EEGnet_CNN_stim_reconst-fork1.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Audio Stimulus Reconstruction From EEG"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Original data type: float32\n",
      "Normalized data type: float16\n",
      "test input shape (135, 60, 500, 1) Nomralized MEAN: 0.54932 min 0.0 max 1.0\n",
      "train input shape (315, 60, 500, 1) Nomralized MEAN: 0.54346 min 0.0 max 1.0\n",
      "test labels shape (135, 1, 500, 1) Nomralized MEAN: 0.22217 min 0.0 max 1.0\n",
      "train labels shape (315, 1, 500, 1) Nomralized MEAN: 0.22144 min 0.0 max 1.0\n"
     ]
    }
   ],
   "source": [
    "# get the Dataset:\n",
    "\n",
    "import numpy as np\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn import preprocessing\n",
    "import scipy.io as sio\n",
    "\n",
    "# load data from the .mat file:\n",
    "# mat_contents = sio.loadmat('/home/amplifier/home/DATASETS/EEG_big.mat')\n",
    "mat_contents = sio.loadmat('/home/amplifier/home/DATASETS/Merged456->2-30Hz,Env=1_TD1_94_.mat')\n",
    "# mat_contents = sio.loadmat('EEG_big4CNN.mat')\n",
    "# mat_contents = sio.loadmat('EEG_big4CNN_1subj.mat')\n",
    "\n",
    "\n",
    "X = mat_contents['X']\n",
    "Y = mat_contents['Y']\n",
    "x_train, x_test, y_train, y_test = train_test_split(X, Y, test_size=0.30)\n",
    "print('Original data type:', x_train.dtype)\n",
    "\n",
    "# convert to float64 for numerical stability:\n",
    "x_train = x_train.astype('float64')\n",
    "y_train = y_train.astype('float64')\n",
    "x_test = x_test.astype('float64')\n",
    "y_test = y_test.astype('float64')\n",
    "\n",
    "# normalize to unit variance and zero mean:\n",
    "for i in range(x_train.shape[0]):\n",
    "    x_train[i,:,:] = preprocessing.scale(x_train[i,:,:], axis=1)\n",
    "    x_train[i,:,:] = preprocessing.minmax_scale(x_train[i,:,:], axis=1)\n",
    "    y_train[i,:] = preprocessing.scale(y_train[i,:], axis=0)\n",
    "    y_train[i,:] = preprocessing.minmax_scale(y_train[i,:], axis=0)\n",
    "\n",
    "for i in range(x_test.shape[0]):\n",
    "    x_test[i,:,:] = preprocessing.scale(x_test[i,:,:], axis=1)\n",
    "    x_test[i,:,:] = preprocessing.minmax_scale(x_test[i,:,:], axis=1)\n",
    "    y_test[i,:] = preprocessing.scale(y_test[i,:], axis=0)\n",
    "    y_test[i,:] = preprocessing.minmax_scale(y_test[i,:], axis=0)\n",
    "\n",
    "# ensure the tensors are sized right:\n",
    "x_train = np.expand_dims(x_train,3)\n",
    "x_test = np.expand_dims(x_test,3)\n",
    "y_test = np.expand_dims(y_test,1)\n",
    "y_test = y_test[:,:,:,None]\n",
    "y_train = np.expand_dims(y_train,1)\n",
    "y_train = y_train[:,:,:,None]\n",
    "\n",
    "# convert to float16 to save space:\n",
    "x_train = x_train.astype('float16')\n",
    "y_train = y_train.astype('float16')\n",
    "x_test = x_test.astype('float16')\n",
    "y_test = y_test.astype('float16')\n",
    "\n",
    "leng = X.shape[2]\n",
    "\n",
    "print('Normalized data type:', x_train.dtype)\n",
    "\n",
    "print('test input shape', x_test.shape, \"Nomralized MEAN:\", np.mean(x_test), \"min\", np.min(x_test),\"max\", np.max(x_test))\n",
    "print('train input shape', x_train.shape, \"Nomralized MEAN:\", np.mean(x_train), \"min\", np.min(x_train),\"max\", np.max(x_train))\n",
    "\n",
    "print('test labels shape', y_test.shape, \"Nomralized MEAN:\", np.mean(y_test), \"min\", np.min(y_test),\"max\", np.max(y_test))\n",
    "print('train labels shape', y_train.shape, \"Nomralized MEAN:\", np.mean(y_train), \"min\", np.min(y_train),\"max\", np.max(y_train))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "# we first define the autoencoder's architecture:\n",
    "\n",
    "# make sure you don't hog all the video memory\n",
    "import tensorflow as tf\n",
    "config = tf.ConfigProto()\n",
    "config.gpu_options.allow_growth = True\n",
    "sess = tf.Session(config=config)\n",
    "from keras import backend as K\n",
    "K.set_session(sess)\n",
    "###################################\n",
    "\n",
    "from keras.layers import Input, Dense, Conv2D, MaxPooling2D, UpSampling2D, Dropout, BatchNormalization\n",
    "from keras.models import Model\n",
    "from keras.models import load_model\n",
    "from keras.callbacks import TensorBoard\n",
    "from keras.initializers import Orthogonal as orth\n",
    "\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "def conv_net():\n",
    "    input_img = Input(shape=(60, leng, 1))  # adapt this if using `channels_first` image data format   \n",
    "    e1 = Conv2D(4, (2, 2), activation='relu', padding='same', kernel_initializer = orth(gain=1.0, seed=None))(input_img)\n",
    "    e2b = BatchNormalization(axis=1)(e1)\n",
    "    e2a = Dropout(0.1)(e2b)\n",
    "    e2 = MaxPooling2D((2, 1), padding='same')(e2a)\n",
    "    \n",
    "    \n",
    "    e3 = Conv2D(8, (2, 2), activation='relu', padding='same', kernel_initializer = orth(gain=1.0, seed=None))(e2)\n",
    "    e4b = BatchNormalization(axis=1)(e3)\n",
    "    e4a = Dropout(0.2)(e4b)\n",
    "    e4 = MaxPooling2D((3, 1), padding='same')(e4a)\n",
    "\n",
    "    \n",
    "    e5 = Conv2D(16, (2, 2), activation='relu', padding='same', kernel_initializer = orth(gain=1.0, seed=None))(e4)\n",
    "    e6b = BatchNormalization(axis=1)(e5)\n",
    "    e6a = Dropout(0.3)(e6b)\n",
    "    e6 = MaxPooling2D((2, 1), padding='same')(e6a)\n",
    "\n",
    "    \n",
    "    e7 = Conv2D(1, (2, 2), activation='relu', padding='same', kernel_initializer = orth(gain=1.0, seed=None))(e6)\n",
    "    \n",
    "    e8 = MaxPooling2D((5, 1), padding='same')(e7)\n",
    "    e9a = Dropout(0.4)(e8)\n",
    "    return Model(input_img, e9a)\n",
    "\n",
    "def correlation_coefficient_loss(y_true, y_pred):\n",
    "    x = y_true\n",
    "    y = y_pred\n",
    "    mx = K.mean(x)\n",
    "    my = K.mean(y)\n",
    "    xm, ym = x-mx, y-my\n",
    "    r_num = K.sum(tf.multiply(xm,ym))\n",
    "    r_den = K.sqrt(tf.multiply(K.sum(K.square(xm)), K.sum(K.square(ym))))\n",
    "    r = r_num / r_den\n",
    "\n",
    "    r = K.maximum(K.minimum(r, 1.0), -1.0)\n",
    "    return 1 - K.square(r)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# define input to the model:\n",
    "x = Input(shape=(60, leng, 1))\n",
    "\n",
    "# make the model:\n",
    "model = conv_net()\n",
    "\n",
    "# compile the model:\n",
    "# model.compile(optimizer='adadelta', loss='binary_crossentropy')\n",
    "model.compile(optimizer='adadelta', loss=correlation_coefficient_loss)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# model.load_weights('/home/amplifier/home/models/EEGnet.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_2 (InputLayer)         (None, 60, 500, 1)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 60, 500, 4)        20        \n",
      "_________________________________________________________________\n",
      "batch_normalization_1 (Batch (None, 60, 500, 4)        240       \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 60, 500, 4)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 30, 500, 4)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 30, 500, 8)        136       \n",
      "_________________________________________________________________\n",
      "batch_normalization_2 (Batch (None, 30, 500, 8)        120       \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 30, 500, 8)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 10, 500, 8)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 10, 500, 16)       528       \n",
      "_________________________________________________________________\n",
      "batch_normalization_3 (Batch (None, 10, 500, 16)       40        \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 10, 500, 16)       0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_3 (MaxPooling2 (None, 5, 500, 16)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_4 (Conv2D)            (None, 5, 500, 1)         65        \n",
      "_________________________________________________________________\n",
      "max_pooling2d_4 (MaxPooling2 (None, 1, 500, 1)         0         \n",
      "_________________________________________________________________\n",
      "dropout_4 (Dropout)          (None, 1, 500, 1)         0         \n",
      "=================================================================\n",
      "Total params: 1,149\n",
      "Trainable params: 949\n",
      "Non-trainable params: 200\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# SEE INSIDE THE MODEL:\n",
    "# conv_net().summary() #or\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5,1,'Channels 11 in test samples 6 and 16')"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f25541fd320>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# inspect channel ___ sample\n",
    "chan = 30\n",
    "samp = 12\n",
    "%matplotlib inline\n",
    "plt.plot(range(0, leng), x_test[samp, chan,:,0], range(0,leng), y_test[samp,0,:,0])\n",
    "ax = plt.gca()\n",
    "ax.set_title(\"Channels 11 in test samples 6 and 16\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /opt/conda/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/base.py:198: retry (from tensorflow.contrib.learn.python.learn.datasets.base) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use the retry module or similar alternatives.\n",
      "Train on 315 samples, validate on 135 samples\n",
      "Epoch 1/1000\n",
      "315/315 [==============================] - 3s 9ms/step - loss: 0.9999 - val_loss: 0.9997\n",
      "Epoch 2/1000\n",
      "315/315 [==============================] - 0s 998us/step - loss: 0.9999 - val_loss: 0.9993\n",
      "Epoch 3/1000\n",
      "315/315 [==============================] - 0s 968us/step - loss: 0.9998 - val_loss: 0.9991\n",
      "Epoch 4/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9999 - val_loss: 0.9992\n",
      "Epoch 5/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9999 - val_loss: 0.9991\n",
      "Epoch 6/1000\n",
      "315/315 [==============================] - 0s 990us/step - loss: 0.9999 - val_loss: 0.9991\n",
      "Epoch 7/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9998 - val_loss: 0.9986\n",
      "Epoch 8/1000\n",
      "315/315 [==============================] - 0s 966us/step - loss: 0.9998 - val_loss: 0.9986\n",
      "Epoch 9/1000\n",
      "315/315 [==============================] - 0s 973us/step - loss: 0.9999 - val_loss: 0.9986\n",
      "Epoch 10/1000\n",
      "315/315 [==============================] - 0s 966us/step - loss: 0.9998 - val_loss: 0.9986\n",
      "Epoch 11/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9996 - val_loss: 0.9986\n",
      "Epoch 12/1000\n",
      "315/315 [==============================] - 0s 983us/step - loss: 0.9999 - val_loss: 0.9986\n",
      "Epoch 13/1000\n",
      "315/315 [==============================] - 0s 902us/step - loss: 0.9999 - val_loss: 0.9986\n",
      "Epoch 14/1000\n",
      "315/315 [==============================] - 0s 910us/step - loss: 0.9999 - val_loss: 0.9986\n",
      "Epoch 15/1000\n",
      "315/315 [==============================] - 0s 931us/step - loss: 0.9998 - val_loss: 0.9986\n",
      "Epoch 16/1000\n",
      "315/315 [==============================] - 0s 910us/step - loss: 0.9997 - val_loss: 0.9986\n",
      "Epoch 17/1000\n",
      "315/315 [==============================] - 0s 908us/step - loss: 0.9998 - val_loss: 0.9986\n",
      "Epoch 18/1000\n",
      "315/315 [==============================] - 0s 933us/step - loss: 1.0000 - val_loss: 0.9986\n",
      "Epoch 19/1000\n",
      "315/315 [==============================] - 0s 950us/step - loss: 0.9998 - val_loss: 0.9986\n",
      "Epoch 20/1000\n",
      "315/315 [==============================] - 0s 970us/step - loss: 0.9998 - val_loss: 0.9986\n",
      "Epoch 21/1000\n",
      "315/315 [==============================] - 0s 918us/step - loss: 0.9998 - val_loss: 0.9986\n",
      "Epoch 22/1000\n",
      "315/315 [==============================] - 0s 928us/step - loss: 0.9999 - val_loss: 0.9986\n",
      "Epoch 23/1000\n",
      "315/315 [==============================] - 0s 933us/step - loss: 0.9997 - val_loss: 0.9986\n",
      "Epoch 24/1000\n",
      "315/315 [==============================] - 0s 905us/step - loss: 0.9997 - val_loss: 0.9986\n",
      "Epoch 25/1000\n",
      "315/315 [==============================] - 0s 927us/step - loss: 0.9999 - val_loss: 0.9986\n",
      "Epoch 26/1000\n",
      "315/315 [==============================] - 0s 933us/step - loss: 0.9998 - val_loss: 0.9986\n",
      "Epoch 27/1000\n",
      "315/315 [==============================] - 0s 933us/step - loss: 0.9998 - val_loss: 0.9986\n",
      "Epoch 28/1000\n",
      "315/315 [==============================] - 0s 964us/step - loss: 0.9998 - val_loss: 0.9986\n",
      "Epoch 29/1000\n",
      "315/315 [==============================] - 0s 927us/step - loss: 0.9998 - val_loss: 0.9986\n",
      "Epoch 30/1000\n",
      "315/315 [==============================] - 0s 946us/step - loss: 0.9998 - val_loss: 0.9986\n",
      "Epoch 31/1000\n",
      "315/315 [==============================] - 0s 913us/step - loss: 0.9998 - val_loss: 0.9986\n",
      "Epoch 32/1000\n",
      "315/315 [==============================] - 0s 908us/step - loss: 0.9999 - val_loss: 0.9986\n",
      "Epoch 33/1000\n",
      "315/315 [==============================] - 0s 945us/step - loss: 0.9995 - val_loss: 0.9986\n",
      "Epoch 34/1000\n",
      "315/315 [==============================] - 0s 915us/step - loss: 0.9997 - val_loss: 0.9986\n",
      "Epoch 35/1000\n",
      "315/315 [==============================] - 0s 942us/step - loss: 0.9997 - val_loss: 0.9986\n",
      "Epoch 36/1000\n",
      "315/315 [==============================] - 0s 931us/step - loss: 0.9998 - val_loss: 0.9986\n",
      "Epoch 37/1000\n",
      "315/315 [==============================] - 0s 935us/step - loss: 0.9996 - val_loss: 0.9986\n",
      "Epoch 38/1000\n",
      "315/315 [==============================] - 0s 921us/step - loss: 0.9993 - val_loss: 0.9986\n",
      "Epoch 39/1000\n",
      "315/315 [==============================] - 0s 946us/step - loss: 0.9994 - val_loss: 0.9985\n",
      "Epoch 40/1000\n",
      "315/315 [==============================] - 0s 940us/step - loss: 0.9996 - val_loss: 0.9986\n",
      "Epoch 41/1000\n",
      "315/315 [==============================] - 0s 943us/step - loss: 0.9996 - val_loss: 0.9986\n",
      "Epoch 42/1000\n",
      "315/315 [==============================] - 0s 898us/step - loss: 0.9998 - val_loss: 0.9985\n",
      "Epoch 43/1000\n",
      "315/315 [==============================] - 0s 934us/step - loss: 0.9996 - val_loss: 0.9985\n",
      "Epoch 44/1000\n",
      "315/315 [==============================] - 0s 947us/step - loss: 0.9999 - val_loss: 0.9985\n",
      "Epoch 45/1000\n",
      "315/315 [==============================] - 0s 905us/step - loss: 0.9994 - val_loss: 0.9985\n",
      "Epoch 46/1000\n",
      "315/315 [==============================] - 0s 939us/step - loss: 0.9999 - val_loss: 0.9985\n",
      "Epoch 47/1000\n",
      "315/315 [==============================] - 0s 920us/step - loss: 0.9998 - val_loss: 0.9985\n",
      "Epoch 48/1000\n",
      "315/315 [==============================] - 0s 913us/step - loss: 0.9995 - val_loss: 0.9985\n",
      "Epoch 49/1000\n",
      "315/315 [==============================] - 0s 936us/step - loss: 0.9996 - val_loss: 0.9985\n",
      "Epoch 50/1000\n",
      "315/315 [==============================] - 0s 908us/step - loss: 0.9996 - val_loss: 0.9985\n",
      "Epoch 51/1000\n",
      "315/315 [==============================] - 0s 910us/step - loss: 0.9997 - val_loss: 0.9985\n",
      "Epoch 52/1000\n",
      "315/315 [==============================] - 0s 913us/step - loss: 0.9997 - val_loss: 0.9985\n",
      "Epoch 53/1000\n",
      "315/315 [==============================] - 0s 923us/step - loss: 0.9997 - val_loss: 0.9985\n",
      "Epoch 54/1000\n",
      "315/315 [==============================] - 0s 964us/step - loss: 0.9994 - val_loss: 0.9985\n",
      "Epoch 55/1000\n",
      "315/315 [==============================] - 0s 937us/step - loss: 0.9998 - val_loss: 0.9985\n",
      "Epoch 56/1000\n",
      "315/315 [==============================] - 0s 937us/step - loss: 0.9996 - val_loss: 0.9985\n",
      "Epoch 57/1000\n",
      "315/315 [==============================] - 0s 880us/step - loss: 0.9997 - val_loss: 0.9985\n",
      "Epoch 58/1000\n",
      "315/315 [==============================] - 0s 939us/step - loss: 0.9999 - val_loss: 0.9985\n",
      "Epoch 59/1000\n",
      "315/315 [==============================] - 0s 922us/step - loss: 0.9997 - val_loss: 0.9984\n",
      "Epoch 60/1000\n",
      "315/315 [==============================] - 0s 939us/step - loss: 0.9997 - val_loss: 0.9984\n",
      "Epoch 61/1000\n",
      "315/315 [==============================] - 0s 933us/step - loss: 0.9995 - val_loss: 0.9984\n",
      "Epoch 62/1000\n",
      "315/315 [==============================] - 0s 953us/step - loss: 0.9997 - val_loss: 0.9984\n",
      "Epoch 63/1000\n",
      "315/315 [==============================] - 0s 951us/step - loss: 0.9998 - val_loss: 0.9984\n",
      "Epoch 64/1000\n",
      "315/315 [==============================] - 0s 935us/step - loss: 0.9998 - val_loss: 0.9984\n",
      "Epoch 65/1000\n",
      "315/315 [==============================] - 0s 921us/step - loss: 0.9998 - val_loss: 0.9984\n",
      "Epoch 66/1000\n",
      "315/315 [==============================] - 0s 906us/step - loss: 0.9997 - val_loss: 0.9984\n",
      "Epoch 67/1000\n",
      "315/315 [==============================] - 0s 900us/step - loss: 0.9995 - val_loss: 0.9984\n",
      "Epoch 68/1000\n",
      "315/315 [==============================] - 0s 917us/step - loss: 0.9995 - val_loss: 0.9984\n",
      "Epoch 69/1000\n",
      "315/315 [==============================] - 0s 922us/step - loss: 0.9997 - val_loss: 0.9984\n",
      "Epoch 70/1000\n",
      "315/315 [==============================] - 0s 918us/step - loss: 0.9998 - val_loss: 0.9984\n",
      "Epoch 71/1000\n",
      "315/315 [==============================] - 0s 937us/step - loss: 0.9994 - val_loss: 0.9984\n",
      "Epoch 72/1000\n",
      "315/315 [==============================] - 0s 896us/step - loss: 0.9998 - val_loss: 0.9984\n",
      "Epoch 73/1000\n",
      "315/315 [==============================] - 0s 903us/step - loss: 0.9997 - val_loss: 0.9984\n",
      "Epoch 74/1000\n",
      "315/315 [==============================] - 0s 929us/step - loss: 0.9993 - val_loss: 0.9984\n",
      "Epoch 75/1000\n",
      "315/315 [==============================] - 0s 960us/step - loss: 0.9997 - val_loss: 0.9984\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 76/1000\n",
      "315/315 [==============================] - 0s 947us/step - loss: 0.9997 - val_loss: 0.9984\n",
      "Epoch 77/1000\n",
      "315/315 [==============================] - 0s 927us/step - loss: 0.9999 - val_loss: 0.9984\n",
      "Epoch 78/1000\n",
      "315/315 [==============================] - 0s 937us/step - loss: 0.9994 - val_loss: 0.9984\n",
      "Epoch 79/1000\n",
      "315/315 [==============================] - 0s 935us/step - loss: 0.9996 - val_loss: 0.9984\n",
      "Epoch 80/1000\n",
      "315/315 [==============================] - 0s 953us/step - loss: 0.9994 - val_loss: 0.9983\n",
      "Epoch 81/1000\n",
      "315/315 [==============================] - 0s 950us/step - loss: 0.9997 - val_loss: 0.9983\n",
      "Epoch 82/1000\n",
      "315/315 [==============================] - 0s 960us/step - loss: 0.9995 - val_loss: 0.9984\n",
      "Epoch 83/1000\n",
      "315/315 [==============================] - 0s 916us/step - loss: 0.9997 - val_loss: 0.9984\n",
      "Epoch 84/1000\n",
      "315/315 [==============================] - 0s 945us/step - loss: 0.9996 - val_loss: 0.9983\n",
      "Epoch 85/1000\n",
      "315/315 [==============================] - 0s 942us/step - loss: 0.9998 - val_loss: 0.9984\n",
      "Epoch 86/1000\n",
      "315/315 [==============================] - 0s 926us/step - loss: 0.9999 - val_loss: 0.9984\n",
      "Epoch 87/1000\n",
      "315/315 [==============================] - 0s 933us/step - loss: 0.9998 - val_loss: 0.9983\n",
      "Epoch 88/1000\n",
      "315/315 [==============================] - 0s 907us/step - loss: 0.9998 - val_loss: 0.9984\n",
      "Epoch 89/1000\n",
      "315/315 [==============================] - 0s 920us/step - loss: 0.9998 - val_loss: 0.9984\n",
      "Epoch 90/1000\n",
      "315/315 [==============================] - 0s 914us/step - loss: 0.9999 - val_loss: 0.9984\n",
      "Epoch 91/1000\n",
      "315/315 [==============================] - 0s 885us/step - loss: 0.9998 - val_loss: 0.9983\n",
      "Epoch 92/1000\n",
      "315/315 [==============================] - 0s 899us/step - loss: 0.9996 - val_loss: 0.9984\n",
      "Epoch 93/1000\n",
      "315/315 [==============================] - 0s 910us/step - loss: 0.9996 - val_loss: 0.9983\n",
      "Epoch 94/1000\n",
      "315/315 [==============================] - 0s 839us/step - loss: 0.9998 - val_loss: 0.9983\n",
      "Epoch 95/1000\n",
      "315/315 [==============================] - 0s 877us/step - loss: 0.9996 - val_loss: 0.9983\n",
      "Epoch 96/1000\n",
      "315/315 [==============================] - 0s 828us/step - loss: 0.9997 - val_loss: 0.9983\n",
      "Epoch 97/1000\n",
      "315/315 [==============================] - 0s 906us/step - loss: 0.9998 - val_loss: 0.9983\n",
      "Epoch 98/1000\n",
      "315/315 [==============================] - 0s 896us/step - loss: 0.9997 - val_loss: 0.9983\n",
      "Epoch 99/1000\n",
      "315/315 [==============================] - 0s 870us/step - loss: 0.9995 - val_loss: 0.9983\n",
      "Epoch 100/1000\n",
      "315/315 [==============================] - 0s 896us/step - loss: 0.9996 - val_loss: 0.9983\n",
      "Epoch 101/1000\n",
      "315/315 [==============================] - 0s 856us/step - loss: 0.9998 - val_loss: 0.9983\n",
      "Epoch 102/1000\n",
      "315/315 [==============================] - 0s 900us/step - loss: 0.9997 - val_loss: 0.9982\n",
      "Epoch 103/1000\n",
      "315/315 [==============================] - 0s 879us/step - loss: 0.9998 - val_loss: 0.9982\n",
      "Epoch 104/1000\n",
      "315/315 [==============================] - 0s 893us/step - loss: 0.9998 - val_loss: 0.9982\n",
      "Epoch 105/1000\n",
      "315/315 [==============================] - 0s 879us/step - loss: 0.9997 - val_loss: 0.9982\n",
      "Epoch 106/1000\n",
      "315/315 [==============================] - 0s 867us/step - loss: 0.9997 - val_loss: 0.9982\n",
      "Epoch 107/1000\n",
      "315/315 [==============================] - 0s 889us/step - loss: 0.9995 - val_loss: 0.9983\n",
      "Epoch 108/1000\n",
      "315/315 [==============================] - 0s 908us/step - loss: 0.9998 - val_loss: 0.9983\n",
      "Epoch 109/1000\n",
      "315/315 [==============================] - 0s 887us/step - loss: 0.9998 - val_loss: 0.9982\n",
      "Epoch 110/1000\n",
      "315/315 [==============================] - 0s 901us/step - loss: 0.9996 - val_loss: 0.9983\n",
      "Epoch 111/1000\n",
      "315/315 [==============================] - 0s 893us/step - loss: 0.9999 - val_loss: 0.9982\n",
      "Epoch 112/1000\n",
      "315/315 [==============================] - 0s 888us/step - loss: 0.9995 - val_loss: 0.9983\n",
      "Epoch 113/1000\n",
      "315/315 [==============================] - 0s 891us/step - loss: 0.9996 - val_loss: 0.9983\n",
      "Epoch 114/1000\n",
      "315/315 [==============================] - 0s 864us/step - loss: 0.9998 - val_loss: 0.9982\n",
      "Epoch 115/1000\n",
      "315/315 [==============================] - 0s 891us/step - loss: 0.9996 - val_loss: 0.9982\n",
      "Epoch 116/1000\n",
      "315/315 [==============================] - 0s 863us/step - loss: 0.9993 - val_loss: 0.9982\n",
      "Epoch 117/1000\n",
      "315/315 [==============================] - 0s 915us/step - loss: 0.9998 - val_loss: 0.9982\n",
      "Epoch 118/1000\n",
      "315/315 [==============================] - 0s 962us/step - loss: 0.9998 - val_loss: 0.9982\n",
      "Epoch 119/1000\n",
      "315/315 [==============================] - 0s 929us/step - loss: 0.9995 - val_loss: 0.9982\n",
      "Epoch 120/1000\n",
      "315/315 [==============================] - 0s 866us/step - loss: 0.9998 - val_loss: 0.9982\n",
      "Epoch 121/1000\n",
      "315/315 [==============================] - 0s 890us/step - loss: 0.9998 - val_loss: 0.9982\n",
      "Epoch 122/1000\n",
      "315/315 [==============================] - 0s 886us/step - loss: 0.9997 - val_loss: 0.9982\n",
      "Epoch 123/1000\n",
      "315/315 [==============================] - 0s 887us/step - loss: 0.9998 - val_loss: 0.9983\n",
      "Epoch 124/1000\n",
      "315/315 [==============================] - 0s 897us/step - loss: 0.9997 - val_loss: 0.9982\n",
      "Epoch 125/1000\n",
      "315/315 [==============================] - 0s 912us/step - loss: 0.9997 - val_loss: 0.9982\n",
      "Epoch 126/1000\n",
      "315/315 [==============================] - 0s 882us/step - loss: 0.9995 - val_loss: 0.9982\n",
      "Epoch 127/1000\n",
      "315/315 [==============================] - 0s 886us/step - loss: 0.9993 - val_loss: 0.9982\n",
      "Epoch 128/1000\n",
      "315/315 [==============================] - 0s 884us/step - loss: 0.9995 - val_loss: 0.9982\n",
      "Epoch 129/1000\n",
      "315/315 [==============================] - 0s 869us/step - loss: 0.9994 - val_loss: 0.9982\n",
      "Epoch 130/1000\n",
      "315/315 [==============================] - 0s 889us/step - loss: 0.9995 - val_loss: 0.9982\n",
      "Epoch 131/1000\n",
      "315/315 [==============================] - 0s 907us/step - loss: 0.9997 - val_loss: 0.9982\n",
      "Epoch 132/1000\n",
      "315/315 [==============================] - 0s 916us/step - loss: 0.9998 - val_loss: 0.9982\n",
      "Epoch 133/1000\n",
      "315/315 [==============================] - 0s 889us/step - loss: 0.9995 - val_loss: 0.9982\n",
      "Epoch 134/1000\n",
      "315/315 [==============================] - 0s 865us/step - loss: 0.9995 - val_loss: 0.9982\n",
      "Epoch 135/1000\n",
      "315/315 [==============================] - 0s 892us/step - loss: 0.9996 - val_loss: 0.9982\n",
      "Epoch 136/1000\n",
      "315/315 [==============================] - 0s 903us/step - loss: 0.9994 - val_loss: 0.9982\n",
      "Epoch 137/1000\n",
      "315/315 [==============================] - 0s 842us/step - loss: 0.9996 - val_loss: 0.9981\n",
      "Epoch 138/1000\n",
      "315/315 [==============================] - 0s 867us/step - loss: 0.9997 - val_loss: 0.9981\n",
      "Epoch 139/1000\n",
      "315/315 [==============================] - 0s 906us/step - loss: 0.9994 - val_loss: 0.9981\n",
      "Epoch 140/1000\n",
      "315/315 [==============================] - 0s 879us/step - loss: 0.9997 - val_loss: 0.9981\n",
      "Epoch 141/1000\n",
      "315/315 [==============================] - 0s 878us/step - loss: 0.9997 - val_loss: 0.9981\n",
      "Epoch 142/1000\n",
      "315/315 [==============================] - 0s 908us/step - loss: 0.9996 - val_loss: 0.9981\n",
      "Epoch 143/1000\n",
      "315/315 [==============================] - 0s 927us/step - loss: 0.9996 - val_loss: 0.9981\n",
      "Epoch 144/1000\n",
      "315/315 [==============================] - 0s 903us/step - loss: 0.9996 - val_loss: 0.9980\n",
      "Epoch 145/1000\n",
      "315/315 [==============================] - 0s 953us/step - loss: 0.9998 - val_loss: 0.9981\n",
      "Epoch 146/1000\n",
      "315/315 [==============================] - 0s 933us/step - loss: 0.9996 - val_loss: 0.9981\n",
      "Epoch 147/1000\n",
      "315/315 [==============================] - 0s 906us/step - loss: 0.9996 - val_loss: 0.9980\n",
      "Epoch 148/1000\n",
      "315/315 [==============================] - 0s 937us/step - loss: 0.9996 - val_loss: 0.9980\n",
      "Epoch 149/1000\n",
      "315/315 [==============================] - 0s 916us/step - loss: 0.9993 - val_loss: 0.9980\n",
      "Epoch 150/1000\n",
      "315/315 [==============================] - 0s 922us/step - loss: 0.9994 - val_loss: 0.9980\n",
      "Epoch 151/1000\n",
      "315/315 [==============================] - 0s 914us/step - loss: 0.9997 - val_loss: 0.9981\n",
      "Epoch 152/1000\n",
      "315/315 [==============================] - 0s 923us/step - loss: 0.9998 - val_loss: 0.9980\n",
      "Epoch 153/1000\n",
      "315/315 [==============================] - 0s 937us/step - loss: 0.9996 - val_loss: 0.9981\n",
      "Epoch 154/1000\n",
      "315/315 [==============================] - 0s 929us/step - loss: 0.9997 - val_loss: 0.9980\n",
      "Epoch 155/1000\n",
      "315/315 [==============================] - 0s 958us/step - loss: 0.9994 - val_loss: 0.9980\n",
      "Epoch 156/1000\n",
      "315/315 [==============================] - 0s 919us/step - loss: 0.9997 - val_loss: 0.9980\n",
      "Epoch 157/1000\n",
      "315/315 [==============================] - 0s 952us/step - loss: 0.9995 - val_loss: 0.9980\n",
      "Epoch 158/1000\n",
      "315/315 [==============================] - 0s 943us/step - loss: 0.9993 - val_loss: 0.9980\n",
      "Epoch 159/1000\n",
      "315/315 [==============================] - 0s 926us/step - loss: 0.9998 - val_loss: 0.9980\n",
      "Epoch 160/1000\n",
      "315/315 [==============================] - 0s 946us/step - loss: 0.9994 - val_loss: 0.9980\n",
      "Epoch 161/1000\n",
      "315/315 [==============================] - 0s 930us/step - loss: 0.9989 - val_loss: 0.9980\n",
      "Epoch 162/1000\n",
      "315/315 [==============================] - 0s 928us/step - loss: 0.9996 - val_loss: 0.9980\n",
      "Epoch 163/1000\n",
      "315/315 [==============================] - 0s 945us/step - loss: 0.9995 - val_loss: 0.9980\n",
      "Epoch 164/1000\n",
      "315/315 [==============================] - 0s 943us/step - loss: 0.9993 - val_loss: 0.9979\n",
      "Epoch 165/1000\n",
      "315/315 [==============================] - 0s 915us/step - loss: 0.9995 - val_loss: 0.9979\n",
      "Epoch 166/1000\n",
      "315/315 [==============================] - 0s 920us/step - loss: 0.9996 - val_loss: 0.9979\n",
      "Epoch 167/1000\n",
      "315/315 [==============================] - 0s 951us/step - loss: 0.9993 - val_loss: 0.9979\n",
      "Epoch 168/1000\n",
      "315/315 [==============================] - 0s 921us/step - loss: 0.9997 - val_loss: 0.9979\n",
      "Epoch 169/1000\n",
      "315/315 [==============================] - 0s 954us/step - loss: 0.9994 - val_loss: 0.9979\n",
      "Epoch 170/1000\n",
      "315/315 [==============================] - 0s 923us/step - loss: 0.9997 - val_loss: 0.9979\n",
      "Epoch 171/1000\n",
      "315/315 [==============================] - 0s 949us/step - loss: 0.9995 - val_loss: 0.9979\n",
      "Epoch 172/1000\n",
      "315/315 [==============================] - 0s 918us/step - loss: 0.9997 - val_loss: 0.9979\n",
      "Epoch 173/1000\n",
      "315/315 [==============================] - 0s 914us/step - loss: 0.9995 - val_loss: 0.9979\n",
      "Epoch 174/1000\n",
      "315/315 [==============================] - 0s 924us/step - loss: 0.9998 - val_loss: 0.9979\n",
      "Epoch 175/1000\n",
      "315/315 [==============================] - 0s 902us/step - loss: 0.9997 - val_loss: 0.9979\n",
      "Epoch 176/1000\n",
      "315/315 [==============================] - 0s 929us/step - loss: 0.9995 - val_loss: 0.9979\n",
      "Epoch 177/1000\n",
      "315/315 [==============================] - 0s 934us/step - loss: 0.9992 - val_loss: 0.9978\n",
      "Epoch 178/1000\n",
      "315/315 [==============================] - 0s 960us/step - loss: 0.9992 - val_loss: 0.9978\n",
      "Epoch 179/1000\n",
      "315/315 [==============================] - 0s 933us/step - loss: 0.9997 - val_loss: 0.9979\n",
      "Epoch 180/1000\n",
      "315/315 [==============================] - 0s 937us/step - loss: 0.9994 - val_loss: 0.9979\n",
      "Epoch 181/1000\n",
      "315/315 [==============================] - 0s 939us/step - loss: 0.9998 - val_loss: 0.9979\n",
      "Epoch 182/1000\n",
      "315/315 [==============================] - 0s 948us/step - loss: 0.9997 - val_loss: 0.9979\n",
      "Epoch 183/1000\n",
      "315/315 [==============================] - 0s 944us/step - loss: 0.9994 - val_loss: 0.9979\n",
      "Epoch 184/1000\n",
      "315/315 [==============================] - 0s 959us/step - loss: 0.9997 - val_loss: 0.9979\n",
      "Epoch 185/1000\n",
      "315/315 [==============================] - 0s 923us/step - loss: 0.9998 - val_loss: 0.9979\n",
      "Epoch 186/1000\n",
      "315/315 [==============================] - 0s 937us/step - loss: 0.9996 - val_loss: 0.9979\n",
      "Epoch 187/1000\n",
      "315/315 [==============================] - 0s 931us/step - loss: 0.9993 - val_loss: 0.9979\n",
      "Epoch 188/1000\n",
      "315/315 [==============================] - 0s 962us/step - loss: 0.9998 - val_loss: 0.9979\n",
      "Epoch 189/1000\n",
      "315/315 [==============================] - 0s 907us/step - loss: 0.9993 - val_loss: 0.9979\n",
      "Epoch 190/1000\n",
      "315/315 [==============================] - 0s 949us/step - loss: 0.9996 - val_loss: 0.9979\n",
      "Epoch 191/1000\n",
      "315/315 [==============================] - 0s 917us/step - loss: 0.9996 - val_loss: 0.9980\n",
      "Epoch 192/1000\n",
      "315/315 [==============================] - 0s 918us/step - loss: 0.9997 - val_loss: 0.9979\n",
      "Epoch 193/1000\n",
      "315/315 [==============================] - 0s 948us/step - loss: 0.9997 - val_loss: 0.9980\n",
      "Epoch 194/1000\n",
      "315/315 [==============================] - 0s 963us/step - loss: 0.9993 - val_loss: 0.9980\n",
      "Epoch 195/1000\n",
      "315/315 [==============================] - 0s 908us/step - loss: 0.9993 - val_loss: 0.9980\n",
      "Epoch 196/1000\n",
      "315/315 [==============================] - 0s 926us/step - loss: 0.9995 - val_loss: 0.9980\n",
      "Epoch 197/1000\n",
      "315/315 [==============================] - 0s 948us/step - loss: 0.9992 - val_loss: 0.9980\n",
      "Epoch 198/1000\n",
      "315/315 [==============================] - 0s 924us/step - loss: 0.9996 - val_loss: 0.9980\n",
      "Epoch 199/1000\n",
      "315/315 [==============================] - 0s 940us/step - loss: 0.9997 - val_loss: 0.9980\n",
      "Epoch 200/1000\n",
      "315/315 [==============================] - 0s 917us/step - loss: 0.9995 - val_loss: 0.9980\n",
      "Epoch 201/1000\n",
      "315/315 [==============================] - 0s 933us/step - loss: 0.9993 - val_loss: 0.9980\n",
      "Epoch 202/1000\n",
      "315/315 [==============================] - 0s 937us/step - loss: 0.9995 - val_loss: 0.9980\n",
      "Epoch 203/1000\n",
      "315/315 [==============================] - 0s 918us/step - loss: 0.9998 - val_loss: 0.9980\n",
      "Epoch 204/1000\n",
      "315/315 [==============================] - 0s 912us/step - loss: 0.9997 - val_loss: 0.9980\n",
      "Epoch 205/1000\n",
      "315/315 [==============================] - 0s 917us/step - loss: 0.9997 - val_loss: 0.9981\n",
      "Epoch 206/1000\n",
      "315/315 [==============================] - 0s 938us/step - loss: 0.9996 - val_loss: 0.9980\n",
      "Epoch 207/1000\n",
      "315/315 [==============================] - 0s 933us/step - loss: 0.9995 - val_loss: 0.9980\n",
      "Epoch 208/1000\n",
      "315/315 [==============================] - 0s 936us/step - loss: 0.9995 - val_loss: 0.9980\n",
      "Epoch 209/1000\n",
      "315/315 [==============================] - 0s 955us/step - loss: 0.9996 - val_loss: 0.9980\n",
      "Epoch 210/1000\n",
      "315/315 [==============================] - 0s 954us/step - loss: 0.9997 - val_loss: 0.9980\n",
      "Epoch 211/1000\n",
      "315/315 [==============================] - 0s 920us/step - loss: 0.9999 - val_loss: 0.9980\n",
      "Epoch 212/1000\n",
      "315/315 [==============================] - 0s 929us/step - loss: 0.9995 - val_loss: 0.9980\n",
      "Epoch 213/1000\n",
      "315/315 [==============================] - 0s 925us/step - loss: 0.9998 - val_loss: 0.9979\n",
      "Epoch 214/1000\n",
      "315/315 [==============================] - 0s 919us/step - loss: 0.9995 - val_loss: 0.9979\n",
      "Epoch 215/1000\n",
      "315/315 [==============================] - 0s 906us/step - loss: 0.9998 - val_loss: 0.9979\n",
      "Epoch 216/1000\n",
      "315/315 [==============================] - 0s 937us/step - loss: 0.9994 - val_loss: 0.9979\n",
      "Epoch 217/1000\n",
      "315/315 [==============================] - 0s 948us/step - loss: 0.9996 - val_loss: 0.9979\n",
      "Epoch 218/1000\n",
      "315/315 [==============================] - 0s 906us/step - loss: 0.9990 - val_loss: 0.9979\n",
      "Epoch 219/1000\n",
      "315/315 [==============================] - 0s 951us/step - loss: 0.9997 - val_loss: 0.9980\n",
      "Epoch 220/1000\n",
      "315/315 [==============================] - 0s 902us/step - loss: 0.9997 - val_loss: 0.9980\n",
      "Epoch 221/1000\n",
      "315/315 [==============================] - 0s 945us/step - loss: 0.9998 - val_loss: 0.9979\n",
      "Epoch 222/1000\n",
      "315/315 [==============================] - 0s 934us/step - loss: 0.9996 - val_loss: 0.9980\n",
      "Epoch 223/1000\n",
      "315/315 [==============================] - 0s 938us/step - loss: 0.9996 - val_loss: 0.9980\n",
      "Epoch 224/1000\n",
      "315/315 [==============================] - 0s 936us/step - loss: 0.9995 - val_loss: 0.9980\n",
      "Epoch 225/1000\n",
      "315/315 [==============================] - 0s 931us/step - loss: 0.9996 - val_loss: 0.9979\n",
      "Epoch 226/1000\n",
      "315/315 [==============================] - 0s 932us/step - loss: 0.9993 - val_loss: 0.9979\n",
      "Epoch 227/1000\n",
      "315/315 [==============================] - 0s 911us/step - loss: 0.9992 - val_loss: 0.9980\n",
      "Epoch 228/1000\n",
      "315/315 [==============================] - 0s 924us/step - loss: 0.9994 - val_loss: 0.9980\n",
      "Epoch 229/1000\n",
      "315/315 [==============================] - 0s 930us/step - loss: 0.9996 - val_loss: 0.9980\n",
      "Epoch 230/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "315/315 [==============================] - 0s 919us/step - loss: 0.9995 - val_loss: 0.9980\n",
      "Epoch 231/1000\n",
      "315/315 [==============================] - 0s 936us/step - loss: 0.9995 - val_loss: 0.9979\n",
      "Epoch 232/1000\n",
      "315/315 [==============================] - 0s 925us/step - loss: 0.9994 - val_loss: 0.9979\n",
      "Epoch 233/1000\n",
      "315/315 [==============================] - 0s 924us/step - loss: 0.9993 - val_loss: 0.9979\n",
      "Epoch 234/1000\n",
      "315/315 [==============================] - 0s 935us/step - loss: 0.9997 - val_loss: 0.9979\n",
      "Epoch 235/1000\n",
      "315/315 [==============================] - 0s 922us/step - loss: 0.9994 - val_loss: 0.9979\n",
      "Epoch 236/1000\n",
      "315/315 [==============================] - 0s 939us/step - loss: 0.9994 - val_loss: 0.9979\n",
      "Epoch 237/1000\n",
      "315/315 [==============================] - 0s 961us/step - loss: 0.9993 - val_loss: 0.9979\n",
      "Epoch 238/1000\n",
      "315/315 [==============================] - 0s 950us/step - loss: 0.9993 - val_loss: 0.9979\n",
      "Epoch 239/1000\n",
      "315/315 [==============================] - 0s 940us/step - loss: 0.9996 - val_loss: 0.9978\n",
      "Epoch 240/1000\n",
      "315/315 [==============================] - 0s 928us/step - loss: 0.9992 - val_loss: 0.9978\n",
      "Epoch 241/1000\n",
      "315/315 [==============================] - 0s 914us/step - loss: 0.9993 - val_loss: 0.9979\n",
      "Epoch 242/1000\n",
      "315/315 [==============================] - 0s 942us/step - loss: 0.9994 - val_loss: 0.9978\n",
      "Epoch 243/1000\n",
      "315/315 [==============================] - 0s 916us/step - loss: 0.9997 - val_loss: 0.9978\n",
      "Epoch 244/1000\n",
      "315/315 [==============================] - 0s 875us/step - loss: 0.9995 - val_loss: 0.9978\n",
      "Epoch 245/1000\n",
      "315/315 [==============================] - 0s 886us/step - loss: 0.9994 - val_loss: 0.9978\n",
      "Epoch 246/1000\n",
      "315/315 [==============================] - 0s 915us/step - loss: 0.9996 - val_loss: 0.9979\n",
      "Epoch 247/1000\n",
      "315/315 [==============================] - 0s 885us/step - loss: 0.9996 - val_loss: 0.9979\n",
      "Epoch 248/1000\n",
      "315/315 [==============================] - 0s 900us/step - loss: 0.9995 - val_loss: 0.9979\n",
      "Epoch 249/1000\n",
      "315/315 [==============================] - 0s 922us/step - loss: 0.9989 - val_loss: 0.9979\n",
      "Epoch 250/1000\n",
      "315/315 [==============================] - 0s 922us/step - loss: 0.9997 - val_loss: 0.9979\n",
      "Epoch 251/1000\n",
      "315/315 [==============================] - 0s 932us/step - loss: 0.9992 - val_loss: 0.9979\n",
      "Epoch 252/1000\n",
      "315/315 [==============================] - 0s 940us/step - loss: 0.9999 - val_loss: 0.9978\n",
      "Epoch 253/1000\n",
      "315/315 [==============================] - 0s 910us/step - loss: 0.9996 - val_loss: 0.9979\n",
      "Epoch 254/1000\n",
      "315/315 [==============================] - 0s 939us/step - loss: 0.9986 - val_loss: 0.9978\n",
      "Epoch 255/1000\n",
      "315/315 [==============================] - 0s 943us/step - loss: 0.9997 - val_loss: 0.9978\n",
      "Epoch 256/1000\n",
      "315/315 [==============================] - 0s 919us/step - loss: 0.9997 - val_loss: 0.9979\n",
      "Epoch 257/1000\n",
      "315/315 [==============================] - 0s 912us/step - loss: 0.9996 - val_loss: 0.9979\n",
      "Epoch 258/1000\n",
      "315/315 [==============================] - 0s 914us/step - loss: 0.9991 - val_loss: 0.9979\n",
      "Epoch 259/1000\n",
      "315/315 [==============================] - 0s 940us/step - loss: 0.9996 - val_loss: 0.9979\n",
      "Epoch 260/1000\n",
      "315/315 [==============================] - 0s 924us/step - loss: 0.9995 - val_loss: 0.9979\n",
      "Epoch 261/1000\n",
      "315/315 [==============================] - 0s 934us/step - loss: 0.9992 - val_loss: 0.9978\n",
      "Epoch 262/1000\n",
      "315/315 [==============================] - 0s 929us/step - loss: 0.9996 - val_loss: 0.9978\n",
      "Epoch 263/1000\n",
      "315/315 [==============================] - 0s 921us/step - loss: 0.9998 - val_loss: 0.9978\n",
      "Epoch 264/1000\n",
      "315/315 [==============================] - 0s 922us/step - loss: 0.9995 - val_loss: 0.9979\n",
      "Epoch 265/1000\n",
      "315/315 [==============================] - 0s 940us/step - loss: 0.9992 - val_loss: 0.9979\n",
      "Epoch 266/1000\n",
      "315/315 [==============================] - 0s 910us/step - loss: 0.9996 - val_loss: 0.9979\n",
      "Epoch 267/1000\n",
      "315/315 [==============================] - 0s 920us/step - loss: 0.9991 - val_loss: 0.9979\n",
      "Epoch 268/1000\n",
      "315/315 [==============================] - 0s 946us/step - loss: 0.9994 - val_loss: 0.9979\n",
      "Epoch 269/1000\n",
      "315/315 [==============================] - 0s 939us/step - loss: 0.9995 - val_loss: 0.9978\n",
      "Epoch 270/1000\n",
      "315/315 [==============================] - 0s 911us/step - loss: 0.9995 - val_loss: 0.9978\n",
      "Epoch 271/1000\n",
      "315/315 [==============================] - 0s 912us/step - loss: 0.9997 - val_loss: 0.9978\n",
      "Epoch 272/1000\n",
      "315/315 [==============================] - 0s 947us/step - loss: 0.9998 - val_loss: 0.9978\n",
      "Epoch 273/1000\n",
      "315/315 [==============================] - 0s 936us/step - loss: 0.9995 - val_loss: 0.9977\n",
      "Epoch 274/1000\n",
      "315/315 [==============================] - 0s 938us/step - loss: 0.9996 - val_loss: 0.9978\n",
      "Epoch 275/1000\n",
      "315/315 [==============================] - 0s 939us/step - loss: 0.9997 - val_loss: 0.9978\n",
      "Epoch 276/1000\n",
      "315/315 [==============================] - 0s 929us/step - loss: 0.9990 - val_loss: 0.9978\n",
      "Epoch 277/1000\n",
      "315/315 [==============================] - 0s 940us/step - loss: 0.9992 - val_loss: 0.9977\n",
      "Epoch 278/1000\n",
      "315/315 [==============================] - 0s 918us/step - loss: 0.9995 - val_loss: 0.9977\n",
      "Epoch 279/1000\n",
      "315/315 [==============================] - 0s 913us/step - loss: 0.9998 - val_loss: 0.9977\n",
      "Epoch 280/1000\n",
      "315/315 [==============================] - 0s 902us/step - loss: 0.9995 - val_loss: 0.9978\n",
      "Epoch 281/1000\n",
      "315/315 [==============================] - 0s 917us/step - loss: 0.9995 - val_loss: 0.9979\n",
      "Epoch 282/1000\n",
      "315/315 [==============================] - 0s 919us/step - loss: 0.9995 - val_loss: 0.9978\n",
      "Epoch 283/1000\n",
      "315/315 [==============================] - 0s 944us/step - loss: 0.9991 - val_loss: 0.9979\n",
      "Epoch 284/1000\n",
      "315/315 [==============================] - 0s 913us/step - loss: 0.9996 - val_loss: 0.9979\n",
      "Epoch 285/1000\n",
      "315/315 [==============================] - 0s 918us/step - loss: 0.9990 - val_loss: 0.9978\n",
      "Epoch 286/1000\n",
      "315/315 [==============================] - 0s 895us/step - loss: 0.9996 - val_loss: 0.9978\n",
      "Epoch 287/1000\n",
      "315/315 [==============================] - 0s 923us/step - loss: 0.9996 - val_loss: 0.9978\n",
      "Epoch 288/1000\n",
      "315/315 [==============================] - 0s 917us/step - loss: 0.9994 - val_loss: 0.9978\n",
      "Epoch 289/1000\n",
      "315/315 [==============================] - 0s 952us/step - loss: 0.9991 - val_loss: 0.9977\n",
      "Epoch 290/1000\n",
      "315/315 [==============================] - 0s 909us/step - loss: 0.9993 - val_loss: 0.9977\n",
      "Epoch 291/1000\n",
      "315/315 [==============================] - 0s 921us/step - loss: 0.9990 - val_loss: 0.9977\n",
      "Epoch 292/1000\n",
      "315/315 [==============================] - 0s 939us/step - loss: 0.9994 - val_loss: 0.9977\n",
      "Epoch 293/1000\n",
      "315/315 [==============================] - 0s 933us/step - loss: 0.9991 - val_loss: 0.9977\n",
      "Epoch 294/1000\n",
      "315/315 [==============================] - 0s 907us/step - loss: 0.9994 - val_loss: 0.9977\n",
      "Epoch 295/1000\n",
      "315/315 [==============================] - 0s 962us/step - loss: 0.9993 - val_loss: 0.9977\n",
      "Epoch 296/1000\n",
      "315/315 [==============================] - 0s 944us/step - loss: 0.9990 - val_loss: 0.9978\n",
      "Epoch 297/1000\n",
      "315/315 [==============================] - 0s 951us/step - loss: 0.9989 - val_loss: 0.9978\n",
      "Epoch 298/1000\n",
      "315/315 [==============================] - 0s 946us/step - loss: 0.9997 - val_loss: 0.9978\n",
      "Epoch 299/1000\n",
      "315/315 [==============================] - 0s 955us/step - loss: 0.9994 - val_loss: 0.9978\n",
      "Epoch 300/1000\n",
      "315/315 [==============================] - 0s 953us/step - loss: 0.9995 - val_loss: 0.9978\n",
      "Epoch 301/1000\n",
      "315/315 [==============================] - 0s 928us/step - loss: 0.9994 - val_loss: 0.9978\n",
      "Epoch 302/1000\n",
      "315/315 [==============================] - 0s 922us/step - loss: 0.9991 - val_loss: 0.9977\n",
      "Epoch 303/1000\n",
      "315/315 [==============================] - 0s 941us/step - loss: 0.9995 - val_loss: 0.9977\n",
      "Epoch 304/1000\n",
      "315/315 [==============================] - 0s 943us/step - loss: 0.9990 - val_loss: 0.9978\n",
      "Epoch 305/1000\n",
      "315/315 [==============================] - 0s 911us/step - loss: 0.9997 - val_loss: 0.9978\n",
      "Epoch 306/1000\n",
      "315/315 [==============================] - 0s 960us/step - loss: 0.9995 - val_loss: 0.9978\n",
      "Epoch 307/1000\n",
      "315/315 [==============================] - 0s 914us/step - loss: 0.9995 - val_loss: 0.9977\n",
      "Epoch 308/1000\n",
      "315/315 [==============================] - 0s 908us/step - loss: 0.9994 - val_loss: 0.9978\n",
      "Epoch 309/1000\n",
      "315/315 [==============================] - 0s 948us/step - loss: 0.9996 - val_loss: 0.9978\n",
      "Epoch 310/1000\n",
      "315/315 [==============================] - 0s 939us/step - loss: 0.9993 - val_loss: 0.9978\n",
      "Epoch 311/1000\n",
      "315/315 [==============================] - 0s 925us/step - loss: 0.9995 - val_loss: 0.9979\n",
      "Epoch 312/1000\n",
      "315/315 [==============================] - 0s 948us/step - loss: 0.9992 - val_loss: 0.9979\n",
      "Epoch 313/1000\n",
      "315/315 [==============================] - 0s 931us/step - loss: 0.9992 - val_loss: 0.9978\n",
      "Epoch 314/1000\n",
      "315/315 [==============================] - 0s 919us/step - loss: 0.9989 - val_loss: 0.9979\n",
      "Epoch 315/1000\n",
      "315/315 [==============================] - 0s 920us/step - loss: 0.9996 - val_loss: 0.9979\n",
      "Epoch 316/1000\n",
      "315/315 [==============================] - 0s 930us/step - loss: 0.9993 - val_loss: 0.9979\n",
      "Epoch 317/1000\n",
      "315/315 [==============================] - 0s 934us/step - loss: 0.9993 - val_loss: 0.9979\n",
      "Epoch 318/1000\n",
      "315/315 [==============================] - 0s 912us/step - loss: 0.9993 - val_loss: 0.9979\n",
      "Epoch 319/1000\n",
      "315/315 [==============================] - 0s 901us/step - loss: 0.9996 - val_loss: 0.9979\n",
      "Epoch 320/1000\n",
      "315/315 [==============================] - 0s 905us/step - loss: 0.9993 - val_loss: 0.9979\n",
      "Epoch 321/1000\n",
      "315/315 [==============================] - 0s 900us/step - loss: 0.9994 - val_loss: 0.9979\n",
      "Epoch 322/1000\n",
      "315/315 [==============================] - 0s 896us/step - loss: 0.9995 - val_loss: 0.9978\n",
      "Epoch 323/1000\n",
      "315/315 [==============================] - 0s 887us/step - loss: 0.9996 - val_loss: 0.9978\n",
      "Epoch 324/1000\n",
      "315/315 [==============================] - 0s 901us/step - loss: 0.9989 - val_loss: 0.9978\n",
      "Epoch 325/1000\n",
      "315/315 [==============================] - 0s 890us/step - loss: 0.9991 - val_loss: 0.9979\n",
      "Epoch 326/1000\n",
      "315/315 [==============================] - 0s 905us/step - loss: 0.9995 - val_loss: 0.9979\n",
      "Epoch 327/1000\n",
      "315/315 [==============================] - 0s 932us/step - loss: 0.9993 - val_loss: 0.9979\n",
      "Epoch 328/1000\n",
      "315/315 [==============================] - 0s 910us/step - loss: 0.9993 - val_loss: 0.9979\n",
      "Epoch 329/1000\n",
      "315/315 [==============================] - 0s 915us/step - loss: 0.9995 - val_loss: 0.9979\n",
      "Epoch 330/1000\n",
      "315/315 [==============================] - 0s 923us/step - loss: 0.9993 - val_loss: 0.9978\n",
      "Epoch 331/1000\n",
      "315/315 [==============================] - 0s 912us/step - loss: 0.9994 - val_loss: 0.9978\n",
      "Epoch 332/1000\n",
      "315/315 [==============================] - 0s 912us/step - loss: 0.9997 - val_loss: 0.9978\n",
      "Epoch 333/1000\n",
      "315/315 [==============================] - 0s 886us/step - loss: 0.9995 - val_loss: 0.9978\n",
      "Epoch 334/1000\n",
      "315/315 [==============================] - 0s 900us/step - loss: 0.9995 - val_loss: 0.9978\n",
      "Epoch 335/1000\n",
      "315/315 [==============================] - 0s 898us/step - loss: 0.9996 - val_loss: 0.9978\n",
      "Epoch 336/1000\n",
      "315/315 [==============================] - 0s 914us/step - loss: 0.9992 - val_loss: 0.9979\n",
      "Epoch 337/1000\n",
      "315/315 [==============================] - 0s 866us/step - loss: 0.9995 - val_loss: 0.9979\n",
      "Epoch 338/1000\n",
      "315/315 [==============================] - 0s 884us/step - loss: 0.9995 - val_loss: 0.9978\n",
      "Epoch 339/1000\n",
      "315/315 [==============================] - 0s 928us/step - loss: 0.9993 - val_loss: 0.9978\n",
      "Epoch 340/1000\n",
      "315/315 [==============================] - 0s 911us/step - loss: 0.9997 - val_loss: 0.9978\n",
      "Epoch 341/1000\n",
      "315/315 [==============================] - 0s 917us/step - loss: 0.9997 - val_loss: 0.9978\n",
      "Epoch 342/1000\n",
      "315/315 [==============================] - 0s 911us/step - loss: 0.9997 - val_loss: 0.9978\n",
      "Epoch 343/1000\n",
      "315/315 [==============================] - 0s 913us/step - loss: 0.9992 - val_loss: 0.9978\n",
      "Epoch 344/1000\n",
      "315/315 [==============================] - 0s 902us/step - loss: 0.9994 - val_loss: 0.9978\n",
      "Epoch 345/1000\n",
      "315/315 [==============================] - 0s 903us/step - loss: 0.9998 - val_loss: 0.9978\n",
      "Epoch 346/1000\n",
      "315/315 [==============================] - 0s 926us/step - loss: 0.9994 - val_loss: 0.9979\n",
      "Epoch 347/1000\n",
      "315/315 [==============================] - 0s 911us/step - loss: 0.9992 - val_loss: 0.9980\n",
      "Epoch 348/1000\n",
      "315/315 [==============================] - 0s 900us/step - loss: 0.9988 - val_loss: 0.9979\n",
      "Epoch 349/1000\n",
      "315/315 [==============================] - 0s 912us/step - loss: 0.9994 - val_loss: 0.9979\n",
      "Epoch 350/1000\n",
      "315/315 [==============================] - 0s 904us/step - loss: 0.9995 - val_loss: 0.9979\n",
      "Epoch 351/1000\n",
      "315/315 [==============================] - 0s 902us/step - loss: 0.9997 - val_loss: 0.9979\n",
      "Epoch 352/1000\n",
      "315/315 [==============================] - 0s 918us/step - loss: 0.9986 - val_loss: 0.9978\n",
      "Epoch 353/1000\n",
      "315/315 [==============================] - 0s 901us/step - loss: 0.9996 - val_loss: 0.9978\n",
      "Epoch 354/1000\n",
      "315/315 [==============================] - 0s 912us/step - loss: 0.9992 - val_loss: 0.9978\n",
      "Epoch 355/1000\n",
      "315/315 [==============================] - 0s 877us/step - loss: 0.9996 - val_loss: 0.9979\n",
      "Epoch 356/1000\n",
      "315/315 [==============================] - 0s 911us/step - loss: 0.9991 - val_loss: 0.9978\n",
      "Epoch 357/1000\n",
      "315/315 [==============================] - 0s 937us/step - loss: 0.9990 - val_loss: 0.9978\n",
      "Epoch 358/1000\n",
      "315/315 [==============================] - 0s 909us/step - loss: 0.9994 - val_loss: 0.9977\n",
      "Epoch 359/1000\n",
      "315/315 [==============================] - 0s 920us/step - loss: 0.9993 - val_loss: 0.9977\n",
      "Epoch 360/1000\n",
      "315/315 [==============================] - 0s 905us/step - loss: 0.9994 - val_loss: 0.9978\n",
      "Epoch 361/1000\n",
      "315/315 [==============================] - 0s 916us/step - loss: 0.9996 - val_loss: 0.9978\n",
      "Epoch 362/1000\n",
      "315/315 [==============================] - 0s 917us/step - loss: 0.9995 - val_loss: 0.9978\n",
      "Epoch 363/1000\n",
      "315/315 [==============================] - 0s 893us/step - loss: 0.9994 - val_loss: 0.9979\n",
      "Epoch 364/1000\n",
      "315/315 [==============================] - 0s 903us/step - loss: 0.9992 - val_loss: 0.9980\n",
      "Epoch 365/1000\n",
      "315/315 [==============================] - 0s 910us/step - loss: 0.9998 - val_loss: 0.9980\n",
      "Epoch 366/1000\n",
      "315/315 [==============================] - 0s 900us/step - loss: 0.9998 - val_loss: 0.9979\n",
      "Epoch 367/1000\n",
      "315/315 [==============================] - 0s 911us/step - loss: 0.9993 - val_loss: 0.9980\n",
      "Epoch 368/1000\n",
      "315/315 [==============================] - 0s 915us/step - loss: 0.9995 - val_loss: 0.9981\n",
      "Epoch 369/1000\n",
      "315/315 [==============================] - 0s 905us/step - loss: 0.9991 - val_loss: 0.9981\n",
      "Epoch 370/1000\n",
      "315/315 [==============================] - 0s 912us/step - loss: 0.9996 - val_loss: 0.9982\n",
      "Epoch 371/1000\n",
      "315/315 [==============================] - 0s 934us/step - loss: 0.9991 - val_loss: 0.9981\n",
      "Epoch 372/1000\n",
      "315/315 [==============================] - 0s 894us/step - loss: 0.9995 - val_loss: 0.9981\n",
      "Epoch 373/1000\n",
      "315/315 [==============================] - 0s 911us/step - loss: 0.9993 - val_loss: 0.9981\n",
      "Epoch 374/1000\n",
      "315/315 [==============================] - 0s 883us/step - loss: 0.9993 - val_loss: 0.9980\n",
      "Epoch 375/1000\n",
      "315/315 [==============================] - 0s 906us/step - loss: 0.9994 - val_loss: 0.9981\n",
      "Epoch 376/1000\n",
      "315/315 [==============================] - 0s 905us/step - loss: 0.9994 - val_loss: 0.9981\n",
      "Epoch 377/1000\n",
      "315/315 [==============================] - 0s 892us/step - loss: 0.9991 - val_loss: 0.9980\n",
      "Epoch 378/1000\n",
      "315/315 [==============================] - 0s 900us/step - loss: 0.9989 - val_loss: 0.9980\n",
      "Epoch 379/1000\n",
      "315/315 [==============================] - 0s 917us/step - loss: 0.9996 - val_loss: 0.9980\n",
      "Epoch 380/1000\n",
      "315/315 [==============================] - 0s 908us/step - loss: 0.9993 - val_loss: 0.9981\n",
      "Epoch 381/1000\n",
      "315/315 [==============================] - 0s 919us/step - loss: 0.9991 - val_loss: 0.9980\n",
      "Epoch 382/1000\n",
      "315/315 [==============================] - 0s 895us/step - loss: 0.9993 - val_loss: 0.9980\n",
      "Epoch 383/1000\n",
      "315/315 [==============================] - 0s 920us/step - loss: 0.9997 - val_loss: 0.9980\n",
      "Epoch 384/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "315/315 [==============================] - 0s 932us/step - loss: 0.9996 - val_loss: 0.9981\n",
      "Epoch 385/1000\n",
      "315/315 [==============================] - 0s 889us/step - loss: 0.9994 - val_loss: 0.9981\n",
      "Epoch 386/1000\n",
      "315/315 [==============================] - 0s 920us/step - loss: 0.9996 - val_loss: 0.9981\n",
      "Epoch 387/1000\n",
      "315/315 [==============================] - 0s 929us/step - loss: 0.9997 - val_loss: 0.9981\n",
      "Epoch 388/1000\n",
      "315/315 [==============================] - 0s 901us/step - loss: 0.9992 - val_loss: 0.9980\n",
      "Epoch 389/1000\n",
      "315/315 [==============================] - 0s 913us/step - loss: 0.9995 - val_loss: 0.9981\n",
      "Epoch 390/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9982\n",
      "Epoch 391/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9996 - val_loss: 0.9981\n",
      "Epoch 392/1000\n",
      "315/315 [==============================] - 0s 980us/step - loss: 0.9994 - val_loss: 0.9981\n",
      "Epoch 393/1000\n",
      "315/315 [==============================] - 0s 918us/step - loss: 0.9991 - val_loss: 0.9980\n",
      "Epoch 394/1000\n",
      "315/315 [==============================] - 0s 911us/step - loss: 0.9994 - val_loss: 0.9980\n",
      "Epoch 395/1000\n",
      "315/315 [==============================] - 0s 921us/step - loss: 0.9995 - val_loss: 0.9980\n",
      "Epoch 396/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9996 - val_loss: 0.9980\n",
      "Epoch 397/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9979\n",
      "Epoch 398/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9994 - val_loss: 0.9980\n",
      "Epoch 399/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9996 - val_loss: 0.9981\n",
      "Epoch 400/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9980\n",
      "Epoch 401/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9986 - val_loss: 0.9979\n",
      "Epoch 402/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9991 - val_loss: 0.9979\n",
      "Epoch 403/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9990 - val_loss: 0.9978\n",
      "Epoch 404/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9979\n",
      "Epoch 405/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9978\n",
      "Epoch 406/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9978\n",
      "Epoch 407/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9977\n",
      "Epoch 408/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9994 - val_loss: 0.9978\n",
      "Epoch 409/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9994 - val_loss: 0.9978\n",
      "Epoch 410/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9994 - val_loss: 0.9979\n",
      "Epoch 411/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9994 - val_loss: 0.9979\n",
      "Epoch 412/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9980\n",
      "Epoch 413/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9988 - val_loss: 0.9980\n",
      "Epoch 414/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9980\n",
      "Epoch 415/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9980\n",
      "Epoch 416/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9996 - val_loss: 0.9981\n",
      "Epoch 417/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9994 - val_loss: 0.9981\n",
      "Epoch 418/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9990 - val_loss: 0.9982\n",
      "Epoch 419/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9982\n",
      "Epoch 420/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9991 - val_loss: 0.9982\n",
      "Epoch 421/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9996 - val_loss: 0.9983\n",
      "Epoch 422/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9982\n",
      "Epoch 423/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9982\n",
      "Epoch 424/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9991 - val_loss: 0.9981\n",
      "Epoch 425/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9987 - val_loss: 0.9982\n",
      "Epoch 426/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9982\n",
      "Epoch 427/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9982\n",
      "Epoch 428/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9994 - val_loss: 0.9982\n",
      "Epoch 429/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9982\n",
      "Epoch 430/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9987 - val_loss: 0.9982\n",
      "Epoch 431/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9982\n",
      "Epoch 432/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9983\n",
      "Epoch 433/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9983\n",
      "Epoch 434/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9983\n",
      "Epoch 435/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9982\n",
      "Epoch 436/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9996 - val_loss: 0.9982\n",
      "Epoch 437/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9989 - val_loss: 0.9983\n",
      "Epoch 438/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9983\n",
      "Epoch 439/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9989 - val_loss: 0.9984\n",
      "Epoch 440/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9996 - val_loss: 0.9984\n",
      "Epoch 441/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9984\n",
      "Epoch 442/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9983\n",
      "Epoch 443/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9994 - val_loss: 0.9984\n",
      "Epoch 444/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9984\n",
      "Epoch 445/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9989 - val_loss: 0.9985\n",
      "Epoch 446/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9994 - val_loss: 0.9984\n",
      "Epoch 447/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9989 - val_loss: 0.9984\n",
      "Epoch 448/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9989 - val_loss: 0.9985\n",
      "Epoch 449/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9990 - val_loss: 0.9984\n",
      "Epoch 450/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9983\n",
      "Epoch 451/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9983\n",
      "Epoch 452/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9983\n",
      "Epoch 453/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9983\n",
      "Epoch 454/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9990 - val_loss: 0.9983\n",
      "Epoch 455/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9982\n",
      "Epoch 456/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9982\n",
      "Epoch 457/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9983\n",
      "Epoch 458/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9986 - val_loss: 0.9983\n",
      "Epoch 459/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9984\n",
      "Epoch 460/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9984\n",
      "Epoch 461/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9990 - val_loss: 0.9984\n",
      "Epoch 462/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9996 - val_loss: 0.9984\n",
      "Epoch 463/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9985\n",
      "Epoch 464/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9991 - val_loss: 0.9985\n",
      "Epoch 465/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9994 - val_loss: 0.9984\n",
      "Epoch 466/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9994 - val_loss: 0.9985\n",
      "Epoch 467/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9985\n",
      "Epoch 468/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9985\n",
      "Epoch 469/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9991 - val_loss: 0.9985\n",
      "Epoch 470/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9985\n",
      "Epoch 471/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9991 - val_loss: 0.9986\n",
      "Epoch 472/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9985\n",
      "Epoch 473/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9985\n",
      "Epoch 474/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9986\n",
      "Epoch 475/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9990 - val_loss: 0.9985\n",
      "Epoch 476/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9997 - val_loss: 0.9986\n",
      "Epoch 477/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9987\n",
      "Epoch 478/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9988 - val_loss: 0.9986\n",
      "Epoch 479/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9990 - val_loss: 0.9986\n",
      "Epoch 480/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9988 - val_loss: 0.9985\n",
      "Epoch 481/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9996 - val_loss: 0.9985\n",
      "Epoch 482/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9986\n",
      "Epoch 483/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9989 - val_loss: 0.9986\n",
      "Epoch 484/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9990 - val_loss: 0.9986\n",
      "Epoch 485/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9991 - val_loss: 0.9986\n",
      "Epoch 486/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9991 - val_loss: 0.9986\n",
      "Epoch 487/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9986\n",
      "Epoch 488/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9990 - val_loss: 0.9986\n",
      "Epoch 489/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9991 - val_loss: 0.9986\n",
      "Epoch 490/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9986\n",
      "Epoch 491/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9994 - val_loss: 0.9986\n",
      "Epoch 492/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9991 - val_loss: 0.9986\n",
      "Epoch 493/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9987\n",
      "Epoch 494/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9987\n",
      "Epoch 495/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9988\n",
      "Epoch 496/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9988\n",
      "Epoch 497/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9988\n",
      "Epoch 498/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9996 - val_loss: 0.9988\n",
      "Epoch 499/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9987\n",
      "Epoch 500/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9989 - val_loss: 0.9987\n",
      "Epoch 501/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9987\n",
      "Epoch 502/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9988 - val_loss: 0.9986\n",
      "Epoch 503/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9987\n",
      "Epoch 504/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9986 - val_loss: 0.9987\n",
      "Epoch 505/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9996 - val_loss: 0.9986\n",
      "Epoch 506/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9990 - val_loss: 0.9986\n",
      "Epoch 507/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9986\n",
      "Epoch 508/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9987\n",
      "Epoch 509/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9987\n",
      "Epoch 510/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9988\n",
      "Epoch 511/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9988\n",
      "Epoch 512/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9990 - val_loss: 0.9988\n",
      "Epoch 513/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9994 - val_loss: 0.9988\n",
      "Epoch 514/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9988\n",
      "Epoch 515/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9990 - val_loss: 0.9988\n",
      "Epoch 516/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9988\n",
      "Epoch 517/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9990 - val_loss: 0.9989\n",
      "Epoch 518/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9989\n",
      "Epoch 519/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9994 - val_loss: 0.9989\n",
      "Epoch 520/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9989 - val_loss: 0.9990\n",
      "Epoch 521/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9990 - val_loss: 0.9989\n",
      "Epoch 522/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9986 - val_loss: 0.9989\n",
      "Epoch 523/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9991 - val_loss: 0.9990\n",
      "Epoch 524/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9990\n",
      "Epoch 525/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9989\n",
      "Epoch 526/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9991 - val_loss: 0.9990\n",
      "Epoch 527/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9990\n",
      "Epoch 528/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9990\n",
      "Epoch 529/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9990 - val_loss: 0.9990\n",
      "Epoch 530/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9990\n",
      "Epoch 531/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9990 - val_loss: 0.9991\n",
      "Epoch 532/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9987 - val_loss: 0.9990\n",
      "Epoch 533/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9995 - val_loss: 0.9990\n",
      "Epoch 534/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9991 - val_loss: 0.9990\n",
      "Epoch 535/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9994 - val_loss: 0.9990\n",
      "Epoch 536/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9989 - val_loss: 0.9990\n",
      "Epoch 537/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9993 - val_loss: 0.9990\n",
      "Epoch 538/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9991 - val_loss: 0.9991\n",
      "Epoch 539/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9989 - val_loss: 0.9991\n",
      "Epoch 540/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9991 - val_loss: 0.9991\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 541/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9991 - val_loss: 0.9991\n",
      "Epoch 542/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9990 - val_loss: 0.9992\n",
      "Epoch 543/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9989 - val_loss: 0.9991\n",
      "Epoch 544/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9988 - val_loss: 0.9991\n",
      "Epoch 545/1000\n",
      "315/315 [==============================] - 0s 1ms/step - loss: 0.9992 - val_loss: 0.9991\n",
      "Epoch 546/1000\n",
      " 40/315 [==>...........................] - ETA: 0s - loss: 0.9985"
     ]
    }
   ],
   "source": [
    "# Training time!\n",
    "# Try 20 epochs using batch size = 3, then switch to 4\n",
    "model.fit(x_train, y_train,\n",
    "                epochs=1000,\n",
    "                batch_size=40,\n",
    "                shuffle=True,\n",
    "                validation_data=(x_test, y_test),\n",
    "                callbacks=[TensorBoard(log_dir='/home/amplifier/home/NEW_DL/logs')])\n",
    "\n",
    "# save the entire model and the weights after the training:\n",
    "model.save('/home/amplifier/home/NEW_DL/models/EEGnet.h5')\n",
    "model.save_weights('/home/amplifier/home/NEW_DL/weights/EEGnet_weights.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(189, 1, 800, 1)\n",
      "(81, 1, 800, 1)\n"
     ]
    }
   ],
   "source": [
    "# let's predict a TRAINING sammple:\n",
    "decoded_train = model.predict(x_train)\n",
    "decoded_test = model.predict(x_test)\n",
    "print(decoded_train.shape)\n",
    "print(decoded_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Pearson's R: [ 0.0345341] p-value [ 0.32930166]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Text(0.5,1,'Original')"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfwAAAEICAYAAAC6UUYcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4xLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvAOZPmwAAIABJREFUeJzsnXd4HMX5+D/v3elULbkbdwOmmWbAmN57AiEkJAHSCJCekG8KqZCQ/FJI7z0QakgIPSH03ovBYBuwsXHvTV26Or8/ZmZv7m7vVG3J8nyeR4+k3b29d3dn5523zDuilMLj8Xg8Hs/QJjLQAng8Ho/H49n2eIXv8Xg8Hs9OgFf4Ho/H4/HsBHiF7/F4PB7PToBX+B6Px+Px7AR4he/xeDwez06AV/jbERE5XkRWldnfKiK7bU+ZSsgxTUSUiMQGWA4lItMHUoaBYrA8g22JiCwTkQ4RuWGgZRmsiMgSEUmKyI0DLYuLiFzZHzKZPjFr+r7T+0O2nQ0ROdncv6yInFzu2C4VvvNStorIOhG5VkTq+k/c/qO/GmGZ818rIt/fVudXStUppd7eVue3iMhjInLJtv6ewY6ITDHt2v4oEWlz/j/GPPNkwXGvOue4WETeFJEWEVkvIveIyDARudc5PlVwjj/183WUHUj28pzjReRqEVlrru1NEfmuiNT25/cAZymlPlxGjpkiMkdE2s3vmWWOHSkid5hnuFxELnD2nSAi80SkUUQ2m+MmljjHRhF5ytlmB19uG7jC2f8TEVkpIs3me79VQr6PmvNc4mw7QUQeFZEmEVlW+Bml1O7AD0tdc8h3uO11i4g8KCJ7d/fzA8Qa0/fdV+oAEfmi0T9NInKNiFSWOG6GiLwkIlvNz0MiMsPZf5mIzDdteqmIXFbw+SNF5AWz/zUROdrZV7YNiUilka3ZyPolZ19cRG4VrU+ViBwfIvvBIvKEeXbrReQLBfu/YGRuE5E3RGRPAKXUQ0qpOmBFuZsM3bfwzzInnAkcBHyjm58bVIjGezU8ACilVpiOps60b4ADnW1Pmm0/cY9TSh0IICLHoTvj85VSw4B9gFvMuc9wzntTwTk+tX2vtGeIyEjgWaAaOMJc2ynAcGD3Hp6r6J3r7nsoInHgLuBGYARwHXCX2R7G74EkMA74IPBHEdnX7HsdOE0pNRyYALwF/DHkHD8G3ihx/uHOM/x/zvargb2VUvXAkcAFIvKegmsZge43FxScsw24BriM/uMnpt1NAjYA1/bjubc7InIa8HXgJGAasBvw3RKHrwHOBUYCo4G7gX+6pwM+gm5PpwOfE5HzzPeMNMf/FN3WfwL8xzw76LoNXQnsAUwFTgC+Kvlei6eADwHrQq5xNHAf8GdgFDAdeMDZfwlwMfBOoA44E9hU4h6URilV9gdYBpzs/P8T4B7n/0rgZ+jRxXrgT0C1s/9sYC7QDCwBTjfbJ6Bv7hZgMfBx5zNXojvO64EW9Esyy9n/NWC12bcQ3RBOR7/sKaAVeNUc+xjwA+BpoMPcyMJruhK40fn/aOAZoBFYCVwIfMKcO2nO/x/nOm4DNgJLgUud81SjX7at6MZyGbCqzL1WwHTz97XoDuwec53PA7sXHHsp8Db6wf8UiJS4nmnm+Ji5Fxmg01zH70LksMd/Av0CrQW+7OyfjVYIjWbf74C42SfAL9EdTRPwGrBfN9vKZeZ8a4CL3PsRImOv20937r+z7Vrg+yWO/wpwZzfOW/IczjFRc282mWf6WfvMzP6PoRVRi9n/SbO9Ft2us+Z5tpp7U/IZdUPe7wPzbHsqccyRwIvmGb8IHOnse4zidy5s2zKc9zDkO05Fv+fibFuB6UMKjq1Fv5t7OttuAK4KObYS+BHwesH2I8w9+xjwVNj70417N9Hcu68WbP8T8BlzHy4J+dzJwLIS57wS533uSVtDK4jWEsfeB3yuYNurwHvM379G93/NwBzgmDCZgOMp6NfcZ4s2LL+O7v83o9/NkaU+GyLnP4AfOv+fBKzrxr2Iod+j9jLH/Ab4rfn7TGBBwf5FwMXdaUOmrZ7q/P//gH+GfHYVcHzBth8CN5SQMWKew0ldXG9wz0v99MjaFZFJwBnoDtbyY2BPtPU/Hd3gv22On43udC9Dj5iONUIB3GwufAJ6RPZDETnJOe+70COz4eiO/XfmnHsBnwMOVdryOA39otxnbtq/lGOFGT6MVl7DgOVdXOMU4F7gt8AYc11zlVJ/Id9SO8tYKf9BvyQT0Q3x/8yIFOA7aItodyPnR8t9dwjno0eyI9D3/AcF+88BZgEHowdWF3V1QqXUt4An0S96nVLqc2UOPwE9Yj0V+Lrk4kMZ4IvoEfQR6Ov+jNl3Kvo574l+dh9Av+RQvq2cjlagp5jvLBuLopftp595HjhNtKv7qFJuxm7ycXSHcxD6mZ5bsH+D2V+PVki/FJGDlVJt6HfSukXrlFJrKP+MuuJk4HalVDZsp7GE7kF3lqOAXwD3iMgo57Cwd67b76FhX+A1ZXozw2tmeyF7Ahml1CJn26vusaJDOI3oAcdX0MaL3RdFD7A/h1buYSwXkVUi8ndjkQWIyNdFpBXdJmvRSsrum41+pv0ayukK0aHXDwKvlDjkH+g+xh4/A22d3mM2vYh+V0eaY/8tIlW9EOVS4N3Acej3dSv6XneXfdHP0vIqMK6gveVhnnMnuh8PDYmIiADHkPO6iPnJOwzYz/lMaBsyXoAJIXKGtdUwDge2iMgzIrJBRP5jdBFoT80kYD/RoaOlps/psbe6ux+4U0Ra0KOMDWhFZm/Yx4EvKqW2KKVa0Df3PPO5i4FrlFIPKqWySqnVSqk3RWQy2or+mlKqUyk1F/gbukOwPKWU+p9SKoMeqVsFnkGPrmaISIVSaplSakkX8l+rlFqglEorpVJdHPtB4CGl1M1KqZRSarORL4xDgTFKqe8ppZJKx9//6lz/+4EfmHuzEt1B9oTblVIvKKXS6MFGYfzyx+bcK4Bf4by8/cR3lVJtSql5wN/t+ZVSc5RSz5n7uQzthjrOfCaF7tD3Rltmbyil1najrbwf+LtSar5RYleWEqqP7ac3fMXE7ezPdeY+PAm8Bz3gugfYLCK/MMqjp7wf+JVSaqVSagvaeghQSt2jlFqiNI+j3X3HlDpZF8+oK0ahvQKleCfwllLqBnP+m4E3gbOcY8LeuZ68h6Bdl00F25rQ7avHxyodwhmOHgRdbmS2XAo8r5SaE3LuTeh3fSpwiDnnTe4BSqmrzPaD0e2tCYKBxB+Az5caQG0DvmKU0mL0fbmwxHF3ADNFZKr5/4PoPicBoJS60fR/aaXUz9H97l69kOeTwLeUUqvMua8EzpXuJ6QWPlv7d1g7AMA85wb0AK7UgOdKtA78u/n/GWCCiJwvIhUi8lG0sVbjnLdUG7IhwUI5S8pYwCS0QfgFYAraW3yzsw+0MbU/2hA7H61fe0R3Ff67jTV9PLojt6PbMeibMcd2hmg30RizfzLajVPIBMB2+pblaIvP4sY52oEqEYkppRYD/4d+WBtE5J8iMqEL+Vd2sd+llMxhTEU3kEbn+r+JjiGCvk73u7tj1bgU3oPCZMnCc3d1H3pK6PlFZE8R+a9JTGlGK+7RAEqpR9DW9O+B9SLyFxGpp+u20pN71ev2052LDuFnSqnhzk/gqVFK3auUOgttBZ2N7lx7kxBZ9vpF5AwReU50IlYj8A5y72ER5Z5RN9gMjO9C1sLnU3j/w965nryHoMMT9QXb6tFhjV4fawZUNh8gZvqPS4HQZDulVKtS6iWj+Najlcippl27xyml1Cto68/GmD+D9lI8W+Y6+xvbXndRSr3LGkSSn3Q4xbw/95AbdJ+HM5ARkS+LTg5rMm2uge63IZepwB3Oe/8G2nAbV/5jAYXP1v4d1g4CjOHwJ+B6ERnr7hORz6Fj+e90Bjib0e/wl9Ahx9OBh9Bem8Jz57UhI6Mrm/27rIwOHcAdSqkXlVKd6PZzpIg0mH2gvcuNzgD+Hd08d0CPXALGsrgWHWsEPfLtAPZ1OsMGlUuAWkl4ks8aYKSIuKOfKegYSHfk+IdS6mh0Q1JoVzGUdsUVbm/DGbUBuzh/l5I57DwrgaUFymCYUso+iLXoAYRlCv1L4bnXmL/LXR+Uvk/dPf8f0SPbPZROVPomjitMKfUbpdQhaHfWnuiQTldtpSf3qk/tZ1ugtAfrYeARHBdgDyh5/SZUcBv6vRtnLIz/kbvnYc+z7DPqgoeAc8q4DNeg3z2XwvsfJlNPl+ZcABxgvEOWAyhOfAMda42JyB7OtgNLHAs6vjsW3SnPRg9wXheRdejY9WwzWArz1tjrKHU/Y+T6kJPQ93KdOfeRwM9FZFuEmMqi8hNPbUb3zcD5InIEOufoUQAROQadK/V+YIRpc02EX3Nef2Pu2Rhn/0rgjIJ+skop1d33dQH5HroDgfVGQXdFxMjmZtNfhEkCVErlKXOl1ONKqUOVUiPRHsO9gBdKnDtoQ0qpreh3uFDOUu2vkNfIfz/cNrYQnZ/S56Vte5Ox/ivgFBGZaVxUf0XHE8cCiMhEJ4Z9NfAxETlJRCJm395Ku7efAX4kIlUicgDaPXFTyPflISJ7iciJphPsRCuRjNm9HpjWjdjGXOA847YpjJfeBJwsIu83o/9RkpsKtB6dIWp5AWgWka+JSLWIREVkPxE51Oy/BfiGiIwQnf/w+a6ur4dcZs49Ge0K+pdzfceaeFMDxbMqCq+jFFeISI3oTOePOecfhk7kaRU95efT9gMicqiIHCYiFeiOoBMdW+2qrdwCXCh6Wk0NJmwURl/aT38iImeLyHnmGYjoWO1xwHO9ON0twKUiMkl0PPDrzr442p26EUiLyBlo955lPTDKPGtLyWdkZH9MRK4sIcsv0IrwOuvuNc/qF+Ze/w/YU0QuMO/IB4AZwH97cd3leAz9bl8qesqTzTd5pPBAY83dDnxPRGpF5Ci0tXaDkf89pu+IiMgYc42vGEvtXnRi3kzz8220G3imUipj2rP97Ch0aO4xpVST2fbJgjbwWeBhI9qF6Nkb9twvoa23bxm5IqLj4hX6X6mS0rMQMJ8JndbVS/6HHrx9D53/ZMMOw4A0us3FROTbFHtQLIvQHrR3mvf+cnR7tfwJ+IHTlsaIyNk9kPF64GLTN4ww57827EAROUVEDjJ9cT36OW/FzLwQkQ+ivV2nqJAp0OazFeazP0MnFN5v9pVrQ1bOy01b2BsdwrzWOXel5HIg4uZZ2wHU39EDw5nmHl6BDks2KqXa0X3vV0VP+Z1kzt3j963HCl8ptdFcmJ2H+jV0rOg50a7DhzBxHqXUC5gEI/To8HFylsH56JdsDTqW9B2l1IPdEKESuAptMa5Dj7C+afb92/zeLCIvlznHFegR+Fb0yxck2JiR7zuAL6MzwOeSG7Vdjc4daBSRO5WOD5+FfpGXGpn+hnZ9Yc693Ox7ANP59CN3obNn56Jdc1eba3gQ3UBeM/sLG8av0TG0rSJSLq/gcfSzfRjtJrTTRL4CXIB2V/2V3EAAdKfwV/S9XY52D1uPULm2ci96MPmIOaaoUy+gt+2nN3xV8t2hdjrMVvSL9xZaud4I/FQp1ZuBx1+B+9GJPi+jlRcAxvV6KXpQsBV97+929r+JttTeNm1zAuWfEWhvwtNhgpgO7Eh0PsbzovN3Hka/w4uNZXUm+h3ZDHwVOFMp1fNpQmVQSiXRyV4fQc82uAgdXkwCiMg3ReRe5yOfQVupG9D349NKKWthTUSHkFrQWfRZdNIrSqmEUmqd/THXmTJ/gx4c28/OBxLk58ucgw4DtqDbwG/ND6bDds+dBJqVUjbWeyzaaPkf2kvSgTMdqxDT2beaa+gzxp19OzpR8x/OrvvRA6FF6Pe4kxIhGXMtn0H3favRA33Xcv41ur0+YNrSc8BhPZDxPnRy3KNGluU4BoGILDCKHHSS7s3oZ7gEnRx8unGTg56BMgp4UcLrYnwV3Y+vRHt9znH2lWxDhu+Y71yO7jt/qvJrCyxEP9+J6PvbgdGHJhT6TXQ/vsHIfYHz2c+hn/sa9EySf6Cnc/YIUarPXgLPACAiCu2uXdzlwR6Pg1Ea/1ZKHTHAcixEd6p3uHkRnhzmHk0EblFKXSQiH0KHxXbIWiiFiMixaOWXAD5grWlP9xE9O+k2tDH8DqXUoyWP9Qp/x8QrfI/H4/H0BF91zuPxeDyenQBv4Xs8Ho/HsxPgLXyPx+PxeHYChuzSm4OB0aNHq2nTpg20GB6Px7NDMWfOnE1KqTFdH+npCV7hb0OmTZvGSy+9NNBieDwezw6FiPS0KqmnG3iXvsfj8Xg8OwFe4Xs8Ho/HsxPgFb7H4/EU8Kkb5vD7R32JC8/Qwit8j8fjKeC+Bev46f0LB1oMj6df8Qrf4/F4PJ6dAK/wPR6Px8EXI/MMVbzC93g8HodEOtv1QR7PDohX+B6Px+PQlkgPtAgezzbBK3yPx+NxaE9mBloEj2eb4BW+x+PxOLR6C98zRPEK3+PxeBzak17he4YmXuF7PB6PQzLts/Q9QxOv8AsQkWtEZIOIzHe2jRSRB0XkLfN7xEDK6PF4th1ZPy3PM0TxCr+Ya4HTC7Z9HXhYKbUH8LD53+PxDEHSWa/wPUMTr/ALUEo9AWwp2Hw2cJ35+zrg3dtVKI/Hs93IOgrfF+HxDCW8wu8e45RSawHM77GlDhSRT4jISyLy0saNG7ebgB6Pp3/IOAo/lfEK3zN08Aq/n1FK/UUpNUspNWvMmDEDLY7H4+khrks/nfVV9zxDB6/wu8d6ERkPYH5vGGB5PB7PNsJN2vMWvmco4RV+97gb+Kj5+6PAXQMoi8fj2Ya4Lv10xlv4nqGDV/gFiMjNwLPAXiKySkQuBq4CThGRt4BTzP8ej2cIkqfwfca+ZwgRG2gBBhtKqfNL7Dppuwri8XgGhPykPW/he4YO3sL3eDweh4xyXfrewvcMHbzC93iGIE++tZGF61oGWowdEu/S9wxVvML3eIYYbYk0H776Bc790zMDLcoOSWaITsvLZBUdfunfnRqv8D2eIcbra5sBv657b8kOUZf+N2+fxz7fvs9XD9yJ8Qrf4xlipNLaKo1GZIAl2TFxlfxQStr710srAWju9Mv/7qx4he/xDDFs3Dnj48+9Is/CH0L3sLoiCsCG5s4BlsQzUHiF7/EMMTJe4feJoTotr65Kz8Le0JIYYEk8A4VX+B7PEGMoWaUDQV4t/SEUw6+Nawt/U6tX+DsrXuF7PEMMXw62b2SHaJa+vapEauhck6dneIXv8Qwx+mrh7+xZ3JkhuniODVUk0n72xs6KV/gezxCjL7H7JxZtZNdv/I+VW9r7UaIdi8wQdennFL638HdWvML3eIYYfbHwH3x9PQC3zlnVX+LscAzlwjvgFf7OjFf4Hs8QI+MoqZ665+urdSZ3a2Lg5mqnM1n+/dJKOlMD43oeqi59O90wMUD31TPweIXv8QwxXCW1sTXBu3//NMs2tXXrs1vbUwC0DmBxlmeWbOayW1/jijvnD8j3ZzKuSz9/8HTl3Qv425NvD4RYfSbtLfydHq/wPZ4hhuuSvnvuGuaubOR3jy5mS1uSOcu3lv3s1rYkAC2J1DaVsRzWu/DyivKybisyJQrvLNnYxrXPLOP797wxEGL1GTuQ8Qp/58UrfI9niOEqKbtYSkTgkute5L1/fIZkmQ5/a7tR+ANo4Td16MFGlakMt71xp+W5VffsvdlRsQMZn6W/8+IVvsczxHBj+B0mXisIb21oBWDl1tIZ+HbBneaOgbPwG01YoTI2MN1TOquImXUI3Cx9K9eOSuDS9/Pwd1q8wvd4hhh5Fr5R+JEIjKuvAigbz7eJcgNp4Td2aEu6MF1u+ebu5SH0laxSxM1gww2PNO7oFr6P4e/0eIXv8QwxXKs0t/65BKVVt5axVDuN9TeQWfpNRr5OxxK9a+5qjvvpYzy+aCNn/vZJ7pu/dpt9fybrKHzHpd/keD0GagZBb1FK+cI7Hq/wPZ6hhmvhW0tdBJJmIFCu9K5VZAOp0Kzr3J0+9tIyncD3h0cXM391M794cNE2+/50VhGPhln4OYXf3Lljuffd0gzewt958Qrf4xliuDH8zW16oRQht/JbqkxhnkDhD6BSsC79jpBBx/NLtwAwYXj1Nvv+bFZRYRS+6y1pS+a8HgOZ49Ab3AJCQ2kFQE/P8Arf4xliuBb+1jatmDJZFXT0ZS18o+iT6Wxetvr2pDFw6ecUfqaggJCdPrgtyChCXfq58Ag0dQxcyKM3uAUD/bLJOy9e4Xs8Qwy3cMxmoxg7UxlSaavwwzv8bFaRTGeD7PiBcv3aWLlr4RcmzG1q3YYKP5slGhGiEcnzlrQ7Cn9HtvD98sk7L17hezxDDLdDt2uftyczgSs/VaI+vFXwI2riwMDF8RudpD1bGtgq29nTRnLWgRO2qWyZrCIqVuHntrcnM1RV6C5zR4vhu1a9t/B3XrzC93h2MLpytYct+JLKZB2XfvjnrRIdXlOh/x+AbO7WRJqOVIaxwyoB2NiSCGSbvetIbvnUEYwdVlm2eFBfyWQhGhFiBRZ+RyrN+AadO7DjWfhDcwVAT8/wCt/j2YG4f8E6dvvm/8rOSXenlVnSWRUo9FIxfKvgG6qNwh+AAi12Wd6zZ04A4Ly/PAdARypLtam8F49FQsMNV969gP+8uqbPMgQufZE8RdmezDCuXg9EmnYwhZ/1Fr4Hr/A9nh2KZxZvAnSN/FKkM4q6yljetkQ6GyjwUln6dn9O4W9/C9/W+n/nAVrhr2vu1LI47vTKWIRkJj+pcEtbkmufWcbnb36lzzJkFEQiQjQqed/RkczQUF1BVUWE5j4WJrpr7mpeXLalr6J2G3fgUiqk4xn6eIXv8exATByhXco2Nh9GJquoiefXoXcr55Wy8K3Lf1jVwCn8u19dw7DKGHvvMoyPHTUNMds7UpnAwq+M6d9J5zpeW9UY/N3TJYELyZrSuoUWflsyTU08RkN1RVAcqLd84Z9zed+fnu3TOXqCj+F7wCt8j2eHoiOplVy5Pjtl5pFHIxJsc2POpdZ4t3HxYVXaO7AtXPpLNraWVTibWhMcu9cYqiqijK6rpC2ZoSOZoTOVoTpuFb6ZReDIt9nJ2u/rGvbpbNZJ2su38KvjUeqrKnbYpL1oRHwMfyfGK3yPZwdgS1uSL/5rbmDZl7O+bQzaVVaugipVeMVur7cKv5+T9l5d2chJP3+c659dVvKY5o5UEFIYXadnC2xuS9CRygSWfWWFnTboTNtzy972Ue5sVq89ECu4h+3JDDUVUW3h72AxfOupqIxFvIW/E+MVfg8QkWUiMk9E5orISwMtz1Dh7lfXcNLPH8srbOLJ53ePLOaOV1Zzw3PLgfw54YWkM7nV3kBbdfku/fAO31rGdUbh9/eqak+Z/IPv/ud1/v3SyqL9Sika21MMNwrfKv6mjlSBha9/u4l7Tc48/b6GIjJKEYtEiDgKXylFRypDTTxKfXX/Wfj9oXyVUl2+OxlH4ft5+DsvXuH3nBOUUjOVUrMGWpChwqU3v8KSjW28sa651+do6kjx4/veLBvbHkq4ZV4LyWQVsWhO4U8YXpW3v1TSVmEMv78XWVnlLMt72a2vBX/PXdmIUor2ZIZ0VgWKvjquBx6tnWlSGeXE8Luw8JN9G6iks4qInZZn8gF0TQAtk7XwlVLBrIKe4Cr5/phe+JuHF7PPt+8ru+CR/c54LJI31dCzc+EVvmdAaXcU1xtre6/wX1i6hT8+toQv3/Jqf4g16IgWvKllLfysIhrJfcDOHQ/2l4rhZwpj+P2r8MOs0JdXbOXdv3+a3z+6OFDagcI3Cn6LqRboTsvT8uUUl7sCYN9d+oqomHh31hb+0e20Jh6lvipGc0eaG59fwTE/eZT3/OHpYKC5rqmzy6RBd6DSHwr/z08sAcqXG85Z+FFv4e/EeIXfMxTwgIjMEZFPhB0gIp8QkZdE5KWNGzduZ/F2PN7emJtP/uDr63t9ntaE7vB39DXLSxERyfu/vYyFn85m81z6ExqqivaHfs4MBHJZ+v1rCRYuhpPJqiDb/Z5564K/beEfq+Dt1Lz6aj0QCSv9u8Eco+Xuo0vfDJiiEQnKFNsBVnVcx/CbO1PBnP+XVzTy8wcWcfvLqzj8Rw932Y7dUEki0/dBlZWtpcxUQfvMK2MRn7S3E+MVfs84Sil1MHAG8FkRObbwAKXUX5RSs5RSs8aMGbP9JdyBWLiuhTN/+xQAB0xq4Nklm3tt8bSazi5WaAoPUbqK4bsZ+tY1DlAbj5bMYk8VWPjfuXtB2YV2uksyneVfL65g+eZ89/fm1kQQC29LpINV8uoDl75R+E1amTdU6yS+XAw/dw/WNnUGSX59HahohQ/RSCRw6dvBio3hKwWrt3YEn7n5hRVBfsUK4+Zv6UyFWvud/WzhW1rK5BUEFn6FT9rbmdk5esd+Qim1xvzeANwBzB5YiXZsfvfo4uDvk/cZRyKdZeG6ll6dq8XEL13LdrCQymQ55w9P89Rbm3p9jsKYfXuiXJa+osKJ4bt/j6iNdzkP32bpA7SV+Z7ucstLK/nabfN4s+DZrm3qDNzQVRWRYOrgcKPYA4XfbBW+HgjYLH2rLLNZxdqmDqaNqgXCl9XtCRmlzOI5OUV5w7NamdeYaXkAqxs7eN8hk/j4MbsC8MoKXQugpTPNxpYE+1/5AH9+4u2i87sWfl+nELqUs/DzXfo+hr+z4hV+NxGRWhEZZv8GTgXmD5Q8K7e084/nVwzU1/cLjh5iz3HDgNxa6D3FdnZ9rLmyTVi2qY1XVjTyrTvn9focrQWdebKM5W1j+H/9yCx+fd5MYk48f0RNvGQM1ypQa0FD+eTA7rLecbe7bGxJBLH3eCwSLJrTYFz6Ncalv7apQOEXuPT/O28tqYzikKkjgL679LNZRUSEaCSX0W6td0ECDwTA5JE1fP2MffIGmo3tSdY2aev/rpCKiNvKwi83c8DN0s+q8t4Az9DFK/zuMw54SkReBV4A7lFK3bctvqixPcniDeGWbiKdIZXJctxPH+Wbd8wLFhfZEYk4neTE4TqxrK1MpnE5rEIslbC1ams7H73mhaIY/5rGjrLx8P5gmXFgwxiaAAAgAElEQVRlj6uv6uLI0hRareUUhY3hnzJjHGfPnBhY+JWxCNXxaMnPWmvTrcPf2+fhsrVEXsXW9mSQkJdKq2Bue0OBS98OGOxAoHBa3qJ1LUQEzj1kEtB3hZ82lfZiEV1a94+PLQn2TR5ZE+QSAIwZVkk0Ipy27y7BtsaOVNmQi2vh91Xhu+758jH8/Ge7/5UPhB6XymR5ZcXWHsmQzSq+8M9XeGZJ7z1Ynu2DV/jdRCn1tlLqQPOzr1LqB9vqu77339f58NUvhMnAgd99gA9f/XxQae3tja3bSoxtTtRJRLOdaGsvXch2SlKp+chX3DmfxxdtDOaCg7aIjrzqES6/o38cNU8v3sSyTcWL2tiYro0x94bCxWLKKvyCGL6dojesqoKKqJS08K1Lv8LJg2jrh9oIW9vCrcmtjiXckcrQ2JEiFhFqnYp6IgSx/1G18WA7QMIo9vXNnYwdVhVY3n0NQ2TMtDxdWjfLj+97E4AzDxjP9LF1QcgBCGoGjDGr++nrSpVNHnWfZbKPSXuPLdwQ/F02hm9cX/Eucly++K+5nPOHZ1jT2FH2OJemjhR3zV3DBX99vtuf8QwMXuEPQiaPqGFdc2fRPOiVWzroTGV57u3cohtLHQWTSGd4YemWokpq5ebnDiSu2qk1i7301qK01k2p+K21st1s92cWbwZgwZreTwd0+eDfnuf4nz1WtN1aroWZ9j2hsAhOMpPltw+/FToVK5PNL7xjXfr1VTFikUiXMXw35t8fFv6WEtPFNrclWbU1p/C3tCYZURtHzH0SkSBTf3xDFVUVhZX2tLzrmjsZV18ZLBjUV3d1VqmgtK4b7rYeGjuLAHJehxE1uUFAWyIdhCrCkvbc9zOZ7lsM6uLrcvW/yi3oY9tPrbOoUljFxf++thagR/UFBmIZZU/v8Ap/EDJ5ZA1KwZrG/NjnlhCrYbPpTJ9ZvIm9Lr+P9//5WX5wzxvB/nmrmtjvO/fzwIJ1PZJhc2uiXzr7clg3/DUXzgo66+/cvaBXi5/YTr6UO9daxG5J1CaTLzCytveWt6WcG9kmo/UlezwsZv/zBxfxn9eKY8S68E7u1bYKfOKIaiqiUrqWvmPhf/7E6UD/KPz1LeEx/C2tySA+35nMsKU9GVjxFqvwp46qCbZVRvNd+huaE4ytr6ImHi2qKtgb0qZwUSwqeQluNlnPVfjW2nePa09mgnyEMNxnWS4Xo6eUG+hY46G2MpefkfcutKfIZFWwIqF9Lt3BV8jccfAKfxBi49mFbrUwN6FVJo8tys35v/aZZby6UmcMP/iGnhP8wtLuL8WplOKoHz/Cqb98os8rjwHcO28t7/7900Vu6LZkmpmTh3Pi3uMCNy30TjF25dK31ozbybUEU/mEG55bzjdu731SXWFi2pV3L+DPj+vYrzv1rLck0hn2m1gPwPF75aZ7hnlv0gUWvrWYp42q1RZ+qUp7xtqsiEY456CJWuY+5je0J9N5XqgjdhsFwG6ja1nb1BnkT7Qk0jz4+vqiwZeN448Zlst/KKylv76lk13qqxAR6ipjPfJo3fDcchasacrblkvak7x7+f5DdY6AHYRAzsJ3Lef2ZDrwNIVNgUul+x7Db+lMMWf5Vg6cPByAcfWVNHeUvm77XnQ4VQjtoKSpPcWB33uAXz+0iFG1OjTRE4W/LRZZ8mwbvMIfhNh50IWJP2ELdthtrhsW4OzfPw0QJPVFo913Jy/d1EZnKsvqxo5+cXd/+qaXmbuysajsbSKdDSwKcdzdvalTbjv5tmSGn97/Jj+7f2FeMp7t8Nx72OpM5bvizvnc/MIKPvuPl3s1yFnndJAbmju59pll/OheHfu1HXF3lOcvH1zEUVc9UnQPEqksk0fUsOyqd3L8njmFHzY9L53J5sXwF63XCaAzJw/XVmuZefgRU2HOelx6m1NhmbN8a97MiW+8Y2+W/PAd7DuxgWWb20gV5BssWp+fk2KVa52jUG0cOpHK0pnS1vS4+srguO62n3QmyxV3zucsUwvCYqflVUQjpDIKEfj08bsHFQvdtmpj+B87aho/PGd/PjBrMu3JTJBfEJa85+ZQ9FbhX/fMMt77x2d4dWUjZ+y3C+MbqkOv++2NrbQm0oEHaoPjbbEGxOKNun08tmhj8Cx6sjhQoUt/+ea2kgs0eQYWr/AHIWG1wiF/iVOLfTHDEpWyWRXEeNf3YMTuzoUPW+Skt1gF+73/vM4l171IIp3NmwJm+9HerETW0plmdJ3u9H//6BJ+9+hiHnx9PU3teuEVO08/T+EbC991q97z2tog3l/Iqysb+cUDC4MBwTOLNwUdqetGnu9YjM2dqaAj7s51/frht1jd2MGKAhmSmWzQLuJdTJsrtPA/ddzunHvIJM46cIJWYiUs/M5UJoiTW4v1ijvn89+QsEF3efTNjVRVRPj1eTP5/InT2X9iA9GIMHVkTRC/H+5Mc7v0pOl5n7cW/jCnNkAkIsSjERLpbDCgHWvi68OqYkVTGEux0QxAC43wjBmExKMREqkMqYyiyrnnAJccvSufOHa34D5VxqJccNgUaitjdCQzQbghbAZIXgy/l0l7W5xEyOE1ceqrK0JDGSf+/HHO/8tzdBgr/Ng9coNFm5C5cot+DuMbqoIBSE+8UZ3OoOblFVs57qeP5YUVPYMHr/AHIbbTLXSVFSqMUbVx7p2/jmWb2kIXjWnuTLG5zdT4LjEXOozL79RZ68fvNYY7564pGni8vqaZ+3uYEwBa4WezimueXspDb2wgkcrkufKvv0jXMeqNwm/tTHPi3vmVDb/wz7kc+L0HAsUCBOVbrTxQfJ9fW9WY9397Mk0ineHyO+fzm0cWM291Eyu3tHPB357na7fpRWDczGu3AtvGlkRgSS3f3M5b67tXWGhLWxKlFKsbO/jR/95gY0simFLl3rMwC79w8Zx9xtfzs/cdSDwWIVZmPfTOdE7hu27rK+9+nbkrGznx54/lWYjl2NiS4N8vrWTZ5jamjKzh7JkT+fKpewXWsZvVftAU7ZY+eZ9xfOSIaXnnsTMGhjkWPuh7kEhngnZtE+oaqivyFtJJZbIlXfzrSgyCMyZpryIqQSKc9URZLj9zBt98xz5Fn62tjNKWzFnUYbMcXKs+1cukPXfq5MjaCoZVFXs2rFdr3uqmQJ6Lj96Vf3z8sLz9dtA0oiYevOs9CeW4Fv68VXqwe+0zy3pyOZ7thFf4g5BSFr6ruEB35AC/efgtmjvTHDipgYXfP50PHjYFgE2tySCpr1TnVsjKLe3BZ9578CSaOlIs3pDvZn3Hb57kkzfM6bHru6k9FUxxsvK5HVewHGpBwtOGls6yC4Mk0hmSmSxTRtZw0VG7Fu23U+VECmL4Ttx/rKOANrXmf9dF177IjG/fH0wdnLe6KVjo55E3NgQyWNY49/pH/3sj73yn/PIJ3uzGqoCb2xJ8+64FHHXVI/z5ibdpT+bWg8+bJ1/Swg9/tWPGTR1GZypLlTl3fo2EKr50y1ze3tiWF+LpSGZKtoHrn13GZbe+xiNvbihavAfyE9+Onj6ar5+xN98+c0bRcTYGXldVoPArIiTT2SB3wrr0x9VXBdtSmSz7fvt+9vvO/cHnmtpT3PDccpRSrG8Or2GhS+tql75NhKuqiIYeW0h1PIpSubBUMp0llcmycks7H/rb82xtS+bd/0QXru8P/e15vv/f14u2uzMtxg6ror4qVmThu0m+m9v0gDESEXYxgyM7CLAeLqXonYXvDJh9QZ/BjVf4g5DKEAt/7spG/vlivnv99xccDOiYYmtnimFVFVTGopy+ny4CsqUtV9hkbTdW8QJ4domeqnbVe/YPFHCpDPTNZZRwGLfOWZVXanRTayLPpW+zoF9ZuTWvw5n9g4c5/EcPlzxvR7CwSYxRIXPd75i7moqocNDk4XkK335HZzpDS2eai4/eVQ8KCpIjn3t7i17opcMu0JMKysRal7P7rNY6yZYPvbGBre3JPAv18YWlF1Wyrvgv/uvVoLqbJefSL18YxyqsMOIFmecuHY5LvxD7PdY7sL65k32+fR83laj26LaZ8Q3FBYfcanWVFVE+ddzuTHEy8d1rgfwYPmgXeiKdDZT2OJPUN76hKmjrVz+1NFBmVp6v3fYaV9w5n3mrm0rG+gOFH4sEnptCC78UtjqgW3ugPZnhR/e+wVOLN/Howg159z8sht+aSPODe16nNZHmqcWb+NtTS4s8K25YYMywSuqrKoqUrTtIvvG5FcF35TyImTwZMkoF19vdWgbrmjq5+YVcG3AHHf2R8OvpX7zCH4QUWvhKKd5tkvD23mVYcFxDTQW7j6nl5RVbeXVVUxDntJnOW9oStHamqarQHVe5qUKWxRtbiccivG/W5ECZuYlHrmXxob893+VCHG7H9HzITIFKpyO13/f7R5dw6c2vmO/WHUginS251r1VttUVUcbUVRbtX7iuheP2HMPkkTV5Ct8OFFo7dVb18OoKGqor8pZadbGd+Na2ZGCl23OUsvBBW05fOHkPLn/nPtTGo6zc2l4Uo7dUlCmMYhV9vIvCOKlMtuSaArFoJC9L3CVRQuEnMzmPgVX8c5bramyX3zmfR9/cUPQZN8QxaUSIhe8o/HLFYKwHwx0ggG4rzR0p1jV1EI9FAo/BuHodh25sT+VVobRW/1rzO5nOlrRibdKeK5d0s4aCXWnQVdDtyXTgYUtlsnlKPkzh//nxJfz1yaVB/X6AuSvyw0xJx0tQX6Vd+p2p/HOXqnBowzV2JkHKGRTZhMLuuvQvu/VVnnwrv5iVpT+KNnn6F6/wByGVBet9uy+OLR9qqa2MsXRTG5msChS+O7UmnVXsPqYu+L8r3t7YxrRRNUQjuaIn7lQ3t7DNm+taWL65uLKciztYCFPYbjzaTYx60lTEc5fPnb86f/qUxVoqVRUR9jVT11w6khmG18RpqK7IU/i5zGUtV311BSNq4nkxYBerNDa0JHh5uVkoxWRAWwVXVxkLFM3sXUcGn504vJpLjtmN3cfW8a8XV3LsTx/lrrmr886vlCpbxMR6Q9z72J5Mo5Tib0++HQwiyln4saiQKjFI60xlQy3ZpvZkkBNg4+FuhcePXfti0WdcC3H62Lqi/cOdQjWux6IQWzDm4Ckj8raPrI3zwOvr+euTS5kysiZQyNabsLapM0+hW4WbDda3z+Ql9/1v3lque2YZ2axCKYxL37mH3TRWJ5gptW87UxFvem4FLxuFvbapM8+lH5bNbpNG3UFkYUEp+7kZ4+vZf1JDMCByrfxSA3w7sO4osPDdwj1dufSveWop075+T56yLzxHd5MnPdsPr/AHISISJCVBzso9eZ9xXHz0rpw9cwJ//9ihQH5yle3HR9Tql99m3+5mFH6pRUxclm5qZdfRetWxwo4hm1Vc+/QyAD4wazJQPI2qkK6Kcrgufdfaty5wV7m9sTY84c0qyaqKKHvvUp9XpAW0tVLjrGNuO/3CZL366phO+nIsI7fTtdbP/QvWsa65M5gPv7ktGQweRtbGA5knDc9ZtrZDnjKyJujw752Xn/jYmcqilJ6jbrH5GJBTjIdO0wOJqaNqaE9kWNvUyffveYNLrn8RpZQpHBP+aleEVNrb2pbk//33dV5f2xxq4a9vSQSK1yqCclXdID87fd8JDUX73az7cgr/rx+ZxU/ee0Bekh/klyn+kHOPdjEKf11zB+uaOwOlPW91EwvXtQQeqaaOFK2OjJ+56WW+c/eCIARQGYvmeVvec/DEMlebw3ozXG+2uyrkuqbOvPsfZuHbkJBbh6MwrJbOZJk6qob/feEYGqorgvvpPpfCGQJfOmVPc235BoUdPDTnhbvKv7duLs7Xz9ibS0/aA8h36ft4/uDDK/xBSlVFlEQqy9JNbdz5irYEz9hvF0SEX593ECfsNRbIn9Nr3cyVsSjDqmKs2qothd3HaAViY5ufvOElHnpdF+R5YtFGVpuOJZNVrNzSwTSr8Ass/LtfXcPPH1wEEBSBWRpSO96lq4Vp3M7etfatB9Uty7piS/h3WfmqKiJEI8Ljl52QNxBqT2SoiWtlrlTpMrz1VRWMqKnIs4zCpjpZa/6UGeO0jK1JEuksFVFhmJM8NcFR+DYG3eC4prMFMU57rz565DROMIOJXR3lbz87bXQty656J4fvOoq2ZDrwgixa3xp0+HWV4bH4WFTIqpylC/CnJ5Zw9VNL2dKWzFP4p5rry2RVkA1vFX5X1ezakxn2HFfHM18/kckji2PzNfHc95RT+DMm1PP+QycXbbdhqw8fPpWPHjkt2G4TBNc2ddLSmQ6y979/zxuc9qsn8hR+mBVrBzaVsUig8E/ce2zJAVQh4xuqAu9KWFhldWMHyUwWEd1e7QBjc2uCr/z7VZqc/JBlm3Jhn8LBaSqj8gYkNv+lnNK2SllEqKqIFCXtWXd8LCJdFi+qdp7fuw6cwAETG4q+v6UPhaY82wav8Acp1sI/67dP8cP/6dG020laNjsWsDtNaFRtnJUmq3/aqFoqYxGWbGxlfXOC+xes55LrX6ItkeYj17zAuX98BtAWRTKTZVezrnhNgYXvutRH11XSUF3BHx5bzPv+9EzJBJ2uLIV8JZ/rIG2HYxX+7mNqi2YLWGxnWDhX2pLMZAMLH3KZ+p2pTF5lt4bqCobXxIPYZzqTLTk7oKoiwh5jc0v6JlK6poBbcc21Sq0F5uZgFK1xHyQfRoM54nbKGhSXAK6pjNKeyLDUCas8YSou2lhyIVZJuHPxnzOJmpDvMfrLR2Yx78pTcfWWLcRTaL1lsorOVIZP3ziHJRtbaU+mGVVbmTfocXGfVVcLuoRhr290XWVeuxldFycei7BkQxttiXTe7AuAhWZaZFNHKtTlbL1FlRWRYCBSmDBYjlg0EoQV3LDFOQdN5IS9xvDkW5v47SOLqYhEiEcjgYV/19w13DpnFV/41ytB23en0hatlliQp2Hb2gYnb8H9jOs1Av2cc0l7+t21BaJG1Ma7HKjbdjRtVA0ThlcH3jnbLkbVxktO//QMHN1vyZ7tSlVFlM5U/hzi6hCFb+N9j37l+DxrcGRtPHC3D6uKMWNCPVc/tZQHjWUP8JJJvLKxfWutTzUK31p7VhG1Ox1IVTzKhOHVvLG2mReXbWXh+hb23qU4fm6Va6XJeK6uiOZ1RJUlrDubgLS5LUlFVDh8t1Hc9PwK3t7YGoQogmOtS9+5P4U5VmEKvyOVYfrYumBQUV9dwfCaCpraU6QyWY7/6WNFiu3gKcN5eUUj08fWMcIkim1tT5FIZ6iqiOQpB3fGgJ1WdsFhUzl2zzFcfuf8IivZ/l9fFeO779qPm19YwczJudh1ocKvq4zRlkznDfo+b5Id60sofKsk0hlFZUxb+m5YprBi47CqCmZMqGf+au09ajTrDxRagJ2pDG+ua+be+etYuqmNaEQY3xAuA+RP+ytn4ZfiI0dM5fU1zZxQUHshFo0we9pInlmyidZEuii8Y2nuSIVWEbQWajwaCe5VbQ8UPmi3/qqtHQyvqQjCO9XxKKsbne8TXUDJWte2rTxmZnDEY5G8cFaYS9+9b3aqnTtIaEukqYgK11x4aFHbqaqIBp6xZIFLf2RNnI0tCRLpTF7IzaKUoqk9xYVHTuOrp+8F5EJzLZ1ppoys4YmvntD1jfJsd7yFP0hxOwtLdUh89eR9tNt1aoHbdFRdZdAp18RjzDZx3xXOKlhuDfG2RDpIwLMDB6uMbTKY666rroiy17ic4n3IGUi4WOV67cdmc82Fs3jyayfw5VP2DBRLufnNm1oTtHSmqKuMBTHUsBCCLWPqWo2FzlTr0ndl6kxlgo4StJIcXh2nJZHm0ptfYXVjR1Gs+uyZE7nstL34yXsPDCy4pvYkncbCz1P4tY6FX6m/OxoRpo6qpTYeK3IpW2XaUB3nkKkj+Nn7DsxLvhtZm69Aa+Ixsiq8xsKwqnAlZV3T1vpa3dhBRypXpz8sn2/2tFHB3xuaE/zqoUXBSoOWzlQmGLC8ua6FDS0JauLdU5S9sfDHN1Rz3UWzOWDS8KJ9+01s4O2NbXnVFwsp5dK3NRMqK6JBnl6p8EgpJo3Q7+JEx7tRUxHlu+/alwMnadd3Mq0rJ1oLv1Ch7za6Ni/htThpL9+lP6qukmhE8ipqticzVFdEOWaPMUV5FFUVUTrNd9tZG1bx28FBKe9cZypLMpNlXH1V8IxtX9HcmerVAM6zffBPZpAydVRtsA64JawD/d0FB/HCt07Ks5iAvFXHaiujfOHkPbjyrBmcYeboAyxYnSuisrqxg0XrW6mNR4MiJtZV+q+XVqKUylN+NfEoZ8/MJTI9+3a+ArBY5TptdA0n7j2O0XWVfP6kPYJzF66O5rK+uZPOlPYKjDXzrNc0dRZNAwtc+k7S328vOCjvmNrKaLDQSVNHinQmSyqj8hRCfXUsSHi8d354JcGG6go+e8J0ZkyoDwYQc1c2cdvLq4jHInnWoJtYVpj9XltZrPBtwSG3KI2LO4Cw1wT62U0eWZ3nLSml8O1A61cP61wMO33swiN35V0HTuDyM4urxx0wKacsXl3VyK8eeqtolbfOdDavwNDGlgRTQmL3YfS3gpg8sppkJktTR6qoYI+lqSMVOvXsVw/p+1IZiwTPp6cWvlWYE4bnBpM18Sj7jK/nqvceEGyLOwq/MLm1MEkxbHlk16UfjQhj6iqLLPxSsldEpUjRF8rflkhz5d0LeHxRft0IOzB126l16euBr1crgxX/ZAYp00bV5FnjEO7Sr3KUocvIPIUfoyYe48KjduWPHzqEK8/SFc2eX7olUEqn/vIJbnhuOftNbMiLiZ5nEqZWbe3Is/BjkQjjnQ7NnT7nYhV+Q8E8attXlVuaduG6lqC+u3V5XnHnfD527Yt51eo6gml5uftz4t7j+M35OaVfXRGl1gyY2pLpwLpxFX51wb38xhl7F8lUmF1eVxnjtpdXARhvRE4G99oK53HXVUaL5in//jGdzV1K4Y8ojOGb61nd2JG3HjtApMS8cTuf/u9mtoV9PtPH1vGb8w8KbUtuQRw3DHHyPmPzqrZtLvBI7VoQNy5FYdvoK5NH5OStK+FlsDH8woGRrRVRGYsEHrKexPAh57FwLXAbbnLvSUVUSKazLFzXwkInrCKS33bisUjRgCBV4NIHGNdQlTcTpz2VCe0z7Dmtoi+cGmgHvVvbk1z7zDI+es0LANz+8io+dcOc0Hfadf17C3/w4p/MIKVw3jGEK/xSjHIUWWGHaqeIbWpN8N6DJ+W9oIftNirv2PeZ6XcL17XkFdUYPSyeV+RmbVNn6BS8po4UFVEpCkeIcbqHVcYDbfnPWb41UPg18Vieley6sTtDFD7kqp6BHvTY60xlsoGsrptcRJg2OqcsLjlmt6I55IXJcO69FZHAoopIcbGYPNmMhW+THdOZbBAnt2usd0W9UVZLN7XRUF0RzB7YbUwt+04ozqeA/M64pTMVdN71JSxhIEjidC19gP0nDufKd+nBY2cqUxSCOmL3/LZUilJu997iPhPXwt3TCUE1tqdoTaRLepgqY9FeW/iH7abDZ0fuPjrYZttiVUWUuz93FLd9+ohA6Z72qyfyqtVVV+SSP6eMrGF8Q1VRfYZ0gUsfYJf6Sl5ctiWYddOWSAeD3EIqnITBRMHUwJFm8FgYPvvSLa9y34J1bDaenOF5Cj98to1ncOGfzCDlkGnFCr8nlpDbkRUmcLn/7z+pgds/fSQfmDWZ/Sc2BBa9xbol17foaU7nHTqZt35wBmOHVRVZlVtCKnu1mJK/hRauHbwUnmPssEp2qa9iyqga1jZ15hWDcY9d7cxRtrHO2oJYq2u9VcejgeWVTGeDRL1Cq3nqyFrzXRVEI8Jdnz2KJ50EpH3GD8s7foQzYFAqZw1mVc7CC4tR11XGSGdzpUzdFfoK3f8Pfek4nvpacRKUzYBXirzktOs+NrvkNDJ34LWuqTPIzC7XtkbUxnn0K8dzyyePCLZddtpefOLY3fLKQG9uTeadp1SGfiH9bRG6s1nqKmNBGOOcg3JFq2wMv7qEQqysiART+sIqBZbjmD3G8OK3Tua0fcc5MuW+54BJwzlk6kji0Uho4Z2aeCxoR3uOG6YTXUMs/MIEy3H1VXSmsvzfP3Xi5rqmzqJZCpZ4NMJTizfx6Rvn8EJBBUzrXfjZAwuDbWubcu/bSuN5rC+h8EsNMjwDj38yg5TC1cGgZ67FQpegS4PjMt57l3qmj63jx+ceQBg2brypJUlHUrsIrSKLRIQrzpzBxpYEf3p8Cc0dqbxEJdDZ3GFyX3/RbG57eVWRS//Zb5wEwKdunMPKLe3UVcYCy31YVYy1Js/wW3fM55CpI9h7l3rak2liBaVQIX/Bldp4zsJPOmV6C63L6niUv194aFCxr7YyRm1ljD988GBWbGkvsvDzByyqaHB1zYWzgul7LjVO2eKqimiwwM8dnzmyaHAUVqkO8hXRvhMaOGaP0Tz51qbQ2vW568vdo/XNiZyF38Vg0rqir79oNtGIcNR0bb3aRMlEKsPG1gTTRtcyvr6K982aVPJc2xrXE1ZbGeP6iw4jHoswcXg1P77vTfYZX8+KzW20JTPMnjYiWAjJpTIW4fMn7sGBk4ZzzB5jivZ3RWEMPsw7F3fyBFwikhuYjaqNs6GlsyjOnsxkiwZ15x06heufXc6Ly7byrTvmsWh9C4fvFu5lsYPksFwVOwi2hbsgtwoe5BJ/82P4uevrb4+Np//wFv4gpbDTL3SndkW52LirKKaVmLZksXXKN7R00m4q1rlcfPSuHG06/7BiLG0lFP5+Exv4zln7Fl1nNCJEI8Louko2tSboSGWCzq/wPKf/6kmeXryJtkSGmni06Fyu8q1xBiqpTGmFD3DC3mOLYtnv2H88nzpu96Jj3UGGUnD6/rtw9PTRQYW8E/ceF1p4xrpsbYdvlzEuVBTlcK3pXUfX8qcPHbHLKukAACAASURBVMJDXzqubJEYN+zRmtAu/dp4tGwNf5dj9xwTKHt9PpOslc6wuTXJmLo4f/rwIZy0z7hSpwi49wvHcOdnj+rW9/YE15qurYxyxO6jOGTqCHZpqGLZVe/klBnjgvyJSSNqmBAyQKqM6Xn4J8/o+jq6J1Oxwq+MRYOSuy7JTJYT9taFtQ6ZNiJvvr4llckWDXBnTKjn/NnaQ3fT8yvIqtLeicJ59p88brfg77C+w00gtgq/oYSFP3pY90JSnu2Pt/B3AF678tSSyUelGFtfWnGMc5RZdyqIjamrZNXWDrIqfKaAdZ2HldJs6QxX+F1/Z5wtbUmGVVUESsomnF122l4Mr6ngW3fM54WlW2hPhmcju9+rFb4eECTT2aDe/ei6OL94/4El5613haso//DBg6mvquDGSw7r8nNWttZA4esQQ2EmfjncAc7YYZXUVsZKegMsrku/uTNNU0eqS+u+HLay3aL1raxr7uzRwNQu79zfuNcY1vZcRVVXFeP6iw/j5hdWcOi0EXzttnl6FkcXi0L1WKYQhV/qvidSWWZOHs6cy09mRE2cO19ZXaTwEyWy4asr8q/32D3DvRPuole7j6nlS6fsyZ8f1ytZul6ro6aP4unFm/PWBli5pZ1oRPLurTtjoCdt2LN98Qp/EHPotBG8vqa5V8po7LAqrnrP/qGdSiQiHDJ1RMnErkKmj63jabOYTVgtgJzCD7Hwk+nQzO+uGD2skqzS1f9sAqOdRjWuvopzD5nEbx9eHMwjD7OgXJd+TTyGiBCPRUhksjy7ZDMTh1fTUF3Bew7uvfv58jP3YfSwOF86Za+yXpVCci59o/Bbk9TEoz1KzHQZW9+9e+yev9Uo/L5kye/SUMX+Exv49UNv0ZHKsO/EnnmitgVuDkTYQNC93obqCqaPreOKM3Xy4cFTR/DPF1ay17jiMExfCBso28JNh+82kguPnMZdc9dw7/x1KFMBwCbexp0ZA5ZEOlzhrzTltL939r7sO6GePUtchzugGVdflZdl7ybS2hDd0k25WQTLt7TTUJ2fl+P+XSoR1zPweIU/iPnnJ47ocvnZcpw3e0rJfbd9+shun+eAScODWF+YYrVx7bD1xdsSGWpH97yZWVd7Ip1L2rNW+a4mk37iiGpWb+2gqiIS2rHnTYsy56iMRnhtZRPPvr2Z9xw8sdvLnpZi0ogavv/u/Xv8uZyFr13LW9qSPRowWPYaN4yF61u67UWpznPpp2nuo4UP8N6DJ3Llf14H4KDJxYVwtjfuM+3Kwh9ecO1jh1UFNef7k7D3xk6dPG7PsZy+33h2aajm3vnriooJhbn0E+lMXtzccumJe7ClLck5B00sWV4Z8i389xYMeN17NnG4ftc2NCcYVRtnc1uSxvZUUalel762J8+2wyv8QYyNZw807nSmcNekbkZNzqIz2axiU1ui9y59J5ZdFWSCa+Vol/udOLyaV1ZuZXxDdWiH6mKVQDwWCYoE9cbz0F+4Mfx5q5p4YtHG4Lp6wi2fOiJ02eFSFCr8po5UUBmut8xwqrhN6SInZHsTNhB0CwJtL+UU5hmz+Rv2/dlvQj2fPHY3Lj5617zj3AI9oEvbJktY+PtPaujWYN5a+DddclheTob9PstEkwOwuS3J+IaqIPRU7r6Vm+LpGVj8k/F0iauIwjouuzqfq3g+fdMc7l+wnoqo9Lg0KeQn09lO+7qLZvPQG+uDkrYTR1Tzv3lrqY3HimYHWB760rEsXOfWis91Zl0NErYldupSayLNWb97CoDj9+q5smyoruiRS95db+D2l1cRj0ZomNA3pbeHkzfQ21yIbUXYYHMvZwGjUkWO+puwtmbzbGzbjUUjfOMdxZUO42YdCks6q8iqvs13txb+uJBQkJsMaAtzNXWk2H9iA6AXHyp33wZbG/Dk8Arf0yUTnUzfUvXRx9RVsslZWe7+Bbq2fiqjqKvseQcw0kkcsgOGg6aM4CCnINGkEdWks4o317WwX4nY8fSxw5juTItzPfhha5FvL2zeg7sC4cQezvfuDe6AzZbC7avSG1Eb53MnTA+m+A0Gbvv0EaQzquQc//qqGM2d6X6v8leKsPfm8yfuwYzxDRxXIrHOUulUxYNcoZywhW26S9qcz/XYnXfoZP754kpEhOsums2Yusq8MF19dUyHFzLZsvfNu/QHL17he7okzyouY63f89pafnZuJqRufM87JrdoTqlKZ67nobuJQq6lNCukuNH2or66AhFdwRDgwiOn8fFjdu3iU32nIhrhPQdNZPSwSv7yhM7K7mpqZnf4yml79fkc/ckhU0eW3X/7Z47ioTfWl41z9yeF74TeFuWdB4zv8rOFMXy7WFRlyDm7iw3huwPAH56zP99/934AwSDk1ZW5aYM18Ri1lVGS7dmi3AeXUus4eAYePw/f0yNKVu4yltTjizYG1dssvekA3MWASuUAzHQSxEZ3cyqQzQP4zlkzOH6vsT2Wq7+IRoT6qopgTvN7D5603ZTPLz4wMy9O3JvcgR2d6WPrQusqbCv6khxaGMPPWfi9777tO+kORCIRKZqm61ryNfEoUTM1tqGmeIC9+xidyBcW9vMMDrzC9/SIUoluV194KABzVzaysSCJrKe1yAspVaqzqiLKMXuYim/dtHaswp/cx0S1/mBETQVrzZoANnFrezGuviooOLNfD4s6ebrPT849gFP7WLwnvg1c+v/4+OF8+8wZXS5h7Jaero5Hgzydg6cUz8b496eO5PaQSpGewYP3vfQAETkd+DUQBf6mlLpqgEXa7pSKiU4cXs3R00dz99zVweIhdhpPXxV+uTDCrz4wkyv/8zondqOyG+RcmYMhzjiiNh7U0B+IRKfbP3MUyXTWJ1ltQ94/azLvnzW56wPLEI9GyWQVmawiGhESZiGdvlj4u46uZdejuw4h1ReUp7aELYw0sjbeq6mlnu2Ht/C7iYhEgd8DZwAzgPNFZMbASrX9uOLMGUUL6xRyxv67sKapk388v4KIwDkHTQTC1wXoCeWm9Y2qq+S35x9UMku/EFtnfiAz9C1uxcOBiHvu0lA16KbReYqxg+yt7UnaEmkazfTXvsTwu4trrY+ojfPvTx3BrZ86ok/eBc/A4S387jMbWKyUehtARP4JnA28PqBSbScK5waHcaApGPLg6+s5ce+xHL7bKP721NJez3f/0il78uRbG0tWC+sNnzlhOlfcOb9s6eHtxZHTR3HfgnXEo5FulTj27JxUG8U+6/sPFWzfvt33eYdO7vaaC57Bidj1uD3lEZFzgdOVUpeY/z8MHKaU+lzBcZ8APgEwZcqUQ5YvX77dZR0olFLcOmcVqYzi3QdNoLoiytJNbew2yJLCOlOZvEVkBor2ZJpbXlzJtNG1A5pA6BncbG1Lcscrq0lnc3H8mniM98+a3O9LC4fx6spGaiujedNbtzUiMkcpNWu7feFOglf43URE3gecVqDwZyulPl/qM7NmzVIvvfTS9hLR4/F4hgRe4W8bvH+m+6wC3CD2JGDNAMni8Xg8Hk+P8Aq/+7wI7CEiu4pIHDgPuHuAZfJ4PB6Pp1t4l34PEJF3AL9CT8u7Rin1gy6O3wj0Nog/GtjUy89uS7xcPWOwygWDVzYvV88YinJNVUqVrzns6TFe4Q9SROSlwRjD8nL1jMEqFwxe2bxcPcPL5eku3qXv8Xg8Hs9OgFf4Ho/H4/HsBHiFP3j5y0ALUAIvV88YrHLB4JXNy9UzvFyebuFj+B7PEENEvgnsZmtG9Nex3TiXAvZQSi3u67k8Hk//4xW+xzPIEZELgS8DuwPNwB3AN5RSjeU+t73xCt/jGdx4l77HM4gRkS8DPwYuAxqAw4GpwIOmHkTh8X59DI/HE4pX+IMQETldRBaKyGIR+fp2/u5rRGSDiMx3to0UkQdF5C3ze4TZLiLyGyPnayJy8DaUa7KIPCoib4jIAhH5wmCQTUSqROQFEXnVyPVds31XEXneyPUvq5xFpNL8v9jsn1bm3PXAd4HPK6XuU0qllFLLgPejlf6HRORKEblVRG4UkWbgQrPtRnOOqIgsE5F2EdksIj8XkYSIrDJy/D/z2UoR+Y+IKBFZIiKrRWSTiHzLkWe2iDwrIo0islZEfhc26OjGPVsmIvNEZK6IvGS2DYY2NtzcyzdNOztioOUSkb3MfbI/zSLyfwMtl/muL5o2P19EbjbvQp/bvWfb4RX+IEMGfhnea4HTC7Z9HXhYKbUH8LD5H7SMe5ifTwB/3IZypYEvK6X2QVu5nzX3ZaBlSwAnKqUOBGYCp4vI4Wir/JdGrq3Axeb4i4GtSqnpwC/NcaU4EqgCbnc3KqVagXuBU8yms4FbgeHATQXn+CEwEXgFGA+ciS4cdaGR6yBHrmbz9zLgWeAk4Nsiso/ZngG+iC6ocoTZ/5ky8pfjBKXUTGee9kA/R4BfA/cppfYGDgTeGGi5lFILzX2aCRwCtKNDOgMql4hMBC4FZiml9kO3qfPon3bv2VYopfzPIPpBd6T3O/9/Ax2v3Z4yTAPmO/8vBMabv8cDC83ffwbODztuO8h4F1rhDRrZgBrgZeAwdIWxWOEzBe4HjjB/x8xxUuJ8HwLWldh3FfAgcCXwRMG+K4Eb0es9LEErhP8CYr4vCZxs5Fpsjr0fOAdQaO/BJnP8C8B5JWT4P+AO538FTO/GfVoGjC7YNqDPEagHlhY+i4GWq0CWU4GnB4Nc6EHkSmCkacf/BU7rj3bvf7bdj7fwBx/2RbKsMtsGknFKqbUA5rddy3VAZDXuwIOA5weDbMZtPhfYgFbCS4BGpVQ65LsDucz+JmBUiVNvAkZLeFx+PLmypStD9oMuA/2KkQvzPY3AZkeuekeutebv1Y5c7UCduc49ReS/IrLOhA9+iLb2e4oCHhCROaKXk4aBf467ARuBv4vIKyLyNxGpHQRyuZwH3Gz+HlC5lFKrgZ8BK9DtpgmYQ/+0e882wiv8wYeEbBusUym2u6wiUgfcBvyfUqq53KEh27aJbEqpjNIu10nAbGCfsMN6Idez6JDBe9yNRhGdgbbcS31+IlrRzwNsTXIxP25H2xO5/gi8ic7Erwe+WeJzXXGUUupg9DV8VkSOLXPs9nqOMeBg4I9KqYOANnJu8oGUS3+ZjoW/C/h3V4eGbOt3uUzOwNnArsAEwLbJUt+9I/VrQxav8Acfg3EZ3vUiMh7A/LYW43aVVUQq0Mr+JqWUjWsPCtkAlJ4m9xg6x2C4Y5m73x3IZfY3AFtKnK8JnbT3W9GJnBXGu/Fvc54byogzBq0gPgGciHbh/wbtGbCd7ySgxZFrvPk7WkKuYeg4f6uI7A18usz3l0Qptcb83oCOR89m4J/jKmCVUup58/+t6AHAQMtlOQN4WSm13vw/0HKdDCxVSm1USqXQeSZH0g/t3rPt8Ap/8DEYl+G9G/io+fuj6Pi53f4Rkxl8ONBk3Yz9jYgIcDXwhlLqF4NFNhEZIyLDzd/V6I7wDeBR4NwScll5zwUeUSawGYZS6idoS/pnaGX7PNo1epJSKlFGtLlKqUlKqYnAT9DW1KnA22glnzByLHTksvK+t4RcXwEuMJ//K/CvMt8fiojUisgw+7eRaT4D/ByV+v/tnXl8FOX9xz/PnrkTQkK4CZecconIIYrggbdVW0VbtbXSw1atR4v1tlqvWu1hbb2P/jypWguKyqEoyBEQEBQkQIBwJRDIvffz+2PmmX3mmZk9kk12kjzv1yuv7M7Ozj47M/t8n+9NDwLYSwgZpm6aBeCbdI+LYw6i5nz2+ekc1x4AkwkhWepvk52vlNz3kjYi3UEE8s/4B+AcAN9B8QXf0c6f/ToUn1wQyqr8Wigm4CUAtqv/C9V9CZSMgh1QTMcT23BcJ0MRWpsAbFD/zkn32ACMgeIn3wRFcN2tbh8EJeCtHIpG7lW3Z6jPy9XXB7XDNZ0BYIH6eLR6HivSMS71vGxU/7aw+zvd11H9rHEAytRr+R6AbjYZVxaUuIt8bpsdxnUfFBfPZijWJq+d7nv5Z/yTlfYkkk4OIeR8KEKBAHgcShbBBCp//BJJl0Ka9CWSzs+FUHyp+6HkZ18uhb1E0vWQGr5EIpFIJF0AqeFLJBKJRNIFkI022pCioiJaWlqa7mFIJBJJh2LdunWHKaXF8feUJIMU+FAaxkCpMV5FlbrQ4usESp3tc6BUHruGUro+3nFLS0tRVlaW6uFKJBJJp4YQsjvdY+iMSJO+wkswNozhac8GHhKJRCKRpBwp8AFQSpcjdtWnCwG8QhVWQakm1SvG/imlqs6HmsZAe32cRCKRSDohUuAnRsINKQghcwkhZYSQsurq6pR8+KQ/LsGEP3ySkmNJJBKJpGsiBX5iJNz4gVL6DKV0IqV0YnGxjDmRSCQSiT2QAj8x7NjQRiKRSBLijTV7cKjOl+5hSNKMFPiJ0d6NMiQSiSQlVNX7MO+dr3Hty2vTPRRJmpFpeQAIIa9DaTBSRAipBHAPADcAUEr/CeADKCl55VDS8n6cnpFKJBJJcviDEQDA0cZgmkciSTdS4AOglM6J8zoFcH07DUcikUhSRkQtn+6Q9twuj7wFJBKJpBMTjigC30nMYo8lXQkp8CUSiaSTQSnFvz7bgap6HwJhxaTvkAK/yyMFfgdCdjaUSCSJsGV/HR76cCtufnOj5sN3OqTA7+pIgd+BYCt1iUQiiYUvGAYANAZC2rwhBb5ECnybw2v1vqAU+BKJJD4h1W/vdjgQCEmTvkRBCnybEwxT7rEU+BKJJD4hdd5wOogm8KWGL5EC3+aEIlEhH5E+fEknIxiOoMEfSvcwOh1Bdd5wOQn8IcW875ACv8sjBb7N4TX8iFTwbUuDP4RIRC7IkuXnr67D6Hs+SvcwOh2+gCLkXQ4CP9Pwpbzv8kiBb3NCnBk/LDV8W9IcCGP0PR/h4UVb0z2UDseSrVUAIBdLKaZZDdpzOR3SpC/RkALf5jz0YVSIdKVJcVPlsQ6ThshM0u+sr0zzSDou9dKsn1KaVA3f7SQyD1+iIQW+zZm/LipEuooP/38b9+OCv6/Agk0doz8RNe+ULEmCumZZ5z2V7D3aBABwOaSGL4kiBX4HojMr+Le+vREz//QpAGBtRQ0AoLren8YRJQ6LiO4i67E2oVYK/JSyYKOyWHY5iXZ/SgVfIgV+ByLciSX+/HWV2Hm4EQC0vt3dst3pHFLCyHTJ1vNF+eF0D6FTweaK7YcaUNMUSPNoJHZBCnyb0zs/Q3vcUXzaraXep/hzg6GO8X35TIqWsGV/La7/v/VdcuHgVkPHV+88kuaRdC6Y3/7rfbV4+tMdAAACqeJ3daTAtzmj++Rrj7tClL4/FNaCi3xq/rDdsRLUL6+swMMfxo/c/82bG7Dw6wPYfqgh1UOzPZluJ4BoZThJamB+ex4ZayKRAt/m8MKkK+Th1zYHtQIhrB643bES+Pe8vwX//GxHXFdMlscFAF2uAA2lVPvOflk2OqWYCfxQKy1Rko6PFPg2h2+Ys2rnEdz7/pY0jqbtqW0KagVCmgMdQwjEM8VvPVgX8/UcryLwP99ejap6JX7h3ve34Kll5akZoE3xBSNaIKo/Re6Mj7YcxOJvDqXkWB0VSqlpo63OHAMkSQwp8G0O78e+f8E3eGllRYfRfFtCbXMQTBHpKCb9gHqNxOm0e7YHALB2V03M92d6FLP235aWY8ZjnwIAXlpZgcc+2tap/fq8RcOfonv6Z6+uw09fKUvJsToqVl01pdtEIgW+zTHTfI524qjb2uYgAqqg7ygLGyuh3D1HEfgVR5pivp8PpWIFUxhmptnOQiMn8Nur9XNtU7DTpwBa3TNSw5dIgW9zgqEIPE79Zapp7HwCP1vVcmubg1rUe0dpB2wl8FkN83gLl1jCrjP7XZmGn+t1pdyHb5XRMvb+jzH2vo9T+ll2w0rgd2ZrkSQxpMC3OYFwBF63/jIdbex8GkqW6sdWBL4yMTUH7BXEdqC22XS7VVoeE2LlVQ14cvF3oJSCUmpYAIhaPa+JBYVIzZU7DnfYFDZfMIx7/rsZ+48p55EJ/MIcT9Ia/utr9qB03kJL4dYY6BjWobYglSb95d9VW973ko6HFPg2JxiOIENNXWIcaewYFeiSgaVn/XvVbmyqrAUAHKqzz/f838b9mPLQUlNhK2pO9/x3M1bvPKK1JS3bfRRPLt6OyqPNeH3NXgy/axEu/scKHGlQvl+TsLDhn/PCf29NE654djUue2ZVyr5Xe/L8F7vw8pe78fqaPQCAZ5fvBAB0y/Ik7bpgueUVRxpNX993NDVCilKKT745pGtiZXeszmVL3ENXvbAG5//ti9YOSWITpMC3OYFQBBmCht9RfNvJwPoE7KiOTuC7TSbz7w7Vp9QXWV5Vn5BPd+UOpRJcebUxV54X+KFwBC9/uRuXPbNKM+nz+737ldIbYf2eY7jj3c0IR6hhIq7khBV/7OmPLtMeX/vS2rhjZnz49QEMu/PDtN833xxQshVK8pRiUhsrjwEABhVna4ujRCktygYAPMp1KOTP43eH6ls1VsaSb6tw3Stl2gKjI2Al2JM5xyvKD+Nb9XodbgigdN5C/GedbA7V0ZEC3+YEwxFkuPQafqAT+nXNfNX7a326SO7th+px5hPL8bel21PymZRSnP7n5bjyOUVj3nesGfU+c+HfrJqIxWsBRIUyAXRCXhT4jf6wrt7+oi0HcctbG9Dg02v4Z//lc+2xlQ+ftZVNhEc/2gZ/KIJ9x9JrmmXnkJ2vPgWZOPW4YvTOz0QgFEmqkmRp9ywA+gViM2fG33XYXPNPhiMNftzwxlcAlHvRzpRX1WPSg4txqM5nuO8YsTT8+esqtZRQALjyudW6+xBAyn53kvQhBb7N8YeMJv3OGLkdEnzVhWpK25JvoznVbNItqzga93jBcESryW8F8/Nu3qdoMtMeXorZT35uui/rL+52GX8ybAFGoRfyoiWiwR8ypO69t2F/TGEinheeRNslsyZp6S7NzKLy2TlqDISR43XB63IgQpPzMbN9mVtEOV504dQcx5qRyLl7dNE2Lb7CZcNOc7sON2r32AsrKlBV78fHWw5a+vCtFgK1TUHc+vZGXPX8GgDWFsTcjI7R20JijRT4NoVSiic++Q71vhCyvXqBn2y07SffHNJM0nYhHKG4673N2KGayMXAt/6FWcj1uvDCigptW6ICDgDu+98WnPTHJbrUL5HDJt34rLTgZjUAzyxfnPl3IyYBeTyxxmJFUOvEZ/zu9b7EjsdKFSfrCdl/rBk3v7UBtU2pCRJlQpgtWBv9IWR5nKhTrSrvfrUv4WOxY9T5QvCHwqj3BXEPV5QqXtR/UwLuDb4UrctpH4H/7PKdWFl+GKf96VP8YcE3AKK/DSfXDldco1gJfPY9mRuk0iL+ITfD1eqxS9KLFPg25UCtD39ZopjQ8oSVdbIa/nWvlOGKZ1enbGypYOvBOry6ajd+/ZpiMhWDohr9IZw3tjcqa6I57EwwJNJZ7f0N+wEAo+75CH/84FvTfao57ZA/p2aWASbIzQQ6M7uHI9RyUgUUDT9Zoc+ObXbcY82JpWcygZ/sfbNsWxXeWb8PN7+1Ian3iTz3+U68+1Wl1vOeF/jZXheG98wDALyhBvMlAn8+jjQE8NSyHfiEq7AXCMcW6IlcB1byGGg/DX/R5gMonbfQcpEVCEXw4Aff4ornlN/zgk3KfR7WBH70/HoF91M4Qk2DD9l72YLQqi01qwgp6bhIgW9T+AlGNKV1BpM+m2Qc6h0YjFD0zIt2Bjzc4Eev/AwcaQygut6PN9bswbaD0UCsijg+Wg832T2jRoOL8OZg3nd/2/xNhn2ZBmVmKmapc5EIjRkYddObG7D1YHLBZOzYZhprogVkWB/0WIsRM7JVgZdMvIAZDyz8Fr95c2O0bn4oDEopGgNhZHmcuHhCH0waWGhIT4xFgDvPNY0Bg9Xr36v2YJkwbl7YJRLAyLvSnI72mSr/ukQpp7ynxrxYk3j/HW5QFn2ssZaDkKjAd5u5n+KX3LVyI7GFo6TjIgW+TeE74+Vl6lfWnaGABjNVu9SJNBiOYMKAAu31SQMLtQXAiQ8uxrx3vsY/uEjpeP7eROamBn908uSrF9aZCFJ2Pczq+2saPqVY8q1eyLRWM/QHI/AFw6YLiWMJmNqbA2FtkZFsJHyqK7MxQRQIReAPRRCOUGR7XSCEoE9BpmnzoFA4ogvGY/CLlzpf0NTk/mMhk4EXdoksfvjzFY5E8NKKXW2ensdSMp0W941VCeKoSZ9o39MswNRMWRC7cFr9trpCt87OjhT4NoX/zfEafm6GK2lNjWGnXGI2FpeDIByhoBQ4riQXa+6YhU9vnYEnLxuPwT1yLN8fb9HD+7ytZC6f775hb602Hn57eVU95jyzCt/sVwL7nlj8nfG7qBcrFKZ47KNtutce+/4YrP79LAzvmWs6hgFqtLkVc18pw/C7FuHDzQcNryWi4b9Vtld7nOx9k2qBz86TPxTRtPkstcJihtthWlnxtvmbMOLuRYYYBn8wolVnrGsOxVxYBUIRZZHBHd/MYvLnj7dhxF2LtOe82f/llbtx7/++wdttnJrGzotVHwmrYEQWAhPi0jzNNHyze8Cg4VtkhqQ7rVPSeqTAtyl8gFoeFyzjdTmS0vD5Ff0Vz662TR1xpuE7HUT7Pm6nAz1yM1BalI1MjxNj+uZbaupm5+D2dzbhPTXwi9cWeV8sTyOn4X+t5oSX5GXoTMv3L/gWX+48opsoxYmPLV7MNKOCTA9K8jJw2vAe2jZeNhVkxo58rle/xz0mXRKPJXAtD3AZAMmWr+U1umQCJq1ggiQQiuCYalHJV7+/1+U01V5ZIN8BIZMhEI6gh2oBqvMFNUuRGTMeW4bj7vwQL62s0LaZWTv+urQczcGwtrjg7w/mWtnfxqmNmsC3cG+YCfynP92hXZ/fzt+EpaorI1ENX7TgWykGmlWJCgAAIABJREFUUuB3fKTAtykRbrLlBYnb6UjKh89rKWsqavDm2sQDo9oSNnm4nQ68s36f+lgv3d1OB76cNwtb/zAbH944Xfea2Tl4fc1e3PTmBoTCEZ22mG8hVHlNnhWFGVqSo5mQv9xxBMu/qza874jQyyCWe8GrpvHxgn3m8BJcekJfAEBeHIEfi7rmIN4q24vr/2+95T58bnWyJn3+e4klfhOF18y1eIRwROsHwdIvM9xOU62Wae7bhNgHfyiMIrU5UV1z0JDuyMPSHlkQrPJ+6+/DNF5+0ciuY70vhDpfsM0a0bDPtNTwTRYCjyzaqqu++f5GJZDPXMM3CToVrq3V/dzE1VEwG4fE/kiBb1P43xxvDvYkqeGLflG7WPXZhPZF+WH8/t2vAcBUS+uZn4EMt1Mz/TLE4CNesIh11K0EPq/Ble0+ivxMN4b2yNHyud8QFkeDi5XqbnywHxC7wQ2bdHnBHqEUD118PNb8fpbhexXnei2PxZPpdqK2OYjfzt+EhV8fsNwvViGgePBavVW/gHg8yrk42CXyB6MCv3u28n0z3A4Ew9QgSJmF52evrtO95g9GUJjtASGKwE+2vW6sxQ/7rvximS0gqxv8GHPvx/jTx9tM35sqzGJF+HGIrN9zzLDNTMM3e3/E4MO36E+gno/L/vUlRty9yHQfib2RAt+msMntL5ePQ2n3bG272+lIqtFIo1CnXfxxpwszDUHU8HkyPWItAv334IW8mHJlZYpsCoS0ToSUAuP7FyDL44IvGEEkQg2BU2xBcqRB1PCtrwdLjeKzAEIRqrgv8jLgFjohZrqNk7TIacOKkZ/p1sziifLb+ZtMyxVbodPwWxg3YlaS9kBtM+a+ug4A0C1bWQixiHh2rdbtPoqb39ygXedAOKKLXGcFqXK8LtT5Qgmbm8f1UwJDRfcGv7i5492vUVZRYxpEuFhN/VvMpQCmCv47WH0fq+1mFi9ew2eWEjZ3bKo8hnvf3wJKqU4JoJRq5/x3s4frFqCs7oPZ4kLSMZAC36YwjdXpILp0GI/TgUAoOjntOdKE0nkLscqig5oo/OzSbtVs4nI5rW9H0Q8vCiBeoLLv/Lc54/GDiX0t072aAmFN4ADAiaWFmsbdHAzDLVgcfjFjMAB9RD9gXHz8cHJ/zQTM+iB8b3xfTB5UCAD48bRSbV/R+pCIwI9QICfDpbNQWFXRE7cv3554ASa9hp8609AWNQAS4DR89XyxBerPXl2Hd4RCPPy93BxUUvpyvC40BUIxK+vxC7fjSpRAUEPZY25h/M5X+3DpP79EYyCE743vo11L/n2DirPREhZtPqCrF8DD13+w+j7xKgjy8ONm9zVb6Fzy9Eq8tLICvmBEZzkJhKPPLxrfG1ee1F977UhjAJ9xLq50V26UJI8U+DaF/QYdhICXO26XXsP/7DslQOe/G8yrlPGpZ4CSXmQHzCauWJHWoiAUBRBfde5jdULN8bqQ5XEZutExGv0hdMvyaM8Lsz1am97GQAhOweIweVB307GLQU7TBhdppmim4RfnevHG3CmoePhcnDYsGsBXkKUX+FYV3UryvNoETsGi2qPjsPK7ippsIgsKs2M2+ENJTfA7TZoMidx0+lDNcsM0/OteLgMAdM/2GPbnYyd8gbDm6mn0h021cTOy1et705sb8LXalREwur4AxeWT7XWaLnYSrXIo8vN/r8d1r5SZvsbfV5sqj2HzvlpD7EIyvvPx/btpj9mCmc0dQS6AUucqCUW4DBqHwQJ19QtrdPtKOhZS4KsQQmYTQrYRQsoJIfNMXr+GEFJNCNmg/v20LcejFaYh+oIXXqdDV3SkTp14rOpcixq+XfqEm1XzEicXHtG8Lro1eA2fpcZNGNANmR6npYZ/uDGgM1nmeF1aRkRtUxCifGPauihERb+z00FAoIzXa1J7n4dfcADWRZUafCEtluO0YcXIcDl1AkIUSntrmvDyygos26avCyB2XowF7/6Z+fhneGJxYs1TFm0+gJmPf4bL/vVlzHoI/MKHfdJGVQgX5RoFfo0amEYpRROn4df5glhRHrVwMddQz7wMBMN6gcYX03n+i2hBJrGBEaAsArK9LtOSxDWNyblTEoG/r94qq8R5f/sCPxFqCVgtXs34xamD8bNTBwHgNXz9b8EfCuuyMXzBsLYYcDtJTDdbSxc9kvQhBT4AQogTwFMAzgYwEsAcQshIk13fpJSOU/+ea8sxRbjKWbyw87od+qIjampWhoVgETUXs6Iy7c2CTfvx7Oe7DNuTqVcuCkZx8pk2pDvyM93Icjt1uck8VXU+9MiNVvfLzXBpxX5+89YGrW87APzzhxOifmYh4CsoSAS306Gl3onlTUWY1SBX1TyttKbGQBhDeuRi/V1n4Jqppcj0OHXXNhiKlvctr6rHnGdX4Z73txiEVaz0NRFxIfM2l9MfC5YWtnpXDYaV6OsPFOVEF1j8Aq+qTr8AZKZ+nppG5d5lwX2ZbieyPC7sqGrQpZu+eM0kFOd6ke11GlxHfCAbH8BmpuEHQhHkmKR0ZnuchkyNVGB27cXeDvUxLBnj+xfonjscRDv/zJIiLpT9goY/6cEleE+1FjodJOb90uAP4Za3NqJ03kLLfST2Qgp8hUkAyimlOymlAQBvALgwnQPiBT6vJWW4nTqzHstPtpoIRA2/LVflvmA4bgtWSil+pdbPF4knjH4ybSDuPHcEAKPfXJzYWc91NtHtOtyI2uYgyqsUE2kkQlFV70dJXlSw5Ga40bsgE0C0gx4AnD26J2aP7qVp62Kks2jSdzkJRvXJB6BkVcRidJ98bL7vLNx/0SgAwJAYxYZqmwNqZDqBx+nAJs4kzSbyxz/ehtP/vNyyAUqsAEPjvsaFTCLwQY2iJYe3qPDa44ml3XT78Q2jbj97OIDoNWb3f6bHhWyvy9BtcFBxNqYN7o5gmGrXii2aMz3R7xAIRzDw9oV47KOtli6BbJP68aN65+NoYwA/eWktZj+53PB6VZ0PZz2xHB/GyJ4AlPLRVXXGtMlTjyvW7dccCGPjXiVQzswSoY3VZHHCrqHow49+ZsSwsGP3ldvp0K6RmbetwRfCf9YrhYgqj5qXApbYCynwFfoA4NWXSnWbyCWEkE2EkPmEkH5mByKEzCWElBFCyqqrjTnciVBV58MFf18BQJmonJzEz3A7dZoA+6HVNccW+I9dOgYAUO9vOw3/kUVbMe3hpfj2QJ3lPrHqpccyHwLA3eePxMUTlPx1ZsIunbcQpfMW4pdCLjoT+Mx3edaTyzH2vo9x+p+XIxKhqGkKIByh6JHLC3wXeuZn6I5DiBL8pzwm8LocBrOoKBidDoJnfzQRr1832ZBdYEaO14Xvje+L5bedhpOHFFnux1/jw0Jq4IMLv8E1L67RBVWZkUwNB7HYTrzrw+DN5kcaA5r1AoCWOw/ogzSnDinCL2YMhtNBsO9Ys86qNXWwck7Yfc9cGZlup6GTJKDEgricDhys9WG+WhnvkgnKz/m0YT1wzdRS5TiBMCgFnlq2w9I3LsZYAMCIXrkIRSiWbq3C1oP1+HjLQRyobcatb29Eoz+E8qoGbDtUbyiWJJ7PiQ8sxqQ/LtGeM2HML0K9Lgee/mwHLnxqhWXmAMOsHC8LCh1UZB6s6A+FLWsKuBxEW+RNHVyE88b00r3ON286+ZFlONoGVg9JapECX8FsJhN/Bf8DUEopHQNgMYCXzQ5EKX2GUjqRUjqxuLjYbJe48G1CCefDdzoIMlz6YC2mUddadE7708dKKdjvT+yHGcOK21TDZ1XIfvXaetMAr7fL9uKUR5dZvj9WlD6DCZ1gOKL7DHHOYkFfYp47oJSkZSlt3bjgsNwMF9xOh05A5XpdunF5XQ7DpBkKR3QBh26nA/lZbkwZ3D3u9+Hp3z0rplDli9+IY3hvw358uq3aMor7grG9lWMkEW3fUg1fpJAT8nzHNfG7XjiuN8IRimkPL8W/V0XdKd2y3Wp2ijJ25sfO9DhMqyi6VM00EI7gkUVbAShaecXD52JoSS7uvWAUJg8q1GVbWNUZEBeAADC8V57u+dxX12HKQ0sxf10lvtpzzODiYYgpsiLsmpZwTaT8oQg+Ussqryg/ElPD5+/BrX+YDQA4c2QJnrxsHG456zgA+qZD7PhWqbpOdeEEKO4BMeDzoGBZqfOl310oiY0U+AqVAHiNvS+A/fwOlNIjlFKmVj0L4IT2GJgSpR8V+JmeaLCWPxTGIdX3eUjwgTYHwgZtLi/D3aY+fDZR7ahuxN4ao0n5tvmbYvo+3Qk0mmFCZ8v+OlQcsTYjMs3GTMM+0hjQfL75mW786rQhmDm8B4pV/zJfJEfUmjLcRr9wSG0Cw2hNw5xYi54mLuPCytdvVZL1rvOUkJSkNHzaMoEvLioKuUVVtk7g64/Xt5t5X4HiXK+60FK+G7MSZbpdyFE1fD42wOUkhmN3z9EHAXpcTp2VxKqdLt/B8dczh+Dnpw5GL5NFACMYiWgpo6IY5bVzs1LFbAw9hOJL29Q+9YfqfVqQrhlOB8GQHjkY3SdPs7IQQnDR+D7aQsug4QeNJn0GIdGgPZeQHgwA+4/pBb7M0rM/ssGxwloAQwkhAwHsA3A5gCv4HQghvSilzCl3AQDzJuspxumI+vBdDqITOOwH53U5DL7zEXcvwvCeuXA5iBapm5vhalMNnxcmR5sC6K82hvEFw4b0IjMS0/CVfd79ah/WVtRY7seyFsw0/JrGABpU10ZBlge3njVMeG/0ZyFOcqYCP0yR43Vpi4hkAuNEYglVXkO0KsBiloXRPdujxRIEkqjDINZsSNSkL1oG+BS7nBgLo2wL94fX5YTHFdXw/75UaSGb5XFqGv6A7lmaAHcJwWZOB8G5x+vN0R6nQ2stC1gvhHgNf8qg7pg6pAibKq0LzzT6o418RAHIu7P+utSY8cBM+j3yzBcUr62OXRbb7XRg8c2nmr7GCkyJ31NpVRz7mIByDocJDaAO1OrnnGTiQyTpQWr4ACilIQC/AvARFEH+FqV0CyHkfkLIBepuNxBCthBCNgK4AcA17TE2QgCq/o6iJn3FnL1PDcyaNLAQNY0Bgx9y68F6raoboAjBOl8QpfMWmlZAay289sA3drnj3c248KkVcd+fSJQ+r3FbBaYB0XQ4s4m8ptGv0/BF8rgUR4dBwzd2dQtFIrqFRTLZBlbjNuP7J0SNUFYC38yk/79fn2w54ceiNRp+N873zUfc60z6wnclMXL4eIHPFhSTBhZqx+MbTCn549Fj9cj1Go4tnmer88K7DNiiqdCkRgCj0R/iFlX688dr9U9yKY5f7lBSCtnvR9TwRWZyjZh4rFrqAspi2ukgBg3/mhfXalkVZrBbwEkIrjipP66eMgCXTewHt5MYlAyZl29/pMBXoZR+QCk9jlI6mFL6oLrtbkrp++rj2ymloyilYymlp1FKt7bHuJyEIDfDhQvG9saL15yIDFWwvLyyAj98fjUAxT8JACvKzauosYk6L9Ol+Sr//Enqa4HzkyZf9jWWJs4jVrZrDUwADyxSKqLxgVBHm4JaL3lTgZ8ZneSdJhq+WIc9GKaCqbrlAp8PeJs+VB/Ad8OsIdrjyyf1hxlmgqt3QaYmrJLz4ev3TUrgc0JR58PnBHMi1/uFayYCUAQ0y0QIhCMY269AKbyjmvR56xAfbAaYp9yJ1ygRYcXOoVnKIGP3kSY8qbZQFjVnq+JIc55dpY5Bua94H/7Zo3vq9s10OzX3jEgsgQ9Ez6EYX2PWepnBcvSdTsW6eN+Fo/HIpWMwa3iJoYOhFPj2Rwp8G8LLGIdD8eH/dc54TCwt1PKI+c5f/QsV0/lPXynDqp1HDD9oNrnxxXnyLAr1tAZ/KKKlXfGxAlbtNgFg7R2na49boxnzDO+Zi0kDlTK2g4pzsP3Bs7XudIBSpIdp+LxmyODPjZkPX8w0CEUiOs3V2YqFC695vnrtSfj01hnac15L/e1Zw3DWqJKEj6uUaE5OwxcvW6wiOjyhsD77obulD994wNvOGqYFTQ4qysbM4cp39Lgc+GjLQQy780Ms/65au6dYKhrvHnA4iO5eMhNEYrqkmJ/+/NUTseDXJ+vGyd6T6XFaViz8x6c7sLNa6Vcgivd4HfaYSZ+PN7hwXG/dPqVF2bq4gtd+ehJ+evJAAPHb13pcDtQ1Bw0LDzHjg4dZJcSFb6+CDJRX6aspJnNvSdKDFPg2R1y0Mw2Qr9DWrzBTe1zTGDAIJE3D54SbVQe5lvDm2j0onbcQNY1+baI/2hREdb0fB2qbTTWbX84YjOevnmiZl90a7j5/pE44up0OnaCp94VQ2xw0ROAzsrhUL1F2F2S6tcVC5dEm7KxuQChMdZp5a4L2RFMzqwsgQgjBAxcdj9F98kxfN8PtTK7TohhYFiulkicYofBwBW54E3iubmFkPE/XnzYEt5ypRJTz941HdWUx4c0a6bDL7HQQPHHZWJwwoJu2P8NMEIkCn1l8GLNGlGA0q6Wg3iMe7l6JZdZniAvvuAI/FIGD6D9nrNrs5yR1AZvjdeoCUacOKcKgYiXlLp7AHdO3AB9uPphUX3s2ZvGeNuvqKDV8+yMFvs0Rg8ZYHvPOw9GuZ/yKv8EfMqT/RAV+VMjnmGi2LeWfnyklSvfUNCNbLU97pMGPEx9cjCkPLTWd6K4/bQhmjdBrqK0JduMx84Hyc29dcxC1TUHLXvS89iZqNt2yPFpZ1ZMfWYaZj3+GUITC49IvMFqKV9AcYxXuKc714vHvj0v42B6TlMJYiAu1dbuPxgxY094XjsDtiFaI5BdbvKnfymfPFk/8fSNWLGQCiBdI3xvfF//5xVTlGHEqHHqc+tdjabnsmvDXosjkHhO1fvGutzLpA6xLXQRup0M7LzOGFaNXfiYqHj4XZ4xUfitsPvjFjMF4/PtjAUTLJcfrojl1cHc1JVVZ3Pxu9nBDMKOIVuJbEPhZJhaOn7y0Vqbm2Rwp8G0I4coCiAJ/woBu4u7I5wKkXlpRgUO1+snLowXtRSfeeDXek4EFdzUFQvC6HCjJy8BBroKYWSqeWfR8a0z6g7nuZcU5xihnvlgO0/DNiqoAej+6eP67ZXtwrCmo0958wbAhKrylmNW6/8vl4/DSj0803Z91f0vs2E5s2Jt4a9MIpQary0aL9/uCYTz3+U4EwxFNcDGLEi8IC7Pia8bs/PPWCI+wiGKxKGwx0TNfbwmJV/BIDMA73BCwtMywz+YXKMU5RoEvpv5RCkx9aAke+0gJ94nVmjocUdrSssXirofOwUs/nqS9Lrrgfjd7OC5R3VTs/MbrhMnOPfO9d8/2mNYZAIB/X3sSAMWFABjL9prVPwhHKOaXVcYcgyS9SIFvc0ThUWQy0RRmebR2nd8cqMMVz63Sve52GX34qfS3MS2gKRDWBL5YFwDQR2jzkyebdBMVlL+eOcSw7bXrJuOqKQMA6IPuGD6+/4Dqw7dyazgFfzBPtyw3AuGILv2t8mizUHinNSZ9o6C6cFwfzBhmHplNCDEEdvH8/pzh2uOzR/fEhr3HEu4sF4pQg6C1sl68uKICDyz8Fm+s3YtQmMLlJJrZlxe+fDtiK9iiJyyY9HnYgvGMESV46OLjcdPpQ3Wv84sMsZ4/fzxmDaqu95uW0QWAq6aq9xW3YDYzaecI728KhLC/1oenlikZMbEE8qZ9tepCSbl3rKwfZpYFFsgbz13DFrgsnc7rduA2ISWVwc7PpIGFWHzzqbhCCBK1WlAdaw5ieZxqj5L0IQW+zUkkUMrldGDpLTO052KuPZuk+Uj1VHbN4xUXj8uBHrle025404aYV567/EQl3SzLnZib4ZYzhxlK0Ga4nLj3/FHY9sBs08ly7imDcPbonhjRKw91qoZvJfAdJosRBps0xTKivHUikXoCVrTE8mIVgPmjyQMw95TB2vMpaqOe3UcaTfcXiUSoQdA6HQSb99UaTLdM2OytaUIwomj4/dRCOny6aKwIdwYzoYs+fJ7XrpsMQFmQzZnUX2eVAaJWgu7ZHrwxd7LhM9iCgN0Dh+p8lu6TX5w6GDv+eI5uwVycY7RUiAsGvnpfJEJjavgX/2MlDtT6LO+dGcOLcd6YXrj/glGG15j7wqrCH4Pdy6xCntflMJw3Bn8uhvTIMfymzCx0APDXJdtx1QtrEs7MkbQvUuDbEL5dpWhSBoDeqhluw91noOzOaJT73FMGmR6PCfwCzpzalKCWlwiirzXb6zLEEYzuk4cnLxtv+v5bzxyGDXefoXNNxONRtTcAI8PjgMNBLLvTFeV48fQPT0Dfbpmoaw6iKRC21OjYGc9wO/D3K/RjZhOk6AsX08JaSksEfq5FPIboImGFkHbHqFDIE4pEDOezORhW2ra+qG/byu6Bz7ZVY29NMwLhCB66+Hj8cHJ/rSMgkFh73iztHPMLBb2AjdVkCIhqoMeV5OriBhhs8cvugZrGgMGawSBCx0pA/1vSxh3DjXC40R/Thw8ARxr9ltUme+Rm4O9XTEB3Ewsf+67BOFY7JvBZ/QoxXoQnnpUqnstkJdeuWGIfpMC3IfwP18zM/cGN0/HprTNQkOXRmfhvP3s4zlSDe3iTIz+RPXLJ8Upd8iSitePBay6K1uAwRD0PK8mznCQcDmI6gcaid0EmfjAxmmpnNVmLsGqDzcGwZWoVW2P9eNpAXftcIGpy94fCOhMvP1G3RuBbaVyxsAo+FM3v7BzHqsfOE44YNWvWCa9s91Hddhb0xsrAfnewHj3yMvDARcfrjkEIwWvXnYSHLj7e8nOjFRqj9+io3tFshMU3nxJ37Ox+sIqMZ4GuvmAYfdRMiHidDXlE8z1gvjhnHKz1GbIe+FRRQCmd3BLrEFtExTPp9ytUejWs36NcO7PAO0a8onlWvx123qsbfKavS9KLFPg2hDfNmcmOgiyPFkzDQwjRhCqvEfGa3mUn9scVJ/VvtQ8/EqFaeg8v8HvkZZgKLRbkVpzrjVmLPBkevXSs9jhWlTaePLXaYHMgbLkAYccys8B61cnVH4roorCdKQraa4mGn0gsAn9ssXCQFeFIxCjwG6Oumuu5DoV8Ixqzz+aZOrgIcywKBwH6rBMGf78P6WH0yYswDTVsYUafWFqIi8b1xj3nj9LqWCRz7sV758yRJTFN9gdrfQYNX4zHqW0OtihwlS3s4lkQ3E4HhvTIxZb9SjdLlvJpVosi1ncBrK8vUySIaT8ySbqRAt+G8IVqYmkNZrBJi/fripqe20ksu4MlyqMfbcPwuxYZhEeu16UT+KwM6FlqYNmX82bii9/NbNVn8zx/9UQtWC8R8jgNP542TQ2JVfqSvXxvcbeTaKlTiS4+zHA5Hbhu+kC888upCb9HrHGujclS4Ce22AuEIzpNLtfr0vW6X8j1exf7rCcaGGgGIQT3nj8SL3KZCWYadSyYQLISgh6XA09ePh7DeuZqkep5GW5suPsMbLz7zLjHF833z1w1Maar5FCdT7M2sAWa2DvgYJ2vRdUm2XsSqbEwQr1XnA6iLbzvPFdfue+MkSU4Xq1BYEWy85LEHsjmOTYkFIntw48FMznzkepGge+I2ehi3e6j2FHdgB9M7Ge5z+trlEYe9b4QmgJhuBwEoQhFSX6GLmDv0hP64l8/OkEbQ2sC2syYNaLEkM8fCz7wysosydranjLU2N6Ynd/mQFjnFnE5CZ66YkJK8pDvONe8dKoVE/p3w6TSQqwRAqXEin9Rd0RiAr/eFzK0DzYLxgSMOeCiwF96y6kxO72JXDNtoO55SwV+OIGGLqyUbbbXmbBrycw6dP7YXlpEPiPb44Q/FMHBOh8K1YDFiyf0wYsrKkwj7lui4bPf+vj+xpRdEbY49Loc2m/xByf2ww9O7IfSeQsBKCl/YnaKyMheebhu+kBcNaUU6/ccxY1vbEh63JL2Rwp8GxJMuYavP4ZScY2CUmqqjV7y9EoAiCnw2YR6rEmp7HfjrKEY2TsPZ44swWtrol29MtyOVhWiSTW8+dsqgGxC/2747oGzTX267PxWC4VanA4HPC6HadpkW+NxOfDWz6dg9pPLsZXrSigKD7dT6bzoT7DSWl1zEAO6R03p+VkeXX0FvmqeuIgQfeesGlxLSbZQ1JAeOeiVn4F5s0fE3benGsCXTFyLmRvl1jOH4YZZQ3HTGxu0+vSZHhfyMwmeWrZD829fedIAnDemF0b2ysft73ytO0ZLNOeCLA8+uGG6lpobC7a4iWUNSMQl5XAQbWHarzDLIPClAcCe2Gcmlmjw+brJWviYjzkv041rppYCMBbJiDZSablZn01MrEVvQZYbZ43qCUKIrspZS4LQ2pKRXPCXlYYPWAdwsUXCb+dv0m23iq5uT0QftBg8SAhR+8orPdDr41gj6n0hXQZAtyy3ltIFAAXZ1nUdzFLhWoNVRoUVWR4Xvrx9Fk4WGhCZwUz6jf7EU1VH9srDHy4ardumnF8nnv7hCXjrZ1MAKPcLax7EFhQuB8EJAwpNF5zxronleHrnJfRbS+Q8itUlJZ0HKfBtCG9uT3bFz8y4OV4X7j5vJP73q5MNKUxM40+mrjqPUkteWSzsV1tk8qZyfuKxm8Af0YsT+HFSi8wQS7IyUu2qaAmiJcVsTF6XYmL+w4JvcPy9H8cM4DMKfI+lT5wX+Oce30vr4Jgqsk0qu6UKVqWvMYm4A0IIfjTZOnYkW+3HkOF2GjJImAZNCDEEsB5tatvStCxuIFZMXioqXMeJ+ZOkifTPUhID4/pFy1gmG/HNUq4KstxwOAiO72uceJlgaInAD0coxt73MQ6rwVuspj+fFcBrLrG06HTgdBAtB7slEfFeCzdAa1LxUoVB4JuMSdHww/jPOqUEqlWKni+oxCjwriGxFHETpxHz5nCrc9QaWpP5EA9278aLTDfjsUvH4JPfGNMEWarfddMHmhYvYvz5B/peCM0pLIhlRiIafioC8qyyIyTpRQp8G3Io6nZGAAAPwUlEQVTZidGUpWR/e8eaFUFslZsNRAVDS3LxWVlOxupdSqAYHymeaWMNHwCumlIKwLyMbTysFgmpau3bGkTBYjYmr9uBt8oqUa9qs/W+EF5dtRt/X7odTVyxpDqT9sF8TYJpQ7rriivxGn5bRnCzTIhU0qcgEz89eSD+ceUJSb/3+xP7YahJ6d6CLA8qHj4Xl53YX9c5ENAvxPimS0DLfpPJwAR+LHHcksWV2LUx3MosIEnbIIP2bE6yk2etahIsiCHwPc7EfPiRCDVE67K2pIyNe48hx+vSmSZ5U7ndNHxAac07vn8BJg80L/UbC6tFgh1M+gaBb6rhO3UBdQ3+EO56bzMAYNGWg1jw6+kAoAXnlXA58ZPUFq0AUJjthS+oxAI4HUQn8MUCM6mi/MGz22Qx4XAQ3HlecpkRySCa9PnfVKo6RCYKczWIrXt5WnKO/3v9yfj3qt245/0tAKSGb1fSP0tJYpJsAM21qgnxxNJCy32Y5hcMRXDzWxswf515h6ugSUrTzmpjHfaCLLcu2j8vgUj4dEIIwdTBRXFTj8wQhSqbzO0QtMf8s0xDMy0cJIyfTyPcvK9Oe8yCMVlxFgDoVxh9PEAtVsNiOALhaJGeeAVgWorL6WjRNUs3oouDX4gNKs7GmSNL8MpPJolvaxPaSsN3Oohuod9Wiz5J67DfbCzRkexqe+rgInz3wNmm9cMZvA//nfX7cOvbG033M7MArBdKqgL6gD1ATH2zn4bfGsTJkOVAt6WPOVFuPWsYLpnQFzOOU+oHmMVoiAK/wRfSFgq8+Z4Jct5yw1s3WGzINjUNMBCKaNddand6vBZBe4Dy23nmqok4waTtdVvAgh/PG9Pbcp+WRunz91ZbLfokrUMKfJvTFhY/JvD54ihmK/LN+2px4xtf6fbbZdJpTSzN2ZkFPs+jl47RFjt2qDXQt1sWHv/BWPRTte+AyYJNjO2obQ6iWc3LbwyENVPvsaYACFEi8xm8tWa4GrOx96ji4vGHwlr1Oem/1RMraI8Rq/FOKnE6CNb8fhYe//5Yy31IC29l3nUhF332JP2zlMQUZvZrC58lCxT63j9WattYEBcv+H/+73X474b9+GxbtL+1WeqSqOHbQfi1JWt+Pwsf3DBdV5jIDkF7jGidBaOG30Oo7nbb/E2IUKWKXjhCtQI6Df4wsj0unQmd1/CZb7+2OYhIhCIYplq8hpzs9SQi8FtTjjlZeuRlxGwU1GINn1sQykWfPencM3MHhvlO2+JnYyaQ65qVhjLHmqM+XTZvf7UnasY3K07Cl/HtCvTIy9AV8AHaP/gqFm4uRkNE7P7HYCl3zJrT4A8aytnyFRu9LgdyvS4cawpqkeXMh2vVoa6rIgp4K4F69ZQB+MeVE9pjSDFpqXuKr1EhF332pGvN1B2I1+dOxqfbqpKuIZ4IZgK/tjmI6Y8uw0CuK1mtKvwb/CG88MUunDGyxLQpSl6GMSPgj987Hr0LUtMVz86wymjdsqyzItqbWGmXVtYX1vxmR1UDinK82FPTpEV0M3gtlBCCvEw36pqD2nsHFGbhqz3HMDROr/quhugusxKo91042nR7e9NSqyJvNZCLPnsiBb5N6VOQiStPSrwLXDKIZl0g2s9812Gjj/7z7Yfxxtq9WFtRY2rSN6vhfcVJ1u1POxNsUWTWCCVdnDGyBE8u3m7aVGjOJMUN8YsZg/FW2V7cqabklRZlo6rej8ueWYWfnTIIq3bWGN4rkp/pRm1zEBsqjwFQmg5dPql/uwWgdRQON+pbB7en+b4ltFjDlwLf9kiB3wUZVJyj1VRnXPPiWsv996kR20qXvegP+bRhxVi2rRqnHmfsKtdVYJkM6WiaY8Wo3vmoePhc09d65GXgxtOHAgCuPKk/pg8tQr0vhKIcLyY/tAQA8K/lO3Xvee/6aYZWroDiBliytQpLtlYBUFK+Jg9KvrZBZ6eKazjUEWhpwgkfpd+SqoWStsc+jkdJu8L3Gk8UvhIboPS6r3j4XF1Hta4G04bETIWOACEEA7pnY3SffPTMz8CaO2aZLlzG9SswrSYnupuSbXDTVTjO5NzZmZZaIHRpeTJoz5ZIgd9FETvoJQJr7MGEm1naV1fjwxun4y+Xj7O9mTYReuRm4OIJfbTncybFdsuIkd5tEW/SGbjrvJF47/pp6R5Gm8Mv+GTQnj2RAr+LwptoTxFM8sWcP5ovjVtVry+3GqvTWlfhuJJcXDiuT/wdOwh8E6QHL0ouiKwtO9p1ZDLcTozpk9rugW3BAxeNxqCillvr+AWf9OHbEynwuyh8GUyxrKeuYUpeVPjvrVF8+ax/OIvOlnQe+IC7ZMvYilH9kigdoSTwDycPwNJbZ7T4/XzxICnw7YlckndRYgWZDSrOwQ61Zn5xjhe7j+gb5ozqnY/Ptx82tEuVdHwmlhbit7OHofxQg+nrn902AwTmwiuznarFSewJ79aSQXv2RAr8LkqG24mXfzJJ1+WMUZQTNevyGj5j7imDMKJXLs49vlebjlGSHn45Y4jla3yAJj+lTx9aZFnUR9L1kEF79kSa9Lswpx5XrPUXf/knk7Se1nzTnGITS0B+phsXjutji5awkjSi3iaDi7PxlA0qxHUEpg8tSvcQ2gWp4dsTqeFLACjC/0iDHze/tRHBcAT/vX4a3v1qny6ADwB+NHmALTrDSezDb844zrTaokTP1j/M1rXG7czIbnn2RKpoEg2+6crYfgW494JRuih9ADhHmvElKj87dRAKstyYIovtJESG29nprWIf3DAdx5XkmHbflKSfzn33SZKCNYAJhKI/VhZ52y3LjU9+cwqmDJaTu0RhTN8CbLj7THS3UZVBSXoZ2TsPI3rlyTx8myIFvkSDtc3l26ryHdDMqq1JJBIJj9vpQL0vJFPzbIgU+CqEkNmEkG2EkHJCyDyT172EkDfV11cTQkrbf5Rty8heSnGQS07oq21jFfnkj1cikSTCrOE9UNMYwJJvD6V7KBIBGbQHgBDiBPAUgDMAVAJYSwh5n1L6DbfbtQCOUkqHEEIuB/AIgMvaf7RtR8/8DEPTFWbSl0E4EokkEc4c1RPPXjURM4f3SPdQJAJSw1eYBKCcUrqTUhoA8AaAC4V9LgTwsvp4PoBZpDMUUI8DC+TzuuStIpFI4uN0EJwxskRm89gQOYsr9AGwl3teqW4z3YdSGgJQC8AQwUYImUsIKSOElFVXV7fRcNuP/oVZAIA/Xnx8mkcikUgkktYgTfoKZktR0YadyD6glD4D4BkAmDhxYoe3g5fkZWDXQ+d0im5wEolE0pWRGr5CJYB+3PO+APZb7UMIcQHIB1DTLqNLM1LYSyQSScdHCnyFtQCGEkIGEkI8AC4H8L6wz/sArlYfXwpgKaUy2VQikUgkHQNp0ofikyeE/ArARwCcAF6glG4hhNwPoIxS+j6A5wG8Sggph6LZX56+EUskEolEkhxEKqltByGkGsDuFr69CMDhFA4nVchxJYddxwXYd2xyXMnRGcc1gFJanMrBSKTAty2EkDJK6cR0j0NEjis57DouwL5jk+NKDjkuSaJIH75EIpFIJF0AKfAlEolEIukCSIFvX55J9wAskONKDruOC7Dv2OS4kkOOS5IQ0ocvkUgkEkkXQGr4EolEIpF0AaTAl0gkEomkCyAFvg0hhMwmhGwjhJQTQua182e/QAipIoRs5rYVEkI+IYRsV/93U7cTQshf1XFuIoRMaMNx9SOELCOEfEsI2UIIudEOYyOEZBBC1hBCNqrjuk/dPpAQslod15tqBUcQQrzq83L19dK2GBc3Pich5CtCyAK7jIsQUkEI+ZoQsoEQUqZus8M9VkAImU8I2areZ1PSPS5CyDD1PLG/OkLITekel/pZv1Hv+c2EkNfV30La7y+JNVLg2wxCiBPAUwDOBjASwBxCyMh2HMJLAGYL2+YBWEIpHQpgifocUMY4VP2bC+DpNhxXCMAtlNIRACYDuF49L+kemx/ATErpWADjAMwmhEwG8AiAJ9RxHQVwrbr/tQCOUkqHAHhC3a8tuRHAt9xzu4zrNErpOC5PO93XEQD+AmARpXQ4gLFQzltax0Up3aaep3EATgDQBODddI+LENIHwA0AJlJKR0OpUHo57HN/ScyglMo/G/0BmALgI+757QBub+cxlALYzD3fBqCX+rgXgG3q438BmGO2XzuM8b8AzrDT2ABkAVgP4CQoFcZc4jWFUr55ivrYpe5H2mg8faEIg5kAFkDp+GiHcVUAKBK2pfU6AsgDsEv8zukelzCWMwGssMO4EG0XXqjeLwsAnGWH+0v+Wf9JDd9+sB8So1Ldlk5KKKUHAED930PdnpaxqubA8QBW22Fsqtl8A4AqAJ8A2AHgGKU0ZPLZ2rjU12sBdG+LcQF4EsBvAUTU591tMi4K4GNCyDpCyFx1W7qv4yAA1QBeVF0gzxFCsm0wLp7LAbyuPk7ruCil+wD8CcAeAAeg3C/rYI/7S2KBFPj2w6wXrV1zJ9t9rISQHAD/AXATpbQu1q4m29pkbJTSMFVMrn0BTAIwIsZnt8u4CCHnAaiilK7jN6d7XCrTKKUToJifryeEnBJj3/YalwvABABPU0rHA2hE1EyeznEpH6b4wi8A8Ha8XU22tcX91Q3AhQAGAugNIBvK9bT67I40r3VapMC3H5UA+nHP+wLYn6axMA4RQnoBgPq/St3ermMlhLihCPv/o5S+Y6exAQCl9BiAT6HEGBQQQlg3Sv6ztXGpr+dD6b6YaqYBuIAQUgHgDShm/SdtMC5QSver/6ug+KMnIf3XsRJAJaV0tfp8PpQFQLrHxTgbwHpK6SH1ebrHdTqAXZTSakppEMA7AKbCBveXxBop8O3HWgBD1WhXDxQz3vtpHtP7AK5WH18NxX/Otl+lRgZPBlDLzIyphhBCoLQo/pZS+me7jI0QUkwIKVAfZ0KZCL8FsAzApRbjYuO9FMBSSmnKNR1K6e2U0r6U0lIo99BSSumV6R4XISSbEJLLHkPxS29Gmq8jpfQggL2EkGHqplkAvkn3uDjmIGrOZ5+fznHtATCZEJKl/jbZ+Urr/SWJQ7qDCOSf8Q/AOQC+g+ILvqOdP/t1KD65IJRV+bVQfG1LAGxX/xeq+xIoGQU7AHwNJWK3rcZ1MhQT4CYAG9S/c9I9NgBjAHyljmszgLvV7YMArAFQDsUM61W3Z6jPy9XXB7XDNZ0BYIEdxqV+/kb1bwu7v9N9HdXPGgegTL2W7wHoZpNxZQE4AiCf22aHcd0HYKt6378KwJvu+0v+xf6TpXUlEolEIukCSJO+RCKRSCRdACnwJRKJRCLpAkiBL5FIJBJJF0AKfIlEIpFIugBS4EskEolE0gWQAl8ikUgkki6AFPgSiUQikXQB/h/AZAWYLEcDKwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fca60680278>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "reconstruct_training_sample = False\n",
    "samp = 20\n",
    "\n",
    "from scipy.stats import pearsonr\n",
    "# now let's see how the model performs:\n",
    "\n",
    "if (reconstruct_training_sample==True):\n",
    "    reconst = decoded_train[samp,:,:,:].reshape(leng,1)\n",
    "    orig = y_train[samp,:,:,:].reshape(leng,1)\n",
    "    R,p = pearsonr(reconst, orig)\n",
    "    title = \"Reconstructed input based on TRAINING data, \" + \"Corr\"+ str(R) + \", P-value\" + str(p)\n",
    "else:\n",
    "    reconst = decoded_test[samp,:,:,:].reshape(leng,1)\n",
    "    orig = y_test[samp,:,:,:].reshape(leng,1)\n",
    "    R,p = pearsonr(reconst, orig)\n",
    "    title = \"Reconstructed input based on TEST data, \" + \"Corr\"+ str(R) + \", P-value\" + str(p)\n",
    "\n",
    "\n",
    "print(\"Pearson's R:\", R, \"p-value\", p)\n",
    "%matplotlib inline\n",
    "plt.subplot(211)\n",
    "plt.plot(reconst)\n",
    "ax = plt.gca()\n",
    "ax.set_title(title)\n",
    "plt.subplot(212)\n",
    "plt.plot(orig)\n",
    "ax = plt.gca()\n",
    "ax.set_title(\"Original\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "R,p = pearsonr(reconst, orig)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[-0.04483732]\n"
     ]
    }
   ],
   "source": [
    "print(R)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}