research/3_feat_select_h5_daily.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "8f2b103a-1f09-486a-92f6-47c2eeba97c0",
   "metadata": {},
   "source": [
    "# Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "77e57da0-b4e5-41b3-93b4-96eac24b8d01",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "from catboost import CatBoostRegressor, EFstrType, Pool, sum_models\n",
    "import gc\n",
    "import h5py\n",
    "import joblib\n",
    "import json\n",
    "import matplotlib.pyplot as plt\n",
    "import mlflow\n",
    "import numpy as np\n",
    "import optuna\n",
    "import os\n",
    "import pandas as pd\n",
    "import pickle\n",
    "import polars as pl\n",
    "import sys\n",
    "\n",
    "from sklearn.metrics import mean_absolute_error\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\",category=DeprecationWarning)\n",
    "\n",
    "from tqdm import tqdm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "1959a435-aecb-4c29-9e0e-86f48847fae9",
   "metadata": {},
   "outputs": [],
   "source": [
    "sys.path.append('../src/')\n",
    "import h5_utils\n",
    "from train import drop_target_nan"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "17c9699d-8caf-4b7c-8987-e7b7f764da02",
   "metadata": {},
   "source": [
    "# Configs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "86233d9e-c714-4c4c-aa90-81a8d2dd9e1e",
   "metadata": {},
   "outputs": [],
   "source": [
    "settings_path = '../configs/settings.json'\n",
    "with open(settings_path, 'r') as f:\n",
    "    config = json.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "1ec2ef4f-01ae-482f-a987-77b731b29afd",
   "metadata": {},
   "outputs": [],
   "source": [
    "folder_daily_h5 = f\"../{config['RAW_DATA_DIR']}/daily/\"\n",
    "\n",
    "metadata_filename = 'metadata.h5'\n",
    "metadata_filepath = f\"../{config['RAW_DATA_DIR']}/{metadata_filename}\"\n",
    "\n",
    "optuna_all_feat_filename = 'optuna_all_sliding.pkl'\n",
    "optuna_all_feat_filenpath = f\"../{config['RAW_DATA_DIR']}/{optuna_all_feat_filename}\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "044abed4-f0d0-453f-93ab-c95726f7a9de",
   "metadata": {},
   "outputs": [],
   "source": [
    "DAY_TEST_START = 400 # train size in day terms"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "73c6a905-2d46-45c4-9307-a4b5d0ae127c",
   "metadata": {},
   "source": [
    "# Daily load trick"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "6cc20fe2-e330-43fb-87fe-b0ab472e49ae",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "481"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "date_ids = h5_utils.load_metadata(metadata_filepath)\n",
    "len(date_ids)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ab5993bb-1d80-4f4f-a5a0-02553193e938",
   "metadata": {},
   "source": [
    "# Validation splits"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "785c4a9e-06c1-4f48-9982-43eff490834d",
   "metadata": {},
   "outputs": [],
   "source": [
    "max_train_size = 70\n",
    "slide_step= max_train_size\n",
    "n_samples=len(date_ids)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0f46495f-fb39-4ed2-bc37-6bb59e90cc81",
   "metadata": {},
   "source": [
    "effective_train_size = max_train_size + gap\n",
    "n_splits = (n_samples - test_size) // (effective_train_size + test_size) + 1\n",
    "print(n_splits)\n",
    "tscv = TimeSeriesSplit(n_splits=n_splits,\n",
    "                max_train_size=max_train_size, \n",
    "                test_size=test_size, \n",
    "                gap=gap)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "18e06c3a-d3a8-4d3b-a62d-c4dc5fd570c6",
   "metadata": {},
   "outputs": [],
   "source": [
    "class SlidingWindowSplit:\n",
    "    def __init__(self, \n",
    "                 data, \n",
    "                 train_days, \n",
    "                 #test_days, \n",
    "                 slide_step=1):\n",
    "        self.data = data\n",
    "        self.train_days = train_days\n",
    "        self.slide_step = slide_step\n",
    "\n",
    "    def __iter__(self):\n",
    "        for i in range(self.train_days, len(self.data) + 1, self.slide_step):\n",
    "            yield np.arange(i - self.train_days, i)\n",
    "\n",
    "    def __len__(self):\n",
    "        return (len(self.data) - self.train_days + 1) // self.slide_step"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "5c4317c3-6ae6-4266-8cf9-0aca2a71d6ba",
   "metadata": {},
   "outputs": [],
   "source": [
    "batch_ids = SlidingWindowSplit(date_ids, \n",
    "                       max_train_size, \n",
    "                       slide_step)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5771cb67-a175-40e9-b85f-d32c1bb5dfc0",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "for i, train_index in enumerate(tqdm(batch_ids, \n",
    "                                    desc=\"Training and Validation\")):\n",
    "    \n",
    "    print(f\"Batch {i}:\")\n",
    "    print(f\"  Train: index={train_index}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "70452b9e-21b9-4fad-8527-4ccb96310825",
   "metadata": {},
   "source": [
    "# Intermediate optuna on Catboost for feature selection"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "710a4edf-393d-4baf-bd90-5e141f7c9574",
   "metadata": {},
   "source": [
    "## Trial"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "c09ff967-3976-42cf-af61-93e725d28ba0",
   "metadata": {},
   "outputs": [],
   "source": [
    "def params_catboost(trial):\n",
    "    \n",
    "    param = {\n",
    "        \"loss_function\" : \"MAE\",\n",
    "        \"eval_metric\" : \"MAE\",\n",
    "        \"iterations\" : 1000,  \n",
    "        #\"used_ram_limit\": \"14gb\",\n",
    "        #\"learning_rate\": trial.suggest_float(\"learning_rate\", \n",
    "        #                              0.2, 0.9), # will optimize it later, catboost has good inner mechanism of choice\n",
    "        \"depth\": trial.suggest_int(\"depth\", 5, 10),\n",
    "        \"l2_leaf_reg\": trial.suggest_int(\"l2_leaf_reg\", \n",
    "                                      3, 50),\n",
    "        \"bootstrap_type\": trial.suggest_categorical(\"bootstrap_type\", \n",
    "                             [\"Bayesian\", \"Bernoulli\", \"MVS\"])\n",
    "    }\n",
    "\n",
    "    if param[\"bootstrap_type\"] == \"Bayesian\":\n",
    "        param[\"bagging_temperature\"] = trial.suggest_float(\"bagging_temperature\", \n",
    "                                                 1, 20)\n",
    "        \n",
    "    elif param[\"bootstrap_type\"] == \"Bernoulli\":\n",
    "        param[\"subsample\"] = trial.suggest_float(\"subsample\", \n",
    "                                         0.1, 0.8)\n",
    "    return param"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "7ed66b73-c131-402b-a1c9-231c65be80a7",
   "metadata": {},
   "outputs": [],
   "source": [
    "def batch_learning(param, train_ids):\n",
    "    '''All the data with such number of features can't fit in memory so we will update model by batches learning'''\n",
    "    batches = SlidingWindowSplit(train_ids, \n",
    "                           max_train_size, \n",
    "                           slide_step)\n",
    "\n",
    "    model_prev = None\n",
    "\n",
    "    for batch_ind in tqdm(batches, desc=\"Batch training:\"):\n",
    "        batch_train_days = date_ids[batch_ind]\n",
    "        \n",
    "        # Load training data\n",
    "        train_data, train_labels = h5_utils.stacked_daily_data(batch_train_days,\n",
    "                                                              folder_daily_h5)\n",
    "\n",
    "        # Preprocess\n",
    "        ## Target handle NaNs\n",
    "        train_data, train_labels = drop_target_nan(train_data, \n",
    "                                                   train_labels)\n",
    "\n",
    "        batch = Pool(train_data, \n",
    "                     label=train_labels)\n",
    "\n",
    "        model_new = CatBoostRegressor(**param, thread_count=-1, random_seed=42)\n",
    "        if model_prev is None:\n",
    "            model_new.fit(X=batch, verbose=1)\n",
    "            model_prev = model_new\n",
    "        else:\n",
    "            # https://catboost.ai/en/docs/concepts/python-usages-examples#batch-training\n",
    "            batch.set_baseline(model_prev.predict(batch))\n",
    "            model_new.fit(X=batch, verbose=0)\n",
    "            model_prev = sum_models([model_prev,model_new])\n",
    "        \n",
    "    return model_prev"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "d75b15b2-a03d-4f89-81dc-c2dd8e4634e4",
   "metadata": {},
   "outputs": [],
   "source": [
    "def objective(trial):\n",
    "    # CV\n",
    "    train_ids = date_ids[:DAY_TEST_START]\n",
    "    test_ids = date_ids[DAY_TEST_START:]\n",
    "    \n",
    "    param = params_catboost(trial)\n",
    "    # Train\n",
    "    final_model = batch_learning(param, train_ids)\n",
    "\n",
    "    # Score\n",
    "    ## Load validation data\n",
    "    valid_data, valid_labels = h5_utils.stacked_daily_data(test_ids,\n",
    "                                                          folder_daily_h5)\n",
    "    ## NaN in target is not scored by the rules\n",
    "    valid_data, valid_labels = drop_target_nan(valid_data, \n",
    "                                               valid_labels)\n",
    "    valid_preds = final_model.predict(valid_data)\n",
    "    ## Calculate metric\n",
    "    mae = mean_absolute_error(valid_labels, valid_preds)\n",
    "    return mae"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "17555e8c-320b-4bd3-be38-ddcf831c7e80",
   "metadata": {},
   "source": [
    "## Search"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f3bc893f-f312-4408-b39d-81e689b2d27c",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Batch training::  50%|█████     | 2/4 [19:46<19:44, 592.18s/it]"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "hours = 10\n",
    "\n",
    "study = optuna.create_study(direction='minimize',\n",
    "                pruner=optuna.pruners.MedianPruner(n_warmup_steps=5))\n",
    "study.optimize(objective, \n",
    "               n_trials=120,\n",
    "               timeout = 60*60*hours\n",
    "              )\n",
    "print('Number of finished trials:', len(study.trials))\n",
    "print('Best trial:', study.best_trial.params)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "cf6be73f-7ba7-4166-968b-560b93634363",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best trial: score 0.6486779503958122, params {'depth': 8, 'l2_leaf_reg': 22, 'bootstrap_type': 'Bernoulli', 'subsample': 0.7294503489264281}\n"
     ]
    }
   ],
   "source": [
    "print(\"Best trial: score {}, params {}\".format(study.best_trial.value, \n",
    "\t\t\t\t\t       study.best_trial.params))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "4cf23b37-775e-4e71-b4b2-86a876e19df9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "5"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(study.trials)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ab5e8c47-b996-4446-b524-5cf1d0b6e090",
   "metadata": {},
   "source": [
    "## Store study (optional)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "80b44a60-18ed-4069-8c7d-83a5a45175c7",
   "metadata": {},
   "outputs": [],
   "source": [
    "joblib.dump(study, optuna_all_feat_filenpath)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "27132162-f84d-4b8d-9f05-0af074d62bc9",
   "metadata": {},
   "source": [
    "## Observe hyper importances"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "74441e8a-9d2d-4efd-bfa8-c78d2595fb1e",
   "metadata": {},
   "outputs": [],
   "source": [
    "study = joblib.load(optuna_all_feat_filenpath)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2e680a4c-d82f-4c1d-a4e8-f7ae3310e237",
   "metadata": {},
   "source": [
    "optuna.visualization.plot_param_importances(study)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "5cc19280-5c09-4b14-bed3-42aaab62dc79",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Get parameter importances\n",
    "param_importances = optuna.importance.get_param_importances(study)\n",
    "\n",
    "# Prepare data for visualization\n",
    "names = list(param_importances.keys())\n",
    "values = list(param_importances.values())\n",
    "\n",
    "# Create bar plot\n",
    "plt.figure(figsize=(10, 5))\n",
    "plt.barh(names, values, color='blue')\n",
    "plt.xlabel('Importance')\n",
    "plt.title('Parameter Importances')\n",
    "plt.gca().invert_yaxis()  # Invert y-axis to display the highest importance at the top\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "188cfa3f-bf9c-4dc4-bdb0-e174232ab6c8",
   "metadata": {},
   "source": [
    "# Selection"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2fd94c80-cb04-4796-a70b-460c26450406",
   "metadata": {},
   "source": [
    "300 features\n",
    "1) exprimentaly can be fited on kaggle as one batch\n",
    "2) can satify time limit to be retrained multiple times during test inference"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "623b74f1-bbd0-4850-958e-ee21d8758894",
   "metadata": {},
   "source": [
    "## Best model fit"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "ff1ea6f6-a5de-4c40-9046-e95e73eb8b02",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b64a46de-4a31-49d9-81e3-e63ef75f7d27",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "%%time\n",
    "data_train = date_ids[:DAY_TEST_START]\n",
    "best_model = batch_learning(study.best_trial.params,\n",
    "                            data_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d04f9a75-9354-4ec0-90b9-93ed8f33f726",
   "metadata": {},
   "source": [
    "## Store best on all features (optional)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "730e9038-89dc-4231-832b-31443fa52157",
   "metadata": {},
   "outputs": [],
   "source": [
    "best_model.save_model('Batched_all_feats_444',\n",
    "           format=\"cbm\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bc46495a-6a1e-446c-be3c-6bc8731a445b",
   "metadata": {},
   "source": [
    "## Catboost importance approach"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "507a2aa8-9c82-4849-af7d-66541415af4f",
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_importance_catboost2 = best_model.get_feature_importance(type=EFstrType.FeatureImportance, \n",
    "\t\t\t\t   prettified=True, \n",
    "\t\t\t\t   thread_count=-1, \n",
    "\t\t\t\t   verbose=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "5bf7c3a5-23ba-4e5d-96a6-b1f36fa159a2",
   "metadata": {},
   "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>Feature Id</th>\n",
       "      <th>Importances</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>417</td>\n",
       "      <td>45.573726</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>422</td>\n",
       "      <td>10.172564</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>440</td>\n",
       "      <td>5.515918</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>191</td>\n",
       "      <td>4.313303</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>442</td>\n",
       "      <td>3.461714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>438</th>\n",
       "      <td>181</td>\n",
       "      <td>0.001182</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>439</th>\n",
       "      <td>185</td>\n",
       "      <td>0.000999</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>440</th>\n",
       "      <td>184</td>\n",
       "      <td>0.000941</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>441</th>\n",
       "      <td>75</td>\n",
       "      <td>0.000112</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>442</th>\n",
       "      <td>374</td>\n",
       "      <td>0.000104</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>443 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    Feature Id  Importances\n",
       "0          417    45.573726\n",
       "1          422    10.172564\n",
       "2          440     5.515918\n",
       "3          191     4.313303\n",
       "4          442     3.461714\n",
       "..         ...          ...\n",
       "438        181     0.001182\n",
       "439        185     0.000999\n",
       "440        184     0.000941\n",
       "441         75     0.000112\n",
       "442        374     0.000104\n",
       "\n",
       "[443 rows x 2 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "feature_importance_catboost2"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2a1fd824-7673-4a55-8883-e5ef73aff67a",
   "metadata": {},
   "source": [
    "## Feature names from h5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "dcc1149c-b7c3-4c7f-94d8-52f452fe0c09",
   "metadata": {},
   "outputs": [],
   "source": [
    "date_id=date_ids[-1]\n",
    "filepath=f'{folder_daily_h5}/{date_id}.h5'\n",
    "with h5py.File(filepath, 'r') as f:\n",
    "    # Load the target column\n",
    "    target = f['data']['target'][:]\n",
    "\n",
    "    features_group = f['data']['features']\n",
    "    feature_names = list(features_group.keys())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "362e632f-bc6e-4cda-b0ca-fd22bda8478c",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['ask_auc_money',\n",
       " 'ask_money',\n",
       " 'ask_price',\n",
       " 'ask_price_diff_1',\n",
       " 'ask_price_diff_10']"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "feature_names[:5]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cb31c6a9-91bc-44b1-9b34-775901c8dea6",
   "metadata": {},
   "source": [
    "## Bar plot of top n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "20394561-2600-4ba5-b61d-14235d69d3cb",
   "metadata": {
    "scrolled": true
   },
   "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>Feature Id</th>\n",
       "      <th>Importances</th>\n",
       "      <th>Feature</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>417</td>\n",
       "      <td>45.573726</td>\n",
       "      <td>target_mock</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>422</td>\n",
       "      <td>10.172564</td>\n",
       "      <td>target_single</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>440</td>\n",
       "      <td>5.515918</td>\n",
       "      <td>wap_shift_n6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>191</td>\n",
       "      <td>4.313303</td>\n",
       "      <td>index_target_mock</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>442</td>\n",
       "      <td>3.461714</td>\n",
       "      <td>weight_tmp</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>393</td>\n",
       "      <td>1.865730</td>\n",
       "      <td>seconds_in_bucket</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>439</td>\n",
       "      <td>1.397131</td>\n",
       "      <td>wap_shift_n3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>145</td>\n",
       "      <td>0.751377</td>\n",
       "      <td>global_rolling18_mean_imb1_reference_price_wap</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>428</td>\n",
       "      <td>0.660389</td>\n",
       "      <td>wap</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>391</td>\n",
       "      <td>0.509950</td>\n",
       "      <td>scale_imbalance_size</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>416</td>\n",
       "      <td>0.508197</td>\n",
       "      <td>stock_id</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>146</td>\n",
       "      <td>0.481616</td>\n",
       "      <td>global_rolling48_mean_target_mock_shift6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>394</td>\n",
       "      <td>0.381346</td>\n",
       "      <td>seconds_in_bucket_group</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>437</td>\n",
       "      <td>0.374136</td>\n",
       "      <td>wap_shift_n1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>2</td>\n",
       "      <td>0.359632</td>\n",
       "      <td>ask_price</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Feature Id  Importances                                         Feature\n",
       "0         417    45.573726                                     target_mock\n",
       "1         422    10.172564                                   target_single\n",
       "2         440     5.515918                                    wap_shift_n6\n",
       "3         191     4.313303                               index_target_mock\n",
       "4         442     3.461714                                      weight_tmp\n",
       "5         393     1.865730                               seconds_in_bucket\n",
       "6         439     1.397131                                    wap_shift_n3\n",
       "7         145     0.751377  global_rolling18_mean_imb1_reference_price_wap\n",
       "8         428     0.660389                                             wap\n",
       "9         391     0.509950                            scale_imbalance_size\n",
       "10        416     0.508197                                        stock_id\n",
       "11        146     0.481616        global_rolling48_mean_target_mock_shift6\n",
       "12        394     0.381346                         seconds_in_bucket_group\n",
       "13        437     0.374136                                    wap_shift_n1\n",
       "14          2     0.359632                                       ask_price"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "feature_id_to_name = {str(id): name for id, name in enumerate(feature_names)}\n",
    "\n",
    "feature_importance_catboost2['Feature'] = feature_importance_catboost2['Feature Id'].map(feature_id_to_name)\n",
    "\n",
    "top_n = 15\n",
    "top_features = feature_importance_catboost2.head(top_n)\n",
    "top_features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "f5a7d5a3-4943-4ecd-9005-88aea1734a9b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABJYAAAUjCAYAAACU568BAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8fJSN1AAAACXBIWXMAAA9hAAAPYQGoP6dpAADMNUlEQVR4nOzdeVxV1f7/8fcBZDwcHBBxQNFQBCccshzB0ktq3rxWGuVsGppjOeR1RFPLITErLS2nzOKWWdqgpmIOaaJipeYYYkmZaeCICvv3Rz/O1yODuMNAfT0fj/24nH3WXvuz18H7/fK+a61jMQzDEAAAAAAAAHCTnAq7AAAAAAAAANyeCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgAAAHBbW7hwoSwWS47H0KFDb8k99+3bp/HjxyspKemW9P93JCUlyWKxaPr06YVdimlbt27V+PHj9eeffxZ2KQCAG3Ap7AIAAACAgrBgwQJVr17d4Vy5cuVuyb327dunmJgYRUREKDAw8Jbc4262detWxcTEqHv37ipevHhhlwMAyAPBEgAAAO4INWvWVIMGDQq7jL/lypUrslgscnG5O//f9IsXL8rd3b2wywAA3ASWwgEAAOCu8MEHH6hRo0by8vKS1WpVZGSkdu/e7dAmISFBTzzxhAIDA+Xh4aHAwEBFRUXp2LFj9jYLFy7U448/Lklq0aKFfdndwoULJUmBgYHq3r17tvtHREQoIiLC/jo+Pl4Wi0VLlizR888/r/Lly8vNzU2HDx+WJH311Vd68MEHZbPZ5OnpqSZNmmjdunWmnj1rueD69evVu3dvlSpVSjabTV27dtX58+f166+/qmPHjipevLjKli2roUOH6sqVK/brs5bXTZ06VZMmTVLFihXl7u6uBg0a5FjT5s2b9eCDD8rb21uenp5q3LixPvvssxxrWrNmjXr27KnSpUvL09NTI0eO1LBhwyRJlStXto9vfHy8pL8+x3/9618qW7asPDw8FBISohdeeEHnz5936L979+6yWq06fPiw2rRpI6vVqoCAAD3//PNKT093aJuenq4JEyYoJCRE7u7uKlWqlFq0aKGtW7fa2xiGoTfeeENhYWHy8PBQiRIl9Nhjj+no0aMOfe3evVsPP/yw/Pz85ObmpnLlyqlt27b6+eefb/6DA4DbAMESAAAA7ggZGRm6evWqw5Fl8uTJioqKUmhoqOLi4rRkyRKdPXtWzZo10759++ztkpKSFBwcrNjYWK1evVovv/yyUlJSdO+99+rUqVOSpLZt22ry5MmSpNdff13ffPONvvnmG7Vt29ZU3SNHjlRycrLmzp2rlStXys/PT++++67+9a9/yWazadGiRYqLi1PJkiUVGRlpOlySpKefflo+Pj56//33NXr0aL333nvq3bu32rZtqzp16ujDDz9Ut27dNGPGDM2ePTvb9a+99pq+/PJLxcbG6t1335WTk5Nat26tb775xt5m48aNeuCBB5Samqq3335by5Ytk7e3t9q1a6cPPvggW589e/ZUsWLFtGTJEn344Yfq27evBgwYIElavny5fXzr1asnSTp06JDatGmjt99+W19++aUGDx6suLg4tWvXLlvfV65c0b///W89+OCD+uSTT9SzZ0/NnDlTL7/8sr3N1atX1bp1a02cOFEPP/ywPv74Yy1cuFCNGzdWcnKyvd0zzzyjwYMHq2XLllqxYoXeeOMN7d27V40bN9Zvv/0mSTp//rxatWql3377Ta+//rrWrl2r2NhYVaxYUWfPnjX5qQFAEWcAAAAAt7EFCxYYknI8rly5YiQnJxsuLi7GgAEDHK47e/as4e/vb3Ts2DHXvq9evWqcO3fO8PLyMmbNmmU//7///c+QZGzYsCHbNZUqVTK6deuW7Xx4eLgRHh5uf71hwwZDktG8eXOHdufPnzdKlixptGvXzuF8RkaGUadOHaNhw4Z5jIZh/PTTT4YkY9q0afZzWWN0/Ri0b9/ekGS88sorDufDwsKMevXqZeuzXLlyxsWLF+3n09LSjJIlSxotW7a0n7v//vsNPz8/4+zZs/ZzV69eNWrWrGlUqFDByMzMdKipa9eu2Z5h2rRphiTjp59+yvNZMzMzjStXrhgbN240JBl79uyxv9etWzdDkhEXF+dwTZs2bYzg4GD768WLFxuSjHnz5uV6n2+++caQZMyYMcPh/PHjxw0PDw9j+PDhhmEYRkJCgiHJWLFiRZ51A8CdhBlLAAAAuCMsXrxYO3bscDhcXFy0evVqXb16VV27dnWYzeTu7q7w8HD7EitJOnfunEaMGKGgoCC5uLjIxcVFVqtV58+f1/79+29J3Y8++qjD661bt+r06dPq1q2bQ72ZmZl66KGHtGPHjmzLvvLr4YcfdngdEhIiSdlmW4WEhDgs/8vSoUMHhz2QsmYiff3118rIyND58+e1fft2PfbYY7JarfZ2zs7O6tKli37++WcdOHAgz+e/kaNHj+rJJ5+Uv7+/nJ2dVaxYMYWHh0tSts/IYrFkm8lUu3Zth2f74osv5O7urp49e+Z6z1WrVslisahz584On4m/v7/q1Klj/x0KCgpSiRIlNGLECM2dO9dhNhwA3Knuzl0BAQAAcMcJCQnJcfPurGVK9957b47XOTn93//W+uSTT2rdunUaM2aM7r33XtlsNlksFrVp00YXL168JXWXLVs2x3ofe+yxXK85ffq0vLy8bvpeJUuWdHjt6uqa6/lLly5lu97f3z/Hc5cvX9a5c+d09uxZGYaR7Zmk//uGvj/++MPhfE5tc3Pu3Dk1a9ZM7u7uevHFF1WtWjV5enrq+PHj6tChQ7bPyNPTM9tm4G5ubg7P9vvvv6tcuXIOvwfX++2332QYhsqUKZPj+1WqVJEk+fj4aOPGjZo0aZL++9//6syZMypbtqx69+6t0aNHq1ixYvl+VgC4XRAsAQAA4I7m6+srSfrwww9VqVKlXNulpqZq1apVGjdunF544QX7+fT0dJ0+fTrf93N3d8+2ObQknTp1yl7LtSwWS471zp49W/fff3+O98gt4LjVfv311xzPubq6ymq1ysXFRU5OTkpJScnW7sSJE5KUbQyuf/68rF+/XidOnFB8fLx9lpIk/fnnn/nu43qlS5fW5s2blZmZmWu45OvrK4vFok2bNsnNzS3b+9eeq1Wrlt5//30ZhqHvvvtOCxcu1IQJE+Th4eHwewUAdwqCJQAAANzRIiMj5eLioiNHjuS57MpiscgwjGzBwfz585WRkeFwLqtNTrOYAgMD9d133zmcO3jwoA4cOJBjsHS9Jk2aqHjx4tq3b5/69+9/w/b/pOXLl2vatGn2WUBnz57VypUr1axZMzk7O8vLy0v33Xefli9frunTp8vDw0OSlJmZqXfffVcVKlRQtWrVbnif3MY3K4S6/jN68803TT9T69attWzZMi1cuDDX5XAPP/ywXnrpJf3yyy/q2LFjvvq1WCyqU6eOZs6cqYULF2rXrl2mawSAooxgCQAAAHe0wMBATZgwQaNGjdLRo0f10EMPqUSJEvrtt9/07bffysvLSzExMbLZbGrevLmmTZsmX19fBQYGauPGjXr77bdVvHhxhz5r1qwpSXrrrbfk7e0td3d3Va5cWaVKlVKXLl3UuXNn9evXT48++qiOHTumqVOnqnTp0vmq12q1avbs2erWrZtOnz6txx57TH5+fvr999+1Z88e/f7775ozZ05BD1O+ODs7q1WrVnruueeUmZmpl19+WWlpaYqJibG3mTJlilq1aqUWLVpo6NChcnV11RtvvKEffvhBy5Yty9cMpVq1akmSZs2apW7duqlYsWIKDg5W48aNVaJECUVHR2vcuHEqVqyYli5dqj179ph+pqioKC1YsEDR0dE6cOCAWrRooczMTG3fvl0hISF64okn1KRJE/Xp00c9evRQQkKCmjdvLi8vL6WkpGjz5s2qVauW+vbtq1WrVumNN95Q+/btVaVKFRmGoeXLl+vPP/9Uq1atTNcIAEUZwRIAAADueCNHjlRoaKhmzZqlZcuWKT09Xf7+/rr33nsVHR1tb/fee+9p0KBBGj58uK5evaomTZpo7dq12Ta3rly5smJjYzVr1ixFREQoIyNDCxYsUPfu3fXkk0/qxIkTmjt3rhYsWKCaNWtqzpw5DuHLjXTu3FkVK1bU1KlT9cwzz+js2bPy8/NTWFiYunfvXlDDctP69++vS5cuaeDAgTp58qRq1Kihzz77TE2aNLG3CQ8P1/r16zVu3Dh1795dmZmZqlOnjj799NNsm4fnJiIiQiNHjtSiRYs0b948ZWZmasOGDYqIiNBnn32m559/Xp07d5aXl5ceeeQRffDBB6pXr56pZ3JxcdHnn3+uKVOmaNmyZYqNjZW3t7fq1Kmjhx56yN7uzTff1P33368333xTb7zxhjIzM1WuXDk1adJEDRs2lCRVrVpVxYsX19SpU3XixAm5uroqODhYCxcuVLdu3UzVBwBFncUwDKOwiwAAAABQdCUlJaly5cqaNm2ahg4dWtjlAACKkNy/+gAAAAAAAADIA8ESAAAAAAAATGEpHAAAAAAAAExhxhIAAAAAAABMIVgCAAAAAACAKQRLAAAAAAAAMMWlsAsAABQNmZmZOnHihLy9vWWxWAq7HAAAAACFxDAMnT17VuXKlZOTU95zkgiWAACSpBMnTiggIKCwywAAAABQRBw/flwVKlTIsw3BEgBAkuTt7S3pr//jYbPZCrkaAAAAAIUlLS1NAQEB9r8R8kKwBACQJPvyN5vNRrAEAAAAIF9bZLB5NwAAAAAAAEwhWAIAAAAAAIApBEsAAAAAAAAwhWAJAAAAAAAAphAsAQAAAAAAwBSCJQAAAAAAAJhCsAQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgAAAAAAAGAKwRIAAAAAAABMcSnsAgAARUvNcavl5OZZ2GUAAAAAd42kl9oWdgmmMWMJAAAAAAAAphAsAQAAAAAAwBSCJQAAAAAAAJhCsAQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgAAAAAAAGAKwRIAAAAAAABMIVgCAAAAAACAKQRLAHCH6N69u9q3b1/YZQAAAAC4ixAsAShyIiIiNHjw4MIuw66o1QMAAAAARQXBEoA70uXLlwu7BAAAAAC44xEsAShSunfvro0bN2rWrFmyWCyyWCw6cuSIevXqpcqVK8vDw0PBwcGaNWtWtuvat2+vKVOmqFy5cqpWrZokaevWrQoLC5O7u7saNGigFStWyGKxKDEx0X7tvn371KZNG1mtVpUpU0ZdunTRqVOncq0nKSkpz2eIj4+XxWLR6tWrVbduXXl4eOiBBx7QyZMn9cUXXygkJEQ2m01RUVG6cOGC/br09HQNHDhQfn5+cnd3V9OmTbVjxw6Hvvfu3au2bdvKZrPJ29tbzZo105EjR3KsY+fOnfLz89OkSZPyO/wAAAAAcFNcCrsAALjWrFmzdPDgQdWsWVMTJkyQJJUoUUIVKlRQXFycfH19tXXrVvXp00dly5ZVx44d7deuW7dONptNa9eulWEYOnv2rNq1a6c2bdrovffe07Fjx7ItaUtJSVF4eLh69+6tV155RRcvXtSIESPUsWNHrV+/Psd6Spcuna9nGT9+vF577TV5enqqY8eO6tixo9zc3PTee+/p3Llz+s9//qPZs2drxIgRkqThw4fro48+0qJFi1SpUiVNnTpVkZGROnz4sEqWLKlffvlFzZs3V0REhNavXy+bzaYtW7bo6tWr2e4dHx9vD9r69u2bY33p6elKT0+3v05LS8vXcwEAAABAFoIlAEWKj4+PXF1d5enpKX9/f/v5mJgY+8+VK1fW1q1bFRcX5xAseXl5af78+XJ1dZUkzZ07VxaLRfPmzZO7u7tCQ0P1yy+/qHfv3vZr5syZo3r16mny5Mn2c++8844CAgJ08OBBVatWLcd68uPFF19UkyZNJEm9evXSyJEjdeTIEVWpUkWS9Nhjj2nDhg0aMWKEzp8/rzlz5mjhwoVq3bq1JGnevHlau3at3n77bQ0bNkyvv/66fHx89P7776tYsWKSZJ+Zda1PPvlEXbp00ZtvvqmoqKhc65syZYrDuAIAAADAzWIpHIDbwty5c9WgQQOVLl1aVqtV8+bNU3JyskObWrVq2UMlSTpw4IBq164td3d3+7mGDRs6XLNz505t2LBBVqvVflSvXl2Scl1ill+1a9e2/1ymTBl5enraQ6WscydPnrTf68qVK/YgSpKKFSumhg0bav/+/ZKkxMRENWvWzB4q5WT79u169NFHtWjRojxDJUkaOXKkUlNT7cfx48dNPScAAACAuxczlgAUeXFxcRoyZIhmzJihRo0aydvbW9OmTdP27dsd2nl5eTm8NgxDFosl27lrZWZmql27dnr55Zez3bds2bJ/q+5rAyCLxZItELJYLMrMzHSoK6d6s855eHjc8J733HOPSpUqpXfeeUdt27Z1CNqu5+bmJjc3t/w9DAAAAADkgBlLAIocV1dXZWRk2F9v2rRJjRs3Vr9+/VS3bl0FBQXlazZR9erV9d133znsI5SQkODQpl69etq7d68CAwMVFBTkcGQFVdfXcysEBQXJ1dVVmzdvtp+7cuWKEhISFBISIumvGVCbNm3SlStXcu3H19dX69ev15EjR9SpU6c82wIAAADA30WwBKDICQwM1Pbt25WUlKRTp04pKChICQkJWr16tQ4ePKgxY8Zk+7a0nDz55JPKzMxUnz59tH//fq1evVrTp0+X9H8zg5599lmdPn1aUVFR+vbbb3X06FGtWbNGPXv2tIdJ19eTNcuoIHl5ealv374aNmyYvvzyS+3bt0+9e/fWhQsX1KtXL0lS//79lZaWpieeeEIJCQk6dOiQlixZogMHDjj05efnp/Xr1+vHH39UVFRUjpt7AwAAAEBBIFgCUOQMHTpUzs7OCg0NVenSpfXQQw+pQ4cO6tSpk+677z798ccf6tev3w37sdlsWrlypRITExUWFqZRo0Zp7NixkmTfd6lcuXLasmWLMjIyFBkZqZo1a2rQoEHy8fGRk5NTjvVcv7dTQXnppZf06KOPqkuXLqpXr54OHz6s1atXq0SJEpKkUqVKaf369Tp37pzCw8NVv359zZs3L8c9l/z9/bV+/Xp9//33euqpp275jCsAAAAAdyeLcf2GIwBwB1u6dKl69Oih1NTUfO1ZdDdJS0uTj4+PAgbHycnNs7DLAQAAAO4aSS+1LewSHGT9bZCamiqbzZZnWzbvBnBHW7x4sapUqaLy5ctrz549GjFihDp27EioBAAAAAAFgKVwAO5ov/76qzp37qyQkBANGTJEjz/+uN56662/1Wd0dLSsVmuOR3R0dAFVDgAAAABFH0vhAOAmnTx5UmlpaTm+Z7PZ5Ofn9w9XVDBYCgcAAAAUDpbCAcBdxM/P77YNjwAAAACgILEUDgAAAAAAAKYQLAEAAAAAAMAUgiUAAAAAAACYQrAEAAAAAAAAUwiWAAAAAAAAYArBEgAAAAAAAEwhWAIAAAAAAIApBEsAAAAAAAAwhWAJAAAAAAAAprgUdgEAgKLlh5hI2Wy2wi4DAAAAwG2AGUsAAAAAAAAwhWAJAAAAAAAAphAsAQAAAAAAwBSCJQAAAAAAAJhCsAQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgAAAAAAAGAKwRIAAAAAAABMIVgCAAAAAACAKQRLAAAAAAAAMIVgCQAAAAAAAKYQLAEAAAAAAMAUgiUAAAAAAACYQrAEAAAAAAAAUwiWAAAAAAAAYArBEgAAAAAAAEwhWAIAAAAAAIApBEsAAAAAAAAwhWAJAAAAAAAAphAsAQAAAAAAwBSCJQAAAAAAAJhCsAQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgAAAAAAAGAKwRIAAAAAAABMIVgCAAAAAACAKS6FXQAAoGipOW61nNw8C7sMQJKU9FLbwi4BAAAAeWDGEgAAAAAAAEwhWAIAAAAAAIApBEsAAAAAAAAwhWAJAAAAAAAAphAsAQAAAAAAwBSCJQAAAAAAAJhCsAQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgDgFkpKSpLFYlFiYmKB9tu9e3e1b9++QPsEAAAAgJtFsASg0EVERGjw4MGFXYZdQdYTEBCglJQU1axZs0D6AwAAAICihGAJwB3h8uXLhV1CjpydneXv7y8XF5fCLgUAAAAAChzBEoBC1b17d23cuFGzZs2SxWKRxWLRkSNH1KtXL1WuXFkeHh4KDg7WrFmzsl3Xvn17TZkyReXKlVO1atUkSVu3blVYWJjc3d3VoEEDrVixIttStH379qlNmzayWq0qU6aMunTpolOnTuVaT1JSUp7PcObMGT311FMqXbq0PDw8VLVqVS1YsEBS9qVw8fHxslgsWrdunRo0aCBPT081btxYBw4ccOjzxRdflJ+fn7y9vfX000/rhRdeUFhYWK41GIahqVOnqkqVKvLw8FCdOnX04Ycf5uMTAAAAAADzCJYAFKpZs2apUaNG6t27t1JSUpSSkqIKFSqoQoUKiouL0759+zR27Fj997//VVxcnMO169at0/79+7V27VqtWrVKZ8+eVbt27VSrVi3t2rVLEydO1IgRIxyuSUlJUXh4uMLCwpSQkKAvv/xSv/32mzp27JhrPQEBAXk+w5gxY7Rv3z598cUX2r9/v+bMmSNfX988rxk1apRmzJihhIQEubi4qGfPnvb3li5dqkmTJunll1/Wzp07VbFiRc2ZMyfP/kaPHq0FCxZozpw52rt3r4YMGaLOnTtr48aNuV6Tnp6utLQ0hwMAAAAAbgZrMwAUKh8fH7m6usrT01P+/v728zExMfafK1eurK1btyouLs4eAEmSl5eX5s+fL1dXV0nS3LlzZbFYNG/ePLm7uys0NFS//PKLevfubb9mzpw5qlevniZPnmw/98477yggIEAHDx5UtWrVcqwnL8nJyapbt64aNGggSQoMDLzhNZMmTVJ4eLgk6YUXXlDbtm116dIlubu7a/bs2erVq5d69OghSRo7dqzWrFmjc+fO5djX+fPn9corr2j9+vVq1KiRJKlKlSravHmz3nzzTft9rjdlyhSHcQYAAACAm8WMJQBF0ty5c9WgQQOVLl1aVqtV8+bNU3JyskObWrVq2UMlSTpw4IBq164td3d3+7mGDRs6XLNz505t2LBBVqvVflSvXl2SdOTIEVO19u3bV++//77CwsI0fPhwbd269YbX1K5d2/5z2bJlJUknT560P8f1dV//+lr79u3TpUuX1KpVK4fnWrx4cZ7PNHLkSKWmptqP48eP37BuAAAAALgWM5YAFDlxcXEaMmSIZsyYoUaNGsnb21vTpk3T9u3bHdp5eXk5vDYMQxaLJdu5a2VmZqpdu3Z6+eWXs903K+C5Wa1bt9axY8f02Wef6auvvtKDDz6oZ599VtOnT8/1mmLFitl/zqo5MzMz27ncnuNaWdd99tlnKl++vMN7bm5uuV7n5uaW5/sAAAAAcCMESwAKnaurqzIyMuyvN23apMaNG6tfv372c/mZTVS9enUtXbpU6enp9sAkISHBoU29evX00UcfKTAwMNdvaru+nvwoXbq0unfvru7du6tZs2YaNmxYnsFSXoKDg/Xtt9+qS5cu9nPXP8e1QkND5ebmpuTk5FyXvQEAAADArcBSOACFLjAwUNu3b1dSUpJOnTqloKAgJSQkaPXq1Tp48KDGjBmjHTt23LCfJ598UpmZmerTp4/279+v1atX28OdrBlAzz77rE6fPq2oqCh9++23Onr0qNasWaOePXvaw6Tr67l2JlFOxo4dq08++USHDx/W3r17tWrVKoWEhJgejwEDBujtt9/WokWLdOjQIb344ov67rvvss1iyuLt7a2hQ4dqyJAhWrRokY4cOaLdu3fr9ddf16JFi0zXAQAAAAA3QrAEoNANHTpUzs7OCg0NVenSpfXQQw+pQ4cO6tSpk+677z798ccfDrOXcmOz2bRy5UolJiYqLCxMo0aN0tixYyXJvu9SuXLltGXLFmVkZCgyMlI1a9bUoEGD5OPjIycnpxzruX5vp+u5urpq5MiRql27tpo3by5nZ2e9//77psfjqaee0siRIzV06FDVq1dPP/30k7p37+6wd9T1Jk6cqLFjx2rKlCkKCQlRZGSkVq5cqcqVK5uuAwAAAABuxGLktXEHANzmli5dqh49eig1NVUeHh6FXY5prVq1kr+/v5YsWXLL7pGWliYfHx8FDI6Tk5vnLbsPcDOSXmpb2CUAAADcdbL+NkhNTZXNZsuzLXssAbijLF68WFWqVFH58uW1Z88ejRgxQh07drytQqULFy5o7ty5ioyMlLOzs5YtW6avvvpKa9euLezSAAAAAMABS+EA3FF+/fVXde7cWSEhIRoyZIgef/xxvfXWW3+rz+joaFmt1hyP6OjoAqr8/1gsFn3++edq1qyZ6tevr5UrV+qjjz5Sy5YtC/xeAAAAAPB3sBQOAG7g5MmTSktLy/E9m80mPz+/f7iiW4OlcCiKWAoHAADwz2MpHAAUID8/vzsmPAIAAACAgsRSOAAAAAAAAJhCsAQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmOJS2AUAAIqWH2IiZbPZCrsMAAAAALcBZiwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgAAAAAAAGAKwRIAAAAAAABMIVgCAAAAAACAKQRLAAAAAAAAMIVgCQAAAAAAAKYQLAEAAAAAAMAUgiUAAAAAAACYQrAEAAAAAAAAUwiWAAAAAAAAYArBEgAAAAAAAEwhWAIAAAAAAIApBEsAAAAAAAAwhWAJAAAAAAAAphAsAQAAAAAAwBSCJQAAAAAAAJhCsAQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgAAAAAAAGAKwRIAAAAAAABMIVgCAAAAAACAKQRLAAAAAAAAMIVgCQAAAAAAAKYQLAEAAAAAAMAUgiUAAAAAAACYQrAEAAAAAAAAUwiWAAAAAAAAYArBEgAAAAAAAEwhWAIAAAAAAIApBEsAAAAAAAAwhWAJAAAAAAAAphAsAQAAAAAAwBSXwi4AAFC01By3Wk5unoVdxh0t6aW2hV0CAAAAUCCYsQQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgAAAAAAAGAKwRIAAAAAAABMIVgCAAAAAACAKQRLAAAAAAAAMIVgCQAAAAAAAKYQLAHATYiIiNDgwYPzbGOxWLRixQr76x9//FH333+/3N3dFRYWdkvrAwAAAIB/EsESABSwlJQUtW7d2v563Lhx8vLy0oEDB7Ru3TotXLhQxYsXvyX3/uyzz3TffffJw8NDvr6+6tChwy25DwAAAABIkkthFwAAdxp/f3+H10eOHFHbtm1VqVKlW3rfjz76SL1799bkyZP1wAMPyDAMff/997f0ngAAAADubsxYAlDoVq5cqeLFiyszM1OSlJiYKIvFomHDhtnbPPPMM4qKitIff/yhqKgoVahQQZ6enqpVq5aWLVvm0F9ERIT69++v/v37q3jx4ipVqpRGjx4twzDyVc8bb7yhqlWryt3dXWXKlNFjjz3m8H5mZqaGDx+ukiVLyt/fX+PHj3d4/9qlcBaLRTt37tSECRNksVgUERGhHj16KDU1VRaLRRaLJdv1OQkMDNTkyZPVs2dPeXt7q2LFinrrrbfs71+9elWDBg3StGnTFB0drWrVqik4ODhb7QAAAABQkAiWABS65s2b6+zZs9q9e7ckaePGjfL19dXGjRvtbeLj4xUeHq5Lly6pfv36WrVqlX744Qf16dNHXbp00fbt2x36XLRokVxcXLR9+3a9+uqrmjlzpubPn3/DWhISEjRw4EBNmDBBBw4c0JdffqnmzZtn69vLy0vbt2/X1KlTNWHCBK1duzbH/lJSUlSjRg09//zzSklJ0aeffqrY2FjZbDalpKQoJSVFQ4cOzdc4zZgxQw0aNNDu3bvVr18/9e3bVz/++KMkadeuXfrll1/k5OSkunXrqmzZsmrdurX27t2ba3/p6elKS0tzOAAAAADgZhAsASh0Pj4+CgsLU3x8vKS/QqQhQ4Zoz549Onv2rH799VcdPHhQERERKl++vIYOHaqwsDBVqVJFAwYMUGRkpP73v/859BkQEKCZM2cqODhYTz31lAYMGKCZM2fesJbk5GR5eXnp4YcfVqVKlVS3bl0NHDjQoU3t2rU1btw4Va1aVV27dlWDBg20bt26HPvz9/eXi4uLrFar/P39ZbPZ5OPjI4vFIn9/f/n7+8tqteZrnNq0aaN+/fopKChII0aMkK+vr33Mjh49KkkaP368Ro8erVWrVqlEiRIKDw/X6dOnc+xvypQp8vHxsR8BAQH5qgMAAAAAshAsASgSIiIiFB8fL8MwtGnTJj3yyCOqWbOmNm/erA0bNqhMmTKqXr26MjIyNGnSJNWuXVulSpWS1WrVmjVrlJyc7NDf/fffL4vFYn/dqFEjHTp0SBkZGXnW0apVK1WqVElVqlRRly5dtHTpUl24cMGhTe3atR1ely1bVidPnvybI3Bj1943K5jKum/WMsJRo0bp0UcfVf369bVgwQJZLJZsoVuWkSNHKjU11X4cP378lj8DAAAAgDsLwRKAIiEiIkKbNm3Snj175OTkpNDQUIWHh2vjxo32ZXDSX8vBZs6cqeHDh2v9+vVKTExUZGSkLl++XCB1eHt7a9euXVq2bJnKli2rsWPHqk6dOvrzzz/tbYoVK+ZwjcVisQc7t1Je9y1btqwkKTQ01P6+m5ubqlSpki10u/Z9m83mcAAAAADAzSBYAlAkZO2zFBsbq/DwcFksFoWHhys+Pt4hWMqazdS5c2fVqVNHVapU0aFDh7L1t23btmyvq1atKmdn5xvW4uLiopYtW2rq1Kn67rvvlJSUpPXr1xfMg0pydXW94cypm1W/fn25ubnpwIED9nNXrlxRUlLSLf82OgAAAAB3L4IlAEVC1j5L7777riIiIiT9FTbt2rXLvr+SJAUFBWnt2rXaunWr9u/fr2eeeUa//vprtv6OHz+u5557TgcOHNCyZcs0e/ZsDRo06IZ1rFq1Sq+++qoSExN17NgxLV68WJmZmQoODi6wZw0MDNS5c+e0bt06nTp1KttSOzNsNpuio6M1btw4rVmzRgcOHFDfvn0lSY8//vjf7h8AAAAAcuJS2AUAQJYWLVpo165d9hCpRIkSCg0N1YkTJxQSEiJJGjNmjH766SdFRkbK09NTffr0Ufv27ZWamurQV9euXXXx4kU1bNhQzs7OGjBggPr06XPDGooXL67ly5dr/PjxunTpkqpWraply5apRo0aBfacjRs3VnR0tDp16qQ//vhD48aN0/jx4/92v9OmTZOLi4u6dOmiixcv6r777tP69etVokSJv180AAAAAOTAYhiGUdhFAEBBioiIUFhYmGJjYwu7lNtKWlraX98ONzhOTm6ehV3OHS3ppbaFXQIAAACQq6y/DVJTU2+4FytL4QAAAAAAAGAKwRKAu8qmTZtktVpzPagJAAAAAPKPPZYA3HHi4+Nzfa9BgwZKTEz8x2rJj6JYEwAAAADkB8ESgLuKh4eHgoKCCrsMB0WxJgAAAADID5bCAQAAAAAAwBSCJQAAAAAAAJhCsAQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmOJS2AUAAIqWH2IiZbPZCrsMAAAAALcBZiwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgAAAAAAAGAKwRIAAAAAAABMIVgCAAAAAACAKQRLAAAAAAAAMIVgCQAAAAAAAKYQLAEAAAAAAMAUgiUAAAAAAACYQrAEAAAAAAAAUwiWAAAAAAAAYArBEgAAAAAAAEwhWAIAAAAAAIApBEsAAAAAAAAwhWAJAAAAAAAAphAsAQAAAAAAwBSCJQAAAAAAAJhCsAQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgAAAAAAAGAKwRIAAAAAAABMIVgCAAAAAACAKQRLAAAAAAAAMIVgCQAAAAAAAKYQLAEAAAAAAMAUgiUAAAAAAACYQrAEAAAAAAAAUwiWAAAAAAAAYArBEgAAAAAAAEwhWAIAAAAAAIApBEsAAAAAAAAwhWAJAAAAAAAAprgUdgEAgKKl5rjVcnLzLOwyipSkl9oWdgkAAABAkcSMJQAAAAAAAJhCsAQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgAAAAAAAGAKwRIAAAAAAABMIVgCAAAAAACAKQRLAAAAAAAAMIVgCbjDRUREaPDgwaavT0pKksViUWJiYoHVhFuje/fuat++fWGXAQAAAOAu4lLYBQC4tZYvX65ixYoVdhn5FhERobCwMMXGxhZ2KXZFsSYAAAAAKAoIloA7XMmSJQu7hEJx+fJlubq6FnYZAAAAAHBHYykccIe7dilcYGCgJk+erJ49e8rb21sVK1bUW2+95dD+22+/Vd26deXu7q4GDRpo9+7d2frct2+f2rRpI6vVqjJlyqhLly46deqUJCk+Pl6urq7atGmTvf2MGTPk6+urlJSUPGvt3r27Nm7cqFmzZslischisSgpKUkZGRnq1auXKleuLA8PDwUHB2vWrFnZrm3fvr2mTJmicuXKqVq1apKkrVu3KiwszP48K1asyLa0L6/nya2mvMTHx8tisWj16tWqW7euPDw89MADD+jkyZP64osvFBISIpvNpqioKF24cMF+XXp6ugYOHCg/Pz+5u7uradOm2rFjh0Pfe/fuVdu2bWWz2eTt7a1mzZrpyJEjOdaxc+dO+fn5adKkSXnWCwAAAABmESwBd5kZM2bYA6N+/fqpb9+++vHHHyVJ58+f18MPP6zg4GDt3LlT48eP19ChQx2uT0lJUXh4uMLCwpSQkKAvv/xSv/32mzp27Cjp/4KsLl26KDU1VXv27NGoUaM0b948lS1bNs/aZs2apUaNGql3795KSUlRSkqKAgIClJmZqQoVKiguLk779u3T2LFj9d///ldxcXEO169bt0779+/X2rVrtWrVKp09e1bt2rVTrVq1tGvXLk2cOFEjRoy4qefJrab8GD9+vF577TVt3bpVx48fV8eOHRUbG6v33ntPn332mdauXavZs2fb2w8fPlwfffSRFi1apF27dikoKEiRkZE6ffq0JOmXX35R8+bN5e7urvXr12vnzp3q2bOnrl69mu3e8fHxevDBBxUTE6NRo0blWF96errS0tIcDgAAAAC4GSyFA+4ybdq0Ub9+/SRJI0aM0MyZMxUfH6/q1atr6dKlysjI0DvvvCNPT0/VqFFDP//8s/r27Wu/fs6cOapXr54mT55sP/fOO+8oICBABw8eVLVq1fTiiy/qq6++Up8+fbR371516dJF//nPf25Ym4+Pj1xdXeXp6Sl/f3/7eWdnZ8XExNhfV65cWVu3blVcXJw9AJIkLy8vzZ8/374Ebu7cubJYLJo3b57c3d0VGhqqX375Rb17976p58mppvx48cUX1aRJE0lSr169NHLkSB05ckRVqlSRJD322GPasGGDRowYofPnz2vOnDlauHChWrduLUmaN2+e1q5dq7ffflvDhg3T66+/Lh8fH73//vv2fbOyZmZd65NPPlGXLl305ptvKioqKtf6pkyZ4jCuAAAAAHCzmLEE3GVq165t/9liscjf318nT56UJO3fv1916tSRp6envU2jRo0crt+5c6c2bNggq9VqP6pXry5J9iVZrq6uevfdd/XRRx/p4sWLBbLp9dy5c9WgQQOVLl1aVqtV8+bNU3JyskObWrVqOeyrdODAAdWuXVvu7u72cw0bNrzp5zHr2rEuU6aMPD097aFS1rmssT9y5IiuXLliD6IkqVixYmrYsKH2798vSUpMTFSzZs3y3Ix9+/btevTRR7Vo0aI8QyVJGjlypFJTU+3H8ePHTT0nAAAAgLsXM5aAu8z1oYTFYlFmZqYkyTCMG16fmZmpdu3a6eWXX8723rVL3bZu3SpJOn36tE6fPi0vLy/TNcfFxWnIkCGaMWOGGjVqJG9vb02bNk3bt293aHf9PQzDkMViyXbOzPOYce1YWyyWfI19TvVmnfPw8LjhPe+55x6VKlVK77zzjtq2bZvnBuZubm5yc3PL38MAAAAAQA6YsQTALjQ0VHv27NHFixft57Zt2+bQpl69etq7d68CAwMVFBTkcGQFO0eOHNGQIUM0b9483X///eratas9QLkRV1dXZWRkOJzbtGmTGjdurH79+qlu3boKCgrK12yi6tWr67vvvlN6err9XEJCwk0/T041FbSgoCC5urpq8+bN9nNXrlxRQkKCQkJCJP01A2rTpk26cuVKrv34+vpq/fr1OnLkiDp16pRnWwAAAAD4uwiWANg9+eSTcnJyUq9evbRv3z59/vnnmj59ukObZ599VqdPn1ZUVJS+/fZbHT16VGvWrFHPnj2VkZGhjIwMdenSRf/617/Uo0cPLViwQD/88INmzJiRrxoCAwO1fft2JSUl6dSpU8rMzFRQUJASEhK0evVqHTx4UGPGjMn2bWm5PU9mZqb69Omj/fv3a/Xq1fbnyZoFdKPnya2mgubl5aW+fftq2LBh+vLLL7Vv3z717t1bFy5cUK9evSRJ/fv3V1pamp544gklJCTo0KFDWrJkiQ4cOODQl5+fn9avX68ff/xRUVFROW7uDQAAAAAFgWAJgJ3VatXKlSu1b98+1a1bV6NGjcq2RKxcuXLasmWLMjIyFBkZqZo1a2rQoEHy8fGRk5OTJk2apKSkJL311luSJH9/f82fP1+jR49WYmLiDWsYOnSonJ2dFRoaqtKlSys5OVnR0dHq0KGDOnXqpPvuu09//PGHfQPyvNhsNq1cuVKJiYkKCwvTqFGjNHbsWEmy77t0o+fJraZb4aWXXtKjjz6qLl26qF69ejp8+LBWr16tEiVKSJJKlSql9evX69y5cwoPD1f9+vU1b968HPdc8vf31/r16/X999/rqaeeuuUzrgAAAADcnSxGfjZVAYA7xNKlS9WjRw+lpqbma8+iu0laWpp8fHwUMDhOTm6eN77gLpL0UtvCLgEAAAD4x2T9bZCamiqbzZZnWzbvBnBHW7x4sapUqaLy5ctrz549GjFihDp27EioBAAAAAAFgKVwAP4xycnJslqtuR63YonZr7/+qs6dOyskJERDhgzR448/bl+mZ1Z0dHSuzxAdHV1AlQMAAABA0cdSOAD/mKtXryopKSnX9wMDA+XiUvQnUp48eVJpaWk5vmez2eTn5/cPV1QwWAqXO5bCAQAA4G7CUjgARZKLi4uCgoIKu4y/zc/P77YNjwAAAACgILEUDgAAAAAAAKYQLAEAAAAAAMAUgiUAAAAAAACYQrAEAAAAAAAAUwiWAAAAAAAAYArBEgAAAAAAAEwhWAIAAAAAAIApBEsAAAAAAAAwhWAJAAAAAAAAphAsAQAAAAAAwBSXwi4AAFC0/BATKZvNVthlAAAAALgNMGMJAAAAAAAAphAsAQAAAAAAwBSCJQAAAAAAAJhCsAQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgAAAAAAAGAKwRIAAAAAAABMIVgCAAAAAACAKQRLAAAAAAAAMIVgCQAAAAAAAKYQLAEAAAAAAMAUgiUAAAAAAACYQrAEAAAAAAAAUwiWAAAAAAAAYArBEgAAAAAAAEwhWAIAAAAAAIApBEsAAAAAAAAwhWAJAAAAAAAAphAsAQAAAAAAwBSCJQAAAAAAAJhCsAQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgAAAAAAAGAKwRIAAAAAAABMIVgCAAAAAACAKQRLAAAAAAAAMMWlsAsAABQtNcetlpObZ2GXcUskvdS2sEsAAAAA7ijMWAIAAAAAAIApBEsAAAAAAAAwhWAJAAAAAAAAphAsAQAAAAAAwBSCJQAAAAAAAJhCsAQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgAAAAAAAGAKwRKAu05gYKBiY2Pz3T4pKUkWi0WJiYm3rCYAAAAAuB0RLAG46+zYsUN9+vQp0D4XLlyo4sWL39Q1NxtwAQAAAEBR41LYBQDAP6106dKFXQIAAAAA3BGYsQSgyFu5cqWKFy+uzMxMSVJiYqIsFouGDRtmb/PMM88oKipKkrR161Y1b95cHh4eCggI0MCBA3X+/Hl72+tnCv34449q2rSp3N3dFRoaqq+++koWi0UrVqxwqOPo0aNq0aKFPD09VadOHX3zzTeSpPj4ePXo0UOpqamyWCyyWCwaP358ns8UERGhY8eOaciQIfZrpP+b+bRq1SoFBwfL09NTjz32mM6fP69FixYpMDBQJUqU0IABA5SRkeHwTBMnTtSTTz4pq9WqcuXKafbs2XnWkJ6errS0NIcDAAAAAG4GwRKAIq958+Y6e/asdu/eLUnauHGjfH19tXHjRnub+Ph4hYeH6/vvv1dkZKQ6dOig7777Th988IE2b96s/v3759h3Zmam2rdvL09PT23fvl1vvfWWRo0alWPbUaNGaejQoUpMTFS1atUUFRWlq1evqnHjxoqNjZXNZlNKSopSUlI0dOjQPJ9p+fLlqlChgiZMmGC/JsuFCxf06quv6v3339eXX36p+Ph4dejQQZ9//rk+//xzLVmyRG+99ZY+/PBDhz6nTZum2rVra9euXRo5cqSGDBmitWvX5lrDlClT5OPjYz8CAgLyrBkAAAAArsdSOABFno+Pj8LCwhQfH6/69esrPj5eQ4YMUUxMjM6ePavz58/r4MGDioiI0OTJk/Xkk09q8ODBkqSqVavq1VdfVXh4uObMmSN3d3eHvtesWaMjR44oPj5e/v7+kqRJkyapVatW2eoYOnSo2rZtK0mKiYlRjRo1dPjwYVWvXl0+Pj6yWCz2Pm6kZMmScnZ2lre3d7Zrrly5ojlz5uiee+6RJD322GNasmSJfvvtN1mtVoWGhqpFixbasGGDOnXqZL+uSZMmeuGFFyRJ1apV05YtWzRz5swcn0WSRo4cqeeee87+Oi0tjXAJAAAAwE1hxhKA20JERITi4+NlGIY2bdqkRx55RDVr1tTmzZu1YcMGlSlTRtWrV9fOnTu1cOFCWa1W+xEZGanMzEz99NNP2fo9cOCAAgICHMKdhg0b5lhD7dq17T+XLVtWknTy5MkCflLJ09PTHipJUpkyZRQYGCir1epw7vp7N2rUKNvr/fv353ofNzc32Ww2hwMAAAAAbgYzlgDcFiIiIvT2229rz549cnJyUmhoqMLDw7Vx40adOXNG4eHhkv5a2vbMM89o4MCB2fqoWLFitnOGYdj3N7qRYsWK2X/OuiZr36eCdO19su6V07n83Du/zwYAAAAAZhAsAbgtZO2zFBsbq/DwcFksFoWHh2vKlCk6c+aMBg0aJEmqV6+e9u7dq6CgoHz1W716dSUnJ+u3335TmTJlJEk7duy46fpcXV0dNtO+VdfkZdu2bdleV69evcD6BwAAAIDrsRQOwG0ha5+ld999VxEREZL+Cpt27dpl319JkkaMGKFvvvlGzz77rBITE3Xo0CF9+umnGjBgQI79tmrVSvfcc4+6deum7777Tlu2bLFv3n0zs30CAwN17tw5rVu3TqdOndKFCxfydc3XX3+tX375RadOncr3vXKzZcsWTZ06VQcPHtTrr7+u//3vf/bADQAAAABuBYIlALeNFi1aKCMjwx4ilShRQqGhoSpdurRCQkIk/bUP0saNG3Xo0CE1a9ZMdevW1ZgxY+x7Il3P2dlZK1as0Llz53Tvvffq6aef1ujRoyUp20bfeWncuLGio6PVqVMnlS5dWlOnTr3hNRMmTFBSUpLuuecelS5dOt/3ys3zzz+vnTt3qm7dupo4caJmzJihyMjIv90vAAAAAOTGYhiGUdhFAEBRsmXLFjVt2lSHDx922ES7KAsMDNTgwYPt34ZnRlpamnx8fBQwOE5Obp4FV1wRkvRS28IuAQAAACjysv42SE1NveGX/LDHEoC73scffyyr1aqqVavq8OHDGjRokJo0aXLbhEoAAAAAUFgIlgDc9c6ePavhw4fr+PHj8vX1VcuWLTVjxoy/1eemTZvUunXrXN8/d+7c3+ofAAAAAIoClsIBwC1w8eJF/fLLL7m+n99vrfsnsRQOAAAAgMRSOAAodB4eHkUyPAIAAACAgsS3wgEAAAAAAMAUgiUAAAAAAACYQrAEAAAAAAAAUwiWAAAAAAAAYArBEgAAAAAAAEwhWAIAAAAAAIApBEsAAAAAAAAwhWAJAAAAAAAAphAsAQAAAAAAwBSXwi4AAFC0/BATKZvNVthlAAAAALgNMGMJAAAAAAAAphAsAQAAAAAAwBSCJQAAAAAAAJhCsAQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgAAAAAAAGAKwRIAAAAAAABMIVgCAAAAAACAKQRLAAAAAAAAMIVgCQAAAAAAAKYQLAEAAAAAAMAUgiUAAAAAAACYQrAEAAAAAAAAUwiWAAAAAAAAYArBEgAAAAAAAEwhWAIAAAAAAIApBEsAAAAAAAAwhWAJAAAAAAAAphAsAQAAAAAAwBSCJQAAAAAAAJhCsAQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgAAAAAAAGAKwRIAAAAAAABMIVgCAAAAAACAKQRLAAAAAAAAMMWlsAsAABQtNcetlpOb5z96z6SX2v6j9wMAAABQMJixBAAAAAAAAFMIlgAAAAAAAGAKwRIAAAAAAABMIVgCAAAAAACAKQRLAAAAAAAAMIVgCQAAAAAAAKYQLAEAAAAAAMAUgiUAAAAAAACYQrAEAAAAAAAAUwiWAAAAAAAAYArBEgAAAAAAAEwhWAIAAAAAAIApBEsAAAAAAAAwhWAJAAAAAAAAphAsAQAAAAAAwBSCJQBFUnx8vCwWi/78888i0U+W7t27q3379gXSV24WLlyo4sWL39J7AAAAAEBBIFgCcEdr3LixUlJS5OPjU9il/GMKOkwDAAAAgNy4FHYBAHArubq6yt/fv7DLAAAAAIA7EjOWgLvEhx9+qFq1asnDw0OlSpVSy5Ytdf78eUnSggULFBISInd3d1WvXl1vvPGGw7U///yznnjiCZUsWVJeXl5q0KCBtm/fbn9/zpw5uueee+Tq6qrg4GAtWbLE4XqLxaL58+frP//5jzw9PVW1alV9+umnDm0+//xzVatWTR4eHmrRooWSkpIc3j927JjatWunEiVKyMvLSzVq1NDnn39+w+e+fvZO1jKz1atXKyQkRFarVQ899JBSUlLyO5SSpJiYGPn5+clms+mZZ57R5cuX7e8FBgYqNjbWoX1YWJjGjx9vf/3nn3+qT58+KlOmjNzd3VWzZk2tWrUqx3v98ccfatiwof7973/r0qVLMgxDU6dOVZUqVeTh4aE6deroww8/lCQlJSWpRYsWkqQSJUrIYrGoe/fuOfabnp6utLQ0hwMAAAAAbgYzloC7QEpKiqKiojR16lT95z//0dmzZ7Vp0yYZhqF58+Zp3Lhxeu2111S3bl3t3r1bvXv3lpeXl7p166Zz584pPDxc5cuX16effip/f3/t2rVLmZmZkqSPP/5YgwYNUmxsrFq2bKlVq1apR48eqlChgj3gkP4KYqZOnapp06Zp9uzZeuqpp3Ts2DGVLFlSx48fV4cOHRQdHa2+ffsqISFBzz//vMMzPPvss7p8+bK+/vpreXl5ad++fbJarabG48KFC5o+fbqWLFkiJycnde7cWUOHDtXSpUvzdf26devk7u6uDRs2KCkpST169JCvr68mTZqUr+szMzPVunVrnT17Vu+++67uuece7du3T87Oztna/vzzz/rXv/6lBg0a6J133pGLi4tGjRql5cuXa86cOapataq+/vprde7cWaVLl1bTpk310Ucf6dFHH9WBAwdks9nk4eGRYx1TpkxRTExMvmoGAAAAgJwQLAF3gZSUFF29elUdOnRQpUqVJEm1atWSJE2cOFEzZsxQhw4dJEmVK1fWvn379Oabb6pbt25677339Pvvv2vHjh0qWbKkJCkoKMje9/Tp09W9e3f169dPkvTcc89p27Ztmj59ukOw1L17d0VFRUmSJk+erNmzZ+vbb7/VQw89pDlz5qhKlSqaOXOmLBaLgoOD9f333+vll1+2X5+cnKxHH33UXneVKlVMj8eVK1c0d+5c3XPPPZKk/v37a8KECfm+3tXVVe+88448PT1Vo0YNTZgwQcOGDdPEiRPl5HTjiaBfffWVvv32W+3fv1/VqlWTlPPzHDx4UK1atdIjjzyiWbNmyWKx6Pz583rllVe0fv16NWrUyH7t5s2b9eabbyo8PNz+Ofn5+eW5CfjIkSP13HPP2V+npaUpICAg3+MAAAAAAARLwF2gTp06evDBB1WrVi1FRkbqX//6lx577DFdvXpVx48fV69evdS7d297+6tXr9o3u05MTFTdunXtYcX19u/frz59+jica9KkiWbNmuVwrnbt2vafvby85O3trZMnT9r7uP/++2WxWOxtskKTLAMHDlTfvn21Zs0atWzZUo8++qhDnzfD09PTHipJUtmyZe215EedOnXk6enpUOu5c+d0/Phxe3CXl8TERFWoUMEeKuXk4sWLatq0qaKiohzGct++fbp06ZJatWrl0P7y5cuqW7duvp9Bktzc3OTm5nZT1wAAAADAtdhjCbgLODs7a+3atfriiy8UGhqq2bNnKzg4WEePHpUkzZs3T4mJifbjhx9+0LZt2yQp12VU17o2EJIkwzCynStWrFi2a7KW0xmGccN7PP300zp69Ki6dOmi77//Xg0aNNDs2bNveF1OcqolPzXcSNYzOzk5ZevvypUr9p/zM6Zubm5q2bKlPvvsM/3888/281lj9tlnnzl8Zvv27bPvswQAAAAA/xSCJeAuYbFY1KRJE8XExGj37t1ydXXVli1bVL58eR09elRBQUEOR+XKlSX9NdMoMTFRp0+fzrHfkJAQbd682eHc1q1bFRISku/aQkND7UFWlutfS1JAQICio6O1fPlyPf/885o3b16+71GQ9uzZo4sXL9pfb9u2TVarVRUqVJAklS5d2mEz8LS0NP3000/217Vr19bPP/+sgwcP5noPJycnLVmyRPXr19cDDzygEydOSPprrNzc3JScnJztM8taxubq6ipJysjIKLiHBgAAAIAcsBQOuAts375d69at07/+9S/5+flp+/bt+v333xUSEqLx48dr4MCBstlsat26tdLT05WQkKAzZ87oueeeU1RUlCZPnqz27dtrypQpKlu2rHbv3q1y5cqpUaNGGjZsmDp27Kh69erpwQcf1MqVK7V8+XJ99dVX+a4vOjpaM2bM0HPPPadnnnlGO3fu1MKFCx3aDB48WK1bt1a1atV05swZrV+//qbCq4J0+fJl9erVS6NHj9axY8c0btw49e/f376/0gMPPKCFCxfav8VuzJgxDhtzh4eHq3nz5nr00Uf1yiuvKCgoSD/++KMsFoseeugheztnZ2ctXbpUUVFReuCBBxQfHy9/f38NHTpUQ4YMUWZmppo2baq0tDRt3bpVVqtV3bp1U6VKlWSxWLRq1Sq1adNGHh4epjc6BwAAAIC8MGMJuAvYbDZ9/fXXatOmjapVq6bRo0drxowZat26tZ5++mnNnz9fCxcuVK1atRQeHq6FCxfaZyy5urpqzZo18vPzU5s2bVSrVi299NJL9qCkffv2mjVrlqZNm6YaNWrozTff1IIFCxQREZHv+ipWrKiPPvpIK1euVJ06dTR37lxNnjzZoU1GRoaeffZZhYSE6KGHHlJwcLDeeOONAhujm/Hggw+qatWqat68uTp27Kh27dpp/Pjx9vdHjhyp5s2b6+GHH1abNm3Uvn17hz2dJOmjjz7Svffeq6ioKIWGhmr48OE5zjBycXHRsmXLVKNGDT3wwAM6efKkJk6cqLFjx2rKlCkKCQlRZGSkVq5caf/Mypcvr5iYGL3wwgsqU6aM+vfvf0vHAwAAAMDdy2IUxMYiAIDbXlpamnx8fBQwOE5Obp43vqAAJb3U9h+9HwAAAIDcZf1tkJqaKpvNlmdbZiwBAAAAAADAFIIlALe16OhoWa3WHI/o6Oib7i+3vqxWqzZt2nQLngAAAAAAbl9s3g3gtjZhwgQNHTo0x/duNGUzJ4mJibm+V758+ZvuDwAAAADuZARLAG5rfn5+8vPzK7D+goKCCqwvAAAAALjTsRQOAAAAAAAAphAsAQAAAAAAwBSCJQAAAAAAAJhCsAQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmuBR2AQCAouWHmEjZbLbCLgMAAADAbYAZSwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgAAAAAAAGAKwRIAAAAAAABMIVgCAAAAAACAKQRLAAAAAAAAMIVgCQAAAAAAAKYQLAEAAAAAAMAUgiUAAAAAAACYQrAEAAAAAAAAUwiWAAAAAAAAYArBEgAAAAAAAEwhWAIAAAAAAIApBEsAAAAAAAAwhWAJAAAAAAAAphAsAQAAAAAAwBSCJQAAAAAAAJhCsAQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgAAAAAAAGAKwRIAAAAAAABMIVgCAAAAAACAKQRLAAAAAAAAMIVgCQAAAAAAAKYQLAEAAAAAAMAUgiUAAAAAAACYQrAEAAAAAAAAUwiWAAAAAAAAYArBEgAAAAAAAEwhWAIAAAAAAIApLoVdAACgaKk5brWc3DwLvN+kl9oWeJ8AAAAAChczlgAAAAAAAGAKwRIAAAAAAABMIVgCAAAAAACAKQRLAAAAAAAAMIVgCQAAAAAAAKYQLAEAAAAAAMAUgiUAAAAAAACYQrAEAAAAAAAAUwiWAAAAAAAAYArBEgAAAAAAAEwhWAIAAAAAAIApBEsAAAAAAAAwhWAJAAAAAAAAphAsAQAAAAAAwBSCJQAAAAAAAJhCsAQANyEiIkKDBw/Os43FYtGKFSvsr3/88Ufdf//9cnd3V1hY2C2tDwAAAAD+SQRLAFDAUlJS1Lp1a/vrcePGycvLSwcOHNC6deu0cOFCFS9evMDv++9//1sVK1aUu7u7ypYtqy5duujEiRMFfh8AAAAAyEKwBAAFzN/fX25ubvbXR44cUdOmTVWpUiWVKlXqlt23RYsWiouL04EDB/TRRx/pyJEjeuyxx27Z/QAAAACAYAlAoVu5cqWKFy+uzMxMSVJiYqIsFouGDRtmb/PMM88oKipKf/zxh6KiolShQgV5enqqVq1aWrZsmUN/ERER6t+/v/r376/ixYurVKlSGj16tAzDyFc9b7zxhqpWrSp3d3eVKVMmWziTmZmp4cOHq2TJkvL399f48eMd3r92KZzFYtHOnTs1YcIEWSwWRUREqEePHkpNTZXFYpHFYsl2fU4CAwM1efJk9ezZU97e3qpYsaLeeusthzZDhgzR/fffr0qVKqlx48Z64YUXtG3bNl25ciVfzw0AAAAAN4tgCUCha968uc6ePavdu3dLkjZu3ChfX19t3LjR3iY+Pl7h4eG6dOmS6tevr1WrVumHH35Qnz591KVLF23fvt2hz0WLFsnFxUXbt2/Xq6++qpkzZ2r+/Pk3rCUhIUEDBw7UhAkTdODAAX355Zdq3rx5tr69vLy0fft2TZ06VRMmTNDatWtz7C8lJUU1atTQ888/r5SUFH366aeKjY2VzWZTSkqKUlJSNHTo0HyN04wZM9SgQQPt3r1b/fr1U9++ffXjjz/m2Pb06dNaunSpGjdurGLFiuXYJj09XWlpaQ4HAAAAANwMgiUAhc7Hx0dhYWGKj4+X9FeINGTIEO3Zs0dnz57Vr7/+qoMHDyoiIkLly5fX0KFDFRYWpipVqmjAgAGKjIzU//73P4c+AwICNHPmTAUHB+upp57SgAEDNHPmzBvWkpycLC8vLz388MOqVKmS6tatq4EDBzq0qV27tsaNG6eqVauqa9euatCggdatW5djf/7+/nJxcZHVapW/v79sNpt8fHxksVjk7+8vf39/Wa3WfI1TmzZt1K9fPwUFBWnEiBHy9fW1j1mWESNGyMvLS6VKlVJycrI++eSTXPubMmWKfHx87EdAQEC+6gAAAACALARLAIqEiIgIxcfHyzAMbdq0SY888ohq1qypzZs3a8OGDSpTpoyqV6+ujIwMTZo0SbVr11apUqVktVq1Zs0aJScnO/R3//33y2Kx2F83atRIhw4dUkZGRp51tGrVSpUqVVKVKlXUpUsXLV26VBcuXHBoU7t2bYfXZcuW1cmTJ//mCNzYtffNCqauv++wYcO0e/durVmzRs7OzuratWuuSwBHjhyp1NRU+3H8+PFbWj8AAACAO49LYRcAANJfwdLbb7+tPXv2yMnJSaGhoQoPD9fGjRt15swZhYeHS/prOdjMmTMVGxurWrVqycvLS4MHD9bly5cLpA5vb2/t2rVL8fHxWrNmjcaOHavx48drx44d9m9yu35pmcVise8PdSvl576+vr7y9fVVtWrVFBISooCAAG3btk2NGjXK1p+bm5vDJuMAAAAAcLOYsQSgSMjaZyk2Nlbh4eGyWCwKDw9XfHy8fX8lSfbZTJ07d1adOnVUpUoVHTp0KFt/27Zty/a6atWqcnZ2vmEtLi4uatmypaZOnarvvvtOSUlJWr9+fcE8qCRXV9cbzpwqCFkzldLT02/5vQAAAADcnQiWABQJWfssvfvuu4qIiJD0V9i0a9cu+/5KkhQUFKS1a9dq69at2r9/v5555hn9+uuv2fo7fvy4nnvuOR04cEDLli3T7NmzNWjQoBvWsWrVKr366qtKTEzUsWPHtHjxYmVmZio4OLjAnjUwMFDnzp3TunXrdOrUqWxL7cz49ttv9dprr9nr3rBhg5588kndc889Oc5WAgAAAICCQLAEoMho0aKFMjIy7CFSiRIlFBoaqtKlSyskJESSNGbMGNWrV0+RkZGKiIiQv7+/2rdvn62vrl276uLFi2rYsKGeffZZDRgwQH369LlhDcWLF9fy5cv1wAMPKCQkRHPnztWyZctUo0aNAnvOxo0bKzo6Wp06dVLp0qU1derUv92nh4eHli9frgcffFDBwcHq2bOnatasqY0bN7LcDQAAAMAtYzFy29UVAG5TERERCgsLU2xsbGGXcltJS0v769vhBsfJyc2zwPtPeqltgfcJAAAAoOBl/W2Qmpoqm82WZ1tmLAEAAAAAAMAUgiUAd5VNmzbJarXmelATAAAAAOSfS2EXAAAFLT4+Ptf3GjRooMTExH+slvwoijUBAAAAQH4QLAG4q3h4eCgoKKiwy3BQFGsCAAAAgPxgKRwAAAAAAABMIVgCAAAAAACAKQRLAAAAAAAAMIVgCQAAAAAAAKYQLAEAAAAAAMAUgiUAAAAAAACYQrAEAAAAAAAAUwiWAAAAAAAAYArBEgAAAAAAAExxKewCAABFyw8xkbLZbIVdBgAAAIDbADOWAAAAAAAAYArBEgAAAAAAAEwhWAIAAAAAAIApBEsAAAAAAAAwhWAJAAAAAAAAphAsAQAAAAAAwBSCJQAAAAAAAJhCsAQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmEKwBAAAAAAAAFMIlgAAAAAAAGAKwRIAAAAAAABMIVgCAAAAAACAKQRLAAAAAAAAMIVgCQAAAAAAAKYQLAEAAAAAAMAUgiUAAAAAAACYQrAEAAAAAAAAUwiWAAAAAAAAYArBEgAAAAAAAEwhWAIAAAAAAIApBEsAAAAAAAAwhWAJAAAAAAAAphAsAQAAAAAAwBSCJQAAAAAAAJhCsAQAAAAAAABTCJYAAAAAAABgCsESAAAAAAAATCFYAgAAAAAAgCkESwAAAAAAADCFYAkAAAAAAACmECwBAAAAAADAFIIlAAAAAAAAmGI6WFqyZImaNGmicuXK6dixY5Kk2NhYffLJJwVWHAAAAAAAAIouFzMXzZkzR2PHjtXgwYM1adIkZWRkSJKKFy+u2NhYPfLIIwVaJADgn1Nz3Go5uXnmq23SS21vcTUAAAAAijJTM5Zmz56tefPmadSoUXJ2drafb9Cggb7//vsCKw4AAAAAAABFl6lg6aefflLdunWznXdzc9P58+f/dlEAAAAAAAAo+kwFS5UrV1ZiYmK281988YVCQ0P/bk0AAAAAAAC4DZjaY2nYsGF69tlndenSJRmGoW+//VbLli3TlClTNH/+/IKuEQAAAAAAAEWQqWCpR48eunr1qoYPH64LFy7oySefVPny5TVr1iw98cQTBV0jAAAAAAAAiqCbDpauXr2qpUuXql27durdu7dOnTqlzMxM+fn53Yr6AAAAAAAAUETd9B5LLi4u6tu3r9LT0yVJvr6+hEoAAAAAAAB3IVObd993333avXt3QdcCAAAAAACA24ipPZb69eun559/Xj///LPq168vLy8vh/dr165dIMUBAAAAAACg6DIVLHXq1EmSNHDgQPs5i8UiwzBksViUkZFRMNUBAAAAAACgyDIVLP30008FXQcAAAAAAABuM6aCpUqVKhV0HQAAAAAAALjNmAqWFi9enOf7Xbt2NVUMAAAAAAAAbh+mgqVBgwY5vL5y5YouXLggV1dXeXp6EiwBAAAAAADcBZzMXHTmzBmH49y5czpw4ICaNm2qZcuWFXSNAAAAAAAAKIJMBUs5qVq1ql566aVss5n+jsDAQMXGxua7/fjx4xUWFva372uxWLRixYq/3U9ukpKSZLFYlJiYKEmKj4+XxWLRn3/+KUlauHChihcvfsvuj+wiIiI0ePDgv9VHUf/cxo8frzJlytzy3+/bWffu3dW+ffvCLgMAAAAAbhsFFixJkrOzs06cOFGQXd6VOnXqpIMHDxZ4v8uXL1dkZKR8fX0dgq1r/frrr+rSpYv8/f3l5eWlevXq6cMPPyzwWoqa5cuXa+LEibf8PikpKXryyScVHBwsJyenvx1m5df+/fsVExOjN998UykpKWrduvU/ct/bzaxZs7Rw4cLCLgMAAAAAbhum9lj69NNPHV4bhqGUlBS99tpratKkSYEUdru6cuWKihUr9rf68PDwkIeHRwFV9H/Onz+vJk2a6PHHH1fv3r1zbNOlSxelpqbq008/la+vr9577z116tRJCQkJqlu3boHXVFSULFnyH7lPenq6SpcurVGjRmnmzJl/u7/Lly/L1dX1hu2OHDkiSXrkkUdksVhM368gfr+LooyMDFksFvn4+BR2KQAAAABwWzE1Y6l9+/YOR4cOHTR+/HjVrl1b77zzTr77OXv2rJ566il5eXmpbNmymjlzZp5LkpKTk/XII4/IarXKZrOpY8eO+u2337K1e/PNNxUQECBPT089/vjj9iVmkrRjxw61atVKvr6+8vHxUXh4uHbt2nWzQyDp/5a0xcXFKSIiQu7u7nr33XeVmZmpCRMmqEKFCnJzc1NYWJi+/PLLfPd7/ZKqrCV+S5YsUWBgoHx8fPTEE0/o7Nmz9jb5GcsuXbpo7NixatmyZa73/uabbzRgwAA1bNhQVapU0ejRo1W8ePF8jdG149GsWTN5eHjo3nvv1cGDB7Vjxw41aNBAVqtVDz30kH7//XeHaxcsWKCQkBC5u7urevXqeuONNxzeHzFihKpVqyZPT09VqVJFY8aM0ZUrV25qjPJy/VgFBgbqxRdfVNeuXWW1WlWpUiV98skn+v333+2/g7Vq1VJCQkK2vlasWKFq1arJ3d1drVq10vHjxx36nTVrlrp27WoqxMhaqjVlyhSVK1dO1apVkyT98ssv6tSpk0qUKKFSpUrpkUceUVJSkn1s2rVrJ0lycnJyCJbyGvfcfr/ze93y5cvVokULeXp6qk6dOvrmm28cnmXLli0KDw+Xp6enSpQoocjISJ05c0bSX2H11KlTVaVKFXl4eKhOnTr5njmXtbT0s88+U506deTu7q777rtP33//vb1N1r+xVatWKTQ0VG5ubjp27Fi2pXCZmZl6+eWXFRQUJDc3N1WsWFGTJk2yv5/XuOfl+++/l5OTk06dOiXpr33rnJyc9Pjjj9vbTJkyRY0aNZL0V/DVq1cvVa5cWR4eHgoODtasWbMc+syqPSYmRn5+frLZbHrmmWd0+fLlfI0bAAAAAJhhKljKzMx0ODIyMvTrr7/qvffeU9myZfPdz3PPPactW7bo008/1dq1a7Vp06ZcAwzDMNS+fXudPn1aGzdu1Nq1a3XkyBF16tTJod3hw4cVFxenlStX6ssvv1RiYqKeffZZ+/tnz55Vt27dtGnTJm3btk1Vq1ZVmzZt8h1A5GTEiBEaOHCg9u/fr8jISM2aNUszZszQ9OnT9d133ykyMlL//ve/dejQIdP3OHLkiFasWKFVq1Zp1apV2rhxo1566SX7+zczlnlp2rSpPvjgA50+fVqZmZl6//33lZ6eroiIiHz3MW7cOI0ePVq7du2Si4uLoqKiNHz4cM2aNUubNm3SkSNHNHbsWHv7efPmadSoUZo0aZL279+vyZMna8yYMVq0aJG9jbe3txYuXKh9+/Zp1qxZmjdvXrYZPzcao5s1c+ZMNWnSRLt371bbtm3VpUsXde3aVZ07d9auXbsUFBSkrl27yjAM+zUXLlzQpEmTtGjRIm3ZskVpaWl64oknTNeQk3Xr1mn//v1au3atVq1apQsXLqhFixayWq36+uuvtXnzZnuAd/nyZQ0dOlQLFiyQ9NdSvJSUFEn5G3cp++93fq8bNWqUhg4dqsTERFWrVk1RUVG6evWqJCkxMVEPPvigatSooW+++UabN29Wu3btlJGRIUkaPXq0FixYoDlz5mjv3r0aMmSIOnfurI0bN+Z7nIYNG6bp06drx44d8vPz07///W+HMPLChQuaMmWK5s+fr71798rPzy9bHyNHjtTLL7+sMWPGaN++fXrvvfdUpkwZ+/V5jXteatasqVKlStmf5+uvv1apUqX09ddf29vEx8crPDxc0l//nVuhQgXFxcVp3759Gjt2rP773/8qLi7Ood+s340NGzZo2bJl+vjjjxUTE5NrHenp6UpLS3M4AAAAAOCmGCbExMQY58+fz3b+woULRkxMTL76SEtLM4oVK2b873//s5/7888/DU9PT2PQoEGGYRhGpUqVjJkzZxqGYRhr1qwxnJ2djeTkZHv7vXv3GpKMb7/91jAMwxg3bpzh7OxsHD9+3N7miy++MJycnIyUlJQc67h69arh7e1trFy50n5OkvHxxx/f8Bl++uknQ5IRGxvrcL5cuXLGpEmTHM7de++9Rr9+/Ryu2717t2EYhrFhwwZDknHmzBnDMAxjwYIFho+Pj/3acePGGZ6enkZaWpr93LBhw4z77rvPMIz8jWVOdWfd/1p//vmnERkZaUgyXFxcDJvNZqxZs+aGY3Ftv/Pnz7efW7ZsmSHJWLdunf3clClTjODgYPvrgIAA47333nPoa+LEiUajRo1yvdfUqVON+vXr21/faIxuJDw83GGsKlWqZHTu3Nn+OiUlxZBkjBkzxn7um2++MSTZf7cWLFhgSDK2bdtmb7N//35DkrF9+/Yb3jM/unXrZpQpU8ZIT0+3n3v77beN4OBgIzMz034uPT3d8PDwMFavXm0YhmF8/PHHxvX/3G807rn9fuf3umt/D7L+re7fv98wDMOIiooymjRpkuMznjt3znB3dze2bt3qcL5Xr15GVFRULiPzf7L+Pb3//vv2c3/88Yfh4eFhfPDBB4Zh/N9nlZiY6HBtt27djEceecQwjL/+Xbm5uRnz5s3L8T75Gfe8dOjQwejfv79hGIYxePBg4/nnnzd8fX2NvXv3GleuXDGsVqvxxRdf5Hp9v379jEcffdSh9pIlSzr8d/OcOXMMq9VqZGRk5NjHuHHjDEnZjoDBcUalEavydQAAAAC486SmphqSjNTU1Bu2NTVjKSYmRufOnct2/sKFC3n+r+PXOnr0qK5cuaKGDRvaz/n4+Cg4ODjH9vv371dAQIACAgLs50JDQ1W8eHHt37/ffq5ixYqqUKGC/XWjRo2UmZmpAwcOSJJOnjyp6OhoVatWTT4+PvLx8dG5c+eUnJycr7pz0qBBA/vPaWlpOnHiRLa9ppo0aeJQ580KDAyUt7e3/XXZsmV18uRJSTc/lnkZPXq0zpw5o6+++koJCQl67rnn9PjjjzssI7qR2rVr23/Omt1Rq1Yth3NZtf/+++86fvy4evXqJavVaj9efPFF+75AkvThhx+qadOm8vf3l9Vq1ZgxY7J9ZnmNkRn5eQ5JDvdwcXFx+H2oXr16tt/Rv6tWrVoO+yrt3LlThw8flre3t338SpYsqUuXLjmM4bXyO+6S4+/3zVx37fhlzWTMGqusGUs52bdvny5duqRWrVo53GPx4sW5Pk9OspaRSX/toRUcHOzwObi6ujrUeL39+/crPT091zrNjPu1IiIiFB8fL0nauHGjWrRooebNm2vjxo3asWOHLl686PDfI3PnzlWDBg1UunRpWa1WzZs3L9u/gTp16sjT09NhDM6dO+ewHPNaI0eOVGpqqv3IrR0AAAAA5MbU5t2GYeS4AfCePXvyvQmy8f+XD13fj3HNsqL83DO381my3sv6z+7du+v3339XbGysKlWqJDc3NzVq1Ohv7UPi5eWV633zW+eNXL9hssViUWZmpr3v3O55M44cOaLXXntNP/zwg2rUqCHprz9UN23apNdff11z58696Vqzarr+XFbtWf85b9483XfffQ79ODs7S5K2bdumJ554QjExMYqMjJSPj4/ef/99zZgxI9f7Xn8fM/LzHNc+w/Xnb3TOrOt/3zIzM1W/fn0tXbo0W9vSpUvn2Ed+xj2n+93MdXmNVV6b02e1+eyzz1S+fHmH99zc3HK9Lj+u/Rw8PDzy/FxutIG+mXG/VkREhAYNGqTDhw/rhx9+ULNmzXTkyBFt3LhRf/75p+rXr28PSuPi4jRkyBDNmDFDjRo1kre3t6ZNm6bt27ff8D5S7r9/bm5uf3tMAQAAANzdbipYKlGihCwWiywWi6pVq+bwx0pGRobOnTun6OjofPV1zz33qFixYvr222/ts5DS0tJ06NAh+74i1woNDVVycrKOHz9ub79v3z6lpqYqJCTE3i45OVknTpxQuXLlJP21GbWTk5N9k+NNmzbpjTfeUJs2bSRJx48ft2+gWxBsNpvKlSunzZs3q3nz5vbzW7dudZhRVJBudixzc+HCBUl/bfB8LWdn578V0OSlTJkyKl++vI4ePaqnnnoqxzZbtmxRpUqVNGrUKPu5Y8eO3ZJ6/q6rV68qISHB/lkfOHBAf/75p6pXr37L7lmvXj198MEH9g2b8yM/416Q112vdu3aWrduXY4zHLM2005OTr6p39/rbdu2TRUrVpT01+bYBw8evKnPoWrVqvLw8NC6dev09NNPZ3vfzLhfK2ufpRdffFF16tSRzWZTeHi4pkyZojNnzjg8+6ZNm9S4cWP169fPfi6nWVF79uzRxYsX7aHYtm3bZLVaHWZxAgAAAEBBuqlgKTY2VoZhqGfPnoqJiXH4VitXV1cFBgY6LD/Ji7e3t7p166Zhw4apZMmS8vPz07hx47J9a1WWli1bqnbt2nrqqacUGxurq1evql+/fgoPD3dYquPu7q5u3bpp+vTpSktL08CBA9WxY0f5+/tLkoKCgrRkyRI1aNBAaWlpGjZs2A1nJtysYcOGady4cbrnnnsUFhamBQsWKDExMceZDQUhv2N5+vRpe/Amyb480N/fX/7+/qpevbqCgoL0zDPPaPr06SpVqpRWrFhh3yT6Vhk/frwGDhwom82m1q1bKz09XQkJCTpz5oyee+45BQUFKTk5We+//77uvfdeffbZZ/r4449vWT1/R7FixTRgwAC9+uqrKlasmPr376/777/fIVRMTEyUJJ07d06///67EhMT5erqqtDQUFP3fOqppzRt2jQ98sgj9m8jTE5O1vLlyzVs2LBcQ4UbjXtuzF53rZEjR6pWrVrq16+foqOj5erqqg0bNujxxx+Xr6+vhg4dqiFDhigzM1NNmzZVWlqatm7dKqvVqm7duuXrHhMmTFCpUqVUpkwZjRo1Sr6+vg7f+HYj7u7uGjFihIYPHy5XV1c1adJEv//+u/bu3atevXqZHvcsFotFzZs317vvvqshQ4ZI+itwu3z5statW6dBgwbZ2wYFBWnx4sVavXq1KleurCVLlmjHjh2qXLmyQ5+XL19Wr169NHr0aB07dkzjxo1T//79s4XFAAAAAFBQbipYyvqDrnLlymrcuHG2pUc365VXXlF0dLQefvhh2Ww2DR8+XMePH5e7u3u2thaLRStWrNCAAQPUvHlzOTk56aGHHtLs2bMd2gUFBalDhw5q06aNTp8+rTZt2jh8Ffo777yjPn36qG7duqpYsaImT56soUOH/q3nuN7AgQOVlpam559/XidPnlRoaKg+/fRTVa1atUDvc638jOWnn36qHj162F9nfVvZuHHjNH78eBUrVkyff/65XnjhBbVr107nzp1TUFCQFi1aZJ/hdSs8/fTT8vT01LRp0zR8+HB5eXmpVq1aGjx4sCTpkUce0ZAhQ9S/f3+lp6erbdu2GjNmjMaPH3/LajLL09NTI0aM0JNPPqmff/5ZTZs21TvvvOPQpm7duvafd+7cqffee0+VKlXK19fU53bPr7/+WiNGjFCHDh109uxZlS9fXg8++GCeM2luNO4Ffd21qlWrpjVr1ui///2vGjZsKA8PD913332KioqSJE2cOFF+fn6aMmWKjh49quLFi6tevXr673//m+97vPTSSxo0aJAOHTqkOnXq6NNPP3XYmyo/xowZIxcXF40dO1YnTpxQ2bJl7bMyzY77tVq0aKHly5fbv3XRYrGoWbNmWrVqlZo2bWpvFx0drcTERHXq1EkWi0VRUVHq16+fvvjiC4f+HnzwQVWtWlXNmzdXenq6nnjiiSL57wQAAADAncNi3OxGPNe5ePGiw1d4SzK1LESSzp8/r/Lly2vGjBnq1avX3ynrrsdY4m4VHx+vFi1a6MyZMypevHhhl/OP6d69u/7880+tWLHCdB9paWny8fFRwOA4Obl53vgCSUkvtTV9PwAAAABFU9bfBqmpqTfMeExt3n3hwgUNHz5ccXFx+uOPP7K9n5GRka9+du/erR9//FENGzZUamqqJkyYIOmvGSq4OYwlAAAAAAD4p5naeGPYsGFav3693njjDbm5uWn+/PmKiYlRuXLltHjx4pvqa/r06apTp45atmyp8+fPa9OmTfL19TVTVoGbPHmyw9edX3u0bt26sMvL5laP5e02HlmSk5NzrdtqtWb7yvbClFedmzZtKuzyioTo6Ohcxyi/Xx7wT+CzBAAAAHA3MLUUrmLFilq8eLEiIiJks9m0a9cu+6bYy5Yt0+eff34rav3HnT59WqdPn87xPQ8Pj2xfhX6nu13H4+rVq3nuXxQYGCgXF1OT9wrc4cOHc32vfPnyBb7R/O3o5MmTSktLy/E9m80mPz+/f7iinN2OnyVL4QAAAABI/8BSuNOnT9u/jchms9nDhqZNm6pv375muiySSpYsqZIlSxZ2GUXG7ToeLi4uCgoKKuwy8uV2qbMw+fn5FZnwKC98lgAAAADuBqaWwlWpUsU+AyQ0NFRxcXGSpJUrV95Vm+UCAAAAAADczUwFSz169NCePXskSSNHjrTvtTRkyBANGzasQAsEAAAAAABA0WRqKdyQIUPsP7do0UI//vijEhISdM8996hOnToFVhwAAAAAAACKrr+9Y/GlS5dUsWJFVaxYsSDqAQAAAAAAwG3C1FK4jIwMTZw4UeXLl5fVatXRo0clSWPGjNHbb79doAUCAAAAAACgaDIVLE2aNEkLFy7U1KlT5erqaj9fq1YtzZ8/v8CKAwAAAAAAQNFlKlhavHix3nrrLT311FNydna2n69du7Z+/PHHAisOAAAAAAAARZepYOmXX35RUFBQtvOZmZm6cuXK3y4KAAAAAAAARZ+pYKlGjRratGlTtvP/+9//VLdu3b9dFAAAAAAAAIo+U98KN27cOHXp0kW//PKLMjMztXz5ch04cECLFy/WqlWrCrpGAMA/6IeYSNlstsIuAwAAAMBt4KZmLB09elSGYahdu3b64IMP9Pnn/6+9O4/Xoqz/x/86gLKfg2JsiqCCCLKpqOEGKoaa5lJqaIhLLikq7h9zAdxQS8WlrDTFzCXL5aPmRiqGu4DHlZBQhD5hmCKLCyrcvz/8cv86sggjejCfz8djHp655ppr3jP3QJ0X18x9byoqKnLWWWdl4sSJufvuu7PTTjt9WbUCAAAAsApZoRlLHTt2zIwZM9KiRYv0798/1157bf7+97+nVatWX1Z9AAAAAKyiVmjGUqlUqrF+33335f3331+pBQEAAADw9VDo5d2LfDZoAgAAAOCbY4WCpYqKilRUVCzWBgAAAMA3zwq9Y6lUKuWggw5K/fr1kyQffvhhjjzyyDRu3LhGv9tvv33lVQgAAADAKmmFgqVBgwbVWP/Rj360UosBAAAA4OtjhYKl66677suqAwAAAICvmS/08m4AAAAAvrkESwAAAAAUIlgCAAAAoBDBEgAAAACFCJYAAAAAKESwBAAAAEAhgiUAAAAAChEsAQAAAFCIYAkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKAQwRIAAAAAhQiWAAAAAChEsAQAAABAIYIlAAAAAAoRLAEAAABQiGAJAAAAgEIESwAAAAAUIlgCAAAAoBDBEgAAAACFCJYAAAAAKESwBAAAAEAhgiUAAAAAChEsAQAAAFBIvdouAIBVS9ehD6RO/Uaf22/qBd/9CqoBAABWZWYsAQAAAFCIYAkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKAQwRIAAAAAhQiWAAAAAChEsAQAAABAIYIlAAAAAAoRLAEAAABQiGAJAAAAgEIESwC14O67706zZs2ycOHCJEl1dXUqKipy8sknl/scccQRGTBgQN5+++0MGDAg66yzTho1apRu3brl5ptvrjFe3759M3jw4AwePDjNmjVL8+bNc8YZZ6RUKn2l5wUAAHyzCJYAasF2222XuXPn5rnnnkuSPProo1lrrbXy6KOPlvuMGTMmffr0yYcffpjNNtss99xzT1566aUcfvjhGThwYJ5++ukaY15//fWpV69enn766Vx++eW59NJLc8011yy1hvnz52fOnDk1FgAAgBUhWAKoBVVVVenZs2fGjBmT5NMQ6fjjj8/zzz+fuXPn5s0338yrr76avn37Zu21185JJ52Unj17Zv31188xxxyT/v37549//GONMdu2bZtLL700nTp1ygEHHJBjjjkml1566VJrGDFiRKqqqspL27Ztv8xTBgAA/gsJlgBqSd++fTNmzJiUSqWMHTs2e+yxR7p27ZrHHnssjzzySFq2bJmNNtooCxYsyHnnnZfu3bunefPmadKkSR588MFMmzatxnjf/va3U1FRUV7v3bt3Jk+enAULFizx+Keddlpmz55dXqZPn/6lni8AAPDfp15tFwDwTdW3b9/89re/zfPPP586deqkS5cu6dOnTx599NHMmjUrffr0SZJcfPHFufTSSzNy5Mh069YtjRs3zpAhQ/LRRx99oePXr18/9evXXxmnAgAAfEOZsQRQSxa9Z2nkyJHp06dPKioq0qdPn4wZM6b8fqUk5dlMP/rRj9KjR4+sv/76mTx58mLjPfXUU4utd+zYMXXr1v1KzgcAAPjmESwB1JJF71n6/e9/n759+yb5NGyaMGFC+f1KSdKhQ4eMHj06TzzxRCZOnJgjjjgib7755mLjTZ8+PSeccEImTZqUm2++OVdccUWOO+64r/CMAACAbxqPwgHUou233z4TJkwoh0hrrLFGunTpkn/+85/p3LlzkuTMM8/M66+/nv79+6dRo0Y5/PDDs+eee2b27Nk1xjrwwAPzwQcfZIsttkjdunVzzDHH5PDDD/+qTwkAAPgGqSiVSqXaLgKAL6Zv377p2bNnRo4cWXiMOXPmfPrtcENuTZ36jT63/9QLvlv4WAAAwKpr0e8Gs2fPTmVl5TL7ehQOAAAAgEIESwAAAAAU4h1LAP8FxowZU9slAAAA30BmLAEAAABQiGAJAAAAgEIESwAAAAAUIlgCAAAAoBDBEgAAAACFCJYAAAAAKESwBAAAAEAhgiUAAAAAChEsAQAAAFCIYAkAAACAQurVdgEArFpeGt4/lZWVtV0GAADwNWDGEgAAAACFCJYAAAAAKESwBAAAAEAhgiUAAAAAChEsAQAAAFCIYAkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKAQwRIAAAAAhQiWAAAAAChEsAQAAABAIYIlAAAAAAoRLAEAAABQiGAJAAAAgEIESwAAAAAUIlgCAAAAoBDBEgAAAACFCJYAAAAAKESwBAAAAEAhgiUAAAAAChEsAQAAAFCIYAkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKCQerVdAACrlq5DH0id+o2Wun3qBd/9CqsBAABWZWYsAQAAAFCIYAkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKAQwRIAAAAAhQiWAAAAAChEsAQAAABAIYIl+AYaNWpUmjVrttLGq6ioyJ133vmFxjjooIOy5557fqExpk6dmoqKilRXV3+hcVYVK+OaAAAAfJnq1XYBwNffjBkzssYaa9R2Gf91LrvsspRKpdouAwAAYKkES8AX1qpVq9ou4b9SVVVVbZcAAACwTB6Fg1XYn/70p3Tr1i0NGzZM8+bN069fv7z33ntJkmuvvTYbb7xx6tevn9atW2fw4MHl/S655JJ069YtjRs3Ttu2bXPUUUdl3rx5yzzW3Xffnc022ywNGjTI+uuvn+HDh+eTTz5Zrjr/81G4RY+j3Xrrrdl2223TsGHDbL755nn11Vfz7LPPplevXmnSpEl23nnnvPXWW4uNNXz48LRo0SKVlZU54ogj8tFHH5W33X///dlmm23SrFmzNG/ePLvttlumTJmy1LoWLFiQQw89NOutt14aNmyYTp065bLLLqvRZ9HjZj//+c/TunXrNG/ePEcffXQ+/vjjcp/58+fnlFNOSdu2bVO/fv107Ngxv/3tb8vbX3nlley6665p0qRJWrZsmYEDB+bf//73cl27ZX3G//ko3KLr+tmlb9++5bGeeOKJbLfddmnYsGHatm2bY489tjwWAADAl0GwBKuoGTNmZMCAATnkkEMyceLEjBkzJnvvvXdKpVKuuuqqHH300Tn88MPz4osv5q677kqHDh3K+9apUyeXX355XnrppVx//fV5+OGHc8oppyz1WA888EB+9KMf5dhjj80rr7ySX//61xk1alTOO++8wvUPHTo0Z5xxRiZMmJB69eplwIABOeWUU3LZZZdl7NixmTJlSs4666wa+zz00EOZOHFiHnnkkdx888254447Mnz48PL29957LyeccEKeffbZPPTQQ6lTp0722muvLFy4cIk1LFy4MOuss05uvfXWvPLKKznrrLPy05/+NLfeemuNfo888kimTJmSRx55JNdff31GjRqVUaNGlbcfeOCBueWWW3L55Zdn4sSJ+dWvfpUmTZok+fRz6tOnT3r27Jlx48bl/vvvz7/+9a/su+++n3uNlvUZf1bbtm0zY8aM8vLcc8+lefPm2W677ZIkL774Yvr375+99947L7zwQv7whz/kscceqxE4ftb8+fMzZ86cGgsAAMAKKQGrpPHjx5eSlKZOnbrYtjZt2pROP/305R7r1ltvLTVv3ry8ft1115WqqqrK69tuu23p/PPPr7HPDTfcUGrduvVyjZ+kdMcdd5RKpVLp9ddfLyUpXXPNNeXtN998cylJ6aGHHiq3jRgxotSpU6fy+qBBg0prrrlm6b333iu3XXXVVaUmTZqUFixYsMTjzpw5s5Sk9OKLL9Y49nPPPbfUWo866qjS97///RrHbdeuXemTTz4pt+2zzz6l/fbbr1QqlUqTJk0qJSmNHj16ieOdeeaZpe985zs12qZPn15KUpo0adJS6yiVlv0ZL6ptjz32WKz9gw8+KG255Zal3XbbrXxtBg4cWDr88MNr9Bs7dmypTp06pQ8++GCJ4w8dOrSUZLGl7ZBbS+1OvWepCwAA8N9t9uzZpSSl2bNnf25fM5ZgFdWjR4/suOOO6datW/bZZ59cffXVmTVrVmbOnJl//vOf2XHHHZe67yOPPJKddtopa6+9dpo2bZoDDzwwb7/99lIfixo/fnzOPvvsNGnSpLwcdthhmTFjRt5///1C9Xfv3r38c8uWLZMk3bp1q9E2c+bMxc65UaNG5fXevXtn3rx5mT59epJkypQp2X///bP++uunsrIy6623XpJk2rRpS63jV7/6VXr16pVvfetbadKkSa6++urF+m+88capW7dueb1169bl2qqrq1O3bt306dNnieOPHz8+jzzySI1rt9FGG5XrXZalfcaf59BDD83cuXNz0003pU6dOuU6Ro0aVaOO/v37Z+HChXn99deXOM5pp52W2bNnl5dF1xkAAGB5eXk3rKLq1q2b0aNH54knnsiDDz6YK664IqeffnoeeuihZe73xhtvZNddd82RRx6Zc845J2uuuWYee+yxHHrooTXeG/SfFi5cmOHDh2fvvfdebFuDBg0K1b/aaquVf66oqFhi29IeYfusRfvvvvvuadu2ba6++uq0adMmCxcuTNeuXWu8h+k/3XrrrTn++ONz8cUXp3fv3mnatGl+9rOf5emnn15qrZ+trWHDhsusbeHChdl9991z4YUXLratdevWy9x3aZ/x008/XQ7NPuvcc8/N/fffn2eeeSZNmzatUccRRxyRY489drF91l133SWOVb9+/dSvX3+ZNQIAACyLYAlWYRUVFdl6662z9dZb56yzzkq7du0yevTotG/fPg899FC23377xfYZN25cPvnkk1x88cXl2SyffafQZ2266aaZNGlSjfc01Ybnn38+H3zwQTnMeeqpp9KkSZOss846efvttzNx4sT8+te/zrbbbpskeeyxx5Y53tixY7PVVlvlqKOOKrd93iyiz+rWrVsWLlyYRx99NP369Vts+6abbprbbrst7du3T716K/5X6pI+4zvuuCMnnHDCYn1vu+22nH322bnvvvuywQYbLFbHyy+/XOufIQAA8M3iUThYRT399NM5//zzM27cuEybNi2333573nrrrXTu3DnDhg3LxRdfnMsvvzyTJ0/OhAkTcsUVVyRJNthgg3zyySe54oor8tprr+WGG27Ir371q2Ue66yzzsrvfve7DBs2LC+//HImTpyYP/zhDznjjDO+ilMt++ijj3LooYfmlVdeyX333ZehQ4dm8ODBqVOnTtZYY400b948v/nNb/L3v/89Dz/88BLDl//UoUOHjBs3Lg888EBeffXVnHnmmXn22WdXqKb27dtn0KBBOeSQQ3LnnXfm9ddfz5gxY8ph3dFHH5133nknAwYMyDPPPJPXXnstDz74YA455JAsWLBgmWMv6zP+rJdeeikHHnhgTj311Gy88cZ588038+abb+add95Jkpx66ql58sknc/TRR6e6ujqTJ0/OXXfdlWOOOWaFzhcAAGBFCJZgFVVZWZm//vWv2XXXXbPhhhvmjDPOyMUXX5xddtklgwYNysiRI/PLX/4yG2+8cXbbbbdMnjw5SdKzZ89ccsklufDCC9O1a9fceOONGTFixDKP1b9//9xzzz0ZPXp0Nt9883z729/OJZdcknbt2n0Vp1q24447pmPHjtluu+2y7777Zvfdd8+wYcOSfPpNd7fcckvGjx+frl275vjjj8/PfvazZY535JFHZu+9985+++2XLbfcMm+//XaN2UvL66qrrsoPfvCDHHXUUdloo41y2GGHld9X1aZNmzz++ONZsGBB+vfvn65du+a4445LVVVVecbY0izrM/6scePG5f3338+5556b1q1bl5dFjy927949jz76aCZPnpxtt902m2yySc4888zPfRwPAADgi6golZbwvdYAfOPMmTMnVVVVaTvk1tSp32ip/aZe8N2vsCoAAOCrtuh3g9mzZ6eysnKZfc1YAgAAAKAQwRKwTDfeeGONr7D/z2XjjTeu7fJWadOmTVvqtWvSpEmmTZtW2yUCAAB8Ib4VDlim733ve9lyyy2XuG211Vb7iqv5emnTpk2qq6uXuR0AAODrTLAELFPTpk3TtGnT2i7ja6levXrp0KFDbZcBAADwpfEoHAAAAACFCJYAAAAAKESwBAAAAEAhgiUAAAAAChEsAQAAAFCIYAkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAopF5tFwDAquWl4f1TWVlZ22UAAABfA2YsAQAAAFCIYAkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKAQwRIAAAAAhQiWAAAAAChEsAQAAABAIYIlAAAAAAoRLAEAAABQiGAJAAAAgEIESwAAAAAUIlgCAAAAoBDBEgAAAACFCJYAAAAAKESwBAAAAEAhgiUAAAAAChEsAQAAAFCIYAkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKAQwRIAAAAAhQiWAAAAAChEsAQAAABAIYIlAAAAAAqpV9sFALBq6Tr0gdSp32ip26de8N2vsBoAAGBVZsYSAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKAQwRIAAAAAhQiWAAAAAChEsAQAAABAIYIlAAAAAAoRLAEAAABQiGAJAAAAgEIESwC1qKKiInfeeWehfdu3b5+RI0d+aeMDAAB8nnq1XQDA18FBBx2Ud999d5UKaZ599tk0bty4tssAAAC+wQRLAF9T3/rWt2q7BAAA4BvOo3AA/+FPf/pTunXrloYNG6Z58+bp169fTj755Fx//fX53//931RUVKSioiJjxoxJkrz44ovZYYcdyv0PP/zwzJs3r8aY1157bTbeeOPUr18/rVu3zuDBg5d6/LPPPjstW7ZMdXX159b62UfhJk+enO222y4NGjRIly5dMnr06CKXAAAAYLmZsQTw/8yYMSMDBgzIRRddlL322itz587N2LFjc+CBB2batGmZM2dOrrvuuiTJmmuumffffz8777xzvv3tb+fZZ5/NzJkz8+Mf/ziDBw/OqFGjkiRXXXVVTjjhhFxwwQXZZZddMnv27Dz++OOLHbtUKmXIkCG5884789hjj6Vjx44rVPvChQuz9957Z6211spTTz2VOXPmZMiQIcvcZ/78+Zk/f355fc6cOSt0TAAAAMESwP8zY8aMfPLJJ9l7773Trl27JEm3bt2SJA0bNsz8+fPTqlWrcv/rr78+H3zwQX73u9+V33V05ZVXZvfdd8+FF16Yli1b5txzz82JJ56Y4447rrzf5ptvXuO4n3zySQ488MCMGzcujz/+eNZZZ50Vrv0vf/lLJk6cmKlTp5b3P//887PLLrssdZ8RI0Zk+PDhK3wsAACARTwKB/D/9OjRIzvuuGO6deuWffbZJ1dffXVmzZq11P4TJ05Mjx49arxAe+utt87ChQszadKkzJw5M//85z+z4447LvO4xx9/fJ588smMHTu2UKi0qJZ11123xv69e/de5j6nnXZaZs+eXV6mT59e6NgAAMA3l2AJ4P+pW7duRo8enfvuuy9dunTJFVdckU6dOuX1119fYv9SqZSKioolbquoqEjDhg2X67g77bRT/u///i8PPPBA4dpLpdISa1iW+vXrp7KyssYCAACwIgRLAP+hoqIiW2+9dYYPH57nnnsuq6++eu64446svvrqWbBgQY2+Xbp0SXV1dd57771y2+OPP546depkww03TNOmTdO+ffs89NBDyzzm9773vdx000358Y9/nFtuuaVQ3V26dMm0adPyz3/+s9z25JNPFhoLAABgeQmWAP6fp59+Oueff37GjRuXadOm5fbbb89bb72Vzp07p3379nnhhRcyadKk/Pvf/87HH3+cAw44IA0aNMigQYPy0ksv5ZFHHskxxxyTgQMHpmXLlkmSYcOG5eKLL87ll1+eyZMnZ8KECbniiisWO/Zee+2VG264IQcffHD+9Kc/rXDt/fr1S6dOnXLggQfm+eefz9ixY3P66ad/4WsCAACwLF7eDfD/VFZW5q9//WtGjhyZOXPmpF27drn44ouzyy67pFevXhkzZkx69eqVefPm5ZFHHknfvn3zwAMP5Ljjjsvmm2+eRo0a5fvf/34uueSS8piDBg3Khx9+mEsvvTQnnXRS1lprrfzgBz9Y4vF/8IMfZOHChRk4cGDq1KmTvffee7lrr1OnTu64444ceuih2WKLLdK+fftcfvnl2Xnnnb/wdQEAAFiaitKSXswBwDfOnDlzUlVVlbZDbk2d+o2W2m/qBd/9CqsCAAC+aot+N5g9e/bnvovVo3AAAAAAFCJYAlgFjR07Nk2aNFnqAgAAsCrwjiWAVVCvXr1SXV1d22UAAAAsk2AJYBXUsGHDdOjQobbLAAAAWCaPwgEAAABQiGAJAAAAgEIESwAAAAAUIlgCAAAAoBDBEgAAAACFCJYAAAAAKESwBAAAAEAhgiUAAAAAChEsAQAAAFCIYAkAAACAQurVdgEArFpeGt4/lZWVtV0GAADwNWDGEgAAAACFCJYAAAAAKESwBAAAAEAhgiUAAAAAChEsAQAAAFCIYAkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKAQwRIAAAAAhQiWAAAAAChEsAQAAABAIYIlAAAAAAoRLAEAAABQiGAJAAAAgEIESwAAAAAUIlgCAAAAoBDBEgAAAACFCJYAAAAAKESwBAAAAEAhgiUAAAAAChEsAQAAAFCIYAkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKCQerVdAACrlq5DH0id+o2WuG3qBd/9iqsBAABWZWYsAQAAAFCIYAkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKAQwRIAAAAAhQiWAAAAAChEsAQAAABAIYKlr1j79u0zcuTI5e4/bNiw9OzZ8wsft6KiInfeeecXHmdppk6dmoqKilRXVydJxowZk4qKirz77rtJklGjRqVZs2Zf2vFhZTrooIOy5557rvRxl+fPwWePXSqVcvjhh2fNNdes8WcMAABgVSBY4iux33775dVXX/1Sj3HEEUekoqJiseDuzTffzMCBA9OqVas0btw4m266af70pz99qbWsCvr27ZshQ4bUdhk1rIo1rWouu+yyjBo1qrx+//33Z9SoUbnnnnsyY8aMdO3adalB8fz583P66aenXbt2qV+/fjbYYINce+21X13xAADAN0692i6AVd/HH3+c1VZb7QuN0bBhwzRs2HAlVbS4O++8M08//XTatGmz2LaBAwdm9uzZueuuu7LWWmvlpptuyn777Zdx48Zlk002+dJq+m/x0UcfZfXVV6/tMr4xqqqqaqxPmTIlrVu3zlZbbfW5++67777517/+ld/+9rfp0KFDZs6cmU8++eTLKhUAAMCMpZVt7ty5OeCAA9K4ceO0bt06l1566TJnaUybNi177LFHmjRpksrKyvIvhp/161//Om3btk2jRo2yzz77lB8xS5Jnn302O+20U9Zaa61UVVWlT58+mTBhQqH6Fz3Sduutt6Zv375p0KBBfv/732fhwoU5++yzs84666R+/frp2bNn7r///uUe97OPAC16xO+GG25I+/btU1VVlR/+8IeZO3duuc/yXsv/+7//y+DBg3PjjTcuMQB78sknc8wxx2SLLbbI+uuvnzPOOCPNmjVbrmv0n9dj2223TcOGDbP55pvn1VdfzbPPPptevXqlSZMm2XnnnfPWW2/V2Pe6665L586d06BBg2y00Ub55S9/WWP7qaeemg033DCNGjXK+uuvnzPPPDMff/zxCl2jpTnooIPy6KOP5rLLLktFRUUqKioyderULFiwIIceemjWW2+9NGzYMJ06dcpll1222L577rlnRowYkTZt2mTDDTdMkjzxxBPp2bNnGjRokF69euXOO+9c7NGsV155JbvuumuaNGmSli1bZuDAgfn3v/+9zJqWZdEjlQ888EA22WSTNGzYMDvssENmzpyZ++67L507d05lZWUGDBiQ999/v7zf/Pnzc+yxx6ZFixZp0KBBttlmmzz77LM1xn755Zfz3e9+N5WVlWnatGm23XbbTJkyZYl1jB8/Pi1atMh55533udf++eefz/bbb5+mTZumsrIym222WcaNG1ejzwMPPJDOnTuX750ZM2Ysdv0X/XzMMcdk2rRpqaioSPv27dO+ffskyV577VVuSz6d2fToo4/m3nvvTb9+/dK+fftsscUWywyk5s+fnzlz5tRYAAAAVoRgaSU74YQT8vjjj+euu+7K6NGjM3bs2KUGGKVSKXvuuWfeeeedPProoxk9enSmTJmS/fbbr0a/v//977n11ltz99135/777091dXWOPvro8va5c+dm0KBBGTt2bJ566ql07Ngxu+6663IFEEtz6qmn5thjj83EiRPTv3//XHbZZbn44ovz85//PC+88EL69++f733ve5k8eXLhY0yZMiV33nln7rnnntxzzz159NFHc8EFF5S3L8+1XLhwYQYOHJiTTz45G2+88RKPs8022+QPf/hD3nnnnSxcuDC33HJL5s+fn759+y53rUOHDs0ZZ5yRCRMmpF69ehkwYEBOOeWUXHbZZRk7dmymTJmSs846q9z/6quvzumnn57zzjsvEydOzPnnn58zzzwz119/fblP06ZNM2rUqLzyyiu57LLLcvXVV+fSSy9doWu0NJdddll69+6dww47LDNmzMiMGTPStm3bLFy4MOuss05uvfXWvPLKKznrrLPy05/+NLfeemuN/R966KFMnDgxo0ePzj333JO5c+dm9913T7du3TJhwoScc845OfXUU2vsM2PGjPTp0yc9e/bMuHHjcv/99+df//pX9t1332XWtDyGDRuWK6+8Mk888USmT5+efffdNyNHjsxNN92UP//5zxk9enSuuOKKcv9TTjklt912W66//vpMmDAhHTp0SP/+/fPOO+8k+TSM3G677dKgQYM8/PDDGT9+fA455JAlzu4ZM2ZMdtxxxwwfPjynn37659Z6wAEHZJ111smzzz6b8ePH53/+539qBJ7vv/9+fv7zn+eGG27IX//610ybNi0nnXTSEse67LLLyoHujBkz8uyzz5YDsuuuu67cliR33XVXevXqlYsuuihrr712Ntxww5x00kn54IMPllrriBEjUlVVVV6W9/MAAAAoK7HSzJkzp7TaaquV/vjHP5bb3n333VKjRo1Kxx13XKlUKpXatWtXuvTSS0ulUqn04IMPlurWrVuaNm1auf/LL79cSlJ65plnSqVSqTR06NBS3bp1S9OnTy/3ue+++0p16tQpzZgxY4l1fPLJJ6WmTZuW7r777nJbktIdd9zxuefw+uuvl5KURo4cWaO9TZs2pfPOO69G2+abb1466qijauz33HPPlUqlUumRRx4pJSnNmjWrVCqVStddd12pqqqqvO/QoUNLjRo1Ks2ZM6fcdvLJJ5e23HLLUqm0fNeyVCqVzj///NJOO+1UWrhwYalUqnl9/3O//v37l5KU6tWrV6qsrCw9+OCDn3st/vO8rrnmmnLbzTffXEpSeuihh8ptI0aMKHXq1Km83rZt29JNN91UY6xzzjmn1Lt376Ue66KLLiptttlm5fXPu0afp0+fPjWu1dIcddRRpe9///vl9UGDBpVatmxZmj9/frntqquuKjVv3rz0wQcflNuuvvrqGp/5mWeeWfrOd75TY+zp06eXkpQmTZq0QjUtsug++stf/lJuGzFiRClJacqUKeW2I444otS/f/9SqVQqzZs3r7TaaquVbrzxxvL2jz76qNSmTZvSRRddVCqVSqXTTjuttN5665U++uijJR530KBBpT322KN05513lpo2bbrYZ7ksTZs2LY0aNWqJ26677rpSktLf//73ctsvfvGLUsuWLRc79iKXXnppqV27djXGWdKf5/79+5fq169f+u53v1t6+umnS3/+859L7dq1Kx188MFLrfXDDz8szZ49u7ws+rzaDrm11O7Ue5a4AAAA//1mz55dSlKaPXv25/b1jqWV6LXXXsvHH3+cLbbYotxWVVWVTp06LbH/xIkT07Zt2xqzBLp06ZJmzZpl4sSJ2XzzzZMk6667btZZZ51yn969e2fhwoWZNGlSWrVqlZkzZ+ass87Kww8/nH/9619ZsGBB3n///UybNq3wufTq1av885w5c/LPf/4zW2+9dY0+W2+9dZ5//vnCx2jfvn2aNm1aXm/dunVmzpyZZPmu5fjx43PZZZdlwoQJqaioWOpxzjjjjMyaNSt/+ctfstZaa+XOO+/MPvvsk7Fjx6Zbt27LVWv37t3LP7ds2TJJauzbsmXLcu1vvfVWpk+fnkMPPTSHHXZYuc8nn3xS4/05f/rTnzJy5Mj8/e9/z7x58/LJJ5+ksrJyua9RUb/61a9yzTXX5I033sgHH3yQjz76aLFvHuzWrVuN9ypNmjQp3bt3T4MGDcpt//nZJJ9+Ho888kiaNGmy2DGnTJlSfqSuiM9e/0WPD/5n2zPPPFM+1scff1zjfl1ttdWyxRZbZOLEiUmS6urqbLvttst8d9jTTz+de+65J3/84x+z1157LXetJ5xwQn784x/nhhtuSL9+/bLPPvtkgw02KG9v1KhRjfWV8Zkmn87eq6ioyI033li+zy655JL84Ac/yC9+8YslvuOsfv36qV+//hc+NgAA8M3lUbiVqFQqJcliIcei9iX1X1IgsrT2RRZtW/Tfgw46KOPHj8/IkSPzxBNPpLq6Os2bN89HH31U6DySpHHjxks97vLW+Xk++0t9RUVFFi5cWB57acdcZOzYsZk5c2bWXXfd1KtXL/Xq1csbb7yRE088sfzemSlTpuTKK6/Mtddemx133DE9evTI0KFD06tXr/ziF78oVOuimj7btqj2Rf+9+uqrU11dXV5eeumlPPXUU0mSp556Kj/84Q+zyy675J577slzzz2X008/fbHPbFnXqIhbb701xx9/fA455JA8+OCDqa6uzsEHH7zYcT/7+S/ps/7sfb1w4cLsvvvuNc65uro6kydPznbbbVe45mTxa1303lnUtjwvkt9ggw2y0UYb5dprr12hP0vDhg0rv7/p4YcfTpcuXXLHHXcs8VwW1bm0vyNWROvWrbP22mvXCC87d+6cUqmUf/zjH194fAAAgCURLK1EG2ywQVZbbbXyzInk09k+S3sPUZcuXTJt2rRMnz693PbKK69k9uzZ6dy5c7lt2rRp+ec//1lef/LJJ1OnTp3yDJCxY8fm2GOPza677pqNN9449evXL78weWWorKxMmzZt8thjj9Vof+KJJ2rUuTItz7UcOHBgXnjhhRohRps2bXLyySfngQceSJLyC53r1Kl5q9etW/cLBTTL0rJly6y99tp57bXX0qFDhxrLeuutlyR5/PHH065du5x++unp1atXOnbsmDfeeGOl1rH66qtnwYIFNdrGjh2brbbaKkcddVQ22WSTdOjQYakvrP5PG220UV544YXMnz+/3PbZF1Jvuummefnll9O+ffvFzntRULWkmla2Dh06ZPXVV69xv3788ccZN25c+X7t3r17xo4dW+Nl6Z+11lpr5eGHHy6/92xZfT9rww03zPHHH58HH3wwe++9d6677rriJ7QEq6222mLXceutt84///nPzJs3r9z26quvpk6dOjVmPAIAAKxMgqWVqGnTphk0aFBOPvnkPPLII3n55ZdzyCGHpE6dOkuc2dOvX7907949BxxwQCZMmJBnnnkmBx54YPr06VPjUbQGDRpk0KBBef7558sh0r777ptWrVol+fQX6RtuuCETJ07M008/nQMOOGC5ZmSsiJNPPjkXXnhh/vCHP2TSpEn5n//5n1RXV+e4445bqcdZZHmuZfPmzdO1a9cay2qrrZZWrVqVH5nbaKON0qFDhxxxxBF55plnMmXKlFx88cUZPXp0+Zu3vgzDhg3LiBEjctlll+XVV1/Niy++mOuuuy6XXHJJkk8/s2nTpuWWW27JlClTcvnll9eY1bIytG/fPk8//XSmTp2af//731m4cGE6dOiQcePG5YEHHsirr76aM888c7FvS1uS/fffPwsXLszhhx+eiRMn5oEHHsjPf/7zJP//zKCjjz4677zzTgYMGJBnnnkmr732Wh588MEccsgh5RBkSTWtbI0bN85PfvKTnHzyybn//vvzyiuv5LDDDsv777+fQw89NEkyePDgzJkzJz/84Q8zbty4TJ48OTfccEMmTZpUY6wWLVrk4Ycfzt/+9rcMGDBgiS/3/k8ffPBBBg8enDFjxuSNN97I448/nmeffXalB7Dt27fPQw89lDfffDOzZs1K8uln1Lx58xx88MF55ZVX8te//jUnn3xyDjnkkJX+9wEAAMAigqWV7JJLLknv3r2z2267pV+/ftl6663LXzn/WRUVFbnzzjuzxhprZLvttku/fv2y/vrr5w9/+EONfh06dMjee++dXXfdNd/5znfStWvXGl9df+2112bWrFnZZJNNMnDgwPLXrK9Mxx57bE488cSceOKJ6datW+6///7cdddd6dix40o9zn9akWu5NKuttlruvffefOtb38ruu++e7t2753e/+12uv/767Lrrrl9a7T/+8Y9zzTXXZNSoUenWrVv69OmTUaNGlWcs7bHHHjn++OMzePDg9OzZM0888UTOPPPMlVrDSSedlLp166ZLly751re+lWnTpuXII4/M3nvvnf322y9bbrll3n777Rx11FGfO1ZlZWXuvvvuVFdXp2fPnjn99NPL34K36PNo06ZNHn/88SxYsCD9+/dP165dc9xxx6Wqqqo8Y2xJNX0ZLrjggnz/+9/PwIEDs+mmm+bvf/97HnjggayxxhpJPg0lH3744cybNy99+vTJZpttlquvvnqJ71xq1apVHn744bz44os54IADljnjqm7dunn77bdz4IEHZsMNN8y+++6bXXbZJcOHD1+p57coHG3btm022WSTJEmTJk0yevTovPvuu+nVq1cOOOCA7L777rn88stX6rEBAAD+U0VpZbzcg6V67733svbaa+fiiy8uz5agGNdy1XLjjTfm4IMPzuzZs82I+S8xZ86cVFVVpe2QW1OnfqMl9pl6wXe/4qoAAICv2qLfDWbPnr3Yl0x9lm+FW8mee+65/O1vf8sWW2yR2bNn5+yzz07y6QwVVoxruWr53e9+l/XXXz9rr712nn/++Zx66qnZd999hUoAAADfYB6F+xL8/Oc/T48ePdKvX7+89957GTt2bNZaa63aLitJcv7556dJkyZLXHbZZZfaLm8xX/a1/Lpdj0WmTZu21LqbNGnypTxi9uabb+ZHP/pROnfunOOPPz777LNPfvOb33yhMY888silnsORRx65kipfuTbeeOOl1nzjjTfWdnkAAABfKY/CfcO88847eeedd5a4rWHDhll77bW/4opq19f1enzyySeZOnXqUre3b98+9eqt+hMSZ86cmTlz5ixxW2Vl5Up/V9jK8MYbbyz1G+JatmyZpk2bfsUVrTwehQMAABKPwrEMa665ZtZcc83aLmOV8XW9HvXq1UuHDh1qu4wvrEWLFqtkeLQs7dq1q+0SAAAAVhkehQMAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKCQerVdAACrlpeG909lZWVtlwEAAHwNmLEEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKAQwRIAAAAAhQiWAAAAAChEsAQAAABAIYIlAAAAAAoRLAEAAABQiGAJAAAAgEIESwAAAAAUIlgCAAAAoBDBEgAAAACFCJYAAAAAKESwBAAAAEAhgiUAAAAAChEsAQAAAFCIYAkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKAQwRIAAAAAhQiWAAAAAChEsAQAAABAIYIlAAAAAAoRLAEAAABQiGAJAAAAgEIESwAAAAAUIlgCAAAAoBDBEgAAAACFCJYAAAAAKKRebRcAwKql69AHUqd+oxptUy/4bi1VAwAArMrMWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKAQwRIAAAAAhQiWAAAAAChEsAQAAABAIYIlAAAAAAoRLAEAAABQiGAJAAAAgEIES/BfbMyYMamoqMi77767SoyzyEEHHZQ999xzpYy1NKNGjUqzZs2+1GMAAAB80wmWgM+11VZbZcaMGamqqqrtUr4yKztMAwAA+G8kWAI+1+qrr55WrVqloqKitkv5r/PRRx/VdgkAAACFCZZgBfzpT39Kt27d0rBhwzRv3jz9+vXLe++9lyS57rrr0rlz5zRo0CAbbbRRfvnLX9bY9x//+Ed++MMfZs0110zjxo3Tq1evPP300+XtV111VTbYYIOsvvrq6dSpU2644YYa+1dUVOSaa67JXnvtlUaNGqVjx4656667avS59957s+GGG6Zhw4bZfvvtM3Xq1Brb33jjjey+++5ZY4010rhx42y88ca59957P/e8Pzt7Z9FjZg888EA6d+6cJk2aZOedd86MGTOW91ImSYYPH54WLVqksrIyRxxxRI2QpX379hk5cmSN/j179sywYcPK6++++24OP/zwtGzZMg0aNEjXrl1zzz33LPFYb7/9drbYYot873vfy4cffphSqZSLLroo66+/fho2bJgePXrkT3/6U5Jk6tSp2X777ZMka6yxRioqKnLQQQd97vnMnTs3BxxwQBo3bpzWrVvn0ksvTd++fTNkyJAa53XuuefmoIMOSlVVVQ477LAkyW233ZaNN9449evXT/v27XPxxRfXGLuioiJ33nlnjbZmzZpl1KhR5ZorKipyyy23ZKuttkqDBg2y8cYbZ8yYMUutd/78+ZkzZ06NBQAAYEXUq+0C4OtixowZGTBgQC666KLstddemTt3bsaOHZtSqZSrr746Q4cOzZVXXplNNtkkzz33XA477LA0btw4gwYNyrx589KnT5+svfbaueuuu9KqVatMmDAhCxcuTJLccccdOe644zJy5Mj069cv99xzTw4++OCss8465YAj+TSIueiii/Kzn/0sV1xxRQ444IC88cYbWXPNNTN9+vTsvffeOfLII/OTn/wk48aNy4knnljjHI4++uh89NFH+etf/5rGjRvnlVdeSZMmTQpdj/fffz8///nPc8MNN6ROnTr50Y9+lJNOOik33njjcu3/0EMPpUGDBnnkkUcyderUHHzwwVlrrbVy3nnnLdf+CxcuzC677JK5c+fm97//fTbYYIO88sorqVu37mJ9//GPf+Q73/lOevXqlWuvvTb16tXL6aefnttvvz1XXXVVOnbsmL/+9a/50Y9+lG9961vZZpttctttt+X73/9+Jk2alMrKyjRs2PBzazrhhBPy+OOP56677krLli1z1llnZcKECenZs2eNfj/72c9y5pln5owzzkiSjB8/Pvvuu2+GDRuW/fbbL0888USOOuqoNG/efLkCrf908sknZ+TIkenSpUsuueSSfO9738vrr7+e5s2bL9Z3xIgRGT58+AqNDwAAUEMJWC7jx48vJSlNnTp1sW1t27Yt3XTTTTXazjnnnFLv3r1LpVKp9Otf/7rUtGnT0ttvv73EsbfaaqvSYYcdVqNtn332Ke26667l9SSlM844o7w+b968UkVFRem+++4rlUql0mmnnVbq3LlzaeHCheU+p556ailJadasWaVSqVTq1q1badiwYStw1p965JFHaoxz3XXXlZKU/v73v5f7/OIXvyi1bNlyucYbNGhQac011yy999575barrrqq1KRJk9KCBQtKpVKp1K5du9Kll15aY78ePXqUhg4dWiqVSqUHHnigVKdOndKkSZOWeIzrrruuVFVVVZo0aVJp3XXXLR1zzDHlazNv3rxSgwYNSk888USNfQ499NDSgAEDlnjOn2fOnDml1VZbrfTHP/6x3Pbuu++WGjVqVDruuOPKbe3atSvtueeeNfbdf//9SzvttFONtpNPPrnUpUuX8nqS0h133FGjT1VVVem6664rlUql0uuvv15KUrrgggvK2z/++OPSOuusU7rwwguXWPOHH35Ymj17dnmZPn16KUmp7ZBbS+1OvafGAgAAfHPMnj27lKQ0e/bsz+3rUThYTj169MiOO+6Ybt26ZZ999snVV1+dWbNm5a233sr06dNz6KGHpkmTJuXl3HPPzZQpU5Ik1dXV2WSTTbLmmmsuceyJEydm6623rtG29dZbZ+LEiTXaunfvXv65cePGadq0aWbOnFke49vf/naN9yD17t27xv7HHntszj333Gy99dYZOnRoXnjhhcLXo1GjRtlggw3K661bty7Xsjx69OiRRo0a1ah13rx5mT59+nLtX11dnXXWWScbbrjhUvt88MEH2WabbbLnnnvm8ssvL1+bV155JR9++GF22mmnGp/Z7373u/JntqJee+21fPzxx9liiy3KbVVVVenUqdNifXv16lVjfWmf/+TJk7NgwYIVquM/P/N69eqlV69ei91Hi9SvXz+VlZU1FgAAgBUhWILlVLdu3YwePTr33XdfunTpkiuuuCKdOnXKa6+9liS5+uqrU11dXV5eeumlPPXUU0myXI9RffbF2KVSabG21VZbbbF9Fj1OVyqVPvcYP/7xj/Paa69l4MCBefHFF9OrV69cccUVn7vfkiypluWp4fMsOuc6deosNt7HH39c/nl5rmn9+vXTr1+//PnPf84//vGPcvuia/bnP/+5xmf2yiuvlN+ztKIW1bqkz/GzGjduvFifz9tvSdf3P6/HsnjpOgAA8GURLMEKqKioyNZbb53hw4fnueeey+qrr57HH388a6+9dl577bV06NChxrLeeusl+XSmUXV1dd55550ljtu5c+c89thjNdqeeOKJdO7ceblr69KlSznIWuSz60nStm3bHHnkkbn99ttz4okn5uqrr17uY6xMzz//fD744IPy+lNPPZUmTZpknXXWSZJ861vfqvEy8Dlz5uT1118vr3fv3j3/+Mc/8uqrry71GHXq1MkNN9yQzTbbLDvssEP++c9/Jvn0WtWvXz/Tpk1b7DNr27Ztkk+/CS/Jcs8Y2mCDDbLaaqvlmWeeqVHz5MmTP3ffLl26LPHz33DDDcvvjPrs9Zg8eXLef//9xcb6z8/8k08+yfjx47PRRhst1zkAAACsKC/vhuX09NNP56GHHsp3vvOdtGjRIk8//XTeeuutdO7cOcOGDcuxxx6bysrK7LLLLpk/f37GjRuXWbNm5YQTTsiAAQNy/vnnZ88998yIESPSunXrPPfcc2nTpk169+6dk08+Ofvuu2823XTT7Ljjjrn77rtz++235y9/+cty13fkkUfm4osvzgknnJAjjjgi48ePL39j2CJDhgzJLrvskg033DCzZs3Kww8/vELh1cr00Ucf5dBDD80ZZ5yRN954I0OHDs3gwYNTp86nefcOO+yQUaNGlb/F7swzz6zxYu4+ffpku+22y/e///1ccskl6dChQ/72t7+loqIiO++8c7lf3bp1c+ONN2bAgAHZYYcdMmbMmLRq1SonnXRSjj/++CxcuDDbbLNN5syZkyeeeCJNmjTJoEGD0q5du1RUVOSee+7JrrvumoYNGy7zRedNmzbNoEGDcvLJJ2fNNddMixYtMnTo0NSpU+dzZwydeOKJ2XzzzXPOOedkv/32y5NPPpkrr7yyxjcL7rDDDrnyyivz7W9/OwsXLsypp5662KyxJPnFL36Rjh07pnPnzrn00ksza9asHHLIIcv9uQAAAKwIM5ZgOVVWVuavf/1rdt1112y44YY544wzcvHFF2eXXXbJj3/841xzzTUZNWpUunXrlj59+mTUqFHlGUurr756HnzwwbRo0SK77rprunXrlgsuuKAclOy555657LLL8rOf/Swbb7xxfv3rX+e6665L3759l7u+ddddN7fddlvuvvvu9OjRI7/61a9y/vnn1+izYMGCHH300encuXN23nnndOrUqUZ48VXacccd07Fjx2y33XbZd999s/vuu2fYsGHl7aeddlq222677Lbbbtl1112z55571ninU5Lcdttt2XzzzTNgwIB06dIlp5xyyhJnGNWrVy8333xzNt544+ywww6ZOXNmzjnnnJx11lkZMWJEOnfunP79++fuu+8uf2Zrr712hg8fnv/5n/9Jy5YtM3jw4M89p0suuSS9e/fObrvtln79+mXrrbdO586d06BBg2Xut+mmm+bWW2/NLbfckq5du+ass87K2WefXeMb4S6++OK0bds22223Xfbff/+cdNJJNd5RtcgFF1yQCy+8MD169MjYsWPzv//7v1lrrbU+t3YAAIAiKkor46UoACzmvffey9prr52LL744hx566Jd6rKlTp2a99dbLc889l549exYaY86cOamqqkrbIbemTv2aodXUC767EqoEAAC+Dhb9bjB79uzP/ZIfj8IBrCTPPfdc/va3v2WLLbbI7Nmzc/bZZydJ9thjj1quDAAA4MvhUTggRx55ZJo0abLE5cgjj1zh8ZY2VpMmTTJ27Ngv4Qy+fNOmTVvmeU2bNi1J8vOf/zw9evRIv3798t5772Xs2LEeRQMAAP5reRQOyMyZMzNnzpwlbqusrEyLFi1WaLy///3vS9229tprp2HDhis03qrgk08+ydSpU5e6vX379qlX7+s9CdSjcAAAQOJROGAFtWjRYoXDo2Xp0KHDShtrVVGvXr3/yvMCAAD4IjwKBwAAAEAhgiUAAAAAChEsAQAAAFCIYAkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCH1arsAAFYtLw3vn8rKytouAwAA+BowYwkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKAQwRIAAAAAhQiWAAAAAChEsAQAAABAIYIlAAAAAAoRLAEAAABQiGAJAAAAgEIESwAAAAAUIlgCAAAAoBDBEgAAAACFCJYAAAAAKESwBAAAAEAhgiUAAAAAChEsAQAAAFCIYAkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKAQwRIAAAAAhQiWAAAAAChEsAQAAABAIYIlAAAAAAoRLAEAAABQSL3aLgCAVUvXoQ+kTv1GNdqmXvDdWqoGAABYlZmxBAAAAEAhgiUAAAAAChEsAQAAAFCIYAkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKAQwRIAAAAAhQiWAFZA3759M2TIkGX2qaioyJ133lle/9vf/pZvf/vbadCgQXr27Pml1gcAAPBVEiwBrGQzZszILrvsUl4fOnRoGjdunEmTJuWhhx7KqFGj0qxZs5V+3PPOOy9bbbVVGjVq9KWMDwAA8FmCJYCVrFWrVqlfv355fcqUKdlmm23Srl27NG/e/Es77kcffZR99tknP/nJT760YwAAAPwnwRJQ6+6+++40a9YsCxcuTJJUV1enoqIiJ598crnPEUcckQEDBuTtt9/OgAEDss4666RRo0bp1q1bbr755hrj9e3bN4MHD87gwYPTrFmzNG/ePGeccUZKpdJy1fPLX/4yHTt2TIMGDdKyZcv84Ac/qLF94cKFOeWUU7LmmmumVatWGTZsWI3t//koXEVFRcaPH5+zzz47FRUV6du3bw4++ODMnj07FRUVqaioWGz/JWnfvn3OP//8HHLIIWnatGnWXXfd/OY3v6nRZ/jw4Tn++OPTrVu35TpPAACAL0qwBNS67bbbLnPnzs1zzz2XJHn00Uez1lpr5dFHHy33GTNmTPr06ZMPP/wwm222We6555689NJLOfzwwzNw4MA8/fTTNca8/vrrU69evTz99NO5/PLLc+mll+aaa6753FrGjRuXY489NmeffXYmTZqU+++/P9ttt91iYzdu3DhPP/10Lrroopx99tkZPXr0EsebMWNGNt5445x44omZMWNG7rrrrowcOTKVlZWZMWNGZsyYkZNOOmm5rtPFF1+cXr165bnnnstRRx2Vn/zkJ/nb3/62XPsuyfz58zNnzpwaCwAAwIoQLAG1rqqqKj179syYMWOSfBoiHX/88Xn++eczd+7cvPnmm3n11VfTt2/frL322jnppJPSs2fPrL/++jnmmGPSv3///PGPf6wxZtu2bXPppZemU6dOOeCAA3LMMcfk0ksv/dxapk2blsaNG2e33XZLu3btsskmm+TYY4+t0ad79+4ZOnRoOnbsmAMPPDC9evXKQw89tMTxWrVqlXr16qVJkyZp1apVKisrU1VVlYqKirRq1SqtWrVKkyZNlus67brrrjnqqKPSoUOHnHrqqVlrrbXK16yIESNGpKqqqry0bdu28FgAAMA3k2AJWCX07ds3Y8aMSalUytixY7PHHnuka9eueeyxx/LII4+kZcuW2WijjbJgwYKcd9556d69e5o3b54mTZrkwQcfzLRp02qM9+1vfzsVFRXl9d69e2fy5MlZsGDBMuvYaaed0q5du6y//voZOHBgbrzxxrz//vs1+nTv3r3GeuvWrTNz5swveAU+338ed1Ew9UWOe9ppp2X27NnlZfr06SujTAAA4BtEsASsEvr27ZuxY8fm+eefT506ddKlS5f06dMnjz76aPkxuOTTx8EuvfTSnHLKKXn44YdTXV2d/v3756OPPlopdTRt2jQTJkzIzTffnNatW+ess85Kjx498u6775b7rLbaajX2qaioKL8f6su0so9bv379VFZW1lgAAABWhGAJWCUses/SyJEj06dPn1RUVKRPnz4ZM2ZMjWBp0WymH/3oR+nRo0fWX3/9TJ48ebHxnnrqqcXWO3bsmLp1635uLfXq1Uu/fv1y0UUX5YUXXsjUqVPz8MMPr5wTTbL66qt/7swpAACArwPBErBKWPSepd///vfp27dvkk/DpgkTJpTfr5QkHTp0yOjRo/PEE09k4sSJOeKII/Lmm28uNt706dNzwgknZNKkSbn55ptzxRVX5LjjjvvcOu65555cfvnlqa6uzhtvvJHf/e53WbhwYTp16rTSzrV9+/aZN29eHnroofz73/9e7FG7oqZNm5bq6upMmzYtCxYsSHV1daqrqzNv3ryVMj4AAMBnCZaAVcb222+fBQsWlEOkNdZYI126dMm3vvWtdO7cOUly5plnZtNNN03//v3Tt2/ftGrVKnvuuediYx144IH54IMPssUWW+Too4/OMccck8MPP/xza2jWrFluv/327LDDDuncuXN+9atf5eabb87GG2+80s5zq622ypFHHpn99tsv3/rWt3LRRRetlHHPOuusbLLJJhk6dGjmzZuXTTbZJJtssknGjRu3UsYHAAD4rIpSqVSq7SIAVqa+ffumZ8+eGTlyZG2X8rUyZ86cT78dbsitqVO/UY1tUy/4bi1VBQAAfNUW/W4we/bsz30XqxlLAAAAABQiWAK+UcaOHZsmTZosdVETAADA8qtX2wUArGxjxoxZ6rZevXqlurr6K6tleayKNQEAACwPwRLwjdKwYcN06NChtsuoYVWsCQAAYHl4FA4AAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBC6tV2AQCsWl4a3j+VlZW1XQYAAPA1YMYSAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKAQwRIAAAAAhQiWAAAAAChEsAQAAABAIYIlAAAAAAoRLAEAAABQiGAJAAAAgEIESwAAAAAUIlgCAAAAoBDBEgAAAACFCJYAAAAAKESwBAAAAEAhgiUAAAAAChEsAQAAAFCIYAkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKAQwRIAAAAAhQiWAAAAAChEsAQAAABAIYIlAAAAAAoRLAEAAABQiGAJAAAAgEIESwAAAAAUIlgCAAAAoJB6tV0AAKuWrkMfSJ36jWq0Tb3gu7VUDQAAsCozYwkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAAAAUIhgCQAAAIBCBEsAAAAAFCJYAgAAAKAQwRIAAAAAhQiWAAAAAChEsAQAAABAIYIlAAAAAAoRLAEsh4qKitx5551f6THHjBmTioqKvPvuu1/pcQEAAJaXYAlgFbXVVltlxowZqaqqqu1SAAAAlqhebRcAwOI+/vjjrL766mnVqlVtlwIAALBUZiwB3xj3339/ttlmmzRr1izNmzfPbrvtlilTpiRJPvroowwePDitW7dOgwYN0r59+4wYMWKpY5199tlp2bJlqqurP/e47du3zznnnJP9998/TZo0SZs2bXLFFVfU6FNRUZFf/epX2WOPPdK4ceOce+65S3wU7vHHH0+fPn3SqFGjrLHGGunfv39mzZqVJCmVSrnooouy/vrrp2HDhunRo0f+9Kc/rfiFAgAAWE6CJeAb47333ssJJ5yQZ599Ng899FDq1KmTvfbaKwsXLszll1+eu+66K7feemsmTZqU3//+92nfvv1iY5RKpRx33HH57W9/m8ceeyw9e/ZcrmP/7Gc/S/fu3TNhwoScdtppOf744zN69OgafYYOHZo99tgjL774Yg455JDFxqiurs6OO+6YjTfeOE8++WQee+yx7L777lmwYEGS5Iwzzsh1112Xq666Ki+//HKOP/74/OhHP8qjjz66xJrmz5+fOXPm1FgAAABWREWpVCrVdhEAteGtt95KixYt8uKLL+Y3v/lNXn755fzlL39JRUXFYn0rKiryxz/+Mf/7v/+bcePGZfTo0VlnnXWW6zjt27dP586dc99995XbfvjDH2bOnDm59957y+MPGTIkl156abnPmDFjsv3222fWrFlp1qxZ9t9//0ybNi2PPfbYYsd47733stZaa+Xhhx9O7969y+0//vGP8/777+emm25abJ9hw4Zl+PDhi7W3HXJr6tRvVKNt6gXfXa5zBQAAvv7mzJmTqqqqzJ49O5WVlcvsa8YS8I0xZcqU7L///ll//fVTWVmZ9dZbL0kybdq0HHTQQamurk6nTp1y7LHH5sEHH1xs/+OPPz5PPvlkxo4du9yh0iL/GfYsWp84cWKNtl69ei1zjEUzlpbklVdeyYcffpiddtopTZo0KS+/+93vyo/7fdZpp52W2bNnl5fp06evwBkBAAB4eTfwDbL77runbdu2ufrqq9OmTZssXLgwXbt2zUcffZRNN900r7/+eu6777785S9/yb777pt+/frVeEfRTjvtlJtvvjkPPPBADjjggC9cz2dnRjVu3HiZ/Rs2bLjUbQsXLkyS/PnPf87aa69dY1v9+vWXuE/9+vWXug0AAGB5CJaAb4S33347EydOzK9//etsu+22SbLYI2WVlZXZb7/9st9+++UHP/hBdt5557zzzjtZc801kyTf+973svvuu2f//fdP3bp188Mf/nC5j//UU08ttr7RRhut0Dl07949Dz300BIfX+vSpUvq16+fadOmpU+fPis0LgAAQFGCJeAbYY011kjz5s3zm9/8Jq1bt860adPyP//zP+Xtl156aVq3bp2ePXumTp06+eMf/5hWrVqlWbNmNcbZa6+9csMNN2TgwIGpV69efvCDHyzX8R9//PFcdNFF2XPPPTN69Oj88Y9/zJ///OcVOofTTjst3bp1y1FHHZUjjzwyq6++eh555JHss88+WWuttXLSSSfl+OOPz8KFC7PNNttkzpw5eeKJJ9KkSZMMGjRohY4FAACwPARLwDdCnTp1csstt+TYY49N165d06lTp1x++eXp27dvkqRJkya58MILM3ny5NStWzebb7557r333tSps/ir6H7wgx9k4cKFGThwYOrUqZO99977c49/4oknZvz48Rk+fHiaNm2aiy++OP3791+hc9hwww3z4IMP5qc//Wm22GKLNGzYMFtuuWUGDBiQJDnnnHPSokWLjBgxIq+99lqaNWuWTTfdND/96U9X6DgAAADLy7fCAXzJ2rdvnyFDhmTIkCG1XcoyLfrmB98KBwAA32y+FQ4AAACAL51gCeALGDt2bJo0abLUBQAA4L+ZdywBfAG9evVKdXX1MvtMnTr1K6kFAADgqyZYAvgCGjZsmA4dOtR2GQAAALXCo3AAAAAAFCJYAgAAAKAQwRIAAAAAhQiWAAAAAChEsAQAAABAIYIlAAAAAAoRLAEAAABQiGAJAAAAgEIESwAAAAAUUq+2CwBg1fLS8P6prKys7TIAAICvATOWAAAAAChEsAQAAABAIYIlAAAAAAoRLAEAAABQiGAJAAAAgEIESwAAAAAUIlgCAAAAoBDBEgAAAACFCJYAAAAAKESwBAAAAEAhgiUAAAAAChEsAQAAAFCIYAkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKqVfbBQCwaiiVSkmSOXPm1HIlAABAbVr0O8Gi3xGWRbAEQJLk7bffTpK0bdu2lisBAABWBXPnzk1VVdUy+wiWAEiSrLnmmkmSadOmfe7/eMCXZc6cOWnbtm2mT5+eysrK2i6Hbyj3IasC9yGrAvfhN1epVMrcuXPTpk2bz+0rWAIgSVKnzqev3auqqvJ/HKh1lZWV7kNqnfuQVYH7kFWB+/CbaXn/sdnLuwEAAAAoRLAEAAAAQCGCJQCSJPXr18/QoUNTv3792i6FbzD3IasC9yGrAvchqwL3IcujorQ83x0HAAAAAJ9hxhIAAAAAhQiWAAAAAChEsAQAAABAIYIlAAAAAAoRLAGQJPnlL3+Z9dZbLw0aNMhmm22WsWPH1nZJ/Bf761//mt133z1t2rRJRUVF7rzzzhrbS6VShg0bljZt2qRhw4bp27dvXn755doplv9KI0aMyOabb56mTZumRYsW2XPPPTNp0qQafdyHfNmuuuqqdO/ePZWVlamsrEzv3r1z3333lbe7B6kNI0aMSEVFRYYMGVJucy+yLIIlAPKHP/whQ4YMyemnn57nnnsu2267bXbZZZdMmzattkvjv9R7772XHj165Morr1zi9osuuiiXXHJJrrzyyjz77LNp1apVdtppp8ydO/crrpT/Vo8++miOPvroPPXUUxk9enQ++eSTfOc738l7771X7uM+5Mu2zjrr5IILLsi4ceMybty47LDDDtljjz3Kv7C7B/mqPfvss/nNb36T7t2712h3L7IsFaVSqVTbRQBQu7bccstsuummueqqq8ptnTt3zp577pkRI0bUYmV8E1RUVOSOO+7InnvumeTTfxVt06ZNhgwZklNPPTVJMn/+/LRs2TIXXnhhjjjiiFqslv9Wb731Vlq0aJFHH3002223nfuQWrPmmmvmZz/7WQ455BD3IF+pefPmZdNNN80vf/nLnHvuuenZs2dGjhzp70M+lxlLAN9wH330UcaPH5/vfOc7Ndq/853v5Iknnqilqvgme/311/Pmm2/WuCfr16+fPn36uCf50syePTvJp7/UJ+5DvnoLFizILbfckvfeey+9e/d2D/KVO/roo/Pd7343/fr1q9HuXuTz1KvtAgCoXf/+97+zYMGCtGzZskZ7y5Yt8+abb9ZSVXyTLbrvlnRPvvHGG7VREv/lSqVSTjjhhGyzzTbp2rVrEvchX50XX3wxvXv3zocffpgmTZrkjjvuSJcuXcq/sLsH+SrccsstmTBhQp599tnFtvn7kM8jWAIgyaePI/2nUqm0WBt8ldyTfFUGDx6cF154IY899thi29yHfNk6deqU6urqvPvuu7ntttsyaNCgPProo+Xt7kG+bNOnT89xxx2XBx98MA0aNFhqP/ciS+NROIBvuLXWWit169ZdbHbSzJkzF/uXKfgqtGrVKknck3wljjnmmNx111155JFHss4665Tb3Yd8VVZfffV06NAhvXr1yogRI9KjR49cdtll7kG+MuPHj8/MmTOz2WabpV69eqlXr14effTRXH755alXr175fnMvsjSCJYBvuNVXXz2bbbZZRo8eXaN99OjR2WqrrWqpKr7J1ltvvbRq1arGPfnRRx/l0UcfdU+y0pRKpQwePDi33357Hn744ay33no1trsPqS2lUinz5893D/KV2XHHHfPiiy+murq6vPTq1SsHHHBAqqurs/7667sXWSaPwgGQE044IQMHDkyvXr3Su3fv/OY3v8m0adNy5JFH1nZp/JeaN29e/v73v5fXX3/99VRXV2fNNdfMuuuumyFDhuT8889Px44d07Fjx5x//vlp1KhR9t9//1qsmv8mRx99dG666ab87//+b5o2bVr+l/iqqqo0bNgwFRUV7kO+dD/96U+zyy67pG3btpk7d25uueWWjBkzJvfff797kK9M06ZNy++XW6Rx48Zp3rx5ud29yLIIlgDIfvvtl7fffjtnn312ZsyYka5du+bee+9Nu3btars0/kuNGzcu22+/fXn9hBNOSJIMGjQoo0aNyimnnJIPPvggRx11VGbNmpUtt9wyDz74YJo2bVpbJfNf5qqrrkqS9O3bt0b7ddddl4MOOihJ3Id86f71r39l4MCBmTFjRqqqqtK9e/fcf//92WmnnZK4B1l1uBdZlopSqVSq7SIAAAAA+PrxjiUAAAAAChEsAQAAAFCIYAkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAoRLAEAAAAQCGCJQAAAAAKESwBAEAtOOigg7LnnnvWdhlLNHXq1FRUVKS6urq2SwFgFSdYAgAAyj766KPaLgGArxHBEgAA1LK+ffvmmGOOyZAhQ7LGGmukZcuW+c1vfpP33nsvBx98cJo2bZoNNtgg9913X3mfMWPGpKKiIn/+85/To0ePNGjQIFtuuWVefPHFGmPfdttt2XjjjVO/fv20b98+F198cY3t7du3z7nnnpuDDjooVVVVOeyww7LeeuslSTbZZJNUVFSkb9++SZJnn302O+20U9Zaa61UVVWlT58+mTBhQo3xKioqcs0112SvvfZKo0aN0rFjx9x11101+rz88sv57ne/m8rKyjRt2jTbbrttpkyZUt5+3XXXpXPnzmnQoEE22mij/PKXv/zC1xiAL4dgCQAAVgHXX3991lprrTzzzDM55phj8pOf/CT77LNPttpqq0yYMCH9+/fPwIED8/7779fY7+STT87Pf/7zPPvss2nRokW+973v5eOPP06SjB8/Pvvuu29++MMf5sUXX8ywYcNy5plnZtSoUTXG+NnPfpauXbtm/PjxOfPMM/PMM88kSf7yl79kxowZuf3225Mkc+fOzaBBgzJ27Ng89dRT6dixY3bdddfMnTu3xnjDhw/PvvvumxdeeCG77rprDjjggLzzzjtJkv/7v//LdtttlwYNGuThhx/O+PHjc8ghh+STTz5Jklx99dU5/fTTc95552XixIk5//zzc+aZZ+b6669f6dccgC+uolQqlWq7CAAA+KY56KCD8u677+bOO+9M3759s2DBgowdOzZJsmDBglRVVWXvvffO7373uyTJm2++mdatW+fJJ5/Mt7/97YwZMybbb799brnlluy3335JknfeeSfrrLNORo0alX333TcHHHBA3nrrrTz44IPl455yyin585//nJdffjnJpzOWNtlkk9xxxx3lPlOnTs16662X5557Lj179lzqOSxYsCBrrLFGbrrppuy2225JPp2xdMYZZ+Scc85Jkrz33ntp2rRp7r333uy888756U9/mltuuSWTJk3KaqutttiY6667bi688MIMGDCg3Hbuuefm3nvvzRNPPFHkUgPwJTJjCQAAVgHdu3cv/1y3bt00b9483bp1K7e1bNkySTJz5swa+/Xu3bv885prrplOnTpl4sSJSZKJEydm6623rtF/6623zuTJk7NgwYJyW69evZarxpkzZ+bII4/MhhtumKqqqlRVVWXevHmZNm3aUs+lcePGadq0abnu6urqbLvttksMld56661Mnz49hx56aJo0aVJezj333BqPygGw6qhX2wUAAABZLGipqKio0VZRUZEkWbhw4eeOtahvqVQq/7zIkh5YaNy48XLVeNBBB+Wtt97KyJEj065du9SvXz+9e/de7IXfSzqXRXU3bNhwqeMv6nP11Vdnyy23rLGtbt26y1UjAF8twRIAAHyNPfXUU1l33XWTJLNmzcqrr76ajTbaKEnSpUuXPPbYYzX6P/HEE9lwww2XGdSsvvrqSVJjVlOSjB07Nr/85S+z6667JkmmT5+ef//73ytUb/fu3XP99dfn448/XiyAatmyZdZee+289tprOeCAA1ZoXABqh2AJAAC+xs4+++w0b948LVu2zOmnn5611lore+65Z5LkxBNPzOabb55zzjkn++23X5588slceeWVn/stay1atEjDhg1z//33Z5111kmDBg1SVVWVDh065IYbbkivXr0yZ86cnHzyycucgbQkgwcPzhVXXJEf/vCHOe2001JVVZWnnnoqW2yxRTp16pRhw4bl2GOPTWVlZXbZZZfMnz8/48aNy6xZs3LCCScUvUwAfEm8YwkAAL7GLrjgghx33HHZbLPNMmPGjNx1113lGUebbrppbr311txyyy3p2rVrzjrrrJx99tk56KCDljlmvXr1cvnll+fXv/512rRpkz322CNJcu2112bWrFnZZJNNMnDgwBx77LFp0aLFCtXbvHnzPPzww5k3b1769OmTzTbbLFdffXV59tKPf/zjXHPNNRk1alS6deuWPn36ZNSoUVlvvfVW/OIA8KXzrXAAAPA1tOhb4WbNmpVmzZrVdjkAfEOZsQQAAABAIYIlAAAAAArxKBwAAAAAhZixBAAAAEAhgiUAAAAAChEsAQAAAFCIYAkAAACAQgRLAAAAABQiWAIAAACgEMESAAAAAIUIlgAAAAAo5P8Div9tRH290AQAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 1000x1600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = top_features.plot(kind='barh', \n",
    "                       x='Feature', \n",
    "                       y='Importances', \n",
    "                       figsize=(10, 16), \n",
    "                       legend=False)\n",
    "\n",
    "plt.gca().invert_yaxis()\n",
    "\n",
    "plt.title('Feature Importances')\n",
    "plt.xlabel('Importance')\n",
    "plt.ylabel('Feature')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ab986f6d-7ba1-4ba8-887c-187cc2135b08",
   "metadata": {},
   "source": [
    "Seconds in bucket group is one of the top-15 features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "666f78bc-c6cc-420d-8e9c-222e5e0d5e98",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['target_mock',\n",
       " 'target_single',\n",
       " 'wap_shift_n6',\n",
       " 'index_target_mock',\n",
       " 'weight_tmp']"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "top_features = feature_importance_catboost2.head(300)\n",
    "top_feature_names = top_features['Feature'].tolist()\n",
    "\n",
    "top_feature_names[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "d75cd141-cc22-4558-8147-883350a21671",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "300"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(top_feature_names)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "be2ecb22-cbb1-49ac-8153-63a4aaebd1fd",
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(f\"../{config['MODEL_DIR']}/top_300_feats_nms.pkl\", 'wb') as f:\n",
    "    pickle.dump(top_feature_names, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0b3b0f8a-7f97-4ee9-b27c-ca3db390570f",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:boosting_mle]",
   "language": "python",
   "name": "conda-env-boosting_mle-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}