoch 79/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0040     \n",
      "Epoch 80/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0038     - ETA: 0s - loss: 0.003\n",
      "Epoch 81/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0038     \n",
      "Epoch 82/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0038     \n",
      "Epoch 83/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0039     \n",
      "Epoch 84/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0039     \n",
      "Epoch 85/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0038     \n",
      "Epoch 86/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0041     \n",
      "Epoch 87/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0040     \n",
      "Epoch 88/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0040     \n",
      "Epoch 89/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0038     \n",
      "Epoch 90/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0039     \n",
      "Epoch 91/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0041     \n",
      "Epoch 92/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0042     \n",
      "Epoch 93/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0039     \n",
      "Epoch 94/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0038     \n",
      "Epoch 95/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0039     \n",
      "Epoch 96/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0036     \n",
      "Epoch 97/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0039     \n",
      "Epoch 98/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0036     \n",
      "Epoch 99/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0038     \n",
      "Epoch 100/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0039     \n",
      "Epoch 101/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0039     \n",
      "Epoch 102/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0037     \n",
      "Epoch 103/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "408/408 [==============================] - 0s - loss: 0.0037     \n",
      "Epoch 104/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0042     \n",
      "Epoch 105/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0038     \n",
      "Epoch 106/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0038     \n",
      "Epoch 107/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0037     \n",
      "Epoch 108/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0036     \n",
      "Epoch 109/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0039     \n",
      "Epoch 110/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0039     \n",
      "Epoch 111/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     \n",
      "Epoch 112/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0039     \n",
      "Epoch 113/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0037     \n",
      "Epoch 114/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0037     \n",
      "Epoch 115/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0036     \n",
      "Epoch 116/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0037     \n",
      "Epoch 117/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0038     \n",
      "Epoch 118/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 119/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     \n",
      "Epoch 120/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0038     \n",
      "Epoch 121/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 122/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     \n",
      "Epoch 123/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0037     \n",
      "Epoch 124/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0036     \n",
      "Epoch 125/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0038     \n",
      "Epoch 126/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     \n",
      "Epoch 127/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0037     \n",
      "Epoch 128/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     \n",
      "Epoch 129/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0039     \n",
      "Epoch 130/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0037     \n",
      "Epoch 131/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0038     \n",
      "Epoch 132/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0037     \n",
      "Epoch 133/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0039     \n",
      "Epoch 134/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0037     \n",
      "Epoch 135/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0037     \n",
      "Epoch 136/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0036     \n",
      "Epoch 137/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 138/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0038     \n",
      "Epoch 139/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0036     \n",
      "Epoch 140/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0036     \n",
      "Epoch 141/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     \n",
      "Epoch 142/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0037     \n",
      "Epoch 143/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 144/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 145/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0036     \n",
      "Epoch 146/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0036     \n",
      "Epoch 147/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 148/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     \n",
      "Epoch 149/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0038     \n",
      "Epoch 150/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0036     \n",
      "Epoch 151/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0037     \n",
      "Epoch 152/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0037     \n",
      "Epoch 153/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 154/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0036     \n",
      "Epoch 155/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     \n",
      "Epoch 156/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     \n",
      "Epoch 157/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0036     \n",
      "Epoch 158/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     \n",
      "Epoch 159/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     \n",
      "Epoch 160/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0037     \n",
      "Epoch 161/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 162/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 163/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     \n",
      "Epoch 164/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0036     \n",
      "Epoch 165/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     \n",
      "Epoch 166/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 167/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 168/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 169/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0036     \n",
      "Epoch 170/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 171/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 172/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     \n",
      "Epoch 173/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 174/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     \n",
      "Epoch 175/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 176/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 177/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 178/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     \n",
      "Epoch 179/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 180/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 181/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 182/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     - ETA: 0s - loss: 0.003\n",
      "Epoch 183/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 184/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     \n",
      "Epoch 185/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 186/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 187/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 188/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 189/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 190/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 191/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 192/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 193/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 194/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 195/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 196/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 197/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 198/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 199/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 200/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 201/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0036     \n",
      "Epoch 202/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 203/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 204/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 205/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 206/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 207/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     \n",
      "Epoch 208/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 209/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 210/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0035     \n",
      "Epoch 211/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 212/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 213/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 214/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 215/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 216/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 217/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 218/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 219/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 220/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 221/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 222/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 223/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 224/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 225/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 226/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 227/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 228/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 229/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 230/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0034     \n",
      "Epoch 231/1000\n",
      "408/408 [==============================] - ETA: 0s - loss: 0.003 - 0s - loss: 0.0031     \n",
      "Epoch 232/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 233/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 234/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 235/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 236/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 237/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 238/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 239/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 240/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 241/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 242/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 243/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 244/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 245/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 246/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 247/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 248/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 249/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 250/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 251/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 252/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 253/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 254/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 255/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 256/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 257/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 258/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 259/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 260/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 261/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 262/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 263/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 264/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 265/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 266/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 267/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 268/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 269/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 270/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 271/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 272/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 273/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 274/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 275/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 276/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 277/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 278/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 279/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 280/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 281/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 282/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0033     \n",
      "Epoch 283/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 284/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 285/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 286/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 287/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 288/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 289/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 290/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 291/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 292/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 293/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0032     \n",
      "Epoch 294/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 295/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 296/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 297/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 298/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 299/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 300/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 301/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 302/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 303/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 304/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 305/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 306/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 307/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 308/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 309/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 310/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 311/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 312/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 313/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 314/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 315/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 316/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 317/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 318/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 319/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 320/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 321/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 322/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 323/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 324/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 325/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 326/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 327/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 328/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 329/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 330/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 331/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 332/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 333/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 334/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 335/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 336/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 337/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 338/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 339/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 340/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 341/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 342/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 343/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 344/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 345/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0031     \n",
      "Epoch 346/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 347/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 348/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 349/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 350/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 351/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 352/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 353/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0030     \n",
      "Epoch 354/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 355/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 356/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 357/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 358/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 359/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 360/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 361/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 362/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 363/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 364/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 365/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 366/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 367/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 368/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 369/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 370/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 371/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 372/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 373/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 374/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 375/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0029     \n",
      "Epoch 376/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 377/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 378/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 379/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 380/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 381/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 382/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 383/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 384/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 385/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 386/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 387/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 388/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 389/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 390/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 391/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 392/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 393/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 394/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 395/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 396/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 397/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 398/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 399/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 400/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 401/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 402/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 403/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 404/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0028     \n",
      "Epoch 405/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 406/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 407/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 408/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 409/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 410/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 411/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 412/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 413/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 414/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 415/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 416/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 417/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 418/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 419/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 420/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 421/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 422/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 423/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 424/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 425/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 426/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 427/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 428/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 429/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 430/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 431/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 432/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 433/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 434/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 435/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 436/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 437/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 438/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 439/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 440/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 441/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 442/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 443/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 444/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025        \n",
      "Epoch 445/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 446/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 447/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 448/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 449/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 450/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 451/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 452/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 453/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 454/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 455/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 456/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 457/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 458/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 459/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 460/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 461/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 462/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 463/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 464/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 465/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 466/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 467/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 468/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 469/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 470/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 471/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 472/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 473/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0027     \n",
      "Epoch 474/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 475/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 476/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 477/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 478/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 479/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 480/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 481/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 482/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 483/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 484/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 485/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 486/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 487/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 488/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 489/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 490/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 491/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 492/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 493/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023        \n",
      "Epoch 494/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 495/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 496/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 497/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 498/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 499/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 500/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0026     \n",
      "Epoch 501/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 502/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 503/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 504/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 505/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 506/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 507/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 508/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 509/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 510/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 511/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 512/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 513/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 514/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 515/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 516/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 517/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 518/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 519/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 520/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 521/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 522/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 523/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 524/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 525/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 526/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 527/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 528/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 529/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 530/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 531/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 532/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 533/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 534/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023        \n",
      "Epoch 535/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 536/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 537/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 538/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 539/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 540/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 541/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 542/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 543/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 544/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 545/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 546/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 547/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 548/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 549/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 550/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 551/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 552/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 553/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 554/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 555/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 556/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 557/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 558/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 559/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 560/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 561/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 562/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 563/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 564/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 565/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 566/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 567/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 568/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 569/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 570/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 571/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024        \n",
      "Epoch 572/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023        \n",
      "Epoch 573/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 574/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 575/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 576/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 577/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 578/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023        \n",
      "Epoch 579/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 580/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 581/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 582/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 583/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 584/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 585/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 586/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 587/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024        \n",
      "Epoch 588/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0025     \n",
      "Epoch 589/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 590/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 591/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 592/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 593/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 594/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 595/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 596/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 597/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 598/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 599/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 600/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023        \n",
      "Epoch 601/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 602/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 603/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 604/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 605/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 606/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 607/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 608/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 609/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 610/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 611/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023        \n",
      "Epoch 612/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 613/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024        \n",
      "Epoch 614/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 615/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 616/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 617/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 618/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 619/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 620/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 621/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 622/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 623/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024        \n",
      "Epoch 624/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 625/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 626/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0024     \n",
      "Epoch 627/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023        \n",
      "Epoch 628/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 629/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 630/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 631/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023        \n",
      "Epoch 632/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 633/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 634/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 635/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 636/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 637/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 638/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 639/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 640/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 641/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 642/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 643/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 644/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 645/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 646/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 647/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 648/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 649/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 650/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 651/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 652/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 653/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 654/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 655/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 656/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023        \n",
      "Epoch 657/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 658/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 659/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 660/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023        \n",
      "Epoch 661/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 662/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 663/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021         \n",
      "Epoch 664/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 665/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 666/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 667/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 668/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 669/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 670/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 671/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 672/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 673/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 674/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 675/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 676/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 677/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 678/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 679/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 680/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 681/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 682/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 683/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 684/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 685/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 686/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 687/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 688/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 689/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 690/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 691/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 692/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 693/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 694/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 695/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 696/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 697/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 698/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 699/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 700/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 701/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 702/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 703/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 704/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 705/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 706/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 707/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 708/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 709/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 710/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 711/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 712/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 713/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 714/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 715/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 716/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 717/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 718/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 719/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 720/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 721/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 722/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 723/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 724/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 725/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 726/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 727/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 728/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 729/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 730/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 731/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 732/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 733/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 734/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 735/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 736/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 737/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 738/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 739/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 740/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 741/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 742/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 743/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 744/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 745/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 746/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 747/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 748/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 749/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 750/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 751/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     e-0 - ETA: 0s - loss: 0.0022   \n",
      "Epoch 752/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 753/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 754/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 755/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019        \n",
      "Epoch 756/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 757/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019        \n",
      "Epoch 758/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 759/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 760/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 761/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 762/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 763/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 764/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 765/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 766/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 767/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 768/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 769/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019     \n",
      "Epoch 770/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 771/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 772/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 773/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 774/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 775/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 776/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 777/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 778/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 779/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 780/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 781/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 782/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019     \n",
      "Epoch 783/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 784/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 785/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 786/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 787/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 788/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023        \n",
      "Epoch 789/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0023     \n",
      "Epoch 790/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 791/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 792/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021         \n",
      "Epoch 793/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 794/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 795/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 796/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 797/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 798/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 799/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 800/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 801/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 802/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 803/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 804/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 805/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 806/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 807/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 808/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 809/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 810/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 811/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 812/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 813/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 814/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 815/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 816/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 817/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 818/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 819/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 820/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 821/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 822/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 823/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 824/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 825/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 826/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 827/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 828/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 829/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 830/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 831/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 832/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 833/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 834/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 835/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 836/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 837/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 838/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019        \n",
      "Epoch 839/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 840/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 841/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019     \n",
      "Epoch 842/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 843/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 844/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 845/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 846/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 847/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 848/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 849/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 850/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 851/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 852/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 853/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 854/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 855/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 856/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 857/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 858/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 859/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019     \n",
      "Epoch 860/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 861/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 862/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 863/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 864/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 865/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 866/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 867/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 868/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 869/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019        \n",
      "Epoch 870/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 871/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 872/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 873/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 874/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 875/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 876/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 877/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020         \n",
      "Epoch 878/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 879/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022        \n",
      "Epoch 880/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 881/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 882/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 883/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 884/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 885/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 886/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 887/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 888/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 889/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 890/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 891/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019        \n",
      "Epoch 892/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 893/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019        \n",
      "Epoch 894/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 895/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 896/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 897/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 898/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019        \n",
      "Epoch 899/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 900/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 901/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019        \n",
      "Epoch 902/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 903/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 904/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 905/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 906/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 907/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 908/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 909/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 910/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019     \n",
      "Epoch 911/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019     \n",
      "Epoch 912/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 913/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 914/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 915/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 916/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 917/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 918/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 919/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 920/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 921/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 922/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 923/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 924/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 925/1000\n",
      "408/408 [==============================] - ETA: 0s - loss: 0.0020    - 0s - loss: 0.0020     \n",
      "Epoch 926/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 927/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 928/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 929/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 930/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 931/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 932/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 933/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 934/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 935/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019     \n",
      "Epoch 936/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 937/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 938/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 939/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 940/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 941/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019     \n",
      "Epoch 942/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 943/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 944/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 945/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 946/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 947/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019        \n",
      "Epoch 948/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 949/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 950/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 951/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 952/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019        \n",
      "Epoch 953/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019        \n",
      "Epoch 954/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 955/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 956/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019        \n",
      "Epoch 957/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 958/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 959/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 960/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 961/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 962/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019        \n",
      "Epoch 963/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 964/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 965/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019     \n",
      "Epoch 966/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 967/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 968/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 969/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 970/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 971/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 972/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019     \n",
      "Epoch 973/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 974/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n",
      "Epoch 975/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 976/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 977/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019        \n",
      "Epoch 978/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 979/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 980/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 981/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0022     \n",
      "Epoch 982/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 983/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 984/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 985/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 986/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 987/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 988/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019     \n",
      "Epoch 989/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019        \n",
      "Epoch 990/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 991/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019        \n",
      "Epoch 992/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0019        \n",
      "Epoch 993/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 994/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021     \n",
      "Epoch 995/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 996/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 997/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020     \n",
      "Epoch 998/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 999/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0020        \n",
      "Epoch 1000/1000\n",
      "408/408 [==============================] - 0s - loss: 0.0021        \n"
     ]
    }
   ],
   "source": [
    "lstm_model = build_model_LSTM(0.0001, time_prediction)\n",
    "lstm_model.fit(trainX, trainY, epochs=1000, batch_size=batch_size)\n",
    "train_predict = lstm_model.predict(trainX)\n",
    "test_predict = lstm_model.predict(testX)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(508, 1) (408, 1) (408, 1, 41)\n"
     ]
    }
   ],
   "source": [
    "#Combine the train and test label\n",
    "y_label = np.append(trainY, testY, axis = 0)\n",
    "print(y_label.shape, train_predict.shape, trainX.shape)\n",
    "\n",
    "#get the class label from the regression format\n",
    "train_predict_class = regression_to_class(train_predict)\n",
    "test_predict_class = regression_to_class(test_predict)\n",
    "real_class = regression_to_class(y_label)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(408, 1) (100, 1) (407, 1)\n",
      "111 78 70 148\n",
      "Performance's Accuracy:  63.6363636364\n",
      "Sensitivity or True Positive Rate:  67.8899082569\n",
      "Specificity or True Negative Rate:  58.7301587302\n"
     ]
    }
   ],
   "source": [
    "#get the prediction, and see the accuracy of our model\n",
    "skip = 3\n",
    "print (train_predict.shape, test_predict.shape, real_class[1:len(train_predict)+skip-3].shape)\n",
    "# tn, fp, fn, tp = confusion_matrix(real_class[len(train_predict):-1,0], test_predict_class[1:,0]).ravel()\n",
    "tn, fp, fn, tp = confusion_matrix(real_class[1:len(train_predict)], train_predict_class[1:]).ravel()\n",
    "print (tn, fp, fn, tp)\n",
    "\n",
    "print (\"Performance's Accuracy: \",(tn+tp)* 100/ (tn + fp + fn + tp))\n",
    "print (\"Sensitivity or True Positive Rate: \", tp*100 / (tp+fn) )\n",
    "print (\"Specificity or True Negative Rate: \", tn*100 / (tn+fp) )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "42 (510, 1)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1bbece76d30>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#----------plot the prediction result\n",
    "# shift train predictions for plotting\n",
    "train_predict_plot = np.zeros((len_data, 1))\n",
    "print(time_prediction, train_predict_plot.shape)\n",
    "train_predict_plot[:train_predict_plot.shape[0], 0] = np.nan\n",
    "train_predict_plot[skip:len(train_predict)+skip] = train_predict\n",
    "\n",
    "# shift test predictions for plotting\n",
    "test_predict_plot = np.zeros((len_data, 1))\n",
    "test_predict_plot[:, :] = np.nan\n",
    "test_predict_plot[len(train_predict) + skip-1:] = test_predict\n",
    "\n",
    "# plot baseline and predictions\n",
    "plot_dataset = np.zeros((len_data, 1))\n",
    "plot_dataset[skip-1:] = y_label\n",
    "plt.plot(plot_dataset)\n",
    "plt.plot(train_predict_plot)\n",
    "plt.plot(test_predict_plot)\n",
    "plt.title('LSTM-features: Prediction Results')\n",
    "plt.ylabel('CPC')\n",
    "plt.xlabel('Dataset')\n",
    "plt.legend(['dataset', 'training', 'testing'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Forecasting"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Forecasting the CPC for n-days"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ 0.  0.  1.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.\n",
      "  0.  0.  0.]\n"
     ]
    }
   ],
   "source": [
    "#--------------forecasting the CPC using LSTM\n",
    "forecast, inp, predict = np.array([]), .0, .0\n",
    "adding = time_prediction*7\n",
    "for i in range(adding):\n",
    "    if i > 0: inp = np.concatenate([inp[-1, 0, 1:], predict[0] ]).reshape((1, 1,time_prediction-1) )\n",
    "    else: inp = np.concatenate([trainX[-1, 0, 1:], train_predict[-1]]).reshape((1, 1, time_prediction-1) )\n",
    "    predict = lstm_model.predict(inp)\n",
    "    forecast = np.append(forecast, predict)\n",
    "\n",
    "#get the forecasting's values\n",
    "forecast = forecast.reshape(int(forecast.shape[0]/14),14)\n",
    "forecast_result = regression_to_class_forecast(forecast[:,-1]).reshape(-1,1)\n",
    "print (forecast_result[:,-1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def regression_to_class_forecast(inp):\n",
    "    len_input = len(inp)\n",
    "    output = np.zeros_like(inp)\n",
    "    for i in range(1,len_input): output[i] = subtract(inp[i], inp[i-1])\n",
    "    return output "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 0.  1.]\n",
      " [ 1.  1.]\n",
      " [ 0.  1.]\n",
      " [ 0.  0.]\n",
      " [ 0.  1.]\n",
      " [ 0.  0.]\n",
      " [ 0.  1.]]\n"
     ]
    }
   ],
   "source": [
    "test_data = regression_to_class_forecast(testY)\n",
    "print (np.append(forecast_result[1:8], test_data[1:8], axis =1))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda root]",
   "language": "python",
   "name": "conda-root-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}