563 lines
132 KiB
Plaintext
563 lines
132 KiB
Plaintext
{
|
||
"cells": [
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "42caa450-1beb-4fe1-a233-98cc1ddf7053",
|
||
"metadata": {},
|
||
"source": [
|
||
"# Calibration — калибровка вероятностей классификатора\n",
|
||
"\n",
|
||
"## Цель работы\n",
|
||
"Изучить методы калибровки вероятностей классификатора и сравнить их влияние на качество предсказаний.\n",
|
||
"\n",
|
||
"## Используемый алгоритм\n",
|
||
"Logistic Regression + CalibratedClassifierCV\n",
|
||
"\n",
|
||
"## Датасеты:\n",
|
||
"1. Встроенный (make_classification)\n",
|
||
"2. Внешний (через pandas / openml)"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 1,
|
||
"id": "279cb38d-792f-4e96-9fdd-d1f7e9e38c26",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"import openml\n",
|
||
"import pandas as pd\n",
|
||
"import matplotlib.pyplot as plt\n",
|
||
"\n",
|
||
"from sklearn.datasets import make_classification\n",
|
||
"from sklearn.preprocessing import LabelEncoder, StandardScaler\n",
|
||
"from sklearn.model_selection import train_test_split\n",
|
||
"from sklearn.linear_model import LogisticRegression\n",
|
||
"from sklearn.calibration import CalibratedClassifierCV, calibration_curve"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 2,
|
||
"id": "8590b5b5-3b27-40ac-9939-ad3a97b63586",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"X, y = make_classification(n_samples=1000, random_state=42)\n",
|
||
"\n",
|
||
"X_train, X_test, y_train, y_test = train_test_split(\n",
|
||
" X, y, test_size=0.2, random_state=42\n",
|
||
")"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 3,
|
||
"id": "d2ae71d6-2668-40bf-8bb5-c956636afffe",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"model = LogisticRegression()\n",
|
||
"model.fit(X_train, y_train)\n",
|
||
"\n",
|
||
"probs = model.predict_proba(X_test)[:, 1]"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 4,
|
||
"id": "cdfa8de5-e650-48e7-841c-19eb87c0f6bb",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"calibrated_model = CalibratedClassifierCV(model, method='sigmoid')\n",
|
||
"calibrated_model.fit(X_train, y_train)\n",
|
||
"\n",
|
||
"calibrated_probs = calibrated_model.predict_proba(X_test)[:, 1]"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 5,
|
||
"id": "5bf0bc01-9d83-416c-a3bc-31a4bbd8ec30",
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"data": {
|
||
"image/png": "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",
|
||
"text/plain": [
|
||
"<Figure size 640x480 with 1 Axes>"
|
||
]
|
||
},
|
||
"metadata": {},
|
||
"output_type": "display_data"
|
||
}
|
||
],
|
||
"source": [
|
||
"prob_true, prob_pred = calibration_curve(y_test, probs, n_bins=10)\n",
|
||
"prob_true_cal, prob_pred_cal = calibration_curve(y_test, calibrated_probs, n_bins=10)\n",
|
||
"\n",
|
||
"plt.plot(prob_pred, prob_true, marker='o', label='Без калибровки')\n",
|
||
"plt.plot(prob_pred_cal, prob_true_cal, marker='o', label='С калибровкой')\n",
|
||
"\n",
|
||
"plt.plot([0, 1], [0, 1], linestyle='--') # идеальная линия\n",
|
||
"\n",
|
||
"plt.xlabel('Предсказанная вероятность')\n",
|
||
"plt.ylabel('Истинная вероятность')\n",
|
||
"plt.title('Калибровочная кривая')\n",
|
||
"\n",
|
||
"plt.legend()\n",
|
||
"plt.grid()\n",
|
||
"plt.show()"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "4cf9051c-64ef-46cd-a1dd-92c607f7c0d9",
|
||
"metadata": {},
|
||
"source": [
|
||
"## Интерпретация результатов\n",
|
||
"\n",
|
||
"- Без калибровки модель может переоценивать или недооценивать вероятности\n",
|
||
"- После калибровки предсказания становятся ближе к реальным вероятностям\n",
|
||
"- Калибровка улучшает интерпретируемость модели"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "5b172d77-26f4-4bf6-9d19-00debebeb04c",
|
||
"metadata": {},
|
||
"source": [
|
||
"## Работа с внешним датасетом (OpenML)"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "29ca3b05-f3a0-43b0-bfa6-3bcf98177a3a",
|
||
"metadata": {},
|
||
"source": [
|
||
"## Работа с внешним датасетом OpenML\n",
|
||
"\n",
|
||
"В качестве внешнего набора данных используется датасет OpenML с идентификатором 40691. \n",
|
||
"Он применяется для проверки работы модели на данных, которые не были сгенерированы внутри программы.\n",
|
||
"\n",
|
||
"Цель данного этапа — загрузить внешний датасет, выполнить предобработку данных, обучить модель логистической регрессии и сравнить вероятностные предсказания до и после калибровки."
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "70df55d6-8846-418d-af3e-06e3589e07f7",
|
||
"metadata": {},
|
||
"source": [
|
||
"## Предобработка данных\n",
|
||
"\n",
|
||
"Так как внешний датасет может содержать не только числовые, но и категориальные признаки, перед обучением модели необходимо привести данные к числовому виду.\n",
|
||
"\n",
|
||
"Для этого используется `pd.get_dummies()`, который преобразует категориальные признаки в набор числовых столбцов."
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 6,
|
||
"id": "0f631ab3-2ee9-42ab-ad33-9cdd39effb08",
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"name": "stdout",
|
||
"output_type": "stream",
|
||
"text": [
|
||
" fixed_acidity volatile_acidity citric_acid residual_sugar chlorides \\\n",
|
||
"0 7.4 0.70 0.00 1.9 0.076 \n",
|
||
"1 7.8 0.88 0.00 2.6 0.098 \n",
|
||
"2 7.8 0.76 0.04 2.3 0.092 \n",
|
||
"3 11.2 0.28 0.56 1.9 0.075 \n",
|
||
"4 7.4 0.70 0.00 1.9 0.076 \n",
|
||
"\n",
|
||
" free_sulfur_dioxide total_sulfur_dioxide density pH sulphates \\\n",
|
||
"0 11.0 34.0 0.9978 3.51 0.56 \n",
|
||
"1 25.0 67.0 0.9968 3.20 0.68 \n",
|
||
"2 15.0 54.0 0.9970 3.26 0.65 \n",
|
||
"3 17.0 60.0 0.9980 3.16 0.58 \n",
|
||
"4 11.0 34.0 0.9978 3.51 0.56 \n",
|
||
"\n",
|
||
" alcohol \n",
|
||
"0 9.4 \n",
|
||
"1 9.8 \n",
|
||
"2 9.8 \n",
|
||
"3 9.8 \n",
|
||
"4 9.4 \n",
|
||
"0 5\n",
|
||
"1 5\n",
|
||
"2 5\n",
|
||
"3 6\n",
|
||
"4 5\n",
|
||
"Name: class, dtype: category\n",
|
||
"Categories (6, str): ['3' < '4' < '5' < '6' < '7' < '8']\n"
|
||
]
|
||
}
|
||
],
|
||
"source": [
|
||
"dataset = openml.datasets.get_dataset(40691)\n",
|
||
"\n",
|
||
"X, y, _, _ = dataset.get_data(\n",
|
||
" target=dataset.default_target_attribute\n",
|
||
")\n",
|
||
"\n",
|
||
"print(X.head())\n",
|
||
"print(y.head())"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 7,
|
||
"id": "5a4f1972-227d-440e-a6af-0eaccec4068c",
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"name": "stdout",
|
||
"output_type": "stream",
|
||
"text": [
|
||
"(1599, 11)\n"
|
||
]
|
||
}
|
||
],
|
||
"source": [
|
||
"X = pd.get_dummies(X)\n",
|
||
"\n",
|
||
"print(X.shape)"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 8,
|
||
"id": "d1005e07-6b9f-4016-996f-c7f393422517",
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"name": "stdout",
|
||
"output_type": "stream",
|
||
"text": [
|
||
"Исходные классы:\n",
|
||
"['3' '4' '5' '6' '7' '8']\n",
|
||
"Положительный класс:\n",
|
||
"8\n",
|
||
"0 1581\n",
|
||
"1 18\n",
|
||
"Name: count, dtype: int64\n"
|
||
]
|
||
}
|
||
],
|
||
"source": [
|
||
"encoder = LabelEncoder()\n",
|
||
"\n",
|
||
"y_encoded = encoder.fit_transform(y)\n",
|
||
"\n",
|
||
"# бинарная классификация\n",
|
||
"positive_class = y_encoded.max()\n",
|
||
"\n",
|
||
"y_binary = (y_encoded == positive_class).astype(int)\n",
|
||
"\n",
|
||
"print(\"Исходные классы:\")\n",
|
||
"print(encoder.classes_)\n",
|
||
"\n",
|
||
"print(\"Положительный класс:\")\n",
|
||
"print(encoder.classes_[positive_class])\n",
|
||
"\n",
|
||
"print(pd.Series(y_binary).value_counts())"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "4ff6cc3d-926b-433a-a36b-4f3f21d26d99",
|
||
"metadata": {},
|
||
"source": [
|
||
"## Подготовка целевой переменной\n",
|
||
"\n",
|
||
"Исходная целевая переменная внешнего датасета содержит несколько классов. \n",
|
||
"Для построения калибровочной кривой задача была приведена к бинарной классификации.\n",
|
||
"\n",
|
||
"В данной работе один класс рассматривается как положительный, а остальные значения объединяются в отрицательный класс."
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 9,
|
||
"id": "7337a1b6-c7f7-40e6-a352-9ff28e437bf8",
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"name": "stdout",
|
||
"output_type": "stream",
|
||
"text": [
|
||
"Исходные значения качества:\n",
|
||
"[np.int64(3), np.int64(4), np.int64(5), np.int64(6), np.int64(7), np.int64(8)]\n",
|
||
"Распределение классов:\n",
|
||
"class\n",
|
||
"1 855\n",
|
||
"0 744\n",
|
||
"Name: count, dtype: int64\n"
|
||
]
|
||
}
|
||
],
|
||
"source": [
|
||
"# target\n",
|
||
"\n",
|
||
"y_numeric = y.astype(int)\n",
|
||
"\n",
|
||
"# бинарная классификация:\n",
|
||
"# 1 — качество 6 и выше\n",
|
||
"# 0 — качество ниже 6\n",
|
||
"y_binary = (y_numeric >= 6).astype(int)\n",
|
||
"\n",
|
||
"print(\"Исходные значения качества:\")\n",
|
||
"print(sorted(y_numeric.unique()))\n",
|
||
"\n",
|
||
"print(\"Распределение классов:\")\n",
|
||
"print(pd.Series(y_binary).value_counts())"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "4c844eb0-e95d-4636-8f21-4032a61bad9c",
|
||
"metadata": {},
|
||
"source": [
|
||
"## Разделение данных на обучающую и тестовую выборки\n",
|
||
"\n",
|
||
"После подготовки признаков и целевой переменной данные разделяются на обучающую и тестовую выборки.\n",
|
||
"\n",
|
||
"Обучающая выборка используется для обучения модели, а тестовая — для проверки качества её предсказаний."
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 10,
|
||
"id": "1033c8a3-4962-4cef-8520-94cbb98e4e55",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"X_train, X_test, y_train, y_test = train_test_split(\n",
|
||
" X,\n",
|
||
" y_binary,\n",
|
||
" test_size=0.2,\n",
|
||
" random_state=42,\n",
|
||
" stratify=y_binary\n",
|
||
")"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "87b7a4be-25f1-4726-a92b-ce3e786ff1c7",
|
||
"metadata": {},
|
||
"source": [
|
||
"## Обучение модели без калибровки\n",
|
||
"\n",
|
||
"На подготовленных данных обучается модель `LogisticRegression`.\n",
|
||
"\n",
|
||
"После обучения получаются вероятностные предсказания с помощью метода `predict_proba()`. Эти вероятности будут использоваться для сравнения с результатами после калибровки."
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 11,
|
||
"id": "b3873576-69c8-4d05-933a-f1ee5ecb39d4",
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"name": "stdout",
|
||
"output_type": "stream",
|
||
"text": [
|
||
"Вероятности без калибровки:\n",
|
||
"[0.32289345 0.50326905 0.09113214 0.26805981 0.84234398 0.96209197\n",
|
||
" 0.12481713 0.10134277 0.19774492 0.83756812]\n"
|
||
]
|
||
}
|
||
],
|
||
"source": [
|
||
"model = LogisticRegression(\n",
|
||
" max_iter=1000,\n",
|
||
" class_weight='balanced'\n",
|
||
")\n",
|
||
"\n",
|
||
"model.fit(X_train, y_train)\n",
|
||
"\n",
|
||
"probs = model.predict_proba(X_test)[:, 1]\n",
|
||
"\n",
|
||
"print(\"Вероятности без калибровки:\")\n",
|
||
"print(probs[:10])"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "46313a16-a495-4615-9e83-eb5eafae57e3",
|
||
"metadata": {},
|
||
"source": [
|
||
"## Калибровка вероятностей\n",
|
||
"\n",
|
||
"Для калибровки используется `CalibratedClassifierCV` с методом `sigmoid`.\n",
|
||
"\n",
|
||
"Калибровка нужна для того, чтобы вероятностные предсказания модели лучше соответствовали реальной частоте событий."
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 12,
|
||
"id": "e453a306-7524-4d92-aea5-f8d04e99ebf1",
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"name": "stdout",
|
||
"output_type": "stream",
|
||
"text": [
|
||
"Вероятности с калибровкой:\n",
|
||
"[0.35424691 0.54027067 0.10630123 0.29422608 0.86167447 0.96691741\n",
|
||
" 0.13873573 0.1192158 0.22601277 0.85601576]\n"
|
||
]
|
||
}
|
||
],
|
||
"source": [
|
||
"base_for_calibration = LogisticRegression(\n",
|
||
" max_iter=1000,\n",
|
||
" class_weight='balanced'\n",
|
||
")\n",
|
||
"\n",
|
||
"calibrated_model = CalibratedClassifierCV(\n",
|
||
" estimator=base_for_calibration,\n",
|
||
" method='sigmoid',\n",
|
||
" cv=5\n",
|
||
")\n",
|
||
"\n",
|
||
"calibrated_model.fit(X_train, y_train)\n",
|
||
"\n",
|
||
"calibrated_probs = calibrated_model.predict_proba(X_test)[:, 1]\n",
|
||
"\n",
|
||
"print(\"Вероятности с калибровкой:\")\n",
|
||
"print(calibrated_probs[:10])"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 13,
|
||
"id": "949ce6b5-5f00-4e58-a882-351831923d5c",
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"data": {
|
||
"image/png": "iVBORw0KGgoAAAANSUhEUgAAArMAAAIjCAYAAAAQgZNYAAAAOnRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjEwLjksIGh0dHBzOi8vbWF0cGxvdGxpYi5vcmcvJkbTWQAAAAlwSFlzAAAPYQAAD2EBqD+naQAAsj9JREFUeJzs3Qd4FNUaBuAvvYceQg899BJ6F5DeVHoHwXaRXqWDIk0EQUFBBAQEBJTeFZDee+81EAikl0127/OfdWMS0jZsyibfe5+97MzOzszOYPhy9j/nWOh0Oh2IiIiIiMyQZVqfABERERFRcjHMEhEREZHZYpglIiIiIrPFMEtEREREZothloiIiIjMFsMsEREREZkthlkiIiIiMlsMs0RERERkthhmiYiIiMhsMcwSUabXoEED9TC4d+8eLCwssGzZsqh1vXv3hrOzc5pfq0mTJqlzIyIiPYZZIjI7t2/fxscff4wiRYrA3t4erq6uqF27NubNm4eQkBCYu+DgYBVa9+/fj/QmNDQU3377LapXr44sWbKo61+iRAkMGDAAN27cSOvTI6JMyDqtT4CIyBjbtm1Dhw4dYGdnh549e6Js2bIIDw/HoUOHMGLECFy+fBk//fTTW13UQoUKqVBsY2OTZmF28uTJ6nn0FmMxbtw4jB49Ok3O68WLF2jWrBlOnz6NVq1aoWvXrqq1+vr161izZo267nIviIhSE8MsEZmNu3fvonPnzips/vXXX8iTJ0/Ua//73/9w69YtFXbflnyNLy2OphIREQGtVgtbW9u33pe1tbV6pAUptTh79izWr1+PDz74IMZrU6dOxdixY9Pd9SKijI9lBkRkNmbOnInAwED8/PPPMYKsQbFixTBo0KCo5V9++QUNGzaEm5ubasktXbo0Fi5cmOhx4qqZNbhz5w6aNm0KJycn5M2bF1OmTIFOp3vjvbNnz8bcuXNRtGhRdewrV66oVssJEybAy8tLfUUv+6hbty7+/vvvGO/PlSuXei6ts7IveUjZQXw1sxL+JEwajuXh4YEvvvgCYWFhMbaT9dKiKq3Y1apVU4FdSjVWrFiR6DU5fvy4+kXhww8/fCPICjmufOb46pCjB2I5j8Sul4RmCe2GFuropCVY3rNgwYKoda9fv8bgwYNRoEAB9X75uzBjxgwViokoY2PLLBGZjS1btqjwVatWrSRtL8G1TJkyaNOmjQpG8v7PPvtMBRxpyTVWZGSk+pq9Ro0aKljv3LkTEydOVGFSQm10EqSlvvSjjz5S4Sp79uzw9/fHkiVL0KVLF/Tv3x8BAQEqmEs4PnHiBCpWrKiCrJz3p59+ivfeew/vv/++2l/58uXjPa9+/fph+fLlaN++PYYNG6aC59dff42rV6/ijz/+iLGttF7LdhJKe/XqhaVLl6qAKQFbrlV8Nm/erP7s0aMHUkLs6yW/rNSvXx/r1q1T1zi6tWvXwsrKSpWbGMoyZNvHjx+rWuqCBQviyJEjGDNmDJ4+fapCMhFlYDoiIjPg5+cnzZ+6tm3bJvk9wcHBb6xr2rSprkiRIjHW1a9fXz0M7t69q471yy+/RK3r1auXWvf5559HrdNqtbqWLVvqbG1tdT4+PjHe6+rqqnv+/HmM40REROjCwsJirHv16pUud+7cur59+0atk33JPiZOnPjG+cu66D+6z507p5b79esXY7vhw4er9X/99VfUukKFCql1Bw8ejFon52hnZ6cbNmyYLiHvvfeeeq+cb1LEvqbRr6Och0FC1+vHH39Ur128eDHG+tKlS+saNmwYtTx16lSdk5OT7saNGzG2Gz16tM7Kykr34MGDJJ0zEZknlhkQkVmQVk3h4uKS5Pc4ODhEPffz81MdmKQFT0oFZDk5pNe+gXzVLctSPrB3794Y28lX8YZyAQNpTTTUgUrrsK+vr2rVrVKlCs6cOZOs89m+fbv6c+jQoTHWSwutiF1DLKUWUtpgIOdYsmRJdU1Mff2NEdf1klZpaVGXlliDS5cuqZKNTp06Ra37/fff1WfKli2buseGR+PGjVVr+sGDB1PknIkofWCZARGZBRl+S8hX80l1+PBh9RX10aNH1VfR0UmYlbpVY1haWqoyh+hkWCpD7Wd0hQsXjnMfUg7wzTff4Nq1a9BoNIlun5j79++r85Ia0ejc3d2RNWtW9Xp08hV8bBICX716leTrL/s1tbg+f86cOdGoUSNVaiA1wUKCrQRcQ/mFuHnzJi5cuPBGGDZ4/vy5yc+XiNIPhlkiMgsSpqTDlbTMJXUsWglCnp6emDNnjuoYJK2i0pIp46SmdMeg6K3CBitXrlT1qe3atVPDiEnHNGmtlfpWOd+3kdSJFOR4cYneiS0uch3FxYsXY7TsJnQ+ce1TWkqTer2EjF7Rp08fnDt3TtUUS7CV+ypB10Du5bvvvouRI0fGuQ/DLxxElDExzBKR2ZCe+DKWqbS01qxZM8FtpbOX9OaXjkvRWyOjjxxgLAlN8nV89HBkmCggeg/9+MiQVtKyu3HjxhjhM3YHJ2Nm+JJhyuS8pHWyVKlSUeufPXumevjL66bQunVrFbolkCclzEprb1ylC7FbihMjwV86dRlKDeR6S8eu6GQEBBnlQsoKiCjzYc0sEZkNaXmT4ayk976EtdikdVNmAYveAhm9dVBKC6TX/NuIPhyU7FuWZXIFaS1MTFznJCMPSDiPztHRUf0pYTQxLVq0UH/G7rEvrdGiZcuWMAX55UFGcpDRGP788883Xpe64eHDh8cImFJK4ePjE7Xu/PnzqvTDGFLSIKM9SIusTMwgresScKPr2LGjuoa7du164/1yDaUumYgyLrbMEpHZkIC0evVq1flHWiGjzwAmQzFJRyD5Gl80adJEBR9pUZSWPWm5W7x4sfpqX4ZrSg4Zl1WG45IhrWQ61x07dqgOVjKma3z1mrFblqVVVobckpApk0AsWrRIdcqS84v+lbusk9ZIaQWWYb3kc8ojtgoVKqjzkRZrCW7SwU2G+ZLaXAl977zzDkxFxqOV6yr1qnJdJcDLLxfSKixBU66rYazZvn37qkAtQVSGAZO6VfmsMvyXoTNZUsn97t69O3744Qe1v9g1u1KyIS3wcn0Nw4wFBQWpkghpDZd65uhlCUSUwaT1cApERMaSIZj69++v8/DwUMNiubi46GrXrq2bP3++LjQ0NGq7zZs368qXL6+zt7dX286YMUO3dOlSNdyTDAll7NBcMvzT7du3dU2aNNE5OjqqIbVkqKzIyMg33jtr1qw3zluG8po2bZoamkqGw6pUqZJu69atbwxXJY4cOaLz8vJSny/6MF2xh+YSGo1GN3nyZF3hwoV1NjY2ugIFCujGjBkT41oIOYYMJZbUYbTiG+5s9uzZuqpVq+qcnZ3V+RUvXlwNWXbr1q0Y265cuVINgybbVKxYUbdr1654h+aK63oZ+Pv76xwcHNR2ss+4BAQEqM9crFgxdbycOXPqatWqpc41PDw8SZ+NiMyThfxfWgdqIiIiIqLkYM0sEREREZkthlkiIiIiMlsMs0RERERkthhmiYiIiMhsMcwSERERkdlimCUiIiIis5XpJk2QaR+fPHkCFxcXo6aMJCIiIqLUISPHBgQEIG/evLC0TLjtNdOFWQmyBQoUSOvTICIiIqJEPHz4EPnz509wm0wXZqVF1nBxXF1dU/x4Go0Gu3fvVlNAyvztZH54D80f76H54z00b7x/5k+TynlGpr2WxkdDbktIpguzhtICCbKpFWYdHR3VsRhmzRPvofnjPTR/vIfmjffP/GnSKM8kpSSUHcCIiIiIyGwxzBIRERGR2WKYJSIiIiKzlelqZpM6HERERAQiIyNNUmNibW2N0NBQk+yPUp853kMrKyt1zhx+joiIMjqG2VjCw8Px9OlTBAcHmywYu7u7q9ETGCzMk7neQynUz5MnD2xtbdP6VIiIiFIMw2ysCRXu3r2rWrVkkF4JAW8bXmSfgYGBcHZ2TnTQX0qfzO0eSviWX8p8fHzU3+fixYubxXkTERElB8NsNBIAJLjIuGbSqmUKsj/Zr729PQOFmTLHe+jg4KCGTrl//37UuRMREWVE5vEvcyozl8BClBD+PSYiosyAqY2IiIiIzBbDLGX4EgF5EBERUcbEMJtCIrU6HL39EpvPP8HJ+35qmVKedHoaPnw4ypcvDzc3N9jZ2am5pImIiChjYgewFLDz0lNM3nIFT/1Co9a5u97CpDal0axsnpQ4JHr37o3ly5e/sT5fvnx49OgRMoPnz5/Dy8sLVapUwXfffYf8+fOrulEPD4+0PjUiIiJKIQyzKRBkP115BrHbYZ/5h6r1C7tXTrFA26xZM/zyyy8x1skwY5nFmDFjUL9+faxcuTKtT4WIiIhSCcsMkjBmZ3B4RJIeAaEaTNx8+Y0gq/bz75+TNl9R2yVlf3JsY8hX6jK4f/RHrly5YmwjA/937NgRWbNmRfbs2dG2bVvcu3cv6vXRo0dHjbErrbqjRo1KsOZUWj3nzp0btTxu3DjVImrY58uXL9GlSxe1LxnurFy5cvjtt9/e2M+yZcvUmL7RHxUrVox6vUGDBhg8eHDU8vXr19XQU9G32bp1K7Jly4ayZcuqoaiKFSuGxYsXxziO7HfhwoVo3ry5Gr6qSJEiWL9+fYxtLl68iIYNG6rXc+TIgY8//liNMxu9FdxwjnKdPD098euvv0a9PmnSpBjndebMGXW9lyxZEuM8/vzzz6jln3/+Wa2L/hmJiIgonbfMHjx4ELNmzcLp06fVrFt//PEH2rVrl+B79u/fj6FDh+Ly5ctqPFgJTxIuUkqIJhKlJ+wyyb4kmnr7h6LcpKTVcF6Z0hSOttYmnZa1adOmqFmzJv755x813emXX36pWnQvXLigglmTJk1U2M2ZM6e6xl27dkWZMmXQs2fPRPf/zTff4Mcff1T31fDVvkwBK1/9Syh2dXXFtm3b0KNHDxQtWhTVqlWLeq8Ed3ldQqqYPXs29u7dG++xRowY8cbYqVIvK8eXcN24cWNVK/vZZ5+pUN+6deuo7caPH4/p06dj3rx5KoR27txZBdhSpUohKCgo6hqdPHlSlS7069dPrY/e4mtoBQ8LC8OCBQvQp08fvPfee2piheiuXbum9id/T2U/cZF9yznFfi8RERGl85ZZ+Ue8QoUK+P7775O0vcxm1LJlS7zzzjs4d+6casWSgLBrl2nCprmTlkkJRPKQ1tF33303RuentWvXqlZWaSGUFlIJbxLIHjx4oH5JENIiWblyZRQsWFC1OErrZGRkZKLHln1OmTIFO3fuVPs1kBZZ6ZAlLZXSCvr555+rILhu3bo3graEaUOLckLB7u+//8aRI0feCIcSiOUXGwmwJUqUwIABA9CrVy/MmDEjxnYdOnRQ75Vtpk6dqmps58+fr15bvXq1CuArVqxQLbxyPaT+Vq7ds2fP3mgFl+skLdlOTk5vlHTIhAVyDz766CN1DeIzc+ZMlC5dWoV+IiIiMqOWWfmqVx5JtWjRIhQuXFi1AAoJTYcOHcK3336rWr9SgoONlWohTYoTd33R+5eTiW63rE9VVCucPUnHNoaEfPkKXfj6+qoWQwn/co2qV6+O8+fP49atW3BxcYnxPglvt2/fjlqeNm2aarENCQlRgTCxVtlNmzap1lgJhxIAo5MgLPuT8Pr48WM1G5W0ZsaeYc3f318FwsRIYB02bBgmTpyoShhiq127dozlOnXqYOPGjTHWSatr7GX55UhcvXpV/YIV/Vxkn/JLgLQa58mTJ8YvDvJ5JIRLq60Ef4PXr1+r1mHpfJfQ380nT55gzpw56h4NGjQo0c9PRESUFiK0Een2wptVB7CjR4+qgBCdBIWE6gwlOMkjemgytATKIzpZlrAUe2xSe+ukNWDXLpoD7q72qrNXXNWuFjKqQRZ7tZ2VpSwlTM4lqXWzsp0ERGn9FPKntJZKkJPyjapVqyIgIEC1/kWv7zSQ2lrDZ5aWRCn3kPIPKemQ5xKU43P48GFVByu1ohIyJbxGb3WUr/MlsElrsITEIUOGqHsS/RpL0JUWTsM6w+eOvo2sk9paadGXczQcx7CN1KXGvnfyXGpRY6+LvV/D+viOa/jTsI3U8P7www/q74y0Rkvgl0As5RXyurTKSolGt27d0LdvX/Va7AAv+/riiy/Qvn17dW2iH8MUDOcq55iZOgLGxfDfeuz/5sl88B6aN94/86XT6fDn7T+x8tpKdEXXVPs5asxxzCrMent7I3fu3DHWybIEVGlFjN4yZvD1119j8uTJb6yXr99jhwupIZWvjqWzj7S4JceIRh4Y/sc1FVyjx1BDdB3e0ANBgQFIiZseERERFdYNYUaGpgoODlbrpSVbvi6XWlOpT43N8F65DhIs5SHbr1mzJt6vwOUY8suE1NpKZylpCZbnUqogDhw4oFrf27RpE7W9tHCWLFkyxrkeO3ZM1eYa1knYlVZdw7J8Nj8/P1V7KgFZ7nfsbYoXL66OJ7WrBlI+IS3G0Y8l9cLRa7OlZEHCpGwjYVQCs9RwG1pnpXZXrqOUTMg2cq2lzEDGsRUffvihCtbSWiuhVs5L9iMhXsgvE1JmIHW60Z04cQIbNmxQf8p+5TPK37vo5/o2ZF9ynaTVXPZNwJ49e3gZzBzvoXnj/TMvYbowbArehAuaC2r5hP0JOOx5M2ulBMkuGTLMJne4JmldNJCgIB3HJHDFDnTydbv09pevj2N3Lkqq96q6qlA9ZetV1dnLQFpkx7cshWZl3ZESpGe/BDvDzX/16pWqRZZgLsFNPquELlkndaTSiip1tdKCKGFLOlTJspQp1KtXTwU5+epbwqCUccQVfoWEPPnqXV6X1lv5qnzgwIE4deqU+vpdArQEtkuXLqmRBmRf0lFLgqu858WLF6rD1vHjx2McR8KitCYaliVgy6gDEqpldIS4tpH7LGFSWqGlBV9+aEqLsbwv+vlv3rxZlRZICYLUyEoL9NKlS6OukdTYymeQVmY5V/k71KlTJ9VpTVp5o19rCYnSMivXW+qCZR9yXlmyZFGjRQgZ/7dGjRrqvOvWrRt1HlIGIucswd7wGeWaxXetjSV/n+XvotzP5P59zijkFxD5+yA1zHL/yPzwHpo33j/zc/3VdYw6NAoPNA9gZWGFT8p+gtwPcqfaz1FjGnbMKsxKq2n0TjhCluUf/7haZYUEC3nEJjci9s2QgCJhRQKaPJKrRfm8aFo2j6qhfeYfAifLSDQokx821in3Va+ct3SEk9ZDIXWx0lIpYU06MQkJ6dJKJyMLyFfbUnYg2zdq1Eh9RS+fefv27SroymsS+uVrcOksJftP6NiG6yWdwLZs2aL+lNZK6aUvHfekdVZawg0lDNLKKu+RsCmt5BKoJfBF36eIfh8kPEq9tGFd7G1kVAIZfUC2kVKGQoUKqXAuw49FJy31UsMr9cASxOUcDLW+co3kOkoolzpjOef3339fBVvD54x+rSWASiusdCCT0BjXeUnIHTt2rLqOUrds+EZA7pHci+ifMfq1fFuGc43r73pmxWth/ngPzRvvn3mUFay7vg4zT85EuDYc7k7umFVvFspkK4PtD7en2j005hgWOmMHM00h8o9uYkNzyT/8ErZkGCUDqUuUzk7SOpbUpC+tZhKm4mqZleAlncxM1ZIlX6vLMeVYpgoplLJ/zzLKPUyJv8/m3CokPztatGjBYG+meA/NG++f+bjvfx/tNrVTHb4a5G+AqbWnIqt91lS/hwnltXTVMitfgUvvegP5h1c6ysjXszLkkXy9Kx2DZJgk8cknn6ivZkeOHKk61fz111+qhU3GLiUiIiKit1PItRBGVBmhwmyP0j0S/GY2vUjTMCt1ldF7yRtqW6Wm09AJR8ZANZAWJgmu8hWydK6RGk/psZ9Sw3IRERERZWQ6nQ6rr62GV24veGb3VOu6luoKc5KmYVaGN0qoykECbVzvOXv2bAqfGWVU6aSqhoiIKM35hflh4pGJ2Pdgn2qRXddqHRxtYo70ZA7MqgMYEREREb29Cz4XMOLACDwJegIbSxt09ewKB+vUGXbL1BhmiYiIiP4VqdWp0YieB4TCzcVezdiZlImOzOkbyhVXVmDu6bmI0EWggEsBzKo/C2VylIG5YpglIiIiArDz0lNM3nIFT/3+Gyc+TxZ7TGxdGs3K6qczN2fBmmCMPDgSBx4dUMtNPZpiUs1JcLZ1hjkzn3GGiIiIiFIwyH668kyMICu8/ULVennd3Nlb2yM8Mhy2lrYYX2O8Gj/W3IOsYMssERERIbOXFkiLbFxdhGWdFBnI6++Wdje7kgOtTquG2bK1soWlhSWm1Z2GlyEvUTK7fvbJjIAts5TuyUDNREREKUVqZGO3yMYOtPK6bGdOXoa8xGd7P8O049Oi1uV0yJmhgqxgmE0p2kjg7j/ApfWwfnhUv0yJioiIwJw5c1C7dm01XazMXCVT4pqrMmXKwMfHR81gItMLyzTBRESUvkhnL1Nulx6c9D6JDls64PCTw9h2ZxseBTxCRsUyg5RwZTOwcxTg/0T9tiDVKDrXvECzGUDpNkgp3t7e+Oqrr9TEEjJzmpubGypWrIjBgwejUaNGMIcelq1bt1bnPnnyZBUEZfpYCbXmSiYAyZs3r5oSd8CAAXBxcUnrUyIiolhk1IIkbyeNU/ePAIHPAOfcQKFagKVVurmmkdpILL64GAvPL1QlBkWyFMHs+rOR3yU/MiqG2ZQIsut6/vulRDT+T/XrO65IkUB779491ZqZNWtWzJo1C+XKlVNfz+/atQv/+9//cO3aNaR3K1euVJ/j5MmTcHY2/4J0IVMvyzTM0uIs0zQTEVH6I3WwUgkb37Q68pp7FntUCz0EzB2tGquipEJjVVK9CHmB0f+MxvGnx9Vy26Jt8UX1L8xyIgRjsMwgMTJjVHhQ0h6h/sCOkXH+52BhWCcttrJdUvZnxGxVn332mZo/+cSJE/jggw9QokQJ1bIpUwQfO3Ys3vf17t0b7dq1i1resWOHCpLyp8GoUaPU/hwdHVGkSBH1tX/sOlYJoXL82I/Xr1+r1ydNmqRaiQ3Cw8NRrFixGNts3boVpUuXRsuWLVULZu7cudXUxbJt9BngpIVTHlmyZEHOnDnV+USf2evVq1fo2bMnsmXLps65efPmuHnzZoyZ5QznZ2VlpVpO5TNK66nYv39/jPOS/ckvCtLKajhOWFgYBg4cqFq/pRSiTp06KoQbRN+Hq6urCrI9eujnuP7zzz+TfF+JiChlSR1sn19ORP3LHbt7l2H5h8qPYPV7r5hBNnpjlTRmpSGtTov+u/urICuTH3xV5yt8WefLDB9kBVtmE6MJBqblNdHl1un/I5heIGmbf/EEsHVKdDNfX1/s3LlTlRg4Ob25vbTWJsU///yDjh074ueff1YB0ECCpQRACX0XL15E//791TppdYz6ZP+GvL1796oQfeTIERWq47NgwQI8e/YsxjqpLf3777/x6aefYtGiRbhz5w769eunSg2++eabqO2WL1+ODz/8UAX3U6dO4aOPPkLBggXVeRkCuoTXzZs3qyApQbVFixa4cuUKbGxs1Day/vr164iMjMShQ4fQuXNnFZSjf24RGBiIVq1awcPDQ10XCaNCPvuGDRvUuRQqVAgzZ85E06ZNcevWrThbYE+fPq3Oh4iI0o+jt1+i77KTCNFEonaxHOhYpQCm77gWozOYtMhObFUSlfa8G0/b7b/jHewcDXi2TLOSA0sLSwzxGoJ5Z+apSRCkvCCzYJjNACRASZj09PRM9j7OnDmj6lUlNHbq1CnGa+PGjYt6LqFu+PDhWLNmTYwwa2ipdXd3V4+EvlKX8P3ll1+qkBm9c5e0jJYsWRLff/+9Co2lSpVSJRMSXKdOnapaWUWBAgXw7bffqm1kewnYsixh1hBiDx8+jFq1aqntV61apd4jLaIdOnRQ6+S9cp6icOHCKjBLS2900vravn17ddylS5fC2lr/n0tQUBAWLlyoAr4h/C5evBh79uxRgXfEiBFvfGZpIZf15tyZjYgoI/nnpg/6rziFUI0W9Urkwk89vGBvY4VW5fO+OQPY/UNvtsi+0Vj1WF9LW7huqn2G58HP8cD/Aaq4V1HL9fLXQ628tWBtmbniXeb6tMkhzfPSQpoU8pd4VfvEt+u2Xl8wnpRjJ0H0r9iT4+7du6pVMTQ0VLVOxrZ27Vp89913uH37tmqplPpPadmMzt/fX/0ZV8twbFOmTME777yjvpqPrWbNmlGtn0K2kTIDCezly5dX62rUqBFjG3mPhHBpZb169aoKndWrV496PUeOHCr0ymsGMrqAlFPIeyS0jh49Oir8GnTr1g379u1TJRJ2dnZR6+U6SHiX0gMDafGtVq1ajGMYSIiWVuZhw4YxzBIRpQN/X3+Oj389jfAILRp5uuH7bpVVkDXUz9YsmiPmG6SzV1IkdTsTOPz4ML449AU0Wg3Wt16PvM76b5EzW5AVme8TG0tCUxK+6leKNtQXgkv9TJxfRVjoX5ftTPg1hAz5JOEuuZ28Lly4oMLc8+fP0bdvXxw8eFC1VIqjR4+qUCejC0jgldZLaZWN/rW/ePLkiXqPobUzPtJyumTJEpw7dw6PHsUcJkRqXOMTPbyagpRJSGu0/CJw+fJl9bm9vLxilEbI6BBSStC1a1c0btxYhWZjSeiVFmwpAXFwcDDpZyAiIuPtufIM/1t1BuGRWjQtkxvzu1SGrXUiXYhk1IKkSOp2b0EmQFhwdgF+vvSzWvbM7qlGMMjM2AHMpFfTSt+jUYkZvnSG5WbTTV5PI1/pS9CUr+flK/DYDB2Z4lOvXj18/fXXanzX+/fvY968eVGvSe2r1ISOHTsWVapUUcFZtolNOj9JmYN0hkqIlBZIHax0/opN3i/hOXpLs9Sz2traomjRolHrjh/X99I0kA5ucl7SmUtKE6TlOPo2L1++VPWx0rnMQIK3nIO8TzrANWzYEH/88UeM/Uq5wvvvv6/OVzqcyX6FnIuck5QyRA+tcg2iH0NIOYK0AEvnLyIiSls7LsqUtadVkG1ZLg8WdE1CkBUFqifybak0VuVL2reub8E7yBt9d/WNCrKdSnbCyhYrUcA1iX1xMiiGWVOToTlk+C3XPDHXS4tsCg3LJSTIylfm8lW3tCZKC6h85S3lAYm1KBpaRKXV9aefflI1sobe/xL2Hjx4oFpj5et12V/00CclAL/++qsKwn369EnwOFIqIL38J0yYEOfr0vFLRkWQocTk3Ldv367qTCVIGuplhZyP1KBKQP3tt98wf/58DBo0KOp827Ztq+pnJQifP38e3bt3V2PVynoDCczS8vr06VPV6ezAgQNv1Bwb6n4l6MsvBDNmzIgqpZBzlXOTjnfSsUyOFxwcrOp7o5OOYdKKbeqWZSIiMs6W808w4LeziNDq0LZiXszrXBE2VkmIQRHhwB8f6zuExynlGquiO/joINpvaY+zz8/C2cZZjR07rsY42Fn9VwaXWbHMICVIYJUejfePQBvwFMEWLnAs1RgW1vqe9ClBhsySr83l62ypzZSQlitXLvXVubQOJpV0aJKe/fK1uwS8Nm3aqOGxJFBKbakMmyWdmKSOVEjnK3ku6yRgJkRajaVcIb7OYTIigQzPJSUPFSpUUCFbShwkTEYnw26FhISo4C6tsRJkZUQDg19++UWtk1EIJGxLy7MEY8NIBoYa3zx58qiQKUOAScewuDpuGcKrjL4gncHee+89lC1bFtOnT1cd1qTFVWb1klZrGdM3dqmE1AbLg4iI0s4fZx9h2Lrz0OqADyrnx8z25VVtbKLCg4F1PYBbewFLG6D6R8DlP+IYZ3Z6io8zK2HWL8wPZXKUUaMVFHDJ3K2x0Vno3rb3kJmRECMtkNIBKHYnJukAJZ2hpHd7Yl+XJ5UEHjmmHMtQh0rJJx3UZLzauXPnptplNNd7mBJ/n82VlIHILzQyRFv0X2rIfPAemre0vH+/n3qIkRsuqKHbO1ctgGnvlYNlUoJsyCtgdSfg4XF9iUGnX4FijdNsBrCwyDCsuroK3Ut1h62VLTL6PfRPIK/FxpZZIiIiypBWH3+AL/64qJ53r1EQU9qUTVqQDXgGrHwfeHYJsM8CdP0dKPjvKDkSXFNh+K19D/Zh251tmFVvFqwsrVQ5Qd+yfVP8uOaIYZaIiIgynBVH72HCpsvqeZ/aHpjQqnTS+i+8ugesaAe8uqtvee2+EXAvi9QSHhmOOafnqFZYsfHWRnQooR8jneLGMEtmRTqQERERJWTJP3fw5Tb9uN8f1SuCMc09kxZkn10Bfn0PCPQGshYCev4JZE+9mbQe+j/E8IPDceXlFbXcu0xvtCv235TzFDeGWSIiIsowFh24raakFf97pyiGNymZtCD76BSw8gMg9DWQqxTQ4483RyZKQbvu7cKkI5MQqAlEFrssmFZnmprRixLHMEtEREQZwvx9N/HNnhvq+eDGxTGokX5SoUTd/htY0w3QBAH5qwJd1wGO8U/LbmpLLi7BvDP6Md4ruVXCzHoz4e6U8CRE9B/z6ZpNREREFAcZmGnOnhtRQXZE05IY3LhE0oLslU3A6o76IFvkHaDHn6kaZIW0wDpYO6BfuX5Y2nQpg6yR2DJLREREZh1kZ+26jh/231bLUh/7cf3/Zo1M0JkVwJZBgE4LlG4LvL8YsE6dSQju+d2DRxYP9bxEthLY9t425HLMlSrHzmjYMktERERmG2Snbb8aFWRlxIIkB9nD84DNn+uDbOWeQPtfUiXIhkaEqtrY9za9h/M+56PWM8gmH1tmiYiIyCyD7OQtV7DsyD21PLVtGfSo6ZGUNwL7JgOHvtUv1x4ENJ4MpMK043de38GwA8Nw6/UtWMACl15cQoVcFVL8uBkdwywRERGZFa1Wh/GbLmHV8Qcqg379Xjl0rlYwCW+MBLYNA07/ol9uPAmoMwSpYdOtTfjq+FcIiQhBDvscmF5vOmrkqZEqx87oGGaJiIjIbERqdRiz8QLWnXqkguys9hXQ3it/4m+MCAf++Ai4/AcAC6D1XMCrd4qfb7AmWIXYzbc3q+Xqeapjet3pyOmQM8WPnVmwZjaD6N27N9q1izmw8v3792Fvb5+03pxERERmEGRH/H5eBVmZlXZup4pJC7LhQcBvnfVB1tIG6PBLqgRZsfPeThVkLS0sMaDiAPzY+EcGWRNjy6wRv1nFxzBnclzbarVa9ZWCtcYalpaW6i+zvbV9ovt1tHHE2xo/fjyDLBERZQgRkVoMXXcem88/gZWlBeZ1rohW5fMm/saQV8CqjsCjE4D829ppJVCsEVLLe8Xew8UXF9GicAtUda+aasfNTNgym0TVV1eP9zHk75j1Ng3WNYh6reaammiyrYn6U5Y/3ftpjG2bbWgW5z7f1sWLF7Fq1Sp8/vnncb7u4eGhgm70x59//hkjhH/99dcoXLgwHBwcUKFCBaxfvz7q9cjISHz44YdRr5csWRLz5ukHfI49/Wzs42TNmjXBFuXoJk2ahIoVK8a5z9evX6vlly9fokuXLsiXLx8cHR1Rrlw5/PbbbzHeExgYqI6VO3fuGOdy7ty5BK+jvMfKygrZsmVTf8p7Bg8eHPV67Ov2888/v7GNXOu5c+e+sd/on7tBgwYx3nP9+nXY2NjE+OxyzYcOHao+p/xiFNd9IyLKiDSRWgxcc1YFWRsrC3zftXLSgmyAN/BLS32Qtc8C9NyU4kE2SBOEOafmqD+F/JyeWHMig2wKYpjNoEaPHo3WrVujVq1a8W4zZcoUPH36VD1ikyC7YsUKLFq0CJcvX8aQIUPQvXt3HDhwICrs5s+fH7///juuXLmCCRMm4IsvvsC6deve6G1qCGdynNihzhRCQ0Ph5eWFbdu24dKlS/joo4/Qo0cPnDhxImqbadOmYffu3er85Dyiv5YQOf+mTZvi2rVrePz4MWrWrBnvtkFBQao13NnZ+a0/04gRI1SJSHQSlH/66Sd1Tx49ehTnfSMiymjCIiLx2aoz2H7RG7ZWlljYzQvNyiZhdizfu8DSpsDzy4BzbqDPDqBAtRQ91+u+19F5a2f8cvkXTDs+LUWPRf9hmUESHe96PMEyg+j2d9wf9VxCX0BAAFxcXKLKDKLb+cFOmNrBgwexa9cu1TorITIuYWFhyJ49O9zd3eN8TcLf3r17o8JbkSJFcOjQIfz444+oX7++ajWcPHly1Hukhfbo0aMqLHbs2DFqvUajUX9Ka6KTkxOyZMli8s8r+x4+fHjUsrRGy+eXc6lWTf+DS1pgW7Vqpc7dEICTQs5fwqm06Lq6usLW1jbebWfOnInSpUsjIiLirT7P33//jSNHjqBfv37quYF8BvnlRH5JISLKDEI1+iD717XnsLW2xE89vNCgpFvib3x2Bfj1PSDQG8jmoZ/VK3vhFDtPafj4/cbvmHFiBsK14cjtmBvtS7RPseNRTAyzSWRMDWv0bSXMRlhHqHUSZt9mv8a0yvbq1QulSpWKN8z6+vqqcBaXW7duITg4GO+++26M9eHh4ahUqVLU8vfff4+lS5fiwYMHCAkJUa/HLgnw9/dXn1tKEeKzdetWFRglIBcsWBCDBg1C3759o16XUB69tVO+bo9OliV8S3iV1lM5DwnkUnIQPWzv2bNHvS7hN6nk/HPkyJHodk+ePMGcOXNU4Jfzj23UqFEYN25c1LKcX8uWLeP8gThs2DBMnDhRlU9EJ59h7dq1qpXY09MzyZ+BiMhcg2z/Fafwz80XsLexxJKeVVGneBJGAHh4EljVHgh9DbiVBnr8AbgkoSU3mQLCAzD56GTsurdLLdfPXx9f1v4SWe3/K6mjlMUwm8H88ccfOHv27Btf90cnX1FL4JNwFBepLxXytX3s4Gdnp+/otmbNGtUa+s0336jWW2l5njVrFo4fP/5GyJNWzbiCvME777yDhQsXqlbQ7du3qxZJqXutWlVfKC/1uJs364c0EXIMKXkwkONKva6UMMj7pAVY6k/lMxpIGcSNGzdUaYS8bih/SIycv+wzMWPHjkWHDh1UbXF8ZQNSJxs93MYO5UJKO6Rc4ZNPPsFXX30V47XPPvsMp06dQpkyZdR9SOiaEhGZs+DwCPRbfgpHbr+Eo60Vfu5VFTWLJt6wgNt/AWu6AdK5On9VoOs6wDF7ip3nrVe3MPDvgXgY8BDWFtYY7DUYPUv3ZOfrVMYwm4FIOJJQJV+zS2iLj9S9SktplSpV4nxdviqXsCQtroav5WM7fPiw+spbApbB7dv66QSjO3nyZIzW3LhIuCxWrJh6Lq3J06dPx/nz56PCrHy1b3jdEMZjn0vbtm2jAq60hktwlc9hIIFaWkzPnDmjArOUGUinq4RIqLx69aoKngmRr/+lc1x8reAiZ86cMT6DhH9DBzYDaQ2X+7dgwQLVSh3XdRo5ciR27typWmhlf8WLF0/w3IiIzE1gWAT6LjuJE3d94WRrhWV9q6GqRxIC6eU/gQ39AK0GKNpQP2qBrVOKnqu0vsqoRHmd8mJW/Vkon6t8ih6P4sYwm4FIjat0GhozZky820jglLAo4S92mJJlac2UoCWtrtLpS4JhnTp14Ofnp0KjlCZICYOEKGlFlNpUaeH99ddfVXA1tPZK6+6SJUuwevVqFbwSIseQcGlomZWv18uWLZvkzy3nImFS6kxl1AH5uv/Zs2cxwuydO3fUecs5V69eHffu6ac/jI98lS/BUUZeaN68uTrH+MyePVuVBuTNm4SetQmQayUd2eIb3UFKQ9q3b6/uX7Nmzd7qWERE6VFAqAa9fzmJ0/dfwcXOGss/rIbKBbMl/sbTy4GtgwGdFijdDnj/J8D6vyEzTSk0IjRqiE2Z+OCHxj8gn3M+ZLEzfZ8QShqG2QxEAqHUWkqgi0+jRo3UZArS619KBaLr06ePGkZKWiynTp2KXLlyqVENJAhKqKtcubIasUB8/PHHqpyhU6dO6usUGRpLWml37NihXpf61MWLF6sOYxLAErJlyxbVUmxtba2OP3/+fNSokfQp/qQWVc5RRh2QOlkZzUACoQRwIfW8H3zwgTq/uOpU4xsSTDpyyS8IUq8rtbPxkfAvwfdtScuslG3ERcoipOVZfrH49NOYw7sREWUEfiEa9Fx6AucfvoarvTVW9quO8vmTUHd6aC6wd6L+eeVeQKtvgVgds03lgs8FjDgwQpUTNC/cXK0rneO/hhNKGxa6pBYPZhASSqRHvQSd2B2gJAzevXtXtS7GHhYpuaRFT44px0oPNY4SFmWcVvkzNgmAUmua2NfvmU16u4dJlRJ/n82VodW/RYsWcZZwUPrHe5ix79/r4HB0//k4Lj32RzZHG/z6YXWUzZdIS6fEl72TgMP/DvlYZwjQaKIM7Gry85eotOLKCsw9PRcRugh4ZvfE2lZr3xihKCPTpPLP0YTyWmxsmc1kpLVVBv+Pi7ToJjT0FBERkam9DAxD959P4OpTf+RwssWq/tXh6Z5weIE2Etg6BDizXL/ceDJQ57+JZ0zpdehrjDs8Dgce6cdZb1KoCSbVmpSpgmx6xzCbyUhda3x++eWXVD0XIiLK3HwCwtBtyTHceBaInM52+K1/dRTP7ZLwmyLCgY39gSt/AhIoW80FvHqlyPmde34OIw6OgHeQN2wtbTGq2ih0KNGBoxWkMwyzRERElOqe+4eiy+JjuO0ThNyudljdvwaK5kpkBsXwIGBtd/0QXJY2wAdLgDLxT4n+Nh4FPEKfnX1UWUEh10KYXX+2Ki+g9IdhNg6ZrIyYMij+PSai9OqpXwi6Lj6Ouy+CkDeLvQqyHjkTGUYr2BdY3Ql4dAKwcQI6r9QPwZVC8rvkR7dS3eAT4oMJNSfASY5J6RLDbDSGgmbpVZ7QjFVE5kD+Hgt2eCKi9OTRq2AVZB/4BiN/Ngf81r8GCmRPZDbMAG/99LTPrwAys1a39UAB/VjkpnTS+yTyO+dHHuc8anmI1xBVGyuj9lD6xTAbjXSMkiGonj9/rpZlmKe3/QssPeFl7FbpWW5OPeHJfO+htMhKkJW/x/L3Ob4Of0REqU0CbM9fTuPx6xAUyuGoWmTzZU2k8cj3LvBrO+DVPcDZXT89bW7TDocVqY3E4ouLsfD8QpTLWQ6/NPsFNpY2sEqhIb7ItBhmY3F318/fbAi0pggWMs6ptPTyNzvzZK73UIKs4e8zEVFaex4CdPv5JLz9w1Akp5MKsu5ZEhk28Nll4Nf3gUBvIJsH0ONPIHvcU7En14uQFxj9z2gcf6qfjl3qYyO0ESrMknlgmI1FwkqePHng5uamxlR7W7KPgwcPol69evy610yZ4z2U82SLLBGlF9LJa/5lK/hrwlDMzRmr+1eHm0siQfbhCWBVeyDUD3ArA/TYCLiY9hd0CbCjDo7Cy9CXcLB2wNjqY9G2WFuTHoNSHsNsPCQImCIMyD5kJikZtN5cghDFxHtIRJR8170DVIusv8YCJXM7Y1X/GmoYrgTd2qcftUATDOSvBnRbBzgkYVpbI8oKFl1YhB/P/wgddCiWtZgaraBo1qImOwalHoZZIiIiShFXnvirmb18g8KRz1GHFX2qJB5kL/8BbOgPaDVA0UZAp18BW9OOJCDDbf314C8VZD8o/oEaP1ZaZsk8McwSERGRyV167KeC7OtgDcrlc0WXPL7I7pTILJOnlwFbZCYvHVDmPeC9nwBr089MaWdlp1pir7y8gpZFWpp8/5S60n/XbCIiIjIr5x6+RtfFx1SQrVQwK5b39oJTYpV2h74FtgzSB1mv3sAHP5ssyEqHrrmn56qyAoPCWQozyGYQbJklIiIikzl93xe9lp5EYFgEqnpkw9LeVWGfUBcUmaho70Tg8Dz9cp2hQKMJ0iPbJOcjU9GOPDgSZ5+fVWPGNivcTI1YQBkHwywRERGZxPE7L9F32UkEhUeiRpHs+LlXVTjZWcc/OpA2Etg6GDizQr/87lSg9kCT3Y2Djw7ii0NfwC/MD842zphYayKDbAbEMEtERERv7citF/hw+SmEaCJRp1hOLO5ZBQ62CTTJRoQBG/sDVzYBFpZA63lA5Z4muRMarQbfnfkOyy4vU8ulc5TG7HqzUcC1gEn2T+kLwywRERG9lYM3fNB/xSmERWhRv0Qu/NjDC/Y2CQTZsED90Ft3/gasbIEPlgCl25psoptP9nyCE94n1HK3Ut0w1GsobOU4lCExzBIREVGy/X3tOT7+9TTCI7VoXMoN33erDDvrBIJssC+wuiPw6CRg4wR0XgUUfcekkx819WiKq75XMbXWVDQq1Mhk+6b0iWGWiIiIkmX3ZW/8b/UZaCJ1aFomN+Z3qQxb6wQGSgp4CqzpBDy/AthnBbpvAPJXeeurHx4ZjmdBz6LKCDqU6ICGBRsip0POt943pX8cmouIiIiMtuPiU3y2Sh9kW5bPgwVdEw6yjmHPYL2ilT7IuuQB+u40SZB9GPAQPXb0QL/d/VRHL0PrLINs5sEwS0REREbZfP4JBvx2FhFaHdpWzIt5nSrCxiqBSPHsMure+BIWr+8D2Qrrg6xbqbe+6rvv7UbHLR3V5AdBEUG473//rfdJ5odlBkRERJRkG888wvDfz0OrAz6onB8z25eHlWUCY8I+OA7r1R1gE+EHnVsZWPT4A3DJ/VZXPCwyDLNOzsLa62vVciW3SphZbybcndx5JzMhhlkiIiJKknUnH2LUxgtqnoPOVQtg2nvlYJlQkL21F1jbAxaaYLx0Kg7X7ptg45Lrra62tL4OPzAc13yvqeUPy36I/1X6H2wsE5tijDIqhlkiIiJK1Krj9zH2j0vqeY8ahTC5TZmEg+yljcDGjwCtBtoijXDUuTOaOmR96yv9/dnvVZDNZpcN0+pOQ518dXj3MjmGWSIiIkrQ8iP3MHHzZfW8T20PTGhVWnWyitepX4CtQ2TUV6DM+4hsvQCRu/aa5Cp/Uf0LwAIY5jUMuZ3erlyBMgZ2ACMiIqJ4LfnnTlSQ/bhekcSD7D9z9FPUSpCt0lc/IcJbTFhw5/UdfH/uezUZgshqn1XVxzLIkgFbZomIiChOC/ffxoyd+trUAe8Uw7AmJeIPshI290wAjnynX647DGg4XsbJAiK1ybrCm29vxpfHvkRIRAgKuBRAm6JteKfoDQyzRERE9Ibv9t3EnD031PPBjYtjUKPi8QdZbSSwZRBw9lf9cpMvgVqfJ/uqBmuCMe34NGy6vUktV3evjlp5a/EuUZwYZomIiCiKfJ0vIXb+X7fU8oimJfG/d4rFf4UiwoAN/YCrmwELS6D1d0DlHsm+ojdf3VSjFdzxuwNLC0t8WuFT9C/XH1aWCUyRS5kawywRERFFBdkZO69j0YHbavmLFp74qF7R+K9OWCCwthtwZ7++LvaDn4HSyS8F2H5nOyYemYjQyFDkcsiFGfVmoKp7Vd4dShDDLBEREakg++W2q/j50F11NSa2Lo0+tQvHf2WCfYFVHYDHpwAbJ6DLaqBIg7e6ktkdsqsJEaSkYFqdacjhkIN3hhLFMEtERJTJSZCdtPkylh/VTwc7tV1ZNZZsvPyfAr++B/hcBRyyAd02APm9kl0f62jjqJ7XyFMDvzT7Rc3oJSUGREnBvylERESZmFarw9g/L6kgK/27pr9fLuEg+/I2sLSJPsi65AH67EhWkJUAve76OjTf2BwP/B9ErffK7cUgS0ZhmCUiIsqkIrU6jNpwAauPP1BBdlb7CuhcrWD8b/C+BCxtBrx+AGQvAvTdBbiVMvq4geGBGHFwBKYemwrfUF/8fuP3t/sglKmleZj9/vvv4eHhAXt7e1SvXh0nTpxIcPu5c+eiZMmScHBwQIECBTBkyBCEhoam2vkSERFlBBGRWgz//Tx+P/0IMivt3E4V0d4rf/xveHAcWNYCCHoO5C6nD7LZEmjBjcfll5fRcWtH7Lq3C9YW1hheZTiGeMlsYURmWDO7du1aDB06FIsWLVJBVoJq06ZNcf36dbi5ub2x/erVqzF69GgsXboUtWrVwo0bN9C7d2817t2cOXPS5DMQERGZG02kFkPXnceW809gZWmB7zpXQsvyeeJ/w829wNruQEQIUKAG0HUt4JDV6LKCNdfX4Nuz30Kj1SCvU17MrD8TFXJVePsPRJlamrbMSgDt378/+vTpg9KlS6tQ6+joqMJqXI4cOYLatWuja9euqjW3SZMm6NKlS6KtuURERKQXHqHFwN/OqiBrY2WB77tWTjjIXtoA/NZZH2SLvQv0+MPoICvOhJ/BzNMzVZBtWKAh1rVexyBL5t0yGx4ejtOnT2PMmDFR6ywtLdG4cWMcPXo0zvdIa+zKlStVeK1WrRru3LmD7du3o0eP+AdnDgsLUw8Df39/9adGo1GPlGY4Rmoci1IG76H54z00f7yHphEWocWgteex75qPCrILulREw5I54v03yvLMMljuGAEL6KAt/R4i23wPWNjIDTH6/lWwrYA7TnfQpFATdC7RWX2ryn8bzYcmlfOMMcex0Em7fxp48uQJ8uXLp1pba9asGbV+5MiROHDgAI4fPx7n+7777jsMHz5cfV0RERGBTz75BAsXLoz3OJMmTcLkyZPjLFmQVmAiIqLMQKMFll63xJXXlrC20KFfSS1KZYsnAuh0KP5sK0o/1XfMupuzIS7k76mf4SuJ5N/p85rzKGdTDlYW+tm7tDotRyqgJAkODlbfxPv5+cHV1TXjjDO7f/9+TJs2DT/88IOqsb116xYGDRqEqVOnYvz48XG+R1p+pS43esusdByTEoXELo6pfrPYs2cP3n33XdjY2KT48cj0eA/NH++h+eM9fDsh4ZH4dPU5XHn9EvY2lljUrRJqF41nQgKdDpZ/TYLVv0E2svYw5K8/GvlluIMk8gvzw8RjE3Hw8UG4lHbBJ2U+Uf8WNm3SlP8WmilNKucZwzfpSZFmYTZnzpywsrLCs2fPYqyXZXd39zjfI4FVSgr69eunlsuVK4egoCB89NFHGDt2rCpTiM3Ozk49YpMbkZrhMrWPR6bHe2j+eA/NH++h8YLDI/DxqnM4euclHG2tsLR3VdQoEk+QjYwAtg4Gzq7ULzedBqua/4O+XTVpzj0/p4bd8g7yho2lDfK55Iv694/3z/zZpFKeMeYYadYBzNbWFl5eXti3b1/UOq1Wq5ajlx3EbnKOHVglEIs0qpYgIiJKtwLDItB76UkVZJ3trLGib7X4g2xEGLC+tz7ISjlB2++Bmv9L8rGkhODniz+j987eKsgWci2E1S1Xo5NnJ9N9IKL0VmYgX//36tULVapUUR26ZGguaWmV0Q1Ez549VV3t119/rZZbt26tRkCoVKlSVJmBtNbKekOoJSIiIsA/VIPeS0/gzIPXcLG3xvK+1VC5YLa4L01YALCmG3D3AGBlC7RfCpRqneTLKBMfjD00FoceH1LLzQs3x8SaE+Fk48RbQRk7zHbq1Ak+Pj6YMGECvL29UbFiRezcuRO5c+dWrz948CBGS+y4ceNU70f58/Hjx8iVK5cKsl999VUafgoiIqL0xS9Yg55Lj+P8Iz9kcbDBrx9WQ/n88QynFewLrGoPPD4N2DoDnVcBRRoYd7wwP5x+dhp2VnYYU20M3i/+vvr3mig1pHkHsAEDBqhHfB2+orO2tsbEiRPVg4iIiN70KigcPZYex6XH/sjmaIOV/aqjTN4scV8q/yfAr+8BPtcAh+xA9/VAPi+jL2vhLIUxve505HfJjxLZSvC2UOaazpaIiIhM42VgGLosPqaCbA4nW/z2UY34g+zL28DSpvog65IX6LMjyUH2RcgLfLLnE5zyPhW1rmHBhgyylDlbZomIiOjtPQ8IRbfFx3HzeSByudhhdb/qKJ7bJe6NvS8Cv74PBD0HshcBem4CshZM0nGOPT2G0QdH42XoSzwKfIRNbTfBypL9VijtMMwSERGZuWf+oapF9o5PENxd7bG6f3UUyeUc98YPjgGrOgJhfoB7OaD7RsDZLdFjRGojsejCIvx4/kfooEOxrMUwu/5sBllKcwyzREREZuzJ6xB0XXwM914GI19WBxVkC+WIZxSBm3uAtT2AiBCgYE2gyxrAIZ6OYdE8D36O0f+Mxknvk2pZOniNrjYaDtYOpv44REZjmCUiIjJTD32D0XXJMTz0DUH+bA74rX8NFMgez1TtF9cDf3wMaCOA4k2ADssB28SndZcxYztt7aSG35LwOqHmBLQq0sr0H4YomRhmiYiIzND9l0Houvg4Hr8OQaEcjirI5s0aT0vpyZ+BbcNkiiGgXAeg3ULAKmkzLOV2zI2q7lVxz++eKivwyOJh2g9C9JYYZomIiMzM3RdB6PLTMXj7h6JITies7l8D7lns39xQZsf85xvgr6n65ar9gOazgDimf4/dGuto4whXW1c1XuzkWpNhZWEFe+s4jkGUxjg0FxERkRm59TwAnX48qoJscTdnrPk4gSC7e9x/QbbeCKDF7ESD7MFHB9FhSwdMOjIpaqp4mcmLQZbSK7bMEhERmYnr3gHotuQYXgSGw9PdRU2IkNPZ7s0NIyOALYOAcyv1y02/Bmp+luC+NVoNvjvzHZZdXqaWHwU8QoAmQLXOEqVnDLNERERm4PITP3RfchyvgjUok9cVKz+sjmxOtm9uqAkFNnwIXNsKWFgBbRcAFbsmuO8ngU8w4uAIXPC5oJa7enbFsCrDYGsVx/6J0hmGWSIionTu4iM/dP/5OPxCNKiQPwtW9K2OLI5xdOAKCwDWdAXuHgSs7IAOvwCeLRPc974H+zD+8HgEhAfAxcYFU2pPQeNCjVPuwxCZGMMsERFROnb2wSv0XHoCAaERqFwwK5b1rQZX+ziCbLAvsPID4MkZwNYZ6PIbULhegvsOjQjF9BPTVZAtl7McZtabifwu+VPuwxClAIZZIiKidOrUPV/0/uUkAsMiUNUjG37pUw3OdnH80+33GPj1PeDFdcAhO9B9PZDPK9H9S6cuCbD77u/DoMqDYJPE4bqI0hOGWSIionTo+J2X6LPsJILDI1GjSHb83KsqnOIKsi9vAyvaAX4PAJe8QM8/gVwl493v7nu7Ea4Nj5r4oJJbJfUgMlcMs0REROnM4Vsv8OHykwjVaFGnWE4s7lkFDrZWb2749AKw8n0gyAfIXlQfZLMWjHOfYZFhmHVyFtZeXwt7K3uUzVGWEyBQhsAwS0RElI4cuOGDj1acQliEFg1K5sKi7l6wlxx79x8g8BngnBsoVAt4eAJY3QkI8wPcywHd/wCcc8W5z/v+9zHiwAhc9b2qlruW6op8LvlS+ZMRpQyGWSIionRi39Vn+HTlGYRHatG4lBu+71YZdje2ATtHAf5P/tvQMQcQGgBow4GCtYCuawD7LHHuc8fdHWoChOCIYGSzy4av6nyFuvnrpt6HIkphDLNERETpwK7L3hiw+gw0kTo0K+OO77pUgu2NrcC6njKdV8yNg1/q/8xTEei+AbB1fGN/MnvXl8e+xLob69RyZbfKqrNXbqfcqfJ5iFILwywREVEa237xKQb+dhYRWh1als+DuZ0qwsZCp2+RjR1ko5NaWes4ZgADYGFhgaz2WWEBC/Qv3x+fVvgU1pb8Z58yHv6tJiIiSkObzj3G0HXnEanVoV3FvJjdoQKsrSz1NbLRSwvi4v8YuH8EKPxf2UCwJhiONvqWWgmwdfPVRUW3iin9MYjSjGXaHZqIiChz23D6EYasPaeCbHuv/PimY0V9kBXS2Ssp/t1OQqzM5NVnVx+ER4arddISyyBLGR1bZomIiNLA2pMPMHrjReh0QJdqBfBVu3KwtLT4b4N4OnS9wTk3br26heEHhuO2321YWljipPdJ1M5XO8XOnSg9YZglIiJKYdLyeuKuL54HhMLNxR43nwdgwqbL6rWeNQthUusyMYPsnf3A1qGJ7NUCOte8+FPzHNO2DUVoZChyOeTCjHozUNW9asp+IKJ0hGGWiIgoBe289BSTt1zBU7/QN17rW7swxrcqpTprKSGvgd3jgLO/6pcdcgAhMnKBRayOYBYIsrDA1OKVse3oJLWmVt5amFZnGnLIe4gyEYZZIiKiFAyyMm5sfOMRVPXI9l+QvboV2DYMCPT+98X+QOOJwO2/3xxn1jUvJhevhB2+52FlYYUBlQagb9m+qsSAKLNhmCUiIkqh0gJpkY0vyEqEnbL1CpoUsoTVzpHAlT/1L+QoBrSZr5/lS5RuA3i21I9aEG0GsM+DnuDmXwMxvsZ4VM5dmfeQMi2GWSIiohQgNbJxlRYY6KBDrYDd0C3oA4T7ARZWQO1BQP1RgI19zI0trRCYrxIOPTmEZh76YbgKuBTAhjYb2BpLmR7DLBERUQp49Co43tfyW/hgmvUS1LO6CMgoWu7lgbYLgDwV4tz+yssrarSChwEP4WLjEjVSAcsKiNgyS0REZFIyjey2i08xfcfVN16zgBY9rfZgpPUaOFmEIUxnA+/KQ1Co1UjAyibOff127TfMPjUbGq0GeZzywNnWmXeMKBq2zBIREZnINW9/TNp8Gcfu+KplGW1L+2/RbFGLx5hhsxhVLG+o5RNaT3xj9z+sbt1Dv2Es/uH+mHh4IvY+2KuWGxRogC9rf4ksdkkcf5Yok2CYJSIiekt+wRp8u/cGfj12X3X8srO2xKcNiqJITicMW3MKH1ltxUDrjbCziECgzh4zIrpgVWQj/NCxCqziCLKXXlxSZQWPAx+rWbyGeQ1Dt1Ld/hv5gIiiMMwSERElkwTXdaceYtau6/AN0k8h27ysO8a2LIX82RyBJ2fRIPc0uPpdU6/9HVkBYzUfQpclP35oXRrNyuaJc793/O6oIJvPOR9m15+NsjnL8h4RxYNhloiIKBlO33+lSgouPvZTy8XdnDGpTRnULpYT0IQAeyYARxbAVRcJnUN23Ko8Dv45m+IbVwdUK5z9jRZZqY81tLy2KdoGIZoQNC/SHK62rrw/RAlgmCUiIjLCc/9QTN95DRvPPFbLLnbWGPxuCTUtrY2VJXDvELB5IOB7W/+Gsh/AotkMFHfOheLx7PPc83Oqk9f8hvORzT6bWtfJsxPvC1ESMMwSERElQXiEFsuO3MV3+24hMCxCretYJT9GNvNETmc7INQf2DEROLVU/waXPEDLOYBni3j3qdVpsezyMnx35jtE6iIx/+x8TKg5gfeDyAgMs0RERIk4cMMHk7dcxh2fILVcoUBWTG5TBhULZNVvcGMXsHUI4K9vrYVXb+DdKYB9/CMP+Ib6YuyhsTj0+JBabu7RHEO9hvJeEBmJYZaIiCgeD14GY+q2K9hz5Zlazulsq1pi21fOD0upeQ16AewcDVz8Xf+GbIWBNt8BhesleE1PeZ/CqIOj8DzkOeys7DC62mh8UPwDjlZAlAwMs0RERLGEhEfih/238OPBO6q8QDpr9a7lgUGNi8PV3kZ6awEX1wM7RgLBLwELS6Dm/4AGXwC2jglez30P9mHo/qGqxMDD1UONVlAye0neA6K0DLORkZGwsrIyxa6IiIjSfPauaduu4olfqFpXu1gOTGpdBsVzu+g38nsMbBsK3NipX3YrA7SdD+TzStIxqrlXQ16nvKjkVgnjaoyDo03C4ZeIUjDMXr16FZ06dcKVK1dQokQJrF27FuXKlXubXRIREaWL2bvyZXXA+Fal0LSMu/7rf60WOP0LsGciEB4AWNoA9UcCtQcD1rYJ7vu673WUyFZC7cfF1gW/tfxNzeTFSRCI3p7l27x5xIgRyJMnDzZv3owKFSpg0KBBJjglIiKi1J29S0Jsy+8OqSArs3cNblwc+4bVV5MaqMD58jawvLW+RVaCbP6qwCeH9GE2gSAbqY3ED+d+QIctHbD2+tqo9VntszLIEqWHltkzZ85g69atqFy5snp4enqa6ryIiIjSZPauL1qUQoHs/371HxkBHPse+HsaEBEKSElAo4lAtf6AZcLldT7BPhj9z2ic8D6hlm+9vsU7SpTewmxAQACyZtUPS5ItWza1TEREZG6zdxWT2btal0Gd4jn/28j7IrBpAPD0nH65yDtA67lANo9E93/kyRGM+WeMGn7LwdoB42uMR+uirVPs8xBlZkaHWSkpMNBqtdi3bx8uXboEjUZj6nMjIiIyqecBoZi+I4HZu0REGHBwFnDoW0AboR8rtunXQMWuwL/TzcYnQhuhygqWXFwCHXSqTnZW/VkokqUI7yRRegmz7dq1i7H88ccfRz1nITsREZnL7F0dvPSzd+VysftvwwfHgc0DgBc39MulWgMtZgMu7kk6zo1XN7D00lIVZDuU6ICRVUfC3to+RT4TESUzzEprLBERkbk4eMMHk6LP3pU/Cya3Lfvf7F0iLBDYNwU48ZMM0AU4uQEtZwOl2xp1rNI5SqtZvNwc3dCscDNTfxQiMkWYXbFihRqOy84u2m+yRERE5jZ7l8GtvcCWIYDfA/1yxe5Ak6mAY/ZEj6HRalRZQesirVEkq76UoGeZnin0iYjIJGG2T58+aNasGdzc3Ix9KxERUarM3rVw/y0sim/2LoNgX2DXWOD8av1y1oJAq7lAsUZJOs7TwKcYcXAEzvucx/6H+7Gu9TrYyNizRJS+w6zMjkJERGSWs3fpNwSubAK2DweCfKTHB1D9E6DhOMDOOUnH+vvB3xh3eBz8w/3hYuOCzyp+xiBLZE5Dc61btw6urq5xvtazJ79eISKi1HXdO0ANtXX0zsu4Z+8yCPAGtg0Drm3VL+csCbRdABSolqTjaCI1mHN6DlZeXamWy+Yoq0YryO+SPwU+FRGlWJidOXMmrKzeHCxafmAwzBIRUWrO3vXt3hv49dh9NQmCzN71Sf2i6uFgaxWzNfbsSn1ZQZgfYGkN1BkK1BsOWCetD4iMGfu/vf/DpZeX1HKP0j0wpPIQ2FixtIDI7MLsqVOnWDNLRERpRoLr76ceYma02bualXHH2JbRZu8y8L0LbBkE3D2gX85bCWizAHAva9QxXW1dYWtlq/78svaXeKfgOyb7PESURjOAERERpbYzD15h4qZEZu8S2kjg+I/AX1MBTTAg472+Mxao8RlglbR//sIjw2EBC9X6am1pjZn1ZiJSF4m8znlT4qMRUWqE2UKFCsVZYkBERJSS/MOBkRsv4Y+zT+Kfvcvg+VX9VLSPT+mXPeoCrecBOYom+XgP/B9g+IHh8MrthVHVRql1uZ1ym/ATEVGahNm7d++a5MBERERJIcNrLT18D9+es0JY5JP4Z+8SEeH6aWhlOlqtBrBz1Y8ZW6knYBkr8CZg592dmHR0EoI0QfAO8sZH5T9CNvtsvGFEGSHMDhw4EMWKFVN/RrdgwQLcunULc+fONeX5ERFRJp+9a/KWy7itZu+yQPl8rpjSrlzM2bsMHp3WT0X7/Ip+uURzoNUcwDXpJQGhEaGYcXIG1t9Yr5Yru1XGjHozGGSJMlKY3bBhAzZv3vzG+lq1amH69OkMs0REZPLZu7I72aCpeygm9awOOzvbmBuHBwN/fwUc+wHQaQHHnECLmUCZ92WYnSQf847fHVVWcPPVTVUn269cPzV+rNTKElH6ZfR/oS9fvkSWLFneWC/jzr548cJU50VERJlQXLN39arpgf/V98Chv/fEnIZW3DkAbBkIvLqnXy7XEWg2HXDKYdRxpaNX/9398Tz4ObLbZ8fXdb9Grby1TPjJiCjdhFkpMdi5cycGDBgQY/2OHTtQpIh+XmoiIiJjZ+/aftEbX227EufsXRqNJuYbQl4De8YDZ1bol13z6aeiLdEkWRdehtwaWXUk1l1fh+l1pyOXYy7eQKKMGmaHDh2qgqyPjw8aNmyo1u3btw/ffPMNSwyIiMgks3eNa1kKzcrGmr3L4No2YOtQINBbv1y1H9BoImAf98yU8bn16hZeh71GFfcqarmpR1M0KdQk7mMSUcYJs3379kVYWBi++uorTJ06Va3z8PDAwoULOfsXERElmV+IBt/uScLsXf+y0/jBauOHwNVN+hXZiwJt5gMetY1uBf7z1p+YdnwaHG0csb71+qiWWAZZIvOTrKr2Tz/9VD2kddbBwQHOzs6mPzMiIsqQtFod1iV19i6h08Hi4jo0vDoalpFBgIUVUHsgUH8UYONg1LGDNcH48tiX2HJni1qu5FYJlhZJH7KLiNKfZHfRlCB7/fp19dzT0xM5c8aaeYWIiCiO2bukpODCo0Rm7zJ4/RDYOhjWt/aqRV3ucrBouwDIW9Hoa3vd97oareCe/z0VYAdUHIAPy33IMEuU2cJsUFAQPv/8c6xYsQJarVatkxnBevbsifnz58PRMY7fqomIKFN7HhCKGTuuY8OZR1Gzdw1qXBy9anm8OXuXkH9fTv0M7J0EhAdCZ2WHq25tULzXd7CxdzS6rGD9zfWYcWIGwiLD4ObopqallZm9iMj8WSanA9iBAwewZcsWvH79Wj02bdqk1g0bNixlzpKIiMySDK+1+OAdNJx9ICrIyuxdfw1vgH51i8QdZF/cBJa1ALYPV0EWBWogot9+3HRvDVjZGH0OUgd77vk5FWTr5KujamQZZIky+aQJ69evR4MGDaLWtWjRQtXOduzYUXUEIyIiijl7F1AhfxZMalMGlQrGMy1spAY48h2wfwYQGQbYOgONJwFVPgQiIwHcNLpF1tCha2z1saiQqwLal2jPsgKizB5mg4ODkTt37jfWu7m5qdeIiChze+gbjKlbr2D3v7N35XCyxahmnmjvlf/NSQ8MnpzTT0XrfVG/XKyxftzYrAX0yyrMJj3Errm+BieensA3Db5R4VVGLehYsuPbfzgiMv8wW7NmTUycOFHVzNrb26t1ISEhmDx5snqNiIgyp/hm75La2CwO8ZQHaEKAAzOAw98BukjAIRvQbAZQvqNRU9Ea+If7Y9KRSdhzf49a3vdgH94t9O7bfjQiykhhdt68eWjatCny58+PChUqqHXnz59XwXbXrl0pcY5ERJSOSUvojksye9dVPH4d8sbsXfG6fwTY/Dnw8pZ+ucz7QPOZgHPyZt+69OKSGq3gceBjWFtaY6jXUDQu2DhZ+yKiDBxmy5Yti5s3b2LVqlW4du2aWtelSxd069ZN1c0SEVHmceOZfvauI7eTOHuXCPXXj1IgoxUIZ3eg1RzAs2Wyw/TKqysx5/QcRGgjkM85H2bXn42yOcsm+3MRUQYfZ1aG3+rfv7/pz4aIiDLk7F1RbuxW48bC/7F+uXIv4N0pgEPWZJ/L1ye+xm/XflPPpSV2cu3JcLU1bmpbIspEYXbz5s0Jvt6mTZu3OR8iIspIs3cZBL0Edo4GLq7TL2fzAFp/BxSp/9bn1KZoG2y6tQmDvQajc8nOnJKWKJMxOsy2a9cu6rl8hSRf70RfjjSixykREZn37F0TW5dG3eIJ1LjKvxGXNgA7RgLBLwGZOrbGZ8A7YwHb5E2yo9VpcePVDXhm91TLUk6wu/1uZLHLkrwPRkSZa9IEmfXL8JByg1u3bkUtJyfIfv/99/Dw8FAdyKpXr44TJ04kuL1M0vC///0PefLkgZ2dHUqUKIHt27cbfVwiIkr67F3D1p3H+z8cUUFWZu+Sutgdg+omHGT9nwC/dQE2fKgPsm6lgQ/3Ak2/SnaQfRX6CgP2DUC3bd1wzVffb0MwyBJlXsmqmTWVtWvXqhnFFi1apILs3Llz1UgJ169fV+PWxhYeHo53331XvSYTN+TLlw/3799H1qzJr7UiIqK4aSK1WH7kHubuvYnAsIio2btGNCsJNxf90IxxkqlozywH9kwAwvwBSxug3gigzhDA2jbZl/texD18t+M7PA95DltLW9zzuxfVOktEmVeahtk5c+aojmR9+vRRyxJqt23bhqVLl2L06NFvbC/rfX19ceTIEdjY6McslFZdIiIyrX9u+qiSgiTP3mXw8jawZRBw7x/9cr4qQNsFgFupZJ+LlBX8fPlnLA1cCi208HD1UKMVlMxeMtn7JKJMHGb9/f1j1MgGBgbGWOfqmrQepNLKevr0aYwZMyZqnaWlJRo3boyjR4/G2/lMJmaQMoNNmzYhV65c6Nq1K0aNGgUrq7h7z4aFhalH7PPXaDTqkdIMx0iNY1HK4D00f7yHSffwVTC+3nEDe64+V8vZnWww/N0S+KBSXjV7V7w/y7QRsDyxCJYHpsMiIhQ6G0doG3wBbZX+gKWV3IRk3TvfUF+MOzIOx7yPqeVmBZthXPVxakYv/lw1H/xv0PxpUjnPGHMcC130HlxJIIHTMHZg9HmvDc+TWjf75MkTVSYgrazRZw4bOXIkDhw4gOPHj7/xHk9PT9y7d0+NafvZZ5+pel35c+DAgWpWsrhMmjRJzU4W2+rVq1XNLxERAeGRwN4nlvjrsQU0OgtYQoe6eXRoll8Lx0SaPVxDHqDig5+RLfiuWn7uUgbnC/RFsF3yJj+I7nDoYewI3QEb2KCVQytUtq3M0QqIMoHg4GDVYOnn55doQ6nRLbN///030op0MpN62Z9++km1xHp5eeHx48eYNWtWvGFWWn6lLjd6y2yBAgXQpEmTJLciv+1vFnv27FG1vobSCDIvvIfmj/cwftIQsfPyM3y78wae+IWqdTWLZMf4Fp4onts54QsbEQbLw3NgeX4eLLQR0NlnQWTjqchWvgsaJGMq2rg00zWD8ylnvF/4fdw+cZs/S80U/xs0f5pUzjPRv/VPjNFhtnDhwioMxjuzSxLlzJlTBdJnz57FWC/L7u7ucb5HRjCQCxi9pKBUqVLw9vZWZQu2tm92LJARD+QRm+wnNcNlah+PTI/30PzxHppg9i6DhyeATQOAF9f1y56tYNHyG1i7xP3zO6l8gn2w6PwijKg6AvbW+k5mE2pNUP+Q3sZt3kMzx/8GzZ9NKuUZY45hmZww6+Pjg7clwVNaVvft2xej5VWWo5cdRFe7du2oocAMbty4oUJuXEGWiIjinr1r8pbLaD7vHxVkba0tMbBRcewdWh/Ny+VJOMiGBQI7RgM/N9EHWSc3oOMKoPMq4C2D7JEnR9B+S3usu7FOTU1LRJQiLbNGltgmSL7+79WrF6pUqYJq1aqpobmCgoKiRjfo2bOnqqv9+uuv1fKnn36KBQsWYNCgQfj8889x8+ZNTJs2TdXMEhFR4rN3/X76IWbuvI6X/87e1bRMboxrWTrh2bsMbu0DtgwG/B7olyt01Y8Z65j9rS59hDYCP5z7AUsuLoEOOhTPVhydPTvzdhJRyg3N9ejRI4SG6murYitYsGCS99OpUyfVyjthwgRVKlCxYkXs3LkTuXPnVq8/ePBAdTgzkPKGXbt2YciQIShfvrwKuhJsZTQDIiKK39kHrzDR2Nm7DIJ9gd3jgHOr9MtZCgKtvwWKNX7rS/4s6BlGHhyJM8/PqOX2JdpjVNVRUSUGREQpEmarVq36xjpjRzMwGDBggHrEZf/+/W+skxKEY8f0Q7QQEVHCfALCMGPnNaw//Ugty+xdgxoXR69aHrCxSkKl2ZVNwLbhQJAM1WUBVP8YaDgesEukc1gSnH1+FoP+GoRXYa/gaO2ISbUmoXnh5rylRJTyYVaGzZIxXomIKH3P3jVv700E/Dt7V3uv/BiZ2OxdBgHewPbhwNUt+uWcJYA2C4CC1U12jnmc8qhJEEplL4VZ9WehkGshk+2biDIPo8OstL5KKUFc080SEVH6mL1r8pYruPU80LjZu4T0izi3Gtg1Bgj1Ayyt9dPQ1h0O2Lz9V/8B4QFwsXVRz92d3PFzk5/hkcUDdlZvjjpDRJTuO4AREZHpPPQNxpfbrmDXZf2QhzmcbFVLbAevAmr2rkS9uqfv4HXn3/HE81TUT0XrXs4k57f/4X6MOzwOU2tNxTsF31HrOCUtEaV6mL179y5LDIiI0pGQ8EgsPHAbPx64jbAILawsLdCzZiEMblwCWRySMFajNhI48ROwbwqgCQak89U7XwA1/gdYJasaLQZNpAZzz8zFiisr1PLa62ujwiwR0dsy+qfU+fPn1SM+bdq0edtzIiKiJH5TtuOSN77adhWPX4eodTWL5FAlBSXd9V/lJ+r5NWDzAODRSf1yoTpAm++AHEVNcg8eBTxSoxVcfHFRLXcv1R1Dvf6blZGIKNXDbLt27d6ooTWUHiRnNAMiIjLN7F1jW5ZC86TM3iUiwoHDc4GDs4DIcEDqWJtMASr3BqINifg29t7fiwmHJyBAo6+T/bL2l2hYsKFJ9k1EZJCs74+ePn0aNRasi4uLaqktUqRIcnZFRERGzt41d+8NrDh6H5FanZq965P6RfFp/aJwsP1vqu8EPT4NbPoceH5Zv1yiGdByDpAln8nuxdWXVzFk/xD1vHyu8phVbxbyOuc12f6JiN5qNIPoncBkatnbt28zzBIRpefZu0R4MLB/GnD0e0CnBRxzAM1nAmU/kB/uJj3fUjlKoVPJTmr82M8rfw4by5Sfy52IMiejw2yePHlw5swZtGjRAtevX0dYWBg6duyIGTNm4KOPPkqZsyQiyuSzd0lJwfl/Z+8qmstJ1cUmafYug7sHgc0DgVd39cvlOgLNpgNOOUx2nrvv7Ubl3JWR0yGnWh5bfWzSSh6IiFIzzEpw7dy5M2rVqoXTp0+jQ4cOGDFiBNq3b48jR45g2bJlvCFERCkwe5eznTUGGzN7l5CxYvdMAE7/+7PZNR/Q6lugRFOT3aPQiFDMPDkTv9/4HdXdq+PHd3+ElaUVgywRpc8wO3v2bHh6eqo62caNG+Ozzz6Do6MjTp06hR49eqTMWRIRZUBS83riri+eB4SqWbmqFc6uhtUyevYuGVrr/hEg8BngnBsoVAuwtAKubQe2DQUCnuq3q/Ih0HgSYO9qss9w1+8uhh8YjhuvbsACFqo+VgeOR05E6TjMWlpaxllOkD17dmzbts1U50VElKHtvPRUzdL11C80al2eLPboUCU/tl/0jpq9q/y/s3dVjm/2riubgZ2jAP8n/61zdgeyFQIeHtcvZy+qH27Lo45JP8PWO1sx5egUhESEILt9dnxd52vUylfLpMcgIkpMskfD9vHxUTWzomTJkpxIgYjIiCD76cozb7RfSrD9bt+tpM/eJUF2XU8ZcTbm+kBv/QOWQO2BQIPRgI2Dye6PhNfpJ6Zj482Narmqe1VMrzsdbo6c5pyIzCDMBgUF4fPPP8eKFSvUSAbCysoKPXv2xPz581XJARERxV9aIC2yCX0R72RrhT1D6yO7k238G0lpgbTIJrQnp5xAown6kgMTkhFtzj4/q8oKPqnwCT4u/7GqkSUiSgtGj4w9dOhQHDhwAFu2bMHr16/VY9OmTWrdsGHDUuYsiYgyCKmRjV5aEJeg8Ehc9w5IeEdSIxu9tCDOHT3Xb2cihmEZHW0cMbv+bPzU5Cd8VvEzBlkiMq+W2Q0bNmD9+vVo0KBB1DoZpsvBwUGNdLBw4UJTnyMRUYYhnb1Msp109kqKpG6XgGBNML46/hVKZCuBXmV6qXXynIjILMNscHBw1Oxf0bm5uanXiIgofnGORpCc7WTUgqRI6nbxkFEKZLQCGbXAzsoOLYu0jBpHlojILMsMatasiYkTJyI09L9Wg5CQEEyePFm9RkRE8ZPht3I4x18La/HvqAayXYJk+C27hIbYstCPKSvbJbOkYP2N9ei6rasKsm4ObljUeBGDLBGZf8vsvHnz0LRpU+TPnx8VKlRQ62TMWXt7e+zatSslzpGIKMMICo9QgTUuhvUTW5dW480m6NFJICww4T3JDF/J6JgVGB6ohtzacW+HWq6drzam1Zmmht8iIjL7MFu2bFncvHkTq1atwrVr19S6Ll26oFu3bqpuloiI4m/tHLPxIl4Ehquht6ytLPDMPyzqdfcs9irINiubJ+FLGPQC+L2PDGkAFKgJ+N2P2RnMNa8+yJZuY/St0Gg16L69O2773YaVhRUGVh6I3mV6w9LC6C/yiIjS7zizMvxW//79TX82REQZ2OoTD7DtwlNYW1pgSa8qKJ8/a5wzgCVIhkTc+BEQ8ATIWQLovl4/hmxcM4Alg42lDd4r/h5WXl2JWfVmoaJbxeR9WCKi9BxmZbIEGVP26tWrarlUqVIYMGCAmuaWiIjedPWpvxpfVoxq5olK/87oVbNoDuMu1z/fALf3AdYOQIflgJ2zfn3husm+7AHhAfAN9UUh10JquWfpnni/+PtwsXXhrSSidM8yOUNzSanB6dOnVc2sPM6cOYNy5cqp14iIKKagsAgMWH0G4RFavFMyFz6sUzh5l+jOAWD/NP3zVnOA3KXf+lJffnEZHbZ0wIB9AxCkCVLrLCwsGGSJKOO2zI4cORJjxozBlClTYqyXEQ7ktQ8++MCU50dEZPYmbLqM2z5ByO1qh286Vox/etqEBHgDG/oBOi1QqTtQsetb1++uuroK35z+BhHaCORzzodnwc9QJEuRt9ovEVG6b5l9+vSpmro2tu7du6vXiIjoP+tPP8KGM48g+fW7zpUSnqI2PpERwPoP9TN65S4LtJj9VpfYL8wPg/8ejBknZ6gg26hgI6xrvY5BlogyR8uszPz1zz//oFixYjHWHzp0CHXrJr9mi4goo7n1PBDj/7ykng9pXALVixhZH2sgpQX3DwG2zvo6WenwlUznfc5j5IGReBL0RHX2Gl5lOLp4dlGlBUREmSLMtmnTBqNGjVI1szVq1FDrjh07ht9//11NnLB58+YY2xIRZUahmkhVJxuiiUTtYjnw2TsxGwCS7OYefacv0eY7IGcy9/OvRecXqSBbwKUAZtWfhTI5yrzV/oiIzC7MfvbZZ+rPH374QT3iek3Ib/mRkZGmOEciIrMzdesVXPMOQE5nW3zbqWLiQ27Fxe8RsPHfYRCr9gfKvn2fhKm1p2LhuYUY4jUEztLSS0SU2WpmtVptkh4MskSUWW298ASrjj+AfHMvQVbGkDVaRDjwe28g5BWQtxLQ9KtkncuZZ2ew4OyCqOWcDjkxvuZ4BlkiytzjzBqEhoaqaWyJiEjv/ssgjNlwUT3/rEFR1C2eK3mXZu8k/ZS19lmADssAazuj3q7VabH00lIVZCN1kSiVo5Tq6EVEhMzeMistrlOnTkW+fPng7OyMO3fuqPXjx4/Hzz//nBLnSERkFsIipE72LALCIlClUDbV6StZrm4Bjn2vf95uEZDNw6i3vwx5iU/3fop5Z+apINuqSCvUzFMzeedCRJTRwuxXX32FZcuWYebMmbC1/W+IGZlIYcmSJaY+PyIiszFjx3VcfOyHrI42+K5LJVhbGf0jFvC9A/z5P/3zWp8Dni2MevtJ75NqEoQjT47A3soeU2pNwbQ60+Bo42j8uRARmQGjf9KuWLECP/30E7p16wYrq//m/paZwK5du2bq8yMiMgt7rjzD0sN31fPZ7Ssgb9ZkDJ+lCdXXyYb5AQVqAI0mGvX2FZdXoN/ufvAJ8VFjxv7W8je8V/w9DrtFRBma0TWzjx8/fmOMWSGdvjQajanOi4jIbDx+HYLhv59Xz2Wq2salcydvR7vGAE/PA445gPZLASsbo95e0LWgqpVtW7Qtvqj+BVtjiShTMDrMli5dWk2aUKhQoRjr169fj0qVKpny3IiI0j1NpBYDfzsLvxANyufPglHNPJO3owu/A6eWysCGwPuLgSz5kvQ2/3B/uNq6qucNCjTAmpZrUCYnx44loszD6DA7YcIE9OrVS7XQSmvsxo0bcf36dVV+sHXr1pQ5SyKidOrbPTdw+v4ruNhZY0GXyrC1TkadrM8NYMsg/fN6I4BiiY86INPQLjy/EL9f/x1rW61FHuc8aj2DLBFlNkb/1G3bti22bNmCvXv3wsnJSYXbq1evqnXvvvtuypwlEVE6dPCGD37Yf1s9n/5BeRTMkYxOVuHBwLqegCYIKFwPaDA60bc8C3qmamN/uvATXoW9wu77u5Nz+kREmXec2bp162LPnj2mPxsiIjPx3D8UQ9edU8+7VS+IluX1LaNG2z4c8LkKOOcGPvgZsPyvY21cDj0+hC/++UKFWEdrR0yqNQnNCzdP3rGJiDJrmA0ODlYjGdjZ2eHBgweqldbT0xO1atUy/RkSEaUzkVodBq89hxeB4fB0d8H4VqWTt6MzvwLnVgEWlvoOX85u8W6q0Wrw/dnv8fMl/Xjentk9Mbv+bBRyjdl/gYgoszG6zGDlypXIkiUL3N3dsXnzZpQrVw6jRo1CvXr1sGjRopQ5SyKidOT7v2/hyO2XcLS1woKulWFvk3Brapy8L+lbZUXDcYBHnQQ3X3VlVVSQ7VyyM1a2WMkgS0SU3EkThgwZomplu3fvjsmTJ8PHx0cF2W+//ZYXlYgytGN3XmLu3hvq+ZftyqKYm7PxOwn119fJRoQCxZsAtYck+pbOnp1R2a0yvqn/DcbWGAs7K+OmtyUiyqiMDrMyfe3nn3+OwYMHIywsDM2b62u15M979+6lxDkSEaULLwPDMGjNWWh1QHuv/Hi/cn7jd6LT6Ucu8L0NuOYH3vsRsHzzR7EmUoN119chUhuplu2t7bGs2TI08Whiio9CRJR5a2ZlYgR7e3s1o4xMZ2uY0tba2hoREREpcY5ERGlOq9Vh2O/n8cw/DEVzOWFK22SO5XpyCXB5I2BpDXRYBjhmf2OTx4GPMeLACFx8cRG+ob74pMInar383CUiIhN0AOvdu7fq/BUaGopPPvlEDdElrbRERBnVkkN3sP+6D+ysLVWdrKNtMn58Pj4N7Byjf/7uVKBA1Tc22Xd/H8YfGY+A8AC42LqgRLYSJjh7IqKMy+ifxjJhgoHUzEbXs2dP05wVEVE6cubBK8zceV09n9i6DErl0c+4ZZSQV8DvvQGtBijVGqjxaYyXwyPD8c2pb7D62mq1XD5XecyqNwt5nfOa5kMQEWVQRofZX375JWXOhIgoHfIL1uDz1WcRodWhVfk86FKtQPLqZP/8DHj9AMjmAbT9XmoGol5+6P8Qww8Ox5WXV9Ry7zK9MbDyQNhY2pjyoxARZUjJKjMgIsoMdDodRm44j8evQ1AohyO+fr9c8upWj8wHrm8HZASCDssB+ywxXg6OCMatV7eQxS4LptWZhnr565nuQxARZXAMs0RE8fj12H3suvwMNlYWmN+lElzsk9FS+uAYsHeS/nnz6UDeilFB2RCMS2YviVn1Z6F0jtJwd3Ln/SAiSsmhuYiIMoNLj/3w5dar6vmY5qVQPn9W43cS9AL4vQ+giwTKdQC8+qjV9/zuoeu2rrjoczFq04YFGzLIEhElA8MsEVEsgWERGLD6DMIjtWhcKjf61PYw/hpptcDG/kDAEyBnCaDVXFUnu+3ONnTa2gmXXl7C1ye+Vi20RESUfCwzICKKRsLlFxsv4t7LYOTNYo/ZHconr072n9nA7b8Aaweg4wqEWFlh+pGJ2Hhzo3q5qntVzKg7g2PHEhGldpgdOnRogq/PmTPnbc6HiChNrTv1EJvPP4GVpQXmd62ErI76iWGMcmc/8Pc0/fNWc3DH1g7DtnXFrde3YAELNQnCx+U/hpWllcnPn4goszE6zM6dOxc1a9aMmvnr0KFD8PLygoODA1sYiMis3XgWgImbL6vnw5qUgFehN2fnSlSAN7Chn7TxApV64Fahaui6rTNCIkKQwz4HZtSbgep5qpv+5ImIMqlklRn88ccfcHNzU89dXFywevVqFClSxNTnRkSUakLCI/G/VWcQqtGibvGc+KReUeN3EhkBrP8QCPIBcpcFWsxCUWt7VHOvhtDIUEyvOx05HXKmxOkTEWVaRncAs7GxQXh4eNSyRqPBhg0bTH1eRESpatLmy7j5PBC5XOzwbaeKsLRMRp3s318B9w/hlmMWBL+3CLDRf2M1s95M/Nj4RwZZIqL0EGYLFy6MNWvWqOcSYiXcLl68GF26dEFwcHBKnCMRUYr68+xjrD31UE3KNa9zReR0tjN+Jzd2Q3doDjY4O6Gzew5Mvbk6aqQCRxtH1scSEaWXMDtq1CiMHj0a9vb26Nixo3p+6tQpFWSrVauWMmdJRJRC7vgEYuwf+vFeBzYsjlpFk1EG8Pohgv78CKNz5cCkXDkQpovAq7BXCNf+9y0WERGlk5rZPn36oFatWrhw4YJqpa1SpYpav2nTJkyfPj0lzpGIKEWEaiIxYPVZBIVHonrh7BjYqLjxO4kIx7X13TE8uz3u29jAysIKAysPRO8yvWFpwaG8iYjSZQewkiVLqkds0kpLRGQuvt5+FVee+iO7ky3mda6khuMyhpQRrNvcCzOtXiLc0gbuDrkwq8EcVHTTT1lLRETpfNKE0NDQGJ3BhKur69ueExFRitt56SmWH72vnn/TsQLcs9gbvQ//i2vxw+vzCLeyQoOspTC16U/Iap+MaW+JiCj1wqzUxo4cORLr1q3Dy5cv33g9MjIy+WdDRJQKHvoGY8T6C+r5x/WL4J2S+qEGjeJ7B1m2jcB0izDcLNEIPdqs5ljbRERpwOiCrhEjRuCvv/7CwoULYWdnhyVLlmDy5MnImzcvVqxYkTJnSURkIuERWgz47SwCQiNQqWBWDG/yZslUQmUFq66uwu7b24B1vYAwf9R0q4yebVcwyBIRmUvL7JYtW1RobdCggeoMVrduXRQrVgyFChXCqlWr0K1bt5Q5UyIiE5i9+zrOP3wNV3trzO9SCTZWSfud3i/MDxMOT8BfD/+CE6xQwecBcjvmANovBaxseG+IiMylZdbX1zdqti+pj5VlUadOHRw8eND0Z0hEZCL7b/jgp4N31POZ7SsgfzbHJL3vgs8FdNzSUQVZGxmt4IUP3CK1wPuLAde8vD9EROYUZiXI3r17Vz339PRUtbOGFtusWdnxgYjSp9dhwMgNl9Tz3rU80Kyse6Lv0eq0WH55OXrt6IUnQU9QwNEdv3q/RNeAQFjUHwkUa5QKZ05ERCYNs1JacP78+aihuL7//ns1gcKQIUNUPS0RUXoTEanFiptWeBWsQdl8rhjTwjPx92gjMOivQZh9ajYidBFoWrAR1j33R5lgf6BwfaD+qFQ5dyIiMnHNrIRWg8aNG+PatWs4ffq0qpstX768sbsjIkpxC/bfwe0ACzjZWWFBl8qws7ZK9D3WltYo4FoAtpa2GFVtFDpc3gcLn2uAszvwwRLAMvF9EBFROh9nVkjHL3kQEaVHh2+9wA8H9HWyU9uUhkdOpwTLCgI1gXC11Y+XPaTyEHxQ/AMUvXMYuLAGkBm9pMOXczKG8iIiovQRZr/77rsEXx84cODbnA8Rkcn4BIRh8Npz0OmAmm5atC6fJ95tfUN98cWhLxCsCcbPTX+GjaUNbKxsUDQ0GNj+bwlVw/GAR23eISIicw6z3377bdTzhw8fIk+ePLC21u/GwsKCYZaI0gWtVoeh686pQFvczQnve/jFu+1J75MYfXA0noc8h72VPa69vIZyucoBof768WQjQoHiTYDag1P1MxARUQqEWcNIBsLFxQUHDhyIGqqLiCi9WHjgNv65+QL2NpaY16kCbp56c+jASG0kFl9cjIXnF6oSgyJZimB2/dkonq24zJAAbP4c8L0NZCkAvPcjYGl0n1kiIkrvNbNEROnNyXu+mLPnhno+pW1ZFHdzxs1Y27wIeYHR/4zG8afH1XLbom3xRfUv4Gjz79izJxYDV/4ELG2ADssAx+yp/TGIiCgJGGaJKEN5FRSOgb+dRaRWh3YV86KDV35ERES8sd0X/3yhgqyDtQPG1RiHNkXb/Pfi49PAri/0z5tMBfJXScVPQERExjD6O7MLFy5EPWSechmaK/q65JCxaj08PNR4tdWrV8eJEyeS9L41a9aoOt127dol67hElLHIz6QR68/jqV8oCud0wpfvlVM/I+IyuvpoVMhVAWtarYkZZENeAet6A1oNUKoNUP2T1PsARESU8i2zFStWVP84yD8aolWrVlHL8mdkZKRR+1u7di2GDh2KRYsWqSA7d+5cNG3aFNevX4ebW/zD39y7dw/Dhw9H3bp1jf0IRJRBLT18D3uvPoettSUWdK0EZ7v/fsT5a/2x494OtCmuD65SH/tr819jhl2tFvjjU8DvAZCtMNB2gfRsTYuPQkREqdEBzBTmzJmD/v37q5nFhITabdu2YenSpWqGsbhIYO7WrRsmT56Mf/75B69fvzbpORGR+Tn/8DWm77iqno9vWQpl8maJeu3IkyNYELAAoUdDkdclL6q468sG3mi1PTofuLEDsLIDOi4H7P/bBxERZZAwa8oJEsLDw9XsYWPGjIlaZ2lpqWYWO3r0aLzvmzJlimq1/fDDD1WYTUhYWJh6GPj7+6s/NRqNeqQ0wzFS41iUMngP07+AUA3+t/oMNJE6NC3thk5eedV9kylpF15YiF+u/KK2K5G1BLLaZI3zv0eLh8dgtXcyJN5GNJkGXc7ScvPT4NNQXPjfoXnj/TN/mlTOM8Yc5606gAUEBGDs2LE4d+4cypUrh2nTpiFLlqS3ZLx48UK1subOnTvGelmWWty4HDp0CD///LM6ZlJ8/fXXqgU3tt27d8PR8d9ey6lgz549qXYsShm8h+mTVDwtu2mJR68skd1OhwZOT7BjxxO81r7GuqB1eBD5QG1X3bY6muma4fKhy5D/RWer8UeD6+NhrYvEw2y1cOZpTmD79jT6RJQQ/ndo3nj/zN+eVMozwcHBqRNmhw0bpj5Uhw4dsGPHDnz++edYsWIFUoqE5x49emDx4sXImTNnkt4jrb5Skxu9ZbZAgQJo0qQJXF31U1am9G8Wco3effdd2NjYpPjxyPR4D9O3304+xLljV2FtaYEfe1VDxQJZ8c/jfzDr6Cz4RfrB2cYZX3h9Ae01bdz/HWojYbWmEyw1r6DLWQLufVajha1zWn0cigf/OzRvvH/mT5PKecbwTXqKh9m9e/eqVtKGDRuib9++qF+/vlHvl0BqZWWFZ8+exVgvy+7u7m9sf/v2bdXxq3Xr1lHrtNJhQz6ItbXqNFa0aNEY77Gzs1OP2ORGpGa4TO3jkenxHqY/V5/648vt19XzUc08UbVILvXcJ9QHfuF+KJ2jNGbXmw13B3dsv7Y97nu4fw5wdz9g4wiLjr/CxilbWnwUSiL+d2jeeP/Mn00q5RljjvFWYVbKBGRILVG4cGG1bAxbW1t4eXlh3759UcNrSTiV5QEDBryxvaenJy5evBhj3bhx41SL7bx581SLKxFlDkFhEapONjxCi4aebuhbW/+zSHQs2RF21nZoUbgFbK1s46+9urMf2P+1/nmrbwE3z1Q6eyIiMhXrt232DQwMVOtCQ0OTdQJSAtCrVy9UqVIF1apVU0NzBQUFRY1u0LNnT+TLl0/Vvso4tGXLlo3x/qxZs6o/Y68nooxt/KZLuOMTBHdXe7St5Ysu2ztjSdMlcLV11Y8/XSyR8af9nwIb+knVLVC5J1Chc2qdOhERpWWYlfBoGM5GxpatVKlS1PP4BidPSKdOneDj44MJEybA29tbjWO7c+fOqE5hDx48UCMcEBEZrD/9CBvPPIalZQRqVv0H445uUOuXX16Ozyt9nviFiowANnwIBPkAucsBzWfy4hIRZZYw+/fff5v8JKSkIK6yArF///4E37ts2TKTnw8RpV+3ngdi/J+XYGHzEoVKbcTeJ7fV+t5leuOTCkmcrevvL4H7hwFbF/14sjYOKXvSRESUfsKssZ28iIhMJVQTiQEynqz9Wbjm+wMvNCHIapcVX9X5CvXy10vaTm7sAg59q3/edj6QI2anUSIiyuBhdvPmzQm+3qZNtDnOiYhMaMrWK7gduhcO+f+AjGNSya0SZtabCXenN0c/iZPfQ2DjR/rn1T4GyrzH+0NElNnCrGHUgbhIzaxMgkBEZGpbzj/B6uMPYGFdFtlsD+ODkm3wv4r/g7Vl0n6MWWgjYLWxHxD6GsjnBTT5kjeJiCgDSNbQXE+fPn1j1i4iopSy89YxjNnop55/Vrc8Pmu4GU42Tkbto8yTNbD0OQ3YZwXa/wJY26bQ2RIRUboPsxxdgIhSQ2hEKL46Ng1/3v4DofbtUcX9XQxpXALWVsaNcGJxbQuK+uzWL7z3I5CtUMqcMBERmUeYlelks2XLBicnJ+TNm1cNp5XU6WWJiJLizus7GHZgGG69vgWdzgKODsH4rkslo4MsXt6G1daB6mlkzc9hVbIZbwARUWYOswULFsRPP/2kZtSRyRJkggNpqW3ZsiV+/fVXuLq6psyZElGmsfn2Znx57EuERIRAG+GM0Med8WP7zsib1cghtDQhwO+9YBEWgJdOJeBa/wtYpdRJExFRmjB6NoJ79+6px+PHj9U0sn5+fti1axdu3LiBkSNHpsxZElGmEKwJxrhD4zD20FgVZBFSHMF3BqF35XfRuHQy6vR3jga8L0LnmBOnPD4DrFJ+PnEiIkpdbz21louLCxo1aoQffvgBu3f/W5NGRJQMl19eVq2ylhaWyBHeBgH3+qBcnvwY1czT+J2dXwuclklVLBDZdhFCbbPznhARZUAmmyf2nXfewZ07d0y1OyLKhKq6V8WwKsPQKOsE3LtdCy52tljQpTJsrY38UfX8GrB1sP55/VHQFWmQIudLRERm2gFMBAcH48GDBwgPD4+xvnz58qY4LyLKBII0QZh9ajb6lumLAq4F1DoPmxaYeOSEej79g/IomMPRuJ2GB6k6WWiCAQmx9UcCkTLFAhERZURGh1kfHx/06dMHO3bsiPN1TppARElx3fc6hh8Yjnv+93Dj1Q2sbL4SPgFhGLr2nHq9e42CaFk+j3EXU6cDtg4FfK4Bzu7A+0sASyuGWSKiDMzoMoPBgwfj9evXOH78OBwcHLBz504sX74cxYsXT3SqWyIinU6HddfXoeu2rirI5nbMjeFVhkOrAwatOYeXQeHwdHfBuJaljb9YZ1YAF9YAFlZA+6WAcy5ecCKiDM7oltm//voLmzZtQpUqVdSQXIUKFcK7776rhuT6+uuv1RBdRERxCQgPwOSjk7Hr3i61XD9/fXxZ+0tktc+KeXtv4uidl3C0tcKCrpVhb2PkIFpPLwDbR+ifNxoPeNTmTSAiygSMDrMyrqybm5t6LhMnSNlBiRIlUK5cOZw5cyYlzpGIMoBHAY/w0Z6P8DDgIawtrDHYazB6lu4JCwsLHLvzEvP23VDbfdmuLIq5ORu381B/fZ1sZBhQohlQa1DKfAgiIjL/MoOSJUvi+vXr6nmFChXw448/qjFnFy1ahDx5jKxvI6JMQ8oJXG1dkdcpL5Y3X45eZXqpIPsyMAyD1pxVZQbtvfLj/cr5ja+T3TwA8L0DZCkAtFsoc26n1McgIiJzb5kdNGgQnj59qp5PnDgRzZo1w6pVq2Bra4tly2RMRyIiPf9wfzhaO8La0ho2VjaY+85cOFg7IItdFvW6VqvDsN/P45l/GIrmcsKUtmWMv3QnfgKubAIsbYAOywBHjidLRJSZGB1mu3fvHvXcy8sL9+/fx7Vr19Q0tzlz5jT1+RGRmbrocxEjDo5Ai8ItMLDyQLXO3ck9xjaL/7mD/dd9YGdtie+7VYajrZE/kh6dBnaN1T9v8iWQv4rJzp+IiMxDsr+Lk/FlpdxAWmQrV67MIEtEUaMVLL+8HD139MTjwMeqs5dMUxvb6fuvMGuXvmRpYusy8HR3Ne4KBvsCv/cGtBqgVBug+se8A0REmZBlciZL+PDDD+Ho6IgyZcqoiRPE559/junTp6fEORKRmfAL88PAvwaqiRAidBFoUqgJ1rRaA0ebmBMf+AVrMPC3s4jQ6tCqfB50qaafMCHJtFrgz08BvwdA9iJA2wWAhYVpPwwREWXMMDtmzBicP38e+/fvh729fdT6xo0bY+3ataY+PyIyE+een0P7Le2x/9F+2FraYnyN8ZhdfzZcbF3eaLkdueE8Hr8OQaEcjvj6/XKqI5hRjnwH3NgJWNkBHZYD9voaXCIiynyMrpn9888/VWitUaNGjH+ApJX29u3bpj4/IjKT8WM/2/sZAjQBKORaSIVYz+yecW674uh97Lr8DDZWFljQpTJc7G2MO9j9I8C+KfrnLWYCeTiFNhFRZpas6WwN48zGHn/W6NYVIsoQpPV1VLVROPLkCCbUnAAnG6c4t7v02A9fbbuqno9pXgrl8hvZohroA6zvC+gigfKdgMq9THH6RESUmcoMZOavbdu2RS0bAuySJUtQs2ZN054dEaVbp7xPqdICg7bF2mJ63enxBtmAUA0GrD6D8EgtGpfKjT61PYw7oDYS2NgPCHgK5PIEWn3LOlkiIjK+ZXbatGlo3rw5rly5goiICMybN089P3LkCA4cOMBLSpTBRWojseTiEvxw/gfkdMiJ9a3XI5t9NvVafN/OSJ3s2D8u4d7LYOTNYo/ZHcob/03OwVnAnf2AdCaTOlnbuEMzERFlLka3zNapUwfnzp1TQVamsN29e7cqOzh69Kgad5aIMq4XIS/wyd5PsODcAmh1WtTIUwN20gkrEetOPcTm809gZWmB+V0rIaujrXEHvv03sP/f0VJazQXc4q7HJSKizMfolllRtGhRLF682PRnQ0Tp1vGnxzHq4Ci8DH2pZvEaW32sKi1IzHXvAEzcfFk9H96kJLwKGTlDl/9TYEM/ad/V18hW6JTcj0BERJk5zGq1WsyaNQubN29WEyY0atRITWfr4OCQsmdIRGlKWmAXnl+IH8//CB10KJa1GL6p/w2KZC2S6HuDwyNUnWyoRot6JXLh43qJvyeGyAh9h6/gF4B7OaD5zOR/ECIiytxlBl999RW++OILODs7I1++fKpW9n//+1/Knh0RpTkLWOD269sqyH5Q/AOsbrk6SUFWTNp8GTefByKXix3mdKwAS0sj62T/mgo8OALIWLVSJ2vz39jWRERERrXMrlixAj/88AM+/lg/ZeTevXvRsmVLNYqBpWWyZ8UlonTcImtpYak6ak2uNRnNPJqhiUeTJL//z7OPse7UIzUx17zOFZHTOfHa2hiu7wQOz9U/lxm+chQ18hMQEVFmkOQUKtPWtmjRIsaMX/KP3JMnT1Lq3IgoDURoIzD39FyMODBCjUJgGEfWmCB7xycQY/+4qJ4PbFgctYrmNO4kXj8A/tD/4ozqnwBl2hn3fiIiyjSS3DIroxdEn75W2NjYQKPRpMR5EVEa8A7yxsiDI3H2+Vm1fOrZKVR1r2rUPkI1kRiw+iyCwiNRo0h2DGxU3LiTiAgHfu8NhL4G8nkB70417v1ERJSpJDnMSgtN7969YWf331eFoaGh+OSTT+Dk9N94jxs3bjT9WRJRijv46CC+OPQF/ML84GzjjEm1JhkdZMW07Vdx5ak/sjvZYl7nSmo4LqPsGQ88Pg3YZwU6LAOsjRzGi4iIMpUkh9levd6cNrJ79+6mPh8iSmUarQbfnfkOyy4vU8ulc5TG7HqzUcC1gNH72nHxKVYcva+ef9OxAnK7Gtlh6/KfwPFF+ufv/QhkLWj0ORARUeaS5DD7yy+/pOyZEFGakLFj99zfo553K9UNQ72GwtbK+NbQh77BGLnhgnr+cf0ieKekm3E7eHkb2DRA/7z2YKBkM6PPgYiIMp9kTZpARBmHBNhT3qcwsdZENCrYKFn7CI/QYsBvZxEQGoFKBbOqyRGMogkB1vUCwgOAgrWAhuOTdR5ERJT5MMwSZTLhkeG45nsN5XOVV8teub2w84OdcLRxTPY+Z+++jvMPX8PV3hrzu1SCjZWRw/XtGAU8uwg45gTaLwWs+KOJiIiShgPEEmUiDwMeoseOHui3ux/uvL4Ttf5tguxf157hp4P6fc3qUAH5sxm5r/NrgDPL1fQM+GAJ4Jon2edCRESZD8MsUSax+95udNzSEVdeXlE1sT4hPm+9z6d+IRi27rx63ruWB5qWcTduB8+vAVuH6J83GA0Ufeetz4mIiDIXfpdHlMGFRYZh1slZWHt9rVqumKsiZtWfBXcnI4NnLBGRWgz67RxeBWtQNp8rxrTwNPLEAoF1PQFNMFCkAVBvxFudDxERZU4Ms0QZ2H3/+xh+YLiqkRUflv0Q/6v0P9hY2rz1vuftu4kT93zhbGeNBV0qw87aKulvlpnFpEX2xXXAJQ/w/hLA0oj3ExER/YthligD23pnqwqy2eyyYVrdaaiTr45J9nv41gss+PuWej7t/XLwyPnfxClJIjWyF9cBFlb6Dl/OuUxyXkRElPkwzBJlYB+X/xhBmiD0Kt0LuZ1ym2SfPgFhGLTmnGpc7Vy1ANpUyGvcDp5eALaP1D9vNAEoVMsk50VERJkTO4ARZSB3/O5g7KGxavgtYW1pjZFVR5osyGq1OgxZew4vAsNQIrczJrYuY9wOQv30dbKRYUCJZkCtgSY5LyIiyrzYMkuUQWy+vRlfHvsSIREhyO2YGwMrmz4oLjxwG4duvYC9jSW+71oZDrZG1snKDF+v7gJZCgLtFgKW/H2aiIjeDsMskZkL1gRj2vFp2HR7k1qunqc6upbqavLjnLzni292X1fPp7Qti+K5XYzbwfEfgaubAel81mEZ4Jjd5OdIRESZD8MskRm79eqWGq3gtt9tWFpY4tMKn6J/uf6wMvHIAK+CwjHwt7PQ6oD3KuVDB6/8xu3g0Slg9zj986ZfAfm9THp+RESUeTHMEpmpvx78hVEHRyE0MhS5HHJhRr0ZqOpe1eTH0el0GP77eTz1C0XhnE6Y2q4sLCwskr6DYF/g996AVgOUbgdU+8jk50hERJkXwyyRmSqetbjq4FU7d218Vecr5HDIkSLH+fnQXey79hy21pZY0LWSGlc2ybRa4I+PAb+HQPYiQJv5gDFBmIiIKBEMs0Rm5GXIy6jQWsC1AFa1WAWPLB6qxCAlnHv4GjN26idcGN+yFMrkzWLcDo7MA27uBqzsgI4rAHvXFDlPIiLKvNiVmMgMyFf9666vQ7MNzXDkyZGo9UWyFkmxIOsfqsHnv52BJlKH5mXd0b1GIeN2cO8wsG+q/nmLWYB7uRQ5TyIiytwYZonSuYDwAIw4OAJTj01V9bHb72xPlfA8esMFPPQNQf5sDpj+QXnj6mQDnwPr+wK6SKB8Z6Byz5Q8XSIiysRYZkCUjl1+eRkjDozAw4CHsLawxmCvwehRukeKH3fV8QfYftEb1pYWWNC1MrI42CT9zdpIYEM/INAbyOUJtJrDOlkiIkoxDLNE6ZC0jK6+thrfnPoGGq0GeZ3yYmb9maiQq0KKH/vKE39M2XpFPR/VzBMVC2Q1bgcHZgJ3DwA2jvo6WVunlDlRIiIihlmi9Om493FMPzFdPW9YoCGm1J6CLHZGdr5KhqCwCAxYfQbhEVo09HTDh3UKG7eD238BB2bon7eaC+QqmSLnSUREZMCWWaJ0qEaeGvig+Aconq04unp2Na5e9S2M//MS7rwIgrurPWZ3qABLSyOO6/8E2NBf2pUBr95AhU4peapEREQKwyxROikrWHt9LZp6NEU2+2xq3aRak1L1HNaffoSNZx9D8ut3XSohu5Nt0t8cqdF3+Ap+oR+1oNm/rbNEREQpjGGWKI29Dn2NcYfH4cCjA/jn8T+Y33B+ig23FV2kVocTd33xPCAUmkgtxv1xSa0f0rgEqhXOnngnr/tHgMBngHNu4OYu4MFRwM4V6LAcsLFP8fMnIiISDLNEaejc83Nq2C3vIG/YWtqiXr56sEDKlxTsvPQUk7dcUVPURufp7ozP3imW8JuvbAZ2jtKXFcTWdgGQo6iJz5aIiCh+DLNEaUCr0+KXS79g/tn5iNRFopBrIcyuPxue2T1TJch+uvKMVLa+4Zp3IPZc8UazsnniD7LrZMzYuN4tOFUtERGlLk6aQJQGZQWf7fsMc8/MVUG2eeHmWNtqbaoEWSktkBbZhKKovC7bxVlaIC2yCb1752j9dkRERKmEYZYolVlaWuKe3z3YWdlhUs1JmFF3BpxsUn4sVhlu689zj98oLYhOYqq8LrW0MWi1wOU/4y4tiP5u/8f6WloiIqJUwjIDolQqK5BaWBliy9XWFXMazIG1pTVKZCth8lERvP1DcdcnSA2xdfdFEO74BKo/H74KibvF9V9WiEReixfwsHgGx/NXgZu+gO8d/ePVPSAyLGknIZ3CiIiIUgnDLFEKexHyAmP+GYPGBRujk6d+7NXSOUq/1T79QjQqoN59EaiC62157qMPryGa+L/md7LSwk37DIUsvFVoLWTxDB4W3urPAhY+sLH4970X4nizhRWgS0IJgYxuQERElEoYZolS0PGnxzHq4Ci8DH2Jq75X0apoqySXFIRFROLBy+A3Wljl8SIwPN73OVpqUD2LPyq5vIKn7QsVXHNHPIFz0ANYBTyChU4b73tDdTZ4YumOwsXLwkJGJcheGMheRP9wzgPMrwj4P42nbtYCcM0LFKqVpM9HRERkCgyzRCkgUhuJRRcW4cfzP0IHHYplLaZGK4gdZLVaHZ5GlQUE4s6/ravyePQqGPFVBTgiFJWcX6Gy8yuUtn+BwhbP4BbxBK4hD2Ad+BQIgf4RhwgrR9zQ5MJ9XW7cUw939fy+NjeeIRt+6F4FReIbzUAmQ1CjGUhXMd2boxg0mw5YWiXrmhERESUHwyyRiT0Pfo7R/4zGSe+Tavn94u/j07JD8eS1FutvP9KXBqiW1iDcexmEUE3cLaUuCEYpuxfwcnmFMhJYLZ/BPfIJXIMfwibEB4iQoRHiOQmZvMDQohrrYe3shgeXvTEl1jizebLY44fWpeMflkuUbgN0XPHmOLPSIitBVl4nIiJKRQyzRCYSqonEtWcv8NnBjgjQ+MIKdsgV1g2b95bB8s2H4niHDlkRiDJWz1HZxRdl7X1RxOoZ8kQ+QZaQR7AJ+3dEgcB/H7E5ZI83sMIxO2AR/5ivEljfLe0eNQOYm4u9mvXLSuayTYwEVs+WMWcAk9ICtsgSEVEaYJglMoKMBvDkdYi+jvXfGtY7/7ayPvELgU4H2OaoAmvXiwh83BWvw3MiJ17Ay8IbFRx9Uc7xpQqseSOfImvoI9ho/PU7lgbSuEbMcnKLFVQNNayFAYdsb3XvJLjWLJojeW+W4Fq47lsdn4iIyBQYZoniGN7KNyg8KqjefhaAo9ctseD2Ydz3DVHjtUZnaf0KOSxeoho0KGnvg7La1/AItEVhl++RLewRrCOC9RvKQAABcVxul7yxgmq04GrnwvtDRESUAIZZStetoEn6GlxmnErGV94h4ZFRna2kjlVaVw0jB8jQV7FnF8lrcQ9VLbxR1OY5yjr4orj1czy0fYJZrhrkiozA8ifPYC9Ns290vLIAshaIuxwgayHA1vEtrxQREVHmxTBL6dLOS0/VtKqxOyhNjN1B6crmeDojzVC1nRGRWjyOKgvQjxigwqsqC4j5vb41IpDP4gUqyvirVt4oa/8SxW18kF/3FNnCnqjXDTThwHdOWbEsq6uKunkiAL/sHrDPVjSOwFoQsLZL4StGRESUOTHMUroMsp+uPPPGSKbefqFq/cLulfWBVoLsup5q6Kvo7bU6CbbremCyw2is8q8ATeR/e7KFBgUsnsPT4hmaWT1DCZvnKGn7AgV03siu8YalqgX4lzyNtqizsoVFNg88yZYfIyxe4IJGP5RA18ItMazGeNjapvyUtERERJQOw+z333+PWbNmwdvbGxUqVMD8+fNRrVq1OLddvHgxVqxYgUuXLqllLy8vTJs2Ld7tyfxKC6RFNq7hVQ3rRm+8iGtPXqPXiaHIGivIClmW8VkHBi+ALVqhiO0LlLT1QUF4I3vEc1jE3nv0+QesHWJ2sspeBBGuhfDX+Xt4p003/PPsCMYdHoeA8AC42LhgSu0paFyosakvAxEREZlLmF27di2GDh2KRYsWoXr16pg7dy6aNm2K69evw83N7Y3t9+/fjy5duqBWrVqwt7fHjBkz0KRJE1y+fBn58uVLk89ApiM1stFLC+LyOliDY/u3YrCtT7zbSGltdgTiC5s1+hX/VQgAts5xdLYyzHLlDlhKhex/dBoNQq4GQmthgeWXl6sgWzZHWcyqPwv5XfK/3QcmIiIi8w6zc+bMQf/+/dGnTx+1LKF227ZtWLp0KUaPHv3G9qtWrYqxvGTJEmzYsAH79u1Dz54yMxGZM+nslRS13DTxTxgQXf5qQJEGMUcLcMqV4Bis8bG0sMSMejOw7vo6fFrhU9hY2Ri9DyIiIspAYTY8PBynT5/GmDFjotZZWlqicePGOHr0aJL2ERwcDI1Gg+zZs8f5elhYmHoY+Pvrx/WU98gjpRmOkRrHyghyOCbtr+Q7XmWBfYlvF9HgC+gK1Ym1MnozbcL2PtiLay+voQiKqHuYwzYHPi33KaAFNFreU3PB/w7NH++heeP9M3+aVM4zxhwnTcPsixcvEBkZidy5c8dYL8vXrl1L0j5GjRqFvHnzqgAcl6+//hqTJ09+Y/3u3bvh6Jh6QyLt2bMn1Y5lzqTWNautFV6rOta4Wk91yGoL3HoRiWI22WGv8Y1nKyDEJjv2XHoNXN5u9HlodBrsDNmJ4+HH1XJfp768hxkA/zs0f7yH5o33z/ztSaU8I42VZlNm8DamT5+ONWvWqDpaqZ+Ni7T6Sk1u9JbZAgUKqDpbV1cZVinlf7OQG//uu+/CxoZfSyeFjcczfL7mvHoevauWPrRa4Mv3K6BpmdywKAZgQx+1TfROXYYuYbZt5qCFZyuj79kD/wcYfXg0roXrf6HqWbInCnkX4j00Y/zv0PzxHpo33j/zp0nlPGP4Jj0p0jTM5syZE1ZWVnj27FmM9bLs7u6e4Htnz56twuzevXtRvnz5eLezs7NTj9jkRqRmuEzt45mzVhXzw9ra6o1xZt1jjzNb7j3AyuqNcWYt1Diz02Fduo3Rx95+ZzsmH52M4IhgZLPLhml1p6G6W3Vs376d9zAD4H+H5o/30Lzx/pk/m1TKM8YcI03DrK2trRpaSzpvtWvXTq3TarVqecCAAfG+b+bMmfjqq6+wa9cuVKlSJRXPmFKLBNZ3S7snPgOYBFbPlsmaASy2WSdnYcWVFeq5V24vzKg7A7mdcrPemYiIKB1L8zIDKQHo1auXCqUyVqwMzRUUFBQ1uoGMUCBDbkntq5ChuCZMmIDVq1fDw8NDjU0rnJ2d1YMyDgmuNYvmSHxDCa6F67718crlKgcLWKB/+f5qtAJryzT/z4OIiIgSkeb/Wnfq1Ak+Pj4qoEowrVixInbu3BnVKezBgwdqhAODhQsXqlEQ2rdvH2M/EydOxKRJk1L9/Mm8vQh5gZwOOdXzZh7NUCJbCRTJUiStT4uIiIjMJcwKKSmIr6xAOndFd+/evVQ6K8rIgjXBmHZ8Gg49PoT1bdZHBVoGWSIiIvMSc6ojokzg1qtb6LqtKzbd3oRXYa9w7OmxtD4lIiIiMueWWaLUoNPp8OetP1WLbGhkKHI55FIzelV1r8obQEREZKYYZinTlBVMOTYF2+5sU8u18tbCtDrTkMMhCR3MiIiIKN1imKVM4ccLP6oga2VhhQGVBqBv2b6wtGCVDRERkbljmKVM4ePyH+PKyytqyK3KuSun9ekQERGRibBpijKkwPBALL+8XNXJCkcbRyxusphBloiIKINhyyxlONICO+LACDwIeKCWe5XpldanRERERCmEYZYyDGmF/e3ab5h9ajY0Wg3yOOVBRbeKaX1aRERElIIYZilD8A/3x8TDE7H3wV61/E6BdzC19lRkscuS1qdGREREKYhhlsze5ReXMezAMDwOfAxrS2sM8xqGbqW6wcLCIq1PjYiIiFIYwyyZPa1Oi2dBz5DPOR9m15+NsjnLpvUpERERUSphmCWzFKmNhJWllXpeLlc5zH1nLirlrgRXW9e0PjUiIiJKRRyai8zOuefn0G5TO1z3vR61rn6B+gyyREREmRDDLJlVOcHSS0vRe2dv3PO/h3ln5qX1KREREVEaY5kBmQXfUF+MPTQWhx4fUsvNPZpjQs0JaX1aRERElMYYZindO/3sNEYeGInnIc9hZ2WH0dVG44PiH3C0AiIiImKYpfTtzLMz6Lurryox8HD1UKMVlMxeMq1Pi4iIiNIJtsxSulYhVwVUda8KNwc3jKsxDo42jml9SkRERJSOMMxSunP2+VmUyl4K9tb2avitBQ0XqOdEREREsXE0A0pXY8f+cO4H9NrRCzNPzoxazyBLRERE8WHLLKULPsE+GPXPKJz0PqmWI7QRMSZGICIiIooLwyyluSOPj2DMoTFq+C0HaweMrzEerYu2TuvTIiIiIjPAMEtpRlpfpaxgycUl/2/vTqCjqu4Hjv+yh0jCFkPYEZBFg1JAKAkYl2AERLAKiMgmAgq0FJAdAWUtKMVaFqkCdalUEFAh/wBSEBEQZalQWURAwLIoggmEkO39z+/2zJwMZIVkZl7y/ZzzGN6b++bdub9J8ps7994RSyypX6G+zI6dLXXK1SEqAACgQEhm4THaE/vB4Q9MItulfhcZdc8oxscCAIBCIZmFx0SERMj01tPlcvplaXdbOyIBAAAKjWQWbpOelS6v73ldmkY0lftq3GeO3Vv9XiIAAABuGEtzwS1OXzotfRP7ypL9S2TCFxMkKS2JlgcAADeNnlkUu00nNjkT2NCAUJncarKEBYbR8gAA4KaRzKLYpGemy5xdc+TdA++a/ahKUWa1guqh1Wl1AABQJEhmUSyuZFyRZxKfkf3n95v9nnf0lGFNh0mAXwAtDgAAigzJLIqFfvlBw0oN5UTyCZkaM1Xur3k/LQ0AAIocySyKzNXMq5KakSrlgsqZ/dH3jJYBjQdIlbJVaGUAAFAsWM0AReJE0gnpmdBTRnw2QjKzMs2xYP9gElkAAFCsSGZx0/7v2P9J1zVd5cAvB+TQL4fkZPJJWhUAALgFwwxww3RIwZ+++pOsOLzC7OuXIfzp3j9J5C2RtCoAAHALklnckGO/HpMXPntBDl84LD7iI882flYGNRkk/r68pAAAgPuQeaDQLMuSMZ+PMYlsxeCKMqPNDImuGk1LAgAAt2PMLArNx8dHXo5+WVpXay0rOq4gkQUAAB5DMosCOXLhiHzy/SfO/QYVG8iCuAVya8ittCAAAPAYhhkg3yEFq4+slulfTpcMK0Nqh9WWxrc2ptUAAIBXIJlFrlLSU2TqjqnyydH/9ci2qtJKqpatSosBAACvQTKLHOl6sbpawfGk4+Lr4ytDmgyRfo37mf8DAAB4C5JZXOfDwx+aYQVpWWkSERIhs+6dJc0qN6OlAACA1yGZxXUupV8yiayuVjC99XSpEFyBVgIAAF6JZBZGRlaG8wsPet3Ry3yLV9tabRlWAAAAvBoDIEs5Xa3g/YPvy5NrnjQTvhzryMbXjieRBQAAXo9kthRLSkuSEZ+NMONjD104JCu/W+npKgEAABQKwwxKqf0/7zerFfx46UczvGB4s+HSo1EPT1cLAACgUEhmS+GwgncPvCtzds0x42Srla0mr8S+IlHhUZ6uGgAAQKGRzJYyb3zzhszbO8/8P65mnLwU85KEBYZ5uloAAAA3hGS2lHmi/hNmbGzfqL7yZIMnzWQvAAAAuyKZLeGyrCzZcXqHRFeNNvvhZcLlk8c+kSC/IE9XDQAA4KaxmkEJdiH1ggzZOEQGbhgoiccTncdJZAEAQElBz2wJtevsLhm1ZZScSzkngb6BkpqR6ukqAQAAFDmS2RI4rOCtfW+ZSV6ZVqbUDqttVitoULGBp6sGAABQ5EhmS5DzV87L2M/HyvbT283+I3UekRd/+6KEBIR4umoAAADFgmS2hH0RgiaywX7BMq7lOOlcrzOrFQAAgBKNZLYEia0RKy80f0FiqsZIvQr1PF0dAACAYsdqBjb2U8pPMnzzcDlz+YzzWO87e5PIAgCAUoOeWZva9t9tZnzsL6m/SEp6iixsu9DTVQIAAHA7klmbycjKkPl758ub+94USyy5vcLtMqrFKE9XCwAAwCNIZm1EhxOM3jJadp/b7fxq2tH3jJZg/2BPVw0AAMAjSGZt4uAvB6X/+v5y8epFCfEPkcnRk6Xdbe08XS0AAACPIpm1iVphtSS8TLhUuaWKzI6dbfYBAABKO5JZL1+toFKZSuLr4ytl/MvIgrgFUiG4ggT5BXm6agAAAF6Bpbm81KYTm6TzR53NRC+HyFsiSWQBAACyIZn1MumZ6TLrq1nyh01/kKS0JPns1GdmBQMAAABcj2EGXuRU8ikZtWWU7Pt5n9l/utHTMrzZcPH3JUwAAAA5IUvyEp/+8KlM/GKiJKcnS2hgqEyNmSoP1HzA09UCAADwaiSzXuBcyjmzfmxaVprcdetdMvve2VK1bFVPVwsAAMDrkcx6gYiQCBndYrScTD4pf2j6BwnwDfB0lQAAAGyBZNZDEo8nSvWy1SUqPMrsd23Q1VNVAQAAsC2SWTdLzUg1qxUsP7xcqpWtJss7LjdjZAEAAGDTpbnmzZsntWvXluDgYGnZsqXs3Lkzz/LLly+Xhg0bmvKNGzeWhIQEsYNjvx6THgk9TCLrIz7S/rb25ssQAAAAYNNk9p///KcMHz5cJk2aJLt375a7775b4uPj5dy5czmW37Ztm3Tv3l369esne/bskc6dO5tt//794s3WHlsr3dZ0k8MXDkvF4IqyMG6hGR/LslsAAAA2TmbnzJkj/fv3l759+8odd9whCxculJCQEFm8eHGO5V977TV5+OGHZeTIkdKoUSOZMmWKNG3aVP7617+KN0rLTJOVKSvlxe0vypWMK3JP5D2youMKia4W7emqAQAA2J5Hx8ympaXJrl27ZOzYsc5jvr6+EhcXJ9u3b8/xHD2uPbnZaU/u6tWrcyx/9epVszkkJSWZ2/T0dLMVNyvTkktZl8ywgv5R/c3m5+vnlmujaDhiRczsixjaHzG0N+Jnf+lu/ltYmOt4NJn9+eefJTMzUypXruxyXPcPHjyY4zlnzpzJsbwez8mMGTPkpZdeuu74+vXrTQ+wOzwe8riczTwrNU/VlHWn1rnlmih6GzZsoFltjhjaHzG0N+Jnf+6KYUpKSoHLlvjVDLTXN3tPrvbM1qhRQx566CEJCwtzyzsLDfzA+IESEMD6sXbkiGHbtm2JoU0RQ/sjhvZG/Owv3c1/Cx2fpHt9MhseHi5+fn5y9uxZl+O6HxkZmeM5erww5YOCgsx2LQ2EO5NLd18PRY8Y2h8xtD9iaG/Ez/4C3JTPFOYaHp0AFhgYKM2aNZONGzc6j2VlZZn9Vq1a5XiOHs9eXuk7hdzKAwAAoOTy+DADHQLQu3dvad68ubRo0ULmzp0rly9fNqsbqF69ekm1atXM2Fc1dOhQiY2NlVdffVU6dOggy5Ytk6+//loWLVrk4WcCAACAUpfMduvWTX766SeZOHGimcTVpEkTSUxMdE7yOnHihFnhwCE6Olr+8Y9/yIQJE2TcuHFy++23m5UMoqL+97WwAAAAKD08nsyqIUOGmC0nmzdvvu5Yly5dzAYAAIDSzeNfmgAAAADcKJJZAAAA2BbJLAAAAGyLZBYAAAC2RTILAAAA2yKZBQAAgG2RzAIAAMC2SGYBAABgWySzAAAAsC2SWQAAANgWySwAAABsi2QWAAAAtkUyCwAAANvyl1LGsixzm5SU5JbrpaenS0pKirleQECAW66JokUM7Y8Y2h8xtDfiZ3/pbs5nHHmaI2/LS6lLZpOTk81tjRo1PF0VAAAA5JO3lStXLq8i4mMVJOUtQbKysuS///2vhIaGio+Pj1veWWjifPLkSQkLCyv266HoEUP7I4b2RwztjfjZX5Kb8xlNTzWRrVq1qvj65j0qttT1zGqDVK9e3e3X1cCTzNobMbQ/Ymh/xNDeiJ/9hbkxn8mvR9aBCWAAAACwLZJZAAAA2BbJbDELCgqSSZMmmVvYEzG0P2Jof8TQ3oif/QV5cT5T6iaAAQAAoOSgZxYAAAC2RTILAAAA2yKZBQAAgG2RzAIAAMC2SGaLwLx586R27doSHBwsLVu2lJ07d+ZZfvny5dKwYUNTvnHjxpKQkFAU1YCbYvi3v/1N2rRpIxUqVDBbXFxcvjGH9/0cOixbtsx8G2Dnzp2LvY4o2hhevHhRBg8eLFWqVDEzrOvXr8/vUxvFb+7cudKgQQMpU6aM+WapYcOGSWpqqtvqC1dbtmyRjh07mm/c0t+Jq1evlvxs3rxZmjZtan7+6tWrJ0uXLhWP0NUMcOOWLVtmBQYGWosXL7b+85//WP3797fKly9vnT17NsfyX3zxheXn52fNmjXL+vbbb60JEyZYAQEB1r59+wiDTWL41FNPWfPmzbP27NljHThwwOrTp49Vrlw569SpU26vO24shg7Hjh2zqlWrZrVp08bq1KkTzWmjGF69etVq3ry51b59e2vr1q0mlps3b7b27t3r9rqj8PF77733rKCgIHOrsVu3bp1VpUoVa9iwYTSnhyQkJFjjx4+3Vq5cqatcWatWrcqz/NGjR62QkBBr+PDhJp95/fXXTX6TmJhouRvJ7E1q0aKFNXjwYOd+ZmamVbVqVWvGjBk5lu/atavVoUMHl2MtW7a0Bg4ceLNVgZtieK2MjAwrNDTU+vvf/04MbBRDjVt0dLT15ptvWr179yaZtVkMFyxYYNWpU8dKS0tzYy1RVPHTsg888IDLMU2KYmJiaGQvIAVIZkeNGmXdeeedLse6detmxcfHW+7GMIObkJaWJrt27TIfMzv4+vqa/e3bt+d4jh7PXl7Fx8fnWh7eF8NrpaSkSHp6ulSsWLEYa4qijuHLL78sERER0q9fPxrXhjH8+OOPpVWrVmaYQeXKlSUqKkqmT58umZmZbqw5bjR+0dHR5hzHUISjR4+aISLt27enUW1iuxflM/5uv2IJ8vPPP5tfnPqLNDvdP3jwYI7nnDlzJsfyehz2iOG1Ro8ebcYYXftDDe+N4datW+Wtt96SvXv3uqmWKOoYavLzr3/9S3r06GGSoCNHjsigQYPMG0v9liJ4d/yeeuopc17r1q31E2LJyMiQ5557TsaNG+emWuNm5ZbPJCUlyZUrV8xYaHehZxa4CTNnzjQTiFatWmUmPcD7JScnS8+ePc1EvvDwcE9XBzcoKyvL9KwvWrRImjVrJt26dZPx48fLwoULaVMb0IlD2pM+f/582b17t6xcuVLWrl0rU6ZM8XTVYEP0zN4E/UPo5+cnZ8+edTmu+5GRkTmeo8cLUx7eF0OHV155xSSzn376qdx1113FXFMUVQy///57OX78uJm1mz0xUv7+/nLo0CGpW7cuDe7lP4e6gkFAQIA5z6FRo0amt0g/9g4MDCz2euPG4/fiiy+aN5XPPvus2deVfS5fviwDBgwwb0p0mAK8W2Qu+UxYWJhbe2UVr5aboL8stUdg48aNLn8UdV/HcuVEj2cvrzZs2JBreXhfDNWsWbNMD0JiYqI0b96cMNkohros3r59+8wQA8f26KOPyv3332/+r0sEwft/DmNiYszQAscbEXX48GGT5JLIen/8dK7BtQmr443J/+Yfwdu18qZ8xu1TzkrgciS6vMjSpUvN0hQDBgwwy5GcOXPG3N+zZ09rzJgxLktz+fv7W6+88opZ1mnSpEkszWWzGM6cOdMsQbNixQrr9OnTzi05OdmDz6J0K2wMr8VqBvaL4YkTJ8wqIkOGDLEOHTpkrVmzxoqIiLCmTp3qwWdRehU2fvq3T+P3/vvvmyWe1q9fb9WtW9es+APPSE5ONktO6qbp4Zw5c8z/f/jhB3O/xk/jeO3SXCNHjjT5jC5ZydJcNqZrq9WsWdMkOLo8yY4dO5z3xcbGmj+U2X3wwQdW/fr1TXld1mLt2rUeqDVuNIa1atUyP+jXbvrLGfb5OcyOZNaeMdy2bZtZ2lCTKF2ma9q0aWbJNXh//NLT063JkyebBDY4ONiqUaOGNWjQIOvChQseqj02bdqU4982R9z0VuN47TlNmjQxMdefwSVLlnikIX30H/f3BwMAAAA3jzGzAAAAsC2SWQAAANgWySwAAABsi2QWAAAAtkUyCwAAANsimQUAAIBtkcwCAADAtkhmAQAAYFskswAAALAtklkABdanTx/x8fHJdbt48SKtCQBwK5JZAIXy8MMPy+nTp122Dz/8kFYEAHgEySyAQgkKCpLIyEiXrWLFii5lli5dKuXLl5fVq1fL7bffLsHBwRIfHy8nT550KffRRx9J06ZNzf116tSRl156STIyMlzKTJ48+boe4M6dO7uU+eKLL+S+++6TkJAQqVChgrnWhQsXzH16/I9//KOz7Jtvvmnqtnv3brOfmZkp/fr1k9tuu03KlCkjDRo0kNdee83l8ceMGSNVq1aVwMBAqVatmowePVqysrIKfL72aF9bZ0cbZX+eTZo0cSmzefNmlx7v8+fPS/fu3U0d9Lk2btxY3n//fZdzLl26ZK5XuXJllzbbu3ev5CZ7ubCwMGnbtq18//33zvuvXr0qL7zwgrnuLbfcIi1btjR1K2y8FyxYIHXr1jXtqO30zjvvuNx/8OBBc+1y5co56+Noo5xeB45NY1zQdi5IPbS9Bw4caNpQn0tUVJSsWbPGGY/cttyuB6B4kcwCKBYpKSkybdo0efvtt02yqQnCk08+6bz/888/l169esnQoUPl22+/lTfeeMMkAnrOte68805nL3DXrl1d7tMk7cEHH5Q77rhDtm/fLlu3bpWOHTuaJPNaH3zwgQwbNkw+/vhjk0QrTUqrV68uy5cvN/WYOHGijBs3zpR1eOihh0wyc+TIEZMML1q0SN59990Cn19UUlNTpVmzZrJ27VrZv3+/DBgwQHr27Ck7d+50lpk+fbqsX7/eXF/bK/t9eVmyZIkpv2XLFjl37px5Dg5Dhgwxbbts2TL55ptvpEuXLqaH/rvvvitwvFetWmViPWLECFN3TRb79u0rmzZtcpZ55plnJD093ZyvdZk7d67zPk2mHa8BfYxWrVo591euXFngNsyvHhrPdu3amTpojDWmM2fOFD8/P4mOjr7u04jsn1AA8BALAAqod+/eVqdOna47vmnTJkt/nVy4cMHsL1myxOzv2LHDWebAgQPm2Jdffmn2H3zwQWv69Okuj/POO+9YVapUcTk2ZswYq3nz5rnWoXv37lZMTEyudY6NjbWGDh1qJSQkWCEhIdbatWvzfZ6DBw+2Hn/88RzvO3r0qKnj4sWLC3x+Tu2mbVSuXDnn/qRJk6y77747z3bNSYcOHawRI0Y499u1a2f179/fuX/s2DHzGHv27Mn1MfT+VatWmf9fvHjRtKfjMX744QfLz8/P+vHHH13O0fiNHTu2wPGOjo52qZfq0qWL1b59e+d+mTJlrPfeey/XNsreVhrXaxWknfOrx7p16yxfX1/r0KFDVl4csblWbnUGUHzomQVQLPz9/eWee+5x7jds2NB8/HrgwAGz/+9//1tefvllKVu2rHPr37+/6eHSXj4H/WhdP/rOjaNnNi/aO/n44487PyK/1rx580yP56233mrqoT2vJ06ccCmjPZ760b4Oh9DH0l7lwpyvPbvZn+tzzz13XT327dvnUkZ7CLPT3uYpU6aY4QU6tEPLrFu3zuVaOtxBPw7/8ccfpTB0+II+ng7TSE5OlhkzZjjrpNetX7++S90+++wzl6EI+cVbb2NiYlyuqfuO+x11157T7PEvrPzaOb966OtJe9r1+d6oX3/91VxbX7c67EJ7lbXHGUDx8C+mxwWAPOnYTh0j+7vf/e66+3ScosPRo0dNkpMbHaeaH/2IXMdJrlixwnxknn2cqX50rsnGq6++aj66Dg0NldmzZ8uXX37p8hiaFGldd+3aZcbg6v/vv//+Ap+vZbUODvrRuCbI2en4TR0C4aCP8fTTTzv39XF1PK5+/K4JrSbnWpe0tDRnGR3mcPjwYZOQ6f3/63jN35///GeJi4szwwPGjx9vxp9+8sknJk76Ebs+b73NThO2ovTWW29J7969TRtqXHX8dPbXQkEUpJ3zUpDXU360/jomW9tehynoc9Kx5fo6AVD0SGYBFAtNRL7++mtp0aKF2T906JBJlBo1amT2dcyqHqtXr16eY0S1V1XHhebmrrvuko0bN5rEODd6viaj2tOpk3m09++xxx4z9+nYSB0LOWjQIGf57D2ODtoTqpv2OGpSrGMmNXEq6PmaWGZ/rhEREdeV0QlJ2cucOnXK5X69VqdOnZwJro7v1MRVxws76KQlHROqyVRCQoJpQ8cEqbxosuW49u9//3t59NFHTW/ib37zG9Mzq+No27Rpc8Px1lutvyZ22Z9P9rr/9re/NdfVcbs6XlXjVJhEtCDtnF899PWk7a7teqO9s76+vs46aM+sTmrLawIegJtDMgugWAQEBJik6C9/+Yv5CFp7RDVZcSQ72oP4yCOPSM2aNeWJJ54wCYAOPdBJOVOnTjU9gjoMQbVu3VrOnDlj/n/lyhUzu14/ytVZ72PHjjW9lJpMasKqCaFO5tFJSuHh4eYcx2oLtWrVMr2bzz//vNx7771SqVIlk2zopCX9uF57gHVm+1dffeXSGzx//nyJjY01iZJOMNuwYYNzxYKCnF9U9FqaSG/bts0MB5gzZ46cPXvWJSHUnmxN1LROOqTi+PHjBXpsTTy1jbVdtYdUh1NoDDWh69GjhxlWob3Pmtz+9NNP5g2EJn4dOnQoULxHjhxpJu/p+doDrL2+2mv66aefOuugbxB0EqD2AuvrIqeE/2blVw+Ns742dCiJtq8mpbrKgq5WoJPeCkrfRDh6ZvU1o28wABSTYhyPC6AUTwDTSTAffvihVadOHSsoKMiKi4szk4myS0xMNBNydOJPWFiY1aJFC2vRokXOST76mLltWheHzZs3m8fR65QvX96Kj4931sUxAcwhKyvLTF7SiWMqNTXV6tOnj6mvnvv888+bSWfZJ2Pp5KDw8HDz+PXq1bOmTZtmHqeg5xfVBLDz58+bxylbtqwVERFhTZgwwerVq5fzsVNSUqwmTZqY44WdAObYQkNDTZtlL5+WlmZNnDjRql27thUQEGAmwD322GPWN998U6h4z58/39yvj1G/fn3r7bffdt6nE660/davX59rGxXFBLD86uFo5759+1qVKlWygoODraioKGvNmjUFngDmaEsfHx8rMjLSvCauXr2aQ8sDKAo++k9xJcoASiftXdOxnDfzjWC6rmj22+x0PVPd9DooGfEGgBvFMAMAXimvyUU6KUiHGAAAQDILwCvlNfNbxy4WZvwiAKDkYpgBAAAAbIsvTQAAAIBtkcwCAADAtkhmAQAAYFskswAAALAtklkAAADYFsksAAAAbItkFgAAALZFMgsAAACxq/8HvkYp49tXhMQAAAAASUVORK5CYII=",
|
||
"text/plain": [
|
||
"<Figure size 800x600 with 1 Axes>"
|
||
]
|
||
},
|
||
"metadata": {},
|
||
"output_type": "display_data"
|
||
}
|
||
],
|
||
"source": [
|
||
"prob_true, prob_pred = calibration_curve(\n",
|
||
" y_test,\n",
|
||
" probs,\n",
|
||
" n_bins=5,\n",
|
||
" strategy='quantile'\n",
|
||
")\n",
|
||
"\n",
|
||
"prob_true_cal, prob_pred_cal = calibration_curve(\n",
|
||
" y_test,\n",
|
||
" calibrated_probs,\n",
|
||
" n_bins=5,\n",
|
||
" strategy='quantile'\n",
|
||
")\n",
|
||
"\n",
|
||
"plt.figure(figsize=(8, 6))\n",
|
||
"\n",
|
||
"plt.plot(prob_pred, prob_true, marker='o', label='Без калибровки')\n",
|
||
"plt.plot(prob_pred_cal, prob_true_cal, marker='o', label='С калибровкой')\n",
|
||
"\n",
|
||
"plt.plot([0, 1], [0, 1], linestyle='--', label='Идеальная линия')\n",
|
||
"\n",
|
||
"plt.xlabel('Предсказанная вероятность')\n",
|
||
"plt.ylabel('Реальная вероятность')\n",
|
||
"plt.title('Calibration Curve')\n",
|
||
"\n",
|
||
"plt.legend()\n",
|
||
"plt.grid()\n",
|
||
"plt.show()"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"id": "61dff203-f0d6-4756-94e4-7eeda7cc6d3e",
|
||
"metadata": {},
|
||
"source": [
|
||
"## Построение калибровочной кривой\n",
|
||
"\n",
|
||
"Калибровочная кривая показывает, насколько предсказанные моделью вероятности соответствуют реальным значениям.\n",
|
||
"\n",
|
||
"Чем ближе линия модели к диагональной идеальной линии, тем лучше откалиброваны вероятности."
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 14,
|
||
"id": "2b772749-d774-4882-9a1a-06f23ed23b1f",
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"name": "stdout",
|
||
"output_type": "stream",
|
||
"text": [
|
||
"Вывод:\n",
|
||
"- Выполнена калибровка вероятностей классификатора.\n",
|
||
"- Использован внешний датасет из OpenML.\n",
|
||
"- После калибровки вероятности стали ближе к реальным значениям.\n",
|
||
"- LogisticRegression показала стабильную работу на внешнем датасете.\n"
|
||
]
|
||
}
|
||
],
|
||
"source": [
|
||
"print(\"Вывод:\")\n",
|
||
"\n",
|
||
"print(\"- Выполнена калибровка вероятностей классификатора.\")\n",
|
||
"print(\"- Использован внешний датасет из OpenML.\")\n",
|
||
"print(\"- После калибровки вероятности стали ближе к реальным значениям.\")\n",
|
||
"print(\"- LogisticRegression показала стабильную работу на внешнем датасете.\")"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "05e7fd9e-0eb8-43a1-a7d7-57067b067d3c",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": []
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"id": "f2af889c-126a-49f7-b2e6-0e051ff08e9c",
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": []
|
||
}
|
||
],
|
||
"metadata": {
|
||
"kernelspec": {
|
||
"display_name": "Python 3 (ipykernel)",
|
||
"language": "python",
|
||
"name": "python3"
|
||
},
|
||
"language_info": {
|
||
"codemirror_mode": {
|
||
"name": "ipython",
|
||
"version": 3
|
||
},
|
||
"file_extension": ".py",
|
||
"mimetype": "text/x-python",
|
||
"name": "python",
|
||
"nbconvert_exporter": "python",
|
||
"pygments_lexer": "ipython3",
|
||
"version": "3.14.4"
|
||
}
|
||
},
|
||
"nbformat": 4,
|
||
"nbformat_minor": 5
|
||
}
|