deep_learning_LSTM.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n",
      "/Users/juan/anaconda/lib/python3.6/importlib/_bootstrap.py:219: RuntimeWarning: compiletime version 3.5 of module 'tensorflow.python.framework.fast_tensor_util' does not match runtime version 3.6\n",
      "  return f(*args, **kwds)\n"
     ]
    }
   ],
   "source": [
    "import numpy as np # linear algebra\n",
    "import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n",
    "import csv\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "import seaborn as sns\n",
    "sns.set_style(\"whitegrid\")\n",
    "from datetime import datetime, date\n",
    "from os import path\n",
    "import os\n",
    "import sys\n",
    "\n",
    "import keras\n",
    "from keras.callbacks import EarlyStopping\n",
    "module_path = os.path.abspath(os.path.join('..'))\n",
    "if module_path not in sys.path:\n",
    "    sys.path.append(module_path)\n",
    "\n",
    "# Any results you write to the current directory are saved as output."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## loading data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_path='data/temporal'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(109490, 531)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X = np.load(path.join(data_path,'X.npy'))\n",
    "X.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(109490,)"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y = np.load(path.join(data_path,'y.npy'))\n",
    "y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import src.lib.balance_data as bl\n",
    "# X, y = bl.balance_data(X,y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train num 98541\n",
      "X_test num 10949\n"
     ]
    }
   ],
   "source": [
    "#split data into train and test dataset (80%/20%)\n",
    "from sklearn.model_selection import train_test_split\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.1, random_state=421)\n",
    "print('X_train num',X_train.shape[0])\n",
    "print('X_test num',X_test.shape[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Import `StandardScaler` from `sklearn.preprocessing`\n",
    "from sklearn.preprocessing import MinMaxScaler,StandardScaler #StandardScaler\n",
    "#Define the scaler \n",
    "# # scaler = MinMaxScaler(feature_range=(-1, 1))\n",
    "# scaler = StandardScaler().fit(X_train)\n",
    "# # Scale the train set\n",
    "# X_train = scaler.transform(X_train)\n",
    "# # Scale the test set\n",
    "# X_test = scaler.transform(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(98541, 518)\n",
      "(10949, 518)\n",
      "(98541, 13)\n",
      "(10949, 13)\n"
     ]
    }
   ],
   "source": [
    "time_feature_beg=13  # temporal feature start position\n",
    "X_train_time = X_train[:,time_feature_beg:]\n",
    "X_test_time = X_test[:,time_feature_beg:]\n",
    "X_train_aux = X_train[:,0:time_feature_beg]\n",
    "X_test_aux = X_test[:,0:time_feature_beg]\n",
    "print(X_train_time.shape)\n",
    "print(X_test_time.shape)\n",
    "print(X_train_aux.shape)\n",
    "print(X_test_aux.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "def case_contro_ratio(y):\n",
    "    num_of_case = y[y==1].shape[0]\n",
    "    num_of_control = y[y==0].shape[0]\n",
    "    ratio = num_of_control/num_of_case\n",
    "    return ratio, num_of_case, num_of_control"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train ratio 10.16, case 8830, control 89711 \n",
      "test cohort ratio 10.02 case 994, control 9955\n"
     ]
    }
   ],
   "source": [
    "print('train ratio %.2f, case %d, control %d ' % (case_contro_ratio(y_train)[0], case_contro_ratio(y_train)[1],case_contro_ratio(y_train)[2]))\n",
    "# print('valid cohort ratio %.2f, case %d, control %d'% (case_contro_ratio(y_valid)[0], case_contro_ratio(y_valid)[1],case_contro_ratio(y_valid)[2]))\n",
    "print('test cohort ratio %.2f case %d, control %d' % (case_contro_ratio(y_test)[0],case_contro_ratio(y_test)[1],case_contro_ratio(y_test)[2]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "# reshape the temporal features\n",
    "def reshape_time_series(X, time_steps, feature_size):\n",
    "     \"\"\"\n",
    "    :param X: Input features\n",
    "    :param time_steps: time intervals, or number of windows (e.g. 7 years in our data )\n",
    "    :param feature_size: size of distinct feature\n",
    "    :return:reshaped features\n",
    "    \"\"\"\n",
    "        \n",
    "    X_new = np.zeros(shape=(X.shape[0],X.shape[1]))\n",
    "    idx=0\n",
    "    for i in range(time_steps):\n",
    "        for j in range(feature_size):\n",
    "            X_new[:,idx] = X[:,i+j*time_steps]\n",
    "            idx+=1\n",
    "    return X_new"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(98541, 518)"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "X_train_time_new = reshape_time_series(X_train_time,time_steps=7,feature_size=74) #74 distinct features\n",
    "X_train_time_new.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10949, 518)"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# X_test_time_new = reshape_time_series(X_test_time,7,14)\n",
    "X_test_time_new = reshape_time_series(X_test_time,7,74)\n",
    "X_test_time_new.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['2000', '2001', '2002', '2003', '2004', '2005']"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "feature_list = [str(i) for i in range(2000,2006)] # years\n",
    "feature_list"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## standarscale the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(98541, 7, 74)\n",
      "(10949, 7, 74)\n"
     ]
    }
   ],
   "source": [
    "# reshape input to be 3D [samples, timesteps, features]\n",
    "timesteps=7\n",
    "X_train_time = X_train_time_new.reshape((X_train_time_new.shape[0], timesteps, int(X_train_time_new.shape[1]/timesteps)))\n",
    "X_test_time = X_test_time_new.reshape((X_test_time_new.shape[0], timesteps, int(X_test_time_new.shape[1]/timesteps)))\n",
    "print(X_train_time.shape)\n",
    "print(X_test_time.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[[119. , 146. , 119. , ...,   0. ,   0. ,   0. ],\n",
       "        [119. , 146. , 119. , ...,   0. ,   0. ,   0. ],\n",
       "        [119. , 146. , 119. , ...,   0. ,   0. ,   0. ],\n",
       "        ...,\n",
       "        [119. , 146. , 119. , ...,   0. ,   0. ,   0. ],\n",
       "        [119. , 146. , 119. , ...,   0. ,   0. ,   0. ],\n",
       "        [119. , 146. ,  86. , ...,   0. ,   0. ,   0. ]],\n",
       "\n",
       "       [[ 90. , 136. ,  90. , ...,   0. ,   0. ,   0. ],\n",
       "        [ 90. , 136. ,  90. , ...,   0. ,   0. ,   0. ],\n",
       "        [ 90. , 136. ,  90. , ...,   0. ,   0. ,   0. ],\n",
       "        ...,\n",
       "        [ 90. , 136. ,  90. , ...,   0. ,   0. ,   0. ],\n",
       "        [ 90. , 140. ,  85. , ...,   0. ,   0. ,   0. ],\n",
       "        [ 94. , 142. ,  94. , ...,   0. ,   0. ,   0. ]],\n",
       "\n",
       "       [[ 68. , 108. ,  68. , ...,   0. ,   0. ,   0. ],\n",
       "        [ 68. , 108. ,  68. , ...,   0. ,   0. ,   0. ],\n",
       "        [ 68. , 108. ,  68. , ...,   0. ,   0. ,   0. ],\n",
       "        ...,\n",
       "        [ 84. , 122. ,  75. , ...,   0. ,   0. ,   0. ],\n",
       "        [ 66. ,  98. ,  66. , ...,   0. ,   0. ,   0. ],\n",
       "        [ 66. ,  98. ,  66. , ...,   0. ,   0. ,   0. ]],\n",
       "\n",
       "       ...,\n",
       "\n",
       "       [[ 54. ,  98. ,  54. , ...,   0. ,   0. ,   0. ],\n",
       "        [ 54. ,  98. ,  54. , ...,   0. ,   0. ,   0. ],\n",
       "        [ 54. ,  98. ,  54. , ...,   0. ,   0. ,   0. ],\n",
       "        ...,\n",
       "        [ 54. ,  98. ,  54. , ...,   0. ,   0. ,   0. ],\n",
       "        [ 80. , 133. ,  68.5, ...,   0. ,   1. ,   0. ],\n",
       "        [ 72. , 123. ,  70. , ...,   0. ,   1. ,   0. ]],\n",
       "\n",
       "       [[100. , 150. , 100. , ...,   0. ,   0. ,   0. ],\n",
       "        [100. , 150. ,  80. , ...,   0. ,   1. ,   0. ],\n",
       "        [ 74. , 136. ,  74. , ...,   0. ,   1. ,   0. ],\n",
       "        ...,\n",
       "        [ 82. , 115. ,  82. , ...,   0. ,   1. ,   0. ],\n",
       "        [ 82. , 112. ,  82. , ...,   0. ,   1. ,   0. ],\n",
       "        [ 82. , 112. ,  82. , ...,   0. ,   1. ,   0. ]],\n",
       "\n",
       "       [[ 88. , 150. ,  88. , ...,   0. ,   0. ,   0. ],\n",
       "        [ 88. , 150. ,  88. , ...,   0. ,   0. ,   0. ],\n",
       "        [ 88. , 150. ,  88. , ...,   0. ,   0. ,   0. ],\n",
       "        ...,\n",
       "        [ 88. , 150. ,  88. , ...,   0. ,   0. ,   0. ],\n",
       "        [ 88. , 150. ,  88. , ...,   0. ,   0. ,   0. ],\n",
       "        [ 88. , 150. ,  88. , ...,   0. ,   0. ,   0. ]]])"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_time.iloc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Import `Sequential` from `keras.models`\n",
    "from keras.models import Sequential\n",
    "# Import `Dense` from `keras.layers`\n",
    "from keras.layers import Dense, Dropout\n",
    "from keras.layers import LSTM\n",
    "from keras.layers import Input\n",
    "from keras.models import Model\n",
    "import keras"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "main_input = Input(shape=(X_train_time.shape[1], X_train_time.shape[2]), name='main_input')\n",
    "lstm_out = LSTM(84, dropout=0.1, recurrent_dropout=0.1)(main_input)\n",
    "\n",
    "# lstm_out = LSTM(84)(main_input)\n",
    "auxiliary_output = Dense(1, activation='sigmoid', name='aux_output')(lstm_out)\n",
    "\n",
    "auxiliary_input = Input(shape=(time_feature_beg,), name='aux_input')\n",
    "x = keras.layers.concatenate([lstm_out, auxiliary_input])\n",
    "\n",
    "# We stack a deep densely-connected network on top\n",
    "x = Dense(8, activation='relu')(x)\n",
    "x= Dropout(0.1)(x)\n",
    "x = Dense(8, activation='relu')(x)\n",
    "x= Dropout(0.1)(x)\n",
    "\n",
    "\n",
    "# And finally we add the main logistic regression layer\n",
    "main_output = Dense(1, activation='sigmoid', name='main_output')(x)\n",
    "#This defines a model with two inputs and two outputs:\n",
    "model = Model(inputs=[main_input, auxiliary_input], outputs=[main_output, auxiliary_output])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "main_input (InputLayer)         (None, 7, 74)        0                                            \n",
      "__________________________________________________________________________________________________\n",
      "lstm_1 (LSTM)                   (None, 84)           53424       main_input[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "aux_input (InputLayer)          (None, 13)           0                                            \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_1 (Concatenate)     (None, 97)           0           lstm_1[0][0]                     \n",
      "                                                                 aux_input[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dense_1 (Dense)                 (None, 8)            784         concatenate_1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "dropout_1 (Dropout)             (None, 8)            0           dense_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dense_2 (Dense)                 (None, 8)            72          dropout_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_2 (Dropout)             (None, 8)            0           dense_2[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "main_output (Dense)             (None, 1)            9           dropout_2[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "aux_output (Dense)              (None, 1)            85          lstm_1[0][0]                     \n",
      "==================================================================================================\n",
      "Total params: 54,374\n",
      "Trainable params: 54,374\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[array([[ 0.07439198, -0.01115449, -0.03859836, ..., -0.06209642,\n",
       "         -0.04029816,  0.03973766],\n",
       "        [ 0.02970695, -0.03725597, -0.04815671, ...,  0.02228572,\n",
       "         -0.07461698, -0.08958131],\n",
       "        [-0.00820258,  0.09913025, -0.04388729, ..., -0.11732537,\n",
       "         -0.04989307,  0.08130426],\n",
       "        ..., \n",
       "        [ 0.10048299,  0.09385218,  0.08753134, ..., -0.08879401,\n",
       "          0.04840863,  0.10997284],\n",
       "        [ 0.05333755, -0.04175213,  0.01096482, ..., -0.01815384,\n",
       "          0.0977878 ,  0.03929863],\n",
       "        [-0.08718094,  0.05484312,  0.00038746, ...,  0.04861846,\n",
       "          0.00084905, -0.04558536]], dtype=float32),\n",
       " array([[ 0.01245363, -0.09115876,  0.00236749, ...,  0.0482492 ,\n",
       "          0.04815456, -0.11101308],\n",
       "        [ 0.01130434, -0.04769856, -0.00449701, ..., -0.0264445 ,\n",
       "          0.08411942,  0.03493295],\n",
       "        [-0.01161938,  0.07425027, -0.0174167 , ..., -0.00021368,\n",
       "          0.02905249,  0.04761047],\n",
       "        ..., \n",
       "        [-0.07629018, -0.03159355, -0.00432858, ...,  0.00140656,\n",
       "          0.03539425,  0.06719942],\n",
       "        [-0.01003201, -0.00657078, -0.00761522, ...,  0.0372538 ,\n",
       "          0.08026908, -0.00271888],\n",
       "        [-0.02502786,  0.02913669, -0.01664479, ..., -0.05754557,\n",
       "          0.03394047,  0.01302586]], dtype=float32),\n",
       " array([ 0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
       "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
       "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
       "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
       "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
       "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
       "         0.,  0.,  0.,  0.,  0.,  0.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,\n",
       "         1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,\n",
       "         1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,\n",
       "         1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,\n",
       "         1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,\n",
       "         1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,\n",
       "         1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  0.,\n",
       "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
       "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
       "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
       "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
       "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
       "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
       "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
       "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
       "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
       "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
       "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
       "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
       "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.], dtype=float32),\n",
       " array([[-0.12161536, -0.15127501,  0.21721144,  0.01333417, -0.07918553,\n",
       "          0.18624757,  0.02508543,  0.13452165],\n",
       "        [-0.09761061, -0.21354429,  0.06814902, -0.09020586,  0.21781676,\n",
       "          0.03634195, -0.09333107, -0.02203974],\n",
       "        [ 0.12118472,  0.02156283, -0.02957004, -0.057936  ,  0.12309365,\n",
       "         -0.1700422 ,  0.23385708, -0.08918627],\n",
       "        [-0.18392313,  0.01850636,  0.23301394,  0.1998478 ,  0.02675249,\n",
       "          0.10887481, -0.12899902,  0.07711972],\n",
       "        [ 0.22338755, -0.1339854 ,  0.20583315, -0.17323358,  0.13123964,\n",
       "          0.23858778,  0.21527128, -0.01944274],\n",
       "        [-0.02897607,  0.1571819 ,  0.08506249, -0.1062724 ,  0.15431847,\n",
       "          0.16813464, -0.08949351, -0.0013672 ],\n",
       "        [ 0.08500086, -0.17475495, -0.04101709, -0.00846265,  0.00606097,\n",
       "         -0.13235083, -0.06453937, -0.07306638],\n",
       "        [-0.11647289,  0.04424785,  0.23318417,  0.18300264,  0.22827743,\n",
       "          0.11129703,  0.12393348,  0.11356817],\n",
       "        [ 0.15745698,  0.01487784,  0.18297036,  0.18262561, -0.07866268,\n",
       "          0.15455808,  0.08858214,  0.08731194],\n",
       "        [-0.21841627, -0.16199024,  0.17179976,  0.09785347,  0.06183283,\n",
       "         -0.06167628, -0.12493975,  0.06106617],\n",
       "        [ 0.13185351,  0.04735367, -0.05594017, -0.01067364,  0.20593105,\n",
       "          0.08797498, -0.0575754 ,  0.13678427],\n",
       "        [ 0.05353911, -0.22156519, -0.02499385, -0.02022628,  0.00094768,\n",
       "         -0.17738238, -0.22352991,  0.21236362],\n",
       "        [ 0.18993907, -0.22820607, -0.02508277,  0.18028258, -0.16951826,\n",
       "          0.05106555,  0.00635757, -0.15997663],\n",
       "        [-0.23876446,  0.16122071, -0.22480626,  0.14542218, -0.02430521,\n",
       "          0.13305499,  0.21003403,  0.08916546],\n",
       "        [-0.17409253, -0.00194089, -0.23129867, -0.21290956, -0.06915769,\n",
       "         -0.08080955, -0.19465216,  0.20603611],\n",
       "        [-0.15098786, -0.14200839, -0.2372354 ,  0.04128142,  0.22666873,\n",
       "         -0.23887025,  0.07037865, -0.10932842],\n",
       "        [-0.06942059,  0.21629231, -0.11078379,  0.07228734,  0.17600133,\n",
       "         -0.06101157,  0.15419172,  0.04136674],\n",
       "        [ 0.01851831, -0.18423481,  0.01659231,  0.00644083, -0.08449483,\n",
       "         -0.18399031, -0.08423471, -0.23331109],\n",
       "        [-0.16857697, -0.05886926,  0.23293968,  0.09559055, -0.02452104,\n",
       "          0.02987508, -0.15074235,  0.23198862],\n",
       "        [-0.07727405,  0.22406997, -0.04531839,  0.0020403 ,  0.14709292,\n",
       "         -0.07862011, -0.1992048 ,  0.21237747],\n",
       "        [ 0.09513147,  0.21987046, -0.06313775,  0.14894046, -0.06581077,\n",
       "         -0.21978895, -0.20132352,  0.13317831],\n",
       "        [ 0.17696099,  0.01985337, -0.0048456 , -0.22495702, -0.21795896,\n",
       "          0.15995462,  0.04392533,  0.0263484 ],\n",
       "        [-0.0616512 , -0.08503045, -0.19442253, -0.16277206, -0.16132668,\n",
       "         -0.00892225,  0.05230163,  0.14566301],\n",
       "        [-0.0811616 ,  0.01189922,  0.08258294,  0.22334327,  0.15538196,\n",
       "          0.11264814,  0.1267315 ,  0.11934568],\n",
       "        [ 0.09296684, -0.08364062, -0.21999195, -0.03831851,  0.14674182,\n",
       "         -0.10737853,  0.23045854, -0.05960812],\n",
       "        [-0.20635179, -0.09337485,  0.14842422, -0.15755966, -0.08066718,\n",
       "          0.16038816,  0.21270125,  0.06871502],\n",
       "        [-0.2275458 , -0.07195923,  0.15454759,  0.20465414, -0.16432354,\n",
       "          0.09355654, -0.06396945, -0.02379382],\n",
       "        [ 0.22409351, -0.15504183,  0.03958245, -0.06952091,  0.06401138,\n",
       "         -0.0811376 ,  0.19676591, -0.22678609],\n",
       "        [-0.09382549,  0.20697181, -0.06542003, -0.12954724, -0.16122055,\n",
       "          0.08606197,  0.16184942, -0.02856766],\n",
       "        [ 0.02843343, -0.17347643,  0.12230919, -0.12161239, -0.02259998,\n",
       "         -0.18806788,  0.11610846,  0.01912744],\n",
       "        [ 0.19597097, -0.07781965, -0.17086177,  0.14181678,  0.03891467,\n",
       "          0.04455058, -0.15934485,  0.22328307],\n",
       "        [ 0.16707177, -0.22843546,  0.16771112,  0.19257407, -0.15349339,\n",
       "          0.09380917, -0.09762225,  0.19514026],\n",
       "        [-0.17908858, -0.0925705 ,  0.08624439, -0.08229752,  0.17861845,\n",
       "         -0.14694148,  0.20605491,  0.06341393],\n",
       "        [ 0.06868111, -0.02587964, -0.05686265,  0.0682797 , -0.10931127,\n",
       "         -0.14169362, -0.13775218, -0.01375291],\n",
       "        [-0.09826876, -0.07803987,  0.08408026, -0.12367634, -0.03019851,\n",
       "          0.17199092, -0.17507371, -0.00516357],\n",
       "        [ 0.12405111,  0.19032381, -0.14833695,  0.17616735, -0.17403069,\n",
       "          0.10920121,  0.12464614, -0.12291616],\n",
       "        [-0.04374988,  0.0279894 ,  0.23142599,  0.2327341 ,  0.03813188,\n",
       "         -0.12770689,  0.18641935, -0.05324464],\n",
       "        [-0.05217123,  0.18837546,  0.04367961, -0.23524487,  0.07133116,\n",
       "          0.16190691, -0.12007912,  0.11651574],\n",
       "        [ 0.1681359 ,  0.05564244,  0.00393616, -0.16984177, -0.19412485,\n",
       "          0.22551532,  0.16950311, -0.08131309],\n",
       "        [-0.23686296,  0.0745682 , -0.20672622, -0.01467414,  0.12685849,\n",
       "          0.04927592,  0.19282244, -0.03593417],\n",
       "        [-0.20430711, -0.17202835,  0.13253902,  0.06131209, -0.22009979,\n",
       "          0.04653503, -0.06046797, -0.04936306],\n",
       "        [-0.11299312, -0.19888321,  0.20765497, -0.04440559,  0.02413698,\n",
       "         -0.08904669, -0.03837529,  0.05812688],\n",
       "        [ 0.07940267,  0.17045231,  0.18201379, -0.1983286 ,  0.18572165,\n",
       "         -0.08221477,  0.01027691, -0.18274872],\n",
       "        [-0.16329083, -0.17440957,  0.00924642,  0.21892692,  0.21159564,\n",
       "         -0.1699664 ,  0.14330204, -0.14936927],\n",
       "        [-0.17896177, -0.0416411 ,  0.05413167, -0.08172075, -0.10116161,\n",
       "         -0.22497861,  0.1346864 , -0.07289961],\n",
       "        [-0.19769859,  0.03055774, -0.01711874,  0.22218026,  0.11356916,\n",
       "         -0.07748966,  0.1845697 ,  0.03675036],\n",
       "        [-0.19614246,  0.02613287,  0.01879485, -0.18034129, -0.00754239,\n",
       "          0.23508267, -0.09150667, -0.22892725],\n",
       "        [ 0.17734466,  0.01236705,  0.09924255,  0.20485435, -0.20968376,\n",
       "         -0.10706779, -0.17152625, -0.23303069],\n",
       "        [ 0.10420345, -0.03868915,  0.09993763, -0.16886513, -0.01365624,\n",
       "          0.02694125,  0.01288839,  0.22717713],\n",
       "        [ 0.19772498, -0.17369066,  0.05021004, -0.01451519, -0.10539573,\n",
       "          0.04388474, -0.23253748,  0.09169333],\n",
       "        [ 0.18795185,  0.0299878 ,  0.03433944,  0.14346348, -0.17407492,\n",
       "          0.03626387,  0.0337026 , -0.04407737],\n",
       "        [ 0.23640229, -0.11425859, -0.00721873, -0.04462883, -0.11163481,\n",
       "          0.13265644,  0.00216049, -0.09357107],\n",
       "        [-0.02530737,  0.04315738, -0.03969137,  0.03765552,  0.09584384,\n",
       "         -0.2260984 ,  0.16795237, -0.11584739],\n",
       "        [-0.1479775 ,  0.15157135, -0.21247402, -0.16803913,  0.07148071,\n",
       "          0.16663618,  0.2230172 ,  0.17990072],\n",
       "        [-0.02915104,  0.22523166,  0.22843097,  0.11432351,  0.09358339,\n",
       "         -0.07704955,  0.00558582, -0.12816477],\n",
       "        [ 0.06693383, -0.17838363,  0.03896214,  0.02690317, -0.00186583,\n",
       "         -0.22681196,  0.15756695, -0.23275103],\n",
       "        [ 0.14229901,  0.2217287 , -0.1132175 , -0.07439254,  0.03358047,\n",
       "          0.04692708,  0.02789707, -0.19657727],\n",
       "        [ 0.18320723,  0.13831054, -0.15042272,  0.07263245, -0.19090647,\n",
       "          0.19109847, -0.16038549,  0.12809409],\n",
       "        [-0.12819332,  0.11343418, -0.04472476, -0.14772958,  0.16235761,\n",
       "          0.12672831,  0.23570259,  0.05117734],\n",
       "        [ 0.12794922, -0.05757958, -0.16307442,  0.00757971,  0.09489347,\n",
       "          0.10401924,  0.01534636,  0.03892772],\n",
       "        [-0.04753439, -0.2041885 ,  0.18853979,  0.18971886,  0.0741183 ,\n",
       "         -0.02766779,  0.16378091,  0.07015552],\n",
       "        [ 0.00750871,  0.14250754,  0.0053045 ,  0.16453128,  0.18470974,\n",
       "         -0.22043616,  0.10090573, -0.07982807],\n",
       "        [ 0.13082834, -0.06975372,  0.1179132 , -0.09702393, -0.18814424,\n",
       "          0.0118062 , -0.07601947, -0.09559244],\n",
       "        [-0.11048241,  0.14288808,  0.20011078,  0.18907322, -0.16239375,\n",
       "         -0.14377467,  0.09210469, -0.01290765],\n",
       "        [ 0.2038172 , -0.08426742, -0.08426964,  0.00556758,  0.12594955,\n",
       "          0.02806343, -0.0610681 , -0.02827306],\n",
       "        [ 0.06481563,  0.18885981,  0.17721801,  0.14555438,  0.08634852,\n",
       "         -0.02132168, -0.13665676, -0.11226703],\n",
       "        [-0.01289107,  0.12525366, -0.17318264,  0.10422914,  0.16436081,\n",
       "          0.06511341,  0.07630609, -0.01618513],\n",
       "        [ 0.04130791, -0.00464778,  0.01687936,  0.1693645 ,  0.10450466,\n",
       "         -0.12154212, -0.14642462,  0.02285542],\n",
       "        [-0.21988925, -0.05991526,  0.12102537, -0.10662849,  0.0237873 ,\n",
       "         -0.0615415 ,  0.21592022, -0.01532443],\n",
       "        [ 0.14498012,  0.21566938, -0.0133356 ,  0.12137456,  0.17188622,\n",
       "         -0.20302886, -0.16625981, -0.10050768],\n",
       "        [-0.1964329 ,  0.00107928,  0.1580147 ,  0.12423293, -0.13053754,\n",
       "          0.07188599, -0.09355152, -0.22489905],\n",
       "        [-0.21692163, -0.1294283 ,  0.17785646,  0.15834866, -0.21603027,\n",
       "         -0.06011559, -0.12105643,  0.14045073],\n",
       "        [ 0.04338674,  0.19461061, -0.11445813, -0.02492473, -0.05693704,\n",
       "         -0.04052135,  0.05710529, -0.11231086],\n",
       "        [-0.18663365,  0.00653476,  0.1143771 , -0.11698668, -0.16526294,\n",
       "         -0.20259064,  0.21841599, -0.05289897],\n",
       "        [ 0.10715728,  0.21302544, -0.04290451,  0.10860471, -0.21303454,\n",
       "          0.1795872 ,  0.21628298,  0.18132739],\n",
       "        [ 0.23690341, -0.01038679,  0.04018055, -0.2095211 ,  0.07283573,\n",
       "         -0.15719451, -0.17018981,  0.12919153],\n",
       "        [ 0.23224999, -0.04349063, -0.21955231, -0.18963358, -0.09432372,\n",
       "          0.10382839,  0.21565486,  0.010124  ],\n",
       "        [-0.18687125, -0.1152399 , -0.08950308,  0.0692593 ,  0.10443802,\n",
       "         -0.07317592, -0.22898892, -0.04548892],\n",
       "        [-0.06972443, -0.21996552, -0.00959618, -0.08459553,  0.01445781,\n",
       "         -0.1313391 ,  0.21939199, -0.09311108],\n",
       "        [-0.13663852,  0.15921034, -0.22198562,  0.23534296,  0.10989685,\n",
       "          0.12794305,  0.11328043,  0.10586621],\n",
       "        [ 0.08853681,  0.23268645, -0.06469023, -0.08027096, -0.11088461,\n",
       "         -0.08525477,  0.10831772, -0.22532752],\n",
       "        [-0.18786731, -0.13264823,  0.0085942 , -0.04621768,  0.0036528 ,\n",
       "          0.05907302, -0.22232896, -0.18095128],\n",
       "        [-0.14563316, -0.06858712, -0.02165066,  0.17470314,  0.19871838,\n",
       "          0.15270825,  0.04040031, -0.03726199],\n",
       "        [-0.00374232,  0.07184719, -0.09781767,  0.16979159, -0.00639951,\n",
       "          0.22449224, -0.08142844,  0.0679002 ],\n",
       "        [-0.13364777,  0.07512404, -0.16975279, -0.10022436, -0.19527173,\n",
       "         -0.05599847,  0.0701652 ,  0.1299613 ],\n",
       "        [-0.0392344 , -0.23764661,  0.16432326, -0.12321458, -0.10758165,\n",
       "          0.09569241,  0.19196807,  0.07715039],\n",
       "        [-0.19304296,  0.02470718, -0.18665747,  0.19304936,  0.0497198 ,\n",
       "          0.06523784,  0.18443795,  0.0200312 ],\n",
       "        [ 0.10111164,  0.01399069, -0.13426265, -0.00312059, -0.00603122,\n",
       "          0.10383864,  0.11820917,  0.05232061],\n",
       "        [ 0.04823653,  0.06900643, -0.22333819, -0.23198521,  0.04169582,\n",
       "          0.16070442,  0.16268818, -0.07840838],\n",
       "        [-0.10641295,  0.11331286,  0.1542054 ,  0.23343118,  0.14299251,\n",
       "          0.0922787 ,  0.21962424,  0.09085096],\n",
       "        [ 0.17083348, -0.10431281, -0.02139954,  0.02785109,  0.12398537,\n",
       "          0.19288664,  0.17272474, -0.11725996],\n",
       "        [-0.18483694,  0.19796155, -0.13371998,  0.0935107 ,  0.02074744,\n",
       "          0.22474791,  0.00490607,  0.0181091 ],\n",
       "        [-0.21012528,  0.0786045 , -0.04555315, -0.02682544, -0.00368078,\n",
       "         -0.06155363, -0.09921251,  0.19527788],\n",
       "        [ 0.03548197,  0.17103662, -0.17325512, -0.03119594,  0.06151019,\n",
       "          0.13550659,  0.09797074,  0.09934177],\n",
       "        [ 0.11489047,  0.04177599, -0.0843524 ,  0.15139236, -0.13149065,\n",
       "         -0.13257891, -0.07310382, -0.1192672 ],\n",
       "        [-0.02646495, -0.23784597,  0.20975329, -0.02378401,  0.22018795,\n",
       "          0.15051596,  0.2203746 , -0.16841824],\n",
       "        [-0.05534135, -0.08271432,  0.07744868,  0.19341113,  0.02297036,\n",
       "          0.1508496 ,  0.1521007 , -0.08189082]], dtype=float32),\n",
       " array([ 0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.], dtype=float32),\n",
       " array([[-0.16593009, -0.27418223, -0.05675423, -0.22963426,  0.43439788,\n",
       "         -0.27865994, -0.12531585, -0.55286992],\n",
       "        [-0.35687646, -0.4971517 ,  0.45055169,  0.58144027,  0.06480515,\n",
       "          0.07423359, -0.31474146, -0.60141933],\n",
       "        [ 0.0119496 ,  0.16972876, -0.12301618, -0.11384398,  0.51907772,\n",
       "          0.01694518, -0.40729177,  0.02381521],\n",
       "        [ 0.29601389,  0.57817394, -0.16593739, -0.44425043, -0.20964864,\n",
       "          0.5259257 , -0.39135194,  0.42416173],\n",
       "        [ 0.52187127,  0.27008456,  0.44082719, -0.18255538,  0.16360897,\n",
       "         -0.51827103,  0.05975443,  0.59352356],\n",
       "        [-0.36435375,  0.35460132, -0.42045739,  0.09495288,  0.14480406,\n",
       "         -0.51083022, -0.02692693,  0.46308762],\n",
       "        [ 0.26812381, -0.02428699, -0.07349676, -0.47904393, -0.45604336,\n",
       "         -0.54231709,  0.20870316, -0.52204794],\n",
       "        [-0.54640746, -0.6107282 , -0.40949783,  0.08738333,  0.16488695,\n",
       "         -0.4202314 , -0.1641264 ,  0.35468495]], dtype=float32),\n",
       " array([ 0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.], dtype=float32),\n",
       " array([[-0.71932614],\n",
       "        [ 0.29694319],\n",
       "        [ 0.30320251],\n",
       "        [ 0.37641311],\n",
       "        [ 0.56708622],\n",
       "        [ 0.55494523],\n",
       "        [-0.6869005 ],\n",
       "        [ 0.31019127]], dtype=float32),\n",
       " array([ 0.], dtype=float32),\n",
       " array([[ 0.20762318],\n",
       "        [-0.0244585 ],\n",
       "        [ 0.18877718],\n",
       "        [-0.22767502],\n",
       "        [-0.166069  ],\n",
       "        [-0.14042522],\n",
       "        [-0.0149366 ],\n",
       "        [-0.22771138],\n",
       "        [ 0.05422434],\n",
       "        [ 0.19591093],\n",
       "        [-0.22129278],\n",
       "        [-0.03517917],\n",
       "        [-0.1230709 ],\n",
       "        [ 0.181214  ],\n",
       "        [-0.11803712],\n",
       "        [ 0.25033388],\n",
       "        [ 0.25388542],\n",
       "        [-0.14791325],\n",
       "        [ 0.12874544],\n",
       "        [-0.01139301],\n",
       "        [-0.18350548],\n",
       "        [-0.05092005],\n",
       "        [-0.01529759],\n",
       "        [ 0.02402624],\n",
       "        [-0.03689839],\n",
       "        [ 0.0474768 ],\n",
       "        [ 0.14632294],\n",
       "        [-0.08187549],\n",
       "        [ 0.19999844],\n",
       "        [-0.1434333 ],\n",
       "        [-0.21717952],\n",
       "        [-0.16060913],\n",
       "        [-0.1622135 ],\n",
       "        [ 0.19254875],\n",
       "        [ 0.05979887],\n",
       "        [ 0.07976651],\n",
       "        [ 0.12950027],\n",
       "        [ 0.25500056],\n",
       "        [-0.02316627],\n",
       "        [-0.23374502],\n",
       "        [-0.10070156],\n",
       "        [-0.08691604],\n",
       "        [-0.01576717],\n",
       "        [-0.15303791],\n",
       "        [ 0.22007746],\n",
       "        [ 0.12900022],\n",
       "        [-0.25468057],\n",
       "        [-0.20513205],\n",
       "        [ 0.1073288 ],\n",
       "        [ 0.08123723],\n",
       "        [-0.07285701],\n",
       "        [-0.14686099],\n",
       "        [ 0.1679163 ],\n",
       "        [ 0.14707741],\n",
       "        [-0.12950994],\n",
       "        [ 0.10304123],\n",
       "        [-0.08980718],\n",
       "        [-0.25953025],\n",
       "        [-0.14976214],\n",
       "        [ 0.22343102],\n",
       "        [ 0.1241827 ],\n",
       "        [ 0.22119245],\n",
       "        [ 0.16639099],\n",
       "        [ 0.12644321],\n",
       "        [ 0.25291756],\n",
       "        [ 0.18977529],\n",
       "        [-0.13618685],\n",
       "        [-0.12056601],\n",
       "        [-0.03248692],\n",
       "        [-0.15144779],\n",
       "        [-0.07976879],\n",
       "        [-0.1677155 ],\n",
       "        [ 0.02384892],\n",
       "        [ 0.25621423],\n",
       "        [-0.18188077],\n",
       "        [ 0.04023573],\n",
       "        [-0.19673535],\n",
       "        [ 0.23627409],\n",
       "        [ 0.24041024],\n",
       "        [-0.14854327],\n",
       "        [ 0.18180215],\n",
       "        [-0.26331559],\n",
       "        [-0.15062565],\n",
       "        [ 0.08323905]], dtype=float32),\n",
       " array([ 0.], dtype=float32)]"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Model output shape\n",
    "model.output_shape\n",
    "# Model summary\n",
    "model.summary()\n",
    "# Model config\n",
    "model.get_config()\n",
    "# List all weight tensors \n",
    "model.get_weights()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 87701 samples, validate on 10840 samples\n",
      "Epoch 1/30\n",
      "87701/87701 [==============================] - 10s 109us/step - loss: 2.1316 - main_output_loss: 1.9741 - aux_output_loss: 0.3151 - main_output_acc: 0.7973 - main_output_mean_pred: 0.2622 - aux_output_acc: 0.9071 - aux_output_mean_pred: 0.1090 - val_loss: 0.4693 - val_main_output_loss: 0.3198 - val_aux_output_loss: 0.2990 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1324 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0899\n",
      "\n",
      "Epoch 00001: val_loss improved from inf to 0.46925, saving model to lstm.hdf5\n",
      "Epoch 2/30\n",
      "87701/87701 [==============================] - 9s 98us/step - loss: 1.8258 - main_output_loss: 1.6748 - aux_output_loss: 0.3020 - main_output_acc: 0.8178 - main_output_mean_pred: 0.2137 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0859 - val_loss: 0.4567 - val_main_output_loss: 0.3074 - val_aux_output_loss: 0.2986 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1083 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0948\n",
      "\n",
      "Epoch 00002: val_loss improved from 0.46925 to 0.45665, saving model to lstm.hdf5\n",
      "Epoch 3/30\n",
      "87701/87701 [==============================] - 8s 96us/step - loss: 1.4588 - main_output_loss: 1.3082 - aux_output_loss: 0.3011 - main_output_acc: 0.8212 - main_output_mean_pred: 0.2051 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0861 - val_loss: 0.4502 - val_main_output_loss: 0.3029 - val_aux_output_loss: 0.2945 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1123 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0832\n",
      "\n",
      "Epoch 00003: val_loss improved from 0.45665 to 0.45016, saving model to lstm.hdf5\n",
      "Epoch 4/30\n",
      "87701/87701 [==============================] - 9s 104us/step - loss: 0.9033 - main_output_loss: 0.7535 - aux_output_loss: 0.2996 - main_output_acc: 0.8729 - main_output_mean_pred: 0.1784 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0866 - val_loss: 0.4479 - val_main_output_loss: 0.3012 - val_aux_output_loss: 0.2934 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1239 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0820\n",
      "\n",
      "Epoch 00004: val_loss improved from 0.45016 to 0.44790, saving model to lstm.hdf5\n",
      "Epoch 5/30\n",
      "87701/87701 [==============================] - 9s 102us/step - loss: 0.8221 - main_output_loss: 0.6726 - aux_output_loss: 0.2989 - main_output_acc: 0.8778 - main_output_mean_pred: 0.1770 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0869 - val_loss: 0.4504 - val_main_output_loss: 0.3033 - val_aux_output_loss: 0.2942 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1445 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0795\n",
      "\n",
      "Epoch 00005: val_loss did not improve\n",
      "Epoch 6/30\n",
      "87701/87701 [==============================] - 9s 100us/step - loss: 0.7599 - main_output_loss: 0.6106 - aux_output_loss: 0.2988 - main_output_acc: 0.8813 - main_output_mean_pred: 0.1692 - aux_output_acc: 0.9103 - aux_output_mean_pred: 0.0863 - val_loss: 0.4520 - val_main_output_loss: 0.3056 - val_aux_output_loss: 0.2929 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1559 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0864\n",
      "\n",
      "Epoch 00006: val_loss did not improve\n",
      "Epoch 7/30\n",
      "87701/87701 [==============================] - 9s 108us/step - loss: 0.7138 - main_output_loss: 0.5650 - aux_output_loss: 0.2977 - main_output_acc: 0.8856 - main_output_mean_pred: 0.1625 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0871 - val_loss: 0.4433 - val_main_output_loss: 0.2975 - val_aux_output_loss: 0.2917 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1421 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0890\n",
      "\n",
      "Epoch 00007: val_loss improved from 0.44790 to 0.44333, saving model to lstm.hdf5\n",
      "Epoch 8/30\n",
      "87701/87701 [==============================] - 9s 105us/step - loss: 0.6796 - main_output_loss: 0.5310 - aux_output_loss: 0.2972 - main_output_acc: 0.8880 - main_output_mean_pred: 0.1567 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0867 - val_loss: 0.4362 - val_main_output_loss: 0.2907 - val_aux_output_loss: 0.2910 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1185 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0939\n",
      "\n",
      "Epoch 00008: val_loss improved from 0.44333 to 0.43623, saving model to lstm.hdf5\n",
      "Epoch 9/30\n",
      "87701/87701 [==============================] - 10s 110us/step - loss: 0.6440 - main_output_loss: 0.4957 - aux_output_loss: 0.2965 - main_output_acc: 0.8912 - main_output_mean_pred: 0.1550 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0885 - val_loss: 0.4350 - val_main_output_loss: 0.2892 - val_aux_output_loss: 0.2917 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1228 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0836\n",
      "\n",
      "Epoch 00009: val_loss improved from 0.43623 to 0.43500, saving model to lstm.hdf5\n",
      "Epoch 10/30\n",
      "87701/87701 [==============================] - 10s 115us/step - loss: 0.6162 - main_output_loss: 0.4684 - aux_output_loss: 0.2956 - main_output_acc: 0.8934 - main_output_mean_pred: 0.1504 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0889 - val_loss: 0.4382 - val_main_output_loss: 0.2935 - val_aux_output_loss: 0.2893 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1417 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0853\n",
      "\n",
      "Epoch 00010: val_loss did not improve\n",
      "Epoch 11/30\n",
      "87701/87701 [==============================] - 9s 100us/step - loss: 0.6129 - main_output_loss: 0.4651 - aux_output_loss: 0.2956 - main_output_acc: 0.8956 - main_output_mean_pred: 0.1464 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0879 - val_loss: 0.4406 - val_main_output_loss: 0.2964 - val_aux_output_loss: 0.2883 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1493 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0904\n",
      "\n",
      "Epoch 00011: val_loss did not improve\n",
      "Epoch 12/30\n",
      "87701/87701 [==============================] - 9s 101us/step - loss: 0.5865 - main_output_loss: 0.4389 - aux_output_loss: 0.2952 - main_output_acc: 0.8987 - main_output_mean_pred: 0.1437 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0877 - val_loss: 0.4297 - val_main_output_loss: 0.2852 - val_aux_output_loss: 0.2891 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1182 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0850\n",
      "\n",
      "Epoch 00012: val_loss improved from 0.43500 to 0.42972, saving model to lstm.hdf5\n",
      "Epoch 13/30\n",
      "87701/87701 [==============================] - 9s 103us/step - loss: 0.5808 - main_output_loss: 0.4331 - aux_output_loss: 0.2954 - main_output_acc: 0.8986 - main_output_mean_pred: 0.1403 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0877 - val_loss: 0.4286 - val_main_output_loss: 0.2841 - val_aux_output_loss: 0.2889 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1138 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0929\n",
      "\n",
      "Epoch 00013: val_loss improved from 0.42972 to 0.42857, saving model to lstm.hdf5\n",
      "Epoch 14/30\n",
      "87701/87701 [==============================] - 9s 101us/step - loss: 0.5429 - main_output_loss: 0.3957 - aux_output_loss: 0.2944 - main_output_acc: 0.9005 - main_output_mean_pred: 0.1363 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0877 - val_loss: 0.4275 - val_main_output_loss: 0.2841 - val_aux_output_loss: 0.2869 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1153 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0873\n",
      "\n",
      "Epoch 00014: val_loss improved from 0.42857 to 0.42751, saving model to lstm.hdf5\n",
      "Epoch 15/30\n",
      "87701/87701 [==============================] - 9s 104us/step - loss: 0.5199 - main_output_loss: 0.3728 - aux_output_loss: 0.2942 - main_output_acc: 0.9017 - main_output_mean_pred: 0.1333 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0878 - val_loss: 0.4335 - val_main_output_loss: 0.2893 - val_aux_output_loss: 0.2885 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1339 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0851\n",
      "\n",
      "Epoch 00015: val_loss did not improve\n",
      "Epoch 16/30\n",
      "87701/87701 [==============================] - 8s 93us/step - loss: 0.5035 - main_output_loss: 0.3571 - aux_output_loss: 0.2927 - main_output_acc: 0.9039 - main_output_mean_pred: 0.1290 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0882 - val_loss: 0.4273 - val_main_output_loss: 0.2835 - val_aux_output_loss: 0.2877 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1105 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0793\n",
      "\n",
      "Epoch 00016: val_loss improved from 0.42751 to 0.42730, saving model to lstm.hdf5\n",
      "Epoch 17/30\n",
      "87701/87701 [==============================] - 8s 94us/step - loss: 0.4912 - main_output_loss: 0.3451 - aux_output_loss: 0.2922 - main_output_acc: 0.9049 - main_output_mean_pred: 0.1274 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0883 - val_loss: 0.4266 - val_main_output_loss: 0.2826 - val_aux_output_loss: 0.2880 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1030 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0738\n",
      "\n",
      "Epoch 00017: val_loss improved from 0.42730 to 0.42663, saving model to lstm.hdf5\n",
      "Epoch 18/30\n",
      "87701/87701 [==============================] - 8s 93us/step - loss: 0.4771 - main_output_loss: 0.3311 - aux_output_loss: 0.2920 - main_output_acc: 0.9056 - main_output_mean_pred: 0.1247 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0874 - val_loss: 0.4294 - val_main_output_loss: 0.2868 - val_aux_output_loss: 0.2852 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1270 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0855\n",
      "\n",
      "Epoch 00018: val_loss did not improve\n",
      "Epoch 19/30\n",
      "87701/87701 [==============================] - 8s 95us/step - loss: 0.4655 - main_output_loss: 0.3197 - aux_output_loss: 0.2915 - main_output_acc: 0.9068 - main_output_mean_pred: 0.1215 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0885 - val_loss: 0.4271 - val_main_output_loss: 0.2841 - val_aux_output_loss: 0.2859 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1147 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0814\n",
      "\n",
      "Epoch 00019: val_loss did not improve\n",
      "Epoch 20/30\n",
      "87701/87701 [==============================] - 8s 96us/step - loss: 0.4610 - main_output_loss: 0.3155 - aux_output_loss: 0.2910 - main_output_acc: 0.9072 - main_output_mean_pred: 0.1190 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0879 - val_loss: 0.4276 - val_main_output_loss: 0.2851 - val_aux_output_loss: 0.2849 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1199 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0853\n",
      "\n",
      "Epoch 00020: val_loss did not improve\n",
      "Epoch 21/30\n",
      "87701/87701 [==============================] - 8s 95us/step - loss: 0.4517 - main_output_loss: 0.3063 - aux_output_loss: 0.2909 - main_output_acc: 0.9081 - main_output_mean_pred: 0.1152 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0886 - val_loss: 0.4244 - val_main_output_loss: 0.2817 - val_aux_output_loss: 0.2854 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1049 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0843\n",
      "\n",
      "Epoch 00021: val_loss improved from 0.42663 to 0.42439, saving model to lstm.hdf5\n",
      "Epoch 22/30\n",
      "87701/87701 [==============================] - 8s 96us/step - loss: 0.4462 - main_output_loss: 0.3009 - aux_output_loss: 0.2905 - main_output_acc: 0.9089 - main_output_mean_pred: 0.1124 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0884 - val_loss: 0.4218 - val_main_output_loss: 0.2793 - val_aux_output_loss: 0.2850 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1013 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0872\n",
      "\n",
      "Epoch 00022: val_loss improved from 0.42439 to 0.42178, saving model to lstm.hdf5\n",
      "Epoch 23/30\n",
      "87701/87701 [==============================] - 8s 96us/step - loss: 0.4398 - main_output_loss: 0.2949 - aux_output_loss: 0.2899 - main_output_acc: 0.9096 - main_output_mean_pred: 0.1085 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0888 - val_loss: 0.4221 - val_main_output_loss: 0.2795 - val_aux_output_loss: 0.2853 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1101 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0874\n",
      "\n",
      "Epoch 00023: val_loss did not improve\n",
      "Epoch 24/30\n",
      "87701/87701 [==============================] - 8s 97us/step - loss: 0.4376 - main_output_loss: 0.2925 - aux_output_loss: 0.2903 - main_output_acc: 0.9098 - main_output_mean_pred: 0.1068 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0888 - val_loss: 0.4188 - val_main_output_loss: 0.2757 - val_aux_output_loss: 0.2863 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.0951 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0855\n",
      "\n",
      "Epoch 00024: val_loss improved from 0.42178 to 0.41880, saving model to lstm.hdf5\n",
      "Epoch 25/30\n",
      "87701/87701 [==============================] - 9s 98us/step - loss: 0.4335 - main_output_loss: 0.2888 - aux_output_loss: 0.2893 - main_output_acc: 0.9102 - main_output_mean_pred: 0.1041 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0889 - val_loss: 0.4165 - val_main_output_loss: 0.2743 - val_aux_output_loss: 0.2844 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.0972 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0888\n",
      "\n",
      "Epoch 00025: val_loss improved from 0.41880 to 0.41648, saving model to lstm.hdf5\n",
      "Epoch 26/30\n",
      "87701/87701 [==============================] - 9s 98us/step - loss: 0.4305 - main_output_loss: 0.2859 - aux_output_loss: 0.2891 - main_output_acc: 0.9103 - main_output_mean_pred: 0.1014 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0884 - val_loss: 0.4175 - val_main_output_loss: 0.2752 - val_aux_output_loss: 0.2847 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1014 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0925\n",
      "\n",
      "Epoch 00026: val_loss did not improve\n",
      "Epoch 27/30\n",
      "87701/87701 [==============================] - 8s 89us/step - loss: 0.4275 - main_output_loss: 0.2832 - aux_output_loss: 0.2886 - main_output_acc: 0.9103 - main_output_mean_pred: 0.1000 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0888 - val_loss: 0.4170 - val_main_output_loss: 0.2751 - val_aux_output_loss: 0.2837 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.1020 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0868\n",
      "\n",
      "Epoch 00027: val_loss did not improve\n",
      "Epoch 28/30\n",
      "87701/87701 [==============================] - 8s 91us/step - loss: 0.4261 - main_output_loss: 0.2819 - aux_output_loss: 0.2884 - main_output_acc: 0.9104 - main_output_mean_pred: 0.0983 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0890 - val_loss: 0.4151 - val_main_output_loss: 0.2730 - val_aux_output_loss: 0.2841 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.0949 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0939\n",
      "\n",
      "Epoch 00028: val_loss improved from 0.41648 to 0.41511, saving model to lstm.hdf5\n",
      "Epoch 29/30\n",
      "87701/87701 [==============================] - 9s 107us/step - loss: 0.4252 - main_output_loss: 0.2807 - aux_output_loss: 0.2890 - main_output_acc: 0.9104 - main_output_mean_pred: 0.0960 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0879 - val_loss: 0.4157 - val_main_output_loss: 0.2739 - val_aux_output_loss: 0.2837 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.0802 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0849\n",
      "\n",
      "Epoch 00029: val_loss did not improve\n",
      "Epoch 30/30\n",
      "87701/87701 [==============================] - 9s 103us/step - loss: 0.4238 - main_output_loss: 0.2798 - aux_output_loss: 0.2879 - main_output_acc: 0.9104 - main_output_mean_pred: 0.0948 - aux_output_acc: 0.9104 - aux_output_mean_pred: 0.0891 - val_loss: 0.4145 - val_main_output_loss: 0.2726 - val_aux_output_loss: 0.2838 - val_main_output_acc: 0.9107 - val_main_output_mean_pred: 0.0858 - val_aux_output_acc: 0.9107 - val_aux_output_mean_pred: 0.0835\n",
      "\n",
      "Epoch 00030: val_loss improved from 0.41511 to 0.41446, saving model to lstm.hdf5\n"
     ]
    }
   ],
   "source": [
    "import keras.backend as K\n",
    "from keras.callbacks import ReduceLROnPlateau,ModelCheckpoint\n",
    "from keras.optimizers import RMSprop\n",
    "def mean_pred(y_true, y_pred):\n",
    "    return K.mean(y_pred)\n",
    "\n",
    "# class_weight = {0 : 1., 1: 7.}\n",
    "opt = RMSprop(lr=0.0001)\n",
    "model.compile(optimizer=opt,\n",
    "              loss={'main_output': 'binary_crossentropy', 'aux_output': 'binary_crossentropy'},\n",
    "              loss_weights={'main_output': 1., 'aux_output': 0.5},metrics=['accuracy',mean_pred])\n",
    "# callback_early_stopping = [EarlyStopping(monitor='val_loss', patience=10,verbose=0, mode='auto')]\n",
    "# And trained it via:\n",
    "reduce_lr = ReduceLROnPlateau(monitor='val_main_output_loss', factor=0.9, patience=30, min_lr=0.000001, verbose=1)\n",
    "checkpointer = ModelCheckpoint(filepath='lstm.hdf5', verbose=1, save_best_only=True)\n",
    "\n",
    "history=model.fit({'main_input': X_train_time, 'aux_input': X_train_aux},\n",
    "          {'main_output': y_train, 'aux_output': y_train},\n",
    "          epochs=30, batch_size=128, callbacks=[reduce_lr, checkpointer],validation_split=0.11)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'aux_output_acc': [0.90708201730949023,\n",
       "  0.91035449995004858,\n",
       "  0.9103544999493689,\n",
       "  0.9103544999493689,\n",
       "  0.91035449995004858,\n",
       "  0.91034309757355136,\n",
       "  0.9103544999493689,\n",
       "  0.91035449994868933,\n",
       "  0.91035449995140783,\n",
       "  0.91035449995072815,\n",
       "  0.91035449995140783,\n",
       "  0.9103544999493689,\n",
       "  0.9103544999493689,\n",
       "  0.9103544999493689,\n",
       "  0.91035449995140783,\n",
       "  0.91035449995004858,\n",
       "  0.9103544999493689,\n",
       "  0.91035449995140783,\n",
       "  0.9103544999493689,\n",
       "  0.91035449995072815,\n",
       "  0.9103544999493689,\n",
       "  0.91035449994868933,\n",
       "  0.9103544999493689,\n",
       "  0.91035449995072815,\n",
       "  0.91035449994868933,\n",
       "  0.91035449995004858,\n",
       "  0.91035449995072815,\n",
       "  0.91035449995072815,\n",
       "  0.9103544999493689,\n",
       "  0.9103544999493689],\n",
       " 'aux_output_loss': [0.31507012483243835,\n",
       "  0.30201621011956886,\n",
       "  0.30113363193627102,\n",
       "  0.29959272544691418,\n",
       "  0.29894738290790029,\n",
       "  0.29878864827715379,\n",
       "  0.29767649653402373,\n",
       "  0.29720353174459496,\n",
       "  0.29651122361941845,\n",
       "  0.29564662225485577,\n",
       "  0.29560186760157603,\n",
       "  0.29520700241226416,\n",
       "  0.29544527633502299,\n",
       "  0.29444521699758747,\n",
       "  0.29419877883682538,\n",
       "  0.29265795057172594,\n",
       "  0.29221215315840304,\n",
       "  0.2919837425465206,\n",
       "  0.29146646772755874,\n",
       "  0.29101249487568254,\n",
       "  0.29089940670708253,\n",
       "  0.29054657557813546,\n",
       "  0.28985318828094775,\n",
       "  0.29033136606825383,\n",
       "  0.28933842759648448,\n",
       "  0.28910583950222951,\n",
       "  0.28856494927896736,\n",
       "  0.28843312358039552,\n",
       "  0.288984246677638,\n",
       "  0.28786332334637216],\n",
       " 'aux_output_mean_pred': [0.10895446852187504,\n",
       "  0.085945330898750408,\n",
       "  0.086067044934070816,\n",
       "  0.086568810897386411,\n",
       "  0.086866961020449884,\n",
       "  0.086283772501044603,\n",
       "  0.087137379942611809,\n",
       "  0.086696692153280824,\n",
       "  0.088479311679863615,\n",
       "  0.088940814362417894,\n",
       "  0.087859900983762618,\n",
       "  0.087696650561731829,\n",
       "  0.087698628038983178,\n",
       "  0.087722541684925007,\n",
       "  0.08782460298803986,\n",
       "  0.088229862689197816,\n",
       "  0.088251538475468919,\n",
       "  0.087411186612703362,\n",
       "  0.088530424624014406,\n",
       "  0.087937675363797888,\n",
       "  0.088574537788260091,\n",
       "  0.088430389716304264,\n",
       "  0.08875900727302348,\n",
       "  0.088753521982977798,\n",
       "  0.08894667709032511,\n",
       "  0.088351309778016793,\n",
       "  0.088791695763485848,\n",
       "  0.089041188592715093,\n",
       "  0.087900036815512908,\n",
       "  0.089134774818985699],\n",
       " 'loss': [2.1316001131899376,\n",
       "  1.8258458735308034,\n",
       "  1.4587775102462401,\n",
       "  0.90328989430554008,\n",
       "  0.82208977101149572,\n",
       "  0.75994859954391714,\n",
       "  0.71379094018714428,\n",
       "  0.67959678791872602,\n",
       "  0.64397333249963029,\n",
       "  0.61620744083644341,\n",
       "  0.61292860782904524,\n",
       "  0.58649543527145853,\n",
       "  0.58080426527903839,\n",
       "  0.5429364575320107,\n",
       "  0.51993038710981743,\n",
       "  0.50345594516764303,\n",
       "  0.49122108036927437,\n",
       "  0.4771387111072134,\n",
       "  0.46547177795431316,\n",
       "  0.46103601343902156,\n",
       "  0.45171455054653897,\n",
       "  0.44618128005039831,\n",
       "  0.43982706326161786,\n",
       "  0.43762420858605949,\n",
       "  0.43348424196522956,\n",
       "  0.43049385946080593,\n",
       "  0.4274669489938448,\n",
       "  0.42611417961933995,\n",
       "  0.42517252101408692,\n",
       "  0.42377590217343869],\n",
       " 'lr': [9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05,\n",
       "  9.9999997e-05],\n",
       " 'main_output_acc': [0.79726570963042642,\n",
       "  0.81784700288684065,\n",
       "  0.82116509504154367,\n",
       "  0.87289768645875432,\n",
       "  0.8777550997158392,\n",
       "  0.88131264182184044,\n",
       "  0.8856341432856869,\n",
       "  0.88798303326073824,\n",
       "  0.89115289449650992,\n",
       "  0.8933877606895978,\n",
       "  0.89557701736854101,\n",
       "  0.89868986670687456,\n",
       "  0.89862145243565761,\n",
       "  0.90054845440826903,\n",
       "  0.90174570415660504,\n",
       "  0.90385514418443591,\n",
       "  0.90489276063054702,\n",
       "  0.90564531761665235,\n",
       "  0.90678555546755002,\n",
       "  0.90723024823181964,\n",
       "  0.90805121948506407,\n",
       "  0.90889499549674013,\n",
       "  0.90957913820958947,\n",
       "  0.90977297864538387,\n",
       "  0.91021767140625531,\n",
       "  0.91033169519297619,\n",
       "  0.91034309757219201,\n",
       "  0.91035449995072815,\n",
       "  0.9103544999493689,\n",
       "  0.9103544999493689],\n",
       " 'main_output_loss': [1.9740650545233482,\n",
       "  1.6748377691523528,\n",
       "  1.3082106948235963,\n",
       "  0.75349353173253708,\n",
       "  0.67261608036223319,\n",
       "  0.61055427545223506,\n",
       "  0.56495269143096538,\n",
       "  0.53099502159060596,\n",
       "  0.49571772071880549,\n",
       "  0.46838412996659706,\n",
       "  0.46512767369115837,\n",
       "  0.43889193478658878,\n",
       "  0.43308162592445992,\n",
       "  0.39571384951524785,\n",
       "  0.37283099782563262,\n",
       "  0.35712696978391273,\n",
       "  0.34511500344209989,\n",
       "  0.33114683979045645,\n",
       "  0.31973854534106166,\n",
       "  0.31552976605555111,\n",
       "  0.30626484686677302,\n",
       "  0.30090799243353306,\n",
       "  0.29490046956698435,\n",
       "  0.29245852559169117,\n",
       "  0.28881502856933111,\n",
       "  0.28594093956289007,\n",
       "  0.28318447466240554,\n",
       "  0.281897618303867,\n",
       "  0.28068039822984986,\n",
       "  0.27984424078476472],\n",
       " 'main_output_mean_pred': [0.26219541437092969,\n",
       "  0.21371363827183729,\n",
       "  0.20508584408796446,\n",
       "  0.17839458236472666,\n",
       "  0.1770367390089933,\n",
       "  0.16916138714546419,\n",
       "  0.16252812572156872,\n",
       "  0.1566984833934609,\n",
       "  0.15500774957928012,\n",
       "  0.15042730187653688,\n",
       "  0.14636588861360633,\n",
       "  0.14371532640573126,\n",
       "  0.14031587736226492,\n",
       "  0.13629094296837274,\n",
       "  0.1332690302398577,\n",
       "  0.12903365459773503,\n",
       "  0.12741482760067591,\n",
       "  0.12468132930963927,\n",
       "  0.12151686313356856,\n",
       "  0.11904558453552698,\n",
       "  0.11519507119967452,\n",
       "  0.11242930700969368,\n",
       "  0.1085018443870098,\n",
       "  0.10677182169303433,\n",
       "  0.10408964615006687,\n",
       "  0.10136778991771744,\n",
       "  0.10002161798140101,\n",
       "  0.098303208398855813,\n",
       "  0.095963647986100353,\n",
       "  0.094793501687938014],\n",
       " 'val_aux_output_acc': [0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154],\n",
       " 'val_aux_output_loss': [0.29898861332133247,\n",
       "  0.29856052974933189,\n",
       "  0.29449042180807389,\n",
       "  0.29335555108710848,\n",
       "  0.29419502367392675,\n",
       "  0.29290376890629422,\n",
       "  0.2917237477548888,\n",
       "  0.29104540473420681,\n",
       "  0.29167916611991684,\n",
       "  0.28931264828931802,\n",
       "  0.28833640393735738,\n",
       "  0.28909876438964338,\n",
       "  0.2889432200206602,\n",
       "  0.28686922597269293,\n",
       "  0.28849118547685909,\n",
       "  0.28768816460542573,\n",
       "  0.28803559829828046,\n",
       "  0.28521370927346151,\n",
       "  0.28594412948812503,\n",
       "  0.28493935017128275,\n",
       "  0.28538353722473792,\n",
       "  0.28502366894725506,\n",
       "  0.28533178915396828,\n",
       "  0.28627148279404729,\n",
       "  0.28442727639666343,\n",
       "  0.28466470804601579,\n",
       "  0.28368331435861621,\n",
       "  0.28414520232000035,\n",
       "  0.28366864886670973,\n",
       "  0.28378075125912461],\n",
       " 'val_aux_output_mean_pred': [0.089902579234534957,\n",
       "  0.094798207986838703,\n",
       "  0.083161381755807742,\n",
       "  0.082028940770898798,\n",
       "  0.079478305220823886,\n",
       "  0.086396201955656282,\n",
       "  0.088989385616075514,\n",
       "  0.093949771350820124,\n",
       "  0.083646623306388779,\n",
       "  0.085347278366669513,\n",
       "  0.09035180755976821,\n",
       "  0.084989970471586254,\n",
       "  0.092933961375173166,\n",
       "  0.087292910969345328,\n",
       "  0.085107612065503518,\n",
       "  0.079329502021694531,\n",
       "  0.073824733187791605,\n",
       "  0.085471656412656014,\n",
       "  0.081440053924203365,\n",
       "  0.085321560403077798,\n",
       "  0.084304043254729122,\n",
       "  0.087174951964198888,\n",
       "  0.08744362632286945,\n",
       "  0.085545487333487755,\n",
       "  0.088788502754100579,\n",
       "  0.092547758808875003,\n",
       "  0.086841927155357448,\n",
       "  0.093860400371885824,\n",
       "  0.084944714105437163,\n",
       "  0.083466662724959451],\n",
       " 'val_loss': [0.46925215624355304,\n",
       "  0.45665076821492606,\n",
       "  0.45016234321347903,\n",
       "  0.44789954173169011,\n",
       "  0.45042467293264243,\n",
       "  0.45203035330860375,\n",
       "  0.44333244535755845,\n",
       "  0.436230772051864,\n",
       "  0.43500227235340105,\n",
       "  0.4381902125488788,\n",
       "  0.44058349919934992,\n",
       "  0.42972186933584317,\n",
       "  0.42856717105281311,\n",
       "  0.42751049326794616,\n",
       "  0.4335230500495742,\n",
       "  0.42730365967398642,\n",
       "  0.42663154135770903,\n",
       "  0.42938314166016245,\n",
       "  0.42705021652348368,\n",
       "  0.42761396503976351,\n",
       "  0.42438744447328069,\n",
       "  0.42177514066555405,\n",
       "  0.42213699217212158,\n",
       "  0.41880481595042884,\n",
       "  0.41648070374538099,\n",
       "  0.41749139999991414,\n",
       "  0.41696148187032045,\n",
       "  0.41510795777134352,\n",
       "  0.41574391535727301,\n",
       "  0.41446361088664768],\n",
       " 'val_main_output_acc': [0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154,\n",
       "  0.9107011068351154],\n",
       " 'val_main_output_loss': [0.31975784847217292,\n",
       "  0.30737050246048675,\n",
       "  0.30291713278232024,\n",
       "  0.3012217673868271,\n",
       "  0.30332716107368468,\n",
       "  0.30557846979021586,\n",
       "  0.29747056998449939,\n",
       "  0.29070807094943479,\n",
       "  0.28916268920546528,\n",
       "  0.29353388761242377,\n",
       "  0.29641529743961742,\n",
       "  0.28517248951640956,\n",
       "  0.28409556209821102,\n",
       "  0.28407587928085748,\n",
       "  0.28927745810294064,\n",
       "  0.28345957728329618,\n",
       "  0.28261374279141865,\n",
       "  0.28677628605568101,\n",
       "  0.28407815116357982,\n",
       "  0.28514429113081902,\n",
       "  0.28169567703760856,\n",
       "  0.27926330768754121,\n",
       "  0.27947109715525076,\n",
       "  0.27566907529021539,\n",
       "  0.27426706537109458,\n",
       "  0.27515904483320086,\n",
       "  0.27511982557078568,\n",
       "  0.27303535590752465,\n",
       "  0.27390958893343093,\n",
       "  0.27257323557600321],\n",
       " 'val_main_output_mean_pred': [0.13237573934437163,\n",
       "  0.10826655690080565,\n",
       "  0.11227373364776704,\n",
       "  0.12390965371774132,\n",
       "  0.14449997438715834,\n",
       "  0.15585401219855374,\n",
       "  0.14212436484674687,\n",
       "  0.11853957121222661,\n",
       "  0.12278891425730998,\n",
       "  0.14174579626079856,\n",
       "  0.14930697106567256,\n",
       "  0.11815488943864498,\n",
       "  0.11380207627681789,\n",
       "  0.11525096109762403,\n",
       "  0.13388682512559574,\n",
       "  0.1105026278548575,\n",
       "  0.1030238479430825,\n",
       "  0.12696987133184481,\n",
       "  0.11466598721887793,\n",
       "  0.11992201727034861,\n",
       "  0.10492639904532486,\n",
       "  0.10126248531565894,\n",
       "  0.11008786688321631,\n",
       "  0.095122993745707055,\n",
       "  0.097246858911980566,\n",
       "  0.1014175876810102,\n",
       "  0.10203292202685592,\n",
       "  0.094853385452634734,\n",
       "  0.080232608890181537,\n",
       "  0.085768887873505312]}"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "history.history"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfUAAAFlCAYAAADyLnFSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3XtYlGXeB/DvHIFhOMhBkzioCHgWwTQzctMo02wlKsBe\nLW333bfTVrtaauXrKppZvVvrlh22tV07Wda64i7VYpJFZjaKhgoeUhRROQ4wwwxzut8/wElyQBCG\nOX0/18U1M88zzzw/bufi63O471sihBAgIiIijyd1dQFERETUOxjqREREXoKhTkRE5CUY6kRERF6C\noU5EROQlGOpERERegqFO5IVefvllbNmyxaU1vPXWW1i8ePFl35eUlIS6uro+qIjI+8ldXQAR9b5H\nH33U1SUQkQsw1InczO7du/F///d/6N+/P44ePYqAgAA88sgj2LhxI06cOIGbb74ZS5cuhc1mw+rV\nq7F//37o9XoIIZCbm4vU1FQsXrwYCQkJuP/++zF69Gj893//N4qKilBVVYV58+bhvvvuu2S/o0eP\nxn333YfCwkLodDosWrQIn376KY4cOYL+/fvjtddeg0qlwvfff4+1a9fCYDBAoVDgscceww033ACz\n2Yzc3Fx88803CA8PR3h4OIKCggAATU1NWLVqFY4cOQKz2YxJkybhiSeegFze8Z+gzZs3Y9OmTTCb\nzWhoaMCvf/1rzJkzBwDw+uuv4x//+Afkcjni4uKwZs0aBAUFdbicyGcIInIr3377rRg+fLg4ePCg\nEEKI+++/X2RlZYmWlhZRW1srRo4cKc6dOyf27t0rHnnkEWG1WoUQQrz++uviN7/5jRBCiCeffFL8\n5S9/EUIIkZiYKDZu3CiEEOKHH34Qo0aNEkaj8ZL9JiYmir/97W/2zxo3bpw4d+6csFqtIiMjQ2zd\nulXU1dWJSZMmieLiYiGEEEeOHBETJkwQp06dEm+//baYN2+eaGlpEXq9XmRkZIgnn3xSCCHE4sWL\nxd///nchhBAWi0UsXLhQvPHGG/b91tbWtqtFp9OJu+++W9TV1QkhhNi3b59ITk4WQghRUFAgbr75\nZqHVaoUQQqxevVq8+uqrHS4n8iU8UidyQ9HR0RgxYgQAIDY2FkFBQVAqlQgLC0NgYCAaGhowbtw4\nhISE4IMPPsDp06exe/duBAYGOvy8adOmAQBGjhwJk8mE5uZm+Pn5XfK+W265xb7PxMREDBgwwF5P\nQ0MDDhw4gNjYWIwdOxYAkJCQgJSUFHz33XfYtWsXbrvtNiiVSiiVSsyaNQtlZWUAgMLCQvzwww/Y\nvHkzAMBoNHb6+wcGBuK1117Dl19+iZMnT6K0tBTNzc0AgF27dmH69OkICQkBACxZsgQAkJub63A5\nkS9hqBO5IaVS2e61o9PUhYWFWLVqFebPn49p06ZhyJAh2Lp1q8PPuxDgEokEACA6mPJBoVA4fH6B\nzWa7ZJkQAhaL5ZLlMpms3XYvv/wy4uPjAQCNjY32Whw5d+4csrKycPfddyM1NRXTp0/Hjh077J97\n8baNjY1obGzscHl0dHSH+yHyNrz7nchDFRUV4cYbb8ScOXMwevRoFBQUwGq1OnWfY8eOxYkTJ3Dg\nwAEAwNGjR7Fnzx5MmDABaWlp2LJlC1paWtDS0oJ///vf9u2uv/56vP322xBCwGQy4YEHHsA777zT\n4X5KSkoQFhaGBx98EGlpafZAt1qtuO666/Cf//wHOp0OALBu3Tq8/fbbHS4n8iU8UifyUNnZ2Vi4\ncCFmzZoFmUyG8ePH4/PPP3d4NN1bwsLC8PLLL2PlypUwGo2QSCR49tlnMXjwYMTGxuLUqVO47bbb\nEBoairi4OPt2Tz31FFatWoVZs2bBbDbjuuuuw69+9asO9zN58mRs3rwZ06dPR0BAAMaMGYOwsDCU\nl5djypQpOHbsGHJycgAAQ4cOxcqVK6FWqx0uJ/IlEtHReTgiIiLyKDz9TkRE5CUY6kRERF6CoU5E\nROQlGOpERERegqFORETkJTy+S5tGo3F1CURERH0qNTXV4XKPD3Wg41/uSmg0ml79PG/BdnGM7eIY\n28UxtotjbBfHOmqXzg5mefqdiIjISzDUiYiIvARDnYiIyEsw1ImIiLwEQ52IiMhLMNSJiIi8BEOd\niIjISzDUnaClpQUfffRRl977ySefYPv27U6uiIiIfAFD3Qmqq6u7HOp33HEHpk2b5uSKiIjIF3jF\niHKd+WveQRTtP9Pl97eYTPDL/7zT90weezUWzBrZ4frXXnsNx44dw7Bhw3DdddehubkZq1atwpYt\nW1BSUgKtVothw4bh2Wefxbp16xAREYEhQ4bgzTffhEKhQEVFBWbMmIEHHnigy3UTERF5fah3h87U\nDKOlBTpLc6fvKzj+IzR573W43pCghyzcD4Oui8Upw3kk3ZOMtXtfR8WZHzHojiQE2Abi0xc+R9U7\nLThT9iP8Kv2hOheEsrJiTPx9Oq62BGPdyj+jJLqit3/FK9ZiMsGv8hNXl+F22C6OsV0cY7s45u3t\ncm1MCuYmZ/bJvrw+1BfMGtnpUfXFNhZ/jC+Pfw8/pbLX9q+KDAIASBUymHRG/PDObsj85LC2WCCs\ntnbvVQ8MgVQmBWRSyBSyXquBiMgZhABsQrT+2AABAYi2dW3PxUXvvfDk58usVhtaLKYeFtOj1U51\ntkbfZ/vy+lDvjrnJmRhhHdTjiQUqKyvx27zfYkrSFERERCBnVg62b9+OPJUBL73xEurq6nDrrbfi\nD1N/jy2NW+yn3z849QH+OGsVAGDymkK80vbcHXDCBcfYLo6xXRxzl3YRQsBitaHFbIPZbEWL2Qqz\nxdb6aLZBbzSjqdmERr0JTfrWx8bmi563/Vh+dmBCjomxUX22L4a6E4SHh8NsNsNoNNqXjRkzBq++\n+iruueceSCQSxMTEoKqqyoVVEpE3MJmtqGkwoLq+7UdrQHV9M6q1BuiaTTBZbDCZrTCZ2x4tNpgt\n1p+OnLsp0F+O4EA/DLk6GMGBfghSKVofAxUIUMohk0khl0kgk7Y9yi56lEohk0kgl7U9tr0uPXwY\nI0aOgKRtHxKJpN0+L7y8sNy+9sLytic/26zD9/e18JCAPtsXQ90J/Pz88M9//rPdssjISHz88ceX\nvPfi/7VPnDjR/ryoqMh5BRKRRxBCoFFvQlV980WBbUC19qfX2qaWDrf3U8qglMugVEjhp5QhSKWA\nUiFr/ZFL7c8Vcin8FDIoFK2PAX5yBAcqERyoRJCq7bHtuVzW+52m6s4qEHdVcK9/ri9iqBMRuZiu\n2YTKGj0qq3Vtj3pU1rQ+1xvMDrdRyKWICA1A7IAgRPYLQGSoqu0xAJH9AhARGgB/Jf/E+xr+ixMR\n9QG9wYzKOhP0+youCfCm5ktvEpPLpBgYocKoIeEYEK5C/34qe2BHhqoQolZecpqaiKFORNQLhBBo\najbjbI0OZ2v0OFujR2Vt6+O5Wj0adBeC+6d7aWRSCa4KVyEprh+iIgMRFaFGVEQgoiLViAgNgEzK\n0KbuYagTEXWR0WRBo96EGq3BHtwXh7ejU+UyqQQDwlQYGh0Kma0ZySMG2wO8f78AyJxwjZp8F0Od\niHyS2WJFbYPRYdeti19fvN5kcdyF68Kp8pGDwzEwItD+ExURiMjQn4K7tUvbkL78NcnHMNSJyOsZ\nWyw4ebYRxyu0OH6mAccrGlB+rhFW2+X7dQX4yREUqETswGAEt90J3i/YH1EXhXd4CE+Vk3tgqDtB\nS0sLtm7dirvuuqvL2+zZswdBQUEYNmyYEysj8n56gxk/VrYG9/EzWhyvaMCZqiZcnN9KuRRDo0MR\nFRmI4EA/e5et4EAlglU/PQ9SKaGQ8/Q4eQ6GuhNcmKWtO6H+8ccfY8aMGQx1oi4ytlhwvr4Z5+ua\ncepck/0o/OdDcgb4yTF8cDjio0MQf3Uo4qNDEB2p5rVs8kpeH+obiz/Gt6f3dvn9XZlY4HKD81+Y\npe3Pf/4zjhw5gvr6egDA008/jaSkJCxZsgTl5eUwGo2YN28ehg4diq+++goHDx7E0KFDERXVd0MK\nErkri9WG6noDztfpcb6uNbzP1zbbn2t1lw66EqRSIDkhsl2AXxUeCClPjZOP8PpQd4X/+Z//wZEj\nR2AwGHDttddizpw5OHnyJJYsWYI333wTe/bswYcffgigdeS4UaNGIS0tDTNmzGCgk08RQqBaa8DJ\ns40oP9uIM9U6e2jXag1wdMlbLpMgsp8Kg6KCMSBMhQFhKkT3VyP+6lBE9gtg323yaV4f6nOTM7s1\n5V1vTrhw5MgRfPvtt8jPzwcANDQ0QK1WY+nSpXjmmWeg0+lw++2398q+iNxds9FsD+8TbY/lZxuh\nN1ravU8iAcKD/TF8cLg9tH/6CURYiD9vSiPqgNeHuitIpVLYbDYMGTIEt99+O2bNmoXa2lp89NFH\nqKqqwsGDB/HKK6+gpaUFU6ZMwS9/+UtIJBKIK51hgciNWK02VDWY8dW+MzhxtgHlZ5tw8mwDquoN\n7d4nlQBRkWqMSwrGoIGtP9EDgtC/XwAUck49THQlGOpOcGGWNr1ej/z8fHz44YfQ6XR4+OGHERkZ\nierqamRnZ0MqlWLBggWQy+UYO3YsXnjhBURHRyM+Pt7VvwJRtwghcPS0Fjs0p/FV8Zm20dPO29f3\nC/JDcmKkPbzjBgYjZkAQ/BQMb6LexFB3AkeztF1sxYoVlyzLzs5Gdna2M8si6nXnavX4cm8Fdmgq\ncKZaBwAIUSuRPESF8aOH2EM8RO3n4kqJfANDnYi6panZhK/3V6JQcxqHTtQBaO33fUPy1bhxfAyS\nEyOxv3gfUlN5xomorzHUieiyzBYr9hw6j8K9Fdhz6DwsVhskEmDM0AjcmBqD68YMhMpf4eoyiXwe\nQ52IHLLZBA6frMMOzWl8vb/SPlnJoIHBuDE1GjeMi0ZEaICLqySiizHUiQgAUNtgwLHTWhw9rcXR\nCi2OndaiUd86XWhYsD9unhiHG1OjMTgqxMWVElFHGOpEPqhB14JjbcF9tO2nrtHY7j39w1QYP3wA\nbkyNxuihkewbTuQBnBbqNpsNy5cvR1lZGZRKJXJzcxEXF2dfv2XLFrz11lsICgpCRkYG7rrrrstu\nQ0TdZ2ix4Ojpehw91XoEfvS0FlV1ze3eExbsj4kjr0JCTCgSYvohPjqEd6wTeSCnhXpBQQFMJhM2\nbdqE4uJirFmzBuvXrwcA1NXV4U9/+hM++eQTBAcH47777sOkSZNw6NChDrchoq5rNprx3aHz+OZA\nJTSHz7ebBzw4UInUYf0xNCYUCdGhGBoTivAQXhsn8gZOC3WNRoO0tDQAQHJyMkpKSuzrKioqkJSU\nhNDQUADA6NGjsX//fhw4cKDDbYioc3qDGd8dOoei/ZXYW1YFc1uQxwwIwjXDByAxth8SYjg+OpE3\nc1qo63Q6qNVq+2uZTAaLxQK5XI64uDgcO3YMNTU1CAwMxK5duzBo0KBOt+mMRqPp1dp7+/O8BdvF\nMVe2i8FkQ1mFAQdPGfDjOSOsbQfk/UMVGBGjxojYAPQPUQAwApazOH3iLE6f6Jva+H1xjO3iGNvF\nse62i9NCXa1WQ6//aV5jm81mD+eQkBAsWbIEjzzyCEJDQzFy5Ej069ev020601sTsAC9O6GLN2G7\nOOaKdmnUm7C75CyKDlRi/9FqWKytcwYMjgrG5LFRmDwmCtH9g/q0pp/j98UxtotjbBfHOmqXzoLe\naaGekpKCHTt2YMaMGSguLkZiYqJ9ncViwaFDh/Dee+/BbDZj/vz5ePzxx2G1Wjvchsgb6A1mbP3q\nR5yobIBEAkglEkilbT8SCWQXPZf+/LkE+PFMAw4cq4G1bU7S+OgQTB7TGuRRkerL7J2IvJ3TQj09\nPR1FRUXIzs6GEAKrV69GXl4empubkZWVBQDIyMiAn58f5s+fj7CwMIfbEHkDk9mKfxWdwEfbj6Cp\n2dyjzxoaE4rrx0ThujFRGBgR2EsVEpE3cFqoS6XSSyYuuXj2sYcffhgPP/zwZbch8mRWqw1ffH8a\n731WipoGIwL95Zg3YzhumhALqUQCm03AJgSsNmF/brM/x0XPWx/DQvzRv5/K1b8WEbkpDj5D5ARC\nCHxbchYb8w/j9HkdlHIpMm8cisypCQhSKV1dHhF5KYY6US/74VgN/vavQyg7VQ+pVIJbro1DdnoS\nx0knIqdjqBP1kuMVWvz934ext6wKAHDdmIH4r+nDETPAtXeiE5HvYKgT9VBljQ7v5pdiZ/EZAK3T\nkd47cwQSY/u5uDIi8jUMdaIrVNtgwLY99dj3wRew2gTio0Nw74wRGJfU39WlEZGPYqgTdYPJbMV3\nh85h+57T2Ft6HjYBREUE4r9uHY7JY6Ig5UxmRORCDHWiyxBC4OhpLbbvOYWd+85AZ2jtZ54QE4ph\nAyVYcOf1kMukLq6SiIihTtShukYjCjWnUbDnNE6fbwIA9Avywx2/GIqp18Qg7qpgaDQaBjoRuQ2G\nOtFFTGYrdh88h+17TmFfWRVsApDLpLh+bBSmXROLcYmRkDHEichNMdTJ5104vV7Qdnpd33Z6PTE2\nFNOuiUVa8tUcMIaIPAJDnXxWfZMRhZoK/Oe7U/bT62HBfph+7VBMHR+D2KuCXVwhEVH3MNTJp1it\nNmhKq/Cf78qx59B5WG0CcpkUk8dG4SaeXiciD8dQJ59w+nwTtu85hS++P436phYArfOPp0+Iw5SU\naAQH8vQ6EXk+hjp5rWajGV/vr0TBd6dw+GQdAEAdoMBtkwfjpgmxiI8OdXGFRES9i6FOXkUIgUMn\n6vCf78rx9f5KtJiskEiAcYmRSJ8Qh4mjroJSIXN1mURETsFQJ69gtliR/81JbCs6gbM1egDAgDAV\nbpoQi6njYzgHORH5BIY6eTSbTeDLfRV4J/8wquoNUMql+EVqNNInxGLUkAgO20pEPoWhTh5rb1kV\n/rbtEH6sbIBcJsXsKfG4a1oib3ojIp/FUCePc6xCi79tO4Tio9WQSIBfpEbjv6YPx4AwnmInIt/G\nUCePca5Wj3fyS/HlvgoArTe/3XfbSAy5OsTFlRERuQeGOrm9Bl0LPtx+BP8uOgmL1YYhV4dg/m0j\nkJzIecuJiC7GUCe3ZTRZkPfVj9j8xVE0Gy3oH6bC3FuH44bkq3kDHBGRAwx1cjtWqw0Fe07jvc9K\nUddoRJBKiV/9chRmXDcICjn7mBMRdYShTm6jvsmIgu9O4bNvy3G+rhlKhQx3TUtA5o0JCAxQuLo8\nIiK3x1Anl7LZBA4cq8anu8rxbclZWG0CSoUM0ycNQnZ6IsJDAlxdIhGRx2Cok0tom1qwfc8pfLa7\n3D4C3KCBwZh+bRx+kRrDI3MioivAUKc+I4TAD8dr8Omucuz6oRIWq4BSLsXU8TG4ddIgJMX1g0TC\nG+CIiK4UQ52crkHXgi++P43Pvj2JM9WtR+UxA4IwfVIcpqbGQK3iCHBERL2BoU5Oc7xCi38UHkfR\ngUpYrDYo2sZln37tIIwYHMajciKiXsZQp15nswn8c+dx/O1fh2C1CVwdqcb0SYMwdXwMx2UnInIi\nhjr1qgZdC176YB++P3we/YL88NuscUgd1p9H5UREfYChTr3mh+M1eOEdDeoajRiXGInfzUlFaJCf\nq8siIvIZDHXqMatN4MOCI/jg81JAIsG9M0fgjl8M5VCuRER9jKFOPVLXaMSL72pw4FgNIvsFYNE9\n4zF8cJiryyIi8kkMdbpie0ur8H/va9CgM2HiyKvwaPY4BLF7GhGRyzDUqdssVhveyT+Mj3ccg1wm\nxX/PHo3brh/Mm+GIiFyMoU7dUlXXjOff+R6l5fUYGBGIJ+aOx9DoUFeXRUREYKhTN+z64Sxe3rQP\neoMZN4y7Gg/dORYqf47RTkTkLhjqdFlmixX//l6L745UQKmQ4ZG7k5E+IZan24mI3AxDnTp18Mda\nrP94P8rP6RAzIAhPzhuPuKuCXV0WERE54LRQt9lsWL58OcrKyqBUKpGbm4u4uDj7+q1bt2LDhg2Q\nSqXIzMzEnDlzYDabsXjxYpw5cwZSqRQrV65EfHy8s0qkTmibWrBh20F88f1pSCTA+IRAPLngBvgr\n+f9AIiJ35bS/0AUFBTCZTNi0aROKi4uxZs0arF+/3r5+7dq12LZtG1QqFWbOnImZM2diz549sFgs\n+OCDD1BUVISXXnoJ69atc1aJ5IDVJvD5tyfxt38fht5gxpCrQ/Bg5hjoak4w0ImI3JzT/kprNBqk\npaUBAJKTk1FSUtJufVJSEpqamiCXyyGEgEQiweDBg2G1WmGz2aDT6SCXM0T60rHTWrz68X4cPa2F\nyl+O32SMxq3XDYZMKoGm5oSryyMiostwWmrqdDqo1Wr7a5lMBovFYg/qhIQEZGZmIiAgAOnp6QgO\nDoZer8eZM2dw6623or6+Hq+99lqX9qXRaHq19t7+PHdnMNnwxf4G7DnaOtf56EEq3DwuBEEB9Sje\nV29/n6+1S1exXRxjuzjGdnGM7eJYd9vFaaGuVquh1+vtr202mz3QS0tLUVhYiO3bt0OlUmHRokXI\nz89HcXExrr/+evz+97/H2bNnce+99yIvLw9+fp1PCpKamtprdWs0ml79PHcmhEDh3gr89dOD0Opa\nEN1fjQcyx2DM0MhL3utL7dIdbBfH2C6OsV0cY7s41lG7dBb0Tgv1lJQU7NixAzNmzEBxcTESExPt\n64KCguDv7w8/Pz/IZDKEhYWhsbERwcHBUCha+z2HhITAYrHAarU6q0SfdupcI9Z/cgAlx2uhVMgw\nb8ZwzJ4yFAq51NWlERHRFXJaqKenp6OoqAjZ2dkQQmD16tXIy8tDc3MzsrKykJWVhTlz5kChUCA2\nNhYZGRkwm81YunSp/U74xx9/HCqVylkl+iRjiwUf/KcMW748DqtNYOLIq/Dr2aMxIIztTETk6ZwW\n6lKpFCtWrGi37OLuaTk5OcjJyWm3XqlU4uWXX3ZWST7v+8Pn8crm/ajRGtA/TIXfzB6NCSOvcnVZ\nRETUS3h7uQ8QQmDzF0exMf8wZFIJ7r4pEXdNS2AXNSIiL8O/6l7OZLbizx8VY4emAhEh/nhqwURO\nwEJE5KUY6l6svtGIVW9/h7LyeiTF9sPS+RMQFuzv6rKIiMhJGOpe6sczDVj5192o0Rrwi5RoPHJ3\nMpQKmavLIiIiJ2Koe6FdP1Tixff2osVkxbwZw3Hn1ATOqEZE5AMY6l5ECIGPtrfeEOenlGHpfddg\n0ugoV5dFRER9hKHuJVrMVqzbVIwv91UgIjQAzyyYiCFXh7i6LCIi6kMMdS9Q12jEqg27ceSUFsPi\nWm+I6xfEG+KIiHwNQ93DHa/QIvevu1HTYMSNqdF4+C7eEEdE5KsY6h6s6EAl/vj+XpjMVtw7cwQy\nbxzKG+KIiHwYQ90DCSGwqeAI3v20FP5KGZ66bwImjhro6rKIiMjFGOoe6O//PozNXxxF/34BeHrB\nRAyO4g1xRETEUPdIu36oRKC/HC8+OgWhQZ3PNU9ERL6Dk2d7GCEEahqMGBAeyEAnIqJ2GOoeRmcw\no8VkRWRogKtLISIiN8NQ9zA1WgMAIIKhTkREP8NQ9zDVbaEeHsLBZYiIqD2Guoe5cKTO0+9ERPRz\nDHUPw9PvRETUEYa6h6lmqBMRUQcY6h6mRmuARAKEhzDUiYioPYa6h6nRGhCq9oNCzn86IiJqj8ng\nQWw2gRqtkafeiYjIIYa6B2nUm2Cx2hjqRETkEEPdg7A7GxERdYah7kF45zsREXWGoe5B7H3Ueec7\nERE5wFD3IBx4hoiIOsNQ9yAMdSIi6gxD3YNUaw2QSoCwYM6jTkREl2Koe5CaBgPCgv0hk/GfjYiI\nLsV08BBWm0BtAweeISKijjHUPYS2yQibTTDUiYioQwx1D8Gb5IiI6HIY6h6iRmsEwNHkiIioYwx1\nD8HR5IiI6HIY6h6Cp9+JiOhyGOoegqFORESXw1D3EDVaA+QyCULVHHiGiIgcY6h7iGqtAWEhAZBK\nJa4uhYiI3JTTQt1ms2HZsmXIysrC3LlzUV5e3m791q1bkZGRgczMTLz33nv25a+//jqysrJwxx13\n4KOPPnJWeR7FYrWhvsnIO9+JiKhTcmd9cEFBAUwmEzZt2oTi4mKsWbMG69evt69fu3Yttm3bBpVK\nhZkzZ2LmzJkoLS3Fvn378P7778NgMOCvf/2rs8rzKHUNRgjBKVeJiKhzTgt1jUaDtLQ0AEBycjJK\nSkrarU9KSkJTUxPkcjmEEJBIJPj666+RmJiIhx56CDqdDk888YSzyvMoNQ0XbpLzd3ElRETkzpwW\n6jqdDmq12v5aJpPBYrFALm/dZUJCAjIzMxEQEID09HQEBwejvr4elZWVeO2111BRUYEHHngAn376\nKSSSzq8jazSaXq29tz+vp3442QwAMDTVuLQ2d2sXd8F2cYzt4hjbxTG2i2PdbRenhbparYZer7e/\nttls9kAvLS1FYWEhtm/fDpVKhUWLFiE/Px+hoaEYMmQIlEolhgwZAj8/P9TV1SE8PLzTfaWmpvZa\n3RqNplc/rzeUNx4FUIeU0YlIHTXQJTW4Y7u4A7aLY2wXx9gujrFdHOuoXToLeqfdKJeSkoKdO3cC\nAIqLi5GYmGhfFxQUBH9/f/j5+UEmkyEsLAyNjY1ITU3FV199BSEEzp8/D4PBgNDQUGeV6DE4mhwR\nEXWF047U09PTUVRUhOzsbAghsHr1auTl5aG5uRlZWVnIysrCnDlzoFAoEBsbi4yMDCiVSuzZswd3\n3nknhBBYtmwZZDKZs0r0GBx4hoiIusJpoS6VSrFixYp2y+Lj4+3Pc3JykJOTc8l2vDnuUjVaA5Ry\nKYIDla4uhYiI3BgHn/EANVojwkMDLnvDIBER+TaGupszW6zQ6lo48AwREV0WQ93NXZhHndfTiYjo\ncroU6mVlZXj88ccBAMePH8c999yDH3/80amFUaufBp5hqBMRUee6FOrPPPMMZs+eDaD1ZrcHH3wQ\nTz31lFMLo1a8852IiLqqS6FuMBgwZcoU++vJkyfDYDA4rSj6yYVQ5zV1IiK6nC6FelhYGN5//33o\n9Xro9XomQBKIAAAWuUlEQVR8+OGHlx3ljXoHB54hIqKu6lKoP/vssygsLMT111+PqVOn4ssvv8Sq\nVaucXRvhotPvIZzMhYiIOtelwWeioqLw6KOPYsSIEWhqakJJSQmuuuoqZ9dGaA11f6UMgQEKV5dC\nRERurktH6i+88AJeeOEFAK3X11999VWsW7fOqYVRqxqtAREceIaIiLqgS6FeWFiIN998EwDQv39/\nbNiwAZ9//rlTCyPAaLKgqdnM6+lERNQlXQp1i8UCo9Fof202m51WEP2Ed74TEVF3dOmaenZ2Nu64\n4w5MnToVQgh89dVXuOeee5xdm89jH3UiIuqOLoV6Tk4OzGYzTCYTgoODceedd6K6utrZtfk8DhFL\nRETd0aVQf+SRR2AwGHDq1CmMHz8ee/bsQXJysrNr83kcIpaIiLqjS9fUT5w4gb///e9IT0/Hr371\nK3z00Ueoqqpydm0+j9fUiYioO7oU6uHh4ZBIJBg8eDDKysowYMAAmEwmZ9fm8y6MJhfOgWeIiKgL\nunT6PSEhAStXrkROTg4WLlyIqqoq3gHfB2q0BgT6y6Hy58AzRER0eV06Ul++fDluvfVWDB06FI88\n8giqqqrw4osvOrs2n3dh4BkiIqKu6NKRukwmw/jx4wEA06ZNw7Rp05xaFAHNRjOajRaGOhERdVmX\njtSp73F2NiIi6i6Gupvine9ERNRdDHU3xYFniIiouxjqbopDxBIRUXcx1N0UT78TEVF3MdTd1IVQ\nD2eoExFRFzHU3VS11oDgQCX8FDJXl0JERB6Coe6GhBCoaTAgIoRH6URE1HUMdTekM5jRYrLyJjki\nIuoWhrob+unOd07kQkREXcdQd0McTY6IiK4EQ90NsTsbERFdCYa6G+LAM0REdCUY6m6IoU5ERFeC\noe6GarRGSCRAOLu0ERFRNzDU3VCN1oBQtR8Ucv7zEBFR1zE13Ix94Bmeeiciom5iqLuZBp0JZouN\noU5ERN3GUHczvEmOiIiuFEPdzdgHnuFNckRE1E1OC3WbzYZly5YhKysLc+fORXl5ebv1W7duRUZG\nBjIzM/Hee++1W1dbW4spU6bg+PHjzirPbXHgGSIiulJyZ31wQUEBTCYTNm3ahOLiYqxZswbr16+3\nr1+7di22bdsGlUqFmTNnYubMmQgJCYHZbMayZcvg7++b457XNvD0OxERXRmnHalrNBqkpaUBAJKT\nk1FSUtJufVJSEpqammAymSCEgEQiAQA899xzyM7ORv/+/Z1VmlvjuO9ERHSlnHakrtPpoFar7a9l\nMhksFgvk8tZdJiQkIDMzEwEBAUhPT0dwcDA++eQThIWFIS0tDW+88UaX96XRaHq19t7+vO44WVEN\niQQ4cewgTkklLqvDEVe2iztjuzjGdnGM7eIY28Wx7raL00JdrVZDr9fbX9tsNnugl5aWorCwENu3\nb4dKpcKiRYuQn5+Pjz/+GBKJBLt27cLhw4fx5JNPYv369YiMjOx0X6mpqb1Wt0aj6dXP665X8z9H\neLA/Jlwz3mU1OOLqdnFXbBfH2C6OsV0cY7s41lG7dBb0Tgv1lJQU7NixAzNmzEBxcTESExPt64KC\nguDv7w8/Pz/IZDKEhYWhsbER7777rv09c+fOxfLlyy8b6N7EahOobTAiISbU1aUQEZEHclqop6en\no6ioCNnZ2RBCYPXq1cjLy0NzczOysrKQlZWFOXPmQKFQIDY2FhkZGc4qxWNom4yw2gSvpxMR0RVx\nWqhLpVKsWLGi3bL4+Hj785ycHOTk5HS4/caNG51VmtviwDNERNQTHHzGjdRojQAY6kREdGUY6m6E\n3dmIiKgnGOpuhKPJERFRTzDU3UgNR5MjIqIeYKi7kRqtAXKZBKFqP1eXQkREHoih7kZqtAaEhQRA\n6mYjyRERkWdgqLsJq9WG+kYjr6cTEdEVY6i7idpGI2yC86gTEdGVY6i7iZ8GnvHNKWeJiKjnGOpu\ngqPJERFRTzHU3QRDnYiIeoqh7iY4mhwREfUUQ91N1Da0jvvOu9+JiOhKMdTdRLXWAKVciuBApatL\nISIiD8VQdxM1WgPCQwMgkXDgGSIiujIMdTdgtlihbWrhqXciIuoRhrobuHA9nTfJERFRTzDU3QDv\nfCciot7AUHcD9j7qIRxNjoiIrhxD3Q1w4BkiIuoNDHU3wNPvRETUGxjqbuDCkTrvficiop5gqLuB\nWq0R/koZAgMUri6FiIg8GEPdDVRrDYjgwDNERNRDDHUXM5osaGo28Xo6ERH1GEPdxTiRCxER9RaG\nuovV1PPOdyIi6h0MdRdjdzYiIuotDHUXq2m4MJocQ52IiHqGoe5iP40mxyFiiYioZxjqLsbT70RE\n1FsY6i5WqzUg0F8OlT8HniEiop5hqLtYTdvAM0RERD3FUHehZqMZeqOFoU5ERL2Coe5CnHKViIh6\nE0PdhWq0HE2OiIh6D0PdhXjnOxER9SaGugvx9DsREfUmhroLMdSJiKg3MdRdpPxcIw6eqAUAhIdw\nNDkiIuo5ubM+2GazYfny5SgrK4NSqURubi7i4uLs67du3YoNGzZAKpUiMzMTc+bMgdlsxtKlS3Hm\nzBmYTCY88MADmDZtmrNKdIkGXQve/awUn31bDptNYOr4GPgrnfbPQEREPsRpaVJQUACTyYRNmzah\nuLgYa9aswfr16+3r165di23btkGlUmHmzJmYOXMmCgoKEBoaiueffx5arRazZ8/2mlA3W2z4V9GP\n+ODzMuiNFlwdqcb9t4/E+OEDXF0aERF5CaeFukajQVpaGgAgOTkZJSUl7dYnJSWhqakJcrkcQghI\nJBJMnz4dt9xyCwBACAGZTOas8vqMEALflpzDhm0HcbZGD3WAAr+ePQozrhsMuYxXP4iIqPc4LdR1\nOh3UarX9tUwmg8VigVzeusuEhARkZmYiICAA6enpCA4Obrftb3/7Wzz22GNd2pdGo+nV2nvr887W\nm/DZ3gacPN8CqQSYmKjGlNHBUPlpsb94X6/soy/1djt7C7aLY2wXx9gujrFdHOtuuzgt1NVqNfR6\nvf21zWazB3ppaSkKCwuxfft2qFQqLFq0CPn5+bj11ltx9uxZPPTQQ5gzZw5mzZrVpX2lpqb2Wt0a\njabHn1ffaMTG/MMo2FMFIYDxwwdgwayRiBkQ1EtV9r3eaBdvxHZxjO3iGNvFMbaLYx21S2dB77RQ\nT0lJwY4dOzBjxgwUFxcjMTHRvi4oKAj+/v7w8/ODTCZDWFgYGhsbUVNTgwULFmDZsmWYNGmSs0pz\nGpPZin/uPI6Pth+BocWK2KuCcP/to5CS1N/VpRERkQ9wWqinp6ejqKgI2dnZEEJg9erVyMvLQ3Nz\nM7KyspCVlYU5c+ZAoVAgNjYWGRkZWLt2LRobG/Hqq6/i1VdfBQC8+eab8Pd37y5fQgh8vb8Sb//r\nEKrqmhEcqMT820bi5olxkPG6ORER9RGnhbpUKsWKFSvaLYuPj7c/z8nJQU5OTrv1Tz/9NJ5++mln\nleQ0b2z5Adu+PgG5TIKMXwzF3TclQh3A+dGJiKhvsYN0D5ktNmzfcwoRoQFY9cB1iIpQX34jIiIi\nJ+C54R46dKIWhhYrJo0eyEAnIiKXYqj30N7SKgBA6jDeDEdERK7FUO8hTel5KOVSjIqPcHUpRETk\n4xjqPVBdb0D5uSaMGhoBP4Xnj35HRESejaHeA3vLeOqdiIjcB0O9BzSl5wEA44dxUhYiInI9hvoV\nslht2H+0GleFqzAwItDV5RARETHUr1TpyTo0Gy1IHTYAEonE1eUQEREx1K+Uhl3ZiIjIzTDUr5Cm\n9DzkMilGsysbERG5CYb6FahtMOBEZSNGxYfD348j7RIRkXtgqF+BfezKRkREboihfgW+t19PZ1c2\nIiJyHwz1brJabSg+Uo3+/QIQ3Z8TuBARkftgqHdT2al66A1mpLArGxERuRmGejexKxsREbkrhno3\n7S09D7lMgjFD2ZWNiIjcC0O9G+qbjDhW0YARg8Oh8le4uhwiIqJ2GOrdwK5sRETkzhjq3aBhVzYi\nInJjDPUustoE9pVVITzEH7FXBbm6HCIioksw1Lvo6Ol6NDWbOSsbERG5LYZ6F+1lVzYiInJzDPUu\n0pSeh0wqwdiESFeXQkRE5BBDvQsadC04elqLYYPCEBjArmxEROSeGOpdsO9INYTgqXciInJvDPUu\n0JSeB8CubERE5N4Y6pdha+vKFhbsh8FRwa4uh4iIqEMM9cs4fkaLBp0JKUnsykZERO6NoX4ZF0aR\nS+H1dCIicnMM9cvYW1oFqQQYl8iubERE5N4Y6p1oajahrLwOSXFhUKuUri6HiIioUwz1ThSXVcPG\nrmxEROQhGOqd0JSxKxsREXkOhnoHbDaBvaVVCFX7YcjVIa4uh4iI6LIY6h04UdmA+qYWjEuKhFTK\nrmxEROT+GOod2Ft2YVY2nnonIiLPwFDvgKa0ChIJMC6JN8kREZFnYKg7oDOYcfhkHRJj+iE4kF3Z\niIjIMzgt1G02G5YtW4asrCzMnTsX5eXl7dZv3boVGRkZyMzMxHvvvdelbfrK/qPVsNkEu7IREZFH\ncVqoFxQUwGQyYdOmTfj973+PNWvWtFu/du1abNiwAe+//z42bNiAhoaGy27TVzSH27qyDef1dCIi\n8hxyZ32wRqNBWloaACA5ORklJSXt1iclJaGpqQlyuRxCCEgkkstu0xeEENhbVoUglRLx0aF9vn8i\nIqIr5bRQ1+l0UKvV9tcymQwWiwVyeesuExISkJmZiYCAAKSnpyM4OPiy23REo9H0Wt1VDRbUNhgx\nOi4Axfv29trneoPebGdvwnZxjO3iGNvFMbaLY91tF6eFulqthl6vt7+22Wz2cC4tLUVhYSG2b98O\nlUqFRYsWIT8/v9NtOpOamtprdb+8sRAAkD55OFJTY3rtcz2dRqPp1Xb2FmwXx9gujrFdHGO7ONZR\nu3QW9E67pp6SkoKdO3cCAIqLi5GYmGhfFxQUBH9/f/j5+UEmkyEsLAyNjY2dbtNXjlUaAQDjEnmT\nHBEReRanHamnp6ejqKgI2dnZEEJg9erVyMvLQ3NzM7KyspCVlYU5c+ZAoVAgNjYWGRkZkMvll2zT\nl5qNZpyqbsHQmFCEBvn16b6JiIh6ymmhLpVKsWLFinbL4uPj7c9zcnKQk5NzyXY/36Yv7T9aw1nZ\niIjIY3HwmYvYh4ZNYlc2IiLyPAz1i0gAhAXJkRjLrmxEROR5nHb63RP9zx1jcM0gM2Qy/l+HiIg8\nD9PrIlKpBFIJp1klIiLPxFAnIiLyEgx1IiIiL8FQJyIi8hIMdSIiIi/BUCciIvISDHUiIiIvwVAn\nIiLyEgx1IiIiL8FQJyIi8hIMdSIiIi/BUCciIvISEiGEcHURPaHRaFxdAhERUZ9KTU11uNzjQ52I\niIha8fQ7ERGRl2CoExEReQmGOhERkZdgqBMREXkJhjoREZGXkLu6AHdhs9mwfPlylJWVQalUIjc3\nF3Fxca4uyy1kZGRArVYDAKKjo/Hss8+6uCLX2r9/P1544QVs3LgR5eXlWLx4MSQSCRISEvC///u/\nkEp98//KF7fLoUOH8Jvf/AaDBg0CAOTk5GDGjBmuLbCPmc1mLF26FGfOnIHJZMIDDzyAoUOH+vz3\nxVG7DBw40Oe/L1arFU8//TROnDgBiUSCP/zhD/Dz8+v294Wh3qagoAAmkwmbNm1CcXEx1qxZg/Xr\n17u6LJdraWmBEAIbN250dSlu4c0338TWrVsREBAAAHj22Wfx2GOPYeLEiVi2bBm2b9+O9PR0F1fZ\n937eLgcPHsT8+fOxYMECF1fmOlu3bkVoaCief/55aLVazJ49G8OGDfP574ujdnnooYd8/vuyY8cO\nAMAHH3yA3bt3449//COEEN3+vvjWfxE7odFokJaWBgBITk5GSUmJiytyD6WlpTAYDFiwYAHmzZuH\n4uJiV5fkUrGxsVi3bp399cGDBzFhwgQAwA033IBvvvnGVaW51M/bpaSkBIWFhbjnnnuwdOlS6HQ6\nF1bnGtOnT8ejjz4KABBCQCaT8fsCx+3C7wtw0003YeXKlQCAyspKBAcHX9H3haHeRqfT2U8xA4BM\nJoPFYnFhRe7B398f999/P9566y384Q9/wMKFC326XW655RbI5T+d4BJCQCKRAAACAwPR1NTkqtJc\n6uftMmbMGDzxxBN49913ERMTg1deecWF1blGYGAg1Go1dDodfvvb3+Kxxx7j9wWO24Xfl1ZyuRxP\nPvkkVq5ciVmzZl3R94Wh3katVkOv19tf22y2dn+kfNXgwYNx++23QyKRYPDgwQgNDUV1dbWry3Ib\nF1/f0uv1CA4OdmE17iM9PR2jRo2yPz906JCLK3KNs2fPYt68efjlL3+JWbNm8fvS5uftwu/LT557\n7jl89tlneOaZZ9DS0mJf3tXvC0O9TUpKCnbu3AkAKC4uRmJioosrcg+bN2/GmjVrAADnz5+HTqdD\nZGSki6tyHyNGjMDu3bsBADt37sT48eNdXJF7uP/++3HgwAEAwK5duzBy5EgXV9T3ampqsGDBAixa\ntAh33nknAH5fAMftwu8LsGXLFrz++usAgICAAEgkEowaNarb3xeO/d7mwt3vR44cgRACq1evRnx8\nvKvLcjmTyYQlS5agsrISEokECxcuREpKiqvLcqmKigr87ne/w4cffogTJ07gmWeegdlsxpAhQ5Cb\nmwuZTObqEl3i4nY5ePAgVq5cCYVCgYiICKxcubLd5S1fkJubi/z8fAwZMsS+7KmnnkJubq5Pf18c\ntctjjz2G559/3qe/L83NzViyZAlqampgsVjw61//GvHx8d3++8JQJyIi8hI8/U5EROQlGOpERERe\ngqFORETkJRjqREREXoKhTkRE5CUY6kTkFJ988gkWL17s6jKIfApDnYiIyEtwHFQiH/fGG28gPz8f\nVqsV119/PXJycvDggw8iJiYG5eXliIqKwvPPP4/Q0FDs2LEDL730Emw2G2JiYrBixQpERETgm2++\nwZo1ayCEQFRUFF588UUAQHl5OebOnYvKykpMmjQJubm5Lv5tibwbj9SJfNjOnTtRUlKCzZs3Y8uW\nLTh//jzy8vJw5MgR3HvvvfjXv/6F+Ph4/PnPf0ZtbS2WLVuGV155BXl5eUhJScGKFStgMpmwcOFC\nPPfcc8jLy0NSUhL+8Y9/AGgd43vdunXIz8/Hzp07cfToURf/xkTejUfqRD5s165dOHDgAO644w4A\ngNFohBACgwYNwsSJEwEAs2fPxsKFCzF58mSMGTMG0dHRAICsrCy88cYbKCsrw4ABAzB8+HAAwO9+\n9zsArdfUx48fj9DQUACt07PW19f39a9I5FMY6kQ+zGq14t5778X8+fMBAI2NjTh37hwef/xx+3su\nzHlts9nabSuEgMVigUKhaLe8qanJPuPhxTMdSiQScFRqIufi6XciH3bttdfin//8J/R6PSwWCx56\n6CGUlJTgxIkTOHz4MADg448/xg033ICxY8di//79qKioAABs2rQJEydOxODBg1FXV4djx44BAP7y\nl7/g/fffd9nvROTLeKRO5MOmTp2K0tJS3H333bBarUhLS8M111yDkJAQ/OlPf8KpU6eQlJSE3Nxc\nqFQqrFixAg8//DDMZjOioqKwatUq+Pn54fnnn8cTTzwBs9mM2NhYrF27Fp999pmrfz0in8NZ2oio\nnYqKCsybNw9ffPGFq0shom7i6XciIiIvwSN1IiIiL8EjdSIiIi/BUCciIvISDHUiIiIvwVAnIiLy\nEgx1IiIiL8FQJyIi8hL/D39JQczQx04jAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a7f1dd240>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfUAAAFlCAYAAADyLnFSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a7ee27f98>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# summarize history for accuracy\n",
    "plt.plot(history.history['main_output_acc'])\n",
    "plt.plot(history.history['val_main_output_acc'])\n",
    "plt.title('main model acc')\n",
    "plt.ylabel('acc')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()\n",
    "# summarize history for loss\n",
    "plt.plot(history.history['main_output_loss'])\n",
    "plt.plot(history.history['val_main_output_loss'])\n",
    "plt.title('main model loss')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAgAAAAFlCAYAAACdqVCOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3Xt0VPW9///nXJIQJiEXBI8WSDRyUS5C4AgUwW/xsGj5\niRirSUCJxHJWjyW0oLQaK4ghYCoIBRR6RH7tOVwEVBqbr9W6oCg9McfqsBKaxGC5hUZbISJKZpLM\nTGa+f0BGqRNumZ2ZYV6PtVgrM3v2ns9+r1nMa9577882+Xw+HyIiIhJVzKEegIiIiHQ9BQAREZEo\npAAgIiIShRQAREREopACgIiISBRSABAREYlCCgAiEjIjRoygoaHhvK9Zu3YtRUVFXTQikeihACAi\nIhKFrKEegIgEj9frZdmyZVRVVeFwOPD5fBQXFzNy5Egee+wx+vfvzw9+8AMA/+O77rqLadOmsXTp\nUm677TZWr15NZWUlGzduxGz+6jfC2rVrOXbsGH/72984fvw4w4YNY9y4cZSWltLQ0MBPf/pT7rjj\nDtxuNyUlJVRUVGCxWBg2bBiFhYUkJCTwwQcfsGTJEkwmE0OHDsXr9fq3/8c//pH169fjdrvp1q0b\njz76KCNGjOhwXxsbG1m0aBGfffYZJ06c4Fvf+ha//OUv6dmzJ0eOHGHRokWcPHkSs9nMQw89xJQp\nUzp8XiQaqQMgcgWpqqri+PHjbN++nd///vdkZWWxYcOG867Ts2dPSkpKWLhwIbt27eK3v/0tzz77\n7Dlf/u3sdjsbNmzg97//Pe+++y6HDh1iy5YtLFy4kLVr1wKwfv16jh8/zmuvvcZrr72G1+vlmWee\nweVy8ZOf/ITHHnuM0tJSRo8eTUtLCwBHjx5l1apVvPDCC5SWlrJkyRLmzp2L0+nscNyvv/46w4cP\nZ/v27ezevZtu3brx2muvAfDwww/z3e9+l9dff50XXniBlStX0tTU1OHzItFIHQCRK8iIESNISkpi\n27Zt/O1vf+O9997DZrNdcL1bb72VKVOmUFBQwObNm0lNTQ34um9/+9skJiYC0Lt3b8aPHw9Av379\nOHXqFAB79+5l/vz5xMTEADBz5kzmzJnDRx99hNVqZezYsQDccccdLFq0CIDy8nKOHz/OrFmz/O9l\nMpk4duxYh2N+4IEH+OCDD/j1r3/N0aNH+etf/8rNN9/MqVOnqKur49577wXgmmuuYdeuXR0+LxKt\n1AEQuYK8/fbb/PCHPwTg9ttvZ/r06f5lJpOJr9/6w+12+//2+XwcPHiQq666isrKyg63Hxsbe85j\nq/WbvyG+3tZvf+x2u7/x/l9f3+v1MnbsWH/X4LXXXmPHjh3079+/w7EsX76c1atXk5KSQk5ODuPG\njcPn8/m3aTKZ/K89fPgwFosl4PPtXQiRaKMAIHIFKS8v5zvf+Q4zZsxg6NCh7Nq1i7a2NgBSUlKo\nrq4G4OTJk3zwwQf+9X7zm9/Q3NzMq6++ym9+8xv2799/2WMYP34827Ztw+124/V62bJlC+PGjWPA\ngAH4fD7eeecdAHbv3s0XX3wBwJgxYygvL+fQoUMAvPPOO9x55520trZ2+D7/8z//wwMPPMBdd91F\nz549effdd2lrayMhIYHBgwdTWloKwN///nemT59OS0tLwOdPnz592fsqEsl0CEDkCpKbm8uCBQuY\nOnUqFouFUaNG8dZbb+H1epk5cyYLFixg8uTJ9OnTh1tuuQWA2tpafvWrX/HKK69w9dVX8/jjj/PI\nI4/w29/+loSEhEsew0MPPcQvfvEL7rrrLjweD8OGDWPhwoXExMTw/PPPs3jxYlauXMmNN95Iz549\nAejfvz9FRUU8/PDD/l/x69evp3v37h2+z5w5c3jmmWdYt24dFouFzMxM/yGDZ599lqeeeopNmzZh\nMplYunQpvXr16vB5kWhk0u2ARUREoo8OAYiIiEQhBQAREZEopAAgIiIShRQAREREopACgIiISBSK\nqssA7XZ7qIcgIiLSpUaOHBnw+agKANBxIS6H3W4P6vauFKpLYKpLYKpLYKpLYKpLYB3V5Xw/fHUI\nQEREJAopAIiIiEQhBQAREZEopAAgIiIShRQAREREopACgIiISBRSABAREYlCCgAh1trayssvv3xR\nr925cye7d+82eEQiIhINFABC7MSJExcdAO6++25uv/12g0ckIiLRIOpmAjyf/7+shvKqjy/69a0u\nF3FvvHXe14y7+Vs8OHVwh8t/9atfcfDgQQYNGsS3v/1tnE4nS5cupbS0lOrqak6dOsWgQYN4+umn\nWbt2LVdddRXXX389GzZsICYmhoaGBqZMmcJDDz100eMWERFRALhMTS4nLZ5WmjzO875u16HD2Mu2\ndri8ub8DS8840r/dj2PNnzLwvuE8s+8/afj4MOl3DyTeew1vrniL45tb+fjAYeI+6Ub3fyRy4EAl\nox+ZxLc8PVi75Dmq+zQEexcvW6vLRdwnO0M9jLCjugSmugSmugR2pddlTN9MZg7/fpe8lwLA1zw4\ndfB5f61/3abKV3nn0AfExcYG7f2790oEwBxjwdXUwl82v4clzkpbqwdfm/ec1yZck4TZYgaLGUuM\nJWhjEBGR6KAAcJlmDv8+N7Wld/qmFJ988gk/Lvsxtw28jauuuorpU6eze/duyro388sXfsnJkyf5\n3ve+x1MTH6H0y1L/IYBtx7axaupSAMaVvM3zZ/8OB7pZR2CqS2CqS2CqS2CqS/AoAIRYz549cbvd\ntLS0+J8bNmwY69at47777sNkMtG3b1+OHz8ewlGKiMiVRgEgxOLi4njttdfOea5Xr168+uqr33jt\n11Pv6NGj/X+Xl5cbN0AREbki6TJAERGRKKQAICIiEoUUAERERKKQAoCIiEgUUgAQERGJQgoAIiIi\nUUgBIMQu5W6A7d5//33q6uoMGpGIiEQDBYAQu5S7AbZ79dVXNTGQiIh0imETAXm9XhYvXsyBAweI\njY2luLiYtLQ0//LS0lI2btxIYmIiWVlZ3Hvvvf5lVVVVrFixgk2bNgFQX1/PY489hslkon///jz5\n5JOYzWaKi4vZt28fNpsNgHXr1pGYmHjZY95U+Sr/+7d9F/36i7kpxYVu7NB+N8DnnnuOjz76iM8/\n/xyAJ554goEDB1JYWEh9fT0tLS3k5eVxww038Kc//YmamhpuuOEGrr322oser4iISDvDAsCuXbtw\nuVxs376dyspKSkpKWL9+PQAnT55kzZo17Ny5kx49ejBr1izGjh1Lnz592LBhA7/73e+Ij4/3b+vp\np59m3rx5jB49mkWLFrF7924mTZpETU0NL774IqmpqUbthuH+4z/+g48++ojm5mbGjBnDjBkzOHr0\nKIWFhWzYsIH333+fHTt2AGdm/BsyZAjjx49nypQp+vIXEZHLZlgAsNvtjB8/HoDhw4dTXV3tX9bQ\n0MDAgQNJTk4GYOjQoVRVVdGnTx/69evH2rVr+dnPfuZ/fU1NDbfccgsAEyZMoLy8nNtvv536+noW\nLVpEY2Mj99xzD/fcc0+nxjxz+Pcv6TaMwbwpxUcffcT//u//8sYbbwDwxRdfkJCQwOOPP87ChQtp\namrizjvvDMp7iYiIGBYAmpqaSEhI8D+2WCx4PB6sVitpaWkcPHiQxsZGbDYbFRUVpKenAzB58mQa\nGs69t73P58NkMgFgs9k4ffo0TqeT+++/n/z8fNra2sjLy2PIkCEMGjTIqF0yhNlsxuv1cv3113Pn\nnXcydepUPvvsM15++WWOHz9OTU0Nzz//PK2trdx2221MmzYNk8mEz+cL9dBFRCSCGRYAEhIScDgc\n/sderxer9czbJSUlUVhYyNy5c0lOTmbw4MGkpKR0uC2z+atzFR0OBz169CA+Pp68vDz/oYIxY8ZQ\nV1d3wQBgt9s7s1tB357L5eLLL7+kvr6ev/zlL2zcuJHm5ma+//3vc+zYMT788EPuuOMOzGYzkydP\npqqqipSUFJYsWcIXX3zBt771rSDtSXAFu85XCtUlMNUlMNUlMNUlsEuti2EBIDMzkz179jBlyhQq\nKysZMGCAf5nH46G2tpatW7fidrvJz89n/vz5HW7rpptu4r333mP06NHs3buXMWPGcPToUebNm0dp\naSler5d9+/aRlZV1wXEF8z7SwToE8NZbb3W4bNSoUd94buTIkeccIgk3ul93YKpLYKpLYKpLYKpL\nYB3V5XyhwLAAMGnSJMrLy8nNzcXn87Fs2TLKyspwOp3k5OQAkJWVRVxcHPn5+ec9ke/RRx9l4cKF\nrFy5kuuvv57JkydjsViYNm0a2dnZxMTEMG3aNPr372/U7oiIiFxRDAsAZrOZoqKic57LyMjw/11Q\nUEBBQUHAdfv06eM/8x3guuuuY/Pmzd943ezZs5k9e3aQRiwiIhI9NBGQiIhIFFIAEBERiUIKACIi\nIlFIAUBERCQKKQCIiIhEIcOuApDATjtd/KnyYzweb6iHckEmkwlbvBVbtxi6x8eQEB+DrVsMtvgY\n4uOsmM2mS96m2+PF2eLG0eLG0Xz2X4sHR7ObllaPAXsRPv72t9N87DgU6mGEHdUlMNUlsCu9Ln16\nJ5I5qHeXvJcCQBf7v/9zhK1/qAv1MDrNZILucVZs8TF0PxsKEuJj6N7NypdffM4faz+gqcWNs/lr\nX/YtHlpdbaEeemjt+yLUIwhPqktgqktgV3Bd4uMsvFT8/2G5jB9Yl0oBoIudOt0CwL9PG0KvlPgL\nvDq02rw+nGd/nTu+/kXe7Pnal7qbE587Ofr3f/71fmYaaIvZhC3+TEBI7dHN/3d7J8F2NjQkxMfQ\nLdaKyfjPfMgcOnSYjIzrQz2MsKO6BKa6BHal1+WaqxK65MsfFAC6nLPlzBflmCHX0Du1e4hHEzxt\nXh/NrR6czW72Ve7nX0fejK1bDHGxFv+NnKJdrOvvjByqWzj/M9UlMNUlMNUleBQAulh7AOgeHxPi\nkQSXxWwi4exhgF5JMfRMCu/uhohItNNVAF3M0eIGID5O2UtEREJHAaCLNbd4iI+zdNkxHhERkUAU\nALqYo8VN925XVvtfREQijwJAF3O2eOjeTe1/EREJLQWALuTz+XCqAyAiImFAAaALtbrbaPP6sCkA\niIhIiCkAdKHms5cAxusQgIiIhJgCQBdqvwRQHQAREQk1BYAu5J8ESB0AEREJMQWALuQ82wHQSYAi\nIhJqCgBdSB0AEREJFwoAXcjpPwdAAUBEREJLAaALOfwdAB0CEBGR0FIA6EI6BCAiIuFCAaAL6SRA\nEREJFwoAXUgdABERCRcKAF1IEwGJiEi4UADoQpoKWEREwoUCQBdytLixmE3ExVhCPRQREYlyCgBd\nqP1WwCaTKdRDERGRKGdYAPB6vSxatIicnBxmzpxJfX39OctLS0uZOnUqM2bM4OWXXz5nWVVVFTNn\nzvQ/rq+vZ/r06cyYMYMnn3wSr9cLwI4dO7j77rvJzs5mz549Ru1K0DhbPDoBUEREwoJhAWDXrl24\nXC62b9/OI488QklJiX/ZyZMnWbNmDZs2bWLz5s2UlZXR0NAAwIYNG3jiiSdobW31v/7pp59m3rx5\nbN26FZ/Px+7duzlx4gSbNm1i27ZtbNy4kZUrV+JyuYzanaBwtrh1AqCIiIQFwwKA3W5n/PjxAAwf\nPpzq6mr/soaGBgYOHEhycjJms5mhQ4dSVVUFQL9+/Vi7du0526qpqeGWW24BYMKECbz77rvs37+f\nESNGEBsbS2JiIv369aOurs6o3em0Nq+P5tY2nQAoIiJhwbBvo6amJhISEvyPLRYLHo8Hq9VKWloa\nBw8epLGxEZvNRkVFBenp6QBMnjzZ3w1o5/P5/MfNbTYbp0+fpqmpicTERP9rbDYbTU1NFxyX3W4P\nwt5d+vaaXWcOW7hbHUEfQziKhn28HKpLYKpLYKpLYKpLYJdaF8MCQEJCAg6Hw//Y6/VitZ55u6Sk\nJAoLC5k7dy7JyckMHjyYlJSUDrdlNn/VqHA4HPTo0eMb23c4HOcEgo6MHDnycnYnILvdftHbO/65\nE/iEa//lqqCOIRxdSl2iieoSmOoSmOoSmOoSWEd1OV8oMOwQQGZmJnv37gWgsrKSAQMG+Jd5PB5q\na2vZunUrq1ev5vDhw2RmZna4rZtuuon33nsPgL179zJq1CiGDRuG3W6ntbWV06dPc+jQoXPeI9z4\nZwGM0yEAEREJPcO+jSZNmkR5eTm5ubn4fD6WLVtGWVkZTqeTnJwcALKysoiLiyM/P5/U1NQOt/Xo\no4+ycOFCVq5cyfXXX8/kyZOxWCzMnDmTGTNm4PP5mD9/PnFxcUbtTqc5ms/OAhivkwBFRCT0DAsA\nZrOZoqKic57LyMjw/11QUEBBQUHAdfv06cOOHTv8j6+77jo2b978jddlZ2eTnZ0dpBEbq7n17CyA\n6gCIiEgY0ERAXUQdABERCScKAF3E2apzAEREJHwoAHQR59kOQHd1AEREJAwoAHSR9g6AZgIUEZFw\noADQRfwdAM0EKCIiYUABoIs4WtoDgDoAIiISegoAXcQ/EZA6ACIiEgYUALqIZgIUEZFwogDQRZyt\nbuJiLVgsKrmIiISevo26iLPZg03tfxERCRMKAF3E0eLWCYAiIhI2FAC6iLPFoxMARUQkbCgAdAGX\nuw1Pm1cdABERCRsKAF1AlwCKiEi4UQDoAs6zkwBpGmAREQkXCgBd4KsOgAKAiIiEBwWALvDVNMA6\nBCAiIuFBAaALOHUfABERCTMKAF1AJwGKiEi4UQDoAg6dBCgiImFGAaALNJ/tAMSrAyAiImFCAaAL\nOM4GAN0LQEREwoUCQBfQSYAiIhJuFAC6gE4CFBGRcKMA0AV0EqCIiIQbBYAu0NziwWw2ERdrCfVQ\nREREAAWALuFocdM9zorJZAr1UERERAAFgC7hbPHo+L+IiIQVBYAu4Gxx6woAEREJK4b9LPV6vSxe\nvJgDBw4QGxtLcXExaWlp/uWlpaVs3LiRxMREsrKyuPfeeztcp6amhieffJLY2FhuvPFGfv7zn2M2\nmykuLmbfvn3YbDYA1q1bR2JiolG7dFm8Xh/NrR5s8QoAIiISPgwLALt27cLlcrF9+3YqKyspKSlh\n/fr1AJw8eZI1a9awc+dOevTowaxZsxg7diy1tbUB11m4cCFPPPEEmZmZrFq1irKyMqZNm0ZNTQ0v\nvvgiqampRu1Gp7W4PPh8EB+nQwAiIhI+DDsEYLfbGT9+PADDhw+nurrav6yhoYGBAweSnJyM2Wxm\n6NChVFVVdbjOp59+SmZmJgCZmZnY7Xa8Xi/19fUsWrSI3NxcXnnlFaN2pVMcze2zAKoDICIi4cOw\nANDU1ERCQoL/scViweM582WYlpbGwYMHaWxspLm5mYqKCpxOZ4fr9O3blz//+c8A7Nmzh+bmZpxO\nJ/fffz/Lly/nxRdfZOvWrdTV1Rm1O5fN2do+C6A6ACIiEj4M+1ZKSEjA4XD4H3u9XqzWM2+XlJRE\nYWEhc+fOJTk5mcGDB5OSktLhOsuWLWPp0qU8//zzjBo1itjYWOLj48nLyyM+Ph6AMWPGUFdXx6BB\ng847LrvdHtT9vND2jp1oBeDLLz4L+nuHs2ja10uhugSmugSmugSmugR2qXUxLABkZmayZ88epkyZ\nQmVlJQMGDPAv83g81NbWsnXrVtxuN/n5+cyfP5+2traA67zzzjusWLGClJQUlixZwoQJEzh69Cjz\n5s2jtLQUr9fLvn37yMrKuuC4Ro4cGbR9tNvtF95e3afACTLS+zBy5IDzv/YKcVF1iUKqS2CqS2Cq\nS2CqS2Ad1eV8ocCwADBp0iTKy8vJzc3F5/OxbNkyysrKcDqd5OTkAJCVlUVcXBz5+fmkpqYGXAfO\nHDKYNWsW8fHxjB49mttuuw2AadOmkZ2dTUxMDNOmTaN///5G7c5lc549B6C7TgIUEZEwYti3ktls\npqio6JznMjIy/H8XFBRQUFBwwXUAJk6cyMSJE7/x/OzZs5k9e3aQRmwM/zkAugxQRETCiCYCMphD\nHQAREQlDCgAGc7aoAyAiIuFHAcBgzlZ1AEREJPwoABjM0XymA6CpgEVEJJwoABisub0DoJkARUQk\njCgAGKy9A6CZAEVEJJwoABjM2eohNsaC1aJSi4hI+NC3ksGczW79+hcRkbCjAGAwZ4sHmwKAiIiE\nGQUAgzlb3MTrBEAREQkzCgAGcnu8uDxedQBERCTsKAAYyD8LoDoAIiISZhQADORsaZ8DQB0AEREJ\nLwoABnKc7QDY1AEQEZEwowBgoOazHYB4dQBERCTMKAAYSB0AEREJVwoABvrqHAAFABERCS8KAAb6\n6ioAHQIQEZHwogBgoPYOgA4BiIhIuFEAMJA6ACIiEq4UAAzk0DwAIiISphQADKSZAEVEJFwpABhI\nMwGKiEi4UgAwkLPFjckE8XEKACIiEl4UAAzkbPHQPc6KyWQK9VBERETOoQBgIGeLm+7xOv4vIiLh\nRwHAQI6zHQAREZFwowBgEJ/PR3OLW1cAiIhIWFIAMEiLqw2vT1cAiIhIeFIAMIhTdwIUEZEwZlgA\n8Hq9LFq0iJycHGbOnEl9ff05y0tLS5k6dSozZszg5ZdfPu86NTU13HPPPcyYMYMlS5bg9XoB2LFj\nB3fffTfZ2dns2bPHqF25LP45AHQSoIiIhCHDAsCuXbtwuVxs376dRx55hJKSEv+ykydPsmbNGjZt\n2sTmzZspKyujoaGhw3UWLlzI448/ztatW0lISKCsrIwTJ06wadMmtm3bxsaNG1m5ciUul8uo3blk\njvZZAHUSoIiIhCHDAoDdbmf8+PEADB8+nOrqav+yhoYGBg4cSHJyMmazmaFDh1JVVdXhOp9++imZ\nmZkAZGZmYrfb2b9/PyNGjCA2NpbExET69etHXV2dUbtzyb7qACgAiIhI+DHs26mpqYmEhAT/Y4vF\ngsfjwWq1kpaWxsGDB2lsbMRms1FRUUF6enqH6/Tt25c///nP3HLLLezZs4fm5maamppITEz0v9Zm\ns9HU1HTBcdnt9qDuZ0fbqznmBOCz4//Abr/wuK40wa7zlUJ1CUx1CUx1CUx1CexS62JYAEhISMDh\ncPgfe71erNYzb5eUlERhYSFz584lOTmZwYMHk5KS0uE6y5YtY+nSpTz//POMGjWK2NjYb7zW4XCc\nEwg6MnLkyKDto91u73B7je564CSDBlzHyJH9gvaekeB8dYlmqktgqktgqktgqktgHdXlfKHgog4B\nHDhwgPnz5wNw6NAh7rvvPg4fPnzedTIzM9m7dy8AlZWVDBgwwL/M4/FQW1vL1q1bWb16NYcPHyYz\nM7PDdd555x1WrFjBf/3Xf3Hq1CnGjRvHsGHDsNvttLa2cvr0aQ4dOnTOe4Rac6vuBCgiIuHrojoA\nCxcuZM6cOQBkZGTwox/9iJ///Oe89NJLHa4zadIkysvLyc3NxefzsWzZMsrKynA6neTk5ACQlZVF\nXFwc+fn5pKamBlwHIC0tjVmzZhEfH8/o0aO57bbbAJg5cyYzZszA5/Mxf/584uLiOlWMYHI0606A\nIiISvi7q26m5udn/pQswbtw4li9fft51zGYzRUVF5zyXkZHh/7ugoICCgoILrgMwceJEJk6c+I3n\ns7Ozyc7Ovphd6HJOdQBERCSMXdQhgNTUVF566SUcDgcOh4MdO3bQs2dPo8cW0ZzqAIiISBi7qADw\n9NNP8/bbb3PrrbcyceJE3nnnHZYuXWr02CJaewdAMwGKiEg4uqifp9deey0/+clPuOmmmzh9+jTV\n1dX8y7/8i9Fji2jqAIiISDi7qA7AihUrWLFiBXDmfIB169axdu1aQwcW6RwtbmKsZmKsllAPRURE\n5BsuKgC8/fbbbNiwAYDevXvz61//mrfeesvQgUU6Z4tH7X8REQlbFxUAPB4PLS0t/sdut9uwAV0p\nnC1u4tX+FxGRMHVR31C5ubncfffdTJw4EZ/Px5/+9Cfuu+8+o8cW0ZytHnomdQv1MERERAK6qAAw\nffp03G43LpeLHj16cM8993DixAmjxxaxPG1eWl1tmgNARETC1kUFgLlz59Lc3MyxY8cYNWoU77//\nPsOHDzd6bBGruVVXAIiISHi7qHMAjhw5wn//938zadIkZs+ezcsvv8zx48eNHlvEcjRrFkAREQlv\nFxUAevbsiclk4rrrruPAgQNcffXVuFwuo8cWsZwt6gCIiEh4u6hvqP79+7NkyRKmT5/OggULOH78\nuK4EOA9ni2YBFBGR8HZRHYDFixfzve99jxtuuIG5c+dy/Phxnn32WaPHFrHUARARkXB3Ud9QFouF\nUaNGAXD77bdz++23GzqoSNfeAdA5ACIiEq4uqgMgl8ahDoCIiIQ5BQADqAMgIiLhTgHAAO3nAOgk\nQBERCVcKAAZw+DsAOgQgIiLhSQHAAM3+cwDUARARkfCkAGAAdQBERCTcKQAYoP0cgPg4BQAREQlP\nCgAGcLa4iY+zYjabQj0UERGRgBQADOBo8WBT+19ERMKYAoABmlvcdI/XCYAiIhK+FACCzOfz4Wjx\n0F3H/0VEJIwpAARZq7sNr9enDoCIiIQ1BYAg898JUB0AEREJYwoAQdZ+HwCbOgAiIhLGFACCTHMA\niIhIJFAACDJHszoAIiIS/gz7mer1elm8eDEHDhwgNjaW4uJi0tLS/MtLS0vZuHEjiYmJZGVlce+9\n93a4zocffsiTTz6JxWIhPT2dpUuXYjabKS4uZt++fdhsNgDWrVtHYmKiUbt0UZyt7fcBUAdARETC\nl2HfUrt27cLlcrF9+3YqKyspKSlh/fr1AJw8eZI1a9awc+dOevTowaxZsxg7diy1tbUB13nuueeY\nM2cOt912G4888ghvv/02EydOpKamhhdffJHU1FSjduOSOc92ALrHqQMgIiLhy7BDAHa7nfHjxwMw\nfPhwqqur/csaGhoYOHAgycnJmM1mhg4dSlVVVYfr3HjjjZw6derMNfYOB1arFa/XS319PYsWLSI3\nN5dXXnnFqF25JO0dAFu8OgAiIhK+DPuWampqIiEhwf/YYrHg8XiwWq2kpaVx8OBBGhsbsdlsVFRU\nkJ6e3uFf0CJvAAAWJElEQVQ66enpFBUVsX79ehITExk9ejROp5P777+f/Px82trayMvLY8iQIQwa\nNOi847Lb7UHdz3/e3sHDXwLQcOwIdtffg/pekSTYdb5SqC6BqS6BqS6BqS6BXWpdDAsACQkJOBwO\n/2Ov14vVeubtkpKSKCwsZO7cuSQnJzN48GBSUlI6XGfp0qVs2bKF/v37s2XLFkpKSnjiiSfIy8sj\nPj4egDFjxlBXV3fBADBy5Mig7aPdbv/G9io/rga+ZPiwm+jfNyVo7xVJAtVFVJeOqC6BqS6BqS6B\ndVSX84UCww4BZGZmsnfvXgAqKysZMGCAf5nH46G2tpatW7eyevVqDh8+TGZmZofrJCUl+TsDvXv3\n5ssvv+To0aNMnz6dtrY23G43+/btY/DgwUbtzkXzXwXQTecAiIhI+DKsAzBp0iTKy8vJzc3F5/Ox\nbNkyysrKcDqd5OTkAJCVlUVcXBz5+fmkpqYGXAeguLiY+fPnY7VaiYmJYcmSJfTp04dp06aRnZ1N\nTEwM06ZNo3///kbtzkXzzwOgqwBERCSMGfYtZTabKSoqOue5jIwM/98FBQUUFBRccB2AUaNGsW3b\ntm88P3v2bGbPnh2kEQeHfyZAdQBERCSMaSKgIHO2eLBaTMRYVVoREQlf+pYKMmerm+7dYjCZTKEe\nioiISIcUAILM0ezRLIAiIhL2FACCrPlsB0BERCScKQAEUZvXR3Nrm04AFBGRsKcAEETNZ68A0CEA\nEREJdwoAQdQ+B4ACgIiIhDsFgCBy+DsAOgQgIiLhTQEgiNQBEBGRSKEAEEROdQBERCRCKAAEUXsH\nwKYOgIiIhDkFgCBSB0BERCKFAkAQOXQOgIiIRAgFgCBSB0BERCKFAkAQ6SoAERGJFAoAQdTeAdBU\nwCIiEu4UAILI3wGIVwAQEZHwpgAQRO0BID5OhwBERCS8KQAEkaPFTXycBYvZFOqhiIiInJcCQBA5\nW9zEx6n9LyIi4U8BIIicLR5s8Wr/i4hI+FMACBKfz4ezxU13dQBERCQCKAAEidvjxdPm0xwAIiIS\nERQAgsTRPgugLgEUEZEIoAAQJM3tcwDoEkAREYkACgBB0t4BsKkDICIiEUABIEiczeoAiIhI5FAA\nCBJnq84BEBGRyKEAECQOdQBERCSCGBYAvF4vixYtIicnh5kzZ1JfX3/O8tLSUqZOncqMGTN4+eWX\nz7vOhx9+SHZ2NtOnT6ewsBCv1wvAjh07uPvuu8nOzmbPnj1G7cpFUQdAREQiiWEBYNeuXbhcLrZv\n384jjzxCSUmJf9nJkydZs2YNmzZtYvPmzZSVldHQ0NDhOs899xxz5szhpZdewuVy8fbbb3PixAk2\nbdrEtm3b2LhxIytXrsTlchm1OxfUfiMgm+YBEBGRCGBYALDb7YwfPx6A4cOHU11d7V/W0NDAwIED\nSU5Oxmw2M3ToUKqqqjpc58Ybb+TUqVP4fD4cDgdWq5X9+/czYsQIYmNjSUxMpF+/ftTV1Rm1Oxfk\naD7bAeimDoCIiIQ/w36uNjU1kZCQ4H9ssVjweDxYrVbS0tI4ePAgjY2N2Gw2KioqSE9P73Cd9PR0\nioqKWL9+PYmJiYwePZo333yTxMRE/2ttNhtNTU0XHJfdbg/qfrZv71jD5wAcOXSA0ycUAoJd5yuF\n6hKY6hKY6hKY6hLYpdbFsACQkJCAw+HwP/Z6vVitZ94uKSmJwsJC5s6dS3JyMoMHDyYlJaXDdZYu\nXcqWLVvo378/W7ZsoaSkhFtvvfWc1zocjnMCQUdGjhwZtH202+3+7e2qeR9w8K8jh5Pao1vQ3iMS\nfb0u8hXVJTDVJTDVJTDVJbCO6nK+UGDYIYDMzEz27t0LQGVlJQMGDPAv83g81NbWsnXrVlavXs3h\nw4fJzMzscJ2kpCR/Z6B37958+eWXDBs2DLvdTmtrK6dPn+bQoUPnvEdXc7aevQpA5wCIiEgEMOzb\natKkSZSXl5Obm4vP52PZsmWUlZXhdDrJyckBICsri7i4OPLz80lNTQ24DkBxcTHz58/HarUSExPD\nkiVL6NWrFzNnzmTGjBn4fD7mz59PXFycUbtzQc5mNxazibgYS8jGICIicrEMCwBms5mioqJznsvI\nyPD/XVBQQEFBwQXXARg1ahTbtm37xvPZ2dlkZ2cHacSd42z10L2bFZPJFOqhiIiIXJAmAgoSZ7Nb\nVwCIiEjEUAAIEkeLR8f/RUQkYigABIHX66O51aMOgIiIRAwFgCBo1hUAIiISYRQAguCraYDVARAR\nkcigABAEzpb2aYDVARARkcigABAE7R0AnQMgIiKRQgEgCBzqAIiISIRRAAiCrw4BqAMgIiKRQQEg\nCL46CVAdABERiQwKAEGgDoCIiEQaBYAg+OokQHUAREQkMigABEH7SYC2eHUAREQkMigABEF7ByA+\nTh0AERGJDAoAQeBUB0BERCKMAkAQqAMgIiKRRgEgCJwtbuJiLVgtKqeIiEQGfWMFgaPFozkAREQk\noigABEFzi4f4OB3/FxGRyKEAEASOFje2eHUAREQkcigAdJLb04bb46W7OgAiIhJBFAA6yT8LoDoA\nIiISQRQAOsl/K2B1AEREJIIoAHSSOgAiIhKJFAA6yT8LoO4EKCIiEUQBoJN0J0AREYlECgCd1N4B\n6K4OgIiIRBAFgE5SB0BERCKRAkAnOdQBEBGRCKQA0EnOZnUAREQk8hj2reX1elm8eDEHDhwgNjaW\n4uJi0tLS/MtLS0vZuHEjiYmJZGVlce+993a4zvz582lsbATg448/5uabb2bVqlUUFxezb98+bDYb\nAOvWrSMxMdGoXQrI2XomAOgqABERiSSGBYBdu3bhcrnYvn07lZWVlJSUsH79egBOnjzJmjVr2Llz\nJz169GDWrFmMHTuW2tragOusWrUKgC+++IK8vDwKCwsBqKmp4cUXXyQ1NdWo3bggZ3P7IQB1AERE\nJHIYdgjAbrczfvx4AIYPH051dbV/WUNDAwMHDiQ5ORmz2czQoUOpqqo67zoAa9eu5f7776d37954\nvV7q6+tZtGgRubm5vPLKK0btynm1dwB0DoCIiEQSw362NjU1kZCQ4H9ssVjweDxYrVbS0tI4ePAg\njY2N2Gw2KioqSE9PP+86n332GRUVFf5f/06nk/vvv5/8/Hza2trIy8tjyJAhDBo06LzjstvtQd3P\nTxs/x2SCmr9UYjKZgrrtSBbsOl8pVJfAVJfAVJfAVJfALrUuhgWAhIQEHA6H/7HX68VqPfN2SUlJ\nFBYWMnfuXJKTkxk8eDApKSnnXefNN9/kjjvuwGKxABAfH09eXh7x8fEAjBkzhrq6ugsGgJEjRwZt\nH+12O2ZLHN27+Rg1alTQthvp7HZ7UOt8pVBdAlNdAlNdAlNdAuuoLucLBYYdAsjMzGTv3r0AVFZW\nMmDAAP8yj8dDbW0tW7duZfXq1Rw+fJjMzMzzrlNRUcGECRP8j48ePcr06dNpa2vD7Xazb98+Bg8e\nbNTudMjZ6sGm4/8iIhJhDPvmmjRpEuXl5eTm5uLz+Vi2bBllZWU4nU5ycnIAyMrKIi4ujvz8fFJT\nUwOu0+7IkSP07dvX/zgjI4Np06aRnZ1NTEwM06ZNo3///kbtToeczW56pXTv8vcVERHpDMMCgNls\npqio6JznMjIy/H8XFBRQUFBwwXXavf766994bvbs2cyePTsIo708Pp8PZ6tHVwCIiEjE0URAneDy\n+PD5dAWAiIhEHgWATmh1+wDNASAiIpFHAaATWtxeQLMAiohI5FEA6IRW15kAoA6AiIhEGgWATvjq\nEIA6ACIiElkUADrhq0MA6gCIiEhkUQDohPYOQLw6ACIiEmEUADqhVR0AERGJUAoAndAeAHQOgIiI\nRBoFgE5o0TwAIiISoRQAOkEdABERiVQKAJ2gmQBFRCRSKQB0QotLHQAREYlMCgCd0Or2Ems1E2NV\nGUVEJLLom6sTWt0+usfr17+IiEQeBYBOaHV76R6n4/8iIhJ5FAA6oUUdABERiVAKAJfJ0+bF0+bT\nLIAiIhKRFAAuk7PFA+gKABERiUwKAJfJ2eIGNAeAiIhEJgWAy6QOgIiIRDIFgMvkUAdAREQimALA\nZWpu7wDEqQMgIiKRRwHgMrV3AGzx6gCIiEjkUQC4TM7m9kMA6gCIiEjkUQC4TM7W9pMA1QEQEZHI\nowBwmXqldMdqMdGnd2KohyIiInLJ9PP1Mv2fzD50b/sHV6d2D/VQRERELpk6AJ1gMZtCPQQREZHL\nYlgHwOv1snjxYg4cOEBsbCzFxcWkpaX5l5eWlrJx40YSExPJysri3nvv7XCd+fPn09jYCMDHH3/M\nzTffzKpVq9ixYwfbtm3DarXy0EMP8Z3vfMeo3REREbmiGBYAdu3ahcvlYvv27VRWVlJSUsL69esB\nOHnyJGvWrGHnzp306NGDWbNmMXbsWGprawOus2rVKgC++OIL8vLyKCws5MSJE2zatIlXX32V1tZW\nZsyYwbhx44iNjTVql0RERK4Yhh0CsNvtjB8/HoDhw4dTXV3tX9bQ0MDAgQNJTk7GbDYzdOhQqqqq\nzrsOwNq1a7n//vvp3bs3+/fvZ8SIEcTGxpKYmEi/fv2oq6szandERESuKIZ1AJqamkhISPA/tlgs\neDwerFYraWlpHDx4kMbGRmw2GxUVFaSnp593nc8++4yKigoKCwv9209M/OoMfJvNRlNT0wXHZbfb\ng7iXwd/elUJ1CUx1CUx1CUx1CUx1CexS62JYAEhISMDhcPgfe71erNYzb5eUlERhYSFz584lOTmZ\nwYMHk5KSct513nzzTe644w4sFkvA7TscjnMCQUdGjhwZlP2DM8UO5vauFKpLYKpLYKpLYKpLYKpL\nYB3V5XyhwLBDAJmZmezduxeAyspKBgwY4F/m8Xiora1l69atrF69msOHD5OZmXnedSoqKpgwYYL/\n8bBhw7Db7bS2tnL69GkOHTp0zutFRESkY4Z1ACZNmkR5eTm5ubn4fD6WLVtGWVkZTqeTnJwcALKy\nsoiLiyM/P5/U1NSA67Q7cuQIffv29T/u1asXM2fOZMaMGfh8PubPn09cXJxRuyMiInJFMSwAmM1m\nioqKznkuIyPD/3dBQQEFBQUXXKfd66+//o3nsrOzyc7ODsJoRUREoosmAhIREYlCCgAiIiJRSAFA\nREQkCpl8Pp8v1IPoKrp2VEREok1Hl01GVQAQERGRM3QIQEREJAopAIiIiEQhBQAREZEopAAgIiIS\nhRQAREREopBhUwFfybxeL4sXL+bAgQPExsZSXFxMWlpaqIcVFrKysvy3dO7Tpw9PP/10iEcUWlVV\nVaxYsYJNmzZRX1/PY489hslkon///jz55JOYzdGZwb9el9raWn74wx+Snp4OwPTp05kyZUpoB9jF\n3G43jz/+OB9//DEul4uHHnqIG264Ieo/L4Hqcs0110T956WtrY0nnniCI0eOYDKZeOqpp4iLi7vk\nz4sCwGXYtWsXLpeL7du3U1lZSUlJCevXrw/1sEKutbUVn8/Hpk2bQj2UsLBhwwZ+97vfER8fD8DT\nTz/NvHnzGD16NIsWLWL37t1MmjQpxKPsev9cl5qaGvLz83nwwQdDPLLQ+d3vfkdycjLLly/n1KlT\n3HXXXQwaNCjqPy+B6jJnzpyo/7zs2bMHgG3btvHee++xatUqfD7fJX9eoitOBondbmf8+PEADB8+\nnOrq6hCPKDzU1dXR3NzMgw8+SF5eHpWVlaEeUkj169ePtWvX+h/X1NRwyy23ADBhwgTefffdUA0t\npP65LtXV1bz99tvcd999PP744zQ1NYVwdKHx3e9+l5/85CcA+Hw+LBaLPi8Eros+L/Bv//ZvLFmy\nBIBPPvmEHj16XNbnRQHgMjQ1Nfnb3AAWiwWPxxPCEYWHbt268YMf/ICNGzfy1FNPsWDBgqiuy+TJ\nk7Fav2qy+Xw+TCYTADabjdOnT4dqaCH1z3UZNmwYP/vZz9iyZQt9+/bl+eefD+HoQsNms5GQkEBT\nUxM//vGPmTdvnj4vBK6LPi9nWK1WHn30UZYsWcLUqVMv6/OiAHAZEhIScDgc/sder/ec/9Ci1XXX\nXcedd96JyWTiuuuuIzk5mRMnToR6WGHj68fjHA4HPXr0COFowsekSZMYMmSI/+/a2toQjyg0/v73\nv5OXl8e0adOYOnWqPi9n/XNd9Hn5yi9+8Qv+8Ic/sHDhQlpbW/3PX+znRQHgMmRmZrJ3714AKisr\nGTBgQIhHFB5eeeUVSkpKAPj0009pamqiV69eIR5V+Ljpppt47733ANi7dy+jRo0K8YjCww9+8AP2\n798PQEVFBYMHDw7xiLpeY2MjDz74ID/96U+55557AH1eIHBd9HmB0tJS/vM//xOA+Ph4TCYTQ4YM\nueTPi+4FcBnarwL46KOP8Pl8LFu2jIyMjFAPK+RcLheFhYV88sknmEwmFixYQGZmZqiHFVINDQ08\n/PDD7NixgyNHjrBw4ULcbjfXX389xcXFWCyWUA8xJL5el5qaGpYsWUJMTAxXXXUVS5YsOecQWzQo\nLi7mjTfe4Prrr/c/9/Of/5zi4uKo/rwEqsu8efNYvnx5VH9enE4nhYWFNDY24vF4+Pd//3cyMjIu\n+f8XBQAREZEopEMAIiIiUUgBQEREJAopAIiIiEQhBQAREZEopAAgIiIShRQARCQs7Ny5k8ceeyzU\nwxCJGgoAIiIiUUjz14rIJXnhhRd44403aGtr49Zbb2X69On86Ec/om/fvtTX13PttdeyfPlykpOT\n2bNnD7/85S/xer307duXoqIirrrqKt59911KSkrw+Xxce+21PPvsswDU19czc+ZMPvnkE8aOHUtx\ncXGI91bkyqUOgIhctL1791JdXc0rr7xCaWkpn376KWVlZXz00Uc88MADvP7662RkZPDcc8/x2Wef\nsWjRIp5//nnKysrIzMykqKgIl8vFggUL+MUvfkFZWRkDBw7kt7/9LXBm3ve1a9fyxhtvsHfvXv76\n17+GeI9FrlzqAIjIRauoqGD//v3cfffdALS0tODz+UhPT2f06NEA3HXXXSxYsIBx48YxbNgw+vTp\nA0BOTg4vvPACBw4c4Oqrr+bGG28E4OGHHwbOnAMwatQokpOTgTO3Df7888+7ehdFooYCgIhctLa2\nNh544AHy8/MB+PLLL/nHP/7B/Pnz/a9pv2+71+s9Z12fz4fH4yEmJuac50+fPu2/u+bX76ppMpnQ\nTOUixtEhABG5aGPGjOG1117D4XDg8XiYM2cO1dXVHDlyhA8//BCAV199lQkTJnDzzTdTVVVFQ0MD\nANu3b2f06NFcd911nDx5koMHDwLw4osv8tJLL4Vsn0SilToAInLRJk6cSF1dHdnZ2bS1tTF+/Hj+\n9V//laSkJNasWcOxY8cYOHAgxcXFdO/enaKiIgoKCnC73Vx77bUsXbqUuLg4li9fzs9+9jPcbjf9\n+vXjmWee4Q9/+EOod08kquhugCLSKQ0NDeTl5fHHP/4x1EMRkUugQwAiIiJRSB0AERGRKKQOgIiI\nSBRSABAREYlCCgAiIiJRSAFAREQkCikAiIiIRCEFABERkSj0/wAsMn8ZaNQZigAAAABJRU5ErkJg\ngg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a7eaacf28>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfsAAAFlCAYAAADs50HhAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a7ebd6940>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# summarize history for accuracy\n",
    "plt.plot(history.history['aux_output_acc'])\n",
    "plt.plot(history.history['val_aux_output_acc'])\n",
    "plt.title('aux model acc')\n",
    "plt.ylabel('acc')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()\n",
    "# summarize history for loss\n",
    "plt.plot(history.history['aux_output_loss'])\n",
    "plt.plot(history.history['val_aux_output_loss'])\n",
    "plt.title('aux model loss')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0.42073906556483609, 0.27739576336395494, 0.28668660750067043, 0.90921545346606991, 0.085843370954340831, 0.90921545346606991, 0.08344433604527722]\n"
     ]
    }
   ],
   "source": [
    "model.load_weights('lstm.hdf5')\n",
    "score = model.evaluate({'main_input': X_test_time, 'aux_input': X_test_aux},\n",
    "          {'main_output': y_test, 'aux_output': y_test},verbose=0)\n",
    "# loss and the accuracy.\n",
    "print(score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "y_score = model.predict({'main_input': X_test_time, 'aux_input': X_test_aux})\n",
    "y_pred = y_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import roc_curve, auc\n",
    "from sklearn import metrics\n",
    "def compute_roc(y_test,y_score,method):\n",
    "    fpr = dict()\n",
    "    tpr = dict()\n",
    "    roc_auc = dict()\n",
    "    n_classes = 2\n",
    "    fpr, tpr, _ = metrics.roc_curve(y_test.ravel(), y_score.ravel())\n",
    "    # Compute micro-average ROC curve and ROC area\n",
    "    roc_auc = auc(fpr, tpr)\n",
    "    plt.figure()\n",
    "    plt.plot(fpr, tpr, label=method+' (AUC = %0.4f)' % roc_auc)\n",
    "    plt.plot([0, 1], [0, 1], color='navy', linestyle='--')\n",
    "    plt.xlim([0.0, 1.0])\n",
    "    plt.ylim([0.0, 1.05])\n",
    "    plt.xlabel('False Positive Rate')\n",
    "    plt.ylabel('True Positive Rate')\n",
    "    plt.title('Receiver operating characteristic example')\n",
    "    plt.legend(loc=\"lower right\")\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfYAAAFlCAYAAAAZGcpRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n+wUNAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a7ed373c8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# main auc\n",
    "compute_roc(y_test,y_score[0],'deep learning')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfYAAAFlCAYAAAAZGcpRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\niV0IIYTQI5LYhRBCCD3y/9Qt/KNBw+5TAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a7edac2b0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# aux auc\n",
    "from sklearn.metrics import precision_score\n",
    "compute_roc(y_test,y_score[1],'deep learning')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/juan/anaconda/lib/python3.6/site-packages/sklearn/metrics/classification.py:1135: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.0"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "precision_score(y_test, (y_pred[0]>0.5).astype(int))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.metrics import recall_score\n",
    "recall_score(y_test, (y_pred[0]>0.5).astype(int))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "             precision    recall  f1-score   support\n",
      "\n",
      "          0       0.91      1.00      0.95      9955\n",
      "          1       0.00      0.00      0.00       994\n",
      "\n",
      "avg / total       0.83      0.91      0.87     10949\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/juan/anaconda/lib/python3.6/site-packages/sklearn/metrics/classification.py:1135: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import classification_report\n",
    "print(classification_report(y_test, (y_pred[0]>0.5).astype(int)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[9955    0]\n",
      " [ 994    0]]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "print(confusion_matrix(y_test, (y_pred[0]>0.5).astype(int)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.1702847794938529"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.metrics import average_precision_score\n",
    "average_precision_score(y_test, y_score[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average precision-recall score: 0.17\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfYAAAFlCAYAAAAZGcpRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3Xl4FFXC9uGn052VJEAUUERQwuICCsQNMSogy4CKiBJA\nlnFj+FQYHQR0EGQQ2eQd5x1UkHEcRhwRRJBFUQygKK8ohs3IpgiRICQBAiF7uru+P850h5iFBGlC\nit99XbmSrjpddfqku546VaeqHZZlWQIAALYQVN0VAAAAZw7BDgCAjRDsAADYCMEOAICNEOwAANgI\nwQ4AgI24qrsCsL+lS5fqn//8pxwOh8LDwzV27Fi1bt260s9/5pln1Lx5cz388MMBq2OnTp0UHBys\nsLAwORwOFRUVqUOHDnrmmWcUFPTb939Xr16tr776Ss8991y5ZcaOHauePXvq5ptv/s3rk0y7rV+/\nXjExMZIkr9er3Nxc9evXT48++ugZWcfJBg0apAceeECtWrXSXXfdpc2bN5/xdQTK1KlT9fbbbysx\nMVEXXXSRf/qgQYN04MABRUVF+d8XrVq10oQJExQeHl7hMvft26c///nPOnbsmCIiIjRt2jTFxsaW\nW37RokVKTEzU7NmzJUlz5szRhx9+6J9/9OhR5eTkaNOmTcrLy9Nzzz2n7du3y+v1atSoUbrjjjsk\nSR999JFmzZolSapbt64mTpyoyy677HSbBjWRBQTQnj17rA4dOlhpaWmWZVnWZ599Zt12221VWsaY\nMWOsN954IwC1K9axY0dr27Zt/scFBQVW3759rXnz5gV0vYFUVrsdOHDAuvbaa60ff/zxjK9v4MCB\n1sqVK639+/dbbdq0OePLD5T8/HzrxhtvtP70pz9ZL730Uol5vtfk4/V6reHDh1tTp0495XL79Olj\nLVu2zLIs877v0aOH5fV6S5XLzMy0xo0bZ1177bXW0KFDy1zW8ePHra5du1qfffaZZVmWNW3aNOu5\n556zLMv8Tzt06GAdPHjQysjIsK6//nrrl19+sSzLsubNm2c99NBDlWgF2Ak9dgRUSEiIJk2apPr1\n60uSWrVqpcOHD6uwsFAhISElyubk5GjSpEnatGmTnE6n7rjjDj311FMlyixatEgLFixQUVGRjh8/\nrkcffVQDBgxQRkaGxowZo8zMTEnSbbfdpieffLLc6ZWpd1xcnH766SelpqbqgQceUGxsrA4cOKB5\n8+YpNTVVM2bMUF5enhwOh4YPH66OHTtKkl5//XUtWbJELpdLTZo00dSpU/Xpp5/qk08+0euvv65V\nq1Zp1qxZcjgccjqdGj16tK6//np/j7d79+5KTEzUK6+8Io/Ho8jISD377LO65pprNHPmTB04cEAZ\nGRk6cOCAYmJi9PLLL6tBgwaV+n8cOnRIkhQZGSlJ2rRpU5Veh9Pp1IQJE7Rv3z4dP35ctWrV0owZ\nM9S0adNKrX/t2rX629/+Jq/Xq4iICP3lL39RZGRkiR5+amqq//HixYu1aNEi5eXlKTIyUkVFRfr9\n73+v7t27S5JmzJghy7I0atQovffee5o/f768Xq/q1KmjcePGKTY2Vt99952ee+45LV26tMw6ffjh\nh2rcuLF+//vf6+GHH9bjjz9ebm/c4XDoxhtv1Lp165SVlaVBgwaVKtO9e3fde++9+umnn9SzZ09J\n5n33l7/8Rdu3b9fVV19dovzKlStVv359jR49Wp9//nmZ6502bZri4+N12223SZISExM1Y8YMSVLD\nhg11yy23aOXKlXrwwQe1fv16BQcHy+1268CBA6pTp86p/i2wGYIdAdWoUSM1atRIkmRZlqZMmaJO\nnTqVCnVJ+vvf/66CggJ99NFH8ng8euihh/TNN9/45+fk5Oi9997TnDlzVLduXW3ZskUPPvigBgwY\noIULF6pRo0Z68803lZubq7Fjx+rEiRPlTo+Kiqqw3mlpaVq7dq1/J+DQoUP6n//5H1133XU6fvy4\nnn32Wf3zn/9Uo0aNlJaWpr59+6ply5basWOHFi9erIULF6p27dqaMmWK3n777RLBO336dM2YMUNt\n2rTRl19+qa+//lrXX3+9f/6ePXv0/PPP691339Wll16qr776So899pg+/vhjSdK3336rDz74QJGR\nkRo2bJgWLFigESNGlPk65s6dq2XLliknJ0cnTpxQXFycZs+erQYNGpzW67j88ssVHR2thQsXSpLG\njx+v//znPxo3btwp3wuHDx/WqFGjNG/ePF155ZVatWqVZsyYoQkTJlT4vB9//FFr1qxRZGSk3n//\nfS1ZskTdu3eXx+PRsmXL9NZbb+mbb77RBx98oP/85z8KDw/Xl19+qeHDh+ujjz5S69atyw11SZo/\nf7569eql1q1bq169elqyZIkGDBhQZtnjx49r5cqV6tSpk6Kjo8td7pYtW1S/fv0Sp3EaNGigQ4cO\nlQr2/v37S5IWL15c5rJ++OEHJSYmKjEx0T/t4MGDuvjii0stW5KCg4P13XffadiwYcrPz9cbb7xR\n7muHPRHsOCtyc3P1zDPP6NChQ+VuaP7v//5Pzz77rJxOp5xOp95++21J0pIlSyRJtWrV0uzZs/X5\n559r37592rlzp3JzcyVJ8fHxGjp0qA4ePKibb75ZI0eOVFRUVLnTy/L0008rLCxMXq9XwcHBuv/+\n+9WtWzelpqbK5XKpTZs2ksxGOyMjQ48//rj/uQ6HQ7t27dJXX32l7t27q3bt2pKkZ599VlLJjXbP\nnj31xBNP6LbbblOHDh1Kne/esGGDbrrpJl166aWSpPbt2ysmJkbJycmSpBtuuMHf477qqqt0/Pjx\nctvd1wvNzc3VU089paCgIP9OxOm8Dkm69NJLNW/ePKWkpOibb75R27Zty13/yTZt2qTmzZvryiuv\nlCR17dpVXbt2VWpqaoXPa9mypf/1/u53v9P06dOVkZGh7du3q0mTJrrsssu0cOFCpaSkqF+/fv7n\nHT9+XMeOHauwx/r9999rx44dmjNnjiTpnnvu0VtvvaX+/fvL4XBIMjtis2bNkvXfu2937NhRgwcP\nrrDHfuONN5a5PqfTWeFrLctbb72lgQMHlnjfWmXcCfzknYjWrVtr/fr1Wrdunf7whz8oMTFR0dHR\nVV43aiaCHQH3yy+/aNiwYYqNjdVbb72lsLAwSVKvXr38ZSZNmiSXy+XfmEqmV+IrK5lec0JCgvr2\n7au4uDh1795da9eulSRdc801/gFqGzZs0P33369XX31V7dq1K3f6r82YMaPcQX0hISFyuczHxePx\nKDY2Vu+9955/flpammJiYrRhw4YSryErK0tZWVkllvXUU0/pvvvu05dffqnFixdrzpw5JYK/rI22\nZVlyu92SVKJNHA6HLMvyH272+XVPMiIiQtOnT1ePHj30r3/9S4888shpvY5169Zp4cKFeuCBB3TX\nXXepTp06pwxmH6fTWWKZlmVp165dioqKKvGai4qKStX95L+7deumFStWaPPmzbr//vslmYGBvXr1\n0qhRo/yP09PT/Tsm5XnnnXfkcrnUp08fSZLb7VZ6errWrVvnP+w9evRo/6H/k1XUYz906JAOHz4s\ny7L8rzktLa3EwLzK8Hg8WrVqld5///0S0y+++GJlZGSoXr16kqT09HRdccUVSktL0+7duxUfHy9J\nuvXWWxUZGamff/5ZrVq1qtK6UXNxuRsC6tixYxo4cKC6du2ql19+uUQoLV261P/TunVrtW/fXkuW\nLJHX61VhYaFGjBihjRs3+ssnJycrJiZGjz32mOLj4/2h7vF4NGPGDL322mu64447NHbsWDVr1kz7\n9u0rd/pv0aZNG6WkpPjrtmPHDnXr1k3p6em6+eab9emnnyo7O1uSNHPmTM2dO9f/XLfbrU6dOik3\nN1f9+/fX888/rz179vhDW5JuuukmrV+/Xvv375ckffXVVzp48KCuvfbacuvkO9zs+ylL7dq1NWbM\nGL366qtKS0s7rdfx5Zdfqnfv3rr//vt1+eWXa82aNfJ4PJVqt2uvvVZ79uzRDz/8IMlcKTBq1ChF\nR0erqKhIP/74oyTp008/rXA5ffv21eLFi7V582Z169ZNktShQwd9+OGHSk9Pl2QOrw8ZMqTC5WRl\nZenDDz/U7NmztWbNGq1Zs0br1q3T3XffXeJ/djouuugiNW7cWB999JEk6YsvvlBQUJBatGhRpeXs\n3r1b0dHR/tNZPp07d9aCBQskmZ2IL774Qh07dlRhYaGeeuoppaSkSDJHf9xud4Wj8WE/9NgRUPPn\nz9fBgwf16aeflthgz507V3Xr1i1R9oknntCLL76oXr16yePxqEePHuratavWrFkjyWy8Fy1apO7d\nuys8PFzXXHONYmJilJKSoiFDhuiZZ57RnXfeqZCQELVs2VJ33nmnjh8/Xub03yImJkZ///vfNX36\ndBUUFMiyLE2fPl2XXHKJLrnkEv3444/+86bNmjXTCy+8oFWrVkmSXC6X/vznP+vpp5/2H6GYPHly\niTEHzZo10/PPP68nnnhCHo9HYWFhmj179inHBVTG3Xffrffee09Tp07Vyy+/XOXXsXPnTo0fP16L\nFy+W0+nU1Vdfrd27d1dq3RdeeKFmzJihMWPG+AcFvvzyy4qKitKoUaP06KOPKiYmpsze8clatWol\nl8ulbt26KTQ0VJI5FfPoo4/qoYceksPhUGRkpF555RU5HI5yB88tWbJEsbGxuummm0pM/3//7/+p\nZ8+elX5d5fnrX/+qcePGadasWQoJCdH//u//+g+X9+rVS5MmTTrlZZ/79u3TJZdcUmr68OHDNWHC\nBPXs2VMej0ejRo1S48aNJUkvvviihg8fLofDoejoaM2ePfuUl+bBXhxWWcf9AABAjcSheAAAbIRg\nBwDARgh2AABshGAHAMBGCHYAAGykxlzulpSUVN1VAADgrIuLi6tS+RoT7FLVXxyqJikpiTY+C2jn\nwKONA482PjtOp1PLoXgAAGyEYAcAwEYIdgAAbIRgBwDARgh2AABshGAHAMBGCHYAAGyEYAcAwEYI\ndgAAbCSgwb5161YNGjSo1PQ1a9aoT58+SkhI0MKFCwNZBQAAzisBu6XsP/7xDy1btkzh4eElphcV\nFWnKlClatGiRwsPD1b9/f3Xq1EkXXnhhoKoCAMB5I2A99saNG2vmzJmlpu/Zs0eNGzdW7dq1FRIS\nori4OG3cuDFQ1QAA4LwSsB57t27dlJqaWmp6dna2oqKi/I9r1aql7OzsSi3z66+T5KpRX1tT8/At\nemcH7Rx4tHHg0cbnprMek5GRkcrJyfE/zsnJKRH0FYmLiyPYA4hvazo7aOfAo40DjzY+O2rEt7vF\nxsYqJSVFx44dU2Fhob799lu1bdv2bFcDAABbOmv93+XLlys3N1cJCQl65pln9PDDD8uyLPXp00cN\nGjQ4W9UAAMDWAhrsjRo18l/Odtddd/mnd+rUSZ06dQrkqgEAOC9xgxoAAGyEYAcAwEYIdgAAbIRg\nBwDARgh2AABshGAHAMBGCHYAAGyEYAcAwEYIdgAAbIRgBwDARgh2AABshGAHAMBGCHYAAGyEYAcA\nwEYIdgAAbIRgBwDARgh2AABshGAHAMBGCHYAAGyEYAcAwEYIdgAAbIRgBwDARgh2AABshGAHAMBG\nCHYAAGyEYAcAwEYIdgAAbIRgBwDARgh2AABshGAHAMBGCHYAAGyEYAcAwEYIdgAAbIRgBwDARgh2\nAABshGAHAMBGCHYAAGyEYAcAwEYIdgAAbIRgBwDARgh2AABshGAHAMBGCHYAAGyEYAcAwEYIdgAA\nbIRgBwDARgh2AABshGAHAMBGCHYAAGwkYMHu9Xo1fvx4JSQkaNCgQUpJSSkxf9myZerdu7f69Omj\nd955J1DVAADgvOIK1IITExNVWFioBQsWaMuWLZo6dapmzZrlnz99+nStWLFCERER6tmzp3r27Kna\ntWsHqjoAAJwXAhbsSUlJio+PlyS1adNGycnJJea3bNlSJ06ckMvlkmVZcjgcgaoKAADnjYAFe3Z2\ntiIjI/2PnU6n3G63XC6zyubNm6tPnz4KDw9Xly5dFB0dfcplJiUlyRWwGkMybYzAo50DjzYOPNr4\n3BSwmIyMjFROTo7/sdfr9Yf6zp079dlnn2n16tWKiIjQqFGjtHLlSv3ud7+rcJlxcXEEewAlJSUp\nLi6uuqthe7Rz4NHGgUcbnx2ns/MUsMFz7dq107p16yRJW7ZsUYsWLfzzoqKiFBYWptDQUDmdTsXE\nxCgrKytQVQEA4LwRsP5vly5dtH79evXr10+WZWny5Mlavny5cnNzlZCQoISEBA0YMEDBwcFq3Lix\nevfuHaiqAABw3ghYsAcFBWnixIklpsXGxvr/7t+/v/r37x+o1QMAcF7iBjUAANgIwQ4AgI0Q7AAA\n2AjBDgCAjRDsAADYCMEOAICNEOwAANgIwQ4AgI0Q7AAA2AjBDgCAjRDsAADYCMEOAICNEOwAANgI\nwQ4AgI0Q7AAA2AjBDgCAjRDsAADYCMEOAICNEOwAANgIwQ4AgI0Q7AAA2AjBDgCAjRDsAADYCMEO\nAICNEOwAANgIwQ4AgI0Q7AAA2AjBDgCAjRDsAADYCMEOAICNEOwAANgIwQ4AgI0Q7AAA2AjBDgCA\njRDsAADYCMEOAICNEOwAANgIwQ4AgI0Q7AAA2AjBDgCAjRDsAADYCMEOAICNEOwAANgIwQ4AgI0Q\n7AAA2AjBDgCAjRDsAADYCMEOAICNEOwAANiIK1AL9nq9mjBhgnbt2qWQkBBNmjRJTZo08c/ftm2b\npk6dKsuyVK9ePb300ksKDQ0NVHUAADgvBKzHnpiYqMLCQi1YsEAjR47U1KlT/fMsy9K4ceM0ZcoU\nzZ8/X/Hx8Tpw4ECgqgIAwHkjYD32pKQkxcfHS5LatGmj5ORk/7y9e/eqTp06mjt3rn744Qfddttt\natq0aaCqAgDAeSNgwZ6dna3IyEj/Y6fTKbfbLZfLpczMTG3evFnjx49X48aNNWzYMLVq1Urt27ev\ncJlJSUlyBazGkEwbI/Bo58CjjQOPNj43VSomDxw4oLffflvHjx+XZVn+6VOmTCn3OZGRkcrJyfE/\n9nq9cv03levUqaMmTZooNjZWkhQfH6/k5ORTBntcXBzBHkBJSUmKi4ur7mrYHu0ceLRx4NHGZ8fp\n7DxV6hz7k08+KUm67rrrdMMNN/h/KtKuXTutW7dOkrRlyxa1aNHCP+/SSy9VTk6OUlJSJEnffvut\nmjdvXuXKAwCAkirV/3W73RozZkyVFtylSxetX79e/fr1k2VZmjx5spYvX67c3FwlJCToxRdf1MiR\nI2VZltq2bavbb7/9dOoPAABOUqlgj4uL05o1a3TLLbcoJCSkUgsOCgrSxIkTS0zzHXqXpPbt22vR\nokVVqCoAADiVSgX7xx9/rLfffrvENIfDoR07dgSkUgAA4PRUKti//PLLQNcDAACcAZUK9ry8PL3y\nyiv66quv5PF4dNNNN+mPf/yjIiIiAl0/AABQBZUaFT9x4kTl5eVp8uTJmjZtmoqKivT8888Hum4A\nAKCKKtVj//7777Vs2TL/4/Hjx6tHjx4BqxQAADg9leqxW5alrKws/+OsrCw5nc6AVQoAAJyeSvXY\nf//73+u+++5Tp06dZFmW1q5dq6FDhwa6bgAAoIoqFex9+vRR69attXHjRnm9Xs2cOVMtW7YMdN0A\nAEAVVXgofu3atZKkDz74QNu3b1etWrUUFRWlHTt26IMPPjgrFQQAAJVXYY/9u+++U8eOHfX111+X\nOf+ee+4JSKUAAMDpqTDYR4wYIankt7idOHFChw4d4ktbAAA4B1VqVPx7772nZ599VkePHlXPnj01\nYsQIvfzyy4GuGwAAqKJKBfv8+fM1ZswYrVixQp07d9by5cv1xRdfBLpuAACgiioV7JJUp04dff75\n57r99tvlcrlUUFAQyHoBAIDTUKlgb9asmf7whz8oNTVV7du31x//+Ee1atUq0HUDAABVVKnr2CdP\nnqzNmzerefPmCgkJUa9evXTbbbcFum4AAKCKKgz2BQsWKCEhQbNnz5akEpe9bd++XU888URgawcA\nAKqkwkPxlmWdrXoAAIAzoMIee79+/SRJw4YN0+eff67OnTvr6NGjWrNmjfr06XNWKggAACqvUoPn\nxo0bp1WrVvkff/3113wfOwAA56BKDZ5LTk7W8uXLJUkxMTF66aWXdNdddwW0YgAAoOoq1WP3er1K\nT0/3Pz5y5IiCgip9CTwAADhLKtVjHzZsmHr37q24uDhZlqVt27Zp7Nixga4bAACookoF+1133aUb\nbrhBW7Zskcvl0rhx41S/fv1A1w0AAFRRpY6nFxYWasmSJVq9erVuuOEGLVy4UIWFhYGuGwAAqKJK\nBfvEiROVm5ur7du3y+Vy6eeff+ZQPAAA56BKBfv333+vP/3pT3K5XAoPD9e0adO0Y8eOQNcNAABU\nUaWC3eFwqLCwUA6HQ5KUmZnp/xsAAJw7KjV4bvDgwXrwwQeVkZGhF198UYmJiXr88ccDXTcAAFBF\nlQr2W2+9Va1atdLXX38tj8ejWbNm6Yorrgh03QAAQBVVKtgfeOABrVy5Us2aNQt0fQAAwG9QqWC/\n4oor9MEHH+iaa65RWFiYf3rDhg0DVjEAAFB1lQr2rVu3atu2bSW+xtXhcGj16tUBqxgAAKi6CoM9\nLS1NL7zwgiIiItSuXTs9/fTTio6OPlt1AwAAVVTh5W5//vOf1bRpU40ePVpFRUWaMmXK2aoXAAA4\nDafssf/zn/+UJLVv31733HPPWakUAAA4PRX22IODg0v8ffJjAABw7qnSl6pztzkAAM5tFR6K/+GH\nH9S5c2f/47S0NHXu3FmWZTEqHgCAc1CFwf7JJ5+crXoAAIAzoMJgv+SSS85WPQAAwBlQpXPsAADg\n3EawAwBgIwQ7AAA2QrADAGAjBDsAADZCsAMAYCMEOwAANkKwAwBgIwELdq/Xq/HjxyshIUGDBg1S\nSkpKmeXGjRunGTNmBKoaAACcVwIW7ImJiSosLNSCBQs0cuRITZ06tVSZd999V7t37w5UFQAAOO8E\nLNiTkpIUHx8vSWrTpo2Sk5NLzN+0aZO2bt2qhISEQFUBAIDzToX3iv8tsrOzFRkZ6X/sdDrldrvl\ncrmUnp6uV199Va+88opWrlxZ6WUmJSXJFbAaQzJtjMCjnQOPNg482vjcFLCYjIyMVE5Ojv+x1+uV\n67+p/PHHHyszM1NDhw5VRkaG8vPz1bRpU917770VLjMuLo5gD6CkpCTFxcVVdzVsj3YOPNo48Gjj\ns+N0dp4CFpPt2rXT2rVr1aNHD23ZskUtWrTwzxs8eLAGDx4sSVq8eLF++umnU4Y6AAA4tYAFe5cu\nXbR+/Xr169dPlmVp8uTJWr58uXJzczmvDgBAgAQs2IOCgjRx4sQS02JjY0uVo6cOAMCZww1qAACw\nEYIdAAAbIdgBALARgh0AABsh2AEAsBGCHQAAGyHYAQCwEYIdAAAbIdgBALARgh0AABsh2AEAsBGC\nHQAAGyHYAQCwEYIdAAAbIdgBALARgh0AABsh2AEAsBGCHQAAGyHYAQCwEYIdAAAbIdgBALARgh0A\nABsh2AEAsBGCHQAAGyHYAQCwEYIdAAAbIdgBALARgh0AABsh2AEAsBGCHQAAGyHYAQCwEYIdAAAb\nIdgBALARgh0AABsh2AEAsBGCHQAAGyHYAQCwEYIdAAAbIdgBALARgh0AABsh2AEAsBGCHQAAGyHY\nAQCwEYIdAAAbIdgBALARgh0AABsh2AEAsBGCHQAAGyHYgd8oP186fFgqKJDcbjPNsiSvVzp+3Mz3\n8Xiqp44Azh+uQC3Y6/VqwoQJ2rVrl0JCQjRp0iQ1adLEP3/FihX697//LafTqRYtWmjChAkKCmI/\nA9Xj2DFp714TxqfLsiSH49TlduyIKDUtKkpq0MDsGBQVSQcOmOnR0VJWlhQebn4uu6xy6wBw/gpY\nsCcmJqqwsFALFizQli1bNHXqVM2aNUuSlJ+fr7/97W9avny5wsPD9ac//Ulr165V586dA1UdwC87\nW0pNlXJySk53u6UjRySn0zyuU0fKzDTBWlQkhYZKwcGmnGWZaZLpqRcVmd64y2WCNzS0uFx+vglo\nj8cs+4cfwhQebsrWrm12Jg4dkk6cMOUl83dWllmG12umBwWZ6ZGRUkyMWVZ4uPnt9Zp65OSYsjk5\nUkSEqX9OTnHdo6PNb5fLzA8JOXvtDuDsCFiwJyUlKT4+XpLUpk0bJScn++eFhITo3XffVXh4uCTJ\n7XYrNDT0lMts1uxvpaaNGnWzHn/8BknSoEFL9MUXKaXK3HRTI7377n2SpH/8I0kvvvhFmcvfvXu4\nQkKc2rnzsLp3f7vMMm+8cbfuuKOpJOmGG/6h9PScUmWGDLlWf/lLR0nS00+v0qJF20uVufzyulq7\ndogkaenSnfrjHz8uc31ffvmQGjWKVmZmntq2fb3MMpMnd9aAAa0lSXfe+Y6Sk9NLlenevZlmz75T\nkjR9+nq99trGUmWcTo/27ImTJG3YkKp+/RaVub733++ruLiGkqTmzWeqqKj08eUnn7xJTz55kyTp\n4YeXavXqvaXKtGt3sRYvTpAkzZ27RRMmfFbm+pKTH1NkZIh++ilTnTr9u8wys2b11O9+11ySdMst\nb2r//ixJJmQ9HvPTrVsrPfHEHZKkmTMT9fHHybIslfiJiYnWmDEPyeORVqz4QQsXfqigIPN8X+gG\nBUkzZw5R06Z1VVhYqISE18qs09Cht+uuu9rIsqTRoxdo586DkqT8fLecTvPRa9Hicj34YC9J0ttv\nb9CqVRvkcJh6O53mt8vl1OzZw3XwoPTpp79o2rSF/p2Pk3vvkyffp1atGiknRxoy5DUVFBSWOgJx\n333X68EHO8jjkaZNW6ENG36UVLwzY1lS06b19e9/D5DXK61e/Z1eeGG1goJKHynYvPkPqls3XKmp\nWbrlljfLbIP//d/u6tXrCklSx47/1t69maXK3HffVZoxo6sk6fnn1+rf/95aqkz9+rX0zTePSpIS\nE3/SI49LcJ5wAAAVhElEQVQsK3N9H388UFdccaGKiry67LLS2wtJGjs2Xo8+at7n/fot0oYNqaXK\nxMc30bx5vSVJr776jV566f/KXNa+fU9Kkr77Lk133TW/zDJvvdVbt95qjla2aTNbx47llyrzyCPt\n9Nxzt0qSRoxYqWXLdpUq07Llhfrkk4GSpEWLtuvpp1eVub6vv35EDRpEKi0tWzfe+EaZZWbM6Kr7\n7rtKktSt29vatetwqTJ3391Sf//77yRJkyat0xtvbCoxv7CwUPXrb9SWLcMkSevWpWjw4CVlrm/5\n8v5q3bqBJJX7f2FbXva2/P3348ssW5GABXt2drYiIyP9j51Op9xut1wul4KCgnThhRdKkubNm6fc\n3Fx16NDhlMssKCgstXH5+ef9SkoyW6UjR46osLCw1POOHs1UUlKSJCklJaXMMpK0adMmBQcHad++\n7HLL7N69W3Xrmo1TTk5OmeUOHjzoX19aWlqZZU6cOOEv8+OPh8pd37Zt25SWFq6srMJyy+zdu1dJ\nSWbe8ePHyyyXkZHhX19qamqZZcLDnf4yO3dmlru+HTt2SDIhVVCQL7fbKlVm//79SkoKliQdPlz2\n/+XYsWP+9e3bt7/c9W3evFkRES6lppZub3Mu26EvvkjRkSPu/77WPOXmFpUqd+DAUSUm/qjCwiD9\n8MNx5ee7FRQkOZ2m/iZQ8xQaukOSFBZ2UF5vUalw9Hql1NQfVVgYqfz8IhUVlVyXzy+//KIdO8wO\n64kTJ/zlTIiav12u4woL2/Hf6WkKCipelu98vGV5dPToDuXluZSefkSFhW7/a/J9HhwO6fPPf9G+\nfUWKjnYrP7+wVDmvV9q166iWLt0rr9ehvXtPKD/f7V9fUJBph2PHsrV8+W55vQ5t335AeXlF/nWc\nbOHC3YqKClZR0Qnl5RX996iCJctySLLkcEibN+9TSEihatXy6MSJE2X+j9PS0vzvg4MHD5ZZJidH\n/jK7d2eU+175/vvvlZNjtjvllUlJSdF/F6WjR8t+nx85csS/vp9/Lv+9WfwZziq3zK5du1SrlgnO\nvLw8FRaWfr8cOPCLf1np6ellLisrK8tfZs+eX8pd39at23TBBaE6cqSg3DJ79uxRUlKef7lllUtP\nT/ev78CBsteXl5fnL7NrV9mfc0navn27CgvNDlR5ZdiWV25bXhkOy7JKb5XPgClTpujaa69Vjx49\nJEm33nqr1q1b55/v9Xr10ksvae/evXr55Zf9vffyJCUl6dpr4+QK2K4IkpKSFBcXV93VKFd2tnT0\nqJSRUXK6L3izs6XcXDNgzeMxh6YdDhNsHo/UsKE5j+3rnVaXHTt26Morr6zeSsicIti/37RHbm5x\ne9WqJV10kZSXZ04HHDtWPN3lMr+dzuJTDUFBZprbLV1wgWlrt1sKCysuY448mEP/v95WhYSY/+EF\nF5jyLpepS3CwmRcdbX5X5bN/rr+X7YA2PjtOp50DFpPt2rXT2rVr1aNHD23ZskUtWrQoMX/8+PEK\nCQnRa6+9xqC5GqCgwJynPZOOHpX27Ss+xH2yCy8054d/PYrc6zXTjh4155uLikxvzus1G/7mzU0o\n4dTCwkx7lad27ZJtmZdn2t0XvmFhJuCDgsyOQVqa9PPPxf8Lt7v4SIFlmeWFhZnl+E43uN3671ET\nM14gJMQEvNdb/PywsOI6NGxYvIMRHs5AQqAsAQv2Ll26aP369erXr58sy9LkyZO1fPly5ebmqlWr\nVlq0aJGuu+46DRlizk0MHjxYXbp0CVR1UAVZWdIPP5Q97+qrTZDm5prBZVFRJed7veb5TqcZnHX8\nuBltXh6PxwT4sWNmYx0cbJ5bVGQ2+BkZxYPKiorMeoODzQY+NpYN+9kUHi5dcknZ82rXli6+uOLn\n+3YIfs2yzI5jQYF08KDZOfCd069Vy/TwCwqK/3Y6Sx91qVvXvI98gwH37QtVw4bFR2yioooHEFb2\n6gWgpgpYsAcFBWnixIklpsXGxvr/3rlzZ6BWjSrwes0h2X37Sl6G5Xab3tnRo2ZjevJBFV8P+5df\nKnd4tLDQXOftW5fHUzyKvKBAatzYlMvONmVdLnOI2OuVmjQxP4zervnKe6/4euVhYWYHoSxut7R7\nt7RtW/HAQslcHRAVJaWnm/eOx2N+HzgQqujo4qMHtWuXDnPfUQXJLM/jKT5qIEn16pXcgQgJ4X2I\nmoEz1ueBjAwTwlLxhuzXfJdLpaWZMM/LM5dVNWliejq7dplQ3rbNlPN6i8+L+v72ek2vKjOzuGcV\nbMbPyek0G9KLLzYb4uo+z42axeWSrrqq5LT8fCklxbzXGjY0O5++ywyzsszAC4fDvJ9TU0053+kk\n3xUHvh69221C23dUwOUqXqbviohf75g4HOYz4jsKEB1tHoeEFC8bqA4Eew2Ul2c2Hr5w9HikHTtM\n4EpmA1S/fnGY+57ju/TryJHiQE5LM4e6vV5pz55w1a8vlTWu68orSx5K9XrNcny9p+Bgs5yLL5Yu\nvzywrx+QTA+/ZcvS04ODpYgIr2JizOMLLjj1stxu8xkJDTXvbd/4j6Cg4vsHFBYWD+jzek1Zl6v4\n77p1S49DqVXL/PYdDbjkkjM/VgX4NYK9GqWnmw9+RETFe/eHD5ueya/57kommR5DRobZINWvbzZC\nRUXmfKXHY85N5+YW35Tk4EHzu23b4uVFROSpYcPy63FyjyUoyByqPNlJVzcCNYrLVXK8SEWfA6l4\nYJ9lmc9aRoY5muVymc9lXp7ZCfCNFfANDjx82HwGg4JMubAwM1aFq31wJvF2Oos8HmnLlpLTCgtN\nwPoOM2ZkmA1AaGjJAWz5+eZwotNpeuZHj5rnREWZkD56tLjnsHu3eU6tWqYH/d9bBgA4Q04OYper\neCxIWSzL7FTv3WvGjtSta3aC69Uz24Tg4OJD96Gh5vNtWWZ6vXrmc52ZacYJWJbZPkhmuxEZyWkt\nlFajgn379pr9Js7P990oxRzG9h3+i4govt2ox1O8l+/rcWdnmw97RIR5ztVXFx9+921Urrmm5LoY\n+QucG3yj+1u1Kr4CYPfu4sGnHo8J9Ojo4m2AL8AzM812IyjIHLXz3TPAd4ngyYP5mjc3ywBqVLDv\n2VOzg93jKb72umFDs+eekmKmbd1aPPAsP9980ENCzA1VmjYtHdJXX13xugh14NzjuwLg1zvi+fnm\n0lDfNftZWWZbEB5upmVnm0P2vstDPZ7i7xLwnevPzzdl69c3y8jJMX9HRJgybBPOHzUq2Bs0sN/A\nk2bNzAC2Q4fM3747evkG7QCwP9/lfj7lXfYnmQ7BySzLd7mq6Rz88os5fB8UZHYCXK7ibUm9emYH\nwXf3QNhTjQp2u2rQwPz4MJAGQGU5HMVXovjOz7vdprOwZ0/xzZ6Cg03Q+w7516tn7gPAoFf7IUIA\nwCZOPqLZrFnx35ZlDucfOWLCPjTUDKz1jcj3XXtfUGDCvm7dkufyUbMQ7ABgcw6HuYImKsoMtj1x\nwoT8/v2mBx8UZELd6TSnBC68sHj0ve/5UvFdJ6OjpZ9/DlWDBubcft26xZf/+cqffPMe3x0AcXbQ\n1ABwHnE4TDCXNYK+sNBcOnvggBmg57uxjm+ex1P8TXt794YqKMg8rlPHlPF9k2JwcPFdJ08WHm5O\nAZx8Zz7fZXsM7jtzCHYAgCQTsqf6dkTfNyweP+5R7drmyp78/OJD9r5vXMzPNyPyfaP6JdOzT083\nf/sO8/vuXBkUVNyzb9my5GBCVA3BDgCoNN8VOyEhlqKizPX5leF2m0v68vKKD8/n5JjTAlFRxffo\nDw42OwUulxkL4BvB73Sacr5r/H039kFpBDsAIOBcrlPft9/jMXfY/O473z3/Tbg7nebUQXi42SHw\nHR3wHdL3BbzHY+7H73QW3677fESwAwDOCU5nydvzFhSYQ/tZWeY6fV9P3e0u7sVLJth9Xy994oQp\n47spT2io6eX7bvBTr56Z73LZ91v4CHYAwDkpNNT8REae+ot5vF5zqD8ry5zTz84uHpTncpmdApfL\nfBFPWWEeHGx2ImJiir+C1/d10zVNDawyAAAlBQWZwXll3ZmvoMD06DMzpV27TLCHhJjz/VLx+Xqv\n1wR77dolR/X7vs2yvKsJzjUEOwDAtnz35w8LO3Uo5+WZb9hMSzPl8/KKdwL27zehf/L1/b6jCF6v\n2Rk4V+7iR7ADACAT2o0bl5zm6/EfOiT99JMJet/gvKwsc/jeN2Lfd+le/fpmZyA6unoO5RPsAACU\nw9fjv+yy4mn5+dLBgybUvV5zF78ffzQ99vBw0+sPDzfhHhZm7uVfq9bZqzPBDgBAFfjC2qdePfPb\n7Tahn50t7d1bfJe/nBzTk7/oouLAD2TQE+wAAJwBLpfptUdGmhD3eMyd+bZsMefnjx41806+617T\npmd+QB7BDgBAAPiCu0kT04vPzTWX22Vnm16776t0GzSQWrQ4c+fjCXYAAALI6TSj5mvXLp5mWWYw\n3p495qY6OTmmXL160qWX/rb1EewAAJxlDocUG2tG3O/ZY3ryERGmB5+ebm7M4/vWvKoi2AEAqCah\nodJVV5m/MzLMoDvf3fYuuOD0BtkFndkqAgCA01GvntS2rfnaWsuSdu48veXQYwcA4Bzi+zKcRo1O\n7/n02AEAsBGCHQAAGyHYAQCwEYIdAAAbIdgBALARgh0AABsh2AEAsBGCHQAAGyHYAQCwEYIdAAAb\nIdgBALARgh0AABsh2AEAsBGCHQAAGyHYAQCwEYIdAAAbIdgBALARgh0AABsh2AEAsBGCHQAAGyHY\nAQCwkYAFu9fr1fjx45WQkKBBgwYpJSWlxPw1a9aoT58+SkhI0MKFCwNVDQAAzisBC/bExEQVFhZq\nwYIFGjlypKZOneqfV1RUpClTpujNN9/UvHnztGDBAh0+fDhQVQEA4LwRsGBPSkpSfHy8JKlNmzZK\nTk72z9uzZ48aN26s2rVrKyQkRHFxcdq4ceMpl+lwBKq2AADYgytQC87OzlZkZKT/sdPplNvtlsvl\nUnZ2tqKiovzzatWqpezs7FMuMyQkKSB1hXHllZJEGwca7Rx4tHHg0caB53Se3vMCFuyRkZHKycnx\nP/Z6vXK5XGXOy8nJKRH0ZYmLiwtMRQEAsJGAHYpv166d1q1bJ0nasmWLWrRo4Z8XGxurlJQUHTt2\nTIWFhfr222/Vtm3bQFUFAIDzhsOyLCsQC/Z6vZowYYJ2794ty7I0efJkbd++Xbm5uUpISNCaNWv0\n6quvyrIs9enTRw888EAgqgEAwHklYMEOAADOPm5QAwCAjRDsAADYyDkX7NyxLvBO1cYrVqzQ/fff\nr379+mn8+PHyer3VVNOa61Rt7DNu3DjNmDHjLNfOHk7Vxtu2bdOAAQPUv39/jRgxQgUFBdVU05rt\nVO28bNky9e7dW3369NE777xTTbW0h61bt2rQoEGlplc596xzzCeffGKNGTPGsizL2rx5szVs2DD/\nvMLCQuuOO+6wjh07ZhUUFFj33nuvlZGRUV1VrbEqauO8vDyrc+fOVm5urmVZlvXUU09ZiYmJ1VLP\nmqyiNvaZP3++1bdvX+ull14629WzhYra2Ov1Wnfffbe1b98+y7Isa+HChdaePXuqpZ413aneyx06\ndLAyMzOtgoIC//YZVTdnzhzrzjvvtO6///4S008n9865Hnsg7liHkipq45CQEL377rsKDw+XJLnd\nboWGhlZLPWuyitpYkjZt2qStW7cqISGhOqpnCxW18d69e1WnTh3NnTtXAwcO1LFjx9S0adPqqmqN\ndqr3csuWLXXixAkVFhbKsiw5uEXoaWncuLFmzpxZavrp5N45F+zl3bHON+907liHkipq46CgIF14\n4YWSpHnz5ik3N1cdOnSolnrWZBW1cXp6ul599VWNHz++uqpnCxW1cWZmpjZv3qyBAwfqX//6lzZs\n2KCvvvqquqpao1XUzpLUvHlz9enTRz179tTtt9+u6Ojo6qhmjdetWzf/TdxOdjq5d84F+5m+Yx1K\nq6iNfY+nTZum9evXa+bMmeyBn4aK2vjjjz9WZmamhg4dqjlz5mjFihVavHhxdVW1xqqojevUqaMm\nTZooNjZWwcHBio+PL9XTROVU1M47d+7UZ599ptWrV2vNmjU6evSoVq5cWV1VtaXTyb1zLti5Y13g\nVdTGkjR+/HgVFBTotdde8x+SR9VU1MaDBw/W4sWLNW/ePA0dOlR33nmn7r333uqqao1VURtfeuml\nysnJ8Q/0+vbbb9W8efNqqWdNV1E7R0VFKSwsTKGhoXI6nYqJiVFWVlZ1VdWWTif3Anav+NPVpUsX\nrV+/Xv369fPfsW758uX+O9Y988wzevjhh/13rGvQoEF1V7nGqaiNW7VqpUWLFum6667TkCFDJJkg\n6tKlSzXXumY51fsYv92p2vjFF1/UyJEjZVmW2rZtq9tvv726q1wjnaqdExISNGDAAAUHB6tx48bq\n3bt3dVfZFn5L7nHnOQAAbOScOxQPAABOH8EOAICNEOwAANgIwQ4AgI0Q7AAA2Mg5d7kbgMBITU1V\n9+7dFRsbK8ncaCQnJ0f33HOPRowYcUbW4bsl5vDhw9WyZUvt2rXrjCwXQOUR7MB5pH79+lq6dKn/\ncVpamrp166aePXv6Ax9AzUawA+exjIwMWZalWrVqac6cOVq5cqU8Ho9uueUWjRo1Sg6HQ3PnztX8\n+fPldDrVsWNHjRo1Srt379YLL7yg3NxcHT16VA8++KAGDx5c3S8HgAh24LySnp6uXr16qaCgQJmZ\nmWrdurVeeeUV7d69W8nJyVq0aJEcDodGjRqlZcuW6fLLL9c777yj999/X+Hh4XrkkUeUnJyspUuX\n6rHHHlP79u21f/9+3X333QQ7cI4g2IHziO9QvNfr1dSpU7Vr1y7ddNNN+utf/6pt27b571mfn5+v\nhg0b6vDhw+rYsaP/Syfmzp0rSbryyiv1xRdf6PXXX9euXbuUm5tbXS8JwK8Q7MB5KCgoSKNHj9Y9\n99yjN998Ux6PR0OGDNGDDz4oScrKypLT6dSiRYtKPC8tLU3h4eEaO3asoqOj1bFjR/Xo0UMffvhh\ndbwMAGXgcjfgPOVyuTR69GjNnj1bV111lZYuXaqcnBy53W49/vjj+uSTT3Tddddp3bp1/ukjR45U\ncnKy1q9frxEjRuiOO+7Qxo0bJUkej6eaXxEAiR47cF679dZb1aZNG23cuFFdu3ZV37595fF4FB8f\nr969e8vhcGjgwIHq16+fvF6vunTpoptvvlnDhw/XgAEDFB0drcsvv1yXXHKJUlNTq/vlABDf7gYA\ngK1wKB4AABsh2AEAsBGCHQAAGyHYAQCwEYIdAAAbIdgBALARgh0AABsh2AEAsJH/D8rNoFq+TMea\nAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a7ef66550>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import src.lib.utility_classfier as uclf\n",
    "uclf.plot_prc(y_test,y_score[0],0.2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}