prediction_Window_9_attention_transformer.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import keras\n",
    "import matplotlib.pyplot as plt\n",
    "from keras.layers.core import Dense, Activation, Flatten, Dropout, Reshape\n",
    "from keras.layers import Conv2D, LSTM, Embedding, Bidirectional, Input, merge, multiply, concatenate, add, GlobalAveragePooling1D, Layer, TimeDistributed, Conv1D, Lambda, Add\n",
    "from keras.models import Sequential\n",
    "from keras.preprocessing.text import Tokenizer\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn import preprocessing\n",
    "from keras.utils import to_categorical\n",
    "from keras import regularizers\n",
    "from keras.models import Model\n",
    "from keras import backend as K\n",
    "import tensorflow as tf\n",
    "\n",
    "from keras.models import *\n",
    "from keras.layers import *\n",
    "from keras.callbacks import *\n",
    "from keras.initializers import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(226, 9)\n",
      "(226, 1)\n"
     ]
    }
   ],
   "source": [
    "with open('dataset/PELM/fixed_sequences_length_9/Group_Phos_Y_neg.fasta', 'r') as f:\n",
    "    s_negatif_txt = f.readlines()\n",
    "\n",
    "s_negatif = np.array([])\n",
    "for i in range(1,len(s_negatif_txt),2):\n",
    "    temp = s_negatif_txt[i]\n",
    "    temp1 = temp[0:9]\n",
    "    temp2 = list(temp1)\n",
    "    s_negatif = np.append(s_negatif, temp2)\n",
    "\n",
    "sequenceL = int(len(s_negatif)/9)\n",
    "s_negatif_new = np.expand_dims(s_negatif, axis=0)\n",
    "s_negatif_new = s_negatif_new.reshape(sequenceL,9)\n",
    "labelneg = np.zeros((sequenceL,), dtype=int)\n",
    "s_negatif_label = np.expand_dims(labelneg, axis=0)\n",
    "s_negatif_label = s_negatif_label.reshape(sequenceL,1)\n",
    "print(s_negatif_new.shape)\n",
    "print(s_negatif_label.shape)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(267, 9)\n",
      "(267, 1)\n"
     ]
    }
   ],
   "source": [
    "with open('dataset/PELM/fixed_sequences_length_9/Group_Phos_Y_pos.fasta', 'r') as f:\n",
    "    s_positif_txt = f.readlines()\n",
    "\n",
    "s_positif = np.array([])\n",
    "for i in range(1,len(s_positif_txt),2):\n",
    "    temp = s_positif_txt[i]\n",
    "    temp1 = temp[0:9]\n",
    "    temp2 = list(temp1)\n",
    "    s_positif = np.append(s_positif, temp2)\n",
    "\n",
    "sequenceL = int(len(s_positif)/9)\n",
    "s_positif_new = np.expand_dims(s_positif, axis=0)\n",
    "s_positif_new = s_positif_new.reshape(sequenceL,9)\n",
    "labelpos = np.ones((sequenceL,), dtype=int)\n",
    "s_positif_label = np.expand_dims(labelpos, axis=0)\n",
    "s_positif_label = s_positif_label.reshape(sequenceL,1)\n",
    "print(s_positif_new.shape)\n",
    "print(s_positif_label.shape)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(493, 9, 20, 1)\n",
      "(394, 9)\n",
      "[ 5.  1.  3. 10. 18. 15.  7. 10.  2.]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/adminsvr/.venv/phos/lib/python3.5/site-packages/ipykernel_launcher.py:16: DeprecationWarning: Both axis > a.ndim and axis < -a.ndim - 1 are deprecated and will raise an AxisError in the future.\n",
      "  app.launch_new_instance()\n"
     ]
    }
   ],
   "source": [
    "training_X = np.concatenate((s_positif_new, s_negatif_new), axis=0, out=None)\n",
    "training_Y = np.concatenate((s_positif_label, s_negatif_label), axis=0, out=None)\n",
    "\n",
    "training_Y_onehot = to_categorical(training_Y)\n",
    "\n",
    "asam = ['A','R','N','D','C','Q','E','G','H','I','L','K','M','F','P','S','T','W','Y','V']\n",
    "tokenizer = Tokenizer(char_level=True)\n",
    "tokenizer.fit_on_texts(asam)\n",
    "sequence_of_int = []\n",
    "for i in range(len(training_X)):\n",
    "    temp = tokenizer.texts_to_sequences(training_X[i])\n",
    "    sequence_of_int = np.append(sequence_of_int, temp)\n",
    "\n",
    "sequence_of_int = sequence_of_int-1\n",
    "sequence_of_int_onehot = to_categorical(sequence_of_int)\n",
    "sequence_of_int_onehot = np.expand_dims(sequence_of_int_onehot, axis=3)\n",
    "sequence_of_int_onehot = sequence_of_int_onehot.reshape(len(training_X),9,20,1)\n",
    "sequence_of_int = sequence_of_int.reshape(len(training_X),9)\n",
    "\n",
    "print(sequence_of_int_onehot.shape)\n",
    "train_X, valid_X, train_label, valid_label = train_test_split(sequence_of_int, training_Y_onehot, test_size=0.2, random_state=13)\n",
    "print(train_X.shape)\n",
    "print(valid_X[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "#LSTM\n",
    "epochs = 100\n",
    "\n",
    "inp = Input(shape=(9,))\n",
    "emb = Embedding(20, 8, input_length=9)(inp)\n",
    "\n",
    "\n",
    "att = LSTM(9, activation = 'softmax')(emb)\n",
    "att = Reshape(target_shape=(9,1))(att)\n",
    "att = concatenate([att,att,att,att,att,att,att,att], axis=2)\n",
    "emb = multiply([emb, att])\n",
    "\n",
    "i = LSTM(20, return_sequences=True)(emb)\n",
    "# i = LSTM(20, return_sequences=True)(i)\n",
    "# i = LSTM(20, return_sequences=True)(i)\n",
    "# i = LSTM(20, return_sequences=True)(i)\n",
    "# i = LSTM(20, return_sequences=True)(i)\n",
    "i = LSTM(20, return_sequences=False)(i)\n",
    "\n",
    "out = Dense(2, activation='softmax')(i)\n",
    "\n",
    "model = Model(inputs=inp, outputs=out)\n",
    "model.compile(loss='categorical_crossentropy', optimizer='adam',metrics=['accuracy'])\n",
    "\n",
    "model.summary()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Bidirectional LSTM\n",
    "epochs = 100\n",
    "\n",
    "inp = Input(shape=(9,))\n",
    "emb = Embedding(20, 8, input_length=9)(inp)\n",
    "\n",
    "\n",
    "att = Bidirectional(LSTM(9, activation = 'softmax'), merge_mode='ave', weights=None)(emb)\n",
    "att = Reshape(target_shape=(9,1))(att)\n",
    "att = concatenate([att,att,att,att,att,att,att,att], axis=2)\n",
    "emb = multiply([emb, att])\n",
    "\n",
    "i = Bidirectional(LSTM(20), merge_mode='concat', weights=None)(emb)\n",
    "\n",
    "out = Dense(2, activation='softmax')(i)\n",
    "\n",
    "model = Model(inputs=inp, outputs=out)\n",
    "model.compile(loss='categorical_crossentropy', optimizer='adam',metrics=['accuracy'])\n",
    "\n",
    "model.summary()\n",
    "\n",
    "# model.add(Bidirectional(LSTM(128, kernel_regularizer=regularizers.l2(0.1), recurrent_regularizer=regularizers.l2(0.1), \n",
    "#                              bias_regularizer=regularizers.l2(0.1)), merge_mode='concat', weights=None))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Dense\n",
    "epochs = 300\n",
    "\n",
    "inp = Input(shape=(9,))\n",
    "\n",
    "emb = Embedding(20, 8, input_length=9)(inp)\n",
    "# pos_emb = Embedding(20, 8, trainable=False, weights=[GetPosEncodingMatrix(20, 8)], input_length=9)(inp)\n",
    "# emb = add([emb, pos_emb])\n",
    "\n",
    "att = Dense(1, activation = 'softmax')(emb)\n",
    "att = concatenate([att,att,att,att,att,att,att,att], axis=2)\n",
    "emb = multiply([emb, att])\n",
    "\n",
    "i = Dense(100)(emb)\n",
    "i = Dense(100)(i)\n",
    "i = Dense(100)(i)\n",
    "i = Dense(100)(i)\n",
    "i = Dense(100)(i)\n",
    "i = Dense(100)(i)\n",
    "i = Flatten()(i)\n",
    "\n",
    "out = Dense(2, activation='softmax')(i)\n",
    "\n",
    "model = Model(inputs=inp, outputs=out)\n",
    "model.compile(loss='categorical_crossentropy', optimizer='adam',metrics=['accuracy'])\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# library Transformer\n",
    "def GetPosEncodingMatrix(max_len, d_emb):\n",
    "    pos_enc = np.array([\n",
    "        [pos / np.power(10000, 2 * (j // 2) / d_emb) for j in range(d_emb)] \n",
    "        if pos != 0 else np.zeros(d_emb) \n",
    "            for pos in range(max_len)\n",
    "            ])\n",
    "    pos_enc[1:, 0::2] = np.sin(pos_enc[1:, 0::2]) # dim 2i\n",
    "    pos_enc[1:, 1::2] = np.cos(pos_enc[1:, 1::2]) # dim 2i+1\n",
    "    return pos_enc\n",
    "\n",
    "class LayerNormalization(Layer):\n",
    "\tdef __init__(self, eps=1e-6, **kwargs):\n",
    "\t\tself.eps = eps\n",
    "\t\tsuper(LayerNormalization, self).__init__(**kwargs)\n",
    "\tdef build(self, input_shape):\n",
    "\t\tself.gamma = self.add_weight(name='gamma', shape=input_shape[-1:],\n",
    "\t\t\t\t\t\t\t\t\t initializer=Ones(), trainable=True)\n",
    "\t\tself.beta = self.add_weight(name='beta', shape=input_shape[-1:],\n",
    "\t\t\t\t\t\t\t\t\tinitializer=Zeros(), trainable=True)\n",
    "\t\tsuper(LayerNormalization, self).build(input_shape)\n",
    "\tdef call(self, x):\n",
    "\t\tmean = K.mean(x, axis=-1, keepdims=True)\n",
    "\t\tstd = K.std(x, axis=-1, keepdims=True)\n",
    "\t\treturn self.gamma * (x - mean) / (std + self.eps) + self.beta\n",
    "\tdef compute_output_shape(self, input_shape):\n",
    "\t\treturn input_shape\n",
    "    \n",
    "class ScaledDotProductAttention():\n",
    "\tdef __init__(self, d_model, attn_dropout=0.1):\n",
    "\t\tself.temper = np.sqrt(d_model)\n",
    "\t\tself.dropout = Dropout(attn_dropout)\n",
    "\tdef __call__(self, q, k, v, mask):\n",
    "\t\tattn = Lambda(lambda x:K.batch_dot(x[0],x[1],axes=[2,2])/self.temper)([q, k])\n",
    "\t\tif mask is not None:\n",
    "\t\t\tmmask = Lambda(lambda x:(-1e+10)*(1-x))(mask)\n",
    "\t\t\tattn = Add()([attn, mmask])\n",
    "\t\tattn = Activation('softmax')(attn)\n",
    "\t\tattn = self.dropout(attn)\n",
    "\t\toutput = Lambda(lambda x:K.batch_dot(x[0], x[1]))([attn, v])\n",
    "\t\treturn output, attn\n",
    "    \n",
    "class MultiHeadAttention():\n",
    "\t# mode 0 - big martixes, faster; mode 1 - more clear implementation\n",
    "\tdef __init__(self, n_head, d_model, d_k, d_v, dropout, mode=0, use_norm=True):\n",
    "\t\tself.mode = mode\n",
    "\t\tself.n_head = n_head\n",
    "\t\tself.d_k = d_k\n",
    "\t\tself.d_v = d_v\n",
    "\t\tself.dropout = dropout\n",
    "\t\tif mode == 0:\n",
    "\t\t\tself.qs_layer = Dense(n_head*d_k, use_bias=False)\n",
    "\t\t\tself.ks_layer = Dense(n_head*d_k, use_bias=False)\n",
    "\t\t\tself.vs_layer = Dense(n_head*d_v, use_bias=False)\n",
    "\t\telif mode == 1:\n",
    "\t\t\tself.qs_layers = []\n",
    "\t\t\tself.ks_layers = []\n",
    "\t\t\tself.vs_layers = []\n",
    "\t\t\tfor _ in range(n_head):\n",
    "\t\t\t\tself.qs_layers.append(TimeDistributed(Dense(d_k, use_bias=False)))\n",
    "\t\t\t\tself.ks_layers.append(TimeDistributed(Dense(d_k, use_bias=False)))\n",
    "\t\t\t\tself.vs_layers.append(TimeDistributed(Dense(d_v, use_bias=False)))\n",
    "\t\tself.attention = ScaledDotProductAttention(d_model)\n",
    "\t\tself.layer_norm = LayerNormalization() if use_norm else None\n",
    "\t\tself.w_o = TimeDistributed(Dense(d_model))\n",
    "\n",
    "\tdef __call__(self, q, k, v, mask=None):\n",
    "\t\td_k, d_v = self.d_k, self.d_v\n",
    "\t\tn_head = self.n_head\n",
    "\n",
    "\t\tif self.mode == 0:\n",
    "\t\t\tqs = self.qs_layer(q)  # [batch_size, len_q, n_head*d_k]\n",
    "\t\t\tks = self.ks_layer(k)\n",
    "\t\t\tvs = self.vs_layer(v)\n",
    "\n",
    "\t\t\tdef reshape1(x):\n",
    "\t\t\t\ts = tf.shape(x)   # [batch_size, len_q, n_head * d_k]\n",
    "\t\t\t\tx = tf.reshape(x, [s[0], s[1], n_head, s[2]//n_head])\n",
    "\t\t\t\tx = tf.transpose(x, [2, 0, 1, 3])  \n",
    "\t\t\t\tx = tf.reshape(x, [-1, s[1], s[2]//n_head])  # [n_head * batch_size, len_q, d_k]\n",
    "\t\t\t\treturn x\n",
    "\t\t\tqs = Lambda(reshape1)(qs)\n",
    "\t\t\tks = Lambda(reshape1)(ks)\n",
    "\t\t\tvs = Lambda(reshape1)(vs)\n",
    "\n",
    "\t\t\tif mask is not None:\n",
    "\t\t\t\tmask = Lambda(lambda x:K.repeat_elements(x, n_head, 0))(mask)\n",
    "\t\t\thead, attn = self.attention(qs, ks, vs, mask=mask)  \n",
    "\t\t\t\t\n",
    "\t\t\tdef reshape2(x):\n",
    "\t\t\t\ts = tf.shape(x)   # [n_head * batch_size, len_v, d_v]\n",
    "\t\t\t\tx = tf.reshape(x, [n_head, -1, s[1], s[2]]) \n",
    "\t\t\t\tx = tf.transpose(x, [1, 2, 0, 3])\n",
    "\t\t\t\tx = tf.reshape(x, [-1, s[1], n_head*d_v])  # [batch_size, len_v, n_head * d_v]\n",
    "\t\t\t\treturn x\n",
    "\t\t\thead = Lambda(reshape2)(head)\n",
    "\t\telif self.mode == 1:\n",
    "\t\t\theads = []; attns = []\n",
    "\t\t\tfor i in range(n_head):\n",
    "\t\t\t\tqs = self.qs_layers[i](q)   \n",
    "\t\t\t\tks = self.ks_layers[i](k) \n",
    "\t\t\t\tvs = self.vs_layers[i](v) \n",
    "\t\t\t\thead, attn = self.attention(qs, ks, vs, mask)\n",
    "\t\t\t\theads.append(head); attns.append(attn)\n",
    "\t\t\thead = Concatenate()(heads) if n_head > 1 else heads[0]\n",
    "\t\t\tattn = Concatenate()(attns) if n_head > 1 else attns[0]\n",
    "\n",
    "\t\toutputs = self.w_o(head)\n",
    "\t\toutputs = Dropout(self.dropout)(outputs)\n",
    "\t\tif not self.layer_norm: return outputs, attn\n",
    "\t\toutputs = Add()([outputs, q])\n",
    "\t\treturn self.layer_norm(outputs), attn\n",
    "\n",
    "class PositionwiseFeedForward():\n",
    "    def __init__(self, d_hid, d_inner_hid, dropout=0.1):\n",
    "        self.w_1 = Conv1D(d_inner_hid, 1, activation='relu')\n",
    "        self.w_2 = Conv1D(d_hid, 1)\n",
    "        self.layer_norm = LayerNormalization()\n",
    "        self.dropout = Dropout(dropout)\n",
    "    def __call__(self, x):\n",
    "        output = self.w_1(x) \n",
    "        output = self.w_2(output)\n",
    "        output = self.dropout(output)\n",
    "        output = Add()([output, x])\n",
    "        output = self.layer_norm(output)\n",
    "        return output\n",
    "    \n",
    "class EncoderLayer():\n",
    "    def __init__(self, d_model, d_inner_hid, n_head, d_k, d_v, dropout=0.1):\n",
    "        self.self_att_layer = MultiHeadAttention(n_head, d_model, d_k, d_v, dropout=dropout)\n",
    "        self.pos_ffn_layer  = PositionwiseFeedForward(d_model, d_inner_hid, dropout=dropout)\n",
    "    def __call__(self, enc_input, mask=None):\n",
    "        output, slf_attn = self.self_att_layer(enc_input, enc_input, enc_input, mask=mask)\n",
    "        output = self.pos_ffn_layer(output)\n",
    "        return output, slf_attn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Transformer\n",
    "epochs = 300\n",
    "d_model = 16\n",
    "d_emb = d_model\n",
    "d_inner_hid = 2*d_model\n",
    "n_head = 4\n",
    "d_k = d_model//4\n",
    "d_v = d_model//4\n",
    "dropout = 0.5\n",
    "\n",
    "inp = Input(shape=(9,))\n",
    "\n",
    "emb = Embedding(20, d_emb, input_length=9)(inp)\n",
    "\n",
    "out, slf_attn = EncoderLayer(d_emb, d_inner_hid, n_head, d_k, d_v, dropout)(emb)\n",
    "out, slf_attn = EncoderLayer(d_emb, d_inner_hid, n_head, d_k, d_v, dropout)(out)\n",
    "out, slf_attn = EncoderLayer(d_emb, d_inner_hid, n_head, d_k, d_v, dropout)(out)\n",
    "out, slf_attn = EncoderLayer(d_emb, d_inner_hid, n_head, d_k, d_v, dropout)(out)\n",
    "out, slf_attn = EncoderLayer(d_emb, d_inner_hid, n_head, d_k, d_v, dropout)(out)\n",
    "out, slf_attn = EncoderLayer(d_emb, d_inner_hid, n_head, d_k, d_v, dropout)(out)\n",
    "out = GlobalAveragePooling1D()(out)\n",
    "\n",
    "out = Dense(2, activation='softmax')(out)\n",
    "\n",
    "model = Model(inputs=inp, outputs=out)\n",
    "model.compile(loss='categorical_crossentropy', optimizer='adam',metrics=['accuracy'])\n",
    "\n",
    "# model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 394 samples, validate on 99 samples\n",
      "Epoch 1/300\n",
      "394/394 [==============================] - 8s 21ms/step - loss: 0.7252 - acc: 0.5660 - val_loss: 0.7823 - val_acc: 0.5152\n",
      "Epoch 2/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.7254 - acc: 0.5254 - val_loss: 0.6471 - val_acc: 0.6263\n",
      "Epoch 3/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.6862 - acc: 0.5736 - val_loss: 0.6269 - val_acc: 0.6263\n",
      "Epoch 4/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.6605 - acc: 0.5635 - val_loss: 0.5133 - val_acc: 0.7677\n",
      "Epoch 5/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.6143 - acc: 0.6853 - val_loss: 0.6158 - val_acc: 0.7576\n",
      "Epoch 6/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.6074 - acc: 0.6878 - val_loss: 0.5848 - val_acc: 0.7677\n",
      "Epoch 7/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5671 - acc: 0.7107 - val_loss: 0.5212 - val_acc: 0.8081\n",
      "Epoch 8/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5321 - acc: 0.7310 - val_loss: 0.6022 - val_acc: 0.7172\n",
      "Epoch 9/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5244 - acc: 0.7462 - val_loss: 0.5688 - val_acc: 0.7576\n",
      "Epoch 10/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5366 - acc: 0.7462 - val_loss: 0.6310 - val_acc: 0.7475\n",
      "Epoch 11/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5351 - acc: 0.7462 - val_loss: 0.6514 - val_acc: 0.7071\n",
      "Epoch 12/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5531 - acc: 0.7157 - val_loss: 0.4761 - val_acc: 0.7778\n",
      "Epoch 13/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5277 - acc: 0.7462 - val_loss: 0.5717 - val_acc: 0.7778\n",
      "Epoch 14/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5219 - acc: 0.7513 - val_loss: 0.5610 - val_acc: 0.7677\n",
      "Epoch 15/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5335 - acc: 0.7411 - val_loss: 0.5797 - val_acc: 0.7677\n",
      "Epoch 16/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5079 - acc: 0.7462 - val_loss: 0.5155 - val_acc: 0.7879\n",
      "Epoch 17/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5201 - acc: 0.7487 - val_loss: 0.6140 - val_acc: 0.7374\n",
      "Epoch 18/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5065 - acc: 0.7563 - val_loss: 0.5218 - val_acc: 0.7778\n",
      "Epoch 19/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5203 - acc: 0.7513 - val_loss: 0.6319 - val_acc: 0.7475\n",
      "Epoch 20/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5017 - acc: 0.7640 - val_loss: 0.5459 - val_acc: 0.7677\n",
      "Epoch 21/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5070 - acc: 0.7538 - val_loss: 0.6100 - val_acc: 0.7475\n",
      "Epoch 22/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4933 - acc: 0.7817 - val_loss: 0.5504 - val_acc: 0.7576\n",
      "Epoch 23/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4964 - acc: 0.7513 - val_loss: 0.5836 - val_acc: 0.7576\n",
      "Epoch 24/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4870 - acc: 0.7513 - val_loss: 0.5740 - val_acc: 0.7677\n",
      "Epoch 25/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5113 - acc: 0.7487 - val_loss: 0.5890 - val_acc: 0.7576\n",
      "Epoch 26/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4969 - acc: 0.7538 - val_loss: 0.5396 - val_acc: 0.7677\n",
      "Epoch 27/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4856 - acc: 0.7563 - val_loss: 0.5385 - val_acc: 0.7778\n",
      "Epoch 28/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4954 - acc: 0.7817 - val_loss: 0.5556 - val_acc: 0.7677\n",
      "Epoch 29/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.5076 - acc: 0.7614 - val_loss: 0.5317 - val_acc: 0.7677\n",
      "Epoch 30/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4907 - acc: 0.7513 - val_loss: 0.5266 - val_acc: 0.7879\n",
      "Epoch 31/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4831 - acc: 0.7614 - val_loss: 0.5551 - val_acc: 0.7475\n",
      "Epoch 32/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4679 - acc: 0.7716 - val_loss: 0.5496 - val_acc: 0.7778\n",
      "Epoch 33/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.5017 - acc: 0.7665 - val_loss: 0.5300 - val_acc: 0.7778\n",
      "Epoch 34/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.5274 - acc: 0.7284 - val_loss: 0.6086 - val_acc: 0.7273\n",
      "Epoch 35/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4943 - acc: 0.7411 - val_loss: 0.5267 - val_acc: 0.7475\n",
      "Epoch 36/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5107 - acc: 0.7462 - val_loss: 0.6223 - val_acc: 0.7273\n",
      "Epoch 37/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5009 - acc: 0.7614 - val_loss: 0.5878 - val_acc: 0.7475\n",
      "Epoch 38/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5195 - acc: 0.7462 - val_loss: 0.5377 - val_acc: 0.7576\n",
      "Epoch 39/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4672 - acc: 0.7792 - val_loss: 0.5674 - val_acc: 0.7576\n",
      "Epoch 40/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4969 - acc: 0.7538 - val_loss: 0.5498 - val_acc: 0.7475\n",
      "Epoch 41/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4824 - acc: 0.7437 - val_loss: 0.6153 - val_acc: 0.7576\n",
      "Epoch 42/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4984 - acc: 0.7640 - val_loss: 0.5922 - val_acc: 0.7576\n",
      "Epoch 43/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5024 - acc: 0.7462 - val_loss: 0.5454 - val_acc: 0.7576\n",
      "Epoch 44/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4909 - acc: 0.7640 - val_loss: 0.5832 - val_acc: 0.7576\n",
      "Epoch 45/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4662 - acc: 0.7614 - val_loss: 0.5753 - val_acc: 0.7475\n",
      "Epoch 46/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4727 - acc: 0.7766 - val_loss: 0.6360 - val_acc: 0.7172\n",
      "Epoch 47/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5003 - acc: 0.7437 - val_loss: 0.5658 - val_acc: 0.7475\n",
      "Epoch 48/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5098 - acc: 0.7284 - val_loss: 0.6142 - val_acc: 0.7475\n",
      "Epoch 49/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4847 - acc: 0.7817 - val_loss: 0.5176 - val_acc: 0.7576\n",
      "Epoch 50/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4737 - acc: 0.7614 - val_loss: 0.5335 - val_acc: 0.7475\n",
      "Epoch 51/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4672 - acc: 0.7792 - val_loss: 0.5935 - val_acc: 0.7475\n",
      "Epoch 52/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4704 - acc: 0.7766 - val_loss: 0.5778 - val_acc: 0.7475\n",
      "Epoch 53/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4824 - acc: 0.7665 - val_loss: 0.5616 - val_acc: 0.7475\n",
      "Epoch 54/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4694 - acc: 0.7640 - val_loss: 0.5679 - val_acc: 0.7778\n",
      "Epoch 55/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4807 - acc: 0.7665 - val_loss: 0.5962 - val_acc: 0.7374\n",
      "Epoch 56/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4766 - acc: 0.7716 - val_loss: 0.5886 - val_acc: 0.7677\n",
      "Epoch 57/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4757 - acc: 0.7614 - val_loss: 0.5435 - val_acc: 0.7879\n",
      "Epoch 58/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4700 - acc: 0.7589 - val_loss: 0.5689 - val_acc: 0.7677\n",
      "Epoch 59/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4576 - acc: 0.7919 - val_loss: 0.6063 - val_acc: 0.7576\n",
      "Epoch 60/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4662 - acc: 0.7893 - val_loss: 0.5518 - val_acc: 0.7677\n",
      "Epoch 61/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4858 - acc: 0.7665 - val_loss: 0.5510 - val_acc: 0.7677\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 62/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4816 - acc: 0.7792 - val_loss: 0.5853 - val_acc: 0.7374\n",
      "Epoch 63/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4913 - acc: 0.7690 - val_loss: 0.5301 - val_acc: 0.7576\n",
      "Epoch 64/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4682 - acc: 0.7665 - val_loss: 0.5333 - val_acc: 0.7576\n",
      "Epoch 65/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4777 - acc: 0.8046 - val_loss: 0.5499 - val_acc: 0.7778\n",
      "Epoch 66/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4677 - acc: 0.7792 - val_loss: 0.5683 - val_acc: 0.7677\n",
      "Epoch 67/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4433 - acc: 0.7843 - val_loss: 0.5916 - val_acc: 0.7677\n",
      "Epoch 68/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4775 - acc: 0.7563 - val_loss: 0.5478 - val_acc: 0.7677\n",
      "Epoch 69/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4630 - acc: 0.7792 - val_loss: 0.5419 - val_acc: 0.7778\n",
      "Epoch 70/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4414 - acc: 0.7843 - val_loss: 0.5873 - val_acc: 0.7677\n",
      "Epoch 71/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4527 - acc: 0.7766 - val_loss: 0.5571 - val_acc: 0.7677\n",
      "Epoch 72/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4353 - acc: 0.7893 - val_loss: 0.6291 - val_acc: 0.7677\n",
      "Epoch 73/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4593 - acc: 0.7843 - val_loss: 0.5663 - val_acc: 0.7778\n",
      "Epoch 74/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4448 - acc: 0.7995 - val_loss: 0.6972 - val_acc: 0.7374\n",
      "Epoch 75/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5212 - acc: 0.7157 - val_loss: 0.5587 - val_acc: 0.7778\n",
      "Epoch 76/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4461 - acc: 0.8096 - val_loss: 0.6484 - val_acc: 0.7576\n",
      "Epoch 77/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.5015 - acc: 0.7640 - val_loss: 0.6489 - val_acc: 0.7374\n",
      "Epoch 78/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4567 - acc: 0.7792 - val_loss: 0.6135 - val_acc: 0.7677\n",
      "Epoch 79/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4524 - acc: 0.7868 - val_loss: 0.6005 - val_acc: 0.7980\n",
      "Epoch 80/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4641 - acc: 0.7817 - val_loss: 0.6085 - val_acc: 0.7879\n",
      "Epoch 81/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4631 - acc: 0.7792 - val_loss: 0.6334 - val_acc: 0.7677\n",
      "Epoch 82/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4543 - acc: 0.7817 - val_loss: 0.5688 - val_acc: 0.7879\n",
      "Epoch 83/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4380 - acc: 0.7766 - val_loss: 0.6446 - val_acc: 0.7374\n",
      "Epoch 84/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4546 - acc: 0.7716 - val_loss: 0.5644 - val_acc: 0.7879\n",
      "Epoch 85/300\n",
      "394/394 [==============================] - 1s 3ms/step - loss: 0.4594 - acc: 0.7690 - val_loss: 0.5924 - val_acc: 0.7576\n",
      "Epoch 86/300\n",
      "394/394 [==============================] - 1s 3ms/step - loss: 0.4293 - acc: 0.7665 - val_loss: 0.6198 - val_acc: 0.7677\n",
      "Epoch 87/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4303 - acc: 0.7919 - val_loss: 0.6375 - val_acc: 0.7576\n",
      "Epoch 88/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4435 - acc: 0.7589 - val_loss: 0.5902 - val_acc: 0.7980\n",
      "Epoch 89/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4370 - acc: 0.7970 - val_loss: 0.6769 - val_acc: 0.7475\n",
      "Epoch 90/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4241 - acc: 0.7893 - val_loss: 0.6217 - val_acc: 0.7677\n",
      "Epoch 91/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4584 - acc: 0.7716 - val_loss: 0.6266 - val_acc: 0.7576\n",
      "Epoch 92/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4276 - acc: 0.8020 - val_loss: 0.6105 - val_acc: 0.7778\n",
      "Epoch 93/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4447 - acc: 0.7919 - val_loss: 0.6213 - val_acc: 0.7576\n",
      "Epoch 94/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4351 - acc: 0.8020 - val_loss: 0.6231 - val_acc: 0.7576\n",
      "Epoch 95/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4608 - acc: 0.7868 - val_loss: 0.5964 - val_acc: 0.8081\n",
      "Epoch 96/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4293 - acc: 0.7944 - val_loss: 0.6423 - val_acc: 0.7576\n",
      "Epoch 97/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4383 - acc: 0.7868 - val_loss: 0.6580 - val_acc: 0.7475\n",
      "Epoch 98/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4509 - acc: 0.7716 - val_loss: 0.5883 - val_acc: 0.7980\n",
      "Epoch 99/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4572 - acc: 0.7792 - val_loss: 0.6437 - val_acc: 0.7576\n",
      "Epoch 100/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4249 - acc: 0.7868 - val_loss: 0.6226 - val_acc: 0.7576\n",
      "Epoch 101/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4283 - acc: 0.7817 - val_loss: 0.6334 - val_acc: 0.7677\n",
      "Epoch 102/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4522 - acc: 0.7766 - val_loss: 0.6482 - val_acc: 0.7576\n",
      "Epoch 103/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4126 - acc: 0.7792 - val_loss: 0.6219 - val_acc: 0.7677\n",
      "Epoch 104/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4324 - acc: 0.7614 - val_loss: 0.6159 - val_acc: 0.7576\n",
      "Epoch 105/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4233 - acc: 0.7868 - val_loss: 0.6514 - val_acc: 0.7273\n",
      "Epoch 106/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4259 - acc: 0.7970 - val_loss: 0.6309 - val_acc: 0.7778\n",
      "Epoch 107/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4413 - acc: 0.7944 - val_loss: 0.6527 - val_acc: 0.7273\n",
      "Epoch 108/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4335 - acc: 0.7868 - val_loss: 0.6375 - val_acc: 0.7576\n",
      "Epoch 109/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4415 - acc: 0.7792 - val_loss: 0.6111 - val_acc: 0.7980\n",
      "Epoch 110/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4216 - acc: 0.7868 - val_loss: 0.7171 - val_acc: 0.7273\n",
      "Epoch 111/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4285 - acc: 0.7893 - val_loss: 0.6274 - val_acc: 0.7980\n",
      "Epoch 112/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4356 - acc: 0.7868 - val_loss: 0.6688 - val_acc: 0.7475\n",
      "Epoch 113/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4460 - acc: 0.7919 - val_loss: 0.7276 - val_acc: 0.7172\n",
      "Epoch 114/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4592 - acc: 0.7944 - val_loss: 0.6230 - val_acc: 0.7475\n",
      "Epoch 115/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4397 - acc: 0.7868 - val_loss: 0.6653 - val_acc: 0.7475\n",
      "Epoch 116/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4302 - acc: 0.7766 - val_loss: 0.6454 - val_acc: 0.7576\n",
      "Epoch 117/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4076 - acc: 0.7995 - val_loss: 0.6776 - val_acc: 0.7677\n",
      "Epoch 118/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4284 - acc: 0.7919 - val_loss: 0.6810 - val_acc: 0.7273\n",
      "Epoch 119/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4311 - acc: 0.7817 - val_loss: 0.6240 - val_acc: 0.7879\n",
      "Epoch 120/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4452 - acc: 0.7970 - val_loss: 0.6439 - val_acc: 0.7677\n",
      "Epoch 121/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4246 - acc: 0.7792 - val_loss: 0.6792 - val_acc: 0.7576\n",
      "Epoch 122/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4453 - acc: 0.7919 - val_loss: 0.5813 - val_acc: 0.7879\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 123/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4473 - acc: 0.7640 - val_loss: 0.6972 - val_acc: 0.7273\n",
      "Epoch 124/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4282 - acc: 0.7944 - val_loss: 0.6288 - val_acc: 0.7778\n",
      "Epoch 125/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4311 - acc: 0.7817 - val_loss: 0.7368 - val_acc: 0.7172\n",
      "Epoch 126/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4159 - acc: 0.8020 - val_loss: 0.6781 - val_acc: 0.7475\n",
      "Epoch 127/300\n",
      "394/394 [==============================] - 1s 1ms/step - loss: 0.4128 - acc: 0.8020 - val_loss: 0.6928 - val_acc: 0.7374\n",
      "Epoch 128/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4223 - acc: 0.7995 - val_loss: 0.6739 - val_acc: 0.7172\n",
      "Epoch 129/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4305 - acc: 0.8046 - val_loss: 0.6711 - val_acc: 0.7576\n",
      "Epoch 130/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4269 - acc: 0.7843 - val_loss: 0.6571 - val_acc: 0.7677\n",
      "Epoch 131/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4189 - acc: 0.7843 - val_loss: 0.6477 - val_acc: 0.7374\n",
      "Epoch 132/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4270 - acc: 0.8046 - val_loss: 0.6949 - val_acc: 0.7273\n",
      "Epoch 133/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4427 - acc: 0.7792 - val_loss: 0.6506 - val_acc: 0.7576\n",
      "Epoch 134/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4169 - acc: 0.7817 - val_loss: 0.6768 - val_acc: 0.7374\n",
      "Epoch 135/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4281 - acc: 0.7741 - val_loss: 0.6472 - val_acc: 0.7374\n",
      "Epoch 136/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4170 - acc: 0.7843 - val_loss: 0.6692 - val_acc: 0.7374\n",
      "Epoch 137/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4138 - acc: 0.7995 - val_loss: 0.6918 - val_acc: 0.7273\n",
      "Epoch 138/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4183 - acc: 0.7944 - val_loss: 0.6638 - val_acc: 0.7273\n",
      "Epoch 139/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4205 - acc: 0.7919 - val_loss: 0.6476 - val_acc: 0.7677\n",
      "Epoch 140/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4126 - acc: 0.7868 - val_loss: 0.7594 - val_acc: 0.7172\n",
      "Epoch 141/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4221 - acc: 0.7817 - val_loss: 0.6219 - val_acc: 0.7778\n",
      "Epoch 142/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4107 - acc: 0.8223 - val_loss: 0.6511 - val_acc: 0.7576\n",
      "Epoch 143/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4155 - acc: 0.7995 - val_loss: 0.6469 - val_acc: 0.7576\n",
      "Epoch 144/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4152 - acc: 0.8046 - val_loss: 0.6289 - val_acc: 0.7677\n",
      "Epoch 145/300\n",
      "394/394 [==============================] - 1s 3ms/step - loss: 0.4291 - acc: 0.7843 - val_loss: 0.6431 - val_acc: 0.7475\n",
      "Epoch 146/300\n",
      "394/394 [==============================] - 1s 3ms/step - loss: 0.4354 - acc: 0.7843 - val_loss: 0.7097 - val_acc: 0.7374\n",
      "Epoch 147/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4068 - acc: 0.7970 - val_loss: 0.6796 - val_acc: 0.7677\n",
      "Epoch 148/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4112 - acc: 0.7919 - val_loss: 0.7269 - val_acc: 0.7475\n",
      "Epoch 149/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4216 - acc: 0.8096 - val_loss: 0.7103 - val_acc: 0.7576\n",
      "Epoch 150/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4296 - acc: 0.8046 - val_loss: 0.6758 - val_acc: 0.7475\n",
      "Epoch 151/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4405 - acc: 0.7970 - val_loss: 0.7075 - val_acc: 0.7475\n",
      "Epoch 152/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4348 - acc: 0.7893 - val_loss: 0.6265 - val_acc: 0.7374\n",
      "Epoch 153/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4139 - acc: 0.8020 - val_loss: 0.6943 - val_acc: 0.7273\n",
      "Epoch 154/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4242 - acc: 0.7868 - val_loss: 0.6837 - val_acc: 0.7172\n",
      "Epoch 155/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4008 - acc: 0.7995 - val_loss: 0.6709 - val_acc: 0.7273\n",
      "Epoch 156/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4080 - acc: 0.8046 - val_loss: 0.6762 - val_acc: 0.7475\n",
      "Epoch 157/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.3974 - acc: 0.8096 - val_loss: 0.6613 - val_acc: 0.7576\n",
      "Epoch 158/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4031 - acc: 0.8071 - val_loss: 0.6705 - val_acc: 0.7475\n",
      "Epoch 159/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4177 - acc: 0.7970 - val_loss: 0.6399 - val_acc: 0.7778\n",
      "Epoch 160/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.3946 - acc: 0.8071 - val_loss: 0.7269 - val_acc: 0.7374\n",
      "Epoch 161/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.3871 - acc: 0.8249 - val_loss: 0.6889 - val_acc: 0.7576\n",
      "Epoch 162/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4154 - acc: 0.7919 - val_loss: 0.7325 - val_acc: 0.7374\n",
      "Epoch 163/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4216 - acc: 0.7741 - val_loss: 0.6836 - val_acc: 0.7273\n",
      "Epoch 164/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4139 - acc: 0.7843 - val_loss: 0.6692 - val_acc: 0.7677\n",
      "Epoch 165/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4188 - acc: 0.7919 - val_loss: 0.7070 - val_acc: 0.7475\n",
      "Epoch 166/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4267 - acc: 0.7919 - val_loss: 0.6635 - val_acc: 0.7677\n",
      "Epoch 167/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4054 - acc: 0.8096 - val_loss: 0.7247 - val_acc: 0.7273\n",
      "Epoch 168/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4006 - acc: 0.8020 - val_loss: 0.6369 - val_acc: 0.7576\n",
      "Epoch 169/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.3982 - acc: 0.8173 - val_loss: 0.6900 - val_acc: 0.7475\n",
      "Epoch 170/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4020 - acc: 0.7944 - val_loss: 0.6566 - val_acc: 0.7879\n",
      "Epoch 171/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4195 - acc: 0.7919 - val_loss: 0.6632 - val_acc: 0.7778\n",
      "Epoch 172/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4146 - acc: 0.7893 - val_loss: 0.6376 - val_acc: 0.7879\n",
      "Epoch 173/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4117 - acc: 0.7893 - val_loss: 0.6642 - val_acc: 0.7778\n",
      "Epoch 174/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4280 - acc: 0.7970 - val_loss: 0.6895 - val_acc: 0.7475\n",
      "Epoch 175/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4131 - acc: 0.8020 - val_loss: 0.7277 - val_acc: 0.7677\n",
      "Epoch 176/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.3979 - acc: 0.8096 - val_loss: 0.7124 - val_acc: 0.7475\n",
      "Epoch 177/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4181 - acc: 0.7995 - val_loss: 0.7216 - val_acc: 0.7576\n",
      "Epoch 178/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4204 - acc: 0.7817 - val_loss: 0.7231 - val_acc: 0.7475\n",
      "Epoch 179/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4190 - acc: 0.8020 - val_loss: 0.7347 - val_acc: 0.7576\n",
      "Epoch 180/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.3861 - acc: 0.8046 - val_loss: 0.6587 - val_acc: 0.7677\n",
      "Epoch 181/300\n",
      "394/394 [==============================] - 1s 2ms/step - loss: 0.4061 - acc: 0.7893 - val_loss: 0.7307 - val_acc: 0.7576\n",
      "Epoch 182/300\n",
      " 32/394 [=>............................] - ETA: 0s - loss: 0.3825 - acc: 0.7812"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-11-761dec573f45>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mmodel_train\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrain_X\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtrain_label\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mepochs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mepochs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbatch_size\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m32\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mvalidation_data\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mvalid_X\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mvalid_label\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m~/.venv/phos/lib/python3.5/site-packages/keras/engine/training.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, **kwargs)\u001b[0m\n\u001b[1;32m   1037\u001b[0m                                         \u001b[0minitial_epoch\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0minitial_epoch\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1038\u001b[0m                                         \u001b[0msteps_per_epoch\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0msteps_per_epoch\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1039\u001b[0;31m                                         validation_steps=validation_steps)\n\u001b[0m\u001b[1;32m   1040\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1041\u001b[0m     def evaluate(self, x=None, y=None,\n",
      "\u001b[0;32m~/.venv/phos/lib/python3.5/site-packages/keras/engine/training_arrays.py\u001b[0m in \u001b[0;36mfit_loop\u001b[0;34m(model, f, ins, out_labels, batch_size, epochs, verbose, callbacks, val_f, val_ins, shuffle, callback_metrics, initial_epoch, steps_per_epoch, validation_steps)\u001b[0m\n\u001b[1;32m    197\u001b[0m                     \u001b[0mins_batch\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mins_batch\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtoarray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    198\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 199\u001b[0;31m                 \u001b[0mouts\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mf\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mins_batch\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    200\u001b[0m                 \u001b[0mouts\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mto_list\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mouts\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    201\u001b[0m                 \u001b[0;32mfor\u001b[0m \u001b[0ml\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mo\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mzip\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mout_labels\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mouts\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/.venv/phos/lib/python3.5/site-packages/keras/backend/tensorflow_backend.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, inputs)\u001b[0m\n\u001b[1;32m   2713\u001b[0m                 \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_legacy_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0minputs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2714\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2715\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0minputs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   2716\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2717\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mpy_any\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mis_tensor\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[0;32min\u001b[0m \u001b[0minputs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/.venv/phos/lib/python3.5/site-packages/keras/backend/tensorflow_backend.py\u001b[0m in \u001b[0;36m_call\u001b[0;34m(self, inputs)\u001b[0m\n\u001b[1;32m   2673\u001b[0m             \u001b[0mfetched\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_callable_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0marray_vals\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mrun_metadata\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrun_metadata\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2674\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2675\u001b[0;31m             \u001b[0mfetched\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_callable_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0marray_vals\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   2676\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mfetched\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2677\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/.venv/phos/lib/python3.5/site-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1437\u001b[0m           ret = tf_session.TF_SessionRunCallable(\n\u001b[1;32m   1438\u001b[0m               \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_session\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_session\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_handle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mstatus\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1439\u001b[0;31m               run_metadata_ptr)\n\u001b[0m\u001b[1;32m   1440\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mrun_metadata\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1441\u001b[0m           \u001b[0mproto_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtf_session\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTF_GetBuffer\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrun_metadata_ptr\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "model_train = model.fit(train_X, train_label, epochs=epochs, batch_size=32, validation_data=(valid_X, valid_label))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAEICAYAAACktLTqAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJztvXl8VdXV///ZmSAJCUNAxQAJk2CAgAxBC4qgOLWOtRUFBZVacaitrT5afbT6LfXpz1pL+9Ba9XGCqEVbKbVY51nLIOIAFEEGAVEgwRAShoSs3x/r7px9zz3n3HNv7nTCer9e93XvPePeZ/icddZee21FRBAEQRDaF1npLoAgCIKQeETcBUEQ2iEi7oIgCO0QEXdBEIR2iIi7IAhCO0TEXRAEoR0i4t6OUUplK6X2KqX6JHLZdKKUGqCUSnj8rlLqVKXUJuP/WqXUiX6WjWNfDyulfh7v+oLgh5x0F0CwUErtNf4WADgA4FDo/w+JqDqW7RHRIQCdEr3s4QARDUrEdpRSMwFMI6KTjW3PTMS2BcELEfcMgohaxTVkGc4kolfclldK5RBRcyrKJgjRkOsxsxC3TIBQSv1SKfUXpdRTSql6ANOUUicopf6tlPpGKbVdKfV7pVRuaPkcpRQppcpD/+eH5r+glKpXSr2vlOob67Kh+WcqpT5TStUppf6glHpXKTXDpdx+yvhDpdR6pdRupdTvjXWzlVL3K6VqlFIbAJzhcXxuU0o9bZs2Vyn129DvmUqpNaH6fB6yqt22tVUpdXLod4FSal6obKsAjLIte7tSakNou6uUUueEpg8D8L8ATgy5vHYZx/YXxvpXh+peo5RaqJTq6efYxHKcdXmUUq8opWqVUl8ppW429vPfoWOyRym1XCl1tJMLTCn1jj7PoeP5Vmg/tQBuV0oNVEq9HtrHrtBx62ysXxaq487Q/DlKqY6hMh9rLNdTKdWolCpxq68QBSKSTwZ+AGwCcKpt2i8BHARwNvjBnA9gDICx4LewfgA+A3BdaPkcAASgPPR/PoBdAEYDyAXwFwDz41j2CAD1AM4NzbsRQBOAGS518VPGvwPoDKAcQK2uO4DrAKwC0AtACYC3+LJ13E8/AHsBFBrb3gFgdOj/2aFlFIBJAPYBqAzNOxXAJmNbWwGcHPr9GwBvAOgKoAzAatuy3wfQM3ROLgmV4cjQvJkA3rCVcz6AX4R+nxYq4wgAHQH8EcBrfo5NjMe5M4CvAdwAoAOAYgBVoXm3AvgIwMBQHUYA6AZggP1YA3hHn+dQ3ZoBzAKQDb4ejwFwCoC80HXyLoDfGPX5NHQ8C0PLjwvNexDAbGM/PwXwXLrvwyB/0l4A+bicGHdxfy3Kej8D8Ezot5NgP2Asew6AT+NY9goAbxvzFIDtcBF3n2U83pj/NwA/C/1+C+ye0vPOsguObdv/BnBJ6PeZANZ6LPs8gGtDv73E/QvzXAC4xlzWYbufAvh26Hc0cX8cwK+MecXgdpZe0Y5NjMf5UgDLXJb7XJfXNt2PuG+IUoYL9X4BnAjgKwDZDsuNA7ARgAr9XwnggkTfV4fTR9wywWOL+UcpNVgp9c/Qa/YeAHcD6O6x/lfG70Z4N6K6LXu0WQ7iu3Gr20Z8ltHXvgBs9igvADwJ4OLQ70tC/3U5vqOUWhJyGXwDtpq9jpWmp1cZlFIzlFIfhVwL3wAY7HO7ANevdXtEtAfAbgClxjK+zlmU49wbLOJOeM2Lhv16PEoptUAptS1UhsdsZdhE3HgfBhG9C34LGK+UGgqgD4B/xlkmAeJzDyL2MMA/gy3FAURUDOAOsCWdTLaDLUsAgFJKIVyM7LSljNvBoqCJFqq5AMCpSqlSsNvoyVAZ8wE8C+AesMukC4CXfJbjK7cyKKX6AfgT2DVREtruf4ztRgvb/BLs6tHbKwK7f7b5KJcdr+O8BUB/l/Xc5jWEylRgTDvKtoy9fr8GR3kNC5Vhhq0MZUqpbJdyPAFgGvgtYwERHXBZTvCBiHvwKQJQB6Ah1CD1wxTs83kAI5VSZyulcsB+3B5JKuMCAD9WSpWGGtf+y2thIvoK7Dp4DOySWRea1QHsB94J4JBS6jtg37DfMvxcKdVFcT+A64x5ncACtxP8nPsB2HLXfA2gl9mwaeMpAFcqpSqVUh3AD5+3icj1TcgDr+O8CEAfpdR1SqkOSqlipVRVaN7DAH6plOqvmBFKqW7gh9pX4Ib7bKXUVTAeRB5laABQp5TqDXYNad4HUAPgV4obqfOVUuOM+fPAbpxLwEIvtAER9+DzUwDTwQ2cfwY3fCYVIvoawEUAfgu+WfsD+BBssSW6jH8C8CqATwAsA1vf0XgS7ENvdckQ0TcAfgLgOXCj5IXgh5Qf7gS/QWwC8AIM4SGijwH8AcDS0DKDACwx1n0ZwDoAXyulTPeKXv9fYPfJc6H1+wCY6rNcdlyPMxHVAZgM4LvgB85nACaEZt8LYCH4OO8BN252DLnbfgDg5+DG9QG2ujlxJ4Aq8ENmEYC/GmVoBvAdAMeCrfgvwOdBz98EPs8HiOi9GOsu2NCNF4IQN6HX7C8BXEhEb6e7PEJwUUo9AW6k/UW6yxJ0pBOTEBdKqTPAkSn7wKF0TWDrVRDiItR+cS6AYekuS3tA3DJCvIwHsAHsaz4dwPnSACbEi1LqHnCs/a+I6It0l6c9IG4ZQRCEdohY7oIgCO2QtPncu3fvTuXl5enavSAIQiD54IMPdhGRV+gxgDSKe3l5OZYvX56u3QuCIAQSpVS0XtoAxC0jCILQLhFxFwRBaIeIuAuCILRDMqoTU1NTE7Zu3Yr9+/enuyhCBtGxY0f06tULublu6VkEQbCTUeK+detWFBUVoby8HJxoUDjcISLU1NRg69at6Nu3b/QVBEEAkGFumf3796OkpESEXWhFKYWSkhJ5mxNaqa4GysuBrCz+ro5p2Pj2Xx5NRok7ABF2IQK5JgRNdTVw1VXA5s0AEX9fdZU/QU2GCDuV5/LLge7d0y/2GSfugiAIbtx2G9DYGD6tsZGne9GWh0Ks5WlqAmpqErufeBBxN6ipqcGIESMwYsQIHHXUUSgtLW39f/DgQV/buPzyy7F27VrPZebOnYvqTHl3E4QA8YVLSrEvvvC2zON9KLih97XZR3eituynTaRr8NZRo0aRndWrV0dM82L+fKKyMiKl+Hv+/JhW9+TOO++ke++9N2J6S0sLHTp0KHE7CghNTU1p3X+s14aQepJ5P2rKyojYJg7/lJQQFRSETysosMqglPN6SrnXAyDKzuZvXZ/583lfTtvy80nEcQGwnNrzANnJes1yYv369aioqMDUqVMxZMgQbN++HVdddRVGjx6NIUOG4O67725ddvz48Vi5ciWam5vRpUsX3HLLLRg+fDhOOOEE7NixAwBw++2343e/+13r8rfccguqqqowaNAgvPceD0DT0NCA7373u6ioqMCFF16I0aNHY+XKlRFlu/POOzFmzBgMHToUV199NSiU5fOzzz7DpEmTMHz4cIwcORKbNm0CAPzqV7/CsGHDMHz4cNwWMid0mQHgq6++woABAwAADz/8MM477zxMnDgRp59+Ovbs2YNJkyZh5MiRqKysxPPPWwMZPfroo6isrMTw4cNx+eWXo66uDv369UNzczMAYPfu3WH/hfZFqu7H2bOBgoLwafq/l2Xex2Xk3W7drN/V1ewrnzbNssgPhYby1r70K65gl0u8bN4MXHopcM018W/DN36eAMn4tNVyd3uCl5X53oQnpuW+bt06UkrRsmXLWufX1NQQEVu048ePp1WrVhER0bhx4+jDDz+kpqYmAkCLFy8mIqKf/OQndM899xAR0W233Ub3339/6/I333wzERH9/e9/p9NPP52IiO655x665ppriIho5cqVlJWVRR9++GFEOXU5WlpaaMqUKa37GzlyJC1atIiIiPbt20cNDQ20aNEiGj9+PDU2Noatq8tMRLR9+3bq378/ERE99NBD1KdPH6qtrSUiooMHD1JdXR0REX399dc0YMCA1vINGjSodXv6e9q0afSPf/yDiIjmzp3bWs94EMs9s0nm/Wi3lgsL+b/5hhDNMp8/nyg3N3J+Xp5lkdst/2R+lIrfgkd7t9y9fG/JoH///hg9enTr/6eeegojR47EyJEjsWbNGqxevTpinfz8fJx55pkAgFGjRrVaz3YuuOCCiGXeeecdTJkyBQAwfPhwDBkyxHHdV199FVVVVRg+fDjefPNNrFq1Crt378auXbtw9tlnA+BOQAUFBXjllVdwxRVXID8/HwDQzTRbXDjttNPQtWtXAGwI3HLLLaisrMRpp52GLVu2YNeuXXjttddw0UUXtW5Pf8+cOROPPvooALbsL7/88qj7E4KJ2323eXO49R5rxEp1NVvMprXc0MD/u3VjS37qVHfLvE8f3sZtt3FDp52DB4Hp09lat1v+yYSI95vMpreM6sQUC336ODdmuJ3ktlJYWNj6e926dZgzZw6WLl2KLl26YNq0aY5x2Hl5ea2/s7OzXV0SHTp0iLqME42NjbjuuuuwYsUKlJaW4vbbb48rHjwnJwctLS0AELG+We8nnngCdXV1WLFiBXJyctCrVy/P/U2YMAHXXXcdXn/9deTm5mLw4MExl00IBm73I8DuGfO3FlHtugFYoJ1wE2WABV6vP3t2+LYBdtecdVbkdDva9ZJqDh2KXv+2EFjL3c33Nnt28ve9Z88eFBUVobi4GNu3b8eLL76Y8H2MGzcOCxYsAAB88sknjm8G+/btQ1ZWFrp37476+nr89a880HzXrl3Ro0cP/OMf/wDAgt3Y2IjJkyfjkUcewb59+wAAtbW1ADj98gcffAAAePbZZ13LVFdXhyOOOAI5OTl4+eWXsW3bNgDApEmT8Je//KV1e/obAKZNm4apU6eK1d7OcbofNdr37RaxMm2auxUf7U1cb3vqVODBB4GyMkAp/n7wQWDx4tRa5LGSzEiawIq728lMxhPQzsiRI1FRUYHBgwfjsssuw7hx4xK+j+uvvx7btm1DRUUF7rrrLlRUVKBz585hy5SUlGD69OmoqKjAmWeeibFjx7bOq66uxn333YfKykqMHz8eO3fuxHe+8x2cccYZGD16NEaMGIH7778fAHDTTTdhzpw5GDlyJHbv3u1apksvvRTvvfcehg0bhqeffhoDBw4EwG6jm2++GSeddBJGjBiBm266qXWdqVOnoq6uDhdddFEiD4+QZGJ1n+j70Y0vvvAWarcGWD9v4ps3c0PoDTfwPvr0sdw1yXLTOqH72pWVAbNmASUl/tZLWhn9OOaT8UlEKGR7pqmpifbt20dERJ999hmVl5enPRwxHp566imaMWNGm7cj10bqcGpcNMMKvfBqWHWb59QAa4YjxvPR5W3LNuL5lJQ4H5dZs6LX2S/w2aAq4p6h7N69m0aOHEmVlZU0bNgwevHFF9NdpJi5+uqracCAAbR+/fo2b0uujdTRlsgXpwdDbm7bYsPj/WRns6i6RdIk62M+BKM9YOKJmvEr7oqXTT2jR48m+zB7a9aswbHHHpuW8giZjVwbqSMri6XHjlJAqN3dk+pqdpG0JR7cCaWcy+VFQUHqfe7Z2XycunUD6us5IseLWOuklPqAiEZHWy6wPndBEJKDV1ihX0Jt9gklHju0sZHFNpUcOsRlramJLuxlZckrh4i7IBzm2BtPzzorMvJFKW649NO46hQVk04OHXKP5EknyY7uE3EXhMMYp7QBjz/OHWzMaA9tNW/ezKGL3bu7i3wqI1Sys/nBU1LCDycndCRdLFay0UUlKaQiuk/EXRDaKX7CGW+4wTn2fMECb9dKTY17jpRkdSS0U1DAD6Y+fYDaWqBr10hRzs0F9u7lsgL+wxOLivw/DHQIpB/3T0EBMH8+sGlT8sO2RdwNJk6cGNEh6Xe/+x1mzZrluV6nTp0AAF9++SUuvPBCx2VOPvlk2BuQ7fzud79Do3GnnXXWWfjmm2/8FF0QWhNfKcUfnQBLW+T2OPLqavdGz5qa6K4VIuCBByIfGl4dmrR13dbxV8rK+O3i8cetOuoc6iUlljWvVHhu9T17/FnltbXO9bALeVkZMG8eb//xxyOXz821ypHKvjgAJBTS5M9//nNETPbYsWPpzTff9FyvsLAw6rYnTJgQlnjMibKyMtq5c2f0gmYoyUyHnO5rI5OZP5+TafkJ0zPDGRMVA15SEpnq1y01ro4/b0vqXB0+qNPxOpXHq35med22YY+395vGOBVpjyFx7rFTU1NDPXr0oAMHDhAR0caNG6l3797U0tJC9fX1NGnSJDruuONo6NChtHDhwtb1tLhv3LiRhgwZQkREjY2NdNFFF9HgwYPpvPPOo6qqqlZxv/rqq2nUqFFUUVFBd9xxBxERzZkzh3Jzc2no0KF08sknE1G42N933300ZMgQGjJkSGtGyY0bN9LgwYNp5syZVFFRQZMnT27N+GiyaNEiqqqqohEjRtApp5xCX331FRER1dfX04wZM2jo0KE0bNgwevbZZ4mI6IUXXqDjjjuOKisradKkSUQUmd9+yJAhtHHjRtq4cSMdc8wxdOmll1JFRQVt2rTJsX5EREuXLqUTTjiBKisracyYMbRnzx468cQTw7Jdjhs3jlauXBlRh3RfG5mKW7ZDL2HUJCv+O1oHIi2c8T5cnHK32z9+MkXq4xdvh610EXhxv+EGogkTEvu54YboB+7b3/52q3Dfc8899NOf/pSIuMeoTne7c+dO6t+/P7W0tBCRs7jfd999dPnllxMR0UcffUTZ2dmt4q5T4jY3N9OECRPoo48+IqJIy13/X758OQ0dOpT27t1L9fX1VFFRQStWrKCNGzdSdnZ2qzh+73vfo3nz5kXUqba2trWsDz30EN14441ERHTzzTfTDcZBqa2tpR07dlCvXr1ow4YNYWX1EnelFL3//vut85zqd+DAAerbty8tXbqUiIjq6uqoqamJHnvssdYyrF27lpyuCyIRdzditX5NizQrKznirvcTTVy9Hi5u9VLKX529esTaO2OlwtpOJH7FXXzuNi6++GI8/fTTAICnn34aF198MQB+CP785z9HZWUlTj31VGzbtg1ff/2163beeustTJs2DQBQWVmJysrK1nkLFizAyJEjcdxxx2HVqlWOScFM3nnnHZx//vkoLCxEp06dcMEFF+Dtt98GAPTt2xcjRowA4J5WeOvWrTj99NMxbNgw3HvvvVi1ahUA4JVXXsG1117bulzXrl3x73//GyeddBL69u0LwF9a4LKyMhx//PGe9Vu7di169uyJMWPGAACKi4uRk5OD733ve3j++efR1NSERx55BDNmzIi6P4Hx8pk7oUPvdBpdPx2S4kXneHFCT3ebX1YGzJnj7O+++mr2h/vZv9/kglOncgNnS4vV0JmMwbRTTcam/A0NVJRyzj33XPzkJz/BihUr0NjYiFGjRgHgRFw7d+7EBx98gNzcXJSXl8eVXnfjxo34zW9+g2XLlqFr166YMWNGXNvR6HTBAKcM3ucQ4nD99dfjxhtvxDnnnIM33ngDv/jFL2Lej5kWGAhPDWymBY61fgUFBZg8eTL+/ve/Y8GCBa3ZKd3QubntCaLaE051BCKn3XCD/21mZ1uNed27u6fRTRREwK5dkdNNcXVL02ueU6dzvXhx9LFL+/Tx3oYXOjw0ltTEmYhY7jY6deqEiRMn4oorrmi12gEr3W1ubi5ef/11bI5ydZ100kl48sknAQCffvopPv74YwCcLriwsBCdO3fG119/jRdeeKF1naKiItTX10ds68QTT8TChQvR2NiIhoYGPPfcczjxxBN916murg6lpaUAgMcff7x1+uTJkzF37tzW/7t378bxxx+Pt956Cxs3bgQQnhZ4xYoVAIAVK1a0zrfjVr9BgwZh+/btWLZsGQCgvr6+NXf9zJkz8aMf/QhjxoxpHRjEiVQOrZgotAWoFJCTw9/l5RxC6GQZOtVx2rTIyBf7ABZeFBRwJIe2SBOdFsCNhobw/0pxhIsWyGiZXZ0sasA7GgcIf4C4bcOLRA+mnTb8+G6S8cnEBlXNc889RwBozZo1rdN27txJxx9/PA0dOpRmzJhBgwcPpo0bNxJR9AbV888/P6xBdfr06TRw4ECaNGkSnX/++fToo48SEdHvf/97OuaYY2JqUNX7IyK699576c4774yoz8KFC6lv3740cuRI+tnPfkYTJkwgIm5Qveyyy2jIkCFUWVlJf/3rX4mIaPHixTRixAiqrKykU089tbU+kydPpoqKCrr88stb628vg1f9li5dSmPHjqXKykoaO3Ys1dfXt64zaNAgeuGFF1zPyerVq5M+tGI8ePlrY0lalegshvZh6DSpzpKYrHNlHveSEvf6xkMsg2mnAwS9QVU4fNi2bRsNHDjQM4xy9erVGXfTeUVaeEVrxNMIGcvHS9xi2X4yomkyRSC9yEQjwsSvuItbRkgrTzzxBMaOHYvZs2cjy63/eIhEJLRyayjzakBzm+f2+j59OvvDifyXC/BuhPRLSUm468Fedrf28awsHmDCdJFcfbW/nCxRTlsYqeq9CsTfKJrOUd4Sip8nQDI+YrkLsbB69eo2xyS75Rp36gCkFLtVvPaZaMu2pCS2uPWcHO9j4VR2p4+5nt3NNGuWZcnqDj8lJUR5ebHXL5Xx44m4VjI1PBJBdcvoeGxB0LS0tLQ++Nty08Xqb/aKqda9HBMp7nl5LKZ+48/1IBhux8Jv+WbNotZj60cQ3barxd98CDiVLxXCmemulbYQSHHfsGED7dy5UwReaKWlpYV27tzZ2qnKJFaRSLSlPWuWP8s41gdKLMu7idX8+bFvw68gtqXtI1U9QjOtfSaR+BV3XyMxKaXOADAHQDaAh4nof2zzywA8AqAHgFoA04hoq9c2nUZiampqwtatW9sU9y20Pzp27IhevXohNze3dZo9FllTUsIdYJxC3rp3T2wYYFkZ+2GnT+ec4enCXme3Y+OGHmHJ7whM5eXOceZlZRxu6EVb1o0Ft3Od6P2kA78jMUU37VnQPwfQD0AegI8AVNiWeQbA9NDvSQDmRduuWzdzQfCDl8vByRKMNQeL/nh1dddWYCxWcrI+eXlWnWN1F8VqubfF+k6FRe12rs1jFGSQKLcMgBMAvGj8vxXArbZlVgHoHfqtAOyJtl0Rd6Et+BGstmQeNAUrmt9dqeTmaYlVpGNx7ZiCF4tox+s3T4Uv3Ot8tQcSKe4Xgl0x+v+lAP7XtsyTAG4I/b4AAAEocdjWVQCWA1jep0+fFB0KIV4yNWLAb6RKPJa6+TEjRez78yvm2dnOkSXxNJpGW1Zbv16Wu7kdHZ1jP7bJPOfJ9rl7vUW1B387UerF/WgAfwPwIdg3vxVAF6/tiuWe2WRyKtRU9LJ0SiurhdGvZazF0836j7Ydc30/dTYzPno19Kb7PCbzAeJ1nNpDpAxRYsU9qlvGtnwnAFujbVfEPbNp6+tzMm/gZOUh9yO6fvednR1e93geSAUF/iNy7P7kaPtsL0Jnx+v8ZIJhkggSKe45ADYA6Gs0qA6xLdMdQFbo92wAd0fbroh7ZpOJ4W6JzL1i/2g3STIeHG0Jl3QbKcj8OLlXEnEeYzknmeK68xp9qb2QMHHnbeEsAJ+FomZuC027G8A5od8XAlgXWuZhAB2ibVPEPbNpi+WejEYzv70t4/1ocUrW9t1EOt6HiV9xTmYDZia67jKxTIkmoeKejI+Ie2bjJqZeVqLGS5RiuclMq9CPBduWT6KSdnl9nERn1izvukUb4zOe86hTK7SVTO0FmmlvE4lGxF1oM24NgV7d3v1EshQWRk/RmmxL3cmyi9dyN+vjNWizm+h4HS8nn3uslqhT6uFEWLPtuRdoJiPiLiQEv4Jnile84qrxGtneab/xWPVamOw5T2J9oHTqFFn2WDvQRPMTt9USTZaFnamWe3tHxF1ICKnK/+03jM9JAJ2sW10WpwyGXm6JWAbYcHs4uXWi8coDk0w/cbIs7MPBv52J+BV3yecueBJL/m2i+PfzxRf87ZQj3YuaGh5Cbvr08Fzk8+ZxeTp1Ag4ejCzn4sXO21u8OPZ62IdgcxvAWdfRTrTh5tpKIvLgu5Gfb/0uKUlsuYW2IeIueDJ7NmDk6/KFUrHvRwuNmwB60djIouw0Vqbb9mKdHg1zvXjENJ6xPv2SjMEndHIyMzmXw9jsQhoRcRc8mToVKC6ObZ1YLV+lgLPO4kx+8Vr/bqIcq9C6TS8p8R6VyFwv00byScabQbsZRLo948d3k4yP+NyDQ7JDBIG2hzomyp8dbVxUJ3+6WxbK9hyOJ5Ey6QPSoCokilTkcmnLJ1ojXqxCG2359i7cfpBImfQh4n6YkSjBcdpOtAgWnUclHcJ+uIprupFImfQh4n4Y4Xaj6ZS1bmNYmu4QPRiyW6/UWbPcBTZaqlkz73lb8quLCyCzkDeY9OBX3H0Ns5cMnIbZE+LDbegypVgGNQUFHDL4+OPO4Yb25U0KCjjszWvoMqfh3QoKIhvv3MrrRna28zB27WHINEGIFb/D7Em0TDvALVLELtSNjcCf/+weR+71nG9s5PjtvLzw6WYUiN+ojFiiRgoK+IHhJ/qkupofHFlZ/F1d7X8/gtDu8GPeJ+MjbpnEkcoGT6+8MrEQbWxSuz/dTyOn+ICFwwFID9X2SXU1x4MrxZ/u3TlG3CsGO5E0NXGvz3nz+P+ll8ZnJc+Z41zmkhKrd6nZmSdaJx+JuxaEcMTnHiCqq4HLL2eBNcnLA668kntpbt7s7TtPFE7+/Fg7xlRXs/h+8QV3Apo9O/6ONVlZznVWih8IgtBe8OtzF3HPYOzit3evc4Mm4N7omGp0OcrK2ibWseLWSCuNrkJ7QxpUA46OPNm8mS3SzZvdhR3IDGEHrHJs3szlT1WjZqZ1+ReEdCPinqHEmh0xmZSUxLdeKn3eyc6sKAhBQ8Q9Q4k3O2GiKSsDdu3i73iIVo9Ehi8mM7OiIAQNEfc04iVsbtkJCwtjs6Szs4FZsyxxjiUdr1KWW8PJ7eFnW15pbp1cT6l05QhCe0bEPcVoQVeKwwjdhM0tj3pTE/D97/sLfVSKfeCLF/P2iDjM0HRdeD3UWd0KAAAgAElEQVQoiMJDEe1uj3nzgPnz3UXefDg4IeGLgpA8JFomhTh1z7dTUsJuEIBj2L0aUZ1wC4PMzeW87LW14WGHbiGEgP9IEy8L3uvykvBFQYgdiZbJQPw0ktbUWNa723BtXpjDnpk0NfG27W8Jbm6TaFa3iZs/PpqfPpnDvwnC4Y6Iewrx20iq3RKxilxWlv8IG+3+cPOlX321/wbJeMMQJXxREJKHiHsK8SvW+iHgJH5exOrK+OILd1/6H//ofzvxhiFK+KIgJA/xuacA3dM0ltQAuocnwGl6k9FJSXpvCkLwEJ97mjCjYXJywqNiABZ23QDZrZv7djZv5vXefZfzr7clMVhhYWSqXj0otSAI7RMR9wRixm0DlrVtt9SJOCpmwwbv7REBDzzAv6dP9xdXXlIS7uaYP59z0lx5Zfj6RPzQkJhyQWif5KS7AO2JWFIG1NTwwBnRILIaWKO5cwoKOJWuk8968WLnwTtuu0183ILQHhGfewLxihmPxpFHAl9/Hd+6SkVPmSsx5YLQPhCfewqwpw/w8qFH45tv4luvrMxfLhWJKReEwwsR9zjQoyFNmxaePmDPnsiGS78cOBD7OrHEhEtMuSAcXoi4x4Ap6k5pAZqagKKi+DMoxkosMeESUy4Ihxci7j7RkTDRcr3U1FhJuubPt8S0pIRDEhNFWVnswiwpcQXh8EHE3SexRMLovC2mmM6Zk7hxTcWdIghCNETcfRLL4BlOaWsTNbKSuFMEQfCDxLn7pE8f5wGY3di8ma33Ll2AnTsTM7KSW7qAhgbg1luBX/6S0/oKgiCI5e6Ts85y7iHq1Wv0qqtYdP/f/2t7yKGXK+bf/wb+8Afgtdfatg9BENoPvsRdKXWGUmqtUmq9UuoWh/l9lFKvK6U+VEp9rJRqV1lLqqu5q76TzzzH492nsRFYuxbYto2taqdQxPnzOQLHTm4uN8L6iWzZs4e/t23zVx9BENo/UcVdKZUNYC6AMwFUALhYKVVhW+x2AAuI6DgAUwDEkDA28/Hylzc1ea978CDHsF96KQ+koYe1KyiwBPuII8LXKS0FHn2UR2TyE9mixX3rVl/VEQThMMCPz70KwHoi2gAASqmnAZwLYLWxDAHQ3t7OAL5MZCHTTSL85QCHSRYUAF27Auefbwm2PUTyrbeAfv38b7e+nr/FchcEQePHLVMKYIvxf2tomskvAExTSm0FsBjA9U4bUkpdpZRarpRavnPnzjiK659Vq9jd8fvfh6cIcMuCeOAA0Ls38PzzkWkFshLYMtHYCNTVhQ+Hp38PGGAtA3C635NOit57NZrlvn8/cOKJwPvvx19uQRCCRaJk62IAjxFRLwBnAZinlIrYNhE9SESjiWh0jx49ErRrZ+67jy3lm24KTxGgY9DtzJnD4nj22Vb+db1OogfKaGkJF3ct5lrc9+3j77ff5s/Gjd7bi+Zz37IFeOcd/giCcHjgR9y3Aeht/O8VmmZyJYAFAEBE7wPoCMChmTB17NrF3wcPhk93ikGvrgbuuMP639bORllZ7Fs/+mj3ZUxx372bv+3irgfIjuZLN90yTmXX24lnwG1BEIKJH3FfBmCgUqqvUioP3GC6yLbMFwBOAQCl1LFgcU+u3yUKXmkCTB96dTUPhBFL4q5Zs9xHRsrPZ8v8wAF+S3Cab34DVkbIgQP52y7u0Xzp2nJvaGCXjx0Rd0E4/Igq7kTUDOA6AC8CWAOOilmllLpbKXVOaLGfAviBUuojAE8BmEFJTBRPxKML2dHT6uu9xb1XL/7W+WJidbs89hi7brKzI+dpYd67F6isDJ9XVgbcfz//Nh8ObuKu6+BX3M1ltTVvbscu7g0NXPd9+6IfgwMHIt+C0o1Zx6DidB0L7QPzvkwHvnzuRLSYiI4hov5ENDs07Q4iWhT6vZqIxhHRcCIaQUQvJbPQ993H2RfNNtlXX2VXyCOPcC/NtWt5uj0OvWNH4MsvgZdeij8lwL59wBNPRB/kYv16/h4yBBg6lEMav/MdnmZa7uPH87fOJhmPW0anGt68mQW+pAR4883w7ZjifugQ0L8/8NBD/KC54grvfZx/PnDttd7LpJING/g8P/RQuksSP0uXcuRUtDYVIXi8/z7QuTPwr3+lrwyB7KH66qv8/fbb1rT//IctyxdeCF/2uOOs32VlbHEfOgS88krbQhz37Yve61SLe2WlJdBauE1x//vfgU8+sUIi43HLjB3LHZ6WL2exaGqyHnB6O+bbTF0dj/y0OhTQ+sQT3vvYssXaXibwZSjY9r770luOtrB+PdDcDHz+ebpLIiSaZ57h73QONhdIcR85kr+XLbOmaQHr0iV82aIi/r7ySv7Wlt78+W0LcVSK0wE4uWY069fzckOGsOulsdFZ3Dt3ZsteT4vVct+zhzs+HXssW4N2MXey3PXvL332SGhqyiyf/f79/J1JD5xY0W6lTDquQmJYt46/O3VKXxkCKe5asM2nor5BdOSJRv9/4onwxF/bt7ctxLFPH+6ENGwYu0SUAnr2DF9m3Tp+2GgLf9s2Z3HXmOJOFJvPvbgYqKoCliyJFHUnn7t92x06eO+juTmzRMj0Z0brJZyp6Dpk0nEVEoP9rTkdBFLcta97yRLgn//k326Nhlrc/QpAVZV3MjBNURGwYgWwciV3NGppYddFYaEllFu2sHW2Zg3/X7rUetuIJu779nEjZkEBsGOHd2NmfT2L+9ixHAK6YgVPr60FXn8d+Oor/t/QwP/37o10+XTs6F3fpiY+xrqZfMMGyzppC/v2cWctO4cOWe43gI/vggXAk09yWUxxX7Wq7eVINvv3W20gGl2HaAPACOlj715OzOeFdvPqe+Obb6x7Q8Q9RrS419dzA+X69dZBtPsvd+yIbdtHHukvg+OePVb6AB3lkp3NrhHzQdLcbEXIzJwJXHcd/3YS95wcThjW2GjVZ/hwvmjcGt2am3n5oiJgdGg8dJ0dcvlyYNKk8GyRkyYBDz9sbX/7dv6OJu7NzfyA0Q3Q/fsDxxzjvY4fnnySe8/qfgmal14CTj2V2yIAfihedBEf80WLwiNl9DKZzPz5wMknA599Zk0Tt0zmc9FFwAknOIcYax5/HJg8GXjqKf6/2kjMIuIeI9qd8vrr/L1kiXUQt2wJXzbWaJiXX+b0vm5x7AD70fbsYWE891xOt6vZsSMyikb7h/U34Czuevq+fVZ9TjuNv832BRMtEMXF1kNJvxKaFm1urvV740Zr+83N3uXR6AdWbW18g3m78c03/PCy3wT6oawjonS4KMCWlGm5B8Hy1Q9nMwWEuGUyH/1Wqe8TJxoa+Fu/aZoPAhH3GNHiOX48i7Ap7m2Nrt+/n5/E06e7L7N3L9+Ye/awZW02qtofLm5EE3ctWOPHs6tnyRLn5U1x796dRVw//Mw2hf79rd/btkUKoh/LHeDj/NFH3svGgn5Q2GPW9X/9rdsqOnbkY7FnD7/pZGUFQ9y1+2vpUmuaiHvmo92hXuKuDSftftXXbO/e6b02Aynuhw7xTZ2Tw64IU9zdiKXVurERWLwYOOoo5/lduvADhshK4atxc+mYljPg3cPVtNx79OA6mqJgogWiqIiPiVvKAzNn/NatkcfLj88d4PXMskSL9Y+Gfpuxd/jQ//W3FvcJE4APPuC2lC5dOE48COKoI57Mh7S4ZTIfLe5ebXb6GtXirv/37SuWe8y0tFjWclUVN2pGO4gNDewX00QTs82brdctk44dgfPOs/536xY+f/Zs50E5Bg8On+bXLVNSwg2lK1c6u0P0haSH19O9b+2Yx2fbtvjFvaYmXKAWLgT+9jdOU/zsszxtyRJg3jz+vXIlh5+uWQM88EDkdt3EXQufXdxPPpkfvu+/z3Xu1q1tN9CCBcB770VOf/FF/iQKbbl/9JFVFy/LvbkZuPvucHeUFw88EO7rnTePH4LtkcceAz7+ODX70m+/TuJeV8ejrOmgjW++4Ye4Pq/l5SLuMdPSYsWojx3LT9doYY1EVhTJhg3+BpjWAqN7fwLAr34FnHKK9d8u7lOn8iAcOt5ej6J07bXA979vLRdN3NevZ2u/Rw+OgT940HkMV+3f0+JeakvGXFYGfOtb3HN34kRg3DhuK7BnXPYTCgnwxWo2Wt94IzcST5gAfO97PO3uu4Ef/Yh///rXwDXXAH/8I+fksb/eRrPc7W6Z44/n708/5beVtor7RRfxMbHzy1/yjZsotm3jc9HczA88wDtaZvly4M47+Q0yGgcO8LH985+taT/+sdWQ39647jrgT39K7T6dxP1f/+KEg2ZnyvXrrWu2Tx8W/ERnlfVLIMVdu2UAFne/6JsoPz82QTCF76KLwgehtos7wAK/ezc/UPQoSj/8odWarsvghBb3pUvZn9+hg+X6cSqzaeEDlrhrN9TgwdwoNHYsR81MncrHT79Carx8ikTWBVpby6Kk3UybN1sRN3rZJUv4om5o4N/NzcCHH/J8uyWqxd3uc3dzy+jIJMCy3JPh19y71/nNLR727OH6nX8+/9dvPqZbxt5WpC19P3XTHdH0OkS8z/Y4eAtReDRZqnASd50XyBy0vr6ej31hIRtmQGTfm1QRSHE33TK9ekX6xt2sUN35qaICeO45//szhadjx3Bxt/vcvTB7xLr1js3P54tm+XLrwRWPuOv0wXZLXv+3N/x6+RRN4dfiXl7uvOzGjZYgrVxpRYloF4FdrGJ1yxx1lPXgKi7meifjRm9oSJy4a5EdM4Yb2bS467o1NUXuS/vo/dRNL6u/9+/nc9Yexf3gwfAOfqnC6f7Q5+zLL60cVjrQQl+bQPpcM4EVdy2OSkVa7wMHOjdYHnkkf8f6JDWt8w4drIeEfV4iKCjguO29e6166X04XST6ItduIO1z1+Ju98Gb/83GV69OUuaFHU3cTX+8+QDVIm6vg25H8OOWyc3lh7p+QCXCLaOx37zJEPfSUm4jWrqUBaq+3urVbK+DXsdP3fSy9oygbvn9g4w9NUeq8BJ3Iuu+qq/nj742ARH3mDDdMkCkuH/+OYcy2hsJ/YYpmigFXHCB9b9Dh3DL3Z7Lpq1otwzgLO7r13PnHjNfTJcu1puMX8sdsLJQAnzx/vCHnAb5Bz8Ann7ammda7rt28cXrJO7FxeHx+H/7W+Qy9gtdi/5773H/gg8+4Nh+7WrQIt/YaLmydB20W6aujtsVhgxxDxl1wqyX7sWrMcX9j38Ebr3V/3btmOI+diy3+QwbxkaKPo66E9fu3dym89Zb/D8WcdcpNcxj5tQg+z//E943Y+5cbmMAuG2mLXWNh8ZG4PTTrbYIk5YWdoW+8gr/jybud90V3vaQKJqauHxnn20ZJObDX1+TpuWu79sTTghv7E4VgRR30y0D8FPSTO27bx/Hqg8dyv/161G0zjf29MD5+dwoMmGCNT8nxxJ3U1T98u67fAO5oQWsSxdLoLt04YdMTQ0PB/jqq8Cjj/K82trwt4eqKr45f/pTTmms/byaI44Afv5zYNo0FnHNwYPc8DttGvdgvfhia55ptWzdysfffDD893+zBdrUxJkmtTW6cSOfA/O4uon7m29yRs8bb+SOZLojlmm5u4k7wNEzq1eHN25FQwuFrpeJKe7XXsuCGG/Yp952aSkf1wsusDqY6SR4+uZ/+WVuG9EPKT/uB739Q4e485dTfn+Tv/wl/K3qr3/laXqe2TaUCj75hHskX3JJ5LzPPuOIJp1mJJq4P/UUR3AlmuZmvkaff97K9mqK+1FHscFpivuIEVbCwliuy0QRSHG3W+6/+U1kg2Bjo5VYTFvw0XLGfPvb1u+LL+Z0wQMGWOvrb+2WicXfrvnWt4DLL3efrwWsqsqqY3Y2C3xtrSWc+gKzi3teHkf0dO/O1ph9qFqdzXLePLYoNPbXTrNfgDlPNx6VlFjH4Yor+HPwID9Au3blTJcA78NMqOYm7hptsWpMn7s+NvoVuKgo8hzE4os1xd0UwYMH+XrSKRc0//mP/22bbNvG5yg/n8s+f741b+xYPtZazO3uxFgsd4CF3mwjcsoo2tAQXnctSHr5VA8yoXuR62vaRB8Xe8rs+npnV8mePYlzp5mY+Yx0Wcz9dO7M16PplunQgUNUs7KiZ3ZNBoEUd7vlHi0vu47m8PI/jhsHjBpl/dex4N26RYp7fj7vP9H+dr1tINLVpH3L+kbQQmMX91gw2w7Mmx3gRmeNfnCavUFNq/noo/mhokd16tDBEuCxY8P9/NHE3Y6TuDtZ7gC3qcTi33QTd/OmNX/H4vIx2bYt3B1mRkp17Wp1xAMihdWvuHftav2OZrl7ibteP5W+el3HpqbItyN9XHQ9zHI7tZ2lQtx1Wcz9FBXx9Wha7gC/tR51VHoatwMr7qblHi3Rl3nBZGVZg2IAloU5fnx4lI2XuCsVKSyJIpq46xth+XK+iGpq4nuDAMLbDuyhiOax0BaS2ctVNxgdcQQLu+4LUF/P62oxq6qyfjulCogm7n7dMl26xN64agrF+vUc+9/SEinuer9uvYSjsXVrZMP2oEH83amT1Ult//7w85CVFb0++/dzW1JVFf+3i/u6dZFC7STu9fXciF9Xxw9p87wQcZvAgQP8O9HCb9bRnmlUH3MncbdfS4cOJbYh3HwzaGqKbKg291NczPeEXdwBvl5F3H1id8s49Qp1Q6cNmD+fv7Wvs7w8XNB0F/2SEkvUzfndu1vRN4lE+9fHjAmfroVLX9B793LZ1q+P/yFjpie2J1gzL1xtuR9xhDWtuJgtEt0gqMV9zx7eZnk5L1NRwb/z85177JntIHob+js729ly1/ssKbEatM8+m4/Dl1/yeVmwIHr9TaH43//l+s2aFV732lprOe3mu+MOTvPsF7vlDlguwLw8FuamJu6YZQrzsceyderm629p4S7uW7bwW2deHr/F6m3k5XEnsltu4QbLn/yEp9vFvb6e7ynTLWKW49e/Zvfe4MHATTfxvReLwK9axffnhg3O881r4q23+DwsWsRl/Ogjtn6//JLra5bbfi3puHPz/F1xBbcjuXHPPeHuSc2jj4YbTablfu+9fAzM+tgtd/OtuLRU3DK+sbtldK9QvwLf2MiNjQC7FF56CbjssnDxPnCARW3oUGu6GX2zYEFiezBqrrySs12aQgrwhVZTwxf0ccdx70MtuvGKe1YWN87OmGFNu+IKfosxbxBtwZgPs+JibtzVDbt2y/2OOzjCITubBeGVV/iB6OWWueYabrDSN1vPnnzD6ptai/vw4dzA9u1vc9rhF17gFAfdurEY7Njhb3gzLRR3383iPmgQd+4y625GWH39NX8vXeq/+3tTE5fHbrn/+tfAP/4R3iaxaxcLg1I8b/p0rrtbutmGBo7yGTQIuOEGvlY/+MCyMP/xD6BfP34rWLaMj4luT9B11ymjgfCObaa46w5omzZZwxrGMrD3unW8P7dRs8xrYsUKfoP65BMud3Mzp5xobubj6CXuuszm+fv0U+9ztWZNZIc+gI+X+RZliruebkbAFBfzZ+dOLqtpuffqJZa7b+xuGYAFXkd42JN0OWH66SdP5geDU+ensWMj3TIAt4T7yfseK127WtE5JqZb5uij+WbWD7i2uIfGjQtv8Dz5ZO4n4GS5m+JeVMTLad+8k1tGv30ccQQ3JDu5TUxx79+fBVsLof7euzdc3JXisEkdhXPGGbzPbt2scvu5mbSonXwyR8QMGcLlcxL3Xr3CBzipq/Pu1avZvp2tXLvlnpPDYxFoFx9gNcYVF/M8fbzdXDO6nDfcwMd47FgWpW++4ftj8mR+EK5bx28A27ZZ65gNkxpTrOx+e9Ml51UmJ6JFuNTU8DWYnR0eKKD97ToUedu28DdMP+JeX+9dVjc3jv36Md0yGjOtgHbLaAvd7papq0tOW4AXgRR37VsrL+eLuLyc47O1YDc1RQ9RdBJmM4cMwBd0377O4p5qunXjm3bnTv6tlBUq2Vbfv/kw7NCB3TV+LHcT/WCsr488jmYdvMRdC6D9e88evqmjvZmZr9F+xN0+5KEun5O4DxjA0w8csLbtJ6mXGQbphj6Wdn9ttE4wupy6Damqitdftoy3oRTvV/cStou77kilMaOB7BE3kyeHt1UlUty1+7Nr10hx79XLMhLMYSqdtqfLrN9OAD4e0cTdHhWl92ViH/3LjnbLaLep3S3jtM1kE0hx37CBRW7zZr5AN29mv9rLL1vLHDpkiZZTut3ZsyO3a7fc9ZB7mSLuRCw2+qbX4u7nTcULU4w7duTjY1pITpa7Xdz1Nvbvd0//4Cbu2gJ3E/eFC8MtdzfMh5yTj3P9+vAb1C7u2vXlJO46p83WrVaUhp+wS31Du2XrBCwhsPtrTXH/+GPrIfvpp3zc7OKuG+Hfecc5kdzBg5YB1NLCv01Bd3LLtLSwv7tPH2ukLyCy7gcOsKvMHNzihReAN96wjnNNDbvnXn45vB1BBy5062aVT6eWHjvWqsPWrZHivmGDdU3ZG5K/+ooFXw9b6YQ+hg0NvL52HdmvHyfLHbDuPe2W0dgtd4DbEbx6gieaQIr7xx/7a9DRN8O0aZYwFxayf94pK6QWpbw8tiImTeL/Tg2qqcb0wWsLVfvKzYE44sF8OHTsyMfowIHIdKe6DDk5kcfCfEC4Hacjj2RhNLttHzjADXX5+VY9hg3jNy/te7/+ehbZWMTd3vWeiDNK/vrX1jQny/3gwfChGe3ibg7p58d61WG4bnn2AUvMddIpLQz6eC9axO6Ve+9lK3zYMO6IZhf3QYN43f37rW3aHyrmMH9Tp1qZPPU8fSy0UO7axee/V6/wbKj2uldXsytp/Hg+9s88w66ziRMtn/3zz/MbwGmnhY8na4q7Fv1Nm1i4R43i49ChA//X56ywkAW4f3/LjWmK+/nncye9aDnzTXG/6y5+S2hoiByeU1vupgsTYPdsVhbnDHJLSzJwIBuJN93kr6E/UQRS3GMdOm/CBOCcc/j3tGnu6X61KOnXwx//OHx6Oi33446zfusL58IL+SLUYXDx4iTugHXh290y+pXfxI+46zroJGLaipkyhW9U/dCaOJGtrilTwt/GYhH3/fvD46C/+YYtR9MicxJ3ILwRVVuS+i3JbJzzI+41NXysdBy6E9nZ/LakQxK1uPfrx7//+EerXNoP/fnn1vnR7qqsLMuF4ZYC2mzU/PzzcEFsarJCNLUomm6lW27htwanuut0EQALsznoud6GOYKXmUlUu2XM86ffIsrL+fiMGMHuJn3OvvUtbjAGrDKZlvX69eFvJX7E/e23eRsvvhhpPGpxnzaNvQb6HjntNL4H+/YNt9aHD7d+9+5thXi6jYWcDAIp7tFucqfl9YH3WleLko7h1n77THDLmINRmzeBvQdqPNjdMnZx126Z4uLI3DpO23ATd/0Q0rHL2t9eUBDZbqAb8MwHVyziDoQLuVMiLj/ivmULv6noNppYLfeaGhZ2tyygGqcOMKZYAywSWiAHDYq03AHLNeMm7qblbs/pD3D4JRDZYae0lI0A/XZlr7t9MJglS6xrQufNsWcX1dTU8LG3hx6a5a+q4mifvXv5fIwbZ7mG9MPNfFAdOhQeNBFN3Gtrrdw2TjmRGhv5LVMPZ6nLWlho/dbHvLjY6j+j6d+f10ul3z2Q4n7ssdFTCQDWxZWfb70y+RF3u3jl5ob73tNBVpZ1ESe685RTgyoQabnn5vK+4xX3Hj3YwtHWpxZ3L3eXfqAA/sVdC5S9Wz4Q7iv2I+5NTeHCY1rufnzufnsQFxdbbhnz9d58uOk8/0B4Jxoncdfb0OI4cCBfQ6a4Ow0iMWBAeP8Ce4OwbpNxEne9z88/54fQxIn833yI6LLq9fft4+tAu2XsmD2dGxv5rS8/P/y4aJeXvcHT9G9HE/f33rOuc6d04GbPbMAqq9nIr+uvrz+nuqQy3j2Q4n700Zy4yryo7ZSVcYgY4N9y1wJlFy8t7OkUd8Dy+8b65hINL8v9pZesmPGcnPCcMm7bcIuWAfgmXbKEX8t19r5ox1XfSNHqrQXY7K2pMS33hQs57G/fPj63urx6/S1buEz6oVJSYs377DOuv1K8rf/8xxpe0OTLLzl/j19x170bTbcMEN5Tub7eyrq5b5+zuOu6620UFnJHr/JydqvZe4Da6dWLy/Lmm8DPfsad/bKzw8dMcBogpbaWrdOCAm5YbWqy2qy05Q5wOTp3ttbXousm7lq4db3eeCNS3O2DqTthlnfFCmuEK30MX3uNv0eNcnb76nLaxd089toN6Cbuqe6pGkhxP3SIby6nRlUdkbFpEw/HNWQIH+xY3DJOlunpp1tDvKWLOXOsYfcSiZfP/fTTebg3vdzpp1sWmYkfyx0AKitZPB94wNpuosS9Z0++6XV/B9Ova4r7ZZdxJ6t9+1iM9FugabkXFFjH4eijOU2Ajurp399K5HbXXbw/ezTG44/zftat82+519VFdl0/8US+hgG2iM10DE7iftRR3LBpDh147rl83nr1cs+MajbAFhezz/z++9kCP+WU8NBip6gnnQajtNQak1aLuymWpaXh6+uGy+7dnV1z+toYMIDreegQXwclJVyvoiLetx59yg2zvDffzL7zgwcta/2DD3j/11/P2xw9msdq1e1Mbpa7eezPPJPn/+xnzmVItbjnRF8k82hp4YPkFFZUV8dP5alT2U+qG1ticcs4WaaxjNyULCZMiGzFTwRulrvdEsrN5Qyc0bbhJe7al266BxIl7h06WC6foqJIPzBguQiWLuUb19ym3k9TEx8D7TYaM4YfAF26sBVaVcW+Xx2LrcdFNbux6xj4DRucu7fbKS7ma5UoMuri00+5cdV8WLmJO2A1NGoee4y/33knPN++Jj+f2wXq61mA9P4vu8zqgWziJO61tRzFc+gQP9BKS8OHRNSUllo9rYHwUFF9vPWYBmZ7gVK8zNq11jlbuJCvx5tu4vOhe/c6GX16fxtnba0AABSbSURBVC0tfAz27Al3sX39NbdrTZ/OH40eF0Kvbw9TNY99v37uvYl1HXfu5AdsKiLvAmm5t7S4x4seOgRcdRWHZpm01XJvz7hZ7vaWfXu+exO/4q5vCjOPiV9xj+WGsAuQ3de5ZQtbwub1kJ8fHjKrHwTaNaK3V1XF21+71jpG9oyRphXp1y2jhc7p+issDB9QZN8+toizs73dYCZuHanMGO3SUquebuMTOw1tqN1P2kdeVeXsNu3VK/zcmA222vXlNtCM3rbTA7m2lh9O9rQdZvkAPmf63GhXjH1bdnJzIy13s0HVL7o+ZmRRMgmkuB865H2jm7ljNCLu7tiFWV+wpnUNeHeWilXcTd9vNHHXN5KXT9VpP06Wu8m770ZeD043rfbv6hjssWN5+ytW8H+lIsXdLKtft4y2XN3EXee2ASzLvbDQX3AB4N6RSou7HoxdvyG4hdjafe5ElrhrARs7lo0B+4NHi7hef+tWbug98kjrOGmL315evW03cd+zJ3I8ZYC3rfenz5NSnMPJXi8nnMTdyXKPhq5PqlwzgRR3nQ3Pqzu6Pcd7W90y7Rm3aBl7oie/lruXJanF03x9jWaRa2sslmRVWoCuuYbjl7dti+xItHdv5PWg49HNm9a+3rHHWvXIymJ/thaNd97hPDXxWO5OvzWFhVYoYdeubMBocfeLm+VeVMSf0tLwB8WwYc7L6wendn/s3ctls4u7Lre9DHbL/aij+Nqyi7ub5W5eL3qda6/ljJJO4l5ebonz0qVc1/HjIy13t9TZubn+3DLR0PW58srUuHkD63M/6ijg9tvZJ+YU0mXPHTNqFI8ReuKJ7tstLgb+67+A885LbHkzHS3GHTrwza0fmrFY7uYN58dyN4lmud9yC9+cM2d6L2dSUsKpZt9/n11Au3YB3/0uDykHcCPlqlWR4n7NNezLveQSTv9rNgYuWsQN9dnZ3CC3axf747OyuONLUxNnp3ziiXCr16/lrjGHMNSYInLkkeGWu1+0uHToYDWs/vjH3EGouNiKalm8mKOA3M53v35c1w0buHHZjHg55RTugKTbGQoLOYrkpJM4Nv+kk/hBqFMZm+mQy8s5wm3mTH5YmL1nzfI7vRW9/z67c2bO5Eyn+kHYoQP3gP7Xv/hhtGQJn7PBgyOHvvOy3LXG6BHKzjiDrw/d6csPxxwDXHopv4G1NWWIL4goLZ9Ro0ZRvIwbRzRpEv+eP5+ooEAPIcCfggKeLvjj3Xf5uHXuzP8PHAg/nvqza5f7NnbvtpZ79ln35RoaIrf76aeJrQ8R0dVXR+7n4Yf5u6SEaNYs/j1xYtv39eCDvK3Nm4kGDeLfPXpY+/3nP6Nv4w9/sJZvbo6cP2WKNf+kk4iGDCE65xyi4cP9l/M//+H1S0utba1a5X99zYcf8rrV1fz/gw/4/8KFkcvq4/H971vTfvtbnrZ7N1FFBdH55/vb78KFvF7fvta0L7+06vLUUzytc2eioiKi7Gw+D3Pn8vw1a4hycohuvZXoV7+KvD7uust5v+Xl1jIHD/orazIBsJx8aGxg3TI6NEvnci8rY6uzrMw9d4zgjGm56/9OLphE+Nzz8yPnJ6P/gN0Ky8mxcqP06mW5DRLRZ0BblKtWWa4ss+OOn5Gy9Ot+797OGU0Tabmb5YmnfWnoUD5uukOVabnb0eUzj7Pef22t80Ambmi3jNmWYqZ1MF1B3bvzvKIia/oDD7BFbyYjM/Gy3PV3SizuBBFIt4x9JKapU0XM24K+YE2RLSyMDOtKhLgrxTeRGdaXCnGvrLR856Wl1g3vd4AXL7ToLFzoryxO6OPn1gFGi2ReHncC0uIeS/tQp068bufO/LBrbo6vfSknh92cup1BP8j8irtebtMmvsb8irtezrwuzWtHj9BVWMgPrdxcPr/DhvE1+Yc/8PyxY/lBrMnNtXoiO6Gv+1gepJlA4C13oe1oYTFvFO1bNPFqUM3OthrjojWQ2m+iZFhD5j6U4htaC2Pv3uz/7NLFuZ6xYqYmNtHHy4/lrhsn3Ybv08JSXGzFgcdquQPcFtW5syW28da/qoqzPT73HLdPAJEDepjlNh+iejn9JtW7t7996oZ1t0Fy9PXXuTN39uvRg89xXh4wciTrRp8+3F5nPlB0fqb2Ju6+LHel1BkA5gDIBvAwEf2Pbf79AHS/xQIARxBRl0QW1MRpJCYhfpws94cf5s4ea9YATz3F07zEXXfjP3Agety1FruHH+ZXZ7fY5Lag99GjB/B//2dlpHzmGSvPynPPeafi9Uu3bvxA27GDxUR3YLr2WhZrP5b7RRex9ej2BqqFpaiobeL+0EMstKee2jYjqWdPPtc6w+fcuc5jCjtZ7mPGcG/rujq+5nTG1mhkZXHDqB79S/PGG+FRMnPn8v4OHrR04ve/5/zyuueuKe5HHMGx517RMmZdgkJUcVdKZQOYC2AygK0AlimlFhFR66BcRPQTY/nrARwXsaEEYnfLCG3DyXI/4wz+/Pa3LO6mZe61HT+977TY9e/PQ9wlA72P0lIePFszebL1O1H71r0nP/+ct//MMzy9vNwaIi4a2dnhPSPt2C33Awc4BDFWwTHbGtry9qvFWvu/Z81yXs5J3LOzgR/9KL79nn565DT7sJRO8fmjR4cPNmK6o9qr5e5HIqsArCeiDUR0EMDTAM71WP5iAE8lonBuiFsmseiL10mU9U3px3Vib5h1Q99EftwV8WKKeyrQ+5k4MTmd4bRbo6jI+l1TE3+bQUFB2/pz6OuipoaNArcHvy5fopPdJRKdnvdwFPdSAEYSVGwNTYtAKVUGoC+A11zmX6WUWq6UWr7TKZm0T2prOa7YHD9ViB8ny12jb0ovl4x9O37FPdGpi5324TW8XSIxO+/Y83snArvlDvAbbLyCY2ZKjXd9gMXdS7idLPdMQ/fytedg17RncY+FKQCeJSKHbkUAET1IRKOJaHSPOEeZqK5m/1hjozV+qlMuGcE/Tj53TTIs9/Ly8EEOkkG3btxYGEsnk7YweDD724cNsx4siezp7CTuQPwNoj17tu3Bpy3y2trgirvuCatTiLu5eoMq7n4aVLcBMNuze4WmOTEFwLVtLZQXt90WmfVN55KRcMj48CPuibTcf/AD9oMnMz9+Xh7nbHdq5EsGN90EXHGFNaAJkBzLXTeoauJtEH7iCf85aZwwfe5BFfcVK7hRulMnTvXrhr4/EhE2m0r8iPsyAAOVUn3Boj4FwCX2hZRSgwF0BfB+Qktow54zJtp0ITpebhl9QcdiuUeLlsnLcw9nSyR+Q+wSQX6+ZQmnyi0DxN+m0FaXmCnuTukSNJks7p06WW8+XlZ5UC33qG4ZImoGcB2AFwGsAbCAiFYppe5WSplBTFMAPB3qHps03EQhFWLRXtGRMG1tUNXrpyJXdSaTSrdMqhqM7egy1Nd7W7ROce5BI6ji7ivOnYgWA1hsm3aH7f8vElcsd2bP5uQ75iOkoICnC/GTl5c4t4zf/OLtlVS6ZdIt7vbfdjLZcveLvvaDJu6BixafOpU7vugWbsklkxjGjOEu+nZibVDNzZU+CGPGcP6VRFruPXtymusRI8KFMl3pqf2K+5AhXPZ+/ZJfpmShM0IGTdwDmVumY0fuHPLQQ+kuSfvBnv5UE6vlfri7ZABOVWtPV9tWCgs5xS4QPjxcuvAr7hUVqRt5KFnoUd+CJu6BtLGkh2rqiNVyF3FPPpngv/Yr7u0BEfcUIj1UU0eslvvh7m9PBZkgpiLumU9gxV0s99QglnvmkQlj/eblWfdgexd3PWpV0MQ9kD53ccukjtxcttr9WO4XXggMH578Mh3udOvG43D+8IfpK4NSLOoNDe1f3INquQdS3MUtk1ry8/1Z7oluRBScycridMnpRot7JrQBJBMt7kGrZyDtX3HLpBa/4i4cXmiL/XCx3IPWnhRIiRS3TGrJz/fnlhEOL0TcM5tASqS4ZVJLYaE0lAqRHC7irhPYis89BYjlnlruvz+5udeFYHK4iPszzwCLF3snSMtEAinuYrmnltNOS3cJhEzkcBH3o47idM5BI5D2rzSoCkL6OVzEPagETiJbWvhbxF0Q0ksQxkc9nAmcRGpxF7eMIKQXLepBi/8+XAicuOv0m2K5C0J6EbdMZhM4iRTLXRAyAxH3zCaw4i6WuyCkFxH3zCZwEiluGUHIDETcM5vASaS4ZQQhMzjmGODII9M31J/gTWDFXSx3QUgvU6YAX30VvJwrhwuBk0hxywiCIEQncBIpbhlBEIToBE7cxXIXBEGITuAkUix3QRCE6ARW3MVyFwRBcCdwEiluGUEQhOgETiLFLSMIghCdwIm7WO6CIAjRCZxEis9dEAQhOoGTSHHLCIIgRCdw4i5uGUEQhOgETiLFchcEQYhOYMVdLHdBEAR3AieR4pYRBEGITuAkUtwygiAI0QmcuIvlLgiCEJ3ASaRY7oIgCNHxJe5KqTOUUmuVUuuVUre4LPN9pdRqpdQqpdSTiS2mhTSoCoIgRCcn2gJKqWwAcwFMBrAVwDKl1CIiWm0sMxDArQDGEdFupdQRySqwuGUEQRCi40ciqwCsJ6INRHQQwNMAzrUt8wMAc4loNwAQ0Y7EFtNC3DKCIAjR8SPupQC2GP+3hqaZHAPgGKXUu0qpfyulznDakFLqKqXUcqXU8p07d8ZVYHHLCIIgRCdREpkDYCCAkwFcDOAhpVQX+0JE9CARjSai0T169IhrR+KWEQRBiI4fidwGoLfxv1domslWAIuIqImINgL4DCz2CUfcMoIgCNHxI+7LAAxUSvVVSuUBmAJgkW2ZhWCrHUqp7mA3zYYElrMVsdwFQRCiE1UiiagZwHUAXgSwBsACIlqllLpbKXVOaLEXAdQopVYDeB3ATURUk4wCi+UuCIIQnaihkABARIsBLLZNu8P4TQBuDH2SijSoCoIgRCdwEiluGUEQhOgETiLFLSMIghCdwIm7WO6CIAjRCZxEis9dEAQhOoGTyLff5u/Bg4HycqC6Oq3FEQRByEgCJe7V1cAjj1j/N28GrrpKBF4QBMFOoMT9ttuAgwfDpzU28nRBEATBIlDi/sUXsU0XBEE4XAmUuPfpE9t0QRCEw5VAifvs2UBubvi0ggKeLgiCIFgEStynTgUuucT6X1YGPPggTxcEQRAsfOWWySRGjQIefxzYtQsoKUl3aQRBEDKTQFnugHRiEgRB8EPgJFLSDwiCIEQncBIpicMEQRCiE1hxF8tdEATBncBJ5E03sWsmPz/dJREEQchcAiXu1dVA375ATg5/S04ZQRAEZwITClldzUnCGhv5v04aBkicuyAIgp3AWO633WYJu0aShgmCIDgTGHGXpGGCIAj+CYy4S9IwQRAE/wRG3GfP5iRhJpI0TBAEwZnAiPvUqZwkrKwMUEqShgmCIHgRmGgZgIVcxFwQBCE6gbHcBUEQBP+IuAuCILRDRNwFQRDaISLugiAI7RARd0EQhHaIIqL07FipnQA2x7l6dwC7ElicdCJ1yUykLpmJ1AUoI6Ie0RZKm7i3BaXUciIane5yJAKpS2YidclMpC7+EbeMIAhCO0TEXRAEoR0SVHF/MN0FSCBSl8xE6pKZSF18EkifuyAIguBNUC13QRAEwQMRd0EQhHZI4MRdKXWGUmqtUmq9UuqWdJcnVpRSm5RSnyilViqlloemdVNKvayUWhf67prucjqhlHpEKbVDKfWpMc2x7Ir5feg8fayUGpm+kkfiUpdfKKW2hc7NSqXUWca8W0N1WauUOj09pY5EKdVbKfW6Umq1UmqVUuqG0PTAnRePugTxvHRUSi1VSn0Uqstdoel9lVJLQmX+i1IqLzS9Q+j/+tD88jYXgogC8wGQDeBzAP0A5AH4CEBFussVYx02Aehum/b/Abgl9PsWAL9Odzldyn4SgJEAPo1WdgBnAXgBgAJwPIAl6S6/j7r8AsDPHJatCF1rHQD0DV2D2emuQ6hsPQGMDP0uAvBZqLyBOy8edQnieVEAOoV+5wJYEjreCwBMCU1/AMCs0O9rADwQ+j0FwF/aWoagWe5VANYT0QYiOgjgaQDnprlMieBcAI+Hfj8O4Lw0lsUVInoLQK1tslvZzwXwBDH/BtBFKdUzNSWNjktd3DgXwNNEdICINgJYD74W0w4RbSeiFaHf9QDWAChFAM+LR13cyOTzQkS0N/Q3N/QhAJMAPBuabj8v+nw9C+AUpZRqSxmCJu6lALYY/7fC++RnIgTgJaXUB0qpq0LTjiSi7aHfXwE4Mj1Fiwu3sgf1XF0Xclc8YrjHAlGX0Kv8cWArMdDnxVYXIIDnRSmVrZRaCWAHgJfBbxbfEFFzaBGzvK11Cc2vA1DSlv0HTdzbA+OJaCSAMwFcq5Q6yZxJ/F4WyPjUIJc9xJ8A9AcwAsB2APeltzj+UUp1AvBXAD8moj3mvKCdF4e6BPK8ENEhIhoBoBf4jWJwKvcfNHHfBqC38b9XaFpgIKJtoe8dAJ4Dn/Sv9atx6HtH+koYM25lD9y5IqKvQzdkC4CHYL3iZ3RdlFK5YDGsJqK/hSYH8rw41SWo50VDRN8AeB3ACWA3mB7e1Cxva11C8zsDqGnLfoMm7ssADAy1OOeBGx4WpblMvlFKFSqlivRvAKcB+BRch+mhxaYD+Ht6ShgXbmVfBOCyUHTG8QDqDDdBRmLzPZ8PPjcA12VKKKKhL4CBAJamunxOhPyy/wdgDRH91pgVuPPiVpeAnpceSqkuod/5ACaD2xBeB3BhaDH7edHn60IAr4XeuOIn3a3KcbRCnwVuRf8cwG3pLk+MZe8Hbt3/CMAqXX6wb+1VAOsAvAKgW7rL6lL+p8CvxU1gf+GVbmUHRwvMDZ2nTwCMTnf5fdRlXqisH4dutp7G8reF6rIWwJnpLr9RrvFgl8vHAFaGPmcF8bx41CWI56USwIehMn8K4I7Q9H7gB9B6AM8A6BCa3jH0f31ofr+2lkHSDwiCILRDguaWEQRBEHwg4i4IgtAOEXEXBEFoh4i4C4IgtENE3AVBENohIu6CIAjtEBF3QRCEdsj/D8C+DcKYiyxoAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot the training accuracy\n",
    "accuracy = model_train.history['acc']\n",
    "val_accuracy = model_train.history['val_acc']\n",
    "loss = model_train.history['loss']\n",
    "val_loss = model_train.history['val_loss']\n",
    "epochs = range(len(accuracy))\n",
    "plt.plot(epochs, accuracy, 'bo', label='Training accuracy')\n",
    "plt.plot(epochs, val_accuracy, 'b', label='Validation accuracy')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.legend()\n",
    "plt.figure()\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}