schneider-electric-classification.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
    "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5",
    "execution": {
     "iopub.execute_input": "2021-05-28T22:44:52.661210Z",
     "iopub.status.busy": "2021-05-28T22:44:52.660825Z",
     "iopub.status.idle": "2021-05-28T22:44:52.689580Z",
     "shell.execute_reply": "2021-05-28T22:44:52.688632Z",
     "shell.execute_reply.started": "2021-05-28T22:44:52.661130Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/kaggle/input/classified-netflix-movies-data/classified_data.csv\n"
     ]
    }
   ],
   "source": [
    "# This Python 3 environment comes with many helpful analytics libraries installed\n",
    "# It is defined by the kaggle/python Docker image: https://github.com/kaggle/docker-python\n",
    "# For example, here's several helpful packages to load\n",
    "\n",
    "import numpy as np # linear algebra\n",
    "import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Input data files are available in the read-only \"../input/\" directory\n",
    "# For example, running this (by clicking run or pressing Shift+Enter) will list all files under the input directory\n",
    "\n",
    "import os\n",
    "for dirname, _, filenames in os.walk('/kaggle/input'):\n",
    "    for filename in filenames:\n",
    "        print(os.path.join(dirname, filename))\n",
    "\n",
    "# You can write up to 20GB to the current directory (/kaggle/working/) that gets preserved as output when you create a version using \"Save & Run All\" \n",
    "# You can also write temporary files to /kaggle/temp/, but they won't be saved outside of the current session"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-28T22:44:52.765365Z",
     "iopub.status.busy": "2021-05-28T22:44:52.765124Z",
     "iopub.status.idle": "2021-05-28T22:44:52.870235Z",
     "shell.execute_reply": "2021-05-28T22:44:52.869285Z",
     "shell.execute_reply.started": "2021-05-28T22:44:52.765342Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Unnamed: 0</th>\n",
       "      <th>show_id</th>\n",
       "      <th>type</th>\n",
       "      <th>title</th>\n",
       "      <th>director</th>\n",
       "      <th>cast</th>\n",
       "      <th>country</th>\n",
       "      <th>date_added</th>\n",
       "      <th>release_year</th>\n",
       "      <th>rating</th>\n",
       "      <th>duration</th>\n",
       "      <th>listed_in</th>\n",
       "      <th>description</th>\n",
       "      <th>cluster</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>s1</td>\n",
       "      <td>TV Show</td>\n",
       "      <td>3%</td>\n",
       "      <td>NaN</td>\n",
       "      <td>João Miguel, Bianca Comparato, Michel Gomes, R...</td>\n",
       "      <td>Brazil</td>\n",
       "      <td>August 14, 2020</td>\n",
       "      <td>2020</td>\n",
       "      <td>TV-MA</td>\n",
       "      <td>4 Seasons</td>\n",
       "      <td>International TV Shows, TV Dramas, TV Sci-Fi &amp;...</td>\n",
       "      <td>In a future where the elite inhabit an island ...</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>s2</td>\n",
       "      <td>Movie</td>\n",
       "      <td>7:19</td>\n",
       "      <td>Jorge Michel Grau</td>\n",
       "      <td>Demián Bichir, Héctor Bonilla, Oscar Serrano, ...</td>\n",
       "      <td>Mexico</td>\n",
       "      <td>December 23, 2016</td>\n",
       "      <td>2016</td>\n",
       "      <td>TV-MA</td>\n",
       "      <td>93 min</td>\n",
       "      <td>Dramas, International Movies</td>\n",
       "      <td>After a devastating earthquake hits Mexico Cit...</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>s3</td>\n",
       "      <td>Movie</td>\n",
       "      <td>23:59</td>\n",
       "      <td>Gilbert Chan</td>\n",
       "      <td>Tedd Chan, Stella Chung, Henley Hii, Lawrence ...</td>\n",
       "      <td>Singapore</td>\n",
       "      <td>December 20, 2018</td>\n",
       "      <td>2011</td>\n",
       "      <td>R</td>\n",
       "      <td>78 min</td>\n",
       "      <td>Horror Movies, International Movies</td>\n",
       "      <td>When an army recruit is found dead, his fellow...</td>\n",
       "      <td>18</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>s4</td>\n",
       "      <td>Movie</td>\n",
       "      <td>9</td>\n",
       "      <td>Shane Acker</td>\n",
       "      <td>Elijah Wood, John C. Reilly, Jennifer Connelly...</td>\n",
       "      <td>United States</td>\n",
       "      <td>November 16, 2017</td>\n",
       "      <td>2009</td>\n",
       "      <td>PG-13</td>\n",
       "      <td>80 min</td>\n",
       "      <td>Action &amp; Adventure, Independent Movies, Sci-Fi...</td>\n",
       "      <td>In a postapocalyptic world, rag-doll robots hi...</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>s5</td>\n",
       "      <td>Movie</td>\n",
       "      <td>21</td>\n",
       "      <td>Robert Luketic</td>\n",
       "      <td>Jim Sturgess, Kevin Spacey, Kate Bosworth, Aar...</td>\n",
       "      <td>United States</td>\n",
       "      <td>January 1, 2020</td>\n",
       "      <td>2008</td>\n",
       "      <td>PG-13</td>\n",
       "      <td>123 min</td>\n",
       "      <td>Dramas</td>\n",
       "      <td>A brilliant group of students become card-coun...</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0 show_id     type  title           director  \\\n",
       "0           0      s1  TV Show     3%                NaN   \n",
       "1           1      s2    Movie   7:19  Jorge Michel Grau   \n",
       "2           2      s3    Movie  23:59       Gilbert Chan   \n",
       "3           3      s4    Movie      9        Shane Acker   \n",
       "4           4      s5    Movie     21     Robert Luketic   \n",
       "\n",
       "                                                cast        country  \\\n",
       "0  João Miguel, Bianca Comparato, Michel Gomes, R...         Brazil   \n",
       "1  Demián Bichir, Héctor Bonilla, Oscar Serrano, ...         Mexico   \n",
       "2  Tedd Chan, Stella Chung, Henley Hii, Lawrence ...      Singapore   \n",
       "3  Elijah Wood, John C. Reilly, Jennifer Connelly...  United States   \n",
       "4  Jim Sturgess, Kevin Spacey, Kate Bosworth, Aar...  United States   \n",
       "\n",
       "          date_added  release_year rating   duration  \\\n",
       "0    August 14, 2020          2020  TV-MA  4 Seasons   \n",
       "1  December 23, 2016          2016  TV-MA     93 min   \n",
       "2  December 20, 2018          2011      R     78 min   \n",
       "3  November 16, 2017          2009  PG-13     80 min   \n",
       "4    January 1, 2020          2008  PG-13    123 min   \n",
       "\n",
       "                                           listed_in  \\\n",
       "0  International TV Shows, TV Dramas, TV Sci-Fi &...   \n",
       "1                       Dramas, International Movies   \n",
       "2                Horror Movies, International Movies   \n",
       "3  Action & Adventure, Independent Movies, Sci-Fi...   \n",
       "4                                             Dramas   \n",
       "\n",
       "                                         description  cluster  \n",
       "0  In a future where the elite inhabit an island ...        7  \n",
       "1  After a devastating earthquake hits Mexico Cit...       11  \n",
       "2  When an army recruit is found dead, his fellow...       18  \n",
       "3  In a postapocalyptic world, rag-doll robots hi...       14  \n",
       "4  A brilliant group of students become card-coun...        3  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data = pd.read_csv('/kaggle/input/classified-netflix-movies-data/classified_data.csv')\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-28T22:44:52.872076Z",
     "iopub.status.busy": "2021-05-28T22:44:52.871745Z",
     "iopub.status.idle": "2021-05-28T22:44:52.883896Z",
     "shell.execute_reply": "2021-05-28T22:44:52.882984Z",
     "shell.execute_reply.started": "2021-05-28T22:44:52.872041Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X: (7787,)\n",
      "y: (7787, 20)\n"
     ]
    }
   ],
   "source": [
    "# X and Y for training\n",
    "X = data.description.values\n",
    "y = pd.get_dummies(data.cluster).values\n",
    "\n",
    "print('X: {}'.format(X.shape))\n",
    "print('y: {}'.format(y.shape))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "# Text - Pre-processing and Model Building"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-29T00:35:48.040709Z",
     "iopub.status.busy": "2021-05-29T00:35:48.040359Z",
     "iopub.status.idle": "2021-05-29T00:35:48.045692Z",
     "shell.execute_reply": "2021-05-29T00:35:48.044886Z",
     "shell.execute_reply.started": "2021-05-29T00:35:48.040676Z"
    }
   },
   "outputs": [],
   "source": [
    "from keras.preprocessing.text import Tokenizer\n",
    "from keras.preprocessing.sequence import pad_sequences\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "\n",
    "from time import time\n",
    "import keras.backend as K\n",
    "from keras.layers import Dense, Input, LSTM, GRU, Dropout, SimpleRNN, Embedding\n",
    "from keras.models import Model\n",
    "from keras.optimizers import SGD, RMSprop, Adam\n",
    "from keras import callbacks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-28T22:44:58.248394Z",
     "iopub.status.busy": "2021-05-28T22:44:58.248082Z",
     "iopub.status.idle": "2021-05-28T22:44:58.831281Z",
     "shell.execute_reply": "2021-05-28T22:44:58.830411Z",
     "shell.execute_reply.started": "2021-05-28T22:44:58.248359Z"
    }
   },
   "outputs": [],
   "source": [
    "# Text Pre-processing\n",
    "token_data = data.description\n",
    "maxlen = 1500 # Max number of frequently occuring words\n",
    "training_samples = 800\n",
    "validation_samples = 450\n",
    "max_words = 10000\n",
    "\n",
    "tokenizer = Tokenizer(num_words=max_words)\n",
    "tokenizer.fit_on_texts(token_data)\n",
    "seq = tokenizer.texts_to_sequences(token_data)\n",
    "word_index = tokenizer.word_index\n",
    "seq_data = pad_sequences(seq, maxlen=maxlen)\n",
    "\n",
    "# Scaling of Data\n",
    "scaler = MinMaxScaler()\n",
    "seq_data = scaler.fit_transform(seq_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-28T22:44:58.833129Z",
     "iopub.status.busy": "2021-05-28T22:44:58.832623Z",
     "iopub.status.idle": "2021-05-28T22:44:58.880441Z",
     "shell.execute_reply": "2021-05-28T22:44:58.879682Z",
     "shell.execute_reply.started": "2021-05-28T22:44:58.833089Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-28T22:44:58.882056Z",
     "iopub.status.busy": "2021-05-28T22:44:58.881699Z",
     "iopub.status.idle": "2021-05-28T22:44:58.927222Z",
     "shell.execute_reply": "2021-05-28T22:44:58.926226Z",
     "shell.execute_reply.started": "2021-05-28T22:44:58.882008Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training Set:- X_train: (6229, 1500) , y_train: (6229, 20)\n",
      "Validation Set:- X_val: (779, 1500) , y_val: (779, 20)\n",
      "Test Set:- X_test: (779, 1500) , y_test: (779, 20)\n"
     ]
    }
   ],
   "source": [
    "# Splitting of Data\n",
    "X_train, X_test, y_train, y_test = train_test_split(seq_data, y, test_size=0.2, random_state=56)\n",
    "X_val, X_test, y_val, y_test = train_test_split(X_test, y_test, test_size=0.5, random_state=56)\n",
    "\n",
    "print(\"Training Set:- X_train: {} , y_train: {}\".format(X_train.shape, y_train.shape))\n",
    "print(\"Validation Set:- X_val: {} , y_val: {}\".format(X_val.shape, y_val.shape))\n",
    "print(\"Test Set:- X_test: {} , y_test: {}\".format(X_test.shape, y_test.shape))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##  DNN Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-28T22:44:58.929046Z",
     "iopub.status.busy": "2021-05-28T22:44:58.928541Z",
     "iopub.status.idle": "2021-05-28T22:44:58.933510Z",
     "shell.execute_reply": "2021-05-28T22:44:58.932401Z",
     "shell.execute_reply.started": "2021-05-28T22:44:58.929004Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import classification_report, confusion_matrix\n",
    "from keras.utils import plot_model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-28T22:44:58.937379Z",
     "iopub.status.busy": "2021-05-28T22:44:58.936989Z",
     "iopub.status.idle": "2021-05-28T22:44:58.943282Z",
     "shell.execute_reply": "2021-05-28T22:44:58.942461Z",
     "shell.execute_reply.started": "2021-05-28T22:44:58.937347Z"
    }
   },
   "outputs": [],
   "source": [
    "# Baseline Model Hyper-parameters\n",
    "\n",
    "BATCH_SIZE = 128\n",
    "b_optimizer = 'rmsprop'\n",
    "EPOCHS = 50\n",
    "size = [X_train.shape[-1], 1000, 500, 20]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-28T22:44:58.945452Z",
     "iopub.status.busy": "2021-05-28T22:44:58.944937Z",
     "iopub.status.idle": "2021-05-28T22:44:58.960921Z",
     "shell.execute_reply": "2021-05-28T22:44:58.959739Z",
     "shell.execute_reply.started": "2021-05-28T22:44:58.945412Z"
    }
   },
   "outputs": [],
   "source": [
    "class DNNModel:\n",
    "    \"\"\"\n",
    "    A deep neural network class class to create custom dnn architecture which could be trained and evaluated\n",
    "    on any dataset.\n",
    "    \"\"\"\n",
    "    \n",
    "    def __init__(self, batch_size, epochs, size, optimizer, loss, activation_fn):\n",
    "        \n",
    "        \"\"\"\n",
    "        Parameters\n",
    "        ----------\n",
    "        \n",
    "        batch_size: size of a batch for training data\n",
    "        epochs:     number of epochs to train for\n",
    "        size:       list of output dimesnions for each layer\n",
    "        optimizer:  an optimizer algorithm could be optimizer object or string\n",
    "        loss:       loss function to be used for optimization\n",
    "        activation_fn: activation function for hidden layers\n",
    "        \"\"\"\n",
    "        \n",
    "        self.batch_size = batch_size\n",
    "        self.epochs = epochs\n",
    "        self.history = None\n",
    "        self.loss = loss\n",
    "        \n",
    "        # Model Building\n",
    "        n_layers = len(size) - 1\n",
    "\n",
    "        # Input Layer\n",
    "        input_data = Input(shape = (size[0], ), name='input')\n",
    "        x = input_data\n",
    "\n",
    "        # Hidden Layers\n",
    "        for n in range(n_layers - 1):\n",
    "            x = Dense(size[n+1], kernel_initializer = 'glorot_uniform', activation = activation_fn, name='dense_layer_{}'.format(n))(x)\n",
    "            if n != n_layers - 2:\n",
    "                x = Dropout(rate = 0.5, name='dropout_layer_{}'.format(n))(x)\n",
    "\n",
    "        # Latent Layer\n",
    "        out = Dense(size[-1], name='Output_layer', activation='softmax')(x)\n",
    "        \n",
    "        self.b_model = Model(inputs=input_data, outputs=out, name='basline_model')\n",
    "        self.b_model.compile(optimizer=optimizer, loss=self.loss)\n",
    "        \n",
    "    def fit(self, training_data, validation_data, verbose):\n",
    "        \"\"\"\n",
    "        This function fits the data to model.\n",
    "        \"\"\"\n",
    "        \n",
    "        self.history = self.b_model.fit(x=training_data[0], y=training_data[1],\n",
    "                                  validation_data = validation_data, epochs = self.epochs, batch_size = self.batch_size,\n",
    "                                  verbose = verbose)\n",
    "        \n",
    "    def plot_training(self):\n",
    "        \"\"\"\n",
    "        Plots the progression of training and validation loss over the epochs\n",
    "        \"\"\"\n",
    "        fig, ax = plt.subplots(1,1, dpi = 100)\n",
    "        ax.plot(self.history.history['loss'], label='Training Loss')\n",
    "        ax.plot(self.history.history['val_loss'], label='Validation Loss')\n",
    "        ax.set_xlabel('Epochs')\n",
    "        ax.set_ylabel(str(self.loss))\n",
    "        plt.legend()\n",
    "        plt.plot()\n",
    "        \n",
    "    def evaluation(self, test_data):\n",
    "        \"\"\"\n",
    "        Evaluates the resulting model on test data and prints relevant information\n",
    "        \"\"\"\n",
    "        \n",
    "        X_test = test_data[0]\n",
    "        y_test = test_data[1]\n",
    "        \n",
    "        y_pred_prob = self.b_model.predict(X_test)\n",
    "        y_pred = np.argmax(y_pred_prob, axis=1)\n",
    "        eval_metric = self.b_model.evaluate(x=X_test, y=y_test)\n",
    "        \n",
    "        print(classification_report(np.argmax(y_test, axis=1), y_pred))\n",
    "        \n",
    "        return (y_pred_prob, y_pred, eval_metric)\n",
    "    \n",
    "    def plot_model(self):\n",
    "        plot_model(self.b_model, to_file='baseline.png', show_shapes=True)\n",
    "        \n",
    "        \n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-28T22:44:58.962621Z",
     "iopub.status.busy": "2021-05-28T22:44:58.962195Z",
     "iopub.status.idle": "2021-05-28T22:45:01.099710Z",
     "shell.execute_reply": "2021-05-28T22:45:01.098856Z",
     "shell.execute_reply.started": "2021-05-28T22:44:58.962585Z"
    }
   },
   "outputs": [],
   "source": [
    "dnn_model = DNNModel(batch_size=BATCH_SIZE, epochs=EPOCHS, size=size, optimizer=b_optimizer, loss='categorical_crossentropy', activation_fn='relu')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-28T22:45:01.101319Z",
     "iopub.status.busy": "2021-05-28T22:45:01.100980Z",
     "iopub.status.idle": "2021-05-28T22:45:01.497297Z",
     "shell.execute_reply": "2021-05-28T22:45:01.496204Z",
     "shell.execute_reply.started": "2021-05-28T22:45:01.101284Z"
    }
   },
   "outputs": [],
   "source": [
    "dnn_model.plot_model()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-28T22:45:01.499192Z",
     "iopub.status.busy": "2021-05-28T22:45:01.498839Z",
     "iopub.status.idle": "2021-05-28T22:45:01.511608Z",
     "shell.execute_reply": "2021-05-28T22:45:01.510758Z",
     "shell.execute_reply.started": "2021-05-28T22:45:01.499150Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from IPython.display import Image\n",
    "Image(filename='baseline.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-28T22:45:01.513479Z",
     "iopub.status.busy": "2021-05-28T22:45:01.512979Z",
     "iopub.status.idle": "2021-05-28T22:45:13.561946Z",
     "shell.execute_reply": "2021-05-28T22:45:13.561077Z",
     "shell.execute_reply.started": "2021-05-28T22:45:01.513348Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/50\n",
      "49/49 [==============================] - 2s 14ms/step - loss: 2.1539 - val_loss: 0.7414\n",
      "Epoch 2/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.6897 - val_loss: 0.4852\n",
      "Epoch 3/50\n",
      "49/49 [==============================] - 0s 5ms/step - loss: 0.4867 - val_loss: 0.3872\n",
      "Epoch 4/50\n",
      "49/49 [==============================] - 0s 6ms/step - loss: 0.3964 - val_loss: 0.3605\n",
      "Epoch 5/50\n",
      "49/49 [==============================] - 0s 5ms/step - loss: 0.3464 - val_loss: 0.3331\n",
      "Epoch 6/50\n",
      "49/49 [==============================] - 0s 5ms/step - loss: 0.3442 - val_loss: 0.3435\n",
      "Epoch 7/50\n",
      "49/49 [==============================] - 0s 5ms/step - loss: 0.3095 - val_loss: 0.2839\n",
      "Epoch 8/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.2901 - val_loss: 0.2433\n",
      "Epoch 9/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.2693 - val_loss: 0.2876\n",
      "Epoch 10/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.2579 - val_loss: 0.3296\n",
      "Epoch 11/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.2463 - val_loss: 0.2291\n",
      "Epoch 12/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.2385 - val_loss: 0.2411\n",
      "Epoch 13/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.2285 - val_loss: 0.2225\n",
      "Epoch 14/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.2243 - val_loss: 0.2473\n",
      "Epoch 15/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.2142 - val_loss: 0.2064\n",
      "Epoch 16/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.2084 - val_loss: 0.2054\n",
      "Epoch 17/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.2129 - val_loss: 0.2508\n",
      "Epoch 18/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1936 - val_loss: 0.2427\n",
      "Epoch 19/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1742 - val_loss: 0.2577\n",
      "Epoch 20/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1868 - val_loss: 0.2061\n",
      "Epoch 21/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1736 - val_loss: 0.2059\n",
      "Epoch 22/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1639 - val_loss: 0.2460\n",
      "Epoch 23/50\n",
      "49/49 [==============================] - 0s 6ms/step - loss: 0.1559 - val_loss: 0.2047\n",
      "Epoch 24/50\n",
      "49/49 [==============================] - 0s 5ms/step - loss: 0.1645 - val_loss: 0.2105\n",
      "Epoch 25/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1542 - val_loss: 0.2087\n",
      "Epoch 26/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1400 - val_loss: 0.2227\n",
      "Epoch 27/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1494 - val_loss: 0.1923\n",
      "Epoch 28/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1412 - val_loss: 0.1974\n",
      "Epoch 29/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1493 - val_loss: 0.2033\n",
      "Epoch 30/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1471 - val_loss: 0.2183\n",
      "Epoch 31/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1341 - val_loss: 0.2442\n",
      "Epoch 32/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1295 - val_loss: 0.1971\n",
      "Epoch 33/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1156 - val_loss: 0.2083\n",
      "Epoch 34/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1260 - val_loss: 0.2386\n",
      "Epoch 35/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1182 - val_loss: 0.2252\n",
      "Epoch 36/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1032 - val_loss: 0.2083\n",
      "Epoch 37/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1068 - val_loss: 0.2583\n",
      "Epoch 38/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1144 - val_loss: 0.1934\n",
      "Epoch 39/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1021 - val_loss: 0.2106\n",
      "Epoch 40/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1045 - val_loss: 0.2120\n",
      "Epoch 41/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1000 - val_loss: 0.2021\n",
      "Epoch 42/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.0991 - val_loss: 0.2663\n",
      "Epoch 43/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.0972 - val_loss: 0.2445\n",
      "Epoch 44/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.0906 - val_loss: 0.2545\n",
      "Epoch 45/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.1032 - val_loss: 0.2147\n",
      "Epoch 46/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.0910 - val_loss: 0.2359\n",
      "Epoch 47/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.0862 - val_loss: 0.2334\n",
      "Epoch 48/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.0883 - val_loss: 0.2053\n",
      "Epoch 49/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.0861 - val_loss: 0.3067\n",
      "Epoch 50/50\n",
      "49/49 [==============================] - 0s 4ms/step - loss: 0.0859 - val_loss: 0.2582\n"
     ]
    }
   ],
   "source": [
    "dnn_model.fit(training_data=(X_train, y_train), validation_data=(X_val, y_val), verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-28T22:45:13.565257Z",
     "iopub.status.busy": "2021-05-28T22:45:13.564993Z",
     "iopub.status.idle": "2021-05-28T22:45:13.754359Z",
     "shell.execute_reply": "2021-05-28T22:45:13.753500Z",
     "shell.execute_reply.started": "2021-05-28T22:45:13.565230Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 600x400 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "dnn_model.plot_training()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-28T22:45:13.756142Z",
     "iopub.status.busy": "2021-05-28T22:45:13.755633Z",
     "iopub.status.idle": "2021-05-28T22:45:13.967650Z",
     "shell.execute_reply": "2021-05-28T22:45:13.966594Z",
     "shell.execute_reply.started": "2021-05-28T22:45:13.756091Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "25/25 [==============================] - 0s 2ms/step - loss: 0.2680\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.97      0.88      0.92        32\n",
      "           1       0.97      0.80      0.88        41\n",
      "           2       0.82      1.00      0.90        23\n",
      "           3       0.96      0.92      0.94        50\n",
      "           4       0.85      0.81      0.83        21\n",
      "           5       0.96      0.82      0.89        33\n",
      "           6       0.87      0.92      0.89        36\n",
      "           7       0.94      0.93      0.94        72\n",
      "           8       0.94      0.83      0.88        36\n",
      "           9       0.88      0.95      0.91        22\n",
      "          10       1.00      0.94      0.97        48\n",
      "          11       0.90      1.00      0.95        37\n",
      "          12       0.88      0.93      0.90        30\n",
      "          13       0.93      1.00      0.96        37\n",
      "          14       0.86      0.86      0.86        29\n",
      "          15       0.86      0.98      0.92        50\n",
      "          16       0.95      0.93      0.94        40\n",
      "          17       0.89      0.94      0.91        33\n",
      "          18       0.89      0.92      0.90        83\n",
      "          19       0.96      0.88      0.92        26\n",
      "\n",
      "    accuracy                           0.92       779\n",
      "   macro avg       0.91      0.91      0.91       779\n",
      "weighted avg       0.92      0.92      0.91       779\n",
      "\n"
     ]
    }
   ],
   "source": [
    "a, b, c = dnn_model.evaluation(test_data=(X_test, y_test))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Sequential Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-29T01:39:18.602909Z",
     "iopub.status.busy": "2021-05-29T01:39:18.602582Z",
     "iopub.status.idle": "2021-05-29T01:39:18.617336Z",
     "shell.execute_reply": "2021-05-29T01:39:18.615033Z",
     "shell.execute_reply.started": "2021-05-29T01:39:18.602874Z"
    }
   },
   "outputs": [],
   "source": [
    "class RecurrentNetwork:\n",
    "     \"\"\"\n",
    "    A recurrent neural network class class to create custom RNN architecture which could be trained and evaluated\n",
    "    on any dataset.\n",
    "    \"\"\"\n",
    "    \n",
    "    def __init__(self, layer, epoch, batch_size, optimizer, loss, input_param, name, size):\n",
    "        \n",
    "        \"\"\"\n",
    "        Parameters\n",
    "        -----------\n",
    "        \n",
    "        layer: type of recurrent layer: SimpleRNN, LSTM, GRU\n",
    "        batch_size: size of a batch for training data\n",
    "        epochs:     number of epochs to train for\n",
    "        size:       list of output dimesnions for each layer\n",
    "        optimizer:  an optimizer algorithm could be optimizer object or string\n",
    "        loss:       loss function to be used for optimization\n",
    "        name:       name of the model\n",
    "        input_param: array of [maclen, max_words]\n",
    "        \"\"\"\n",
    "        \n",
    "        self.BATCH_SIZE = batch_size\n",
    "        self.EPOCH = epoch\n",
    "        self.history = None\n",
    "        self.loss = loss\n",
    "        \n",
    "        # Model Building\n",
    "        inputs = Input(name='inputs',shape=[input_param[0]])\n",
    "        emb = Embedding(input_param[1],500,input_length=input_param[0])(inputs)\n",
    "        x = emb\n",
    "        \n",
    "        n_layers = len(size)\n",
    "        \n",
    "        for i in range(n_layers-1):\n",
    "            x = layer(units = size[i], return_sequences = True, name='LSTM_layer_{}'.format(i+1))(x)\n",
    "        \n",
    "        x = layer(units = size[-1])(x)\n",
    "        x = Dense(200, activation='relu')(x)\n",
    "        out = Dense(20, activation='softmax')(x)\n",
    "        \n",
    "        self.model = Model(inputs = inputs, outputs = out, name = name)\n",
    "        self.model.compile(optimizer=optimizer, loss=self.loss)\n",
    "        \n",
    "    \n",
    "    def fit(self, training_data, validation_data, verbose, callbacks):\n",
    "        \"\"\"\n",
    "        Fits the model on training_data along with specified callbacks\n",
    "        \"\"\"\n",
    "        \n",
    "        self.history = self.model.fit(x=training_data[0], y=training_data[1],\n",
    "                                  validation_data = validation_data, epochs = self.EPOCH, batch_size = self.BATCH_SIZE,\n",
    "                                  verbose = verbose, callbacks=callbacks)\n",
    "        \n",
    "    def plot_training(self):\n",
    "        \"\"\"\n",
    "        Plots the progression of training and validation loss over the epochs\n",
    "        \"\"\"\n",
    "        \n",
    "        fig, ax = plt.subplots(1,1, dpi = 100)\n",
    "        ax.plot(self.history.history['loss'], label='Training Loss')\n",
    "        ax.plot(self.history.history['val_loss'], label='Validation Loss')\n",
    "        ax.set_xlabel('Epochs')\n",
    "        ax.set_ylabel(str(self.loss))\n",
    "        plt.legend()\n",
    "        plt.plot()\n",
    "        \n",
    "    def evaluation(self, test_data):\n",
    "        \"\"\"\n",
    "        Evaluates the resulting model on test data and prints relevant information\n",
    "        \"\"\"\n",
    "        \n",
    "        X_test = test_data[0]\n",
    "        y_test = test_data[1]\n",
    "        \n",
    "        y_pred_prob = self.model.predict(X_test)\n",
    "        y_pred = np.argmax(y_pred_prob, axis=1)\n",
    "        eval_metric = self.model.evaluate(x=X_test, y=y_test)\n",
    "        \n",
    "        print(classification_report(np.argmax(y_test, axis=1), y_pred))\n",
    "        \n",
    "        return (y_pred_prob, y_pred, eval_metric)\n",
    "    \n",
    "    def plot_model(self):\n",
    "        \"\"\"\n",
    "        Plots the model architecture\n",
    "        \"\"\"\n",
    "        \n",
    "        plot_model(self.model, to_file='recurrent.png', show_shapes=True)\n",
    "         "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-29T01:39:19.755501Z",
     "iopub.status.busy": "2021-05-29T01:39:19.755182Z",
     "iopub.status.idle": "2021-05-29T01:39:19.759474Z",
     "shell.execute_reply": "2021-05-29T01:39:19.758475Z",
     "shell.execute_reply.started": "2021-05-29T01:39:19.755472Z"
    }
   },
   "outputs": [],
   "source": [
    "input_param = [maxlen, max_words]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-29T03:28:13.535428Z",
     "iopub.status.busy": "2021-05-29T03:28:13.535101Z",
     "iopub.status.idle": "2021-05-29T03:28:13.541219Z",
     "shell.execute_reply": "2021-05-29T03:28:13.540372Z",
     "shell.execute_reply.started": "2021-05-29T03:28:13.535395Z"
    }
   },
   "outputs": [],
   "source": [
    "seq_checkpoint = callbacks.ModelCheckpoint(filepath = 'lstm.hdf5', save_weights_only= True,\n",
    "                                                  save_best_only= True, monitor = 'val_loss', mode = 'min', verbose = 1)\n",
    "seq_early_stopping = callbacks.EarlyStopping(monitor = 'val_loss', patience = 5, verbose = 1, mode = 'min')\n",
    "reduce_lr = callbacks.ReduceLROnPlateau(monitor = 'val_loss', patience = 3, verbose = 1, mode = 'min', factor = 0.1)\n",
    "seq_callbacks = [seq_checkpoint, seq_early_stopping, reduce_lr]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-29T03:28:27.234827Z",
     "iopub.status.busy": "2021-05-29T03:28:27.234454Z",
     "iopub.status.idle": "2021-05-29T03:28:29.467635Z",
     "shell.execute_reply": "2021-05-29T03:28:29.465090Z",
     "shell.execute_reply.started": "2021-05-29T03:28:27.234789Z"
    }
   },
   "outputs": [],
   "source": [
    "# Deep LSTM Network\n",
    "optimizer = SGD(lr=0.1, momentum=0.9)\n",
    "lstm_rnn = RecurrentNetwork(layer = LSTM, epoch=100, batch_size=128, optimizer = optimizer, loss='categorical_crossentropy', \n",
    "                              input_param = input_param, name='LSTM_RNN', size = [1000, 500, 250])\n",
    "lstm_rnn.plot_model()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-29T03:28:31.359013Z",
     "iopub.status.busy": "2021-05-29T03:28:31.358641Z",
     "iopub.status.idle": "2021-05-29T03:28:31.368024Z",
     "shell.execute_reply": "2021-05-29T03:28:31.366965Z",
     "shell.execute_reply.started": "2021-05-29T03:28:31.358974Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 108,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Image(filename='recurrent.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-29T03:28:35.211918Z",
     "iopub.status.busy": "2021-05-29T03:28:35.211585Z",
     "iopub.status.idle": "2021-05-29T04:00:11.616297Z",
     "shell.execute_reply": "2021-05-29T04:00:11.615517Z",
     "shell.execute_reply.started": "2021-05-29T03:28:35.211887Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "49/49 [==============================] - 115s 2s/step - loss: 2.9533 - val_loss: 2.9072\n",
      "\n",
      "Epoch 00001: val_loss improved from inf to 2.90723, saving model to lstm.hdf5\n",
      "Epoch 2/100\n",
      "49/49 [==============================] - 111s 2s/step - loss: 2.9235 - val_loss: 2.9038\n",
      "\n",
      "Epoch 00002: val_loss improved from 2.90723 to 2.90375, saving model to lstm.hdf5\n",
      "Epoch 3/100\n",
      "49/49 [==============================] - 111s 2s/step - loss: 2.9218 - val_loss: 2.9072\n",
      "\n",
      "Epoch 00003: val_loss did not improve from 2.90375\n",
      "Epoch 4/100\n",
      "49/49 [==============================] - 111s 2s/step - loss: 2.9244 - val_loss: 2.9028\n",
      "\n",
      "Epoch 00004: val_loss improved from 2.90375 to 2.90284, saving model to lstm.hdf5\n",
      "Epoch 5/100\n",
      "49/49 [==============================] - 111s 2s/step - loss: 2.9228 - val_loss: 2.9030\n",
      "\n",
      "Epoch 00005: val_loss did not improve from 2.90284\n",
      "Epoch 6/100\n",
      "49/49 [==============================] - 111s 2s/step - loss: 2.9136 - val_loss: 2.9027\n",
      "\n",
      "Epoch 00006: val_loss improved from 2.90284 to 2.90272, saving model to lstm.hdf5\n",
      "Epoch 7/100\n",
      "49/49 [==============================] - 111s 2s/step - loss: 2.9126 - val_loss: 2.9042\n",
      "\n",
      "Epoch 00007: val_loss did not improve from 2.90272\n",
      "Epoch 8/100\n",
      "49/49 [==============================] - 111s 2s/step - loss: 2.9271 - val_loss: 2.9021\n",
      "\n",
      "Epoch 00008: val_loss improved from 2.90272 to 2.90207, saving model to lstm.hdf5\n",
      "Epoch 9/100\n",
      "49/49 [==============================] - 111s 2s/step - loss: 2.9319 - val_loss: 2.9014\n",
      "\n",
      "Epoch 00009: val_loss improved from 2.90207 to 2.90144, saving model to lstm.hdf5\n",
      "Epoch 10/100\n",
      "49/49 [==============================] - 111s 2s/step - loss: 2.9202 - val_loss: 2.9012\n",
      "\n",
      "Epoch 00010: val_loss improved from 2.90144 to 2.90116, saving model to lstm.hdf5\n",
      "Epoch 11/100\n",
      "49/49 [==============================] - 111s 2s/step - loss: 2.9143 - val_loss: 2.9014\n",
      "\n",
      "Epoch 00011: val_loss did not improve from 2.90116\n",
      "Epoch 12/100\n",
      "49/49 [==============================] - 111s 2s/step - loss: 2.9272 - val_loss: 2.9009\n",
      "\n",
      "Epoch 00012: val_loss improved from 2.90116 to 2.90091, saving model to lstm.hdf5\n",
      "Epoch 13/100\n",
      "49/49 [==============================] - 111s 2s/step - loss: 2.9176 - val_loss: 2.9032\n",
      "\n",
      "Epoch 00013: val_loss did not improve from 2.90091\n",
      "Epoch 14/100\n",
      "49/49 [==============================] - 111s 2s/step - loss: 2.9218 - val_loss: 2.9011\n",
      "\n",
      "Epoch 00014: val_loss did not improve from 2.90091\n",
      "Epoch 15/100\n",
      "49/49 [==============================] - 111s 2s/step - loss: 2.9199 - val_loss: 2.9032\n",
      "\n",
      "Epoch 00015: val_loss did not improve from 2.90091\n",
      "\n",
      "Epoch 00015: ReduceLROnPlateau reducing learning rate to 0.010000000149011612.\n",
      "Epoch 16/100\n",
      "49/49 [==============================] - 111s 2s/step - loss: 2.9179 - val_loss: 2.9050\n",
      "\n",
      "Epoch 00016: val_loss did not improve from 2.90091\n",
      "Epoch 17/100\n",
      "49/49 [==============================] - 111s 2s/step - loss: 2.9166 - val_loss: 2.9038\n",
      "\n",
      "Epoch 00017: val_loss did not improve from 2.90091\n",
      "Epoch 00017: early stopping\n"
     ]
    }
   ],
   "source": [
    "lstm_rnn.fit(training_data=(X_train, y_train), validation_data=(X_val, y_val), verbose=1, callbacks=seq_callbacks)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-29T02:23:22.448486Z",
     "iopub.status.busy": "2021-05-29T02:23:22.448116Z",
     "iopub.status.idle": "2021-05-29T02:23:22.608440Z",
     "shell.execute_reply": "2021-05-29T02:23:22.607495Z",
     "shell.execute_reply.started": "2021-05-29T02:23:22.448454Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 600x400 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "lstm_rnn.plot_training()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-29T04:01:07.526016Z",
     "iopub.status.busy": "2021-05-29T04:01:07.525614Z",
     "iopub.status.idle": "2021-05-29T05:56:10.777910Z",
     "shell.execute_reply": "2021-05-29T05:56:10.777022Z",
     "shell.execute_reply.started": "2021-05-29T04:01:07.525983Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "49/49 [==============================] - 71s 1s/step - loss: 2.9980 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00001: val_loss improved from inf to 2.99753, saving model to shallow_lstm.hdf5\n",
      "Epoch 2/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9977 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00002: val_loss improved from 2.99753 to 2.99753, saving model to shallow_lstm.hdf5\n",
      "Epoch 3/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9978 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00003: val_loss did not improve from 2.99753\n",
      "Epoch 4/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9977 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00004: val_loss improved from 2.99753 to 2.99753, saving model to shallow_lstm.hdf5\n",
      "Epoch 5/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9977 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00005: val_loss improved from 2.99753 to 2.99753, saving model to shallow_lstm.hdf5\n",
      "Epoch 6/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9977 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00006: val_loss improved from 2.99753 to 2.99753, saving model to shallow_lstm.hdf5\n",
      "Epoch 7/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9975 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00007: val_loss improved from 2.99753 to 2.99753, saving model to shallow_lstm.hdf5\n",
      "Epoch 8/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9976 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00008: val_loss improved from 2.99753 to 2.99753, saving model to shallow_lstm.hdf5\n",
      "Epoch 9/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9975 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00009: val_loss improved from 2.99753 to 2.99753, saving model to shallow_lstm.hdf5\n",
      "Epoch 10/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9979 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00010: val_loss improved from 2.99753 to 2.99753, saving model to shallow_lstm.hdf5\n",
      "Epoch 11/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9977 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00011: val_loss improved from 2.99753 to 2.99753, saving model to shallow_lstm.hdf5\n",
      "Epoch 12/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9976 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00012: val_loss improved from 2.99753 to 2.99752, saving model to shallow_lstm.hdf5\n",
      "Epoch 13/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9976 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00013: val_loss improved from 2.99752 to 2.99752, saving model to shallow_lstm.hdf5\n",
      "Epoch 14/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9980 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00014: val_loss improved from 2.99752 to 2.99752, saving model to shallow_lstm.hdf5\n",
      "Epoch 15/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9980 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00015: val_loss improved from 2.99752 to 2.99752, saving model to shallow_lstm.hdf5\n",
      "Epoch 16/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9977 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00016: val_loss improved from 2.99752 to 2.99752, saving model to shallow_lstm.hdf5\n",
      "Epoch 17/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9980 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00017: val_loss improved from 2.99752 to 2.99752, saving model to shallow_lstm.hdf5\n",
      "Epoch 18/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9977 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00018: val_loss improved from 2.99752 to 2.99752, saving model to shallow_lstm.hdf5\n",
      "Epoch 19/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9978 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00019: val_loss improved from 2.99752 to 2.99752, saving model to shallow_lstm.hdf5\n",
      "Epoch 20/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9980 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00020: val_loss improved from 2.99752 to 2.99751, saving model to shallow_lstm.hdf5\n",
      "Epoch 21/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9977 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00021: val_loss improved from 2.99751 to 2.99751, saving model to shallow_lstm.hdf5\n",
      "Epoch 22/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9977 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00022: val_loss improved from 2.99751 to 2.99751, saving model to shallow_lstm.hdf5\n",
      "Epoch 23/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9978 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00023: val_loss improved from 2.99751 to 2.99751, saving model to shallow_lstm.hdf5\n",
      "Epoch 24/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9976 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00024: val_loss improved from 2.99751 to 2.99750, saving model to shallow_lstm.hdf5\n",
      "Epoch 25/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9978 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00025: val_loss improved from 2.99750 to 2.99750, saving model to shallow_lstm.hdf5\n",
      "Epoch 26/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9977 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00026: val_loss improved from 2.99750 to 2.99750, saving model to shallow_lstm.hdf5\n",
      "Epoch 27/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9974 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00027: val_loss improved from 2.99750 to 2.99749, saving model to shallow_lstm.hdf5\n",
      "Epoch 28/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9976 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00028: val_loss improved from 2.99749 to 2.99749, saving model to shallow_lstm.hdf5\n",
      "Epoch 29/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9977 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00029: val_loss improved from 2.99749 to 2.99748, saving model to shallow_lstm.hdf5\n",
      "Epoch 30/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9977 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00030: val_loss improved from 2.99748 to 2.99748, saving model to shallow_lstm.hdf5\n",
      "Epoch 31/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9976 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00031: val_loss improved from 2.99748 to 2.99747, saving model to shallow_lstm.hdf5\n",
      "Epoch 32/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9977 - val_loss: 2.9975\n",
      "\n",
      "Epoch 00032: val_loss improved from 2.99747 to 2.99746, saving model to shallow_lstm.hdf5\n",
      "Epoch 33/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9975 - val_loss: 2.9974\n",
      "\n",
      "Epoch 00033: val_loss improved from 2.99746 to 2.99745, saving model to shallow_lstm.hdf5\n",
      "Epoch 34/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9978 - val_loss: 2.9974\n",
      "\n",
      "Epoch 00034: val_loss improved from 2.99745 to 2.99744, saving model to shallow_lstm.hdf5\n",
      "Epoch 35/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9977 - val_loss: 2.9974\n",
      "\n",
      "Epoch 00035: val_loss improved from 2.99744 to 2.99743, saving model to shallow_lstm.hdf5\n",
      "Epoch 36/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9976 - val_loss: 2.9974\n",
      "\n",
      "Epoch 00036: val_loss improved from 2.99743 to 2.99741, saving model to shallow_lstm.hdf5\n",
      "Epoch 37/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9977 - val_loss: 2.9974\n",
      "\n",
      "Epoch 00037: val_loss improved from 2.99741 to 2.99740, saving model to shallow_lstm.hdf5\n",
      "Epoch 38/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9973 - val_loss: 2.9974\n",
      "\n",
      "Epoch 00038: val_loss improved from 2.99740 to 2.99738, saving model to shallow_lstm.hdf5\n",
      "Epoch 39/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9975 - val_loss: 2.9974\n",
      "\n",
      "Epoch 00039: val_loss improved from 2.99738 to 2.99736, saving model to shallow_lstm.hdf5\n",
      "Epoch 40/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9976 - val_loss: 2.9973\n",
      "\n",
      "Epoch 00040: val_loss improved from 2.99736 to 2.99734, saving model to shallow_lstm.hdf5\n",
      "Epoch 41/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9976 - val_loss: 2.9973\n",
      "\n",
      "Epoch 00041: val_loss improved from 2.99734 to 2.99731, saving model to shallow_lstm.hdf5\n",
      "Epoch 42/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9976 - val_loss: 2.9973\n",
      "\n",
      "Epoch 00042: val_loss improved from 2.99731 to 2.99728, saving model to shallow_lstm.hdf5\n",
      "Epoch 43/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9974 - val_loss: 2.9972\n",
      "\n",
      "Epoch 00043: val_loss improved from 2.99728 to 2.99725, saving model to shallow_lstm.hdf5\n",
      "Epoch 44/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9976 - val_loss: 2.9972\n",
      "\n",
      "Epoch 00044: val_loss improved from 2.99725 to 2.99721, saving model to shallow_lstm.hdf5\n",
      "Epoch 45/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9973 - val_loss: 2.9972\n",
      "\n",
      "Epoch 00045: val_loss improved from 2.99721 to 2.99717, saving model to shallow_lstm.hdf5\n",
      "Epoch 46/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9975 - val_loss: 2.9971\n",
      "\n",
      "Epoch 00046: val_loss improved from 2.99717 to 2.99712, saving model to shallow_lstm.hdf5\n",
      "Epoch 47/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9975 - val_loss: 2.9971\n",
      "\n",
      "Epoch 00047: val_loss improved from 2.99712 to 2.99707, saving model to shallow_lstm.hdf5\n",
      "Epoch 48/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9973 - val_loss: 2.9970\n",
      "\n",
      "Epoch 00048: val_loss improved from 2.99707 to 2.99701, saving model to shallow_lstm.hdf5\n",
      "Epoch 49/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9972 - val_loss: 2.9969\n",
      "\n",
      "Epoch 00049: val_loss improved from 2.99701 to 2.99695, saving model to shallow_lstm.hdf5\n",
      "Epoch 50/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9973 - val_loss: 2.9969\n",
      "\n",
      "Epoch 00050: val_loss improved from 2.99695 to 2.99687, saving model to shallow_lstm.hdf5\n",
      "Epoch 51/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9968 - val_loss: 2.9968\n",
      "\n",
      "Epoch 00051: val_loss improved from 2.99687 to 2.99679, saving model to shallow_lstm.hdf5\n",
      "Epoch 52/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9971 - val_loss: 2.9967\n",
      "\n",
      "Epoch 00052: val_loss improved from 2.99679 to 2.99669, saving model to shallow_lstm.hdf5\n",
      "Epoch 53/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9971 - val_loss: 2.9966\n",
      "\n",
      "Epoch 00053: val_loss improved from 2.99669 to 2.99659, saving model to shallow_lstm.hdf5\n",
      "Epoch 54/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9969 - val_loss: 2.9965\n",
      "\n",
      "Epoch 00054: val_loss improved from 2.99659 to 2.99648, saving model to shallow_lstm.hdf5\n",
      "Epoch 55/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9969 - val_loss: 2.9963\n",
      "\n",
      "Epoch 00055: val_loss improved from 2.99648 to 2.99635, saving model to shallow_lstm.hdf5\n",
      "Epoch 56/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9967 - val_loss: 2.9962\n",
      "\n",
      "Epoch 00056: val_loss improved from 2.99635 to 2.99621, saving model to shallow_lstm.hdf5\n",
      "Epoch 57/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9966 - val_loss: 2.9960\n",
      "\n",
      "Epoch 00057: val_loss improved from 2.99621 to 2.99605, saving model to shallow_lstm.hdf5\n",
      "Epoch 58/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9963 - val_loss: 2.9959\n",
      "\n",
      "Epoch 00058: val_loss improved from 2.99605 to 2.99587, saving model to shallow_lstm.hdf5\n",
      "Epoch 59/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9965 - val_loss: 2.9957\n",
      "\n",
      "Epoch 00059: val_loss improved from 2.99587 to 2.99567, saving model to shallow_lstm.hdf5\n",
      "Epoch 60/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9962 - val_loss: 2.9954\n",
      "\n",
      "Epoch 00060: val_loss improved from 2.99567 to 2.99545, saving model to shallow_lstm.hdf5\n",
      "Epoch 61/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9956 - val_loss: 2.9952\n",
      "\n",
      "Epoch 00061: val_loss improved from 2.99545 to 2.99520, saving model to shallow_lstm.hdf5\n",
      "Epoch 62/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9959 - val_loss: 2.9949\n",
      "\n",
      "Epoch 00062: val_loss improved from 2.99520 to 2.99492, saving model to shallow_lstm.hdf5\n",
      "Epoch 63/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9955 - val_loss: 2.9946\n",
      "\n",
      "Epoch 00063: val_loss improved from 2.99492 to 2.99461, saving model to shallow_lstm.hdf5\n",
      "Epoch 64/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9950 - val_loss: 2.9943\n",
      "\n",
      "Epoch 00064: val_loss improved from 2.99461 to 2.99427, saving model to shallow_lstm.hdf5\n",
      "Epoch 65/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9946 - val_loss: 2.9939\n",
      "\n",
      "Epoch 00065: val_loss improved from 2.99427 to 2.99388, saving model to shallow_lstm.hdf5\n",
      "Epoch 66/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9944 - val_loss: 2.9935\n",
      "\n",
      "Epoch 00066: val_loss improved from 2.99388 to 2.99346, saving model to shallow_lstm.hdf5\n",
      "Epoch 67/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9941 - val_loss: 2.9930\n",
      "\n",
      "Epoch 00067: val_loss improved from 2.99346 to 2.99299, saving model to shallow_lstm.hdf5\n",
      "Epoch 68/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9935 - val_loss: 2.9925\n",
      "\n",
      "Epoch 00068: val_loss improved from 2.99299 to 2.99245, saving model to shallow_lstm.hdf5\n",
      "Epoch 69/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9933 - val_loss: 2.9919\n",
      "\n",
      "Epoch 00069: val_loss improved from 2.99245 to 2.99189, saving model to shallow_lstm.hdf5\n",
      "Epoch 70/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9926 - val_loss: 2.9912\n",
      "\n",
      "Epoch 00070: val_loss improved from 2.99189 to 2.99123, saving model to shallow_lstm.hdf5\n",
      "Epoch 71/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9925 - val_loss: 2.9905\n",
      "\n",
      "Epoch 00071: val_loss improved from 2.99123 to 2.99054, saving model to shallow_lstm.hdf5\n",
      "Epoch 72/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9915 - val_loss: 2.9898\n",
      "\n",
      "Epoch 00072: val_loss improved from 2.99054 to 2.98976, saving model to shallow_lstm.hdf5\n",
      "Epoch 73/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9910 - val_loss: 2.9889\n",
      "\n",
      "Epoch 00073: val_loss improved from 2.98976 to 2.98893, saving model to shallow_lstm.hdf5\n",
      "Epoch 74/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9904 - val_loss: 2.9880\n",
      "\n",
      "Epoch 00074: val_loss improved from 2.98893 to 2.98804, saving model to shallow_lstm.hdf5\n",
      "Epoch 75/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9895 - val_loss: 2.9870\n",
      "\n",
      "Epoch 00075: val_loss improved from 2.98804 to 2.98704, saving model to shallow_lstm.hdf5\n",
      "Epoch 76/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9884 - val_loss: 2.9860\n",
      "\n",
      "Epoch 00076: val_loss improved from 2.98704 to 2.98596, saving model to shallow_lstm.hdf5\n",
      "Epoch 77/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9874 - val_loss: 2.9847\n",
      "\n",
      "Epoch 00077: val_loss improved from 2.98596 to 2.98472, saving model to shallow_lstm.hdf5\n",
      "Epoch 78/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9867 - val_loss: 2.9834\n",
      "\n",
      "Epoch 00078: val_loss improved from 2.98472 to 2.98343, saving model to shallow_lstm.hdf5\n",
      "Epoch 79/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9852 - val_loss: 2.9820\n",
      "\n",
      "Epoch 00079: val_loss improved from 2.98343 to 2.98202, saving model to shallow_lstm.hdf5\n",
      "Epoch 80/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9828 - val_loss: 2.9804\n",
      "\n",
      "Epoch 00080: val_loss improved from 2.98202 to 2.98045, saving model to shallow_lstm.hdf5\n",
      "Epoch 81/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9820 - val_loss: 2.9789\n",
      "\n",
      "Epoch 00081: val_loss improved from 2.98045 to 2.97891, saving model to shallow_lstm.hdf5\n",
      "Epoch 82/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9817 - val_loss: 2.9773\n",
      "\n",
      "Epoch 00082: val_loss improved from 2.97891 to 2.97727, saving model to shallow_lstm.hdf5\n",
      "Epoch 83/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9783 - val_loss: 2.9754\n",
      "\n",
      "Epoch 00083: val_loss improved from 2.97727 to 2.97542, saving model to shallow_lstm.hdf5\n",
      "Epoch 84/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9777 - val_loss: 2.9735\n",
      "\n",
      "Epoch 00084: val_loss improved from 2.97542 to 2.97348, saving model to shallow_lstm.hdf5\n",
      "Epoch 85/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9764 - val_loss: 2.9713\n",
      "\n",
      "Epoch 00085: val_loss improved from 2.97348 to 2.97135, saving model to shallow_lstm.hdf5\n",
      "Epoch 86/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9734 - val_loss: 2.9690\n",
      "\n",
      "Epoch 00086: val_loss improved from 2.97135 to 2.96897, saving model to shallow_lstm.hdf5\n",
      "Epoch 87/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9720 - val_loss: 2.9663\n",
      "\n",
      "Epoch 00087: val_loss improved from 2.96897 to 2.96635, saving model to shallow_lstm.hdf5\n",
      "Epoch 88/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9712 - val_loss: 2.9635\n",
      "\n",
      "Epoch 00088: val_loss improved from 2.96635 to 2.96346, saving model to shallow_lstm.hdf5\n",
      "Epoch 89/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9678 - val_loss: 2.9601\n",
      "\n",
      "Epoch 00089: val_loss improved from 2.96346 to 2.96014, saving model to shallow_lstm.hdf5\n",
      "Epoch 90/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9638 - val_loss: 2.9566\n",
      "\n",
      "Epoch 00090: val_loss improved from 2.96014 to 2.95656, saving model to shallow_lstm.hdf5\n",
      "Epoch 91/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9610 - val_loss: 2.9525\n",
      "\n",
      "Epoch 00091: val_loss improved from 2.95656 to 2.95250, saving model to shallow_lstm.hdf5\n",
      "Epoch 92/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9558 - val_loss: 2.9480\n",
      "\n",
      "Epoch 00092: val_loss improved from 2.95250 to 2.94799, saving model to shallow_lstm.hdf5\n",
      "Epoch 93/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9508 - val_loss: 2.9432\n",
      "\n",
      "Epoch 00093: val_loss improved from 2.94799 to 2.94318, saving model to shallow_lstm.hdf5\n",
      "Epoch 94/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9529 - val_loss: 2.9382\n",
      "\n",
      "Epoch 00094: val_loss improved from 2.94318 to 2.93818, saving model to shallow_lstm.hdf5\n",
      "Epoch 95/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9433 - val_loss: 2.9327\n",
      "\n",
      "Epoch 00095: val_loss improved from 2.93818 to 2.93272, saving model to shallow_lstm.hdf5\n",
      "Epoch 96/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9421 - val_loss: 2.9275\n",
      "\n",
      "Epoch 00096: val_loss improved from 2.93272 to 2.92753, saving model to shallow_lstm.hdf5\n",
      "Epoch 97/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9389 - val_loss: 2.9221\n",
      "\n",
      "Epoch 00097: val_loss improved from 2.92753 to 2.92210, saving model to shallow_lstm.hdf5\n",
      "Epoch 98/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9321 - val_loss: 2.9169\n",
      "\n",
      "Epoch 00098: val_loss improved from 2.92210 to 2.91692, saving model to shallow_lstm.hdf5\n",
      "Epoch 99/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9244 - val_loss: 2.9123\n",
      "\n",
      "Epoch 00099: val_loss improved from 2.91692 to 2.91232, saving model to shallow_lstm.hdf5\n",
      "Epoch 100/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9316 - val_loss: 2.9084\n",
      "\n",
      "Epoch 00100: val_loss improved from 2.91232 to 2.90842, saving model to shallow_lstm.hdf5\n"
     ]
    }
   ],
   "source": [
    "# Shallow LSTM network\n",
    "seq_checkpoint = callbacks.ModelCheckpoint(filepath = 'shallow_lstm.hdf5', save_weights_only= True,\n",
    "                                                  save_best_only= True, monitor = 'val_loss', mode = 'min', verbose = 1)\n",
    "shallow_seq_callbacks = [seq_checkpoint, seq_early_stopping, lr_scheduler]\n",
    "\n",
    "shallow_lstm = RecurrentNetwork(layer = LSTM, epoch=100, batch_size=128, optimizer = optimizer, loss='categorical_crossentropy', \n",
    "                              input_param = input_param, name='LSTM_RNN', size = [1000])\n",
    "\n",
    "shallow_lstm.fit(training_data=(X_train, y_train), validation_data=(X_val, y_val), verbose=1, callbacks=shallow_seq_callbacks)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-29T05:58:59.006379Z",
     "iopub.status.busy": "2021-05-29T05:58:59.006051Z",
     "iopub.status.idle": "2021-05-29T05:58:59.168459Z",
     "shell.execute_reply": "2021-05-29T05:58:59.167486Z",
     "shell.execute_reply.started": "2021-05-29T05:58:59.006345Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAkQAAAIECAIAAAB32qZhAAAABmJLR0QA/wD/AP+gvaeTAAAgAElEQVR4nOzde1wT57Y4/GcSEq4hIKh4AZGmCqLSixW5GPEAEs+hCkiLVfCGHP2gnlal1W1xA0p3dddbP9KNW6lYqlbAoogFRFQuUdhg641yqIKA9RQrIBAitxDy/vH8nHd2gmEICUlgff8yMyvDyoxkMc/MPIuQyWQIAAAA0GcMbScAAAAADBUUMwAAAHoPihkAAAC9B8UMAACA3jPQdgLqUVJScujQIW1nAQAAembbtm1ubm7azkINRsiZ2e+//37+/HltZwH0XmlpaWlpqbaz0CHnz59/+vSptrMAmnL+/Pnff/9d21moxwg5M8PS09O1nQLQbx988AGC/0gUBEFs3br1ww8/1HYiQCMIgtB2CmozQs7MAAAAjGZQzAAAAOg9KGYAAAD0HhQzAAAAeg+KGQDqcfv2bYIgYmNjtZ3IcDt9+jTxyieffEJdJZFIpk2btmrVKvyyuroah61YsUJuI/v27SMI4ty5c8OU9ODV19cnJCT4+fkZGhoSBJGbm6sYY2ZmRig4duyYXJhQKPT19eVyuRwOZ8GCBXl5eYqbUh4jFAoJgrhy5Qp1YVVVFflD582bN+RPrGegmAGg9/BXW3x8vLYTkZeYmFhbW/vXv/5Vbvm5c+fu37+vlZRU5urqumXLlry8vJ6enqFs58qVK15eXvn5+SKRSCwWFxUVCQQCuRtoB4zx9PT09fX99NNP+/r6hpLMSALFDAD1mDNnjkwmG4VnZtgPP/wgk8mOHDlCLunp6fniiy8CAgJ4PB41curUqTKZ7PPPPx/2HIfE3t5+06ZNubm5ERERSsI8PDxk/27jxo3k2p6eng0bNkil0m3btjU2Nra0tOzdu1cmk0VGRorFYvoxCKGoqKgHDx5QK5yjoyP+iVwuV92fXg9AMQMAaERGRsbz58/JMUbS7Nmzg4KCLl++XFJSopXEVFNaWoqHGdlstsobyc/Pr6+v5/P5Bw8etLa2trCwiI6ODgoKampqyszMpB+DEPL29p4wYYLiGOaoBcUMAPWQu2aWm5tLEMSRI0dKS0u9vLxMTU2trKxWr1794sUL8i1kzLVr19zd3U1MTMaNGxcREdHY2EjGJCUlEQQhN8ENXnjx4kWEUHx8/Pz58xFCu3fvJi+Z4DBPT0+CIO7evavhj96/lJQUY2NjgUCguCo+Pp7BYOzatUvJ2zs7O+Pi4pycnIyMjLhcrre3t9wlIjp7GCEkk8lOnjzp7u7O4XCMjY1dXFy++eYbbfVxLCoqQgitXLmSujA0NBQhVFhYSD8GIcRkMpcuXVpQUPDkyRMNZ60foJgBoEHl5eVeXl6FhYUdHR0vXrxISUkJDAyUi7l165afn19JSUlnZ2djY2NSUhKfz6cOKOmj3t7e4uLid955h8ViKa51cnIKDQ0tKCi4evVqv2/v6enx9fWNjY2tqqrq7u4WiUTXr19fvHix4omI8j0sk8nCwsLCw8NLSkrEYnFXV9f9+/c3b968YcMGNX5YqqqqKh6Px2azJ06cuHz58jt37lDXVldXI4RmzpxJXTh79mxyFc0YDN/lcf36dbV/Cn0ExQwADTp79uzatWsfPXrU0dEhFArt7OyKioru3btHjUlPTw8LC3v06BG+1D9r1qyqqqr9+/fT/BHR0dHFxcUIIXxlBcOrhEKhTCZ766231Puh6KisrBSLxS4uLq8LiIuLY7PZr7tylpCQcPPmTTs7u6ysrLa2tidPnsTGxuK5tZ49e0aNVL6HT58+febMmVmzZmVnZzc3N4vF4sLCQhcXlxMnTmhokLO5ubmmpkYikTQ0NKSmprq6umZkZJBrRSIRQmjMmDHUt+CXbW1t9GMwfGT1a7RWc6CYAaBBixYtSkxM5PF4xsbGHh4en332GUJIrpjNnTv35MmTPB7P1NR0/vz5Fy9eZLFY+j5xdkNDA0LI2tr6dQH29vYRERHl5eXU73oSvq8hNTXV39/f3Nzc1tY2JiZm/fr1XV1dly5dokYq38PJyclMJvPKlSuLFy8eM2aMqakpn88/e/YsQoh6/UldvL29s7KyGhoaRCJRWVlZcHCwRCJZv359e3s7Duh3eFNuIZ0YbOzYsejVrgZQzADQIC8vL+pLBwcHhBD51YYtWrSIOt+rg4PDtGnTampqhiVBTWltbUUIcTgcJTHR0dEmJia7d+9WvL+8urrayspK7mEpf39/pDDUpnwP//rrr1Kp1NbW1sDAgMlkMhgMBoPh7OyMENLEpabMzEx/f38bGxsOh/Pee++lpaUtXLiwpaXlxo0bOADfZyh3Va+lpYVcRTMGMzc3R692NYBiBoAGGRsbU1/iojWouw8YDAZCSO7rvrOzUx3ZaZCFhQV6NWL2OjY2Nlu2bKmsrDx9+rTiWpoTuivfw3i/SaVSqVTa19dHHYMd4uNidBAE4enpiRAih0bxUwoVFRXUMPzIHfkAA50YDI86WlpaaiZ9PQPFDAAty8vLo5a3x48fP3z48I033sAvx40bhxCqra2lvkXumj8ueL29vRrPlbYJEyYghJqampSH7dixg8vlxsbGSiQS6nIej9fU1FRWVkZdmJ2djRS+0JVzdHQ0MTFpbW2VKRiGgVyZTCYUChFCNjY2eAmfz0cInTlzhhqGazleRTMGw7sX72oAxQwALSsrKwsPD6+urn758qVQKAwMDJRIJMHBwXjtjBkzEEJHjhwpKCjo7Oysq6vbvn07vimfhO8OKC4ubm5uHv78+zVjxgwzM7MBnwqwtLSMioqqra09deoUdTluLBcSEpKdnS0SiZ4+fbp3797jx48bGhouWbKEfhrh4eEdHR0+Pj6XL19ubGzs6empr6//6aefli1bdu3atcF/LGX2798fFRVVVlaG7zS5fft2SEjIjRs3uFwuORbq4+ODb1HZvn17U1NTW1tbfHx8RkaGtbV1QEAA/RgM795ROHNV/xT/YNFHqampI+azAC0KDg4ODg5W7b3l5eUIoZiYGPwyJycHIXT48GFqDF549OhR6svg4GAmk0n9rXR0dGxvbyffFRQURF1rYGAQFhaGELpw4QIO6O3tnTRpkuLvtYeHB0Lozp07qn0imUyGEEpNTVUe8/3336NXM4BQCQQCIyOjrq4ucsmjR48QQkuXLqWGtbe347NP6ka6u7vd3d0Vv68SExPJN9LZw319fWvWrOn3qy8nJ4d8F50dJffgFyk9PR0H7NixQ3GtgYGB3A7MycmRO9wEQagQI3s1t0hdXZ3cci6X6+rqquSzkOgcX30BZ2YAaJmHh0dOTs7cuXONjY2tra3Dw8OLiorMzMzIgG+//XbdunVWVlZGRkZubm75+flyw01MJvP8+fOenp6mpqbDnv5rhYWFdXV14bFBJczMzBSfnmaz2fn5+TExMdOnT2ez2RwOZ+HChTk5OdSpoeggCCI5OTk1NdXHx8fS0pLNZjs4OAQEBFy4cMHHx4cMw5fWDAwMBrVxObt27UpISJg/f761tTWLxbKzswsNDS0rK5Pr0y0QCAoKCry9vTkcDr5/NTc3V4UYqVSamZnJ5/OnTJkylLRHDm1XU/WAMzOgFkM5M1NBv+cWOgUN4cysu7t77NixgYGBGstOPaRSqYWFhbOzM3mHiF7A86Eo7nYZnJkBAIDKPvroI7kWMPiZ6EuXLuHRRZ1VUVHR2tq6c+dOmvdP6ogDBw7MnDkTX1zEyBYwcs9WjxKjq5iNvI5TuEEUdcBE63QwJaAtkZGRDg4Oe/bs0XYiygiFwilTpixfvlzbiQzCzZs3r169euDAAblLa6PZ6CpmWqSzHaeGYkR+KDBYoaGh5FAPtQUMQojFYj18+BCPQ+qsyMjIurq6IV4wG2a40Yyfnx91IdkCRiaTlZaWais3bdGn4zd0uOOUtrMA4P8RCATwHxIAtYAzMwAAAHpvdBUzXes4JZVKjx49+u6771paWlpYWMyZM+fQoUMdHR3kRlToR6X1D0VHUVHRypUreTyeoaHh2LFj33///Zs3b5JrS0pKCILYvHmz3LvS0tIIgjhw4AB+KRuoTxX5MQsLCxcsWMDhcObMmUMzQwCAfhldw4z9Ki8v37lzZ3d3N0Koo6MjJSWlrq6O2gQPIXTr1q2oqCipVIoQ6uzsTEpKEgqF5eXl1IeBVPCXv/zlq6++Il/+/PPPP//8M5vNVvweHywtfqgBPXv2bMGCBeTLpqamy5cv5+bmXrt2DT8+5ebm9t5776WkpOzbt4+azD/+8Q9TU9P169ejV32qqFP+4D5V9+7dO378uNzH/PTTT/FUT4oT2gIARobRdWbWLy12nLp48aKpqemPP/7Y2tr68uXLu3fvRkVFUb++Ve5Hpd02WsoRBOHr65uVlfX777/39PT8+eefaWlphoaG+/btI2P+53/+p729nXrjQGVlZWFh4apVq/AMtvT7VKWnp69ateq3337r7e395ZdfaH46AIB+gWKmzY5TkydPnjhx4pIlS7hcromJiYuLy1dfffW62XcGRZfbaI0fP/7LL79MSUlxdXU1MTEZP378hx9++PLlywcPHpAxH374oY2NzT/+8Q9yCf73li1b8Ev6farmzZuXlJQ0bdo0uIkZgBEMhhlV7zj18OHDIf7ow4cPL1u2jMfj+fn5ubi4uLm5vf3220PcJqbFDzWgW7duLVy4ULEBB7WtCZvN3rhxY2xsbFFREZ/PF4vF33///aJFi5ycnHAA2acKvZr5Bb1q/CHXp8rHx2dQD8OeP39evx6e1bSQkJCQkBBtZwHAAKCYabPjlIuLS1VV1a1bt27dulVcXBwXF2dtbX3u3LlZs2bR/+n90uU2Wvv27evp6YmJiQkLC5s0aZKhoSFBEI6OjnLtQjZu3Pi3v/3tH//4B5/P//7770Ui0ccff0yuJftUKW5frkxaWVkNKr158+Zt3bp1UG8ZwUJCQj755BM3NzdtJwI0YiT9mQLFjJa8vLw9e/aQf7CrseOUgYEBn8/HNz50dHRMnz49PDxcro2ThmjuQyn3+PHj8ePHU+dhqampefTokVyPwfHjx4eEhJw7d+7Zs2eJiYlvvvnm4sWLybWOjo6//PLLH3/8Idd7d+gmT54sN6PraBYSEuLm5gY7ZKQaScUMrpnRoqGOU+7u7seOHausrOzs7Gxra8vNzW1ubn78+LFef6gB2dnZPX/+PCEhoa2tra2tLTs7+z//8z/7vc/w448/lkgk69ate/DgwZYtW6ijf8PZpwoAoAc0NYPx8KI5a76udZwyNDRUPCL/8z//Q252wDZLeApXb29vufy1+KGUzCqL73uUq4gIobfffnvmzJlWVlaKHxA3tTI3NxeJRNTldPpUqTAn/TDPmq/70AiaVR0oGknHF87MaNFQx6l//etfmzZtmjFjBt6sh4dHUlLS4cOH9fpDDWjp0qVnzpyZPXu2sbHxhAkTNmzYcO3atX7rOkII969at24dh8OhLqfZpwoAMFpou5qqh+b6mel+xykV6NGH+uSTTwiCqK6uHp4fB2dmctAI+ssdKBpJxxfOzICOkkqlV65cSUxM5PP55G0pQAedPn2anM+M2s8MISSRSKZNm7Zq1Sr8ErcHIghixYoVchvZt28fQRDnzp0bpqQHr76+PiEhwc/PD99/m5ubqxhjZmZGKDh27JhcmFAo9PX15XK5HA5nwYIFeXl5iptSHoMbVuD+nCSynxlBEPPmzRvyJ9YzUMyALoqNjTUwMBAIBN3d3Tt27NB2Opqi0R46utCgJzExsba29q9//avc8nPnzt2/f18rKanM1dV1y5YteXl5io9IDsqVK1e8vLzy8/NFIhGefEcgEKSnpw8qxtPT09fX99NPP4UZ2khQzIDumjhx4qFDh6h35AOd9cMPP8j+vZ9ZT0/PF198ERAQwOPxqJFTp06VyWSff/75sOc4JPb29ps2bcrNzY2IiFAShjuNUeHrvlhPT8+GDRukUum2bdsaGxtbWlrwXVGRkZFisZh+DEIoKirqwYMH1ApH9jNT+/MqegGeMxvAiOw4pfsfKjY2diQ1BB+dMjIynj9/To4xkmbPnv32229nZGSUlJTo0ePYZLvLrKwslTeSn59fX1/P5/MPHjyIl0RHR9+5cycjIyMzM3PlypU0YxBC3t7eEyZMOHbs2Eh6Vmwo4MwMAFV0dnbGxcU5OTkZGRlxuVxvb2+5Cxgq99DRdIMeFfoKqSYlJcXY2FggECiuio+PZzAYu3btUvL2AfcwnW5HiEaroOFUVFSEECILEhYaGooQIpta0IlBCDGZzKVLlxYUFMjN3zZqQTEDYNB6enp8fX1jY2Orqqq6u7tFItH169cXL16seKlfZbdu3fLz8yspKens7GxsbExKSsJzVKpr+5rW29tbXFz8zjvvsFgsxbVOTk6hoaEFBQVXr17t9+3093B5ebmXl1dhYWFHR8eLFy9SUlICAwPJtTKZLCwsLDw8vKSkRCwWd3V14VZBGzZsUOOHpaqqquLxeGw2e+LEicuXL79z5w51bXV1NUJo5syZ1IWzZ88mV9GMwfBdHnLz8oxaUMwAGLSEhISbN2/a2dllZWW1tbU9efIkNjaWIIitW7c+e/aM5kaU99DRXIMelfsKDUplZaVYLHZxcXldQFxcHJvNft2VM/p7WHm3I/qtgtSlubm5pqZGIpE0NDSkpqa6urpmZGSQa0UiEXo1dQ4Jv2xra6Mfg+GDqKEPonegmAEwaPiqe2pqqr+/v7m5ua2tbUxMzPr167u6ui5duqSWH6GtBj3q0tDQgBCytrZ+XYC9vX1ERER5eTn1u55Efw8r73ZEv1WQWnh7e2dlZTU0NIhEorKysuDgYIlEsn79erJhRb/Dm3IL6cRgY8eORa92NYBiBsCgVVdXW1lZyT3K4+/vjxQGglTWb4OempoatWx8GLS2tiKE5OZtkRMdHW1iYrJ7927F+8vp72Hl3Y7IVkEGBgZMJpPBYDAYDGdnZ6TQKkgtMjMz/f39bWxsOBzOe++9l5aWtnDhwpaWlhs3buAAfJ+h3FW9lpYWchXNGMzc3By92tUAihkAqhiw55nmeuhoeuNqgRuC4xGz17GxsdmyZUtlZeXp06cV19LsKqe82xHZKkgqlfb19VGHW4f4uBgdBEF4enoihMihUfyUQkVFBTUMP3JHPsBAJwbDo45y7SZGLShmAAwaj8dramqS69STnZ2NKF83Q+yhk5eXRx1ZGrYGPeoyYcIEhJBcjzpFO3bs4HK5sbGxEomEupzOHqbD0dHRxMSktbVVcfajYRizlclkQqEQIWRjY4OX4NlNz5w5Qw3DtZyc+JRODIZ3L97VAIoZAIP2wQcfIIRCQkKys7NFItHTp0/37t17/PhxQ0PDJUuW4Jgh9tDRVoMedZkxY4aZmdmADwBYWlpGRUXV1taeOnWKupzOHqZjOFsF7d+/PyoqqqysDN9pcvv27ZCQkBs3bnC5XHIs1MfHB9+isn379qampra2tvj4+IyMDGtr64CAAPoxGN69o3Dmqv4NbWpHXaG5iYbBqEJzouHu7m7cm0ZOYmIiNUy1HjoabdAjo9FXiArRmIj2+++/R69mAKESCARGRkZdXV3kEtweaOnSpdSw9vZ2fKJJ3QidPUyn2xGdVkE094ncg1+k9PR0HNDvvGsGBgZyOzAnJ0fuyBIEoUKM7NXcInV1dXLLuVyuq6urks9ConN89QWcmQEwaGw2Oz8/PyYmZvr06Ww2m8PhLFy4MCcnhzpxERpaDx1tNehRo7CwsK6uLjw2qISZmZni09M09/CAaLYKwpfWDAyGNCPSrl27EhIS5s+fb21tzWKx7OzsQkNDy8rK5Pp0CwSCgoICb29vDoeDb1XNzc1VIUYqlWZmZvL5/ClTpgwl7ZFD29VUPeDMDKiFLrSA0akGPWgIZ2bd3d1jx44NDAzUWHbqIZVKLSwsnJ2dyTtE9AKeD0Vxt8vgzAwAAFT20UcfybWAwc9EX7p0SUnzcV1QUVHR2tq6c+dOmvdP6ogDBw7MnDkTX1zEyBYwcs9WjxJQzAAAmhIZGeng4LBnzx5tJ6KMUCicMmXK8uXLtZ3IINy8efPq1asHDhyQu7Q2msGs+QCAIQkNDcXT4CpisVgPHz4c5nwGKzIyMjIyUttZDA5uNCO3ELeA0Uo+ugCKGQC6Rfcb9ACgg2CYEQAAgN6DYgYAAEDvQTEDAACg96CYAQAA0Hsj6gaQtLQ0bacA9NvTp08R/Ef6d9D7EegHLT+0rSZ4BhAAAACDMmJmACFkcBMwAJqEp9SDsz0ANAqumQEAANB7UMwAAADoPShmAAAA9B4UMwAAAHoPihkAAAC9B8UMAACA3oNiBgAAQO9BMQMAAKD3oJgBAADQe1DMAAAA6D0oZgAAAPQeFDMAAAB6D4oZAAAAvQfFDAAAgN6DYgYAAEDvQTEDAACg96CYAQAA0HtQzAAAAOg9KGYAAAD0HhQzAAAAeg+KGQAAAL0HxQwAAIDeg2IGAABA70ExAwAAoPegmAEAANB7UMwAAADoPShmAAAA9B4UMwAAAHoPihkAAAC9B8UMAACA3oNiBgAAQO9BMQMAAKD3oJgBAADQewbaTgCAkaaoqKikpIR8WVVVhRDav38/ucTNzY3P52shMwBGLkImk2k7BwBGlGvXrvn4+LBYLAZDfuSjr69PIpHk5+d7e3trJTcARiooZgCoWV9fn42NTWNjY79rra2tnz17xmQyhzkrAEY2uGYGgJoxGIyVK1ey2WzFVWw2OzQ0FCoZAGoHxQwA9fvoo496enoUl/f09Hz00UfDnw8AIx4MMwKgEfb29vX19XILbW1t6+vrCYLQSkoAjGBwZgaARoSFhbFYLOoSFou1Zs0aqGQAaAKcmQGgEVVVVU5OTnILKyoqnJ2dtZIPACMbnJkBoBGOjo7Ozs7U87AZM2ZAJQNAQ6CYAaApq1atIm9cZLFYq1ev1m4+AIxgMMwIgKb8/vvvU6ZMwb9iBEE8fvzY3t5e20kBMDLBmRkAmmJra+vq6spgMBgMhqurK1QyADQHihkAGhQWFkYQBIPBCAsL03YuAIxkMMwIgAY1NTXZ2NgghP74449x48ZpOx0ARi4ZRWpqqrbTAQAAAAaWmppKrV/9tICBkgaAupSUlBw5cmTz5s3z58/Xdi464fDhwwihrVu3ajsRoN9CQkLklvRTzD788MNhSQaAUeHIkSN/+9vfOByOthPRCenp6Qi+ZMCQKRYzuAEEAI2DSgaApkExAwAAoPegmAEAANB7UMwAAADoPShmAOio27dvEwQRGxur7UT0nkQimTZt2qpVq/DL6upqgiAIglixYoVc5L59+wiCOHfu3LDnSFd9fX1CQoKfn5+hoSFBELm5uYoxZmZmhIJjx47JhQmFQl9fXy6Xy+FwFixYkJeXp7gp5TFCoZAgiCtXrqjxA6oMihkAQB7+koqPj9d2IuqRmJhYW1v717/+VW75uXPn7t+/r5WUVObq6rply5a8vLx+W5nTd+XKFS8vr/z8fJFIJBaLi4qKBAIBvteUfoynp6evr++nn37a19c3lGTUAooZADpqzpw5MpkMzsyGqKen54svvggICODxeNTlU6dOlclkn3/+ubYSU429vf2mTZtyc3MjIiKUhHl4eMj+3caNG8m1PT09GzZskEql27Zta2xsbGlp2bt3r0wmi4yMFIvF9GMQQlFRUQ8ePJCrgloBxQwAMJJlZGQ8f/6cHGMkzZ49Oygo6PLlyyUlJVpJTDWlpaV4mJHNZqu8kfz8/Pr6ej6ff/DgQWtrawsLi+jo6KCgoKampszMTPoxCCFvb+8JEyYojmEOPyhmAOgouWtmubm5BEEcOXKktLTUy8vL1NTUyspq9erVL168IN9Cxly7ds3d3d3ExGTcuHERERGNjY1kTFJSEkEQ58+fp/4svPDixYsIofj4eDxfye7du8krLjjM09OTIIi7d+9q+KOrU0pKirGxsUAgUFwVHx/PYDB27dql5O2dnZ1xcXFOTk5GRkZcLtfb21vuEhGd44IQkslkJ0+edHd353A4xsbGLi4u33zzjbamxi0qKkIIrVy5krowNDQUIVRYWEg/BiHEZDKXLl1aUFDw5MkTDWc9AChmAOiT8vJyLy+vwsLCjo6OFy9epKSkBAYGysXcunXLz8+vpKSks7OzsbExKSmJz+dTh4ZGj97e3uLi4nfeeYfFYimudXJyCg0NLSgouHr1ar9v7+np8fX1jY2Nraqq6u7uFolE169fX7x4seKJiPLjIpPJwsLCwsPDS0pKxGJxV1fX/fv3N2/evGHDBjV+WKqqqioej8dmsydOnLh8+fI7d+5Q11ZXVyOEZs6cSV04e/ZschXNGGzevHkIoevXr6v9UwwKFDMA9MnZs2fXrl376NGjjo4OoVBoZ2dXVFR07949akx6enpYWNijR4/wRftZs2ZVVVXt37+f5o+Ijo4uLi5GCOFrJBheJRQKZTLZW2+9pd4PpTmVlZVisdjFxeV1AXFxcWw2+3VXzhISEm7evGlnZ5eVldXW1vbkyZPY2FiCILZu3frs2TNqpPLjcvr06TNnzsyaNSs7O7u5uVksFhcWFrq4uJw4cUJDg5zNzc01NTUSiaShoSE1NdXV1TUjI4NcKxKJEEJjxoyhvgW/bGtrox+D4f8PWh+thWIGgD5ZtGhRYmIij8czNjb28PD47LPPEEJyxWzu3LknT57k8Ximpqbz58+/ePEii8WSG1ccJRoaGhBC1tbWrwuwt7ePiIgoLy+nfteT8H0Nqamp/v7+5ubmtra2MTEx69ev7+rqunTpEjVS+XFJTk5mMplXrlxZvHjxmDFjTE1N+Xz+2bNnEULU60/q4u3tnZWV1dDQIBKJysrKgoODJRLJ+vXr29vbcUC/w5tyC+nEYGPHjkWvdrUWQTEDQJ94eXlRXzo4OCCEyC8pbNGiReRVLhwzbdq0mpqaYUlQt7S2tqYwZQAAACAASURBVKKB5saMjo42MTHZvXu34v3l1dXVVlZWeBiN5O/vjxSG2pQfl19//VUqldra2hoYGDCZTNx83NnZGSGkiUtNmZmZ/v7+NjY2HA7nvffeS0tLW7hwYUtLy40bN3AAl8tFCMld1WtpaSFX0YzBzM3N0atdrUVQzADQJ8bGxtSXuGgN6j4CBoOBEJL74u7s7FRHdjrHwsICvRoxex0bG5stW7ZUVlaePn1acS31zwIllB8XvLelUqlUKu3r66OO3A7xcTE6CILw9PRECJFDo/gphYqKCmoYfuSOfICBTgyGRx0tLS01kz5dUMwAGGny8vKo5e3x48cPHz5844038Evc8Lq2tpb6Frmr97jg9fb2ajxXDZswYQJCqKmpSXnYjh07uFxubGysRCKhLufxeE1NTWVlZdSF2dnZSOELXTlHR0cTE5PW1laZgmEY/pXJZEKhECGEm54jhPh8PkLozJkz1DBcy/EqmjEY3r14V2sRFDMARpqysrLw8PDq6uqXL18KhcLAwECJRBIcHIzXzpgxAyF05MiRgoKCzs7Ourq67du345vySfg6f3FxcXNz8/Dnr0YzZswwMzMb8FkCS0vLqKio2traU6dOUZd/8MEHCKGQkJDs7GyRSPT06dO9e/ceP37c0NBwyZIl9NMIDw/v6Ojw8fG5fPlyY2NjT09PfX39Tz/9tGzZsmvXrg3+Yymzf//+qKiosrIyfKfJ7du3Q0JCbty4weVyybFQHx8ffIvK9u3bm5qa2tra4uPjMzIyrK2tAwIC6MdgePfKDcZqAfVvBNxjWvFvBwCAaobyO1VeXo4QiomJwS9zcnIQQocPH6bG4IVHjx6lvgwODmYymdRfc0dHx/b2dvJdQUFB1LUGBgZhYWEIoQsXLuCA3t7eSZMmKX5ReHh4IITu3Lmj2ieSyWTBwcHBwcEqv10FAoHAyMioq6uLXPLo0SOE0NKlS6lh7e3t+JwVIfTDDz/ghd3d3e7u7opfm4mJieQb6RyXvr6+NWvW9PsNnJOTQ76Lzu6Ve/CLlJ6ejgN27NihuNbAwCA1NVUuQ7n/JARBqBAjezW3SF1dnZK01Q4hJJcJnJkBMNJ4eHjk5OTMnTvX2NjY2to6PDy8qKjIzMyMDPj222/XrVtnZWVlZGTk5uaWn58vN3DEZDLPnz/v6elpamo67OmrWVhYWFdXFx4bVMLMzEzx6Wk2m52fnx8TEzN9+nQ2m83hcBYuXJiTk0OdGooOgiCSk5NTU1N9fHwsLS3ZbLaDg0NAQMCFCxd8fHzIMHxpzcDAYFAbl7Nr166EhIT58+dbW1uzWCw7O7vQ0NCysjK57t4CgaCgoMDb25vD4eC7XnNzc1WIkUqlmZmZfD5/ypQpQ0lbDaiVDc7MAFCvYf6d6vcsQacM/5lZd3f32LFjAwMDh/OHqkAqlVpYWDg7O5N3iOgFPB8KeS47bJDun5mRc8NoYuOnTp1SnMhHeYy22nDQafTQ19f33Xffubm5WVtbW1pazpkz55tvvpG7gj0gsh2GotcNjNAHRxNoHX4m+tKlS3h0UWdVVFS0trbu3LmT5v2TOuLAgQMzZ87EFxe1a0jns0BzXF1d//zzT+UxYWFh+LlL7Oeff/75558vX76cnZ2tX78PAGhUZGTkN998s2fPnu+//17bubyWUCicMmXK8uXLtZ3IINy8efPq1au5ublyl9a0AorZAHAbjuH/ufb29sHBwe+///6PP/544sQJxYA7d+6cPXvW0NDwn//855IlS5hM5tWrV9etW5ebm5ufn+/r6zuoH+ft7Z2fn6+m3HWXto4m0C4Wi/Xw4UNtZzGAyMjIyMhIbWcxOLjRjLaz+H+gmOmo0tJS/I+srKx+A3799VeE0Nq1a1evXo2XLFu27MGDB3Fxcb/++utgixkYGQQCge58uQAwnFS8ZiYbqJ0BebHkxo0b7u7upqamtra2+/btw2uPHj06ffp0IyMjR0fH13V1U9LDgk4CCKHW1tbNmzdPmDDB2Nh4zpw5ly9fVvwpA8ao0IYDIdTS0hIZGWljY0Nuls4FnkEhn39URH16US09O+BoavpoAgCGino3CM07r/r6+vp91iEiIoKMwXdVhYSEyN1meujQITwFJ4nBYFCfq8Bv/OCDD5Q8KEMngc7OTrm5vQmCCAkJQZQHMujE9Pusz4oVKwwNDalv5PP5g/rR9G3atAn9+8MoWG9vr7Ozs6Gh4alTp1paWtra2s6fP29ubm5nZ/fy5UsybMAnV/BVcW9vbyU5wNFU+WjCHcJyhv9uRjAiIYW7GVUpZikpKQghxXYGCKFbt27hGPw1gRD65JNP6urqxGLx+fPnWSwWl8vlcDhJSUnPnz9vbm7evn07QmjVqlXkxsk3rlmzhtrDAiEUHR1NP4G///3vCKHp06fn5+e3t7c/fvyYHI8mv4PoxPT79YcQ2rhxI7XdA0Lo7t27cpu9du1ae3t7bW3tli1b5DZL3+uKmUwme/bs2cqVK/HMQ1hAQEBtbe2gtq/kFi8yWziaKh9NKGZyoJgBtVBPMVu4cCGTyfzjjz+oC/ElnB07duCX+GsCj+CTli1bhhA6ePAguaS3t5fL5eLL8tQ3zp07l/qwRU1NDYvFcnR0pJ+Aq6srQRAVFRXUGG9vb+p3EJ2Yfr/+Fi1aRH1LQkICQui7777DL+fOnau4WXwRS73FLDs7G09NRHJwcDh37tygtk+/mMHRJNE/mlDM5EAxA2qhWMxUuQGEbGdA/paiV/NDy7UzWLBgAfUlfkScOtcAk8mcNGmS4j3o/fawIO9HopNAdXX1pEmTcJMFkkAgoM6ERiemX8rbPdTU1Chu1s/P73XdbFVTVlbm7+8/ZcqU7OxsNzc3JpN5+/btjz/+ePny5ebm5osXLx7U1ujczQhHkzTYo5mWlkY/eGR7+vQpgh0CNECVYka2M1BcJdfOwMjIiPoSf6MpLlRsI6SuBDRkwDYcio95ydR9j9m3337b19d39OhRsm4tXLgwOTl5zpw5x48fH2wxowOOJmmwRxNfYwMk2CFA7VS5m3EY2hko72FBJwEej/d///d/eLSKJDePBp0YFbzxxhtPnz6trKykLlTvaRlSaJqH4cKgazOdw9FUYRRlpIJhRqAWir9lqhSzYWhnoLyHBZ0Eli1bJpPJli1bdv36dbFYXFtbu2nTJrn06MSoAG82ODi4oKDg5cuX9fX1W7duzcvLG+Jm5bz99tsIoS1btuTk5LS1tYnF4uvXr69du5ZcpTvgaAIANI5a6+jfmj9gO4N+JzzFd7s9ePCAutDZ2XnSpEnkSzo9LOgkoNGbuZW3e+jo6Jg9e7bcZvHcZRcvXhxw92IDNnp48eIFvrojx8bG5unTp+R2aN6a3693331XyaeGo0nnaMINIHLgzAyoBVLLRMM02xkMhfIeFnQSMDIyunHjRmRk5Pjx442MjN5+++0LFy4IBALqT6ETowJjY+MbN25s2LBh3LhxRkZG77777qVLl/Bth2rsLG5paVlWVrZ9+/bp06cbGhqy2ew33ngjMjLyl19+ketEpXVwNAEAmkbIKIOPaWlpISEhMpgOR936+vrmzJlz9+7dxsZGKysrbacDhmRQRxN+p+Tgk9rXzRQDAE24TSi1uZrOtYAZGbZv33769On6+vqOjo67d+9++OGHd+7c8fLygkqmj+BoAqD7oJhpxG+//RYWFmZvb29qavr222//+OOPZmZmhw4dQgjdvXv3df3DCIIICAjQdu5AnpKjCTRKIpFMmzZt1apV+CXZe2/FihVykfv27SMI4ty5c8OeI110OhSamZkpficcO3ZMLkwoFPr6+uL5dxYsWNDv7Uh0YgY0bPkIhUKCIHCfT9VRL6DBxWp1efjw4Zo1a3g8nqGh4dixY4ODg8kbJe7cuaPkcCxdulS7mQNFSo7mgIb5d6q4uBghtHfvXp3d+KBuAPn6668NDAwePXqEX5I3KxEEce/ePWrkl19+ibTR75i+8ePHU3/T+53Tx9TUVPE7ITExkRqj2DyMIIi0tLTBxtAxnPn4+vrOmjVLKpXSzA2pZTorAABNUMzk0C9m3d3d48aNowbjYjZ16lSEkL+/PzVY94uZq6vrpk2bcnNzIyIilBQz3CHsdbq7u/HMO9u2bWtsbGxpadm7dy9CyNramrw9mE4MTcOZDz4toz8bn2Ixg2FGAIAuysjIeP78OTnGSJo9e3ZQUNDly5dLSkq0kphqSktL8TAjm81WeSP5+fn19fV8Pv/gwYPW1tYWFhbR0dFBQUFNTU2ZmZn0Y9RFjfl4e3tPmDBBcQyTPihmAOiEzs7OuLg4JycnIyMjLpfr7e0tdwkhKSlJsYkaXnjx4kWEUHx8/Pz58xFCu3fvJq9wIErbNiVd5VTeOFJTzzxFKSkpxsbG/T5cER8fz2Awdu3apeTtA+5Pmt3sZDSa7Q2boqIihJDcQ6ihoaEIocLCQvoxOpgPk8lcunRpQUGB3AS/9EExA0D7enp6fH19Y2Njq6qquru7RSLR9evXFy9ePJQ/VOXcunXLz8+vpKSks7OzsbExKSmJz+eLxWJ1bV+9ent7i4uL33nnHRaLpbjWyckpNDS0oKDgdfOK0d+f5eXlXl5ehYWFHR0dL168SElJCQwMJNfKZLKwsLDw8PCSkhKxWNzV1XX//v3Nmzdv2LBBjR+WqqqqisfjsdnsiRMnLl++XO4Se3V1NUJo5syZ1IX4oX68imaMbuYzb948hND169cHmyQGxQwA7UtISLh586adnV1WVlZbW9uTJ09iY2MJgti6deuzZ89obiQ6Olrxsha5Nj09PSwsjNpVrqqqav/+/UPfuFAolMlkctOvDFFlZaVYLMZt7foVFxfHZrM///zzftfS359nz55du3YttZtdUVHRvXv38NrTp0+fOXNGsdneiRMnNDTI2dzcXFNTI5FIGhoaUlNTXV1dMzIyyLUikQghNGbMGOpb8Mu2tjb6MbqZD/4vpPKOhWIGgPbhh4hTU1P9/f3Nzc1tbW1jYmLWr1/f1dV16dIltfyIuXPnnjx5ksfjmZqazp8//+LFiywWS10zg6tdQ0MDQsja2vp1Afb29hEREeXl5dTvVhL9/blo0aLExEQej2dsbOzh4YE7p5PFLDk5mclkXrlyZfHixWPGjDE1NeXz+WfPnkUIqf36E0LI29s7KyuroaFBJBKVlZUFBwdLJJL169eTDYn6Hd6UW0gnRjfzGTt2LHp16FUAxQwA7auurrayssLDLCR/f3+k0tBQv/rtKldTU6OWjatda2srQojD4SiJiY6ONjEx2b17t2LXIfr7U3k3O7LZnoGBAZPJZDAYDAYDN7dT+dKOEpmZmf7+/jY2NhwO57333ktLS1u4cGFLS8uNGzdwAJfLRQodM1paWshVNGN0Mx9zc3P06tCrAIoZADpBsWuaHAaDgV51+SF1dnaq5adrdOMqsLCwQK9GqF7HxsZmy5YtlZWVp0+fVlw74P7ElHezI5vtSaVSsls6XjUMzfYIgvD09EQIkUOjPB4PIVRRUUENu3//PrmKZoxu5oNHHVWe8hSKGQDax+PxmpqaysrKqAuzs7MR5Rd+3LhxCKHa2lpqjNzVclyTent7FX+E8q5yQ9y42k2YMAEh1NTUpDxsx44dXC43NjZWIpFQl9PZn3QMQ+9GJWQymVAoRAjZ2NjgJbix+5kzZ6hhuJaTPd/pxOhmPvhw40OvYn4keGgaAPWi+Tt18OBBhJC9vf1PP/3U1tb2+++/79mzh8FgGBoaNjQ04Bg8JGhjY3Pjxo2Ojo7a2tpt27bh3+ILFy7gmP/93/9FCP3Hf/xHU1MTuXHc1AYhhO90EIvFxcXF+Hay6OjoIW58sGg+NC2RSMzMzNzc3KgL8UPTcrPk4Cdw8fAg+dA0nf1JpwFQcnIyQmjOnDlZWVnPnz/v7u6uq6u7fPlyUFBQfn6+antg06ZNqL+Hpvft27d9+/Z//etfTU1N7e3t5eXleFJmLpfb1taGY7q7u+3s7NCrB5BbW1v7fUh5wBg6hj+fU6dOIYSSk5PppIdgBhAAhhPN36nu7m53d3fFvzXlpg4KCgqirjUwMAgLC6PWm97eXrkGQDJ6XeVU3riMRs88KvozgAgEAiMjo66uLnJJv8Wsvb0dn1ZSixmd/UmnmNFptkdzDwzYoXDHjh2Kaw0MDOS+snNychSnhlIhZsCchzkfmUy2ceNGhFBdXZ2S3UhCMAMIADqIzWbn5+fHxMRMnz6dzWZzOJyFCxfm5OTgX2/St99+u27dOisrKyMjIzc3t/z8fLmBGiaTef78eU9PT8VZ9ZR3lRvixjUhLCysq6sLjw0qYWZmpvj0NM39OSCavRvxpTUDA4NBbVzOrl27EhIS5s+fb21tzWKx7OzsQkNDy8rKqF1OEEICgaCgoMDb25vD4eAbU3Nzc1WIGTDnYc5HKpVmZmby+Xw895UqqJUNzswAUC9d+J3q9xREWwY1N+PYsWMDAwM1ndIQSaVSCwsLZ2dn8g4R3aeDOeP5WejProngzAwAoBfwM9GXLl0iJ8vXTRUVFa2trTt37qR5/6Qu0MGcDxw4MHPmTHxZTjVQzAAAOioyMtLBwWHPnj3aTkQZoVA4ZcqU5cuXazuRQdC1nG/evHn16tUDBw7IXVoblCEN8gIAgOawWKyHDx9qO4sBREZGRkZGajuLwdG1nHGjmSFuBIoZACOcQCAY+jcFADoOhhkBAADoPShmAAAA9B4UMwAAAHoPihkAAAC9188NIEO50x8AQPX06VMEv1MUpaWlCHYI0ABmbGws+UIkEqnQihQA8Drm5uZSqfT58+fjx4/Xdi46YfLkyZMnT9Z2FkDvzZgxQyAQ2NrakksIuGcXAI3Cc9ClpaVpOxEARjK4ZgYAAEDvQTEDAACg96CYAQAA0HtQzAAAAOg9KGYAAAD0HhQzAAAAeg+KGQAAAL0HxQwAAIDeg2IGAABA70ExAwAAoPegmAEAANB7UMwAAADoPShmAAAA9B4UMwAAAHoPihkAAAC9B8UMAACA3oNiBgAAQO9BMQMAAKD3oJgBAADQe1DMAAAA6D0oZgAAAPQeFDMAAAB6D4oZAAAAvQfFDAAAgN6DYgYAAEDvQTEDAACg96CYAQAA0HtQzAAAAOg9KGYAAAD0HhQzAAAAeg+KGQAAAL0HxQwAAIDeg2IGAABA7xEymUzbOQAwoqSkpBw6dEgqleKXTU1NCCFra2v8kslkbtu2bdWqVVrLD4CRCIoZAGr28OHD6dOnKwn47bffpk2bNmz5ADAawDAjAGo2bdo0FxcXgiAUVxEE4eLiApUMALWDYgaA+q1atYrJZCouNzAwWL169fDnA8CIB8OMAKjfH3/8YWtr29fXJ7ecIIjff/990qRJWskKgBEMzswAUL+JEye6u7szGP/2+8VgMDw8PKCSAaAJUMwA0IiwsDC5JQRBwE2MAGgIDDMCoBEtLS3jx4+XSCTkEgMDg2fPnllZWWkxKwBGKjgzA0AjLC0tfX19ydtAmEymn58fVDIANASKGQCaEhoaSt4DIpPJQkNDtZsPACMYDDMCoCkdHR1WVlZdXV0IISMjo6amJlNTU20nBcDIBGdmAGiKiYlJYGAgi8VisViBgYFQyQDQHChmAGjQihUrJBKJRCJZsWKFtnMBYCQz0HYC6vH06dNbt25pOwsA5EmlUhMTE5lMJhKJ0tLStJ0OAPLc3d0nT56s7SzUQTYipKamantHAgCA/klNTdX297d6jJAzM0wGN7OMUGlpaSEhIXp6fAsLCwmC4PP5atzmBx98gBBKT09X4zbBKNTvdNh6akQVMwB00Pz587WdAgAjHxQzADRLboZGAIAmwK8ZAAAAvQfFDAAAgN6DYgYAAEDvQTEDI9zt27cJgoiNjdV2IlojkUimTZtGdp+prq4mCIIgCMXnuPft20cQxLlz54Y9R7rq6+sTEhL8/PwMDQ0JgsjNzVWMMTMzIxQcO3ZMLkwoFPr6+nK5XA6Hs2DBgry8PMVN0YkZ0LDlIxQKCYK4cuWKCkmOBNp+NkA98HNm2s4CaMpQjm95eTlCKCYmRq0Z0VJcXIwQ2rt3r9q3HBwcHBwcTDP466+/NjAwePToEX756NEj/LtPEMS9e/eokV9++SVC6IcfflBzuuozfvx46tdXTk6OYky/04YlJiZSY3Jzc8mGBuTeSEtLG2wMHcOZj6+v76xZs6RSKc3c0Ah6zgzOzMAIN2fOHJlMNmrPzHp6er744ouAgAAej0ddPnXqVJlM9vnnn2srMdXY29tv2rQpNzc3IiJCSZiHh4fcN93GjRvJtT09PRs2bJBKpdu2bWtsbGxpacF/cERGRorFYvox9A1bPlFRUQ8ePBidDyBCMQNgJMvIyHj+/Llih+vZs2cHBQVdvny5pKREK4mpprS0FA8zstlslTeSn59fX1/P5/MPHjxobW1tYWERHR0dFBTU1NSUmZlJP0Zd1JiPt7f3hAkTFMcwRwMoZmCEk7tmlpubSxDEkSNHSktLvby8TE1NraysVq9e/eLFC/ItZMy1a9fc3d1NTEzGjRsXERHR2NhIxiQlJREEcf78eerPwgsvXryIEIqPj8ePS+/evZu8UoLDPD09CYK4e/euhj86QgilpKQYGxsLBALFVfHx8QwGY9euXUre3tnZGRcX5+TkZGRkxOVyvb295S7J0NmfCCGZTHby5El3d3cOh2NsbOzi4vLNN9/ItDSlS1FREUJo5cqV1IW421xhYSH9GB3Mh8lkLl26tKCg4MmTJ+pNUvdBMQOjUXl5uZeXV2FhYUdHx4sXL1JSUgIDA+Vibt265efnV1JS0tnZ2djYmJSUxOfzVRhi0qLe3t7i4uJ33nmHxWIprnVycgoNDS0oKLh69Wq/b+/p6fH19Y2Nja2qquru7haJRNevX1+8eLHiH/7K96dMJgsLCwsPDy8pKRGLxV1dXffv39+8efOGDRvU+GGpqqqqeDwem82eOHHi8uXL79y5Q11bXV2NEJo5cyZ14ezZs8lVNGN0M5958+YhhK5fvz7YJPUdFDMwGp09e3bt2rWPHj3q6OgQCoV2dnZFRUX37t2jxqSnp4eFhT169EgsFhcVFc2aNauqqmr//v00f0R0dLTiDSB4lVAolMlkb731lno/lKLKykqxWOzi4vK6gLi4ODab/borZwkJCTdv3rSzs8vKympra3vy5ElsbCxBEFu3bn327Bk1Uvn+PH369JkzZ2bNmpWdnd3c3CwWiwsLC11cXE6cOKGhQc7m5uaamhqJRNLQ0JCamurq6pqRkUGuFYlECKExY8ZQ34JftrW10Y/RzXzw/yv9Gj1WCyhmYDRatGhRYmIij8czNjb28PD47LPPEEJyxWzu3LknT57k8Ximpqbz58+/ePEii8WSG1fUcQ0NDQgha2vr1wXY29tHRESUl5dTv1tJ+D6C1NRUf39/c3NzW1vbmJiY9evXd3V1Xbp0iRqpfH8mJyczmcwrV64sXrx4zJgxpqamfD7/7NmzCCG1X39CCHl7e2dlZTU0NIhEorKysuDgYIlEsn79+vb2dhzQ7/Cm3EI6MbqZz9ixY9GrQz+qQDEDo5GXlxf1pYODA0KI/HLBFi1aRJ1T3MHBYdq0aTU1NcOSoHq0trYihDgcjpKY6OhoExOT3bt39/X1ya2qrq62srLCw1Ykf39/pDC0pXx//vrrr1Kp1NbW1sDAgMlkMhgMBoPh7OyMENLEpZ3MzEx/f38bGxsOh/Pee++lpaUtXLiwpaXlxo0bOIDL5SKE5K7qtbS0kKtoxuhmPubm5ujVoR9VoJiB0cjY2Jj6EhetQf3djacPlisAnZ2d6shObSwsLNCrEarXsbGx2bJlS2Vl5enTpxXX0mwRonx/4r0klUqlUmlfXx91xLWnp4fO9oeCIAhPT0+EEDk0ip9SqKiooIbdv3+fXEUzRjfzwaOOlpaWQ0lSH0ExA6B/eXl51PL2+PHjhw8fvvHGG/jluHHjEEK1tbXUt8hddccFr7e3V+O5vsaECRMQQk1NTcrDduzYweVyY2NjJRIJdTmPx2tqaiorK6MuzM7ORoP8Qnd0dDQxMWltbVV80HUYhm1lMplQKEQI2djY4CW4t9yZM2eoYbiWk23n6MToZj74cONDP6pAMQOgf2VlZeHh4dXV1S9fvhQKhYGBgRKJJDg4GK+dMWMGQujIkSMFBQWdnZ11dXXbt2/HN+WT8PX54uLi5ubm4c8fJ2lmZjbgMwCWlpZRUVG1tbWnTp2iLsctQENCQrKzs0Ui0dOnT/fu3Xv8+HFDQ8MlS5bQTyM8PLyjo8PHx+fy5cuNjY09PT319fU//fTTsmXLrl27NviPpcz+/fujoqLKysrwnSa3b98OCQm5ceMGl8slx0J9fHzwLSrbt29vampqa2uLj4/PyMiwtrYOCAigH6Ob+eDDLTc4PCqoNnGIroHprEY2NU5nlZOTgxA6fPgwNQYvPHr0KPVlcHCw3OxBjo6O7e3t5LuCgoKoaw0MDMLCwhBCFy5cwAG9vb2TJk1S/HXz8PBACN25c0e1TyQbzHRWAoHAyMioq6uLXIKns1q6dCk1rL29HZ9rIsp0Vt3d3e7u7opfGtSpmOjsz76+vjVr1vT7/UOdj4rObpF70IqUnp6OA3bs2KG41sDAQG7SppycHMWpoVSIGTDnYc5H9mpukbq6OiW7kYRgOisARjwPD4+cnJy5c+caGxtbW1uHh4cXFRWZmZmRAd9+++26deusrKyMjIzc3Nzy8/PlBnyYTOb58+c9PT37nZ1veISFhXV1deGxQSXMzMwUn55ms9n5+fkxMTHTp09ns9kcDmfhwoU5OTnUqZjoIAgiOTk5NTXVx8fH0tKSzWY7ODgEBARcuHDBx8eHDMOX1gwMhtQxeNeuXQkJCfPnz7e2tmaxWHZ2dqGhoWVl2VHw4wAAIABJREFUZR9++CE1TCAQFBQUeHt7czgcfLdqbm6uCjED5jzM+Uil0szMTD6fP2XKlEHuOf2n7WqqHnBmNrIN8/Ht92xDp9A/M+vu7h47dmxgYKCmUxoiqVRqYWHh7OxM3iGi+3QwZzw/C/2pohGcmekpXe4GMmBvi5cvX545c2bJkiX29vaGhoYTJ04MCQmRm0qADtwBhPoXsaKXL1/GxcXNnj3bxMTE3Nx81qxZ//3f/11cXCyTyVpbWxX7WchZvnw5orQacXJyUvwRhw4dwmt37tw52I8A6MPPRF+6dImcLF83VVRUtLa27ty5k+b9k7pAB3M+cODAzJkz8cXO0WZ0FbNBwc2B4uPjh+fHubq6btmyJS8v73U3K3/55ZehoaFZWVn19fU9PT0NDQ1paWmurq79tnQaCpFI5ObmFhsb++DBg87Ozvb29oqKihMnTvD5/F9//XWwW2OxWFVVVbdu3ZJbnpyc3O8cS0DtIiMjHRwc9uzZo+1ElBEKhVOmTMF/BukLXcv55s2bV69ePXDggNyltVECipmuGLC3BYfDwcXs8ePHHR0dd+7c8fX1lUgkmzZtUm8mX3/99YMHD2xtbdPT0//888+Ojo7KysqTJ08uWLCAwWBYWFhQT+3x3/sLFiygLqR2d1ywYAGHw0lOTqb+iLKysoqKivfff1+9mYN+sVishw8ffv/999pORJnIyMi6urohXjAbZrqWM2404+fnp+1EtENXDgMoLS3F/8jKyuo3QO62qLfeeuvixYuTJ09+/Phxc3OzlZWVujLBs7r985//XLx4MV7i5OTk5OS0du1aFbZmamr64Ycfpqamfv311yYmJnjht99+a2Zm9sEHH/Q7i5LWCQQCmZYmdAcAqGZUn5lJpdKjR4++++67lpaWFhYWc+bMOXToUEdHB3p9/w6y4cWNGzfc3d1NTU1tbW337duHN3j06NHp06cbGRk5OjoOQ388ExMTOzs7AwMD6s1yQ28vgqeNGOycPUqsXbu2vb2dfDy2o6Pj3LlzH3zwAfXOQAAAGIpRfWb2l7/85auvviJf/vzzzz///DObzd68ebPyN5aWln766ad4ZoeOjo6//OUvhoaGz549+/vf/44Dfvvtt+XLl7/55psanRn9t99+e/DgQUBAgJGRkRo3+/777//www8bN27829/+5uXlNfSS4+HhMX369JMnT+IWkefPnxeJROvWrRuF08cBADRkVJ+ZXbx40dTU9Mcff2xtbX358uXdu3ejoqLwd7eS/h0IodTU1M2bN9fV1YnF4vPnz7NYrLi4uMTExKSkpOfPnzc3N2/fvr2vr+/w4cOaS/7ly5crVqzgcrkHDx6kLh96e5GPPvooKiqqsrLy/ffft7CwmDFjxpo1ay5cuKA4ES19a9asKSoqevz4MULo5MmTb775Jp6eDgAA1GJUF7PJkydPnDhxyZIlXC7XxMTExcXlq6++et1UBVQCgeDw4cNTpkwxNTVdtmzZkiVL2traYmNjw8PDx44dO2bMmP3793O53MrKSg1l/vLly6VLl1ZVVV28eNHe3l7t2//qq68eP3585MiR4ODg3t7e7777LigoyM3NTeVzqdWrVzMYjOTk5JqamqKiItUuvwEAwOuM6mHGw4cPL1u2jMfj+fn5ubi4uLm5vf3223TeuGDBAupL/LA9dfYHJpM5adKkP//8U70JYy0tLf/1X/91//797OzsIc55qoSdnd3HH3/88ccfI4Tq6uo+/vjjS5cuffbZZ8ePH1dhaxMmTBAIBN99911vby+DwVi9erUKGxmdT8/0C98uBDsEANKoPjNzcXGpqqpKSUmZOnVqcXGxQCBwdnZ+8ODBgG+Uu0aF7w1RXDiUcbnXaWhoWLBgwYMHD3JycuRqqubY29ufPn2awWAMOCuSEmvXrv39998PHTq0aNGiiRMnqjE9AAAY1WdmCCEDAwM+n4/Pbzo6OqZPnx4eHo57Xmi9f4eimpoaX1/f5ubmK1eu9DsDrOa0tbX19fXJ9QYclCVLllhbWzc1Na1bt061LQzDDaL6Ap+TwQ4BQ6Q7c5cM3aguZu7u7qtWreLz+VOnTu3p6bl27VpzczPZX5Hav0ONT3GprKKiwtfXt6ur6+rVq3PnztXcD4qIiGAwGAEBAdOmTZs0adLLly9v3779l7/8BSE0lJ/LYrEaGxvVlyYAAPz/RnUx++WXX/ADwlTkBBxvvvnmpEmTrl+/bm1tjZdo9EHa0NBQauc98oHl9PR03EPryJEjuDWtq6ur3Hvv3LlD3r7o6el58+ZN6pJ+Xbt2TfGPsr1790ZHRzc2NmZmZipeG+NwOAcOHBj0BwMAAM0b1dfM/vWvf23atGnGjBm4x4eHh0dSUhJ5P70u9O/QiqSkpBMnTvj7+/N4PENDQ2NjY0dHx02bNt27d2/OnDnazg4AAPpBjIxpe9LS0kJCQkbGZwGK4PjKgWtmQC1we0+5pmh6alSfmQEA6JNIJNOmTcPTuCBKi58VK1bIRe7bt48gCOp807pmwI5LmFAo9PX15XK5HA5nwYIFeXl5mojBDTpwKzKgOs22SxsuutycU3nLMbnu9aBfw3x8FSd/0bWN02/OqUZff/21gYHBo0eP8EuyQRpBEPfu3aNGfvnll2gwLSKH3/jx46m/hjk5OYoxubm5cr1UCIJIS0vTRIyvr++sWbOkUqkmPqwSCJpzAgBGlZ6eni+++CIgIIDH41GXT506VSaTff7559pKTDUDdlzq6enZsGGDVCrdtm1bY2NjS0sL/vsjMjJSLBarNwYhFBUV9eDBAxg3HgooZhr31ltvKflr4uLFi9pOEICBZWRkPH/+nBxjJM2ePTsoKOjy5cuKNwbrstLSUjzMyGaz+w3Iz8+vr6/n8/kHDx60tra2sLCIjo4OCgpqamrKzMxUbwxCyNvbe8KECceOHdP0Bx/BoJiBEaWzszMuLs7JycnIyIjL5Xp7e8tdikhKSiIIguxHQ12I/7AYsPvPtWvX3N3dTUxMxo0bFxERQX14TuWNI3X07tGolJQUY2NjgUCguCo+Pp7BYOzatUvJ2wc8LuTuLS0t9fLyMjU1tbKyWr16tdxz+jKZ7OTJk+7u7hwOx9jY2MXF5ZtvvpFp4M6goqIihNDKlSupC0NDQxFChYWF6o1BCDGZzKVLlxYUFDx58kTNn2TUgGIGRo6enh5fX9/Y2Niqqqru7m6RSHT9+vXFixer8Q/eW7du+fn5lZSUdHZ2NjY2JiUl8fl86njRiNTb21tcXPzOO++wWCzFtU5OTqGhoQUFBVevXu337fSPS3l5uZeXV2FhYUdHx4sXL1JSUgIDA8m1MpksLCwsPDy8pKRELBZ3dXXdv39/8+bNGzZsUOOHxaqrqxFCM2fOpC6cPXs2uUqNMdi8efMQQtevX1fjpxhVoJiBkSMhIeHmzZt2dnZZWVltbW1PnjyJjY0lCGLr1q34eXM6lHf/SU9PDwsLe/TokVgsLioqmjVrVlVV1f79+4e+8aH37tGcyspKsVjs4uLyuoC4uDg2m/26K2f0j8vZs2fXrl376NGjjo4OoVBoZ2dXVFR07949vPb06dNnzpyZNWtWdnZ2c3OzWCwuLCx0cXE5ceKE2gc5RSIRejUNEAm/bGtrU28Mhg+9fo3W6hQoZmDkwNfPU1NT/f39zc3NbW1tY2Ji1q9f39XVdenSJbX8iLlz5548eZLH45mams6fP//ixYssFktuXHHkaWhoQAiRU+Eosre3j4iIKC8vz8jIUFxL/7gsWrQoMTGRx+MZGxt7eHh89tlnCCGymCUnJzOZzCtXrixevHjMmDGmpqZ8Pv/s2bMIIer1J7Xod+hSbqG6YrCxY8eiV7saqACKGRg5qqurrays8HANyd/fHykM6ahs0aJF1GnAHBwcpk2bVlNTo5aN6yzcx47D4SiJiY6ONjEx2b17t2KzCPrHxcvLi/rSwcEBIdTe3o5f/vrrr1Kp1NbW1sDAgMlkMhgMBoPh7OyMEFL7pSYul4sQkrti19LSQq5SYwxmbm6OXu1qoAIoZmBEGXAWcNwMQe4Ll5xdeog0unEtsrCwQK9GzF7HxsZmy5YtlZWVp0+fVlxLc3Z2Y2NjxXeR5zF4x0qlUqlU2tfXRx2k7enpobN9+vATCBUVFdSF9+/fJ1epMQbDo46WlpZq/BSjChQzMHLweLympibcwYeEe7CRXxzjxo1DCNXW1lJj5K66K+n+k5eXRx0jevz48cOHD9944w21bFxnTZgwASHU1NSkPGzHjh1cLjc2NlYikVCX0zkudDg6OpqYmLS2tio+4qL2kV7cFoo69zdCCNdpsiOuumIwvHvxrgYqgGIGRg48Y2FISEh2drZIJHr69OnevXuPHz9uaGi4ZMkSHDNjxgyE0JEjRwoKCjo7O+vq6rZv3y73tB+1+4/cjygrKwsPD6+urn758qVQKAwMDJRIJLitwdA3rrNmzJhhZmY24GMDlpaWUVFRtbW1p06doi6nc1zoCA8P7+jo8PHxuXz5cmNjY09PT319/U8//bRs2bJr164N/mMp4+Pjg28/2b59e1NTU1tbW3x8fEZGhrW1dUBAgHpjMLx75QZjwSCoOHOIjtHl6azA0NE8vt3d3f32LE1MTKSGBQUFUdcaGBiEhYUhhC5cuIADent7J02aJPdrkpOTgxAKDg6Wm5rI0dGxvb19iBuXyWQeHh4IoTt37tDZIcM/nZVAIDAyMurq6iKX4Oms5OZja29vx6eniDKdFZ3jgnfv4cOHqVvDC48ePYpf9vX1rVmzpt/vMep8VHT2pNyDX6T09HTqT1echkpu8id1xchkso0bNyKE6urqlKStdgimswJAB7HZ7Pz8/JiYmOnTp7PZbA6Hs3DhwpycHPw1Qfr222/XrVtnZWVlZGTk5uaWn58vN+CjpPuPh4dHTk7O3Llzcdug8PDwoqIiMzMztWxcl4WFhXV1deGxQSXMzMwUn56meVwGRBBEcnJyamqqj4+PpaUlm812cHAICAi4cOGCj48PGYYvrRkYDLVZo0AgKCgo8Pb25nA4+ObV3NxcuQnm1RUjlUozMzP5fP6UKVOGmPbope1qqh5wZjay6cLx7ffUQVuG/8ysu7t77NixgYGBw/lDVSCVSi0sLJydnck7RPQCng9l+KdmRnBmBgAYVfAz0ZcuXSIny9dNFRUVra2tO3fupHn/pI44cODAzJkz8cVFoBooZgAAWiIjIx0cHPbs2aPtRJQRCoVTpkxZvny5thMZhJs3b169evXAgQNyl9bAoAx1WBkAMEqwWKyHDx9qO4sBREZGRkZGajuLwfHw8JBBF/Uhg2IGAC0CgQC+cQDQWTDMCAAAQO9BMQMAAKD3oJgBAADQe1DMAAAA6D0oZgAAAPTeiLqbUb8ekwSDBcdXDuwQAEjEyLjb+OnTp7du3dJ2FgD04/DhwwihrVu3ajsRAPrh7u4+efJkbWehBiOkmAGgs/CUsmlpaf9fe/ceFcWR9g+8muEyXEZA8UJUREIU8YJJ/KkIGXABwcRXBNmQENAkhDWLusd4iZ4EAiiuugeV82oObtaoa0RFDOKNi6AioLiyZ3UVXYIIYoi4AeU2MNyG+f1Ra7+9Aw4NDAzNfD9/ZapremrKTD9UdXU92m4IwHCGe2YAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4CGYAACB4+tpuAMBwU1NT09DQwL5samoihJSVlbElI0aMsLKy0kLLAIYvRqlUarsNAMPK4cOHP/30UzUVDh069MknnwxaewB0AYIZgIbV19ePHj26vb2926MGBgbV1dXm5uaD3CqA4Q33zAA0zNzc/N1339XX72YOX19f/7333kMkA9A4BDMAzQsODlYoFF3LOzs7g4ODB789AMMephkBNK+lpcXKyoou/eAyMTGpqakxNjbWSqsAhjGMzAA0TywW+/v7GxgYcAsNDAwCAgIQyQAGAoIZwIAICgpSWQPS3t4eFBSkrfYADG+YZgQYEB0dHWPHjn3x4gVbYmFhUV1d3e3CEADoJ4zMAAaEvr7+hx9+yM40GhgYBAcHI5IBDBAEM4CB8uGHH7Izje3t7R9++KF22wMwjGGaEWCgKJXKiRMn/vLLL4QQa2vrX375hWEYbTcKYHjCyAxgoDAMExISYmhoaGhouHLlSkQygIGDkRnAALp7966TkxP9j5kzZ2q7OQDDlq7cjt6zZ09BQYG2WwG6yMzMjBCydetWbTcEdJGzs/P69eu13YrBoCvTjAUFBTdv3tR2K0AwTp8+XVlZqZFTTZo0ydbWViOn0pbKysrTp09ruxXQazdv3tSdP+J1ZWRGCJk/f35ycrK2WwHCwDDMF1988f777/f/VDSTmZ2dXf9PpS2nTp0KDAzEz0dwfvvb32q7CYNHh4IZgFYIOowBCIWuTDMCAMAwhmAGAACCh2AGAACCh2AGoGF///vfGYaJjo7WdkO0o729fcqUKStWrKAvS0tLGYZhGKZrxoCdO3cyDHPy5MlBbyNfFRUV+/fv9/b2NjIyYhgmIyOj22r5+fleXl7m5uYSicTNze3SpUsDUSc/P59hmMzMTI18teEHwQxg+KDXu9jYWC22ISEhoby8/JtvvlEpP3ny5N27d7XSpD6bN2/e2rVrL1261NbW9qo6mZmZ7u7u2dnZDQ0NMpksNzfXx8dHZeWnRuq4urp6eXlt2rSps7NT4990GEAwA9CwOXPmKJVK3RyZtbW1bd++fdmyZfb29tzyyZMnK5XKr7/+WlsN6xtbW9vVq1dnZGSEhYV1W6GtrW3VqlUKhWL9+vXV1dW1tbXbtm1TKpXh4eEymUyzdQghGzduvHfvHp6R6BaCGQBoTEpKyq+//srOMbJmzZrl7+9/4cIFYT3De/PmTTrNaGho2G2F7OzsiooKqVS6e/duKysrCwuLiIgIf3//mpqas2fParYOIcTDw8Pa2vrAgQMD/cWFCMEMQMNU7pllZGQwDBMfH3/z5k13d3dTU9NRo0atXLmSm7eTrXP58uUFCxaYmJiMGTMmLCysurqarXPw4EGGYVR24qCFqamphJDY2Nh33nmHEBIZGcm8RKu5uroyDHPnzp0B/urk6NGjxsbGPj4+XQ/Fxsbq6el99dVXat4ul8tjYmKmTZsmFovNzc09PDxUbhHx6UxCiFKpPHTo0IIFCyQSibGxsZOT07fffjsQ+9Dm5uYSQj766CNuYXBwMCHk2rVrmq1DCBGJRL6+vjk5OU+ePNHwNxE+BDOAwVBYWOju7n7t2rXm5uYXL14cPXrUz89Ppc6NGze8vb0LCgrkcnl1dfXBgwelUil3lmmI6+joyMvLe+utt9iUpFzTpk0LDg7OycnJysrq9u1tbW1eXl7R0dHFxcWtra0NDQ1XrlxZvHhx14GI+s5UKpUhISGhoaEFBQUymaylpeXu3btr1qxZtWqVBr8sVVpaSgiZMWMGt3DWrFnsIQ3WoebPn08IuXLliga/xfCAYAYwGI4fP/7JJ588fPiwubk5Pz/fxsYmNzf3n//8J7dOcnJySEjIw4cP6f3/mTNnFhcX79q1i+dHRERE5OXlEULo7RaKHsrPz1cqlbNnz9bsl1Lx4MEDmUxGswR0KyYmxtDQ8FV3zvbv33/9+nUbG5vz58/X19c/efIkOjqa7iv27Nkzbk31nXns2LHExMSZM2empaU9f/5cJpNdu3bNycnpL3/5i8YnORsaGgghI0eO5BbSl/X19ZqtQ9F/RGHN1g4OBDOAwbBo0aKEhAR7e3tjY2MXF5cvv/ySEKISzObOnXvo0CF7e3tTU9N33nknNTXVwMBAQDv8VlVVEUKsrKxeVcHW1jYsLKywsDAlJaXrUbquISkpacmSJSNGjJg4cWJUVNRnn33W0tJy7tw5bk31nXn48GGRSJSZmbl48eKRI0eamppKpdLjx48TQrj3nzSi26lLlUJN1aFGjx5NXnY1cCGYAQwGd3d37ku6YWNjYyO3cNGiRdwEnnZ2dlOmTHn06NGgNFAD6urqCCESiURNnYiICBMTk8jIyK7ry0tLS0eNGkWn0VhLliwhXaba1Hfm/fv3FQrFxIkT9fX1RSKRnp6enp7e9OnTCSEav9Vkbm5OCFG5Y1dbW8se0mAdasSIEeRlVwMXghnAYDA2Nua+pEGrV0sS9PT0CCEqMUAul2uidZphYWFBXs6Yvcq4cePWrl374MGDY8eOdT3KMxm3+s6kXaRQKBQKRWdnJ3e6Vc3jYn1Dn0AoKiriFtLH6diHEzRVh6KzjpaWlhr8FsMDghnAUHHp0iVueCsrKyspKXn99dfpyzFjxhBCysvLuW9RWQhAA15HR8eAt7U71tbWhJCamhr11TZv3mxubh4dHd3e3s4tt7e3r6mpuXXrFrcwLS2NdLmgq+fg4GBiYlJXV6fsQuNztlKplBCSmJjILaRxmh7SYB2Kdi/tauBCMAMYKm7duhUaGlpaWtrU1JSfn+/n59fe3h4QEECPOjo6EkLi4+NzcnLkcvnjx483bNhAF+Wz6JKBvLy858+fD377HR0dzczMenwAwNLScuPGjeXl5UeOHOGW0+RbgYGBaWlpDQ0NlZWV27Zt++6774yMjJYuXcq/GaGhoc3NzZ6enhcuXKiurm5ra6uoqLh48eLy5csvX77c+6+ljqenJ11+smHDhpqamvr6+tjY2JSUFCsrq2XLlmm2DkW7V2UyFgghpOsfL8NSQEBAQECAtlsBgkEISUpK6tt7CwsLCSFRUVH0ZXp6OiFk79693Dq0cN++fdyXAQEBIpGI+/N0cHBobGxk3+Xv7889qq+vHxISQgg5c+YMrdDR0TF+/PiuP3AXFxdCyO3bt/v2jZKSknheK3x8fMRicUtLC1vy8OFDQoivry+3WmNjIx1oEkJOnDhBC1tbWxcsWND1GpWQkMC+kU9ndnZ2fvzxx91e7tLT09l38ekTlQe/WMnJydxPV/lXYxhG5X8eTdVRKpWff/45IeTx48dqms3SqeseRmYAQ4WLi0t6evrcuXONjY2trKxCQ0Nzc3PNzMzYCt9///2nn346atQosVjs7OycnZ2tMgclEolOnz7t6upqamo66M0nhJCQkJCWlhY6N6iGmZlZ16enDQ0Ns7Ozo6Kipk6damhoKJFIFi5cmJ6eTi/f/DEMc/jw4aSkJE9PT0tLS0NDQzs7u2XLlp05c8bT05OtRm+t6ev3N0Gxj49PTk6Oh4eHRCKhy1AzMjJUcpRrqo5CoTh79qxUKp00aVI/mz0MaTuaDhKd+gsF+o/0Y2TWB90OOIYO/iOz1tbW0aNH+/n5DXST+kmhUFhYWEyfPp1dISIIdD8UdizbI5267mFkBgAaQ5+JPnfuHJ1dHLKKiorq6uq2bNnCc/3kEBEXFzdjxgx6cxFUIJj17MiRI133xBv6mpqaEhMTly5damtra2Rk9NprrwUGBt6+fVulWltb2x//+Mfp06eLxeKRI0f6+vr2YQc/NmcVJZFIbGxs3n333bi4OO7ugqALwsPD7ezstm7dqu2GqJOfnz9p0qQPPvhA2w3phevXr2dlZcXFxancWgOqv/PFMGTt2LFj+/bt7MuqqqpTp06dOXPm3Llz7D6wHR0d7733XnZ2Nn3Z2tp67ty5zMzMrKwsumVt38hkMplM9vPPP6enp2/dujUhIeFVN9Jh+DEwMCgpKdF2K3oQHh4eHh6u7Vb0jouLi3IA9koeNjAyG7YkEklwcPD58+fLysqam5tv377t5eXV3t6+evVqts7Ro0ezs7PHjx9/4cKFhoaGX3755Ztvvmltbf3ss88UCkVvP9HDw4NOXre0tFRUVCQnJ7u7uzc2Nq5YseLixYsa/XLDjY+Pj1KpXLdunbYbAiBUCGbD1ubNm3/44YclS5ZMnjzZ2Nh49uzZqamplpaWZWVl7ENIdMu7b7/99r333pNIJK+99lpMTMyyZctKSkquXr3a5482MjKysbEJCAi4evXqunXrOjs7169fj/S4ADBwEMxU1dXVrVmzxtra2tjYeM6cORcuXOhaR9lTtiQ+WZcUCsW+ffvefvttS0tLCwuLOXPm7Nmzp7m5mf+n9JaJiYmNjY2+vj67bvvf//43ebkPN4u+zMnJYUv6kw1r586d48aNKykp4b5doB0IAEOXNpZQagHPJapyuVzlys4wTGBgIOE8JtnZ2dntHaCwsDD2PHSldVBQkJGREbeOVCpl62zatKnrSbgPfvb4Kb1VXFysp6fn7+/PltBdXFNTU7nV6I4D77//PlvS4+OldOkaO82oYuXKlYSQP//5z/y/mtY7kAzu0vwhjv/SfBhSsDRfd+3bt+/OnTtTp07Nzs5ubGwsKyv7/e9/T3/JLP7ZktRnXUpNTTU1Nf3xxx/r6uqampru3LmzceNG9glZjedkampqCgoKMjc33717N1tIV4KsWbPm4sWLMpns6dOnUVFRNE0Gd1vufmbDog94slv2CbQDAWBI03Y0HSQ8/0KZN28ewzBFRUXcQg8PD8IZmS1cuFAkEj19+pRb5/79+4SQzZs305d0YLFo0SJunf379xNC/vrXv7LneeONN9rb27ttCZ9P4U8mk3l4eJiYmFy7do1b3tLS8v/+3/9T+V9ixYoV5OWSBJ7Uj8wiIyMJIdu3b6cvBdGBBCMzDozMBEqnRmZYmv9fSktLx48fT1MfsXx8fLj7k7LZksjLnzd5mX5CJVuS+qxLe/fuXb58ub1AWzKUAAAgAElEQVS9vbe3t5OTk7Oz85tvvtmHT+lRbW3te++9d/fu3bS0NJXdj4yMjK5evRobG3vq1KnKysrx48dv2LBh7NixR48epTkANaKiooK8TCpIhNOBgYGBdIYZKGE9XwwUu1H1sIdg1mtstqSuh1SyJanPuuTk5FRcXHzjxo0bN27k5eXFxMRYWVmdPHly5syZvfoU9aqqqry9vcvLy9PT07t9dMzU1HTHjh07duxgS373u98RQt5++23+n6JGa2vrpUuXuCcUSgeuW7fO2dmZT81hr6CgID4+XmW+HYa+vXv3arsJgwfB7L/Y29vfunXr/v373MFZRkYGt46Dg8M//vGPp0+fqmSA7QN9fX2pVEpHS83NzVOnTg0NDaX5nDTyKY8ePfLy8nr+/HlmZma3+5F3VVpaeuzYMZFIpLJBe59t2bLl2bNnU6ZMYW+5CaUDnZ2dVbZ51WXx8fHoDcFJTk7WdhMGDxaA/Jfly5crlcrly5dfuXJFJpOVl5evXr1aJQeSprIlLViw4MCBAw8ePJDL5fX19RkZGc+fPy8rK9PUpxQVFbm6utbW1mZlZamJZMuWLbtw4cLz588bGxvPnDnj4eEhl8s///xzOkHXN62trT///POPP/74m9/8Jj4+nmGYuLg4mjdSI1+NGugOBAAhGcT7c9qk2aX5PWZL4pN1SWXROfWHP/yB/6eoFxoa+qp/dO4i+7Fjx6oc9fDwaG5u5p6K59L8bpmZmR05coRbWRAdSLAAhAMLQARKpxaAYGT2X8Ri8dWrV8PDw8eOHSsWi998880zZ86wOxlSPLMl9ehvf/vb6tWrHR0dafIqFxeXgwcPsnPcmvqUHp0/f97f33/MmDGmpqZvv/32vn37MjIyVG5W9ZaJicmECRO8vb3/9Kc/PXr0iD5nxhpmHQgAQwGj1I0NEWjSBJ2aQYb+oEl+cZeIOnXqVGBgoI5cK4YTnbruYWQGAAOrvb19ypQp9PlFwkkYFBQUpFJz586dDMOcPHly0NvIC8+0SoSQ/Px8Ly8vc3NziUTi5uZGF/T2qk5+fj7DMDQbJ/CBYCZId+7cYV6N7kcFQkEvW7GxsYI7OU8JCQnl5eXffPONSvnJkyfv3r2rlSb1zY4dO2gmioqKira2NppWad68eSoLnjMzM93d3bOzsxsaGmQyWW5uro+Pj8rwqMc6rq6uXl5emzZtwg7dPCGYAcAAamtr2759+7Jly+zt7bnlkydPViqVX3/9tbYa1gd80iq1tbWtWrVKoVCsX7++urq6trZ227ZtSqUyPDxcJpPxr0MI2bhx471793RkkrD/EMwEafbs2WpW9aSmpmq7gQD/kZKS8uuvv7JzjKxZs2b5+/tfuHBBQFtl8kmrlJ2dXVFRIZVKd+/ebWVlZWFhERER4e/vX1NTQ3c95VmHEOLh4WFtbX3gwAEtfFUBQjAD6Du5XB4TEzNt2jSxWGxubu7h4aFyk+PgwYMMw5w+fbprIf2bIzY2lu7MEhkZyU4UE04SnMuXLy9YsMDExGTMmDFhYWHV1dX9PznpX1qfXjl69KixsbHKkmAqNjZWT0/vq6++UvP2HnuYT7YgMmD5gLqmVcrNzSWEqGRsCA4OJoRcu3aNfx1CiEgk8vX1zcnJ6e0OdroJwQygj9ra2ry8vKKjo4uLi1tbWxsaGq5cubJ48WIN/il948YNb2/vgoICuVxeXV198OBBqVTKnYka4jo6OvLy8t566y0DA4OuR6dNmxYcHJyTk5OVldXt2/n3cGFhobu7+7Vr15qbm1+8eHH06FE/Pz/2qFKpDAkJCQ0NLSgokMlkLS0td+/eXbNmzapVq/r5BX/66ad79+4tXbpULBbTktLSUkLIjBkzuNVmzZrFHuJZh5o/fz4h5MqVK/1spy5AMAPoo/3791+/ft3Gxub8+fP19fVPnjyJjo5mGOaLL7549uwZz5NERETk5eURQuhdE4o9mpycHBIS8vDhQ7pGYObMmcXFxbt27er/yfuZ1oenBw8eyGQyJyenV1WIiYkxNDR81Z0z/j2sPlvQAOUD6jatUkNDAyFk5MiR3Jr0ZX19Pf86FP0HEtBMrBYhmAH0Eb0zn5SUtGTJkhEjRkycODEqKuqzzz5raWk5d+6cRj5i7ty5hw4dsre3NzU1feedd1JTUw0MDFTmFYeyqqoqQoiVldWrKtja2oaFhRUWFqakpHQ9yr+HFy1alJCQYG9vb2xs7OLi8uWXXxJC2GB2+PBhkUiUmZm5ePHikSNHmpqaSqXS48ePE0K496h6pampydfXt7i4ODU11dbWli3vdupSpZBPHYrmmqDdCOohmAH0UWlp6ahRo+hEEIsm71aZLOqzRYsWcROv2NnZTZky5dGjRxo5+SCgKV4lEomaOhERESYmJpGRkV3XoPPvYfXZgth8QPr6+iKRSE9PT09Pj24m3rfbUbW1tV5eXjdv3uyaVolubK1yx662tpY9xLMONWLECPLfmXLhVRDMAPquxxRfdHtllcu0XC7XyKcP6Mk1wsLCgrycVXuVcePGrV279sGDB8eOHet6lGcSNfXZgth8QAqForOzkzvd2quESlRVVZWbm9u9e/fS09Pd3NxUjtInEIqKiriF9HE69uEEPnUoOutoaWnZ20bqIAQzgD6yt7evqamhGWdYaWlphHNJGjNmDCGkvLycW0flfj6NSR0dHV0/4tKlS9zZp7KyspKSktdff10jJx8E1tbWhJCamhr11TZv3mxubh4dHd3e3s4t59PDfDg4OJiYmNTV1XV9jqW3c7aPHj1ycXGpqKjIzMzsNkEgHaglJiZyC2mcZsdwfOpQtOtoN4J6CGYAfUQ3vgsMDExLS2toaKisrNy2bdt3331nZGS0dOlSWsfR0ZEQEh8fn5OTI5fLHz9+vGHDBpUHAemd/7y8PPZZJdatW7dCQ0NLS0ubmpry8/P9/Pza29vZ3MH9PPkgcHR0NDMz6/EBAEtLy40bN5aXlx85coRbzqeH+dBUPiA+aZU8PT3p8pMNGzbU1NTU19fHxsampKRYWVmxW/PwqUPRrlOZaIXu9XW7fYHRqVQI0H+ERwqY1tbWbq9oCQkJ3GoqaU719fVDQkIIIWfOnKEVOjo6xo8fr/KrpPluAgICRCIR95CDg0NjY2M/T67kkdaHqz8pYHx8fMRicUtLC1tCEwb5+vpyqzU2NtKBJiHkxIkTtJBPD/PJFsQzH1CPfcIzrVJ6errKvxrdt1qlhT3WUSqVn3/+OSHk8ePHavv4lXTquoeRGUAfGRoaZmdnR0VFTZ061dDQUCKRLFy4MD09nV6AWN9///2nn346atQosVjs7OycnZ2tMpUkEolOnz7t6urKPnjLcnFxSU9Pnzt3Lk1zExoampuba2ZmppGTD46QkJCWlhY6N6iGmZlZ16enefZwj3jmA6K31vT19Xt18q58fHxycnI8PDwkEgldhpqRkaGSgYFPHYVCcfbsWalUOmnSpH42SSdoO5oOEp36CwX6j2g7OWe3Aw5t6c/IrLW1dfTo0X5+fpptksYpFAoLC4vp06ezK0S0ju51wo5T+0CnrnsYmQHAAKLPRJ87d05NOvKhoKioqK6ubsuWLTzXTw6CuLi4GTNm0BuH0CMEMwAYWOHh4XZ2dlu3btV2Q9TJz8+fNGnSBx98oO2G/Mf169ezsrLi4uJUbq3Bq/R3dhgAQD0DA4OSkhJtt6IH4eHh4eHh2m7F/3FxcVEitXdvIJgBDEU+Pj64lgHwh2lGAAAQPAQzAAAQPAQzAAAQPAQzAAAQPB1aAFJZWXnq1ClttwIEAxkRWbQr8PMRnMrKygkTJmi7FYNF209tDxJ2b1YAAN2hOzuAMEos/wUYSHTDPQxrAAYU7pkBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDgIZgBAIDg6Wu7AQDDTW5ubkFBAfuyuLiYELJr1y62xNnZWSqVaqFlAMMXo1Qqtd0GgGHl8uXLnp6eBgYGenqqMx+dnZ3t7e3Z2dkeHh5aaRvAcIVgBqBhnZ2d48aNq66u7vaolZXVs2fPRCLRILcKYHjDPTMADdPT0/voo48MDQ27HjI0NAwODkYkA9A4BDMAzfvwww/b2tq6lre1tX344YeD3x6AYQ/TjAADwtbWtqKiQqVw4sSJFRUVDMNopUkAwxhGZgADIiQkxMDAgFtiYGDw8ccfI5IBDASMzAAGRHFx8bRp01QKi4qKpk+frpX2AAxvGJkBDAgHB4fp06dzx2GOjo6IZAADBMEMYKCsWLGCXbhoYGCwcuVK7bYHYBjDNCPAQPn5558nTZpEf2IMw5SVldna2mq7UQDDE0ZmAANl4sSJ8+bN09PT09PTmzdvHiIZwMBBMAMYQCEhIQzD6OnphYSEaLstAMMZphkBBlBNTc24ceMIIU+fPh0zZoy2mwMwfCl1VUBAgLb7HgBAkwICArR9ZdUanU4BM3/+/C+++ELbrQBhCAwMXLdunbOzc2/fmJubyzDMO++8MxCt0paCgoL4+PikpCRtNwT+z969e7XdBG3S6WA2YcKE999/X9utAGEIDAx0dnbuw/8wixcvJoRIJJIBaJQ2xcfH4+czpCQnJ2u7Cdqk08EMYBAMvzAGMARhNSMAAAgeghkAAAgeghkAAAgeghnAgPv73//OMEx0dLS2G6Id7e3tU6ZMWbFiBX1ZWlrKMAzDMEFBQSo1d+7cyTDMyZMnB72NvDQ1NSUmJi5dutTW1tbIyOi1114LDAy8fft215r5+fleXl7m5uYSicTNze3SpUu9rZOfn88wTGZm5kB9mWEHwQxgOKPXxNjYWC22ISEhoby8/JtvvlEpP3ny5N27d7XSpL7ZsWNHcHDw+fPnKyoq2traqqqqTp06NW/evIyMDG61zMxMd3f37OzshoYGmUyWm5vr4+OjstSwxzqurq5eXl6bNm3q7OwcpK8ncAhmAANuzpw5SqVSN0dmbW1t27dvX7Zsmb29Pbd88uTJSqXy66+/1lbD+kAikdBgVlZW1tzcfPv2bS8vr/b29tWrV7N12traVq1apVAo1q9fX11dXVtbu23bNqVSGR4eLpPJ+NchhGzcuPHevXs6vuCePwQzABhAKSkpv/76KzvHyJo1a5a/v/+FCxcKCgq00rA+2Lx58w8//LBkyZLJkycbGxvPnj07NTXV0tKyrKzs+fPntE52dnZFRYVUKt29e7eVlZWFhUVERIS/v39NTc3Zs2f51yGEeHh4WFtbHzhwQAtfVYAQzAAGnMo9s4yMDIZh4uPjb9686e7ubmpqOmrUqJUrV7548YJ9C1vn8uXLCxYsMDExGTNmTFhYWHV1NVvn4MGDDMOcPn2a+1m0MDU1lRASGxtLdx6JjIxkXqLVXF1dGYa5c+fOAH91cvToUWNjYx8fn66HYmNj9fT0vvrqKzVvl8vlMTEx06ZNE4vF5ubmHh4eKreR+HQmIUSpVB46dGjBggUSicTY2NjJyenbb79V9ntnWhMTExsbG319fVNTU1qSm5tLCPnoo4+41YKDgwkh165d41+HECISiXx9fXNycp48edLPduoCBDMA7SgsLHR3d7927Vpzc/OLFy+OHj3q5+enUufGjRve3t4FBQVyuby6uvrgwYNSqZQ7EzXEdXR05OXlvfXWWwYGBl2PTps2LTg4OCcnJysrq9u3t7W1eXl5RUdHFxcXt7a2NjQ0XLlyZfHixV0HK+o7U6lUhoSEhIaGFhQUyGSylpaWu3fvrlmzZtWqVf38gj/99NO9e/eWLl0qFotpSWlpKSFkxowZ3GqzZs1iD/GsQ82fP58QcuXKlX62UxcgmAFox/Hjxz/55JOHDx82Nzfn5+fb2Njk5ub+85//5NZJTk4OCQl5+PAhXSMwc+bM4uLiXbt28fyIiIiIvLw8Qgi9JUPRQ/n5+Uqlcvbs2Zr9UioePHggk8mcnJxeVSEmJsbQ0PBVd872799//fp1Gxub8+fP19fXP3nyJDo6mmGYL7744tmzZ9ya6jvz2LFjiYmJM2fOTEtLe/78uUwmu3btmpOT01/+8pf+THI2NTUFBQWZm5vv3r2bLWxoaCCEjBw5kluTvqyvr+dfh6L/QAKaidUiBDMA7Vi0aFFCQoK9vb2xsbGLi8uXX35JCFEJZnPnzj106JC9vb2pqek777yTmppqYGCgMq84lFVVVRFCrKysXlXB1tY2LCyssLAwJSWl61G69iEpKWnJkiUjRoyYOHFiVFTUZ5991tLScu7cOW5N9Z15+PBhkUiUmZm5ePHikSNHmpqaSqXS48ePE0K496h6pampydfXt7i4ODU1lZt2tdupS5VCPnWo0aNHk5fdCOohmAFoh7u7O/elnZ0dIaSxsZFbuGjRIvYuF60zZcqUR48eDUoDNaCuro70tDtlRESEiYlJZGRk1zXopaWlo0aNolNtrCVLlpAu03HqO/P+/fsKhWLixIn6+voikYjm/p4+fTohpG+3o2pra728vG7evJmWliaVSrmHzM3NCSEqd+xqa2vZQzzrUCNGjCAvuxHUQzAD0A5jY2PuSxq0erUkQU9PjxCiEgPkcrkmWqcZFhYW5OWs2quMGzdu7dq1Dx48OHbsWNej3FiuhvrOpF2kUCgUCkVnZyd3urWtrY3P+bmqqqrc3Nzu3buXnp7u5uamcpQ+gVBUVMQtpI/TsQ8n8KlD0VlHS0vL3jZSByGYAQxdly5d4oa3srKykpKS119/nb6kqavLy8u5b1FZLEADXkdHx4C3tTvW1taEkJqaGvXVNm/ebG5uHh0d3d7ezi23t7evqam5desWtzAtLY10ueir5+DgYGJiUldX1zWjY2/nbB89euTi4lJRUZGZmdltjjo6UEtMTOQW0jjNjuH41KFo19FuBPUQzACGrlu3boWGhpaWljY1NeXn5/v5+bW3t7NJ0h0dHQkh8fHxOTk5crn88ePHGzZsoIvyWXRZQV5eHvsg1GBydHQ0MzPr8QEAS0vLjRs3lpeXHzlyhFv+29/+lhASGBiYlpbW0NBQWVm5bdu27777zsjIaOnSpfybERoa2tzc7OnpeeHCherq6ra2toqKiosXLy5fvvzy5cv8z1NUVOTq6lpbW5uVlbVgwYJu63h6etLlJxs2bKipqamvr4+NjU1JSbGyslq2bBn/OhTtOpWJVujeQKWwHvICAgJ0OcU49BYhJCkpqW/vLSwsJIRERUXRl+np6YSQvXv3cuvQwn379nFfBgQEiEQi7g/WwcGhsbGRfZe/vz/3qL6+fkhICCHkzJkztEJHR8f48eO7/uRdXFwIIbdv3+7bN6I5pvnU9PHxEYvFLS0tbMnDhw8JIb6+vtxqjY2NdKBJCDlx4gQtbG1t7TZmJCQksG/k05mdnZ0ff/xxtxfA9PR09l099kloaOirLqTcd6Wnp6v8qzEMo/I/D586SqXy888/J4Q8fvxYbR//h45f0zAyAxi6XFxc0tPT586da2xsbGVlFRoampuba2Zmxlb4/vvvP/3001GjRonFYmdn5+zsbJV5KpFIdPr0aVdXV/ap3kEWEhLS0tJC5wbVMDMz6/r0tKGhYXZ2dlRU1NSpUw0NDSUSycKFC9PT0+klnj+GYQ4fPpyUlOTp6WlpaWloaGhnZ7ds2bIzZ854enqy1eitNX39/qYs9vHxycnJ8fDwkEgkdBlqRkaGSlZuPnUUCsXZs2elUumkSZP62SSdoO1oqjU6/lcM9Bbpx8isD7odcAwd/Edmra2to0eP9vPzG+gm9ZNCobCwsJg+fTq7QkTr6F4n7Di1Rzp+TcPIrC+OHDnSdRshQcjLy/vss8+mTZtmbGw8cuRILy+vCxcudK3GJ4GFemyaD0oikdjY2Lz77rtxcXHcDZlg2KPPRJ87d47OLg5ZRUVFdXV1W7Zs4bl+chDExcXNmDGD3jiEHiGY6ZDS0lKpVPr9998XFxe3tLTU1tZmZ2f/z//8z549e7jV+CSw6C2ZTPbzzz+np6dv2rTp9ddfV1nHBcNbeHi4nZ3d1q1btd0QdfLz8ydNmvTBBx9ouyH/cf369aysrLi4OJVba/AqCGY6RE9Pz9vbOzExsaSkRC6Xl5eX040SvvrqK/ZJIJ7JKXjy8PCgMwAtLS0VFRXJycnu7u6NjY0rVqy4ePGihr8eDFUGBgYlJSU//PCDthuiTnh4+OPHj/t/w0xTXFxclEqlt7e3thsiGAhmOsTOzi4jIyMoKOiNN94Qi8W2tra7du1yc3NrbW3917/+RevwTE7RW0ZGRjY2NgEBAVevXl23bl1nZ+f69euRdVANHx8fpVK5bt06bTcEQBgQzHpWV1e3Zs0aa2trY2PjOXPmdHuTSdlTggk+iSoUCsW+ffvefvttS0tLCwuLOXPm7Nmzp7m5mf+n9AHdzpxdFc0zOUV/Eojs3Llz3LhxJSUl3LcLtwMBYEjQwqKToYHnyh+5XK6yszjDMIGBgYSQ5ORkWqezs1Pl6k+FhYWx56GL04KCgoyMjLh1pFIpW2fTpk1dT8J9VqbHT+FPoVDQR1DJy0EAtXz5ckLI9evXuZXpPngLFy5kS3p8Iofe7WenGVWsXLmSEPLnP/+Z/1fTegeSwV3NOMTxX80IgwarGUGdffv23blzZ+rUqdnZ2Y2NjWVlZb///e/pL5nFP8GE+kQVqamppqamP/74Y11dXVNT0507dzZu3Mg+VKSpNBbFxcUMw4hEogkTJmzfvj08PJy7uINncop+JhChz82wuxwJqwMBYCjSdjTVGp5/xcybN49hmKKiIm6hh4cH4YzMFi5cKBKJnj59yq1z//59QsjmzZvpSzqwWLRoEbfO/v37CSF//etf2fO88cYb7e3t3baEz6fwwd4eo958803uOIw+Q/qvf/2L+xa6E9Jbb73F/1PUj8wiIyMJIdu3b6cvBdGBBCMzDozMhiAdH5kNlaU7Q1Zpaen48eNptgiWj48Pd0s3NsEEefnzJi937FZJMKE+UcXevXuXL19ub2/v7e3t5OTk7Oz85ptv9uFT1HNwcFAqlQqF4t///ndaWtqGDRs8PT3v378/efJk0pvkFP1RUVFBXuZqIsLpQAzgWLQrTp06pe2GwP+prKycMGGCtluhNQhmGsAmmOh6SCXBhPpEFU5OTsXFxTdu3Lhx40ZeXl5MTIyVldXJkydnzpzZq0/hQyQSvfbaazTP4dq1a5OTk+kyfTY5BXdPvG6TU/RZa2srfQr77bffpiVC6cD4+Pj4+Hg+NXUEvXkMQwe7CbUOQjDrgb29/a1bt+7fv88dnGVkZHDrODg4/OMf/3j69Gn/xy76+vpSqZRur9fc3Dx16tTQ0FCaAkODn8LV2tpKOBmnpFLprl27EhMTf/e737F1uk1O0Wdbtmx59uzZlClT2FtuQunApKQkld3zdNapU6cCAwOVWAg6lOj4XiFYANKD5cuXK5XK5cuXX7lyRSaTlZeXr169WiVthKYSTCxYsODAgQMPHjyQy+X19fUZGRnPnz8vKyvT1Kf88Y9/3Lx5c2Fh4fPnz1taWsrLy//3f//3m2++IYSwmZn4J6foldbW1p9//vnHH3/8zW9+Ex8fzzBMXFwcTbWlka9GDXQHAsDQNWh354YazS7N7zHBBJ9EFSqLzqk//OEP/D9FvQ0bNnT79o8++kilVT0mp+C5NL9bZmZmR44c4VYWRAcSLADhwAKQIUjHF4BgZNYDsVh89erV8PDwsWPHisXiN99888yZMz4+Ptw6PBNM9Ohvf/vb6tWrHR0dab4PFxeXgwcP7t27V1OfEhkZ+e2337q5uY0ZM8bAwGDs2LGLFy8+ceKEyj5DfJJT9JaJicmECRO8vb3/9Kc/PXr0iD5nxhJKBwLAkMUodXXWm84v93P/XNAddHiKe2YU7pkNQTp+TcPIDAAABA/BbJi4c+cO82r9WbsBoHHt7e1TpkxZsWIFfclmvwsKClKpuXPnToZhTp48Oeht5KWpqSkxMXHp0qW2trZGRkavvfZaYGDg7du3u9ZUnyMwPz+fYRiajRP6BsEMYMihl7bY2FjBnZynhISE8vJyupKW6+TJk/ShRqHYsWNHcHDw+fPnKyoq2traqqqqTp06NW/ePJWnd3rMEejq6url5bVp0yakkugzBLNhYvbs2WrW+aSmpmq7gQD/0dbWtn379mXLlqk8hj958mSlUvn1119rq2F9IJFIaDArKytrbm6+ffu2l5dXe3v76tWr2To8cwRu3Ljx3r17OnvHq/8QzABgUKWkpPz666/sHCNr1qxZ/v7+Fy5cENC2YZs3b/7hhx+WLFkyefJkY2Pj2bNnp6amWlpalpWV0R1NCe8cgR4eHtbW1gcOHNDSVxE8BDMATZLL5TExMdOmTROLxebm5h4eHio3Qg4ePMgwzOnTp7sW0gF0bGwsfYY9MjKSvetJOBndLl++vGDBAhMTkzFjxoSFhVVXV/f/5KR/Oep65ejRo8bGxirPt1CxsbF6enpfffWVmrf32MN8Ut+RAUtuZ2JiYmNjo6+vb2pqSkt45ggUiUS+vr45OTm92moVWAhmABrT1tbm5eUVHR1dXFzc2tra0NBw5cqVxYsXa/DP7Rs3bnh7excUFMjl8urq6oMHD0qlUu5s1RDX0dGRl5f31ltv0aywKqZNmxYcHJyTk5OVldXt2/n3cGFhobu7+7Vr15qbm1+8eHH06FE/Pz/2qFKpDAkJCQ0NLSgokMlkLS0td+/eXbNmzapVqz913pYAAAXGSURBVPr5BX/66ad79+4tXbpULBbTEpoOcMaMGdxqs2bNYg+x5s+fTwi5cuVKP9ugmxDMADRm//79169ft7GxOX/+fH19/ZMnT6KjoxmG+eKLL549e8bzJBEREXl5eYQQemeFYo8mJyeHhIQ8fPiQriOYOXNmcXHxrl27+n/yfuao4+nBgwcymczJyelVFWJiYgwNDV9154x/D6tPfTdAye2ampqCgoLMzc13797NFvLMEUgIoZ0voFnWIQXBDEBj6N37pKSkJUuWjBgxYuLEiVFRUTQ1wblz5zTyEXPnzj106JC9vT3dnCU1NdXAwEBlXnEoq6qqIoRYWVm9qoKtrW1YWFhhYWFKSkrXo/x7eNGiRQkJCfb29sbGxi4uLjQpBBvMDh8+LBKJMjMzFy9ePHLkSFNTU6lUevz4cUII9z5WrzQ1Nfn6+hYXF6emptra2rLl3U5ddltIkyLRLoLeQjAD0JjS0tJRo0bRySLWkiVLSJcJpT5btGgRe5eLEGJnZzdlypRHjx5p5OSDoK6ujhAikUjU1ImIiDAxMYmMjOy6Tp1/D6tPfccmt9PX1xeJRHp6enp6ejQzRt9uWdXW1np5ed28eTMtLU0lvwT/HIEjRowgL7sIegvBDECTuJGmWzRXgMplWi6Xa+TTB/TkGmFhYUE4KYe6NW7cuLVr1z548IDmHlLRYw9T6lPfscntFApFZ2cnd7q1D9kBq6qq3Nzc7t27l56e7ubmpnKUzRHILew2RyCddbS0tOxtA4AgmAFokL29fU1NDU2fxkpLSyOcy9aYMWMIIeXl5dw6Kvf8aUzq6Ojo+hGXLl3izlCVlZWVlJS8/vrrGjn5ILC2tiaE1NTUqK+2efNmc3Pz6Ojo9vZ2bjmfHubDwcHBxMSkrq6u60OZvZ2zffTokYuLS0VFRWZmJptKiYsO1BITE7mF3eYIpN1Cuwh6C8EMQGPoTq+BgYFpaWkNDQ2VlZXbtm377rvvjIyMli5dSus4OjoSQuLj43NycuRy+ePHjzds2KDyVDtdHZCXl8c+q8S6detWaGhoaWlpU1NTfn6+n59fe3s7m1+4nycfBI6OjmZmZj0+AGBpablx48by8vIjR45wy/n0MB+aSm5XVFTk6upaW1ublZXFTc7OxT9HIO0WlUlU4Ksf6WOETcdz/0BvER75zFpbW7u9oiUkJHCr+fv7c4/q6+uHhIQQQs6cOUMrdHR0jB8/XuV3SpO3BQQEqGSbc3BwaGxs7OfJlTxy1HH1J5+Zj4+PWCxuaWlhS2j2O19fX261xsZGOtAkhJw4cYIW8ulhPqnveCa367FPQkNDuz2Jyrv45AhUKpWff/45IeTx48c8e1KFjl/TMDID0BhDQ8Ps7OyoqKipU6caGhpKJJKFCxemp6fTixTr+++///TTT0eNGiUWi52dnbOzs1Wmm0Qi0enTp11dXdkHb1kuLi7p6elz586lOdtCQ0Nzc3PNzMw0cvLBERIS0tLSQucG1TAzM+v69DTPHu4Rz+R29Naavr5+r07eFZ8cgQqF4uzZs1KpdNKkSf38OB2l7WiqNTr+Vwz0FtF2puluBxza0p+RWWtr6+jRo/38/DTbJI1TKBQWFhbTp09nV4gMKLqPCTsG7QMdv6ZhZAYAg4o+E33u3Dk6uzhkFRUV1dXVbdmyhef6yX6Ki4ubMWMGvSkIfYBgBgCDLTw83M7ObuvWrdpuiDr5+fmTJk364IMPBuGzrl+/npWVFRcXp3JrDfjr71wwAEBvGRgYlJSUaLsVPQgPDw8PDx+cz3JxcVH2e49jHYdgBiAMPj4+uN4BvAqmGQEAQPAQzAAAQPAQzAAAQPAQzAAAQPB0egHIzZs38VQH8Ld3716aTwsqKyvJy50SYYi4efOmLu/ryOjs+qg9e/YgoysADCfOzs7r16/Xdiu0Q3eDGQAADBu4ZwYAAIKHYAYAAIKHYAYAAIKHYAYAAIL3/wEw9Yyr2rJX2QAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 113,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "shallow_lstm.plot_model()\n",
    "Image(filename='recurrent.png')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-29T05:56:34.076214Z",
     "iopub.status.busy": "2021-05-29T05:56:34.075892Z",
     "iopub.status.idle": "2021-05-29T05:56:34.221073Z",
     "shell.execute_reply": "2021-05-29T05:56:34.220129Z",
     "shell.execute_reply.started": "2021-05-29T05:56:34.076185Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 600x400 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "shallow_lstm.plot_training()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-05-29T06:00:42.890299Z",
     "iopub.status.busy": "2021-05-29T06:00:42.889970Z",
     "iopub.status.idle": "2021-05-29T06:08:45.736465Z",
     "shell.execute_reply": "2021-05-29T06:08:45.735449Z",
     "shell.execute_reply.started": "2021-05-29T06:00:42.890263Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9231 - val_loss: 2.9078\n",
      "\n",
      "Epoch 00001: val_loss improved from 2.90842 to 2.90777, saving model to shallow_lstm.hdf5\n",
      "Epoch 2/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9230 - val_loss: 2.9078\n",
      "\n",
      "Epoch 00002: val_loss improved from 2.90777 to 2.90777, saving model to shallow_lstm.hdf5\n",
      "Epoch 3/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9230 - val_loss: 2.9078\n",
      "\n",
      "Epoch 00003: val_loss did not improve from 2.90777\n",
      "Epoch 4/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9230 - val_loss: 2.9078\n",
      "\n",
      "Epoch 00004: val_loss did not improve from 2.90777\n",
      "Epoch 5/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9230 - val_loss: 2.9078\n",
      "\n",
      "Epoch 00005: val_loss did not improve from 2.90777\n",
      "Epoch 6/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9230 - val_loss: 2.9078\n",
      "\n",
      "Epoch 00006: val_loss did not improve from 2.90777\n",
      "Epoch 7/100\n",
      "49/49 [==============================] - 69s 1s/step - loss: 2.9230 - val_loss: 2.9078\n",
      "\n",
      "Epoch 00007: val_loss did not improve from 2.90777\n",
      "Epoch 00007: early stopping\n"
     ]
    }
   ],
   "source": [
    "# Training it for another 100 iterations\n",
    "shallow_lstm.fit(training_data=(X_train, y_train), validation_data=(X_val, y_val), verbose=1, callbacks=shallow_seq_callbacks)"
   ]
  },
  {
   "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}