19_recurrent_neural_nets/06_sentiment_analysis_pretrained_embeddings.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sentiment analysis with pretrained word vectors"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In Chapter 15, Word Embeddings, we discussed how to learn domain-specific word embeddings. Word2vec, and related learning algorithms, produce high-quality word vectors, but require large datasets. Hence, it is common that research groups share word vectors trained on large datasets, similar to the weights for pretrained deep learning models that we encountered in the section on transfer learning in the previous chapter.\n",
    "\n",
    "We are now going to illustrate how to use pretrained Global Vectors for Word Representation (GloVe) provided by the Stanford NLP group with the IMDB review dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:14:54.294593Z",
     "start_time": "2021-02-23T17:14:52.919376Z"
    }
   },
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "\n",
    "from pathlib import Path\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "from sklearn.metrics import roc_auc_score\n",
    "\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras.callbacks import ModelCheckpoint, EarlyStopping\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Dense, GRU, Embedding\n",
    "from tensorflow.keras.preprocessing.sequence import pad_sequences\n",
    "from tensorflow.keras.preprocessing.text import Tokenizer\n",
    "import tensorflow.keras.backend as K\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:14:54.335324Z",
     "start_time": "2021-02-23T17:14:54.295781Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using CPU\n"
     ]
    }
   ],
   "source": [
    "gpu_devices = tf.config.experimental.list_physical_devices('GPU')\n",
    "if gpu_devices:\n",
    "    print('Using GPU')\n",
    "    tf.config.experimental.set_memory_growth(gpu_devices[0], True)\n",
    "else:\n",
    "    print('Using CPU')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:14:54.343605Z",
     "start_time": "2021-02-23T17:14:54.336878Z"
    }
   },
   "outputs": [],
   "source": [
    "sns.set_style('whitegrid')\n",
    "np.random.seed(42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:14:54.356444Z",
     "start_time": "2021-02-23T17:14:54.344420Z"
    }
   },
   "outputs": [],
   "source": [
    "results_path = Path('results', 'sentiment_imdb')\n",
    "if not results_path.exists():\n",
    "    results_path.mkdir(parents=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load Reviews"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We are going to load the IMDB dataset from the source for manual preprocessing."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Data source: [Stanford IMDB Reviews Dataset](http://ai.stanford.edu/~amaas/data/sentiment/)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Dowload extract, and place the content in a newly created `data` folder so that your directory structure looks as follows:\n",
    "```\n",
    "19_recurrent_neural_nets\n",
    " |-data\n",
    "     |-aclimdb\n",
    "          |-train\n",
    "              |-neg\n",
    "              |-pos\n",
    "              ...\n",
    "          |-test\n",
    "          |-imdb.vocab\n",
    "```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:14:54.364726Z",
     "start_time": "2021-02-23T17:14:54.357689Z"
    }
   },
   "outputs": [],
   "source": [
    "path = Path('data', 'aclImdb')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:14:54.629329Z",
     "start_time": "2021-02-23T17:14:54.366279Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "50003"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "files = path.glob('**/*.txt')\n",
    "len(list(files))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:14:56.111490Z",
     "start_time": "2021-02-23T17:14:54.631782Z"
    }
   },
   "outputs": [],
   "source": [
    "files = path.glob('*/**/*.txt')\n",
    "outcomes = set()\n",
    "data = []\n",
    "for f in files:\n",
    "    if f.stem.startswith(('urls_', 'imdbEr')):\n",
    "        continue\n",
    "    _, _, data_set, outcome = f.parent.as_posix().split('/')\n",
    "    if outcome == 'unsup':\n",
    "        continue\n",
    "    data.append([data_set, int(outcome == 'pos'),\n",
    "                 f.read_text(encoding='latin1')])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:14:56.139785Z",
     "start_time": "2021-02-23T17:14:56.112577Z"
    }
   },
   "outputs": [],
   "source": [
    "data = pd.DataFrame(data, columns=['dataset', 'label', 'review'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:14:56.178373Z",
     "start_time": "2021-02-23T17:14:56.140713Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 50000 entries, 0 to 49999\n",
      "Data columns (total 3 columns):\n",
      " #   Column   Non-Null Count  Dtype \n",
      "---  ------   --------------  ----- \n",
      " 0   dataset  50000 non-null  object\n",
      " 1   label    50000 non-null  int64 \n",
      " 2   review   50000 non-null  object\n",
      "dtypes: int64(1), object(2)\n",
      "memory usage: 1.1+ MB\n"
     ]
    }
   ],
   "source": [
    "data.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:14:56.191606Z",
     "start_time": "2021-02-23T17:14:56.179291Z"
    }
   },
   "outputs": [],
   "source": [
    "train_data = data.loc[data.dataset=='train', ['label', 'review']]\n",
    "test_data = data.loc[data.dataset=='test', ['label', 'review']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:14:56.196955Z",
     "start_time": "2021-02-23T17:14:56.192500Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    12500\n",
       "1    12500\n",
       "Name: label, dtype: int64"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data.label.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:14:56.206787Z",
     "start_time": "2021-02-23T17:14:56.198458Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    12500\n",
       "1    12500\n",
       "Name: label, dtype: int64"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_data.label.value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Prepare Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Tokenizer"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Keras provides a tokenizer that we use to convert the text documents to integer-encoded sequences, as shown here:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:14:58.986540Z",
     "start_time": "2021-02-23T17:14:56.208686Z"
    }
   },
   "outputs": [],
   "source": [
    "num_words = 10000\n",
    "t = Tokenizer(num_words=num_words, \n",
    "              lower=True, \n",
    "              oov_token=2)\n",
    "t.fit_on_texts(train_data.review)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:14:58.989919Z",
     "start_time": "2021-02-23T17:14:58.987445Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "88586"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vocab_size = len(t.word_index) + 1\n",
    "vocab_size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:15:02.913481Z",
     "start_time": "2021-02-23T17:14:58.991167Z"
    }
   },
   "outputs": [],
   "source": [
    "train_data_encoded = t.texts_to_sequences(train_data.review)\n",
    "test_data_encoded = t.texts_to_sequences(test_data.review)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:15:02.916314Z",
     "start_time": "2021-02-23T17:15:02.914473Z"
    }
   },
   "outputs": [],
   "source": [
    "max_length = 100"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Pad Sequences"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We also use the pad_sequences function to convert the list of lists (of unequal length) to stacked sets of padded and truncated arrays for both the train and test datasets:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:15:03.128188Z",
     "start_time": "2021-02-23T17:15:02.917344Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(25000, 100)"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_padded = pad_sequences(train_data_encoded, \n",
    "                            maxlen=max_length, \n",
    "                            padding='post',\n",
    "                           truncating='post')\n",
    "y_train = train_data['label']\n",
    "X_train_padded.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:15:03.343442Z",
     "start_time": "2021-02-23T17:15:03.130324Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(25000, 100)"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test_padded = pad_sequences(test_data_encoded, \n",
    "                            maxlen=max_length, \n",
    "                            padding='post',\n",
    "                           truncating='post')\n",
    "y_test = test_data['label']\n",
    "X_test_padded.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load Embeddings"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Assuming we have downloaded and unzipped the GloVe data to the location indicated in the code, we now create a dictionary that maps GloVe tokens to 100-dimensional real-valued vectors, as follows:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:15:09.664986Z",
     "start_time": "2021-02-23T17:15:03.344584Z"
    }
   },
   "outputs": [],
   "source": [
    "# load the whole embedding into memory\n",
    "glove_path = Path('..', 'data', 'glove', 'glove.6B.100d.txt')\n",
    "embeddings_index = dict()\n",
    "\n",
    "for line in glove_path.open(encoding='latin1'):\n",
    "    values = line.split()\n",
    "    word = values[0]\n",
    "    try:\n",
    "        coefs = np.asarray(values[1:], dtype='float32')\n",
    "    except:\n",
    "        continue\n",
    "    embeddings_index[word] = coefs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:15:09.668022Z",
     "start_time": "2021-02-23T17:15:09.665862Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loaded 399,883 word vectors.\n"
     ]
    }
   ],
   "source": [
    "print('Loaded {:,d} word vectors.'.format(len(embeddings_index)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There are around 340,000 word vectors that we use to create an embedding matrix that matches the vocabulary so that the RNN model can access embeddings by the token index:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:15:09.770408Z",
     "start_time": "2021-02-23T17:15:09.669142Z"
    }
   },
   "outputs": [],
   "source": [
    "embedding_matrix = np.zeros((vocab_size, 100))\n",
    "for word, i in t.word_index.items():\n",
    "    embedding_vector = embeddings_index.get(word)\n",
    "    if embedding_vector is not None:\n",
    "        embedding_matrix[i] = embedding_vector"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:15:09.774652Z",
     "start_time": "2021-02-23T17:15:09.772236Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(88586, 100)"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "embedding_matrix.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Define Model Architecture"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The difference between this and the RNN setup in the previous example is that we are going to pass the embedding matrix to the embedding layer and set it to non-trainable, so that the weights remain fixed during training:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:15:09.783915Z",
     "start_time": "2021-02-23T17:15:09.775763Z"
    }
   },
   "outputs": [],
   "source": [
    "embedding_size = 100"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:15:10.504198Z",
     "start_time": "2021-02-23T17:15:09.784990Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:Layer gru will not use cuDNN kernel since it doesn't meet the cuDNN kernel criteria. It will use generic GPU kernel as fallback when running on GPU\n",
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding (Embedding)        (None, 100, 100)          8858600   \n",
      "_________________________________________________________________\n",
      "gru (GRU)                    (None, 32)                12864     \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 8,871,497\n",
      "Trainable params: 12,897\n",
      "Non-trainable params: 8,858,600\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "rnn = Sequential([\n",
    "    Embedding(input_dim=vocab_size, \n",
    "              output_dim= embedding_size, \n",
    "              input_length=max_length,\n",
    "              weights=[embedding_matrix], \n",
    "              trainable=False),\n",
    "    GRU(units=32,  dropout=0.2, recurrent_dropout=0.2),\n",
    "    Dense(1, activation='sigmoid')\n",
    "])\n",
    "rnn.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:15:10.518063Z",
     "start_time": "2021-02-23T17:15:10.505197Z"
    }
   },
   "outputs": [],
   "source": [
    "rnn.compile(loss='binary_crossentropy',\n",
    "            optimizer='RMSProp',\n",
    "            metrics=['accuracy', \n",
    "                     tf.keras.metrics.AUC(name='AUC')])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:15:10.524265Z",
     "start_time": "2021-02-23T17:15:10.519172Z"
    }
   },
   "outputs": [],
   "source": [
    "rnn_path = (results_path / 'lstm.pretrained.h5').as_posix()\n",
    "\n",
    "checkpointer = ModelCheckpoint(filepath=rnn_path,\n",
    "                               verbose=1,\n",
    "                               monitor='val_AUC',\n",
    "                               mode='max',\n",
    "                               save_best_only=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T17:15:10.532489Z",
     "start_time": "2021-02-23T17:15:10.525759Z"
    }
   },
   "outputs": [],
   "source": [
    "early_stopping = EarlyStopping(monitor='val_AUC',\n",
    "                               patience=5,\n",
    "                               mode='max',\n",
    "                               restore_best_weights=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T18:05:39.642268Z",
     "start_time": "2021-02-23T17:15:10.534115Z"
    },
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.6505 - accuracy: 0.6087 - AUC: 0.6565\n",
      "Epoch 00001: val_AUC improved from -inf to 0.80524, saving model to results/sentiment_imdb/lstm.pretrained.h5\n",
      "782/782 [==============================] - 111s 141ms/step - loss: 0.6505 - accuracy: 0.6087 - AUC: 0.6565 - val_loss: 0.7026 - val_accuracy: 0.6586 - val_AUC: 0.8052\n",
      "Epoch 2/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.5053 - accuracy: 0.7570 - AUC: 0.8316\n",
      "Epoch 00002: val_AUC improved from 0.80524 to 0.86949, saving model to results/sentiment_imdb/lstm.pretrained.h5\n",
      "782/782 [==============================] - 111s 143ms/step - loss: 0.5053 - accuracy: 0.7570 - AUC: 0.8316 - val_loss: 0.4681 - val_accuracy: 0.7776 - val_AUC: 0.8695\n",
      "Epoch 3/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.4538 - accuracy: 0.7872 - AUC: 0.8679\n",
      "Epoch 00003: val_AUC improved from 0.86949 to 0.88737, saving model to results/sentiment_imdb/lstm.pretrained.h5\n",
      "782/782 [==============================] - 117s 149ms/step - loss: 0.4538 - accuracy: 0.7872 - AUC: 0.8679 - val_loss: 0.4352 - val_accuracy: 0.7931 - val_AUC: 0.8874\n",
      "Epoch 4/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.4303 - accuracy: 0.8000 - AUC: 0.8823\n",
      "Epoch 00004: val_AUC improved from 0.88737 to 0.89225, saving model to results/sentiment_imdb/lstm.pretrained.h5\n",
      "782/782 [==============================] - 109s 139ms/step - loss: 0.4303 - accuracy: 0.8000 - AUC: 0.8823 - val_loss: 0.4404 - val_accuracy: 0.7947 - val_AUC: 0.8922\n",
      "Epoch 5/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.4138 - accuracy: 0.8062 - AUC: 0.8915\n",
      "Epoch 00005: val_AUC improved from 0.89225 to 0.89876, saving model to results/sentiment_imdb/lstm.pretrained.h5\n",
      "782/782 [==============================] - 109s 139ms/step - loss: 0.4138 - accuracy: 0.8062 - AUC: 0.8915 - val_loss: 0.4284 - val_accuracy: 0.7994 - val_AUC: 0.8988\n",
      "Epoch 6/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.4064 - accuracy: 0.8123 - AUC: 0.8959\n",
      "Epoch 00006: val_AUC improved from 0.89876 to 0.90227, saving model to results/sentiment_imdb/lstm.pretrained.h5\n",
      "782/782 [==============================] - 109s 139ms/step - loss: 0.4064 - accuracy: 0.8123 - AUC: 0.8959 - val_loss: 0.4015 - val_accuracy: 0.8176 - val_AUC: 0.9023\n",
      "Epoch 7/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3959 - accuracy: 0.8201 - AUC: 0.9016\n",
      "Epoch 00007: val_AUC improved from 0.90227 to 0.90520, saving model to results/sentiment_imdb/lstm.pretrained.h5\n",
      "782/782 [==============================] - 108s 138ms/step - loss: 0.3959 - accuracy: 0.8201 - AUC: 0.9016 - val_loss: 0.4228 - val_accuracy: 0.8072 - val_AUC: 0.9052\n",
      "Epoch 8/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3884 - accuracy: 0.8230 - AUC: 0.9054\n",
      "Epoch 00008: val_AUC improved from 0.90520 to 0.90554, saving model to results/sentiment_imdb/lstm.pretrained.h5\n",
      "782/782 [==============================] - 118s 151ms/step - loss: 0.3884 - accuracy: 0.8230 - AUC: 0.9054 - val_loss: 0.4152 - val_accuracy: 0.8064 - val_AUC: 0.9055\n",
      "Epoch 9/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3814 - accuracy: 0.8264 - AUC: 0.9090\n",
      "Epoch 00009: val_AUC improved from 0.90554 to 0.90921, saving model to results/sentiment_imdb/lstm.pretrained.h5\n",
      "782/782 [==============================] - 107s 137ms/step - loss: 0.3814 - accuracy: 0.8264 - AUC: 0.9090 - val_loss: 0.3939 - val_accuracy: 0.8167 - val_AUC: 0.9092\n",
      "Epoch 10/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3766 - accuracy: 0.8275 - AUC: 0.9112\n",
      "Epoch 00010: val_AUC improved from 0.90921 to 0.90976, saving model to results/sentiment_imdb/lstm.pretrained.h5\n",
      "782/782 [==============================] - 108s 138ms/step - loss: 0.3766 - accuracy: 0.8275 - AUC: 0.9112 - val_loss: 0.3896 - val_accuracy: 0.8194 - val_AUC: 0.9098\n",
      "Epoch 11/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3698 - accuracy: 0.8348 - AUC: 0.9147\n",
      "Epoch 00011: val_AUC improved from 0.90976 to 0.91103, saving model to results/sentiment_imdb/lstm.pretrained.h5\n",
      "782/782 [==============================] - 142s 182ms/step - loss: 0.3698 - accuracy: 0.8348 - AUC: 0.9147 - val_loss: 0.3808 - val_accuracy: 0.8267 - val_AUC: 0.9110\n",
      "Epoch 12/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3644 - accuracy: 0.8345 - AUC: 0.9174\n",
      "Epoch 00012: val_AUC improved from 0.91103 to 0.91111, saving model to results/sentiment_imdb/lstm.pretrained.h5\n",
      "782/782 [==============================] - 106s 136ms/step - loss: 0.3644 - accuracy: 0.8345 - AUC: 0.9174 - val_loss: 0.3912 - val_accuracy: 0.8201 - val_AUC: 0.9111\n",
      "Epoch 13/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3600 - accuracy: 0.8369 - AUC: 0.9195\n",
      "Epoch 00013: val_AUC improved from 0.91111 to 0.91143, saving model to results/sentiment_imdb/lstm.pretrained.h5\n",
      "782/782 [==============================] - 111s 142ms/step - loss: 0.3600 - accuracy: 0.8369 - AUC: 0.9195 - val_loss: 0.3763 - val_accuracy: 0.8279 - val_AUC: 0.9114\n",
      "Epoch 14/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3553 - accuracy: 0.8398 - AUC: 0.9217\n",
      "Epoch 00014: val_AUC improved from 0.91143 to 0.91288, saving model to results/sentiment_imdb/lstm.pretrained.h5\n",
      "782/782 [==============================] - 108s 138ms/step - loss: 0.3553 - accuracy: 0.8398 - AUC: 0.9217 - val_loss: 0.3769 - val_accuracy: 0.8277 - val_AUC: 0.9129\n",
      "Epoch 15/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3505 - accuracy: 0.8433 - AUC: 0.9239\n",
      "Epoch 00015: val_AUC improved from 0.91288 to 0.91290, saving model to results/sentiment_imdb/lstm.pretrained.h5\n",
      "782/782 [==============================] - 110s 141ms/step - loss: 0.3505 - accuracy: 0.8433 - AUC: 0.9239 - val_loss: 0.3991 - val_accuracy: 0.8195 - val_AUC: 0.9129\n",
      "Epoch 16/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3488 - accuracy: 0.8429 - AUC: 0.9247\n",
      "Epoch 00016: val_AUC did not improve from 0.91290\n",
      "782/782 [==============================] - 110s 140ms/step - loss: 0.3488 - accuracy: 0.8429 - AUC: 0.9247 - val_loss: 0.4081 - val_accuracy: 0.8132 - val_AUC: 0.9118\n",
      "Epoch 17/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3450 - accuracy: 0.8452 - AUC: 0.9264\n",
      "Epoch 00017: val_AUC improved from 0.91290 to 0.91368, saving model to results/sentiment_imdb/lstm.pretrained.h5\n",
      "782/782 [==============================] - 110s 140ms/step - loss: 0.3450 - accuracy: 0.8452 - AUC: 0.9264 - val_loss: 0.3795 - val_accuracy: 0.8306 - val_AUC: 0.9137\n",
      "Epoch 18/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3430 - accuracy: 0.8460 - AUC: 0.9272\n",
      "Epoch 00018: val_AUC did not improve from 0.91368\n",
      "782/782 [==============================] - 108s 138ms/step - loss: 0.3430 - accuracy: 0.8460 - AUC: 0.9272 - val_loss: 0.3891 - val_accuracy: 0.8221 - val_AUC: 0.9127\n",
      "Epoch 19/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3376 - accuracy: 0.8477 - AUC: 0.9296\n",
      "Epoch 00019: val_AUC did not improve from 0.91368\n",
      "782/782 [==============================] - 106s 135ms/step - loss: 0.3376 - accuracy: 0.8477 - AUC: 0.9296 - val_loss: 0.3822 - val_accuracy: 0.8267 - val_AUC: 0.9134\n",
      "Epoch 20/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3368 - accuracy: 0.8516 - AUC: 0.9300\n",
      "Epoch 00020: val_AUC improved from 0.91368 to 0.91385, saving model to results/sentiment_imdb/lstm.pretrained.h5\n",
      "782/782 [==============================] - 105s 135ms/step - loss: 0.3368 - accuracy: 0.8516 - AUC: 0.9300 - val_loss: 0.3994 - val_accuracy: 0.8138 - val_AUC: 0.9139\n",
      "Epoch 21/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3340 - accuracy: 0.8532 - AUC: 0.9312\n",
      "Epoch 00021: val_AUC did not improve from 0.91385\n",
      "782/782 [==============================] - 105s 135ms/step - loss: 0.3340 - accuracy: 0.8532 - AUC: 0.9312 - val_loss: 0.3741 - val_accuracy: 0.8297 - val_AUC: 0.9131\n",
      "Epoch 22/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3310 - accuracy: 0.8509 - AUC: 0.9324\n",
      "Epoch 00022: val_AUC improved from 0.91385 to 0.91487, saving model to results/sentiment_imdb/lstm.pretrained.h5\n",
      "782/782 [==============================] - 107s 137ms/step - loss: 0.3310 - accuracy: 0.8509 - AUC: 0.9324 - val_loss: 0.3795 - val_accuracy: 0.8306 - val_AUC: 0.9149\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 23/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3276 - accuracy: 0.8534 - AUC: 0.9339\n",
      "Epoch 00023: val_AUC did not improve from 0.91487\n",
      "782/782 [==============================] - 107s 137ms/step - loss: 0.3276 - accuracy: 0.8534 - AUC: 0.9339 - val_loss: 0.3904 - val_accuracy: 0.8244 - val_AUC: 0.9135\n",
      "Epoch 24/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3266 - accuracy: 0.8554 - AUC: 0.9343\n",
      "Epoch 00024: val_AUC did not improve from 0.91487\n",
      "782/782 [==============================] - 108s 138ms/step - loss: 0.3266 - accuracy: 0.8554 - AUC: 0.9343 - val_loss: 0.3725 - val_accuracy: 0.8280 - val_AUC: 0.9141\n",
      "Epoch 25/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3221 - accuracy: 0.8580 - AUC: 0.9361\n",
      "Epoch 00025: val_AUC did not improve from 0.91487\n",
      "782/782 [==============================] - 123s 158ms/step - loss: 0.3221 - accuracy: 0.8580 - AUC: 0.9361 - val_loss: 0.3825 - val_accuracy: 0.8276 - val_AUC: 0.9142\n",
      "Epoch 26/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3206 - accuracy: 0.8590 - AUC: 0.9367\n",
      "Epoch 00026: val_AUC did not improve from 0.91487\n",
      "782/782 [==============================] - 127s 162ms/step - loss: 0.3206 - accuracy: 0.8590 - AUC: 0.9367 - val_loss: 0.3769 - val_accuracy: 0.8284 - val_AUC: 0.9144\n",
      "Epoch 27/100\n",
      "782/782 [==============================] - ETA: 0s - loss: 0.3195 - accuracy: 0.8608 - AUC: 0.9372\n",
      "Epoch 00027: val_AUC did not improve from 0.91487\n",
      "782/782 [==============================] - 125s 160ms/step - loss: 0.3195 - accuracy: 0.8608 - AUC: 0.9372 - val_loss: 0.3782 - val_accuracy: 0.8279 - val_AUC: 0.9142\n"
     ]
    }
   ],
   "source": [
    "training = rnn.fit(X_train_padded,\n",
    "                   y_train,\n",
    "                   batch_size=32,\n",
    "                   epochs=100,\n",
    "                   validation_data=(X_test_padded,\n",
    "                                    y_test),\n",
    "                   callbacks=[early_stopping,\n",
    "                              checkpointer],\n",
    "                   verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T18:05:51.382125Z",
     "start_time": "2021-02-23T18:05:39.643359Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.914964528"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_score = rnn.predict(X_test_padded)\n",
    "roc_auc_score(y_score=y_score.squeeze(), y_true=y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-23T18:05:52.075335Z",
     "start_time": "2021-02-23T18:05:51.383093Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA+4AAAEdCAYAAACSfT0gAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAACcXklEQVR4nOzdd3hTZfvA8W+6996DMgtlChQQGZWyRwEFREQUBKuAijhBfwqiuHh9UVFfRARBEWVZRhEVEBFEthalKKt0QSfdbdom5/fHKYHSFlpom477c125kpx5n9PAkzvP0iiKoiCEEEIIIYQQQog6ycTYAQghhBBCCCGEEKJikrgLIYQQQgghhBB1mCTuQgghhBBCCCFEHSaJuxBCCCGEEEIIUYdJ4i6EEEIIIYQQQtRhkrgLIYQQQgghhBB1mCTuQgghbmjSpEksWLDA2GHUWUuWLGHEiBHVdrzQ0FA+//zzajtefbZp0yY6d+5cK+fq3LkzmzZtqpVzCSGEEFUlibsQQtRx6enpzJ8/n9DQUNq3b89dd93Fww8/zP79+6v1PBUlSUuWLOGZZ56p1nPdqqr+iHDy5EmCgoK4//77azCq6rVhwwYeeOCBGj3HnDlzaN26Na1bt6Zdu3b079+fd955h7y8vNs6bnUn2sOGDWPnzp3Vdrxb9frrrzNo0CAURSmzLjMzkw4dOrBu3brbPo/8aCOEEKIiZsYOQAghxI09+eST5Ofns3DhQpo0aUJaWhqHDx8mIyOjVs7v5ORUK+epCevWreOBBx4gIiKCs2fP0qJFC2OHdFMuLi61cp677rqLd999l+LiYo4cOcL//d//kZeXx2uvvVZm2+LiYkxNTdFoNNVy7sLCQiwsLG66nZWVFVZWVtVyztsxbtw4vvrqKw4fPkz37t1Lrdu6dStmZmYMGzbMSNGVVdn7K4QQov6QGnchhKjDsrKyOHLkCM899xw9e/bE19eXjh07MnXqVIYPH27YrrCwkEWLFtG3b1/uuOMOxowZw6+//mpYf/DgQVq3bs2BAwcYN24cnTp14t577+Xvv/82rJ87dy55eXmGmtglS5YAZWu5Q0ND+eijj5gzZw6dO3cmJCSE7du3k5WVxezZs+ncuTODBg1i3759pa7lzJkzhIeH07lzZ3r27MkzzzxDSkqKYf2cOXN47LHHWLVqFX369KFbt27MnTuX/Px8w/pDhw6xZs0aQ4zx8fEV3ruCggK2bdvGuHHjGDx4MBs2bCi1Pj4+ntatW/PDDz8wZcoUOnXqxLBhw0q1ZNDpdLz00kuEhobSsWNHBg0axGeffYZery/3nIcPH6Zdu3alrgtg8eLFhIWFAZCdnc3zzz9Pz5496dChA/379+eLL74odX+vrXX95ptvGDx4MB06dODOO+9k6tSpFBcXV3jdlWVhYYG7uzve3t6EhYURFhbGrl27gKvN/zdt2sSAAQPo0KEDeXl5ZGdn88orr9CzZ086d+7Mgw8+yIkTJ4Abf4ZCQ0NZsmQJc+fOJTg4mOeeew6A//znPwwePJiOHTsSGhrKu+++i1arNcR4fQ3+lbgiIyMZMGAAnTt3ZsaMGaSnp5e6to0bNzJs2DA6dOjA4MGD+eKLL0r9zS5cuMCkSZMM63/++ecb3qs2bdrQvn17Nm7cWGbdhg0bGDJkCHZ2diQlJTF79my6detGt27dCA8PJyYmptT2e/bsYdy4cXTs2JEePXrw+OOPo9VqmTRpEgkJCbz77ruG+3fFjz/+SFhYGO3btyckJIT//e9/pWr/K7q/H330Ef369aN9+/b06tWLF1544YbXKYQQou6SxF0IIeowGxsbbGxs2L17d6mE5npz587l8OHDvPfee2zdupV77rmH6dOnc+rUqVLbvffeezz77LNs2rQJZ2dnnnvuORRFoXPnzrz00ktYW1uzb98+9u3bxyOPPFLh+VavXk2HDh347rvvGDp0KC+++CLPPvssISEhREREEBwczPPPP2+IOTk5mYkTJ9KqVSs2bNjAypUrycvLY/r06aUSqiNHjnD69Gm++OILFi9ezE8//cTq1asBePnll+ncuTP33nuvIUZvb+8KY9yxYwc+Pj60adOGUaNGERERQVFRUZntFi9ezKRJk9i8eTMdOnTgmWeeITc3FwC9Xo+npyfvv/8+27dv5+mnn+bTTz8tN4ED6NatG/7+/kRERBiW6fV6IiIiGDt2LADvv/8+//77L59++inff/89b775Jp6enuUe78SJEyxYsICZM2eyY8cOvvjiC/r06VPhNd8OKyurUvcnPj6ebdu28cEHH7B582YsLCwIDw8nKSmJTz/91PB3fvjhh0lOTr7pZ2jlypU0b96cjRs3GrpeWFtb8+abb7J9+3bmzZvH9u3b+d///nfDOBMSEti+fTsfffQRK1asIDo6mvfff9+wft26dSxevJinnnqK7du38+KLL/LZZ5/x9ddfA+rf44knnkCv1/Ptt9/y5ptvsmTJEgoLC2943rFjx/LDDz+Qk5NjWPb3338THR3N2LFjyc/P56GHHsLS0pIvv/ySb775Bnd3d6ZMmWL48Wnv3r3MmDGDu+66i02bNrFq1Sq6deuGXq9nyZIleHl5MXPmTMP9A/jrr7+YNWsWAwcOZOvWrTz77LMsW7aMr776qlR819/fH374gRUrVjBv3jx+/PFHli5dSseOHW94jUIIIeowRQghRJ22Y8cOpVu3bkr79u2V++67T3n77beVP/74w7D+woULSuvWrZWEhIRS+02fPl2ZN2+eoiiK8vvvvyuBgYHK3r17DeuPHDmiBAYGKhcvXlQURVE2btyo3HHHHWXO/+CDDyqvvfaa4X2/fv2U2bNnG97n5OQogYGByuuvv25YFhcXpwQGBipRUVGKoijK+++/rzz00EOljpuRkaEEBgYqf/75p6IoivLiiy8qffv2VYqKigzbvPzyy8rDDz9cYSw3MnHiRGX58uWKoiiKXq9X+vXrp+zYsaNMjGvXrjUsu3TpkhIYGKgcPny4wuMuWrSoVEwffvihMnz4cMP75cuXK0OGDDG837Nnj9KuXTslPT1dURRFeeyxx5Q5c+ZUePx+/foZ4v7hhx+ULl26KNnZ2ZW65sp68cUXlfDwcMP7P//8U+nevbsya9YsRVHUa2rbtq2SkpJi2Oa3335T7rjjDiU/P7/UsUaOHKksW7ZMUZSKP0P9+vVTHnvssZvG9fXXXysDBgwwvL/+eB9++KHSvn17JSsry7Dsk08+KbVPSEiI8t1335U67sqVK5WhQ4cqiqIov/76q9KmTZtS/14OHz6sBAYGKhs3bqwwtuzsbKVTp07KN998Y1g2f/58w3HXr1+vDBw4UNHr9Yb1xcXFSvfu3ZXIyEhFURRl/PjxytNPP13hOa7921/xzDPPKJMmTSq17MMPP1T69OlTar/r7++KFSuUQYMGKYWFhRWeTwghRP0hfdyFEKKOGzx4MHfffTdHjhzh+PHj7Nu3jxUrVjB79mwef/xx/v77bxRFKdV0HtTm83feeWepZdc2v/Xw8AAgLS0NLy+vKsV07XFsbW2xtrYmMDDQsMzNzQ3A0IT577//5siRI+UOXBYbG2uoCWzZsiVmZleLJg8PD/78888qxQZqU+hjx47x3nvvAaDRaAgLC2P9+vUMHjy4wmu5ck+ubXq9du1a1q9fT2JiIlqtlqKiInx9fSs89z333MPixYs5duwYXbp0YePGjQwYMABnZ2cAJkyYwKxZs/j777/p1asX/fr1K9Nv+oq77roLHx8f+vfvT+/evenduzcDBw7Ezs6u3O2nTZvG0aNHAfDx8SEyMrLCOH/99Vc6d+5McXExxcXF9O/fn1deecWw3tPT0/B3BPVvmJ+fT8+ePUsdR6vVEhcXV+F5rmjfvn2ZZTt27GDVqlXExsaSl5eHTqersBvCFT4+Ptjb2xvee3h4kJaWBqh/t4sXLzJv3rxSffWLi4sNTcvPnj2Lp6cnPj4+hvWdOnXCxOTGjRDt7OwYPHgwGzduZPz48Wi1WiIjI3n88ccB9f7Ex8fTpUuXUvvl5+cb7k90dDT33nvvDc9zvXPnzhESElJqWdeuXfnoo4/IyckxfBauv79Dhgxh9erVhs9Onz596N+/v/R9F0KIekoSdyGEqAcsLS3p1asXvXr14oknnuDll1/mo48+4pFHHkFRFDQaDRs2bCiV9AJlBva6dv2VgcZuliiV5/rzaDSaGx5br9cTEhLCiy++WOZYrq6uNzyuUs5I3jezfv16dDod/fr1Myy7cpyLFy+WamJ/o7i3b9/Om2++yYsvvkjnzp2xs7NjzZo1Nxzp3MXFhdDQUDZu3EizZs3YvXs3S5cuNawPCQlh9+7d7N27l99//53HHnuMIUOG8NZbb5U5lp2dHd999x2HDx/mt99+49NPP+W///0vGzZsKLd5/cKFCykoKChzXeUJDg7m9ddfx8zMDA8PD8zNzUutt7GxKfVer9fj5ubGmjVryo3zZqytrUu9/+OPP3jmmWeYOXMmffr0wcHBgd27d/POO+/c8DjXx3ntZ+TK3+21116rcHT7W/k8XTFu3DgmTpzI6dOnOXXqFHl5eYwaNcpw7jZt2rB48eIy+zk6Ot7yOa/8+76Z6++vt7c3O3bs4MCBA/z222+88847fPzxx6xbt67M31Y0PKGhoaxevRo/Pz9jhyKEqCbSx12IBmTTpk3MmTPH2GGIWtCyZUuKi4spLCwkKCgIRVFISUkhICCg1KOivtPlMTc3R6fT1Ui87dq148yZM/j4+JSJsTJJX1ViLC4uJiIigmeffZaIiAjDY/PmzbRu3brC/unlOXr0KJ06deLBBx+kXbt2BAQEEBsbe9P97rvvPr7//nu+/fZb3NzcuOuuu0qtd3FxYfTo0bz99tssXLiQ7777rsI+1mZmZvTs2ZNnn32WLVu2kJ+fz549e8rd1tPT03Bfb9QqANRE78p21yfD5WnXrh2pqamYmJiU+Rte+fGlKp+hY8eO4enpycyZM+nYsSNNmzYlMTGxUvtWxM3NDU9PT2JjY8vEGBAQAKj/dpKSkrh48aJhv6ioqEr9gBUcHEyzZs3YuHEjGzZsIDQ01HDt7dq1IzY2Fmdn5zLnvTIzQ1BQEAcOHKjw+OXdvxYtWhhaUVxx9OhRvLy8bvpvx9LSkrvvvpuXXnqJDRs2cPr0aY4dO3bT62zoJk2aRLdu3W46rkFDtWnTJoKCgujcuTOdO3emf//+hjEgbve4EyZMqNI+GRkZPP300/To0YMePXrw7LPPGsaRSE9P5/7776dHjx4EBwczfvz4Mv8WrlVYWMjcuXPp0qULvXr1YuXKlaXW63Q6Fi9eTO/evencuTOjR48mKysLgAMHDhAaGkrv3r3Zvn27YZ+srCzuueeeUmNbCGEskrgLUcMq+oIwadIk1q9fX2rZwYMH6du3r+G9oiisXr2aESNGcMcdd9C3b1+eeuop/vnnn1uK5crI2J07dzaMeHztl9dbFRoaym+//ValfQ4cOMCQIUPo1KmTYTTlipw9e5aHHnqIrl27MnDgQH766SfDusLCQp566ilCQ0Np3bo1Bw8eLPcYhYWFDBkypNT9BTV5GDt2LJ07dyYsLIwjR45U6Tpq2uXLl3nooYfYvHkzp06dIi4uju+//57ly5fTs2dP7OzsaNasGWFhYcydO5cdO3YQFxfHiRMn+Pzzz/nxxx8rfS5fX1+0Wi379+8nPT3dMKBWdXjggQfIzs5m9uzZ/Pnnn8TFxfHbb7/xyiuvVOkLka+vLydOnCA+Pp709PRyk609e/Zw+fJlxo0bR2BgYKnHsGHD2LRpU6VbGTRt2pS///6bX375hZiYGD7++GMOHz580/169eqFk5MTH330Effee2+pZtgffPABO3fuJCYmhrNnz/Ljjz/i7+9fbhPmn3/+mVWrVnHy5EkSEhLYtm0bubm5RpnW7q677qJLly7MmDGDX375hbi4OI4fP86HH35o+HdTlc9Q06ZNSUpKYsuWLcTFxfH111+zbdu2247zySefZPny5XzxxRecO3eOf//9l4iICD799FPDdTRv3pwXXniB6Ohojh8/zltvvXXTFgpXjBkzho0bN3Lw4EHGjBljWB4WFoarqyszZszg0KFDxMXFcfjwYd5++23DyPLTp09nx44dLF68mDNnzhgGYrxyn3x9fTl69ChJSUmG7hqPPPIIhw8fZsmSJZw/f54tW7awYsUKpk2bdsM4N23axPr16/nnn3+Ii4tj06ZNmJubG37AaKzi4+M5cuQIGo3GMItCbamO2SCqyx133MHx48cN/4YXLVrEyZMnaz2O999/n6ysLHbt2sXOnTtJS0szzEZha2vLm2++yYEDBzh8+DCPPvoo06dPr/A+LlmyhAsXLvDzzz+zevVqli9fzt69ew3rP/zwQ44fP863337LsWPHePfdd7G0tATgzTffZOnSpSxfvpz58+cbfkB77733ePTRR6v0A7MQNUUSdyFq0O1+QVi4cCGrV6/m5Zdf5tChQ/zwww8MGDCAX3755ZZjWrp0qaGftKurK6+//votH+tWpaen88QTTzBr1iwOHTpE+/btmT17drnbFhcXM2PGDPr168ehQ4dYsGABzz//POfPnzds06VLF959913c3d0rPOfnn39eqkk2qL/0T58+nalTp3LkyBGmTZvG9OnTyczMrJ4LrQa2trbccccdrF69mkmTJjFixAgWL15seL7irbfe4t5772XRokUMHTqUxx9/nMOHD5fqx3szXbp04f777+eZZ56hZ8+eLF++vNquw9PTk7Vr12JiYsK0adMYPnw4r732GhYWFlXqc/vII49gbm7O8OHD6dmzZ7k1tBs2bKBHjx6GPuXXGjp0KAkJCZX+oWn8+PEMHTqU5557jrFjx5KQkMCUKVNuup9Go+Hee++luLi4TJ9mCwsLFi9ezKhRo5gwYQK5ubmlmtJfy97enp07dzJlyhSGDh3KihUreOONNwgODq5U/NVJo9GwbNkyevTowSuvvMLQoUN5+umnOX/+vGFsgKp8hkJDQ5k6dSpvvvkmI0eO5LfffuOpp5667TjHjRvHm2++yebNmxk1ahQTJ07k22+/NTQZNjEx4aOPPkJRFMaNG8eLL77I9OnTK/05vOeee8jPz8fT07PUCP/W1tasWbMGf39/Zs2aZZhtITMzEwcHB0DtJvHRRx/x66+/Mnr0aB588EF+//13ww87Tz31FBcvXmTAgAGGsQTatWvHBx98YJgS7r333iM8PJwHH3zwhnE6ODiwYcMGJk6cSFhYGD/88ANLlizB39+/yve0IYmIiKBTp07cc889pWZ/ALUbzRNPPMGdd95Jjx49Sk2DuW7dOoYOHUrnzp0ZNmyYYSrN1q1bc+HCBcN2c+bMMfzffOXH+GXLltGrVy/mzp1LZmYmjz32GHfeeSfdunXjscce49KlS4b9MzIymDt3Lr1796Zbt27MmDEDgBEjRrB7927DdkVFRfTo0YPo6Ojbvift2rWjRYsWnD171rDsjz/+4P777yc4OJiRI0eW+lF806ZN9O/fn86dOxMaGsqWLVs4e/Ys8+bN448//qBz586V/j8qPj6e/v37Y2dnh729PQMHDuTMmTOA2mKkefPmmJiYoCgKJiYmZGZmVlhGR0REMGPGDBwdHWnRogXjxo3ju+++AyAzM5PVq1fzxhtv4Ovri0ajITAw0JC45+XlERgYSJs2bTA3NycjI4OoqCji4+MZNmzYLd1XIaqdccbEE6JxWLJkiTJ+/HjlzTffLDWCs6Koo2OvW7eu1LLff//dMFLw+fPnlTZt2hhG3K6MjRs3Ki+++GKF6/v166fs37/f8H7Pnj3KoEGDDO+1Wq3y9ttvKyEhIUrPnj2VV155xTCCdFpamhIeHq507dpV6datmzJhwgRFp9Mpzz33nNK6dWulQ4cOyh133GEYXfpGvvnmG2X8+PGG97m5uUqHDh2UM2fOlNn2n3/+Ue64445SIzVPmTJFWbx4cZlt+/Tpo/z+++9llsfGxipDhgxR9uzZU2ok5t27dyvDhg0rte2gQYPK/F2EuBWvvvqqMnnyZGOHIYS4xoABA5SvvvpKOXHiRKmZE4qLi5WwsDBl4cKFSm5urlJQUGCYXWL79u1K7969lT///FPR6/VKTEyMEh8fryiKogQGBioxMTGG47/44ovKf//7X0VR1DI9KChIeffddxWtVqvk5+cr6enpyo4dO5S8vDwlOztbefLJJ5Xp06cb9n/00UeVWbNmKRkZGUphYaFy8OBBRVEUZdmyZYZZHxRFUX766SdlxIgRFV5nv379lLi4uHLXbdy4Ubn//vsN7//880+la9euyrlz5xRFUWfX6N69u7Jnzx5Fp9Mp+/btU7p3766kpaUpubm5SufOnZWzZ88qiqIoSUlJyr///lvucRVFUbZs2XLDOHfv3q1MmzZNycjIUDIyMpRJkyYpK1euLLXNiBEjlHbt2imBgYHKyy+/XO5xrsxScu1MGN9//73h3IcOHVK6du2qfPrpp8pdd92lDBo0SPnqq68M244bN06Jjo5WoqOjlV69eimFhYXK+PHjDfdEiLpABqcTogZt3ryZyZMn06lTJ8aPH09qamqpUZpv5MCBA3h5edXYvLv5+fls376dTp06GZYtWrSIuLg4IiIiMDMz47nnnuPjjz/m2WefZeXKlXh6ehr6Z/75559oNBoWLVrE0aNHeeONN0r14w0LCyM8PJywsLAy5z59+nSpkbxtbGxo0qQJZ86cKdMEWClnIClFUTh9+nSlr/WNN97gmWeeKTNQm6IoZY5f1WMLcb3s7Gz++usvNm/eXGp+cSGEcR05coTExESGDh2Ki4sL/v7+bNu2jcmTJxMVFUVycjIvvPCCodvElVrjDRs2MG3aNEN5XJXuBiYmJjz11FOGFh1WVlalZraYPn06Dz30EADJycns3buXgwcPGgY0vDLjxMiRI/nkk08MMwls2bKFkSNH3vK9+PPPPwkODkan05GXl8eDDz5I06ZNAfW7S9++fQ2zGfTq1Yv27dvzyy+/MHjwYExMTDh9+jQ+Pj54eHgYWtyUJywsrNzvAVe0bdvW0HoAoGfPnjzwwAOlttm6dStarZaffvqJoqKico+Tl5cHUGrGCXt7e3JzcwG4dOkS2dnZxMTEsGvXLmJiYpg8eTJNmzalV69evPbaa4bBPRctWsTatWu588470Wq1TJ06lcLCQp588skKZwARojZIU3khasi1XxDat29v+IJQWRkZGTds+n2rZs6cSXBwMF27dmX//v1MnToVUBPW9evX89JLL+Hk5ISdnR2PPfaYYTopMzMzUlJSSExMxNzcnODg4BuOdLx169YKC+u8vLxShSuoo1JfKWCv1bx5c1xcXFi+fDlFRUXs27ePw4cPG0bOvpmffvqJ4uJiBg4cWGZd586dSU5OZtu2bRQVFfHdd98RGxtb6WMLUZ4ZM2bw+OOPM2bMGO6++25jhyOEKBEREUGvXr1wcXEB1ObnV5pSX7x4ER8fn3LHOrh48SJNmjS5pXM6OzsbmmOD+qP5q6++Sr9+/ejSpQsTJ04kKysLnU7HpUuXcHR0LHcWAk9PT7p06cIPP/xAVlYWe/fuva3EvVOnToYpRvfv38/p06f573//C0BiYiI7duwgODjY8Dh69CgpKSnY2NiwePFivvnmG3r37k14eHipJvZVNWvWLJo2bcqxY8c4evQo/v7+PP/882W2s7S0ZMSIESxbtoxTp06VWX9lpoRrx0zJycnB1tYWuDrDysyZM7GysqJNmzYMHz7c0PUwKCiIL7/8kvXr19OiRQs2btzI448/zv/93/8xc+ZM3nrrLV544YXbmpVCiNslNe5C1JCKviBMnjwZAFNT0zK/HBcXFxu+NDg5OZGSklLtcX388cfcdddd6HQ6du3axaRJk4iMjMTExIT8/PxS/XEVRTEM4jV16lTD9GOg9v0NDw+/pRhsbGzKDEiWm5trKGCvZW5uzscff8wbb7zB8uXLad++PUOGDKlUf9S8vDwWLVrEsmXLyl3v7OzMJ598wjvvvMOCBQvo3bs3d911V5VGYhfiel9++aWxQxBCXKegoIDvv/8evV5Pr169AHXQ0qysLE6dOoW3tzcXL14sVQ5f4e3tXeFsEtbW1qUGYUxJSSlVhlz/A/eKFSs4f/4869atw93dnejoaEaPHo2iKHh5eZGZmUlWVpZhXIRr3XPPPYapLu+4445qK6vc3NwYPHgwa9eu5dlnn8Xb25tRo0bxxhtvlLt9nz596NOnDwUFBbz//vu88sorfP3115WatvB6//zzD/Pnzzck3hMmTChT436t4uJi4uLiaNOmTanljo6OuLu7c+rUKcPf99SpU7Rs2RLA0MqvMjG+9dZbPP3001hZWfHvv//Svn17LCwsKC4uJj09vcx4OULUFqlxF6IGXPmCcPjwYcPc26tWreLUqVOGX4q9vb3LjKQeHx9vmMKpZ8+eXLp0iRMnTtRIjKampgwaNAgTExOOHj2Ks7MzVlZWREZGcuTIEY4cOcLRo0c5fvw4oNaIz5kzh127drF06VJWrlx5w2mNbqRVq1alfjHPy8sjNjbWUMBer02bNnz11VccPHiQzz//nPj4+Ep1Ibhw4QIJCQlMnDiRXr168eSTT5KSkkKvXr2Ij48H1GaIGzdu5NChQ7z77rucP3++xronCCGEMI6dO3diampKZGSkYYrI7du3ExwcTEREBB07dsTd3Z333nuPvLw8tFqtYeqxsWPHsmLFCv766y8URTGULaCWT9u2bUOn07F3796bzjqRm5uLpaUlDg4OZGRk8NFHHxnWeXh40LdvX1577TUyMzMpKioqdbwBAwZw8uRJVq9ezejRo6vt3ly+fJmffvrJUAaPHDmSn3/+mV9//RWdTodWq+XgwYNcunSJ1NRUdu3aRV5eHhYWFtjY2GBqagqAq6srSUlJVZpmr3379qxfv56CggIKCgr49ttvDUn2H3/8wZEjRygsLKSgoIBly5aRmppaYRk9evRo/ve//5GZmcnZs2dZv34999xzDwBNmjQhODiYpUuXUlhYyNmzZ9m+fTv9+vUrdYz9+/ej1WoNy/38/Pj99985ffo0hYWFhqkdhTAGSdyFqAE3+4IAGKalioqKQlEUzp8/zxdffGEYvbRp06Y88MADPPvssxw8eJDCwkK0Wi2RkZEV1iBXhaIo7Ny5k6ysLFq0aIGJiYlhNOa0tDQAkpKS+PXXXwF1WqoLFy6gKAp2dnaYmpoaRkJ2c3MjLi6u0uceOHAgp0+f5ocffkCr1fLxxx/TunXrCqe4OnXqFFqtlvz8fD7//HOSk5NLtQy4cm9AHWlXq9WiKAqtWrViz549hr/BG2+8gaurKxEREXh7ewNw8uRJioqKyMnJ4Z133ikzUrQQQoj677vvvuPee+/Fx8cHd3d3w2PixIls3boVRVFYunQpFy5coF+/fvTt25fvv/8ewDBTx7PPPkuXLl2YOXOmYWTzl19+mZ9//png4GC2bt3KgAEDbhjHww8/jFar5c4772T8+PFlypt3330XMzMzhg4dyl133cWqVasM66ysrBg0aBDx8fHldv+qiiujv18ZJd/FxYVXXnkFUCsWPvnkEz799FN69uxJSEgIn3/+OXq9Hr1ez8qVK+nTpw/du3fn8OHDzJs3D4A777yTli1b0rt3b0Of9S1btjB8+PAK43jzzTdJSEggJCSEvn37EhcXx9tvvw2oZfuCBQvo0aMHffv2Ze/evSxbtszQ0uD6Yz/11FP4+/vTr18/Jk2axNSpU0tNAfvf//6XhIQEevToQXh4OLNmzTLM3nDlfO+++y4vv/yyYdkrr7zC/PnzmTJlCvPmzTP8SCGEMWgU6awhRLWbOnUqrVq1Ys6cOaWWb9++nYULF/LLL79gZmbGhg0bWLlyJRcvXsTV1ZVx48Yxbdo0Q0KslMzjvm7dOuLj43FwcKBr167MnDmTVq1alTnvpk2bOHTokKHQu15oaCipqamGgsfX15fw8HBDP7krSXRkZCSXL1/G09OTCRMm8NBDD/HFF1+wevVq0tPTcXBwYPz48cycORNQf6h44403yMnJMUyvNnz4cB577LEK++D99ttvLFiwgMTERDp16sRbb71lmK5p6dKlHDlyxDCV1DvvvMOGDRsoLi6ma9euvPLKK6UGBwoNDS3TemHXrl2G411x8OBBnn/++VLzuj7zzDOGPm59+vThlVdekWZwQggh6qSPPvqImJgY/vOf/9xwu9DQUFavXl2mHBRC1F+SuAvRgNwscRdCCCFE/ZSRkcE999zDu+++S7du3W64rSTuQjQ8NdpUfu/evQwePJiBAweW27Q3MzOTmTNnEhYWxtixY/n3338N60JDQwkLC2PUqFGlmsQKIYQQoubcTtld0b4ZGRlMmTKFQYMGMWXKFEMzYyFE5axbt467776bPn363DRpB3jooYfKHeBOCFF/1ViNu06nY/DgwYa5n8eOHct///vfUoNPvfPOO9ja2vLEE09w9uxZFixYYOjLExoayoYNGwwjcgshbi46OpqEhISb9rETQojy3E7ZfaN93333XZycnAgPD2fZsmVkZmaWO+WTEEIIIcpXYzXuUVFRBAQE4O/vj4WFBcOHD2fXrl2ltjl79ix33nknAC1atCAhIYHU1NSaCkmIBi8oKEiSdiHELbudsvtG++7atcswCvbo0aPZuXNnrV6XEEIIUd/V2DzuSUlJeHl5Gd57enoSFRVVaps2bdrw008/ERwcTFRUFImJiVy6dAk3NzdAHeBLo9Ewfvx4xo8ff9NzRkVFYW5uXr0XUkdotVosLS2NHUadIvekLLknZdXmPfnoo4944oknauVct0M+J6XVl/sRFBRU4+e4nbL7RvumpaXh4eEBqFNepaen3zSWhlymQ/353NUWuR9lyT0pq7bvSX0o1+VzUlZ9uSdVLddrLHEvrwW+RqMp9T48PJyFCxcyatQoAgMDCQoKwsxMDWnt2rV4enqSlpbGlClTaN68+U379Jibm9fKFxtjiI6ObrDXdqvknpQl96Ss2rwnjzzySL24//I5KU3ux1W3U3ZXZt+qaMhlOsjn7npyP8qSe1JWbd+T+lCuy+ekrIZ6T2oscffy8uLSpUuG90lJSYZf26+ws7PjrbfeAtQvC/379zeMfnlljkZXV1cGDhxIVFRUpQbjEEIIIcStuZ2yOz8/v8J9XV1dSU5OxsPDg+TkZBm/RgghhKiiGuvj3qFDB2JiYoiLi6OwsJDIyEhCQ0NLbZOVlUVhYSEA69evJzg4GDs7O/Ly8sjJyQEgLy+P/fv3lztntRBC1CXBwcHGDkGI23I7ZfeN9g0NDSUiIgKAiIgI+vfvX6vXJYQQt0LKdVGX1FiNu5mZGa+++irTpk1Dp9MxZswYWrVqxdq1awGYMGECZ8+e5cUXX8TExISWLVuycOFCQO0LN3PmTEAd4XbEiBH07du3pkIVQgghBLdXdle0L6jN659++mk2bNiAt7c3H3zwgdGuUQghhKiPaixxBwgJCSEkJKTUsgkTJhhed+7cmR9//LHMfv7+/mzZsqUmQxNCCCFEOW617K5oXwBnZ2fDdK9CCCGEqLoaayovhBCNzbx584wdghBCCCGqiZTroi6RxF0IIarJ/PnzjR2CEEIIIaqJlOuiLpHEXQghqomPj4+xQxBCCCFENZFyXdQlkrgLIUQ1uXjxorFDEEIIIUQ1kXJd1CU1OjidEEKIhk1RFAp1egqL9WivPIp0aIvVZU3dbHG0Njd2mEIIIYQQ9Zok7kIIUU26dOli7BAqlKstJjEjn0tZBfxzIZd/ChLIL9JRUKRTnwvV5/wiHfmFegquWadup0dbrENbpC9JynWGRP1GQtt4sGJyt1q6SiGEEKL61OVyXTQ+krgLIUQ1OXr0qFHOq9MrpGRrScjIJzEj3/Csvi4gMSOfzPyi6/ZKKvVOowFrc1OszE1Lnk2wtlBf21ma4WqrLrMwM8HSzBRLMxMszU2wNDXB0rzkvVnZ9e19HWvvRgghhBDVyFjluhDlkcRdCCGqSXh4OMuWLbvpdsU6Pem5hSRna0nN0ZKr1VGk01Oo06vPxepzkU5Be+V18bXrFbTFOpKztWotemYBxXql1DkcrMzwcbLG18ma4ABn9bWzNV4OVqQkxhIU2NKQmFuVJN4ajaambo0QQghR71S2XBeiNkjiLoQQ1aBIp+ezzz5j1vxFpOYUklKSlKdma0nJufK6kJQcLZfzClGUmx/zCgtTE8xNNViYmWBuqj4szUxws7OkW1MXfJys8HGyNiTq3o5W2FtV3K88uiCJ5u521XDVQgghRMP12WefSeIu6gxJ3IUQ4jo6vUJGXiGX84q4nFdIem4hl3MLSc8rec69ujyj5DmroBiA4R/uK3Usa3NT3OwtcLOzJMDVhq5NnXGzs8Td3hJ3O3W5vZU55qYaQ0JubmpyTZKukZpwIYQQQohGThJ3IUSDVqjTcymzgMt5hVzOKyQjr8iQcF9JzDNKni/nqsuyCooqrBG3NjfFxdYCJxtzXGwtaOJig4utBc42Fjz9Dix9sAtudpbqw94SWwtTSbyFEEIIIcRtkcRdCFHnFOv0xKbncTYll7MpOZxLySFHW0xhsUKx/kqfb4WiCl4X69V+4oU6PQVFeiCm3PPYWJjibGOBs605zjYW+Dnb4GJjjpONhZqM21rgYnM1SXe2scDawrTCuMclJODj410zN0UIIYQQtSohIcHYIQhhIIm7EMJosguKOFeSnJ9NyeFsci5nUnK4kJZLke5qlbebnSVONuaGpuNXnu3MzQyvzUxNDH3Br31dmJNBYFNfnK9LwJ1szLE0qzgJvxVHjx7Fx8enWo8phBBCCOOQcl3UJZK4CyFum6IoFBTpyS0sJk+rI7ewmFxtMbmFOvKuPBcWk6Mt5lJmAWeS1UQ9KUtrOIapiYYAVxtautsxsK0nLdztaOFuS3N3OxytKx5o7Waio6MJCgqojsu8qZEjR6JUZdQ5IYQQQtRZUq6LukQSdyFEuQqKdCRlFXAps4BLWQUkZRWQlKVVX2cWkJ5bWCpR11eyXLO3NKOFhx29W7rTwsO2JEG3o4mLDRZmJjV7UUIIIYQQQtRDkrgL0cBdqQ3P1haRq9WRq1Vrvq88ZxUUk1ySoCdla0kqSdQz84vKHMva3BQvRys8HSxp6+OAnaUZNhZm2FqaGp5tr3tvY2FWsp0ptpZmMl+4EEIIIYQQVSSJuxD1kF6vkJyt5XxqLhfScjmflktceh6JqZkou9PVZurXJOg3qw030YC7vSWeDlY0cbWhezMXvByt8LC3xMvRCi8HKzwdrbC3NJOk+wY+/fRTY4cghBBCiGoi5bqoSyRxF6KO0usVkrILSpLzPGJSc4lJyyUmNY8L6bklo6WrLExN8HOxxgo9bvbm+DpZldR8q7Xd6rNa4339Mnsrc1xtLTAzlWbqtys8PNzYIQghhBCimki5LuoSSdyFMCKdXuFiZr6amKddTdAvpJWfnDdxtaGpqy19WrkR4GZLM1dbAlxt8HGyxtREUzIQW5ARr6hx02g0MoiNqPf27t3LwoUL0ev1jBs3rswX1+zsbJ5//nkSExPR6XQ88sgjjBkzhnPnzjF79mzDdnFxcTz11FNMnjyZJUuWsG7dOlxcXAB45plnCAkJqdXrEkKIqpJyXdQlkrgLUcOKdHoSLucTk5ZLbHqeWmOeptaex6XnU6i7Jjk3M6GJy9XkvKmbLU1dbWnqZoO3o5qcCyFETdHpdCxYsICVK1fi6enJ2LFjCQ0NpWXLloZt1qxZQ4sWLVi6dCnp6ekMGTKEsLAwmjdvzubNmw3H6du3LwMHDjTsN3nyZKZOnVrr1ySEEEI0BJK4C1ENMvOLiEvPI7bkce3r+Mv56K7pZG5jYUqAqy2BnvYMaOtJ05Ja86autng5WGEiybkQwkiioqIICAjA398fgOHDh7Nr165SibtGoyE3NxdFUcjNzcXR0REzs9JfJw4cOIC/vz++vr61Gr8QQgjRUEniLkQlFBbrScjINyTkcel5xF0uSc7T8sgqKC61vbONOU1cbOjg60hYRx81MXdTE3R3O0sZ4K2BGjFihLFDEOK2JCUl4eXlZXjv6elJVFRUqW0mTpzI9OnT6dOnD7m5uSxevBgTk9JjZERGRpb597BmzRoiIiJo3749c+bMwdHRseYuRAghqoGU66IukcRdiGvkF+o4m5LDv0nZnE7O4XRSDqeTs4lLzys1MvuVweCauNjQ2d+ZJi42+LvYlDxbY29lbryLEEazdetWY4cgxG0pry/n9T807tu3j6CgIFavXk1sbCxTpkwhODgYOzs7AAoLC9m9ezfPPvusYZ8JEyYwY8YMNBoNH3zwAW+//TZvvfXWDWPRarVER0dXw1XVTQUFBQ36+qpK7kdZck/Kqu178u6779b5v4F8TsqqL/ekquNSSeIuGqVcbXFJgq4m5meScvg3OZv4y/lc+d5qbqqhmZst7X0cGdXJhyautobE3NNemrSLssLCwiR5F/Wal5cXly5dMrxPSkrCw8Oj1DabNm0iPDwcjUZDQEAAfn5+nDt3jo4dOwLq4Hbt2rXDzc3NsM+1r8eNG8fjjz9+01gsLS0b9GCbMphoaXI/ypJ7UlZt35P6UK7L56SshnpPJHEXDV5mfhF/xGVwPPYyUfGZ/HMpm4SMfMN6C1MTmrvb0snPibFd/An0tKOVpx0BrraYyxRp4noFmZCZAFkJkBlf8pwAWfFs27YN1j4Adu5g6wF2HmDrrj6uvLZyhFvpKqEoUFwAhXlQlKs+K7qKt634QFiln4H4PNAXqw9FV/Jad3WZ/rplih40JmBiqj5feZiYgsa04nWe7cHGperXK4yiQ4cOxMTEEBcXh6enJ5GRkbz33nultvH29ubAgQMEBweTmprK+fPn8fPzM6yPjIxk+PDhpfZJTk42/ACwc+dOWrVqVfMXI4QQt2nbtm3GDkEIA0ncRYOi1yucScnh2IXLHIu9zPHYDE4n5wBqrhToYU/XAGcmdPenpYe9mqC72NSdOcwVBQpzoVhb0QY32FmDRldYvfHo9ZCbDBlxkHnlEa8+tNlqEmrlCFZO6rO1UwXvHcHcpnTCqihQlK9eb1Gu+lyYB4U5UJRX8j736uuifNAVqslrcYF6j4oLoLjwuvda0GlLngtpoZjCzy5gYQsWduqzpf3V95Z2JcuvWacxgeyLhoT8aqKeAIXZ1912E7D3BoeSQbgux0D8IchNpdy/l6llSSJ/Jbl3BxOzkmvPvZqUX38PivLU5LkaNKuWo1RSq8EwcV1tnlHcBjMzM1599VWmTZuGTqdjzJgxtGrVirVr1wJXm7zPnTuXsLAwFEXhueeeM0zzlp+fz2+//caCBQtKHXfRokWcOnUKAF9f3zLrhRBCCHFjkriLei0zr4hjcWqCfjz2Mn/EZpCtVQeKc7Ixp0sTZ0Z28qFLgDMd/RyN1/e8ME9NgHOuPJIgN0V9vn5ZUd4tn6YNwGYbsHZWH1ZOavJs7XzN85XlJa/NrCA78WpCXipJTwB9UemTWDqAoz9YOUD6ebUGuiBDTbhvxMS8pLbZ5GpSesMfIq7f30yN1cxSTX7NLK++v/Js5QhmFlffm5iTl5aEhbUpaHPUc+Yklbwuedzsxw5bD3D0BdeW0PxuNUF39AUHP/XZzgtMS/4rfVQDM35TX+t1kJem/m1zkyEn5epnIDdFfc5OhIt/qjXe5jYlPx7YqK9t3a++trBVH1dem9uo60xu9F94xbX6cYkX8W/STK0RNzFVj2N4lNSgX/v+Sk26olcf+pJnRade55XX169T9ODe8JqqNXQhISFl5lifMGGC4bWnpycrVqwod19ra2sOHjxYZvmiRYuqN0ghhBCikZHEXdQLGXmFXEjLU+dCT8vjfGouh88lE5d5DgATDbT2cmDkHT50aeJMZ38Hmtnp0OSlqclT3mk4mQr5GVcTCkUpaVKsXH1/7WtFf817PeiK1SRWV6Q2HdYVVfD+ynaFaq10TkrZWtorbFyvNqn273G1ObW5TcU3o6Jm1oqe5PizeNiZq9eZf1lNqNPPqa/zM6A4v/x9DccuqT129APfYGg7Wn3t6A9O/uprqwpGgtYVX03iCzLU1/kZ1ywrea/or9ZuW5Qkq+Ulp9e/N721H10uRkfjdKN+TsWFJUl8rvqszVH/nvZe4OCj/gBQSaUG9jIxVf+edh4V72AkOZpoCJSEWgghhLiR8gbsFMJYJHEXdYKiKKRka4lJzSE+OY2LqRlcSs8g5XIGaZczKdbmYaUpxJJCrCjC01rPI1Z5tA9U8LfKx02TjVnBZUhMgzOpkJdecf/fm9KU9NEteb7y3tRcrYE0NVdrj03NSp6vX25eUitsDi4twM5TbQ5t53k1SbfzBFu3W05GK5LmEI3HjZLUovySZDqjJJm/rC67kqw7+Nx6TKZmYOuqPuoTMwswc6mWftjLli0jPDy8GoISQgghhLFJuS7qEkncRa0rSo8lMepncs/sxz7lGLaFqZjptThSSHdNMd3L2+n6Sk8dkAvkatSEy8ZNrb12bQFNeqivryyzKUkmbVzVJuKm5pSfnGtubdCw+sTcWn04eBs7kgbpsccekwJeCCGEaCCkXBd1iSTuombpdWTF/MHFv/agv3AA98t/4KZPIQDIVSw5adqaPLueWFnbYWtri52dPQ52djg6OGBqYQ1m1mBuVe7zv/EpBHbsrjZJFkIIIYQQQogGShJ3Ua30BdlcPLmP9FN7sUg8jF/OXziQjwNwSXHhlFU7sj0mYdeyFy079qCbs/0tn0uXUiRJuxBCCCGEEKLBk8RdXJWXDpdOwOXz6oBdusKrg60ZXheVGpRNV1xIZk4el7PzMM2Kw6/wDL7o8VY0nNE04ZD9AIr9euAWFEKb1m3pbSkfOdFwbdmyxdghCCGEEKKaSLku6hLJohojvU6dxutSFCT9BZf+Up+zEm68n4k5iqkFOo0phYop+TpT8nUatIoZxZiSae7CWfdJmDW7iyYd+9LK15vAht5nXIhrdO3a1dghCCGEEKKaSLku6hJJ3Bs6bQ4k/Q1JJ9QE/dIJSD55da5wjSm4t4amvcGzPXi1B9dW6gBmpuZkFcGx+BwOxmRxKOYyUfEZFOkUTDTQ1seB7k1d6dHchW5NXWhta2HcaxXCyHx9fWXqGCGEEKKBkHJd1CU1mrjv3buXhQsXotfrGTduXJlRGTMzM3nppZeIjY3F0tKSN998k8DAwErtK8qhK1KT9PjDEHcIEo6qc3hT8h+OlSN4doAuD6sJumd7cG+jDvZWIldbzK+nUzl0PplDMWmcTMxCr4CZiYaOfo5M69Oc7s1c6BrgjINV9U5lJoQQQgghhBCirBpL3HU6HQsWLGDlypV4enoyduxYQkNDadmypWGbpUuXEhQUxMcff8zZs2dZsGABq1atqtS+AshJVhP0+MPqI+EYFOer6+w8wa8bdLr/ak26o3+5050V6fTsO53Kd8cT+OlkEvlFOizNTOjSxJknQ1vRo5kLnZs4Y20hA8EJIYQQQgjR0CmKQlpuIRfScjmfmseFtFziL+dzfzd/ejR3NXZ4jVKNJe5RUVEEBATg7+8PwPDhw9m1a1ep5Pvs2bOGmvQWLVqQkJBAamoqcXFxN9230dEXq4n5ldr0+EOQEauuMzEH747Q9WE1WffvXmGSfoWiKPwRl0HE8QS2RV0kLbcQR2tz7u3iy4iOPnQJcMLSTBJ1Iari0UcfNXYIQgghhKgmDb1cVxSFlBwtF9LyiEnNJSYtl5g0NUm/kJpHtrbYsK2JBvycbRjWwduIETduNZa4JyUl4eXlZXjv6elJVFRUqW3atGnDTz/9RHBwMFFRUSQmJnLp0qVK7VserVZLdHR09V2EkWl0hdjH7cbp/DZap0aBvhCAImt38l07kB8winy3DhQ4BaKYXW3uzsVcuHiq3GMmZBXx87lsdp/L4WJ2MeYmGu70tyG0uzNdfW0wN9VAYTLnTifXxiXeloKCggb1964Ock/Kqs17Mnv27Hpx/+VzUlp9uR9BQUHGDkEIIRqVZcuWGTuEKtPrFbILiknPKyQ9V31czi0kPa/kueRxMbOAC2m55BbqDPuammjwd7YmwNWW4AAXAlxtaOpqS1M3W3ydrLEwMzHilYkaS9zLG8hBc10NcHh4OAsXLmTUqFEEBgYSFBSEmZlZpfYtj6WlZcP4YpN+Ho6uhONfQV4aODcjreW9uHYaAn7dMHf0wxxwqOThUnO0bPszke/+SOTPuAw0GujZ3JXZg30Z0t6r3vZVj46Obhh/72ok96Ss2rwnXbt25ejRo7Vyrtshn5PS5H4IIYQoT10s1xVF4VJWASfiMzmRkMmxs0kU783gckmifjmvCJ2+/AH1LMxMcLW1wNnGAg8HS7o3c6Gpqw1N3Wxp6mqLr7M15qaSnNdVNZa4e3l5cenSJcP7pKQkPDw8Sm1jZ2fHW2+9Bagfwv79++Pn50d+fv5N921w9Do4/SMc/hzO7FSbubceBsGPQPN+JP/zD65V+GKZkVfIz/8ks/mPRH49nYpOr9DW24GXhrVhZCdfvBytbn4QIUSVHDt2zNghCCGEEKKa1IVyPTmrgKj4TKISMvkrIZOo+ExSc7SAWkPubWeGr6slLdztCG5qgYuNBc62FrjYmuNsY4GrrSXOtua42FpgbW5aqcpQUTfVWOLeoUMHYmJiiIuLw9PTk8jISN57771S22RlZWFlZYWFhQXr168nODgYOzu7Su3bYOQkw7FVcHQVZMaBnReEvKCO/O7oW+nDFOv0/BmfyS//prD33xSi4jPQK+DrZM1jfZszurMvgZ72NXghQgghhBBCiFuVkq3lREIGJ+KzOJGQQVR8JsnZapJuooFWHvaEBLrT0c+RDn6OtPV24PyZf6XVWCNRY4m7mZkZr776KtOmTUOn0zFmzBhatWrF2rVrAZgwYQJnz57lxRdfxMTEhJYtW7Jw4cIb7ttgKApc2K/WrkdvBX0RNOsLg96ANsPBtHJN1y9m5rP33xT2/pvKr6dTyCooxkQDnfydeDK0FSGt3bnDzwkTE/llTYja4O0tA7YIIYQQDUVNlOtXBoQ7m5zLmZQczibncDYlh9NJOVzKKgDUhrct3O3o3dKNDn6OdPB1pK2PAzYWNTqTt6jjavSvHxISQkhISKllEyZMMLzu3LkzP/74Y6X3rffyMyBqHRz5HFJOqfOqd39UbQ7vdvMfJgqKdByOSeeXf1LYezqFf5NyAPB0sGRIey/6BrrTu6UbTjYWNXwhQojyJCYmGjsEIYQQQlST2ynXi3V6YtPzOJuSy9mUHM6UJOhnk3PIKrg6WruthSktPOzo2cKVdj4OdPRzoq2PA3aWkqSL0uQTUdMy4+Gf7+FUJMTsU2vXfTrDyI+g/RiwsLnh7oXFetYfjeO7Qxf56+sYCor0WJia0L2ZC2O7+hES6EGgp530VxGiDpg/fz7z5883dhhCCCFEo6Yt1pFwOZ+4y/nEpeehVxRae9rTxtsBR+vKD8pc2XL9cm4hJxLUweL+TszkdFIOMWm5FOmuDhLnYa/2Qx91hy8t3G1p4WFHSw87vBys5Hu8qBRJ3KubosClE/DPdjVZv1QyjZ1LC7jzcWh3L/h2qcRhFHafSuaNyGjOp+bi62DO/d2aEBLoTo/mLtJURog66LXXXpPEXQghhKhhOr06snpcep76uJxPfHoecZfziEvPJym7gHImqQLU8Z+CvO0J8nYwPAJcbMrtWlpeuZ6Wo+VEyUBx6nMWCRn5hvVNXGxo7WVP/yBPWnrY0cLdlubudlX6wUCI8kj2Vx10RWpt+j/b1dr1zDhAA/7dYcB8aD0c3AMrfbgzydks2BbN3n9TaO5uy8op3fDSp8nAE0IIIWrc3r17WbhwIXq9nnHjxhEeHl5qfXZ2Ns8//zyJiYnodDoeeeQRxowZA0BoaCi2traYmJhgamrKpk2bAMjIyGD27NkkJCTg6+vL+++/j6OjY61fmxCi/lEUhX+SstkVncyuExdJ33aJhIz8UrXZGg14OVjh72zDXS1d8Xe2oYmLDf4uNvi7WKNBQ/SlLE5dzCb6YhbRF7P4+Z8Uw7Rp1uamtPZSk/m2JUl9ay91UOef/0nmr/hMQ7KemFlgOG8zN1s6N3HioZ4BdPB1pJ2PI442kqCLmiGJ+60qyITTP6nJ+umdoM0EMytoEaqOCh84BOyqNoVdZn4RH+w8zeoDMVhbmPLKiLY81DMAc1MToqPTauhChBBCCJVOp2PBggWsXLkST09Pxo4dS2hoKC1btjRss2bNGlq0aMHSpUtJT09nyJAhhIWFYWGhjq+yatUqXFxcSh132bJl9OzZk/DwcJYtW8ayZct4/vnna/XahBD1R2GxnkPn09kZncTO6CTiL6s12s2dLWjn78qQ9t74u1jj76wm5z5OVliamd7wmF6OVvRrffW7eUGRjtNJOWoif0lN5refuMjaQ7Gl9puy8rB6bjdbgpu60MHXkfa+jrTzdcDBSpJ0UXskcb8Ve96GvYtAXww2btA2TJ1zvXm/m/ZZL49Or/DN4Vje+/FfLucVcn+3Jjw3KBBXO8saCF4IUVOOHDli7BCEuC1RUVEEBATg7+8PwPDhw9m1a1epxF2j0ZCbm4uiKOTm5uLo6IiZ2Y2/TuzatYsvv/wSgNGjRzNp0iRJ3IUQpWTkFfLzP8nsjE5m7z8pZGuLsTQzoU8rN2b2a0n/Nh6kJZyvthaoVuam6ojtfldb/yiKwsXMAqIvZnHqUjaXWkYwIvRO2vo4YC9JujAySdyr6uAy2PMWtB0Nd04Hv25gcuNf+G7k93NpvLb1JNEXs+jezIV5YW1p5yPNB4UQQtS+pKQkvLy8DO89PT2Jiooqtc3EiROZPn06ffr0ITc3l8WLF2NiYmJYP3XqVDQaDePHj2f8+PEApKWl4eGh1nR5eHiQnp5+01i0Wi3R0dHVcVl1UkFBQYO+vqqS+1FWY7gn8ZmFHIzP4/e4PE4mF6BXwNnalF5NbOjhZ8MdPtZYmZkAuaQlnK+1e+ID+HjB32lFOGiTiT+fXOPnvFWN4XNSVfXlnlT1RyhJ3Kvi5Bb4/gW1dn3sittK2OMv5/HW9lNEnriIr5M1Hz/QhWEdvGRUSSHqseDgYJSKRsMRoh4o7/N7fbm0b98+goKCWL16NbGxsUyZMoXg4GDs7OxYu3Ytnp6epKWlMWXKFJo3b063bt1uKRZLS8sGPbZLdHR0g76+qpL7UVZDuifFOj0XM9XB5GLT8zidnMPP/yRzLiUXgDZe9sy4248BbT3p6OtY7kBxUPv3pG3btnW+XG9In5Pq0lDviSTulRX7O2x6FPyCYcznt5y05xUWs3TPWT7dew6NBmYPCCS8b3OsLW79RwAhhBCiOnh5eXHp0iXD+6SkJENN+RWbNm0iPDwcjUZDQEAAfn5+nDt3jo4dO+Lp6QmAq6srAwcOJCoqim7duuHq6kpycjIeHh4kJyeX6QMvhKj/MvOKiC1JzK884i+rzwmX8ynWX02ALUxN6NHchYd7NqV/kAd+zlXvaipEYyOJe2Wk/AtfjwcHX5jw7S31Ywf44e9LzN/yNxczCxjZyYc5Q9vg42RdzcEKIYQQt6ZDhw7ExMQQFxeHp6cnkZGRvPfee6W28fb25sCBAwQHB5Oamsr58+fx8/MjLy8PvV6PnZ0deXl57N+/nxkzZgDqaPMRERGEh4cTERFB//79jXF5QohKKtLpycwvKvXIuvI6r/TyxMx8YtPyyCooLnUMF1sL/F1s6OjnxIiO3uoo7yWDyXk7WmFmalLB2YUQ5ZHE/WayL8FXY8DUHB7cCLaut3SYz/ed5/VtJ2nn48CHEzrTranUNgjR0MybN8/YIQhxW8zMzHj11VeZNm0aOp2OMWPG0KpVK9auXQvAhAkTmDFjBnPnziUsLAxFUXjuuedwcXEhLi6OmTNnAuro9CNGjKBv374AhIeH8/TTT7Nhwwa8vb354IMPjHaNQgi16XpMWi4nS6ZHO3Uxi8SMAjVBLygir1B3w/2tzU1xtDbH0docbycrujRxpomLDX6GadisG8RgblKui7pEEvcb0WbDmnGQlwaTt4FLsyofQlEU3v3hH/635yxD23uxePwdWJlLs3ghGqL58+cbOwQhbltISAghISGllk2YMMHw2tPTkxUrVpTZz9/fny1btpR7TGdnZ1atWlW9gQohKiUzr8gw3Zn6yObfpGy0xXoAzEw0tPSwo4mrjSEZv/7hcN17C7PGUVsu5bqoSyRxr4iuCNY9BEl/wwPfgm+XKh+iWKfnpe9OsO5IPBN7NGHBqPaYVjDYhhCi/vPx8SExMdHYYQghhKininV6jsdl8Ms/KRz45yKOB3OwNDPBytwUK3MTLM1MDa+tzE2xMqwzNWyXV6gzJOmnLmWTkJFvOL6LrQVB3vZMujOAIG8HgrwdaOlh12gS8aqScl3UJZK4l0dRYMuTcHY3jPwIWg2s8iEKinQ88fVxdkYnMat/K54e0EpGjBeigbt48aKxQxBCCFHPJGTks/ffFH75J4X9Z1LJ1hZjaqKhqZM5Wk0BBUV6Cop0aItLnov0FOr0NzymqYmGFu62BDd15kGvAIK87Wnr7YC7vaV8H60CKddFXSKJe3l+Xgh/roW750KXSVXePTO/iEdXHeHwhXReH9WOST2bVn+MQgghhBCi3iko0nHofLqarP+bwunkHAC8Ha0Y3tGbkEB37mrpRmLMmQqntNLpFbTFOgqK9IbngiIdBUU6LMxMaOFuJ10zhWhgJHG/3pEVsHcRdHkIQl6s8u5JWQU8vOIQZ1NyWDKhMyM6+tRAkEKIuqhLl6p3qRFCCNGwKYrC+dRcfilJ1H8/l0ZBkd4wJdr4bv6EBLrT0sOuVG34jRpom5posLEww8ai5uNvzKRcF3WJJO7X+ud7iHwWWg2C4Yuhik2JzqXk8NCKQ1zOLeSLKd3p1dKthgIVQtRFR48eNXYIQgghjKxYp+fUpWyOxV7m2IXLHI65bOhn3tzNlvu7NSEk0J0ezV2wsZCv4nWZlOuiLpH/La6IPwLrp4B3Jxi7EkyrdmtOxGcyeeUhAL4J70kHP8eaiFIIUYeFh4ezbNkyY4chhBCiFqXnFnI89jJHL1zmWOxl/ozLJL9InU7Nw96SLk2cefzuFoS0cqeJq42RoxVVIeW6qEskcQdIOwtf3wf2nvDAerC0q9Lu+06n8tiXR3C2teDLqT1o5mZbQ4EKIeqyzz77TAp4IYRowHR6hX+TrtSmZ3As9jLnU3MBdVq1tj4OjO/mT+cmTnQNcMbXyVoGg6vHpFwXdYkk7jkp8NUY9fWDm8DOvUq7b4tKZPa3f9DC3Y5Vj3TH08GqBoIUQgghhBC1qVin51xqLicTs/g7MZO/E7OIis8kR1sMgKutBZ2bOHNfsD9dmjjR0c8JawsZEE4IUTMad+JemKvWtGdfgsnbwLVFlXZffSCGeVv+pluAC589HIyjtXkNBSqEEEIIIWpKXmEx0RezOXkxi5OJmZxMVOdA1xar065ZmJnQxsue0Z196NLEma4BzjRxsZHadCFErWncifvxryDxGNz/NfgFV3o3RVFY/NO/fLj7DAPberJkQmeZckMIQUJCgrFDEEKIRuOPuAx2RydhamKChZkJ5qYaLMxMsDA1wdz0yjITLEuer6w3NzUhLbfQUJN+8mIW51NzURT1uI7W5rTzceChngG09XGgrbcjLdxtMTM1Me4Fi1on5bqoSxp34n5+LzgFQJvhVdrtkz1n+XD3Ge4L9uPNezrIf+RCCEAdfdbHR6aAFEKImpScVcA7O/5h47H42z6Wn7M1bb0dGNXJV03SfRzwcbSSmnQBSLku6pbGm7jr9XDhN2g9tEq75Rfq+OzXc/Rv48E7YzrKf+xCCIORI0eiXKmyEUIIUa20xTpW7Ivho92nKdIpTL+7BTP7tcTa3JQinR5tsZ4inZ7CkueryxTDssJiPYU6PQ5W5rT1dsDRRro5iopJuS7qksabuKecgvx0COhVpd02HY8nI6+Ix0JaSNIuhBBCCFHDFEVhZ3Qyb0Se5EJaHgPbevLysCCaXjOLj6mJqXRbFEI0aI03cb+wX31uWvnEXa9XWLHvPO19HejW1LmGAhNCCCGEEACnk7JZsO0kv55OpaWHHV9O7U6fVlWbAUgIIRqCxpu4x/wKjv5qH/dK+vVMKmdTcvnvfZ2ktl0IUcann35q7BCEEKJByMwrYvHOf/ny9wvYWpgyL6wtD94ZgLmMK1T9FAUiZoA2C4a+C46+xo6o+imKen156ZB/WW11m59R8r5kWal1lyH/Mp8+1B7O/gzN7wb57i+MrHEm7oqi9m9v0b9K/whX7DuPu70lwzt612BwQoj6Kjw83NghCCFEvabTK6w9FMt7P/5DZn4RE7o34ZmBgbjaWRo7tIbr4FL482swMVMHbh78JnR+sOEkqj++Ar9/AvriirexdABrZ/Vh4wLOzcDSnnCzH+DL0eDXHUJehJZVyx2EqE6NM3FP/RdyU6rUTP5Mcja//JvCMwMDsTSTPlRCiLI0Go0MYiOEELfo93NpvLb1JNEXs+jRzIVXw9rSzsfR2GE1bBf/hJ9ehcChMORN2PwEbHkCTm6GsA/qf+370VXw24fQdhT4dQNrFzUxt3ZWX1s7g7UTmJY/SKFGo0E5tBz2LYY1Y8Cns5rABw6p/gS+WAsaUzCtpvRMr1NbGRRkqo/8jKuvCzJBXwRmVurD3LqcZ0swswZzq9LP1RWfqLLGeedj9qnPVRiYbuX+GCzMTHigR5MaCkoIIYQQomHLL9SRlqslPbeQtNxC0nMKSc8tZM9fl9gfew5fJ2s+fqALwzp41a1uiWd2wQ8vQ2YcmJiqtdOGhymYmF/3/pr15lYQPBWCRhj7KkrT5sCGR8DGFUZ9DLau8PA2OPwZ7JwPn/RUk/k7JtbPWua4w7D9OWgRCmNXqn+XW9FtKnSeBFHfwN7/wNr7wasjhLwArYeDyS1239AVQ/xhOLsLzuyExD8ARU3ezaxKEmWrkgS64mfvrGz4Q1OSkGdcTcy1WbcW182MWAzBj9TMscUNNc7E/cJ+sPcGl+aV2jwjr5CNx+IZfYcPbtJUSwghRAO2d+9eFi5ciF6vZ9y4cWW6gGRnZ/P888+TmJiITqfjkUceYcyYMVy8eJEXXniB1NRUTExMuO+++3j44YcBWLJkCevWrcPFxQWAZ555hpCQkFq/NlGzLmUWcPB8GokZBaTnatXEvOSRVpKg5xfpyt3X2kzD7AGBPBbSvHKjwysK/P0dXDoBwVPAqYYqVrIuwg9z1XO5toQuD4OiU5td64rUWk198XWP65aln4NvJ0Lb0TBsEdh51EysVfX9C5B2Fh7eoibtoCahPR6DVgMhYiZsnnm19t2hGuYzL9bCvz/AuT1w1xOV/i5eZdlJsG6SGvOYz289ab/CzAK6PASdJsCJ9bB3EXz7IHi0g5DnIWhU5RL4zHj1R6AzO+HcL6DNBI2J2hS/z7NqMl5coN6n4oLrXpc8F+WrffBL3ttoC8DeDawc1bG7rBwrflg7XX1tYn71HEX5Jc8FUJx/9blYe826fNBpoZn8320sjS9xVxS1xr1Z30r/erj2UBwFRXqm9GpWw8EJIeqzESPqWG2KEFWk0+lYsGABK1euxNPTk7FjxxIaGkrLli0N26xZs4YWLVqwdOlS0tPTGTJkCGFhYZiamjJnzhzatWtHTk4OY8aMoVevXoZ9J0+ezNSpU411aaIG5BUWc/BcOr+eTuXX0ymcTs4xrLMyN8HV1hIXWwtcbC1o6W6nvrazwNXWApeSda4ly+LPnaZt21aVO3HCMdgxF+J+V9//tgS6Toa+z4G9V/VcnK4YDi+H3W+ArhD6vQy9ZqmJVZWPVQT734df3lUT1iFvqQmgMWuxo9bDH2ug7/Pqd+LruTSHyZFwaJla+/7xnTD07VuLW6+H2AMQ9S2cjFBrg0FN4CdvA5dq/n5dXAjrHlLPM/UntWn8LSpTrpuawx0PQIf74O9NagK/fjK4t1HvZbt7Sv9IUFSgXvuZnWrCnhKtLrf3gbYjoeUAaB6iNtm/RWejowkKCrq1nS1sbvm8ovY1vsQ97SzkJFW6mXyRTs/qAzHc1cKVIG+HGg5OCFGfbd261dghCHFboqKiCAgIwN/fH4Dhw4eza9euUom7RqMhNzcXRVHIzc3F0dERMzMzPDw88PBQaxLt7Oxo3rw5SUlJpfYV9ZtOr/B3YqYhUT964TJFOgVLMxO6N3NhXLAfvVq60czNFhuLqn3FrFSz+KyLsGuBOpCarTuMXALN+8Gv78HRlXD8K+j+KPSefVvJGvFHYdvTcClKHch42CJwbXHrxzM1V5O6oFGw5UmImK7W2o54H5wrP7tRtUk/B9tmg38PCJlT8XYmJnDn42rt++aZatx/R5TUvldioObkU2qyfmK92sXA3BaCwqDjferf58t7YFWY+gNBdd6HH0p+1Bm7Erza39ahKizXTc3U62g/Rv0x4pdFsHEq7HkLej0NRXlqsn7+V7Xm2tQCAu6CzhPVZN29Tf3sfiCMqvEl7hdK+rc37V2pzXf8dYmLmQW8Pur2/uELIRq+sLAwSd5FvZaUlISX19UaS09PT6KiokptM3HiRKZPn06fPn3Izc1l8eLFmFzXRDQ+Pp7o6Gg6depkWLZmzRoiIiJo3749c+bMwdHxxoOOabVaoqOjq+Gq6qaCgoJ6cX1JOUUcT8zn+EX1ka3VA9DcxYJRbRzo4mNNWw8rLM1MgELITORCZtXPc6P7oSkuwOWfr3GLXg2KjvSgh0gLehi9uS1czIGWj2HuNRz3v5bj8NsS9IeWk976AdJbT1C3qSSTwmw8ov6H09nvKLZ2I+muhWT7hUJyISRX09+q539xdt+Ee9QnaD7uQXKHx7nccmy5Tblr5DOiK6Lp7sewUOBcpzkU/3u6cvv1+A/OLuvxOPE/lCXduNTlGbICyg7QZpafikPsjzjG7MAq418UjSm5Xt3JDJpGtm9fFDNrKAIywar3YprseRLd8sFc6PcJxbY3/zHgZvfE8dxWfA4vJ7XNg6SYBsFt3r8ZM2bwySef3Hgj07bQ73Ps4/fg9vcKrLY8AUChnR85TYeT692TXI8u6rUDpAFpp24rrmvVl/9LalN9uSdVbSnR+BL3mP1g66H2U6qEFfvPE+BqQ2ibOtIfSQhRZ23bts3YIQhxW8qbFeH6mtB9+/YRFBTE6tWriY2NZcqUKQQHB2NnZwdAbm4uTz31FC+99JJh2YQJE5gxYwYajYYPPviAt99+m7feeuuGsVhaWt568896IPp2mrfWEEVRiL+cz5EL6RyJucyBs2mcS80FwNPBkkHtfOgb6MZdLdxwt6/eMX/KvR+KAn9thJ/mQVY8BI2EgQtwc2mGW5kjBEG3gZAcjenPC3H/eznu5zZB76eh26M3bhKsKGqt8I8vQ14q9Hgc834v4WdVQy0t27aDvpNh22y8ji/GK2UfjPwIPNqU2qxGPiM/vQrpJ+G+1bRqG1q1fdu1hz4PQcQMfA++hu/lQxD2PljYQvQ2tXb9/C+g6MGnC9z5Dpr292Jn54FduQcMgmZNMV01ilb7noYp28HR74Yh3PCexB+FY4ugeT/c7vsQt9vt1w7s2bOn8n+Dtu1gwHSIOwj2nli4NMcFuI22H5VSF/8vMbaGek9qNHG/1QFuAEJDQ7G1tcXExARTU1M2bdp0+wEpijowXdNelWqeciz2MsdjM5gf1hYTE2nOIoQQomHz8vLi0qVLhvdJSUmG5u9XbNq0ifDwcDQaDQEBAfj5+XHu3Dk6duxIUVERTz31FGFhYQwaNMiwj5vb1TRr3LhxPP744zV/MeKmtMU6/krI4tiFyxy9cJmjsZdJydYCYGdpRnBTZybeGUCfVm608rCr3VHe44+WNHk+CF4d4N5PK9da0iMIxn+l9oPf/YaaqB74RO3/3uVhdZCxa6Wehshn1PnLfbvCgxvAu1P5x65OTv4wcb36g8H3L8KnfaDPc2oz/+tjrC5ndsH+D6DrFHV6tFvh2kJNsH//H+x+HT7qro4BUJyvDozW5zm1CblbJccr8OkMD30Hq0fDFyPUY9/KIHjZSepgcfbeMHbF7Q9Gd6tMTCCgp3HOLRq8Gkvcb2eAGwsL9T+sVatWGUagrRaXYyArodL921fuj8He0oyxwf7VF4MQQghRR3Xo0IGYmBji4uLw9PQkMjKS9957r9Q23t7eHDhwgODgYFJTUzl//jx+fn4oisLLL79M8+bNmTJlSql9kpOTDT8A7Ny5k1atKvmlXlSrlGwtx2Ivc+zCZY5cuMyJ+EwKdWrT9yYuNvRu6UaXAGe6NnGmtZc9psaotMhKhJ2vqVNv2XqoNdF3PFD1RMy3C0zaBBd+g12vq9OC7f8Q7n4ROt6vzmH963tqImtuDcP/qw5wV5sJn0ajJrktQtXkfc+ban/pkR+BX9fqPVdOMnz3OLgHweA3b+9YJqbqiPCBg9UfR6ydodP9ap/5W/lxx7crTLomeZ8cWbk+9FcUF8L6h9WR1qfd3mB0QtRlNZa4384ANzXmyvztTfvcdNOLmflsP3GRKXc1xc6y8fUoEEJUXXnNjIWoT8zMzHj11VeZNm0aOp2OMWPG0KpVK9auXQtcbfI+d+5cwsLCUBSF5557DhcXF44cOcLmzZsJDAxk1Ci1Nu/KtG+LFi3i1Cm1T6evry8LFiww2jU2Nr/8m8Lm4wkcjb3MhbQ8ACxMTejg58jkXk3p0sSZLgFOeNhbGTVOTXGBOur6vsXqFGq9Z6vTY1na396BA+5Sa3HP7lIT+M0zYd/7auJ+OQY6jodBbxh3ijZbNxj7OXQYpw4a9/kAuHMGGp8x1XN8vV5N2rVZ8NDm6htJ3K0V3Leqeo7lFwwPboSv7i0ZsG5b5WcI+OEldeT2MZ+rrTOqkZTroi7RKDX0idyxYwe//vorCxcuBCAiIoKoqCheffVVwzY5OTlMnz6d8+fPGwa4ufvuuwG1qbyjoyMajYbx48czfvz4m57zjz/+wNKy4j5X3gcXYHfxAKdHbb/pL4Irj6az4e8MPr/HHy9780pccc0qKCjAysq4hWpdI/ekLLknZdXmPVm3bh333XdfrZzrdsjnpLT6cj8aYn+9G2mofRSvqOnru5iZz4KtJ/n+r0u42loQ3NSZrgHOdA1wob2vA5ZmRmpKfIVeD2lnIP4wJByh6O+tmOenqE24By4A56bVf05Fgeit6g8EKGrNc/M6Nid1QZY6/dqRzym09cZi4Kvq1GOmt1GJtP9D+OkVtVVBtzo+JeOFA/DVGHD0VWver/tBpcy/m+NfqT/G3PWk+gNMNVu2bFmZrr51TUP/v/JWNNR7UmNVybc7wM3atWvx9PQkLS2NKVOm0Lx5c7p163bDc950IJsdf0HzPgS1bXvD4+QX6vhh3S4GtfWiX/eON9y2tjTUD+DtkHtSltyTsmrznsyfP5958+bVyrluh3xOSpP7IRqSYp2e1Qcu8N6P/1CsV3h+cGse7dMcCzOTm+9ck/LSIf4IJBxRk/X4o6AtGX7e0gGtc1vMx69SxyGqKRqNOnd225E1d47bZeUAI/4L7cegj3hanYLtl3fVPvodx6tTy1VFwlHY9Rq0GQHBj9RIyNUqoKfa93/NWLXm/eFtYOde/rYJR2HbM9D8bug/v0bCeeyxx+p84i4ajxr7X7yyA9wMGjSozAA3oE5BA+Dq6srAgQPLTEdTZRmxkBlbqYFNNh2PJzO/iEd6N7u9cwohhBBG8vPPP6PX640dhqgucYcgYqY6CFcF/ojLYNTH+1mw7STBTV34aXYIM/u1rP2kXVcEicfh0GewKRw+7ALvNoOvx8HeRWp/6/b3wKiPYcZBePECcSHv12zSXt807cX5Qavg/rVqMr95JizpCsdWq/e3MgqyYMNUsPNS57yvL/OGN+0FD6yDyxdg9UjITS27TU4yfPMg2Huq87XfTosEIeqJGvuU384AN3l5eej1euzs7MjLy2P//v3MmDHj9gKK2a8+32RgOr1eYcW+87T3daBbU+fbO6cQQghhJJGRkSxcuJBBgwYxZswYWrRoYeyQxK3KvwzrHobsRDi7G+7/Sh3Qq0RmfhH/+eEfvjp4AXc7Sz5+oAvDOnjV7ijwoPYZ3zZbHRCuuEBdZusB/t2h84Pg100dRdyy/MnBxHU0GmgzDFoPhX9/gF/ehi1Pwi+LoM8zcMfEikegVxSIfBYyLsDk7fVvwLZmfeCBb+Dr8bB6FDy0BWxd1XW6Ilg/Wf13MfXH+ndtQtyiGkvcb2eAm7i4OGbOnAmoo9OPGDGCvn373l5AMfvUUS89btxM/tczqZxNyeW/93Wq/QJPCFGvbdmyxdghCGHwn//8h5ycHLZt28bcuXPRaDTce++9DB8+3DC/uqgnIp+D3GQY9QnseRtWDIWRS1A63sfWqIu8vu0kaTlaHu7ZlGcHBWJvZYSxec7vVX9cUHRqk2y/buqAY47+9aemt67SaKD1EHUU9zM71c/AtqfVUfF7z1Z/FDG7boynP7+BE+vg7pfq7/Rkze+GCWvh6/vhy5LkHeCHl9Xpncd8Dt4126VVynVRl9Rou5KQkBBCQkoP+jFhwgTDa09PT1asWFFmP39//+r/h3Jhn1rbbnLj5mIr9p3H3d6SER1vYQ5JIUSj1rVrNU/fI8RtsrOzY9CgQRQUFLB69Wp++uknPv/8cyZNmsSkSZOMHZ6ojBMb4K8N0O9l6DxRTd7WT4bvwtm+80eeThlNez9nVjzcjQ5+jrUfn6KoTeJ3zAHXlmqi5SqtO2qERgOtBkLLAWrLi1/eUeegNyTwk8DcClLPqLXtAb3VvvH1WYtQmPA1rJ0AX47G2SsUjn8KPZ+ADmNr/PRSrou6pHF0CMlMUJtvdX/shpudSc7ml39TeHZgoPEHcRFC1Du+vr4ydYyoM3bv3s3GjRuJjY1l1KhRrF+/HldXV/Lz8xk2bJgk7vVBZryamPl1g97PAKC1dGaZzyKcz73Kg9kbuMM3Ea8pX2Nqa4SkvVirJojHv4TAoXDvMrU/tqhZGg207K8mtef2qAn89ufUBL7X0/DHGrUJ/b3Lande+prScgCMXwPfTsTr4p/QrC8MeK1WTi3luqhLGkfifqGkf/tNBj1ZuT8GCzMTHujRpBaCEkIIIWrOjh07mDx5cpkZWaytrXnzzTeNFJWoNL0eImaArhju+RRMzfjtTCr/F/EX51JzGdFxLiObDMJ391z4vL9a0+3euvbiy06CdZMg7iD0eU5tEXCTVo2immk00KKf2qQ85lfY8w7seFFdd//X6pRqDUXgILj/a7L2LMFBBqMTjVTj+NTH7ANLR/BsX+EmGXmFbDwWzz13+OJqV/Fc8EIIIUR98OSTT5aazaWgoIDU1FT8/Pzo2bOe9nltTA4uhfO/QNgHHM914eNth9kZnUwTFxtWPdKdkEB3oAs06QDfPgif9Ycxy9W+0DUt4Rh8MxEKMtQRvdvfW/PnFBXTaNRa6GZ91cGYc1OgzXBjR1X9Wg0kodgPhyuD1AnRyDSOn0Yv7FcH5rhBc6G1h+IoKNIzpXfT2otLCNGgPProo8YOQQiDWbNmlRpk1cTEhFmzZhkxIlFpydEoO+dz2S+UB4+14Z5PfuPIhcs8OzCQH2f3LUnaSzS5E8L3gGtzWHs/7P2P2u+8pkStg5VDwcQMHvlBkva6pmkvaDfa2FE0GFKui7qk4de4Z1+CtDPQ5eEKNynS6Vl9IIZeLV1p4yV9s4QQt2bZsmXGDkEIA51Oh4XF1amiLCwsKCqq5PzPwmiUYi05X09Br7di4JlxYJfDS8PaMLFHALaWFXxtc/SDKTtg61Ow+3VI+htGfQQWttUXmF4HO+fDbx+qg57dtwps3arv+ELUQVKuV05RURHx8fEUFBQYOxRAjSc6OtrYYdwWKysr/Pz8MDe/OktIw0/cY/apz017V7jJjr8ucTGzgDdGV9yUXgghbqZr164cPXrU2GEIAYCLiwu7du2if//+AOzcuRNnZ2cjRyUqotcr/Hgyiext/8e4gmheMJvDU6Pu4r5gf6zMKzHAmIUN3PsZeHWAn+ZB2mm1n7NTNYzbk38ZNk5TpyLrNg2GvA2mRphyTohaJuV65cTHx2Nvb0/Tpk3rxHTa+fn5WFtbGzuMW6YoCmlpacTHx9OsWTPD8oafuF/YDxb24FXxPI8r9p+nqasN/Vp7VLiNEELczLFjx4wdghAGr732Gs899xyvv/46iqLg7e3NO++8Y+ywxHWKdXoiT1zk45/P4Jh8hG8tN3DOfwxvPPxi1We40Wig1yzwaAsbpsKyfnDf6psOzntDKf+qTfAzYmHE+xA85daPJUQ9I+V65RQUFNSZpL0h0Gg0uLq6kpKSUmp5w0/cY/ar/b8qGH3yWOxljsdm8NrIdpiYyIdNCCFEw9CkSRPWrVtHbm4uiqJgZ2dn7JDENYp0Ct8ejuWTPWe5kJZHZw9TvnD+HI1lAM0f/ABuZ1raVgPh0V3q3NerR0KPx9Wad2tn9WHldM1rx4pH6P5nB2x6FMws4eGt6nhBQghRDknaq1d597NhJ+45KZD6D9wxocJNVuw7j72VGWO7+tViYEKIhsjb29vYIQhRyp49ezh9+jRardaw7IknnjBiRKKgSMe3h+P4aFcsKbk6Ovg68umkrgz69zU0URfh/h1gaX/7J3JrpSbvETPgwEc33tbSAaydSif1Jqbw1ybw7qjOoe3kf/sxCVHPSLleP1y+fJnJkycDkJqaaqixBli/fn2p8V6ud+LECTZv3sz//d//1Uaot6XCxP3XX38lNzeXIUNKTyuyZcsWXF1d6dXrNppd1ZYLJf3bA8rv356Ykc/3f13ikV5NKx7wRQghKikxMdHYIQhh8Oqrr1JQUMDBgwcZN24cP/zwAx06dDB2WI1WXmExa36PZdmv50jJ1tLOw4r/jO9K31ZuaKK3wp9fq/OhN+lRfSe1coT716hzwRdkqtO35V++5nHd+yvrk0+q6+54AIb9R+0/L0QjJOV6/eDs7MzmzZsBWLJkCebm5jz++OOG9cXFxZiZlZ/rdejQod6UjRVmq0uWLGHp0qVllvfs2ZMnnniifiTuMfvB3BZ87ih39Z5/UtDpFe7vXg0DtwghGr358+czf/58Y4chBADHjx9n69athIWF8cQTTzBlyhSefPJJY4fV6GQXFLH6wAU+33ee9NxCerV0ZcmEzjhqkwkKdFdnv9k6C7w7QciLNROEqRnYuqoPIUSlSblef82ZMwdHR0dOnjxJu3btGDZsGG+++SYFBQVYWVnx5ptv0rx5cw4ePMiKFSv49NNPWbJkCYmJicTHx5OYmMjDDz/MQw89ZOxLMagwcc/Pz8fFxaXMcnd3d/Ly8mo0qGpzYb/6y3UFI58mZ6tTFjRxkV+ShRC377XXXpMCXtQZlpaWAFhbW5OUlISzszPx8fFGjqrxyMwrYuVv51m5P4bM/CLubu3Ok6Gt6BqgjuwfHZ2szre++QkoylNHhDeruDmnEKL2SbledRuPxrPuSFy1HvO+YH/G3EK35piYGL744gtMTU3Jycnhq6++wszMjN9++43FixezZMmSMvucP3+e1atXk5OTw9ChQ5kwYUKpKdmMqcLEvbCwsNxmBUVFRaX6ytVZuWlqU6/2YyrcJCVbi7ONOeamtzEAjBBCCFEH9evXj6ysLKZOncq9996LRqNh3Lhxxg6rwUvPLWTFvvOs+i2GbG0xA9t68mRoSzr6OZXd+MgKOPMTDH0X3FvXeqxCCNGQDRkyBFNTdTrN7OxsXnzxRS5cuIBGo6GoqKjcfUJCQrCwsMDFxQUXFxfS0tLw8vKqzbArVGHiPnDgQF555RVeeeUVbGzUGum8vDzeeOMNBg4cWGsB3rLY39TnG8zfnpqjxc3OspYCEkIIIWqHXq+nZ8+eODg4MHjwYPr164dWq8Xe/uaDnu3du5eFCxei1+sZN24c4eHhpdZnZ2fz/PPPk5iYiE6n45FHHmHMmDE33DcjI4PZs2eTkJCAr68v77//Po6OjtV/4UaUkq1l+a/n+PL3C+QX6Rja3osn+rWirY9DudtbZMfCT/8HLUKh26O1HK0QQtSMMV39bql2vCZcO5f7Bx98QI8ePfj444+Jj4+vsAn8tQPZmZqaUlxcXONxVlaFVc1PP/00rq6u9OvXj3vvvZd77rmH0NBQXFxcePrpp2sxxFsUsw/MrMGnS4WbpOYUSuIuhKg2R44cMXYIQgBgYmJSas52CwuLSiXtOp2OBQsWsHz5ciIjI9m2bRtnzpwptc2aNWto0aIFW7Zs4csvv+Sdd96hsLDwhvsuW7aMnj178uOPP9KzZ0+WLVtWvRdsRJcyC3ht69/0eXc3n/16joFtPfnx6b58MrFrhUk7uiJ8fp8PphYw6mMwkZZ/QtRFUq43HNnZ2Xh6egLw3XffGTmaW1NhjbuZmRnPPfccTzzxBBcuXAAgICAAKyurWgvutsTsB/9uN+wvlpqjLb/pmhBCCFHP9erVix9++IFBgwZVen7dqKgoAgIC8PdXp/4aPnw4u3btomXLloZtNBqNYW743NxcHB0dMTMz488//6xw3127dvHll18CMHr0aCZNmsTzzz9fzVdcu4p0et7afoqvfr+ATlG4p7MvM+5uQXN3u5vvvPc/WKefhHFfgINPjccqhBCN3bRp05gzZw4rV67kzjvvNHY4t6TCxP3HH38s9V6j0ZCdnU2bNm2ws6tEoWRM+Zch6S+4e+4NN0vN1uIuNe5CiGoSHByMoijGDkMIAFauXEl+fj5mZmZYWFigKAoajYZjx45VuE9SUlKpvnyenp5ERUWV2mbixIlMnz6dPn36kJuby+LFizExMbnhvmlpaXh4eADg4eFBenr6TePXarVER0dX6ZprS7Fe4e29yey/kMvgVvaM7+CEt7052tQ4olMr2Emvwyb5KI4x3+MY+yOX/QaSZNIG6ug11raCgoI6+/c2FrknZdX2PQkODubkyZO1dr5bURc+J0VFReTn5xs1hiumTZtmKO8AQ1xt2rQhIiLCsN1jjz1Gfn4+HTt25P333yc/P59p06aV2mf9+vWl3le3a5vyV0aFifvPP/9cZllGRgb//PMPCxcupGfPnlWPrrZcOAAoN+zfnldYTG6hDjd7GcFVCCFEw3P8+PEq71PeD0/X19bv27ePoKAgVq9eTWxsLFOmTKnwR6vK1vSXx9LSkqCgoFvev6YU6fTM+uY4+y/k8sqItkzt3ezGO6T8A3+uhah1kJUAlo7Q5SFSAh6sk9dnLNHR0XI/riP3pCxj3JO6/jeoC5+T6OjoKiehNSk/P79OxVNdKkzc33rrrXKXJyQk8PTTTxt+gaiTLuwHU0vw7VrhJqnZhQDSx10IIUSDdPjw4XKXd+vWrcJ9vLy8uHTpkuF9UlKSoab8ik2bNhEeHo5GoyEgIAA/Pz/OnTt3w31dXV1JTk7Gw8OD5OTkcqebrQ+KdXqe/uYPtp+4xP8ND6o4ac9LhxMb1IQ98RhoTKFlfxj0BrQeCubW6KUmVQghRBVUmLhXxNfXt06NrleumH3g1w3MK+6Pn5KjTmknTeWFENVl3rx5xg5BCIPPP//c8Fqr1RIVFUW7du1YvXp1hft06NCBmJgY4uLi8PT0JDIykvfee6/UNt7e3hw4cIDg4GBSU1M5f/48fn5+ODg4VLhvaGgoERERhIeHExERQf/+/WvmomtQsU7PrG//IPLERV4eFsS0Ps2v26AQTv+oJuv//gD6IvDsAIMWQodxYO9pnMCFELdMynVRl1Q5cT979mypYfLrnIJMuBQFfW886E1qSeIuNe5CiOoyf/58Y4cghMHSpUtLvb948SKLFi264T5mZma8+uqrTJs2DZ1Ox5gxY2jVqhVr164FYMKECcyYMYO5c+cSFhaGoig899xzhhr08vYFCA8P5+mnn2bDhg14e3vzwQcf1MAV15xinZ7Z6/4kMuoiLw1rw6N9S5J2RVFr1P/8Rq1hz08HWw/o8Rh0uh+8Ohg3cCHEbZFyXdQlFSbujz/+eJllmZmZpKSk3LTgN6rY30HRQ0CvG25mSNylj7sQopr4+PiQmJho7DCEKJeXlxenT5++6XYhISGEhISUWjZhwgTDa09PT1asWFHpfQGcnZ1ZtWpVFSOuG4qLdcxbu4e4k3/xWRdzBmr+ga2xkHEB0s5ARqzaPa/NcOg0QZ2X3bTK9SJCiDpIynVRl1RYsjzyyCOl3ms0GhwdHcnMzGTLli107ty5xoO7JTH7wMRcbSp/A1f6uLvaSo27EKJ6XLx40dghCGHw+uuvGwaH0+v1REdH07p1ayNHVUcV5UPySTUJz4iFyxcgIxYlIxZ9WgwLFS1YAidLHjau4BQAPl2gz7PQdjRYOxn1EoQQ1U/KdVGXVJi4d+/e3fA6OjqarVu3smPHDnx9fRk8eHCtBHdLLuwHv2CwsLnhZik5BTjZmGNhZlJLgQkhhBC1p3379obXpqamDB8+nK5dKx60tdHKS4flAyD97NVl1s4oTk2IKvDkYFFL2rRpR99uXcE5ABz9wbKOT4srhBCNyKRJkwgPD6dPnz6GZV988QUxMTHldneYNGkSL7zwAh06dODRRx/lvffew8HBodQ2S5YswcbGhqlTp1Z43p07d9K0aVNatmwJwAcffEC3bt246667qufCrlNh4n7+/HkiIyOJjIzEycmJYcOGoSgKX375ZY0EUi30Okj8A3rPvummqdmF0r9dCFGtunTpYuwQhDAYPHgwlpaWmJqaAqDT6RrsFDm3TFcE6x+GzDgYvVTtk+7UBJ2FPc+v/5NN5xN4fnBr+vZraexIhRBGIOV6/TB8+HC2b99eKnHfvn07L7zwwk33/eyzz275vDt37uTuu+82JO6zZs265WNVRoXVzUOHDuX3339n6dKlrF27lkmTJmFiUsdrpwtzQdFB0xv3bwe1j7ubnfRvF0JUn6NHjxo7BCEMJk+eTEFBgeF9QUEBU6ZMMWJEddAPL8P5vRD2AdwxAbzaq0n7hj/ZdDyBZwcGMlOSdiEaLSnX64fBgwfz888/U1iodoVOSEggOTmZrVu3cu+99zJ8+HA+/PDDcvcNDQ0lPT0dgP/9738MHjyYyZMnc/78ecM269atY8yYMYwcOZInn3yS/Px8jh07xu7du3n33XcZNWoUsbGxzJkzhx07dgBw4MABRo8eTVhYGHPnzjXEFhoayocffsg999xDWFgYZ8+eLRtUBSqscV+yZAmRkZE89NBD9OnTh+HDh6MoSqUPbBSFOWBiBv49brppao6W9r6OtRCUEKKxCA8PZ9myZcYOQwhAnQLO1tbW8N7W1pb8/HwjRlTHHP0CDn0KPZ+AOx4AQKdXeGFDFJuOJfDMwECe7N/KuDEKIYxKyvVb8MdaOP5V9R6z84Pqj6sVcHZ2pmPHjuzdu5cBAwbwww8/MHToUB577DGcnJzQ6XRMnjyZU6dO0aZNm3KP8ddff7F9+3YiIiLQ6XTcc889tGvXDoCBAwdy3333AbB48WI2bNjApEmTCA0N5e6772bIkCGljqXVapkzZw5ffPEFzZo144UXXuDrr79m8uTJhni/++471qxZw4oVK1i4cGGlbkOFVegDBw7k/fff5/vvv6d79+588cUXpKWlMW/ePPbt21epg9e6whzw6QwWtjfdNDWnEHd7aSovhKg+t9PcSojqZm1tzd9//214/9dff2FlZWXEiOqQCwcg8jlo0R8GvAaAXq8wZ2MUG4/FM3tAIE9J0i5Eoyflev1xpbk8wI4dOxgxYgTff/8999xzD6NHj+b06dM3rN0+cuQIAwYMwNraGjs7O0JDQw3rTp8+zQMPPEBYWBhbt2696Qwt58+fx8/Pj2bNmgFwzz33cOTIEcP6QYMGAepYNAkJCZW+xpvOV2JjY8PIkSMZOXIkGRkZ7Nixg2XLltG7d+9Kn6TWFObddBo4gPxCHTnaYunjLoQQosF66aWXmDVrFh4eHgCkpKSwePFiI0dVB2TEwbcPqgPNjV0BpmZq0r4pivVH45nVvxWzBkjSLoQQt+SOCTesHa8pAwYM4O233+bvv/9Gq9Xi4ODAihUr2LBhA46OjsyZMwetVnvDY1yZieV6c+bM4ZNPPqFNmzZs2rSJQ4cO3fA4N2ulbm5uDoCJiQk6ne6G216rSp3WnZycuP/++1m9enVVdqtFCjTtc9Otrszh7i6JuxBCiAaqY8eOfP/998yfP5958+axffv2UiPNN0qFufDNBNAVwoRvDFO4vREZzboj8TzVvxWzBwYaN0YhhBBVZmtrS/fu3XnppZcYMmQIubm5WFtbY29vT2pqKnv37r3h/t26deOnn36ioKCAnJwcfv75Z8O63Nxc3N3dKSoqYuvWraXOmZubW+ZYzZs3JyEhgQsXLgCwefNmunW78VTllVHHR5urKg00uXn/9pSSxN3NXganE0JUn6o0dxKipq1Zs4b8/HwCAwNp3bo1eXl5rFmzxthhGY+iQMR0uPSXWtPuptaqJ2UVsPpADPd382e21LQLIa4h5Xr9MmLECE6dOsWQIUNo06YNbdu2Zfjw4bz00ks3nSGgXbt2DBs2jFGjRvHUU0+Vmj511qxZjBs3jkceeYTmzZsblg8bNozPP/+c0aNHExsba1huaWnJW2+9xaxZswgLC0Oj0TBhwu23QrhpU/l6xdwaLO1vullqdkniLjXuQohqdPToUXx8fIwdhhCAOgruxIkTDe8dHR1Zv359qWWNyt5FcHIzDHwdWg00LP76YCw6RWH63S0qbCYphGicpFyvXwYOHMg///xjGIj17bffLne7a6c33717t+H19OnTmT59epntH3jgAR544IEyy7t27WroV3/9+Xr27ElERESZfa49X4cOHao01XrDqnG3tKvUZqk56nD8krgLIarTyJEjjR2CEAZ6vb5UPzudTkdRUZERIzKi6K3w80LoeD/c9aRhcWGxnq8PxXJ3oDsBrjcf2FYI0bhIuS7qkoZV427lVKnNrvRxd5V53IUQQjRQvXv3ZtasWYbmed988w19+/Y1clRGkPQ3bHoMfLuq87VfU6v+w9+XSMnW8tBdTY0XnxBCCFEJDStxr8Q0cKAm7o7W5liamdZwQEIIIYRxPP/883z77besXbsWRVEICgoiJSXF2GHVrtw0WHu/2o1u/BowLz0d3uoDMQS42hDSyt1IAQohhBCVU6NN5ffu3cvgwYMZOHAgy5YtK7M+Ozubxx9/nJEjRzJ8+HA2btxY6X1vR0q2FjepbRdCVLNPP/3U2CEIYWBiYsIdd9yBn58ff/31FwcOHKBFixbGDqv26Ipg3UOQnQT3fw0O3qVWn0zM4nDMZSbdGYCJifRtF0KUJeV65d1sCjRRNeXdzxqrcdfpdCxYsICVK1fi6enJ2LFjCQ0NpWXLloZt1qxZQ4sWLVi6dCnp6ekMGTKEsLAwTE1Nb7rv7UjN0Ur/diFEtQsPDzd2CEJw/vx5IiMjiYyMxMnJiWHDhgFUaQCcBuH7F+HCPrhnGfh1LbP6y99jsDI3YVxXfyMEJ4SoD6RcrxwrKyvS0tJwdXWVQT6rgaIopKWlYWVVupVYjSXuUVFRBAQE4O+vFojDhw9n165dpZJvjUZDbm4uiqKQm5uLo6MjZmZm/Pnnnzfd93ak5hTS1sehWo4lhBBXaDQa+cVZGN3QoUMJDg5m6dKlBAQEAPDFF18YN6jadng5HPkces2CTuPLrM7MK+K74wnc09kXRxtzIwQohKgPpFyvHD8/P+Lj4+tMd6yioiLMzev3/+1WVlb4+fmVWlZjiXtSUhJeXl6G956enkRFRZXaZuLEiUyfPp0+ffqQm5vL4sWLMTExqdS+5dFqtURHR988tsw8OriZVmrbuqKgoKBexVsb5J6UJfekrNq+J/Xh/svnpLT6cj+CgoIqtd2SJUuIjIzkoYceok+fPgwfPrxxffGM2afWtrcaBP3nlbvJ+qNxFBTpmXRn09qNTQghGiBzc3OaNWtm7DAMoqOjK11m1ic1lriX9yXh+qYT+/btIygoiNWrVxMbG8uUKVMIDg6u1L7lsbS0vOkfqaBIR17ROVoHeBMUVD01+LWhoX4Ab4fck7LknpRV2/ekPtx/+ZyU1tDux8CBAxk4cCB5eXns3LmTL774grS0NObNm8fAgQPp3bu3sUOsOZdj4NtJ4NwMxiwHk7KD0Or1Cl/+foFuTZ2l9Z0QQoh6o8YSdy8vLy5dumR4n5SUhIeHR6ltNm3aRHh4OBqNhoCAAPz8/Dh37lyl9r1VKdnqVHAyOJ0QorqNGDHC2CEIYWBjY8PIkSMZOXIkGRkZ7Nixg2XLlt00cd+7dy8LFy5Er9czbty4Mn08ly9fztatWwF1PJuzZ89y4MAB0tPTmT17tmG7uLg4nnrqKSZPnsySJUtYt24dLi4uADzzzDOEhIRU8xUD+z+A4gKY8A1YOZa7yS+nU7iQlsdzg1pX//mFEA2KlOuiLqmxxL1Dhw7ExMQQFxeHp6cnkZGRvPfee6W28fb25sCBAwQHB5Oamsr58+fx8/PDwcHhpvveqitzuMvgdEKI6nYlmRGirnFycuL+++/n/vvvv+F2lRlYdtq0aUybNg2A3bt388UXX+Dk5ISTkxObN282HKdv374MHDjQsN/kyZOZOnVqDVzdNS5fAPfW4FZxi7rVv8Xgbm/J4HZeFW4jhBAg5bqoW2psOjgzMzNeffVVpk2bxrBhwxg6dCitWrVi7dq1rF27FoAZM2Zw/PhxwsLCmDx5Ms899xwuLi4V7lsdUnMKAUnchRDVLywszNghCHFbrh1Y1sLCwjA4bEUiIyPLrZE6cOAA/v7++Pr61mS4ZWVfBHufCldfSMtlz78pPNC9CRZmNTojrhCiAZByXdQlNVbjDhASElKmKdyECRMMrz09PVmxYkWl960Ohhp3e0nchRDVa9u2bcYOQYjbUpXBYfPz8/n111955ZVXyqwrL6Ffs2YNERERtG/fnjlz5uDoWH5T9isqO+DstQIz4sl0CCKpgv0+O5yGCdDNperHrm71ZVDE2iL3oyy5J2XV9j3Ztm1bnf8byOekrPpyT6o6vk6NJu51UWpJH3dXW+njLoQQQlyrKoPD/vzzz3Tp0gUnJ6dSywsLC9m9ezfPPvusYdmECROYMWMGGo2GDz74gLfffpu33nrrhrFUZsDZ0ifOg8IsXALa4VLOfvmFOnZ+u5MhHbzp3bVD5Y9bQxraoIi3S+5HWXJPyjLGPanrfwP5nJTVUO9Jo2snlpqjxcHKDCvzsiPNCiGEEI1ZVQaHjYyMZPjw4WWW7927l3bt2uHm5mZY5ubmhqmpKSYmJowbN44TJ05Uf/DZF9XnCprKb/kzgayCYh7u2bT6zy2EEELUsEaXuKfkaKWZvBCiRjSqubJFg3TtwLKFhYVERkYSGhpaZrvs7GwOHz5M//79y6wrL6FPTk42vN65c2e1jVtTSlai+uzgXWaVoiis+u0Cbbzs6dbUufrPLYRokKRcF3VJo0vcU7MLZWA6IUSNWLZsmbFDEOK2VGZgWYCffvqJXr16YWNjU2r//Px8fvvtNwYNGlRq+aJFiwgLCyMsLIzff/+duXPnVn/wN6hxP3rhMicvZvHwXU0rbPovhBDXk3Jd1CWNr497jpYgbwdjhyGEaIAee+yxMnNeC1Hf3GxgWYB7772Xe++9t8y+1tbWHDx4sMzyRYsWVW+Q5clKUJ/LqXFffeAC9lZmjLqj4hHnhRDielKui7qk0dW4p+RocbOTgemEEEKIBiXrIlg6gKV9qcXJWQVsP3GR+4L9sbFodPUVQgghGohGlbgXFOnILijGXfq4CyGEEA1LdiLYl61tX3sojmK9wqQ7A4wQlBBCCFE9GlXinpZbCCB93IUQNWLLli3GDkGIxisrERxKN4Uv0un5+tAFQgLdaepma6TAhBD1lZTroi5pVIl7Sskc7pK4CyFqQteuXY0dghCNV9bFMon7j38nkZSl5eG7pLZdCFF1Uq6LuqRRJe6pVxJ3aSovhKgBvr6+xg5BiMZJVww5SWWayq86EEMTFxtCAsufi14IIW5EynVRlzSuxD3nSo27DE4nhBBCNBi5yaDoStW4n7qUxaHz6Tx4ZxNMTWQKOCGEEPVbI03cpcZdCCGEaDCySuZwvyZxX33gApZmJtwX7G+koIQQQojq08gS90LsrcywMjc1dihCiAbo0UcfNXYIQjRO2Ynqc0lT+cz8Ir47lsDoO3xxspFWdkKIWyPluqhLGlXinpKtxV1q24UQNWTZsmXGDkGIxum6GveNR+PJL9IxqacMSieEuHVSrou6pHEl7jlaaSYvhKgxMvqsEEaSlQAm5mDjhl6v8OXvF+ga4Ex7X0djRyaEqMekXBd1SaNK3FNztLjZS5M5IUTNOHbsmLFDEKJxyr6oNpM3MeHXM6mcT83lIaltF0LcJinXRV3SuBL3bKlxF0IIIRqcrERwUPu3f3kgBjc7S4a2977JTkIIIUT90WgSd22xjqyCYknchRA1xttbEgUhjCIrERx8iEvPY9epZB7o7o+FWaP5iiOEqCFSrou6pNGUamk5hQC420viLoSoGYmJicYOQYjGR1FKmsr7sO5IHCYaDQ/0kGbyQojbJ+W6qEsaTeKeki1zuAshatb8+fONHYIQjU9BJhTlgYM351Jyaepqg5ejlbGjEkI0AFKui7qk0STuqTlXEncZnE4IUTNee+01Y4cgROOTfXUquNQcLa7yA70QoppIuS7qkkaYuEuBLoQQQjQYWQnqs70PabmF8gO9EEKIBqkRJe7Sx10IIYRocLKu1Lh7k5ajxcVWEnchhBANj5mxA6gtKdla7C3NsDI3NXYoQogG6siRI8YOQYjbtnfvXhYuXIher2fcuHGEh4eXWr98+XK2bt0KgE6n4+zZsxw4cAAnJydCQ0OxtbXFxMQEU1NTNm3aBEBGRgazZ88mISEBX19f3n//fRwdHasn4JKm8sU2nlzO+xtXW/mBXghRPaRcF3VJI6px1+Imte1CCCFEhXQ6HQsWLGD58uVERkaybds2zpw5U2qbadOmsXnzZjZv3swzzzxDt27dcHJyMqxftWoVmzdvNiTtAMuWLaNnz578+OOP9OzZk2XLllVf0FkJYONGeqEGkLFshBBCNEyNJnFPydZKYS6EqFHBwcHGDkGI2xIVFUVAQAD+/v5YWFgwfPhwdu3aVeH2kZGRjBgx4qbH3bVrF6NHjwZg9OjR7Ny5s7pCVpvKO3gbpn2VwemEENVFynVRlzSapvKpOVoCPe2NHYYQQghRZyUlJeHl5WV47+npSVRUVLnb5ufn8+uvv/LKK6+UWj516lQ0Gg3jx49n/PjxAKSlpeHh4QGAh4cH6enpN41Fq9USHR190+2apZynyNqd4ydPA5CTepHo6Iyb7mdsBQUFlbq+xkLuR1lyT8oyxj2p638D+ZyUVV/uSVBQUJW2b0SJeyF3tZBf4YUQQoiKKIpSZplGoyl3259//pkuXbqUaia/du1aPD09SUtLY8qUKTRv3pxu3brdUiyWlpaV+1KzNQ2rFr2wdfUCLtGlXSAtPexu6Zy1KTo6uspf2hoyuR9lyT0pyxj3pK7/DeRzUlZDvSeNoql8YbGezPwimQpOCFGj5s2bZ+wQhLgtXl5eXLp0yfA+KSnJUFN+vcjISIYPH15qmaenJwCurq4MHDjQUFvv6upKcnIyAMnJybi4uFRPwMVayEsDBx9DU3npFieEqC5Srou6pFEk7mm56hzuMhWcEKImzZ8/39ghCHFbOnToQExMDHFxcRQWFhIZGUloaGiZ7bKzszl8+DD9+/c3LMvLyyMnJ8fwev/+/bRq1QqA0NBQIiIiAIiIiCi1320pGVEee2/ScrWYmWhwsDKvnmMLIRo9KddFXdIomsqnZKuJu/wKL8T/t3f/MVXd9x/HXxcQLL9KBfnl+LovGU0IcV1tyWLNSnKddh0wCIw2zCWtkbFubsaRratanXPtbGq2zJgtmbXZ+ocj65dZUW6bFZmrpqsr7ZqwZHSLzVwQEeSicC8/7sXL/f5x4VZ6Qady7jmc+3wkBu6558L7fjzw5n0+v2Ck/Px8Xbx40ewwgNuWkJCg3bt3q6GhQYFAQLW1tSoqKlJzc7Mkqb6+XpLU3t6utWvXKjk5Ofxat9utLVu2SAqtTl9RUaGHH35YktTY2Kht27appaVFeXl5OnDgwMIEHN7DPV/uf/t1T0qi4uLmHtoPALeKvA4riYnCfdA7XbjT4w7AQH19fWaHANyxsrIylZWVzTo2U7DPqKmpUU1NzaxjBQUFOn78+Jxf85577tErr7yysIFKoa3gJCk9X4PeUWWmcIMewMIhr8NKYmKo/KAnNO9tOXPcAQCwj08MlWctGwCAXcVE4X55psedhA7AQKtXrzY7BCC2jPRJS5KlpXfL7fUrkylxABYQeR1WYuhQ+dOnT+v555/X1NSU6urq1NjYOOv5w4cP68SJE5JC8+E++ugjvfPOO8rIyJDT6VRKSori4uIUHx+vo0eP3nYcg16fUhLjdVdi/B29HwC4kffff9/sEIDYMtIrpedLDofcXp8yU7hBD2DhkNdhJYb1uAcCAe3du1eHDx+Wy+VSW1ubzp07N+uchoYGtba2qrW1VU1NTSotLZ21H+wrr7yi1tbWOyrapdAe7qwoD8Bon7w5CcBgnj4pLU8TkwGN+gP0uANYUOR1WIlhhXtXV5dWrlypgoICJSYmqry8XB0dHfOe73K5VFFRYUgslz0TDJMHYLiXXnrJ7BCA2DLSF1pRfpQ93AEsPPI6rMSwofL9/f3Kzc0NP87JyVFXV9ec546Pj+vMmTPatWvXrOObN2+Ww+HQ448/rscff/ym39Pn86m7uzvieK/bo/+5e8mczy0WExMTizp+I9AmkWiTSNFuk8XQ/lwnsy2W9iguLjY7BGuZmgr1uKfnyz29lg1D5QEAdmVY4R4MBiOOORxz76166tQprV69etYw+ebmZuXk5MjtdmvTpk0qLCxUaWnpDb9nUlLSnH/YeP6vR/+bn7Wo/+jp7u5e1PEbgTaJRJtEinabLIb25zqZjfZYpMYGpalJKS1fbm+ox52h8gAAuzJsqHxubq4uXboUftzf36/s7Ow5z3W5XCovL591LCcnR5KUmZmp9evXz9tbfzOTgSldHZtkqDwAw/X29podAhA7Ri6GPqbnaZAedwAGIK/DSgwr3FetWqXz58+rp6dHfr9fLpdLTqcz4jyPx6POzk6tW7cufGxsbExerzf8+dtvv62ioqLbimPmLjyFOwCjsfosEEUze7hfN8edHncAC4m8DisxbKh8QkKCdu/erYaGBgUCAdXW1qqoqEjNzc2SpPr6eklSe3u71q5dq+Tk5PBr3W63tmzZIim0On1FRYUefvjh24pj5i48q8oDMNpXvvKVOacJATDAyHRPWFq+3N4rWrokTsls+wpgAZHXYSWG7uNeVlamsrKyWcdmCvYZNTU1qqmpmXWsoKBAx48fX5AYLntChTs97gAA2MhIn+SIl1Kz5fb2KzMlad61dAAAWOwMGypvFZdnetwp3AEAsA9Pn5SaI8XFa3DUz1ZwAABbs33hPjNUPiuNhA7AWL/+9a/NDgGIHSO9Unq+JMnt9SmTG/QAFhh5HVZi/8Ld41dyYrySEw2dFQAAamxsNDsEIHaM9EnpeZKkoVG/MlO4QQ9gYZHXYSX2L9y9PhamAxAVzK8FosjTJ6XlKxgMyu310+MOYMGR12Elti/cL3t8LEwHAICd+DySb0RKz5fHd03+wBQ97gAAW7N94T7o9bFgDQAAdjJy3R7uXvZwBwDYX4wU7vS4AzBeRUWF2SEAscFzMfQxLU/u6UVoGSoPYKGR12Eltl6xbTIwpStjkxTuAKLixIkTZocA3LHTp0/r+eef19TUlOrq6iIWZzp8+HD4Wg8EAvroo4/0zjvvaHx8XE8//bQGBwcVFxenxx57TE888YQk6eDBg3r11Ve1bNkySVJTU5PKyspuP8jretwH+6Z73BkqD2CBkddhJbYu3IdGQ8k8i8XpAERBZWUlSR6LWiAQ0N69e/Wb3/xGOTk5+upXvyqn06nPfOYz4XMaGhrU0NAgSfrTn/6k3/72t8rIyJDf79czzzyjkpISeb1e1dbWau3ateHXPvnkk9q8efPCBDrSG/qYni/3uQFJ4iY9gAVHXoeV2Hqo/GVPaPjccpI5gChoa2szOwTgjnR1dWnlypUqKChQYmKiysvL1dHRMe/5LpcrPJQ0OztbJSUlkqTU1FQVFhaqv7/fmEA9fdLSDGnJXeE57svocQewwMjrsBJ7F+7T896Wp5HMAQC4mf7+fuXm5oYf5+TkzFt8j4+P68yZM9qwYUPEcxcuXFB3d7fuu+++8LEjR46osrJS27dv1/Dw8J0FOtInpedLktxen9KXJigxwdZ/0gAAYpyth8oPTve4M3wOAICbCwaDEcfm28f41KlTWr16tTIyMmYdHx0d1datW7Vjxw6lpqZKkurr6/Xtb39bDodDBw4c0AsvvKB9+/bdMBafz6fu7u45n/v0wEcKJGWop7tb/+4bVFqiY95zrWpiYmLRxWwk2iMSbRLJjDax+v8B10mkxdImxcXFt3S+vQv36eFzFO4AomGuogdYTHJzc3Xp0qXw4/7+fmVnZ895rsvlUnl5+axjk5OT2rp1qyorK2f1xGdlZYU/r6ur01NPPXXTWJKSkub/o8Z1Rfp0qYqLizV5Zlh59yTe8h9AZuvu7l50MRuJ9ohEm0SKdpsshrzOdRLJrm1i63Flg16f7loSr5QkW9+fAGARhw4dMjsE4I6sWrVK58+fV09Pj/x+v1wul5xOZ8R5Ho9HnZ2dWrduXfhYMBjUzp07VVhYqE2bNs06f2BgIPz5yZMnVVRUdPtBBiYl78DHQ+VHfezhDsAQ5HVYie0L9yzmtwOIkm9+85tmhwDckYSEBO3evVsNDQ368pe/rEcffVRFRUVqbm5Wc3Nz+Lz29natXbtWycnJ4WPvv/++WltbdfbsWVVVVamqqkpvvfWWJGn//v2qrKxUZWWlzp49q+3bt99+kN5+ScHr5rj72cMdgCHI67ASW3dFD3p9rCgPAMAtKCsri9hjvb6+ftbjmpoa1dTUzDr24IMP6p///OecX3P//v0LF+DIxdDHtHwFpoIaGvMrixXlAQA2Z+se98seH/PbAQCwk5nCPT1PV8b8CgZFjzsAwPZsXbgPev3KSiOZA4iO48ePmx0CYH+evtDH9BXhPdyZ4w7ACOR1WIltC/drgSldGfPT4w4gah544AGzQwDsb6RXik+S7rpHbm9o29fMFHI9gIVHXoeV2LZwHxoNDZ9bzl14AFGyYsUKs0MA7G+kT0rPkxwODY7ObPtKrgew8MjrsBLbFu6Xp+/CL2eoPAAA9uHpk9JCK8oPzfS4M7oOAGBz9i3cPaFkzlB5AABsZOTidXu4+xXnkDLuWmJyUAAAGMu2hfugd2b4HIU7gOj4xje+YXYIgL0Fg9OFe56kUK5flpKouDiHyYEBsCPyOqzExoX7dI87Q+UBRMmhQ4fMDgGwt/ErUsAXHirv9vpYmA6AYcjrsBL7Fu4en5YuiVNKYrzZoQCIEaw+CxgsvIf7x0Pl2QoOgFHI67AS+xbuXp+yUpPkcDB8DkB0/O1vfzM7BMDePlm4e30sTAfAMOR1WImNC3c/K8oDAGAnnunCPS00x93t9SszhR53AID92bZwv+zxsTAdgKjKy8szOwTA3kb6JDmktFxNTAbk8V1jD3cAhiGvw0psW7jPDJUHgGi5ePGi2SEA9jbSK6VmS/FLNDQa2j2GofIAjEJeh5XYsnC/FpjS0Jhfy7kLDyCK9uzZY3YIgL15+sLD5GcK92UMlQdgEPI6rMSWhfvQmF/BIFvBAYiuH//4x2aHANjbSF94Ybrwtq/cpAdgEPI6rMSWhfugJ3QXnqHyAADYiOfidSvKTw+VZx93AEAMsGfhPn0XnlXlAQCwiclxafzKxyvKj4ZyPfu4AwBigS0L98uemeFzFO4Aoue9994zOwTAviL2cPcrMSFOqUkJJgYFwM7I67ASWxbuzHsDAMBmPH2hj+E57n5lpSTK4XCYGBQAANFhaOF++vRpPfLII1q/fr0OHToU8fzhw4dVVVWlqqoqVVRUqLi4WFevXv2vXnsjg16fkrgLDyDKHnzwQbNDAO6YEbn76tWr2rRpkzZs2KBNmzZpeHj41gOb6XFPm+5xH/WxFRwAQ5HXYSWGFe6BQEB79+7V4cOH5XK51NbWpnPnzs06p6GhQa2trWptbVVTU5NKS0uVkZHxX732Rga9fmWlJnEXHgCAW2BU7j506JDWrFmjN998U2vWrLnlG/KSrhsqPz3H3etnfjsAIGYYVrh3dXVp5cqVKigoUGJiosrLy9XR0THv+S6XSxUVFbf12k8a9PrYCg4AgFtkVO7u6OhQdXW1JKm6ulonT5689eA8fVJimpSUJklye33s4Q4AiBmGjSXv7+9Xbm5u+HFOTo66urrmPHd8fFxnzpzRrl27bvm11/P5fOru7taOh9IlSd3d3XfyFizHbu9nIdAmkWiTSNFqk3/84x+Lpv0XS5zRshjaIyEhQUVFRYZ+D6Nyt9vtVnZ2tiQpOztbQ0NDN41lJqeHffrJ0L/pYy9Xh4bML4b/u/ks5tiNQHtEok0iRbNNFkteXwwxRttiaJNbzeuGFe7BYDDi2HxD10+dOqXVq1crIyPjll97vc997nO3FCMAAPiYGbl7PuR0AAA+ZthQ+dzcXF26dCn8uL+/P3y3/ZNcLpfKy8tv67UAAGBhGJW7MzMzNTAwIEkaGBjQsmXLjAgfAADbMqxwX7Vqlc6fP6+enh75/X65XC45nc6I8zwejzo7O7Vu3bpbfi0AAFg4RuVup9OpY8eOSZKOHTs263UAAODmDBsqn5CQoN27d6uhoUGBQEC1tbUqKipSc3OzJKm+vl6S1N7errVr1yo5OfmmrwUAAMYxKnc3NjZq27ZtamlpUV5eng4cOBD9NwcAwCLmCM41KQ0AAAAAAFiCYUPlAQAAAADAnaNwBwAAAADAwgyb446F4XQ6lZKSori4OMXHx+vo0aNmhxR127dv15///GdlZmaqra1NknT16lV973vfU29vr1asWKFf/OIXuvvuu02ONHrmapODBw/q1VdfDa/W3NTUpLKyMjPDjKq+vj49/fTTGhwcVFxcnB577DE98cQTMX2tzNcmsXyt+Hw+bdy4UX6/X4FAQI888oi2bt0a09cJoou8Tl6fC3k9Enk9Enk9Uizldea4W5zT6VRLS0tMb53T2dmp5ORk/fCHPwwnsxdffFEZGRlqbGzUoUOHNDw8rB/84AcmRxo9c7XJwYMHlZycrM2bN5scnTkGBgZ0+fJllZSUyOv1qra2Vr/85S919OjRmL1W5muTN954I2avlWAwqLGxMaWkpGhyclJf+9rXtHPnTr355psxe50gusjr5PW5kNcjkdcjkdcjxVJeZ6g8LK+0tDTiDllHR4eqq6slSdXV1Tp58qQJkZlnrjaJddnZ2SopKZEkpaamqrCwUP39/TF9rczXJrHM4XAoJSVFknTt2jVdu3ZNDocjpq8TINrI65HI65HI65HI65FiKa9TuC8CmzdvVk1NjX7/+9+bHYpluN1uZWdnSwr9EhsaGjI5Ims4cuSIKisrtX37dg0PD5sdjmkuXLig7u5u3XfffVwr065vEym2r5VAIKCqqio99NBDeuihh7hOEHXk9Uj8DM4tln9XX4+8Hom8/rFYyesU7hbX3Nys1157TS+99JKOHDmizs5Os0OCRdXX16u9vV2tra3Kzs7WCy+8YHZIphgdHdXWrVu1Y8cOpaammh2OJXyyTWL9WomPj1dra6veeustdXV16V//+pfZISGGkNfx34r139UzyOuRyOuzxUpep3C3uJycHElSZmam1q9fr66uLpMjsobMzEwNDAxICs33ieW5gjOysrIUHx+vuLg41dXV6e9//7vZIUXd5OSktm7dqsrKSm3YsEES18pcbcK1EpKenq7Pf/7zOnPmTMxfJ4ge8vrc+BmMxO9q8vpcyOvzs3tep3C3sLGxMXm93vDnb7/9toqKikyOyhqcTqeOHTsmSTp27JjWrVtnbkAWMPPLSZJOnjwZc9dKMBjUzp07VVhYqE2bNoWPx/K1Ml+bxPK1MjQ0pJGREUnSxMSE/vKXv6iwsDCmrxNED3l9fvwMRorl39USeX0u5PVIsZTXWVXewnp6erRlyxZJobkbFRUV+ta3vmVyVNHX1NSkd999V1euXFFmZqa++93v6otf/KK2bdumvr4+5eXl6cCBA8rIyDA71KiZq03effddffjhh5KkFStWaO/eveG5PbHgvffe08aNG3XvvfcqLi50T7KpqUmf/exnY/Zama9N2traYvZa+fDDD/XMM88oEAgoGAzqS1/6kr7zne/oypUrMXudIHrI6yHk9Ujk9Ujk9Ujk9UixlNcp3AEAAAAAsDCGygMAAAAAYGEU7gAAAAAAWBiFOwAAAAAAFkbhDgAAAACAhVG4AwAAAABgYQlmBwDAXMXFxbr33nvDj8vLy9XY2LggX/vChQt66qmn1NbWtiBfDwAA3Bh5HbAnCncgxi1dulStra1mhwEAABYAeR2wJwp3AHNyOp169NFH9de//lWS9LOf/UwrV65Ub2+vduzYoaGhIS1btkz79u1Tfn6+BgcH9aMf/Ug9PT2SpD179ig7O1uBQEDPPvusPvjgA+Xk5OhXv/qVli5dauZbAwAg5pDXgcWNOe5AjJuYmFBVVVX43+uvvx5+LjU1VS0tLfr617+un/70p5Kkn/zkJ6qurtaJEydUWVmp5557TpL03HPPqbS0VMePH9drr72moqIiSdJ//vMfbdy4US6XS2lpafrjH/8Y/TcJAECMIK8D9kSPOxDjbjSkrqKiQlJofty+ffskSR988IEOHjwoSaqqqtL+/fslSWfPntWLL74oSYqPj1daWpqGh4f1qU99SsXFxZKkkpIS9fb2Gvp+AACIZeR1wJ7ocQdwRxwOxw2fT0xMDH8eHx+vQCBgdEgAAOA2kdcBa6JwBzCvN954Q5L0+uuv6/7775ck3X///XK5XJKkEydO6IEHHpAkrVmzRr/73e8kSYFAQF6v14SIAQDAfMjrwOLFUHkgxs3MhZvxhS98Qd///vclSX6/X3V1dZqamtLPf/5zSdKzzz6rHTt26OWXXw4vYiNJO3fu1K5du/SHP/xBcXFx2rNnj5YvXx79NwQAQAwjrwP25AgGg0GzgwBgPU6nUy0tLVq2bJnZoQAAgDtEXgcWN4bKAwAAAABgYfS4AwAAAABgYfS4AwAAAABgYRTuAAAAAABYGIU7AAAAAAAWRuEOAAAAAICFUbgDAAAAAGBh/w8CuJLXlQGIBAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 1008x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df = pd.DataFrame(training.history)\n",
    "best_auc = df.val_AUC.max()\n",
    "best_acc = df.val_accuracy.max()\n",
    "\n",
    "fig, axes = plt.subplots(ncols=2, figsize=(14,4))\n",
    "df.index = df.index.to_series().add(1)\n",
    "df[['AUC', 'val_AUC']].plot(ax=axes[0], \n",
    "                            title=f'AUC | Best: {best_auc:.4f}', \n",
    "                            legend=False, \n",
    "                            xlim=(1, 33),\n",
    "                            ylim=(.7, .95))\n",
    "\n",
    "axes[0].axvline(df.val_AUC.idxmax(), ls='--', lw=1, c='k')\n",
    "df[['accuracy', 'val_accuracy']].plot(ax=axes[1], \n",
    "                                              title=f'Accuracy | Best: {best_acc:.2%}', \n",
    "                                              legend=False, \n",
    "                                              xlim=(1, 33),\n",
    "                                      ylim=(.7, .9))\n",
    "axes[1].axvline(df.val_accuracy.idxmax(), ls='--', lw=1, c='k')\n",
    "axes[0].set_xlabel('Epoch')\n",
    "axes[0].set_ylabel('AUC')\n",
    "axes[1].set_xlabel('Epoch')\n",
    "axes[1].set_ylabel('Accuracy')\n",
    "fig.suptitle('Sentiment Analysis - Pretrained Vectors', fontsize=14)\n",
    "fig.legend(['Train', 'Validation'], loc='center right')\n",
    "\n",
    "sns.despine()\n",
    "fig.tight_layout()\n",
    "fig.subplots_adjust(top=.9)\n",
    "fig.savefig(results_path / 'imdb_pretrained', dpi=300);"
   ]
  }
 ],
 "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": "3.8.8"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": true,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": true
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}