Skip to content

Instantly share code, notes, and snippets.

@SharanSMenon
Last active December 5, 2021 22:40
Show Gist options
  • Save SharanSMenon/c29a13172d95a1f4fcea30a16e7a7b47 to your computer and use it in GitHub Desktop.
Save SharanSMenon/c29a13172d95a1f4fcea30a16e7a7b47 to your computer and use it in GitHub Desktop.
A simple implementation of evolution strategies to train a classification network. Based on https://openai.com/blog/evolution-strategies/
Display the source blob
Display the rendered blob
Raw
{
"cells": [
{
"cell_type": "markdown",
"id": "35194e0e-1d70-43d1-a4d2-3cf5aeaf922e",
"metadata": {},
"source": [
"# Implementing Evolution Strategies for classification\n",
"\n",
"This notebook implements the algorithm discussed in the following blog post\n",
"\n",
"[https://openai.com/blog/evolution-strategies/](https://openai.com/blog/evolution-strategies/)\n",
"\n",
"We will use this algorithm to train a neural network to classify on scikit learn's `make_moons` dataset"
]
},
{
"cell_type": "code",
"execution_count": 1,
"id": "9c47fc44-973f-4518-abb9-a4d19d1cee4a",
"metadata": {},
"outputs": [],
"source": [
"import torch\n",
"import torch.nn as nn\n",
"from torch.nn import functional as F"
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "3a36b483-f921-4724-a120-88420823f3fe",
"metadata": {},
"outputs": [],
"source": [
"import numpy as np\n",
"import matplotlib.pyplot as plt"
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "7293a500-4194-4958-8367-7777979b02f0",
"metadata": {},
"outputs": [],
"source": [
"from sklearn.datasets import make_moons"
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "e743c05e-6173-4812-bc04-db15c5ce44f0",
"metadata": {},
"outputs": [],
"source": [
"X, Y = make_moons(noise=0.15, random_state=1, n_samples=500)"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "144f1c56-03a1-4c87-80ce-ca412a49ec73",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": 5,
"id": "eab9d1d7-8ac0-4da3-8aeb-0107dff77dc4",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"<matplotlib.collections.PathCollection at 0x12f8b0d60>"
]
},
"execution_count": 5,
"metadata": {},
"output_type": "execute_result"
},
{
"data": {
"image/png": "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",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"plt.scatter(X[:,0], X[:,1], c=Y)"
]
},
{
"cell_type": "markdown",
"id": "f9c16543-2be0-4c60-be33-b5692acac74c",
"metadata": {},
"source": [
"Create training and testing sets. Then convert them into PyTorch Tensors"
]
},
{
"cell_type": "code",
"execution_count": 6,
"id": "3f231db9-abf4-4da4-b2b1-6c2fe3dd3c1f",
"metadata": {},
"outputs": [],
"source": [
"from sklearn.model_selection import train_test_split"
]
},
{
"cell_type": "code",
"execution_count": 7,
"id": "ce5ae456-361c-4a8a-8a18-9f9c7cde70b5",
"metadata": {},
"outputs": [],
"source": [
"X_train, X_test, y_train, y_test = train_test_split(X, Y, test_size=0.36, random_state=42)\n",
"X_train, X_test, y_train, y_test = torch.FloatTensor(X_train), torch.FloatTensor(X_test), torch.LongTensor(y_train), torch.LongTensor(y_test)"
]
},
{
"cell_type": "code",
"execution_count": 8,
"id": "ded34ad7-00dd-45cf-80a0-09897177f17d",
"metadata": {},
"outputs": [],
"source": [
"n_input_dim = X_train.shape[1]"
]
},
{
"cell_type": "markdown",
"id": "d7cfb9c8-b043-4bf7-a1f8-bba1eea2b040",
"metadata": {},
"source": [
"## Model\n",
"\n",
"Create the model with and initialize its weights. We use `model.float()` to ensure that it accepts `float` datatypes and not `double`."
]
},
{
"cell_type": "code",
"execution_count": 9,
"id": "1867d3c2-5851-4c76-9444-47f24adc0877",
"metadata": {},
"outputs": [],
"source": [
"def weights_init(m):\n",
" classname = m.__class__.__name__\n",
" if classname.find('Linear') != -1:\n",
" m.weight.data.normal_(0.0, 0.02)"
]
},
{
"cell_type": "code",
"execution_count": 10,
"id": "f1e31148-6d05-4582-b184-2999219a783b",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"Sequential(\n",
" (0): Linear(in_features=2, out_features=40, bias=True)\n",
" (1): ReLU()\n",
" (2): Linear(in_features=40, out_features=20, bias=True)\n",
" (3): ReLU()\n",
" (4): Linear(in_features=20, out_features=2, bias=True)\n",
" (5): Sigmoid()\n",
")"
]
},
"execution_count": 10,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"model = nn.Sequential(\n",
" nn.Linear(n_input_dim, 40),\n",
" nn.ReLU(),\n",
" nn.Linear(40, 20),\n",
" nn.ReLU(),\n",
" nn.Linear(20, 2), \n",
" nn.Sigmoid()\n",
")\n",
"model = model.float()\n",
"model.apply(weights_init)"
]
},
{
"cell_type": "code",
"execution_count": 11,
"id": "59274e62-85d7-4144-b51e-7cb8b43bb4c8",
"metadata": {},
"outputs": [],
"source": [
"## TESTING. SOMETIMES THE MODEL HAS FLOAT/DOUBLE ISSUES\n",
"_ = model(X_train)\n",
"## IF THIS CELL RUNS FINE, THE REST OF THE PROGRAM WILL BE FINE"
]
},
{
"cell_type": "markdown",
"id": "88fd9410-fd3c-4a37-b17f-3c12e44051fa",
"metadata": {},
"source": [
"OpenAI's blog mentions that we have an initial set of parameters, and the ES algorithm optimizes this set. Our model has a total of 982 parameters, a small model."
]
},
{
"cell_type": "code",
"execution_count": 12,
"id": "cfbff143-4b04-426e-9126-eb8d1d24a3f0",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"torch.Size([982])"
]
},
"execution_count": 12,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"mother_params = model.parameters()\n",
"mother_vector = nn.utils.parameters_to_vector(mother_params)\n",
"mother_vector.shape # torch.Size([982])"
]
},
{
"cell_type": "markdown",
"id": "20052687-208c-4304-b975-22fbdd760e1c",
"metadata": {},
"source": [
"As with other evolution strategies, we are trying to maximize the loss, so we take the reciprocal of `CrossEntropyLoss`, so that a higher loss means that the model is improving"
]
},
{
"cell_type": "code",
"execution_count": 13,
"id": "4eb38437-def5-45eb-8fd7-7a056bd91011",
"metadata": {},
"outputs": [],
"source": [
"loss_func = nn.CrossEntropyLoss()\n",
"def loss(y_pred, y_true):\n",
" return 1/loss_func(y_pred, y_true)"
]
},
{
"cell_type": "markdown",
"id": "2251cb75-1dfa-4a9c-bf76-8226cc67bcf5",
"metadata": {},
"source": [
"We implement our fitness function, which calculates the fitness of a given set of parameter"
]
},
{
"cell_type": "code",
"execution_count": 14,
"id": "85cd162f-a805-48ff-8d5c-fb226eb95581",
"metadata": {},
"outputs": [],
"source": [
"def fitness_func(solution):\n",
" nn.utils.vector_to_parameters(solution, model.parameters())\n",
" return loss(model(X_train), y_train) + 0.00000001"
]
},
{
"cell_type": "markdown",
"id": "458355a6-2842-4c21-b9c6-53f36505a758",
"metadata": {},
"source": [
"In ES, our \"population\" is just slightly different versions of the `mother_vector`. That's the purpose of the jitter function. It slightly alters the mother vector."
]
},
{
"cell_type": "code",
"execution_count": 15,
"id": "4735c389-9b23-4b8a-9c2a-ea840c932173",
"metadata": {},
"outputs": [],
"source": [
"def jitter(mother_params, state_dict):\n",
" params_try = mother_params + SIGMA*state_dict\n",
" return params_try"
]
},
{
"cell_type": "markdown",
"id": "38ac303e-9818-4b46-8009-341c6fd97061",
"metadata": {},
"source": [
"This takes a population of randomly initialized parameters, and jitters them slightly. It then calculates the fitness of the jittered population and returns that."
]
},
{
"cell_type": "code",
"execution_count": 16,
"id": "ce1e7fa0-435f-4298-bf63-03bbe0733715",
"metadata": {},
"outputs": [],
"source": [
"def calc_population_fitness(pop, mother_vector):\n",
" fitness = torch.zeros(pop.shape[0])\n",
" for i, params in enumerate(pop):\n",
" p_try = jitter(mother_vector, params)\n",
" fitness[i] = fitness_func(p_try)\n",
" return fitness"
]
},
{
"cell_type": "markdown",
"id": "740649f8-4c5d-4438-aecb-75aae3494142",
"metadata": {},
"source": [
"A utility function to return the test accuracy of our model"
]
},
{
"cell_type": "code",
"execution_count": 17,
"id": "005c29ec-9c87-4f9e-b202-5970164eb741",
"metadata": {},
"outputs": [],
"source": [
"def test(mother_params):\n",
" nn.utils.vector_to_parameters(mother_params, model.parameters())\n",
" return (((torch.max(model(X_test), 1)[1] == y_test).sum())/len(y_test)).item()"
]
},
{
"cell_type": "markdown",
"id": "629d3c9d-b3f3-4baf-b378-99e86390e6ee",
"metadata": {},
"source": [
"### Hyperparameters"
]
},
{
"cell_type": "code",
"execution_count": 18,
"id": "5ca96f06-7658-4b3a-95e3-ad512b53ef83",
"metadata": {},
"outputs": [],
"source": [
"SIGMA = 0.01\n",
"LR = 0.001\n",
"POPULATION_SIZE=50\n",
"ITERATIONS = 500"
]
},
{
"cell_type": "code",
"execution_count": 19,
"id": "cb7ae4b8-5ab7-4edb-831e-fe36b134c15b",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Number of parameters: 982\n"
]
}
],
"source": [
"n_params = nn.utils.parameters_to_vector(model.parameters()).shape[0]\n",
"print(f\"Number of parameters: {n_params}\") # Number of parameters: 982"
]
},
{
"cell_type": "markdown",
"id": "2f849c42-d134-41cc-b7dc-fb070d25b3bc",
"metadata": {},
"source": [
"## Training"
]
},
{
"cell_type": "code",
"execution_count": 20,
"id": "886ba6b5-5b79-4cc5-b9f6-c91bc3102e4a",
"metadata": {},
"outputs": [],
"source": [
"from tqdm.notebook import tqdm # You can get rid of this code."
]
},
{
"cell_type": "markdown",
"id": "0cdff4ee",
"metadata": {},
"source": [
"Training Code is very simple. We disable autograd, for faster training. \n",
"We run for the specified number of iteration. In each iteration, we generate a random population, calculate the fitness. Then, we normalize the fitness so that we can update the mother vector based on the fitness of each population. We update the mother vector with the entire population, with the highest fitness having the greatest preference."
]
},
{
"cell_type": "code",
"execution_count": 21,
"id": "17e06b20-9c21-4a65-940e-0fca4e9b8ed3",
"metadata": {},
"outputs": [
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "664f58978695472e9b3fd582155fa47f",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
" 0%| | 0/500 [00:00<?, ?it/s]"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Iteration: 0, Reward: 1.4428613185882568, Accuracy: 48.88888895511627\n",
"Iteration: 50, Reward: 2.35758376121521, Accuracy: 84.44444537162781\n",
"Iteration: 100, Reward: 2.4251275062561035, Accuracy: 87.77777552604675\n",
"Iteration: 150, Reward: 2.4666550159454346, Accuracy: 88.33333253860474\n",
"Iteration: 200, Reward: 2.5329749584198, Accuracy: 90.55555462837219\n",
"Iteration: 250, Reward: 2.6997175216674805, Accuracy: 93.33333373069763\n",
"Iteration: 300, Reward: 2.810124397277832, Accuracy: 96.66666388511658\n",
"Iteration: 350, Reward: 2.8788623809814453, Accuracy: 97.22222089767456\n",
"Iteration: 400, Reward: 2.9336156845092773, Accuracy: 98.33333492279053\n",
"Iteration: 450, Reward: 2.9613733291625977, Accuracy: 97.22222089767456\n"
]
}
],
"source": [
"with torch.no_grad(): # We do not need autograd, it makes the program faster and fixes a few autograd related errors\n",
" for iteration in tqdm(range(ITERATIONS)):\n",
" pop = torch.from_numpy(np.random.randn(POPULATION_SIZE, n_params)).float()\n",
" fitness = calc_population_fitness(pop, mother_vector)\n",
" normalized_fitness = (fitness - torch.mean(fitness)) / torch.std(fitness)\n",
" mother_vector = mother_vector + (LR / (POPULATION_SIZE * SIGMA)) * torch.from_numpy(np.dot(pop.t().numpy(), normalized_fitness.numpy()))\n",
" if iteration % 50 == 0:\n",
" reward = fitness_func(mother_vector)\n",
" acc = test(mother_vector)\n",
" print(f\"Iteration: {iteration}, Reward: {reward}, Accuracy: {acc * 100}\")"
]
},
{
"cell_type": "code",
"execution_count": 22,
"id": "cb962be7-da75-4c97-ad56-1f4beda63d30",
"metadata": {},
"outputs": [],
"source": [
"preds = model(X_test)\n",
"_, preds_classes = torch.max(preds, 1)"
]
},
{
"cell_type": "code",
"execution_count": 23,
"id": "8a47f56b-965c-4249-837f-97a8f1da8293",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"<matplotlib.collections.PathCollection at 0x12faee490>"
]
},
"execution_count": 23,
"metadata": {},
"output_type": "execute_result"
},
{
"data": {
"image/png": "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",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"# This graph should look very similar to the original dataset. If it does, then our model has learned the data.\n",
"plt.scatter(X_test[:,0], X_test[:,1], c=preds_classes.detach().numpy())"
]
},
{
"cell_type": "code",
"execution_count": 24,
"id": "05fe8e4d-e6c8-4c47-9e92-576794827ad5",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"X_train Accuracy: 97.81249761581421%\n",
"X_test Accuracy: 97.22222089767456%\n"
]
}
],
"source": [
"print(f\"X_train Accuracy: {(((torch.max(model(X_train), 1)[1] == y_train).sum())/len(y_train)).item() * 100}%\")\n",
"print(f\"X_test Accuracy: {(((preds_classes == y_test).sum())/len(y_test)).item() * 100}%\")"
]
},
{
"cell_type": "markdown",
"id": "0fee437d",
"metadata": {},
"source": [
"The evolution strategy clearly works, our model got 97% test accuracy."
]
},
{
"cell_type": "markdown",
"id": "efde5b13",
"metadata": {},
"source": [
"## Decision Boundary Function\n",
"\n",
"Utility function to plot the decision boundary of the model. The model has a very accurate decision boundary, consistent with the shape of the dataset"
]
},
{
"cell_type": "code",
"execution_count": 25,
"id": "c532f921-8555-4b3b-a043-fcb97b8965cc",
"metadata": {},
"outputs": [],
"source": [
"def boundary(X, ax): # determine boundary between different colored dots\n",
" x_min, x_max = X[:, 0].min()-0.1, X[:, 0].max()+0.1\n",
" y_min, y_max = X[:, 1].min()-0.1, X[:, 1].max()+0.1\n",
" spacing = min(x_max - x_min, y_max - y_min) / 100\n",
" XX, YY = np.meshgrid(np.arange(x_min, x_max, spacing),np.arange(y_min, y_max, spacing))\n",
" data = np.hstack((XX.ravel().reshape(-1,1),YY.ravel().reshape(-1,1)))\n",
" data_t = torch.FloatTensor(data)\n",
" db_prob = model(data_t)\n",
" _, clf = torch.max(db_prob, 1)\n",
" Z = clf.reshape(XX.shape)\n",
" return(ax.contourf(XX, YY, Z, cmap=plt.cm.Accent, alpha=0.6))"
]
},
{
"cell_type": "code",
"execution_count": 26,
"id": "058d63ca-f9fc-40b6-bdd1-e69837142de6",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"<matplotlib.collections.PathCollection at 0x12fb48fa0>"
]
},
"execution_count": 26,
"metadata": {},
"output_type": "execute_result"
},
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAzoAAAFlCAYAAAAj9p2/AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/MnkTPAAAACXBIWXMAAAsTAAALEwEAmpwYAACbIklEQVR4nOzdd3xcZ5X4/88tM6NR792yZFmWe49bmpM4Tu8QEiC0AMsCy9J290u2/r7LsrCFLfDdwgIJsEAoCSmkOU5x7703Wb33NtLMLb8/xpIlz8hNI03Reb9eEPvemXuPLGnuPfc8z3kU27YRQgghhBBCiFiihjsAIYQQQgghhAg1SXSEEEIIIYQQMUcSHSGEEEIIIUTMkURHCCGEEEIIEXMk0RFCCCGEEELEHEl0hBBCCCGEEDFHD8VBFEX5MXA/0Gzb9vwg+9cCLwPnL2x60bbt/3u5YyanJ9s5BTmhCE8IIYQQQggRo84ePdtq23bWpdtDkugAzwHfB356mddssW37/qs9YE5BDt995bvjjUsIIYQQQggRwx6c8WBVsO0hGbpm2/ZmoD0UxxJCCCGEEEKI8ZrMOTqrFUU5pCjKG4qizJvE8wohhBBCCCGmmFANXbuS/cB027Z7FUW5F3gJKLv0RYqifBb4LEBWfsAwOyGEEEIIIYS4KpNS0bFtu9u27d4Lf34dcCiKkhnkdT+wbXu5bdvLU9JTJiM0IYQQQgghRAyalERHUZRcRVGUC39eceG8bZNxbiGEEEIIIcTUE6r20r8E1gKZiqLUAn8NOABs2/4v4APAHyqKYgAe4Anbtu1QnFsIIaKJbducfbOao8+fxhw0mf1IKXMeLUVzyLJmQgghRCiFJNGxbfvJK+z/Pv7200IIMaW9++c7OPVKJYbHAKD5aDunXqngkZ/eiapJsiOEEEKEilxVhRBikrSf7eTky+eHkxwAw2PQfLiNqk31YYxMCCGEiD2S6AghxCSp290UdLuv36B6S90kRyOEEELENkl0hBBikrhSXKiaErBddai4M9xhiEgIIYSIXZLoCCHEJJlxR2HQeTiqpjDn0dIwRCSEEELELkl0hBBikuhxOo/89E4Sst04EnSciQ6cSQ7u/f6tJOUnhDs8IYQQIqaEpOuaEEKIq5O9IINPbfsAzUfbML0WOYsypbW0EEIIMQEk0RFCiEmmqAo5CzPDHYYQQggR0+QxohBCCCGEECLmSKIjhBBCCCGEiDmS6AghhBBCCCFijiQ6QgghhBBCiJgjiY4QQgghhBAi5kiiI4QQQgghhIg5kugIIYQQQgghYo4kOkIIIYQQQoiYI4mOEEIIIYQQIuZIoiOEEEIIIYSIOZLoCCGEEEIIIWKOJDpCCCGEEEKImCOJjhBCCCGEECLmSKIjhBBCCCGEiDmS6AghhBBCCCFijiQ6QgghhBBCiJgjiY4QQgghhBAi5kiiI4QQQgghhIg5kugIIYQQQgghYo4kOkIIIYQQQoiYI4mOEEIIIYQQIuZIoiOEEEIIIYSIOZLoCCGEEEIIIWKOJDpCCCGEEEKImCOJjhBCCCGEECLmSKIjhBBCCCGEiDmS6AghhBBCCCFijiQ6QgghhBBCiJgjiY4QQgghhBAi5kiiI4QQQgghhIg5kugIIYQQQgghYo4kOkIIIYQQQoiYo4c7ACGEuB7nNlSz818P0lPXR0Z5Kmv+ZCkFN+SEOywhhBBCRAip6Aghos6JF8/y1le20HaqE2+vj4Z9Lbz8iY3U7W4Kd2hCCCGEiBCS6Aghoopt22z7h/0YA+ao7caAybbv7LuqY7Sd6mDbP+5n8zf3UL+3Cdu2JyJUIYQQQoSRDF0TQkQVb6+PgY7BoPvaTnde8f0Hnj3Ojn86gOmzsC2bo8+fYfZDJdz2zVUoihLiaIUQQggRLlLREUJEFUe8jhYX/BlNYl78Zd/b29TP9n88gDFgYps22GB4DE6+fJ6Gfc0TEa4QQgghwkQSHSFEVFE1laWfnovu1kZt190aK/948WXfW7WpDiXIp54xYHD2zeoQRimEEEKIcJOha0KIqLPiCwuxLTjwo2OYXgtnooM1X1vCrPuKL/s+VVeDDk9TVAXNKc99hBBCiFgiiY4QIuooqsKqP17Eii8swNvrw5XsRFGvPL+m5I5C3vvLnQHbNYdK+UMzJiJUIYQQQoSJPMIUQkQtVVeJS3VdVZIDEJfi4q5/vRk9TsMRr6O7NTSXxqqvLiGzPG2CoxVCCCHEZJKKjhBiSim9s4hPbf8A59+pxfRZFN9aQGLu5ZsYCCGEECL6SKIjhJhy4lJczHm0NNxhCCGEEGICydA1IYQQQgghRMyRREcIIYQQQggRcyTREUIIIYQQQsQcSXSEEEIIIYQQMUcSHSGEEEIIIUTMkURHCCGEEEIIEXOkvbQQE6ivxUPl+7WoukrJ7YXEpbjCHZIQQgghxJQgiY4QE+TQz06y9e/3oqgqigLv/cVO1n/3JmbeNT3coQkhhBBCxDxJdISYAB0VXWz79j7MQQuwhrdv+OpWCrbm4E6LC19wIubYtk3j/hbq9zUTn+Vm5l1FOOId4Q5LCCGECCtJdISYAKdeOY9pWIE7VIWKjTXM+2DZ5AclYpLps/j9H7xL3e5mLJ+J5tTY/Ld7ePTn68makx7u8IQQQoiwkWYEQkwA02tim3bgDsvG9AZJgETU66rpYfs/7eetr27hxIvnMAbNSTnv0edPU7e7CcNjYBk2vn6DwS4vr3/+fWw7yM+gEEIIMUVIRSfGWKZF44FWjEGDvKXZONzyLQ6H0vVFHPrpSQzP6Jtd24aS2wrCFFV06Gvx0LC/GXd6HPnLslFUJdwhXVHV5jpe+/z7WIaN5bOoeLuGfT84xuMv3IMzYWKHkB379ZmAnzPw/zt2nu8mbUbKhJ5fCCGEiFRyFxxDmo+18crT7+DrN1AUBcuyueNbqyl/oCTcoU05uYuzmPvYTI6/cA5jwEBRQHNqrPjSIpLyE8MdXkSybZvt/3SAgz8+jurUwLaJS3XxyM/Wkzo9KdzhjckyLTZ8fduoZMPXb9BV3cPBZ0+w4osLJziAsXYo2JZUdIQQQkxdIUl0FEX5MXA/0Gzb9vwg+xXg34B7gX7gE7Zt7w/FuYWfMWjy0sc2MtA5OGr7O/9nO9nz0uWpbhjc+jcrmPVACWfeqEJzqJQ/VDLl5kz0tw3g7faSXJSIqgWOlDW9JuferqHzfDfGgMHB505geq3h4X2Gx+DVz7zLR996EP/HSORpP9uF4TECtpuDJqdfq5zwRGf2ozPY+d2DGAOjqzpxaS7SSuX3XgghxNQVqorOc8D3gZ+Osf8eoOzC/1YC/3nhvyJEqrfWYxqBw1dMw+LYb85y058tC0NUU5uiKOQvzyZ/eXa4Q5l0A12DvPXlLdTubETRVPQ4jdv+diVl9xQPv6a3sZ9ff+B1Bru9+DwGSpAKhG1BT30v7We7yChLndwv4irpcfqYlZPJGDq68KOzqdhQQ8vxdnz9BnqchqKp3Pu9WyI2ORRCCCEmQ0iuwrZtb1YUpfgyL3kI+Kntnxm7U1GUVEVR8mzbbgjF+QUMdnmDDmGxDRtP28DkBySmtN9/9j0aD7Vi+fzttQ2Pwdtf30ZyQSI5CzMBePfPd9DX7Blu2mATPFlQNRVvr2+yQr9mqdOTSClKov1sJ/aI30HdrbPwqfIJP7/u0njsl3dRva2B+r1NJGTHM+v+YlmcVgghxJQ3WV3XCoCaEX+vvbBNhEjhqhysIF2+HPE6JXcUhiEiMVV1nu+m+WjbhSTnImPQZP8PjwP+eS1VW+qDd6YLImtuZA/5u++/biMhOx5Hgo4jXkdzacx6oJjZD8+YlPMrqsL0m/NZ/ZUlLPxIeUQmOd5eH0d/dYYt39rLyZcrJq0rnRBCiKkropoRKIryWeCzAFn5WWGOJrok5Sey+BNzLnT68s8X0N0aWfPSmXHHtDBHJ6aS3ub+4LUZG1pPdVz1cRRNQXOo3PZ3q9BdWsjimwip05P4xOZHqd3RSF+Lh7wlWaQWJ4c7rIjRWdXDbz7wOj6PieExcMTr7PjnA3zoxXuJz3SHOzwhhBAxarISnTpg5N124YVto9i2/QPgBwBlC8qkXdA1uvFPl1KwModjz5/G22dQ/mAJ5Q+WoOqyXJKYPBnlqZgDwZ/We3u8gH84WtFNeVRvqR813EvVFbIXZODOcJOQ7WbhR8vJLE+bjLDHzf815Yc7jIj0zjPbGegcHP5e+/oNTK/J1u/sY/0/3hTe4IQQQsSsyUp0XgG+qCjK8/ibEHTJ/JyJUXxrAcW3yqhAET6qpvoHxQaZMzZ4IdEBuP3vVvPrx97A2+vF1+9/yp+Q5eaB/7kdd1rc5AUsJpTpNanf0zwqoQWwDJtzb9XAP4YnLiGEELEvVO2lfwmsBTIVRakF/hpwANi2/V/A6/hbS5/F3176k6E4rxAi8jjidfQ4DaM/sKqTlJcw6s8ff+8RKt6upuN8N5mz0ii5o1AqkLFGUWCM5m+qJl3hhBBCTJxQdV178gr7beALoTiXECKyqZrK0qfnsf+Hx0Ytoqm7NVb+8eJRr9VdGrPulwVtY5nmUCm+tYDKTXXYxsURyarTv7aUEEIIMVHk0akQIuRWfmkRS5+e6+9A5lSJS3Vx858vZ9Z9xeEOTYTB7X+3mqT8BBwJ/p8HR7xORlkqq7+25KqPYds2/mdmQgghxNWJqK5rQojoY9s23l4fjnjdPz8Hf7vjVV9Zwg1fXIS314sr2Tm8T0w9CVluPvb2w1Rtqaez0j9MsXBN7lUtaNpZ2c37f72Lmh2NqLpK+QPF3PwXN+BKck5C5EIIIaKZJDpChEBfcz97//soNdsaSMyJZ+ln51F0Y+x34Dr5UgVb/34vA51eNKfK4k/OYdWXF6Oo/htYzaFKYwEBgKqrlNw29ppeno4BarY2oDpUim7Ox5ngYKBzkF8/9jqD3V5sC0zT5NQr52k73cnjL957VYmSEEKIqUsSHSHGqbepn1/c/yreHh+Wz6L9TBf1+5q5+ZnlLPhwebjDmzDn363l3T/fgXGhlbRlWBz48XEs0+bGP1ka5uhENDn26zO8/ze7UXXF37zAsrnn+7fSfqYT34A5qmOb6bVoP9tF48FW8pbIemtCCCHGJmNJhBinff91dDjJGWJ4TLZ+e981rf5uWzaV79ex458PcPjnpxjs9l75TWG0898ODic5QwyPyaGfnMT0yqr34up0nu9m0/+3G3PQxNdn4Ov14es3eP0L79N4uG3MNZk6znVNcqRCCCGijVR0hBin6q31o5KcixQ6KrrImpN+xWMYAwYvfmQDbac78fUb6G6N7f+wn0f/dz3ZCzJCH3QIdNf0Bt1uWzaD3V5Z8V5clRMvVWAZgb8/iqKgqvhblV+S7Ng2pM9MmawQhRBCRCmp6AgxTgk58UG3W4aJO/3q5qcc+PEJWk504Os3AH9lxNvr4/UvbYrYTlOZc9KCbtfjNOLSXJMcjYgEA12DHHz2OO/++Q6O/uoMvn7fFd9jeAwsM/Bn3DIsNKeK5tJQRlypVKdK5uxUchZlhjJ0IYQQMUgSHSHGaeln5qG7tVHbVIdKwQ05JI6RBF3q5EvnMIMMc+tv9tBVHbxyEm5rvr4UPW701627NVZ9ZbF0WJuC2s918ZPbfsf2fzrA0efPsPmbe/jpupfpa+6/7Ptm3DkNPS5wcIHptTj7RjU+j0FCdjyqrqC7deY+VsrDP7lTGhEIIYS4IrkbEWKcim8t4MY/XYbu1nEmOtBcGgU3ZHP3v99y1ccY6lIWTKSuHp+3JItHfnonecuycCTopJUms+7v17DoqdnhDk2EwTvf2M5gt3d4mJnRb9Df6mHrt/dd9n35y7OZeXcRenxgsuPrN7C8FgNdg9z8Fzfw+aMf5vZvrsaZ4LjsMZuPtfHml7fw/MOvseVbe+htunyyJYQQIjYpkTospmxBmf3dV74b7jCEuGrGgEH72S7iM90k5l5dJWfI/h8dY+d3L5ncr0DajBSe2vBQiCMVIrSMQZP/nP8LbCvweuJIcPCHh5+87Ptt26ZmWwMnfneOUy+fhyCXpdSSZD628eErxnL+vVre+OImTK+/W5vq8C9Q+sTL95EyLelqvyQhhBBR5MEZD+6zbXv5pduloiNEiOhxOtnzM645yQFY9NRs8pZl+xfd1BUcCTpxKS7u/f6tl32fbfsn/gebzC3EZFEURs2jGUlzXPkyoygKRTflc9OfLUNzBn+9t/fK831s2x5ueT7UktryWQz2eNnx3YNXfL8QQojYIl3XhIgAmlPj4Z+so2FfCw37m0nMjad0fVHQuQtDTrx4lq3f2c9glxfNEbhY51Rk2zZNh1oZ6Bgkd0kWcanSFGEyaE6N4tsKqXyvFsuwR2xXmf3ojKs+TnyWm/hMNz11faO2K7pCye0FV3x/f4uHgc4gbdktqNnecNVxCCGEiA2S6ExRlmmB7V+tXEQGRVHIX55N/vLsK762YmMN7/3VLgzPhcU6ff7FOm3LZs3XY2exTtuysW37qpobdFb18NLH38bTPoCiKJg+i5VfWsTyz82fhEjF7X+3mheefJPexn5s0wZFIWtuGqu/uuSqj6EoCuu+cyOvfuZdTJ+JbdhoLg1XkoOVf7z4iu93JDjG7FIYlyJJrxBCTDWS6EwxA52DvPdXuzi3oRrbtMlblsUd31pN2gxZkyKa7Py3g8NJzhDDY3LwJydZ+aVFaE5tjHdGB0/7AO/95S4qNlZjW1C4Opfb/27VmHMsbNvmlU+9Q09d7/CQJYDd3z9EzoIMpt2YN0mRT13xGXF89M2HqN3RSGdVD5mz08hdknnN3dGmrc7lw6/ez6GfnaSjopvCVTnMf3LWVSUqzgQHM+6Yxvl3ajC9F38QdLfGkqfnXPPXJIQQIrpJojOF2LbNCx9+i46K7uEFLuv3NvPrD7zBx999ZMoN82k708mBHx2no6KL/OXZLP7kXBKyInuRS9u26W/x0FPbF3y/aTPY4yM+I3oTHduy+e2H3qSrumd4GFTtjkZ+/ejrfPy9R3EmBnbcaj3RQV9z/6gkB/zJ36GfnpREZ5IoqsK0G/PG/e+dWpLMrX+14rreu+7ba3j9i+9Tt7sZzaliei0WfLiceY+XjSsmIYQQ0UcSnSmkfk8z3TW9w0kOALa/Y9LxF86y9Ol54QtuktVsa+DVP3gX02thmzZNR9o4+vyZiOzMZAwY7P5/Rzj6y9MMdg36G1KN0XtAd2vEpTonM7yQq97WQG9T/6i5HrZl4/OYnHq1ggVPlge8Z7DbO+bcpIHOwVF/9/b6qN5aD8C0G/NwJUX3v5cYzZno4OHn7qS7tpee+j4yZqVOuYc4Qggh/CTRiSK2bXP8N2fZ/8PjDHQOMm1NLqu/uoSUoqu7Me+s7A46ft0cMGk71RniaCOXbdu888yOUUO/LK/FoOFl+z/u555/v3yns8lk2za/+9jbNB9pGzUUJxjdrbHma0uifrHOzoquoF3kDI9B+5muoO/JWZjhn3d2CT1Oo/SuouG/n91QxYavbvUnRQrYhs2d/3wTZXdPD90XICJCcmEiyYWJIT+ubdlUvl/H2beqcCY6mPfBMjJnp4X8PEIIIcZPEp0osu07+zn8v6cwPAYAZ16vompTPR9+/QGS8hKu+P6MWalA4FNv3a2RvSAjxNFGroGOweALCFpQs61x8gO6jPq9zbSe6BgzyVFU0Fw6yQUJrPjSImbdVzy5AU6A9FlpqJqKeUnZSo/XyZqbHvQ9jngHt/zFDWz+5h7/WkS2P8lJLkxk/hP+IUt9LR42fGXr6LWKgA1f3Ur+0iwSsq+9LbiYWizT4vefe5+6nY34+g0UFY49f4abvrGchR8NrDQKIYQIr+h+9DuFDHQNcuinJ4eTHPDPx/B5DA786PhVHSNnUSZZc9NGrVOhqP4JvHMeKQ15zJFKd+uMNT/amXT5FdcnW8vRdszLrJHjiHew/p9u5KNvPRQTSQ5A4aocUouTR/+cagquJCdll/ka5z8xi0d/fhezH55B0U153PSNZXzopftwxPu/p2dfr2Ss9ZHPvFYVyi9BxKjzG2uHkxwA2wJjwGTLt/bg6RgIc3RCCCEuJYlOlGg/0xl0IT3LZ1G3p+mqjqEoCg8/t475T87ClexEd2uU3jWdD/3uvqATvGOVw60z486igH9P3a2x+JOR1ZkpqTABzTF2YwHTZ5E1L3iVI1opisKjv1jPnMdm4khwoMdpzLx7Ok+8dC8O9+WL0LmLMln/Tzfx8E/uZOFHZ496vc9jBB0SZxkWvhEPEIQYy5k3KoeTnJFUXaN2R2RVg4UQQsjQtaiRmJcQfPiSAqnTk6/6OI54B7f+1Yrr7mgUK+741mo87QM0HmhBdaiYgyazHyll0VOzwx3aKMVrC3ElOzA8BrY1uhyhxamU3T094ponhIIrycnt31zF7d9cdU3vM70mLcfbcbh10meljmptXHxbIbu/dxjDHD10TXOoFN925cUohXDE6ygqAd39FMVfKRZCCBFZ5JM5SiQXJFK4MofanY2j14eI01j22anTLS1UnIkOHv3f9XSe76a7vpeMWWkR2Vpac6h88Nd3s+Hr22jY1+xPdhSFhGw3S56ey6KPRVZiFk7nNlTz9p9uw7b9wzoTctw8+D+3D68RlVmextzHyzj+m7MYA4Z/Ho9bZ+5jpWTNia2qmJgY8x4v49TL5wPmeaHAtDXSwlwIISKNMtYq0uFWtqDM/u4r3w13GBNuoHOQw/97iuot9STmJ7Dkk3PIWZgZ9LXeXh/vPLODirerQVGIS3Vx+zdXUXJ74SRHLcJhsMeLbdmywnsQHee7+eX9r46+AVUgIcvNJ7c+NtyJzrZtanc2cerlCmzbZvZDMyhcnXvNi1qKqWvvfx9l178eRNXV4ZbmD/7oDvKXZ4c5MiGEmLoenPHgPtu2l1+6XSo6YdTfNsAvH/g9A50DmIMWigoVG6q54ztrKL+/JOD1zkQH9/z7LXj7fPj6fMRnueUGbQqR9V7GdvSXpwKbNtjg7TOo3dFI0U35gH/+z7TVuUxbnRuGKEU0sC2bmh0NdJ7vJr0slYIVOaM+Z5f/wXzmPFpKzbYGHPE602/JR4+TS6kQQkQi+XQOo30/OIqnfWB4Ac+hDj7v/+UuZt41Hc0RvFeEM8GBM2HqNA8IJ8u06K7tw5XkwJ0eB/hvhI78/BQHnzvBYI+PopvzWfO1xSTlh37NjlDqb/Vg+iwSc+NjLkHua/ZgG8Gq0zaedumGJa6Op2OAF554i576PizTRtUUUkuSefTn60c9aEjIcjP74RlhjFQIIcTVkEQnjCrfrR1OckayTIuOii4yy2URunA693Y17z6zA5/HwDZtClbmcte/3MTOfznIiRfPDS84evqVCqo21fHRNx8kPjPy5vl01fTwxpc203ayAxSFxLx47v6Xm8ccIhmNpt9awPl3agM6YlmGLUOKxFV7/29201nVM/y5bOLveLntO/u4/ZurwxucEEKIaybtpcMoLi34XAvLsHElyzClcGo53s5bX96Cp30Qw2Niei1qdzby0ic2+iezey7OBbEt8PX5OPSzU2GMODjTZ/HbD71Fy9E2TK+FOWjSVdnDix99m/622Kl0lN1bTEpxEnrcxVbcultn/pNlEV9pE5HBtmzOvVkd8PDJ9FqceuV8mKISQggxHlLRCaMlT8+l5fjWUTfNqq6QszCDpLyEMEYmDvzoeEA7b8tn0XbKv57RpftMr0Xd7shbR6Nqcx3eXm9AO1zLsDj50jmWPh0bHft0l8YHf30PR35xijOvVeJIcLDwo+WUri8Kd2giilzawn2IFXRY5BjHsG06znVheEwy56Sh6vI8UQghQsVVc22jUSTRCaPS9UUs/fQ89v3gGJpTxTIs0ktTuPf7t4Y7tCmvu7Y36E2P5lCDDjdUNIW0kqtfz2iy9Db0Bb1JMwdNuqt7wxDRxHG4dZY+PS9mkjcxuRRVoXB1LrU7Gkf97iuactXrLHWc7+b3n32XnoZ+FNX/ebH+n26i+DbpjCmEEOO1qLqNH+zz4FSvfp66JDphpCgKq768mMWfmEPLsXbis9xkzEoNd1gCKFiVS9ORVszBS6o6hkXW3HSaj7djjajqaE6VxZ+cO9lhXlHOwkyUIA+UHfG6zF0R4hK3f3MVv37sdXweA8NjosfrOBMd3PqXV15g2TIsXvjwW/S3eOBCnuQDXv/iJj7yxoOkFMXewr5CCDEZXDWZzLZPUf1MN+szfwWaJDpRJS7VxbQbZbG5SLL447M5+svTDJqDwxUR3a2z8KPl3PD5Bbz9Z9uofK8ORYH4LDd3/P1qMspSwxt0EDkLMylYmUvdzsbhNWY0p0pSfgKld8mwLiFGSilK4mPvPsKpV87TdqqDrLnplD9YgiP+yhfV6m0N+PqM4SRniGXaHPv1GdZ8fekERS2EELFrqIoT/0I3O0p3E+fMQckNMvf2d8HfL4mOEEG40+P48O/vZ9f3DlO1qY64FBdLnp5L+UMlKIrC/f95G95eHz6PQXxmXES3a77/v27j4HPHOfars5g+k1n3FbP8DxegObUrv1mISdZwoIU9/+8wHRXd5CzMZMUXF5A+M3XSzu9KcrLwI+XX/D5P2wDBFuC2fBa9Df2hCE0IIaaMS6s4O0ozcBelkjp71Rjv+H9Bt0qiI8QYErLjuf1vV2HbNpXv1XHsV2c4+VIF5Q+VUP5ACc5EB87EyF/PSHOoLPvMfJZ9Zn64QxHisirfq+X1L24arj521/Ry/p0aPvCru8mamx7m6C4vb1k2thmY6DjidYpuyQ9DREIIEZ2CVXHcuYmXSXLGJomOEFew+W/3cOzXZzE8/jVa6vc2c/KlCh5+dh2KGrmVHCGiiW3bvPfXu4aTHPB3QfP1G2z5+708+rP1YYzuylKnJzHnsVJO/q5i+LNCj9NIKU6i7J7pYY5OCCEi37VXca5MEh0hLqPzfDdHnz+DOXjx5svwGDTub6Fqcz3Fay/fjanzfDe9zf1kzk4jLiX4ukmxzBgw2P29wxx/4RyWaTHz7ums/upi3Glx4Q5NRBhfn0FfU/AhXk0HWyc5mutz2/9dSeGKHA7/7yl8HoNZDxSz8KOzZZioEEKMsKi6Lej2UFVxRpJER4jLqNnRGLRrma/foPK92uFEZ6BzkNZTHSTmxJNanMxA5yCvfuZdWo63ozpUTK/Jss/MY+WXF0f0fJ5Qsm2blz6+cVT3uuO/PkvN1no+8uZD6C65+YtWtm3TsK+Z2p1NuNPjKLtv+rgTeT1OQ9VVLMMM2OdOn5jE2NfvwzJtXEnjX6DZ0zFAb2M/xbcVMuuBkhBEJ4QQsWVkxUbVAm+u1qeFpoozkiQ6QlyGK8UZdHia6lCJS3dh2zbb/+kAB5894V9I1GeRsyADVVNoOtLmX3PnwlCc/T86QfqsNGbdVzzJX0V4NO5voeV4+6gW3ZZh0d86wNk3qpj98IwwRieul2VY/P4P36d2ZyPGgIHu0tj67b089ON142pZruoq854o49jzZ0YNX9PdGsv+ILRrI/U29fP2n2yjbpd/kd/MOenc+Q83Xld7f9Nr8s6f7+DM7yvRnBqmz2LxJ+ew5utLpsxDDSGEuJLAeTfBUpDQVHFGkkRHiMsoub0weKKjKcx5tJRTr5zn0E9OYA6aw8PbGg+0YplWQJtZw2Nw4EfHpkyi03ysHSvI5Gxfv0HTodaYTHQs06LjXDfORJ2k/CDtL2PAyZcqqN3RgOHx/7wP/ff1L7zP0zs+OK55azf92TK8vT5Ov1qJ6lCxDYvFn5zL/CdnhSR28H+PfvuhN+mp7xtuHtB8tI3fPvEmH3/vkWuuTG399j7Ovl6F6bUwL6ytdegnJ0jOT2DBdXRvE0KIWDIR826uhSQ6QlyGw63z8E/u5NVPv4MxaKIoCrZtc+c/3kjKtCRe+8P3h2/0hliGNcbRwNM+ONEhR4zkaYmouoJ5yZesu/0TtGPN+fdqeftPtmEOmlimTcasVO77j7Uk5SeEO7SQOvabMwE/8wA+j0HLsXayF2Rc97E1p8ad37mRm7+xnN6mfpILE3EmhLazYfWWBjztA6M7pNlgDpqcfKmCxR+fc9XHsgyLY78aXYECf/K37wfHJNERQkxpoeyedr0k0RHiCnIXZfL0zg/SeKAFc9Aib3n28PySgc6rT1xUXWH6rZdvXhBLpt+cT1yqC2PAvHhTqYDm0JjzSGl4gwux9nNdvDGiLTJAy/F2fvfUBp7a+LAMYbpGcaku4lInpnlHV03P8CLAIxkDJp0V3dd0LGPAxPQFf7Dh6Ri4rviEECLahbuKM1KQadZCiEupmkr+8hym3Zg3ahL99FvyUfTAm1hnogMtToULuzSniivFxYovLJiskMNO1VU++Ot7KFyZg6orqLpK9vwMPvCru3Elj3/ydyQ58r+nAm54bdOmr9lD4/6WMEU1MeZ9sAzdHdhIwuHWyZoX2WvdAGTPTUfVAn9nHfE6OYszr+lYjgSdpLzgFbvcxVnXFZ/wV8rObahm338f5fx7tf6hwEKIqLCouo3Ne2uofsZfxTmUlRO2JAekoiPEuKz80mIq3q7B2+vD9Foo6oXhN/94I4m58Rz40XG6a3spXJPH4k/MIT5jarVVTsyN55Gfrcfb58O2QtPdKhKNnO8xigp9LZ7JD2gCzX5kBuc2VFO7w9+MQHPpKCrc9x9ro2JdqdylWWTNTR/VDVB1qLgz4ii7t/iajqUoCmv/ZoV/kdNBE2xQNAXdpXHTN5ZNQPSxr6/Fw28++Aae9gGMARM9TiMxN4EP/vruCavyCSGuzVDFJphIqOKMpNh2kItzBChbUGZ/95XvhjsMIa6ov9XDwedOULujkeRpSSx9eu645imI6HPoZyfZ9p19AXNXNKfKx959ZMyn/tHKtm0a97dQu6sJd7qLsnuLo6pKd3F9p7NYPpuZ91xY3+k621g3Hmxhz38coaOim5yFGdzwhYWkl6aEOOqp4fefe4/z79VijxheqOoq5Q+VcOc/3BjGyIQQMDTvJp67X2gM2iJ6W/FO4pw6yiTPxXnqqaf22ba9/NLtkugIIcQ4+fp9/OK+39PT2Id1ofOW7taZ+8GZrP3rFWGOTojoYFs235/9v0Gro7pb5/NHPxyGqIQQMHrejZX5BrtSMoK2iJ7sBGfIWImODF0TQohxcsQ7eOLl+9j/w2OcfbMKZ6KDRR+fQ/mDsnCkECERoQ9lhZgKIqF72vWSREcIIULAlexk9VeXsPqrS8IdihBRSVEVim7Op3pzHfaI/gOqrjBjfVH4AhNiioqk7mnXS7quCSGEECIi3P63q4jPcuOI9z+HdSToJOTEc8ufB4xIEUJMoEjrnna9pKIjhBBCiIiQlJ/Ax999hLNvVtNxrouM8lRK1xehOQNbmgshxsdVE7ylfrRXcUaSREcIIYQQEUOP05n98IxwhyFETBuad+NUHQH74n8TfXNxxiKJjhBCCBEDbMum8v06f0OMBAdzPlBK9jxpdS/EVKHjIV5pp99OwyA+6GuG5t1UfqOT9Vm/Ai0w0dlRmhzVVZyRJNEREcO2bSrfq+PIL09jDhjMfriUWQ+WoDlkKpkQQlyObdm89vn3qdnWgK/fABWO/foMa76+hMWfnBvu8IQQE8pijv4K0/UdWGiomNSaN3DM9wg2F4d9juyetqt0N3GuHJTcxICjuSEmkhyQREdEkM3f3MOxX53F8BgANBxo5cRL53j4uXVBF6Uaj+YjbWz7x/00H2kjISeeFX+0kFn3FYf0HCJSWKSplTjop8MqwUdsLd4pBEDl+3UXkxwAC4wBk23/sJ9ZD84gPuP6FkMVQkS+Gfr7FOk70RQDDf9nQIG2F6+dwGnjnoAqjn/eTXrMJDOXI4mOiAidld0c/eUZzMGLK8sbHoOmg61Ubaqn5PbCkJ2r+Vgbv33yzeFV7Ae7vWz8s214Wj0s+vickJ1HhF+C0sIK53/hVDzYKKgYnDbWU2HcEe7QxFWwbZtzb1Vz7FdnsEyb2Y/MoPyBElRdqryXOvN65cUkZwTVoVKzrZ7yB2XOixCxaoa+CV3xjdqmKz6K9a24K1YEVHGifd7NtZBER0SE2p2NKEHuXXz9Buffqw1porPjuwcxBsxR2wyPyY5/Ocj8D5fLULmYYXOD8wfEKV2oysXFBsv0jXRZRbRZZWGMTVyNd/7Pdk6/XoVx4Qa+YX8LZ16r5IH/uR1FUcIcXeSwbRufJzDJAVAU/+T+IT6PwcnfneP8u7UkZLtZ8JFymccjRJRz0B90u84Alc90jOieNjWqOCNJonOVmo+00XKineRpSRSuzEFR5SIbSq4U15j/pt4+X9Dt16vlaBsEWWTbMiz6Wzwk5cvQpliQotTiUnpHJTkAGl6m61tp80qiE8laT3Vw+veVox5KGB6Dul1N1O5oZNqavDBGFzls2+adb2yn8r26MV6hMP2WfMD/WfrrR1+nu64Xw2OiqAqnXj7P7X+3WrqcCRGlXDWZ+Kbl4nLXB+zrr85g14w9U66KM5IkOldgDJq88vQ7NB5sAfwrNyfmJvDYL9YTn+kOc3Sxo+S2gjH3VbxVjbfPhzMhsDPI9UgqTKS/dSBgu21BXJorJOcQ4acrA9hB1kRWFHDgCUNE4lrUbG/AtgKfSPj6Daq21Euic0H11gZOv1Y1atjvEEeCzgM/vH24onPk56foqukdfq1t2RgDJu/91U5m3jMd3SVr1QgRyZz0oCsD9NsZuGqymW2f4gf7POT+5Ebm/OXvUBwmqmZjW2D5FH7/Rhbu6VOvijOSjNG5gt3fO0TD/hYMj4nhMfH1GXRVdbPxGzsmLYbuul5qtjfQ2xS8NBkL9DidGeumBd2naCqV79WG7Fwr/2gRetzoC7ru1pj/RBkOt+T+saLTKkIh8ObPsB00mgvDEJG4FnEprqBzcTSXijtdJtYPOfVyxfDQvpG0OI3b/241BTfkDG87+1Z10IRIURR/pVsIEZEc9LHC+d/cHvdNbnZ9lzsd/x8tjW/6F/Xc9CyblDp+/T/zOHc8nY7WOM4dT+e3P1xAm1E0pZMckIrOFR3/zdmAC4Nl2FRvrscYNCf0CZgxYPDmH2+maksDmlPF9JqU3VfMur9fM67JuMagScWGajrOd5MxK5WSO6ZFxLwUxxgVG9u2A+bUjEfx2gJu/7vVbPnWXrw9XhRVYf6Ts7jpz5aF7Bwi/ExcHPM9zDzHS6gYqIqNYTvptzOoMVeGOzxxBaXri9j0/+0O2K6oCrMfKglDRJFJ0RRQCBiOq+pqwPUpLtkZ9BiWaeFMDE3FXIipIlmpJV87ANg0mIvpsosm7Fw3OH9EslqLply4F9K9PL7yeX5xy1y6W3JIW7sOG9jTBDRdeFM5pE5YRNFDEp0rML1W0O02NpZhwQQmOpv/bi9VW+oxB63hZOvs61WkFiez4gvX90S6t7GfXz/2OoM9Xnx9Bo4EnfiMOD7423snvP2ot8/HiRfPUbujkZSiJBZ8eBYpRUnD+2feXcTJlwKfTtqGNTzGPFRmPzyD8gdLGOgcxJnoQHPKkI1YVGuuoscqYLq+DafSS5M5nzpzGRZyUxfpnIkOHvrxHfz+c+9hDFqg+Icd3v1vt5CQHXwhvKlo9sOlnHmtargt/zDLpuim0cP7Fn5sNnV7moY7TgIoKiTlJ5I+K/Wy57Ftm5Zj7ZiDJtkLMuQzU0xpM/W3KNXfu9DK2Wa6vp1K4yZOGfeH/FwJShNJav3FJOcCxWGz4q529tU+FvJzxhJJdK6gZF0hp145j22MflyWNTstZHNGgrEtmxMvnMMcHJ1oGQMmh3968roTnXf/cid9LR5s0//1+PoMegb72PJ3e7jruzePO+6xeDoGeP6h1/C0D2B4TFRd4fD/nuKBH9w2PNZ+2po8ZtxeSMU7tf6Ltgq6U2PlVxZPyI2NoioyBGYK6LKncdj3RLjDENchb1k2T+/4IE2HWrFMm9zFmXKDfYnCVTnMf6KMo788jWXaqLoCNtz977fgiB99jSq5rZCln57Hvh8cRXNo2LaNOz2OB394+S52rSc7eOXT7zDY5a+Ao8Bd3705pN0whYgWCUoLM/V30ZSLDxd0fJToW6k3l9Fjh3b+YFKLgpqvwiUffaoKaTk6hG5kf0ySROcKbvzTZdRsa2Sw24vhMdDjNFSHyrp/uHFCz2v6LCxf8GqSt/f6upDZlk3VprrhJGeIZdic21B9Xce8Wnv+4wh9zZ7hr8kybCzD4O0/2cYntz6GoigoisJd/3oz1VsbOPtGFbpbY86jpdL6dApw1WSG7dyD01rDdm5xZaqukrcsO9xhRCxFUbjlL25g3ofKqNpcj8OtM/PuojEf4qz68mIWfrSchgMtxKfHkbs067JJjjFo8uJHNjDQOThq+xt/tImPvPkgKdOSxninELEpWztGsNatCiY52jF6jOtLdMa6Dm443MaSB42ARMcwNZraZRjvlUiicwUJWW4+tvEhTr58nqaDLaSVpjD3AzMnvBKguzTSy1JoO9U5eocC+SMml4bOxLbLPrehOmjiNtA5SHdt7/DFUlEUpt+cz/SbQztUTUSmodWaf7DPg1Od/OFkXsvHZ+1+DhVJMi2iW0ZZKhllqVf12vhMN6V3Xt18gqpNdZhBPrst0+b4b8+y+itLriVMIaKebWvYQe6ZbBSsS7ORq3Cl6+Da937FnvRslt7UhMPp/100LRWfEcfp2hXX/gVMMZLoXAVHvIMFT85iwZOzJuT4tm1Ts72ByvfqcCU5mP1IKSlFSdz+t6v43cc3YnpNbNNGdahoTo2b/3z5dZ1HURVKbivg/HujqzqqQ2Xm3RM3iQ4IGEIxxLZs6XQ2RblqMtm8t4b4F7pZn/kr0MIwb8b0Uf3CPSz6FpxUyqW6I8QlBjoHg7b5tnwWniBt+oWIdQ3mQmY7fh9kj0KDueiajnWl6+CuxGQOZeWgHE/AyFrN7Ok7cDk81LXO5Nj5mxn0yXzFK5E7zDCzLZvXvvA+NVsb8PUbqA6VfT84xh3fXkP5AyU8+er9HPjRMVpPdpK7KJMlT88hKT/xus9329+uouWxNxjoGsTnMXC4dRKy3NedPF2tRU+Vs+Vbe0dPgtUUchZmynpEU8zIp1frNz3LjtIM4pw5KLnX/3N9vezGXgZSdmP+cyub134qhNUdGwd9WDgwkbWZRPQquCEnaKLjiNcpCnGTGCGiwSApHPY9zkLHr4fXalOwOOp7FI+dflXHuNrroBuG20NXNUFV0/yQfi1TgWLbQZaIv9aDKMrdwL/hH0H4Q9u2v33J/k8A/wgMLd38fdu2f3i5Y5YtKLO/+8p3xx1bpDv7ZhVv/8k2fJd0GtPjND69+/EJaXhg+iwqNtbQWdFF+qxUSm4rHFe76iG9jf0cfO44DftaSC9LYcnT80gvTQH8Cd3G/7Od07+v9E+WBRJy4nn0f9eTmCNPJKaKoadXd7/QiJX5hv9JVZhXa+48uRNPdScru9pQW++h6FvJQV93tRWfNPU8Cx3P41Y6UIAWq5xD3ifxIT/nIjq991c7OfFixXBnN92tkTUnncd+eVdIrh1CRCMHveRoxwGbZnMuXq5uvlokXgdjwVNPPbXPtu2Ap/bjTnQURdGA08Cd+Hs/7AGetG37+IjXfAJYbtv2F6/2uFMl0XntC+9z7s3ARgDORAd3/Uv0dLXpPN/N84++huExsXwWigqaS+PBH91B4crc4dd1VffQdLiVxNwE8pZdfhKsiB0Xn17Fs37Ts+xKySDOqZO2dl24QwP8yY7d2MuA12BlVxt6x32j9lumxZuP5fLZZZev+LiVNm5x/RO64h3eZtoaPXYe2wa/MmHxi6nBtm0a9rXQXdtL1rz0q56TE4rznnurmiO/OI0xYFD+0AzmfmDmhK4jJ0SsifTrYLQbK9EJxdC1FcBZ27YrABRFeR54CDh+2XcJgDFbpdr4585Ei63f2Yev14d9Yc6qbYHhMXn3z3fysY0PD78upShp1No5IvZdOgb5UFYO7gh7epU6exXM9ic8hxp1yNo7av+A12D9prYrzueZrm1DYfRaB5pikkgzyUot3XZ0PLgQkae/bYAXP7qBntpeUMA2bYpuzuee79064Qs+K4rCzLunM/Pu6RN6HiFiVTRcB2NVKBKdAqBmxN9rgWDLjj+mKMot+Ks/X7Ftu+bSFyiK8lngswBZ+VkhCC3yzfvATCrerglY7E1RoHDlRHRXmxi1OxqHk5yRumt6GOzx4koKviK3iF1jjUGO5KdXQwnPpYYSoIGU3fDMCjY/VhN0Pk+i2hywqBv4u/HEK+1Rleg46aFA249L6abdKqXZmg1Ez8OXWPP2n26js6ILa8SabtVb6tn/P8e44fMLwhiZEAIuXvOCiabrYKyZrGYErwK/tG17UFGUPwB+Atx+6Yts2/4B8APwD12bpNjCatqNeSz8aDmHfnoSRfFP0Ad44Ae3R9XCeM4kR9D1fRRViaqvQ4RGrD29Glnx2cHuMas77dYMMtWzaMro3wUVky67IByhX5c0tYIVzv8BLHTFYLq9nW4rn13ez2ERhu54U5y3z0fNtoZRSQ74F5A++svTkugIEWYjr3mqFvhAaH1a9F8Ho1UoEp06YNqIvxdysekAALZtt4346w+BfwjBeWPGTf9nGfOfnEXNtnqciQ5K7pg2IU0IJtLiT8xh578eHNVVTXOqlN1XHJPjuG3LpuKdGs78vhLNpTH3AzMpWBE9FbiJEo1VnGvhv0CNXd2pNlYxQ9+EYpuoir/EadgOmsz5eOxoWavHYqnjZ6PmGemKl2S1jiJtO5XmrZMeUd3uJrZ9Zx9tZzpJyk9k1VcWMfOuqTOMaqzFowF8A8aY+4QQEyv4NS/YrXXsXAejTSgSnT1AmaIoJfgTnCeAD498gaIoebZtN1z464PAiRCcN6akTk8idXp5uMO4bos/OYeOii5O/q4CzaVhei0KVuRw2/8NNooxutm2zet/tInqzfX+bnkKnHm9kiWfnMvqr02NxfPGWsE5lqo4Y7lSdWe3/reUZrxAZsIhTNtFTcc6ziQtC3fYVy1JaUJXAtdH0RUfhfq+SU90anc18sqn3sEY8D9EaT/TyYavbsX7f33MfWzmpMYSLnGpLlKnJ9F+tmvUdlVXmLFu2hjvEkJMpFgbuRCrxp3o2LZtKIryReAt/O2lf2zb9jFFUf4vsNe27VeALymK8iBgAO3AJ8Z7XhFZVE3ljm+tYdWXF9N+povkaYkx23SgdnvjxSQHwPY3Xtj/o+PM+1AZyYWTvx7MZLnSCs6xWMUZS7Dqzs4PNgKwiQeABwDwWj4+u6wjROvzTDz7MvNwLHvy5+hs+87+4SRniDFgsu0f9jPn0dIp07nxzn+8kRc/sgHLsDC9Frpbw5XsZPVXp8bDFSEiRayPXIg1IZmjY9v268Drl2z7qxF//gbwjVCcS0S2hOx4ErJje72QindqAtY9An8Dieot9cx/clYYopp4V1rBGZhyT7Qure6s3bwx8EWm74rd2iJJr53NoJ2ERhsjcwjDdlJjTv73tf1MZ9Dtg52D+PoMnInRNcz3euUszOSptx/m6POn6ajoJn95NnMeLY3Zr9/b6+PIL05RsbGG+Iw4Fn18DoWrcq/8RiEmkFRxos9kNSMQImY4Ex2ouhIwMVjRFBwJsfkrNfQEK/GlXoycjRxKD1zBGSBtin7YD1V3DpMasM9u7L1it7bIorDP+wlWuf7TP9cIExuFFrOcGnPFpEeTmJdAx7mugO16nI7ujr35f5eTmBvPqi8vDncYE87b5+P5h35PT0M/5qC/mle1pZ7VX1vCkk/ODXN0YioINjxbqjiho2CRl3GWjJR6+geSqWqah2G6JuRcsXlXJsQEmvNIKQd+dBzLuKSNsA0ld8TuePkkRxLQS2ZynKziHMSY/x5X0a0t0vTY+bwz8FfkaEdx0UO7VUK3HZ6f7VVfXsTbf7ptVKMT3a2z9DPzgnY3EtHv6C9P09t4MckB//DgHf90gLkfmCnLFYgJM1SxcaqNAfvifyNVnFDQVB/rlv2E5IRWdM2LYTpYUraRjfs+Tmdv6Js6SaIjxDVKLUnmtm+u4t2/2ImqqwyN7rn/v2+Lum55YnJcqVtbJLJw0GCGf/5H2b3FeHt9bP/H/Qz2+NBdGks/PTdiWirblk3rqQ5UTSW9LGXKzBmaSBUbawLmZYF/Ee3mw21MuzEvDFGJWHbpvJtgQ7N3lCZLFScE5kzfTkpiM7rmnwLg0H3Yto8181/k9Z1/GPLzSaITA2p3NbL7+4foqu4lb2k2K7+0iLSS5HCHFdPmPFLKjHXTqN3eiOZUKVyTF5NttEf63m4V+Xi/fulzl+Na1E/TkaNRVd2JBPMeL2PuB2bi7fPhiNcjppJTu6uRN/5oM8aAgW1BfEYc9/3XWrLmpIc7tKgWn+kGBbhkNT3btIlLm5jhLWLqCjbvJtjQbDeXqdyLq1aSd3g4yRmiKJDo7sDt6sYzGNr7V0l0otypVyrY8PVt2Kb/itBTe55zb1XxxMv3k1GWGt7gYpwryUnpXUXhDmNCXfqUa1txBnF6/JSdi3M9FCwWl22krHCv/2+rVHZtL+bQ2zlRU92JBIqqRNSQpf5WD69++t1RjUm6a3t58SMbeHr7B9DjLn95bTvdSV9LP1lz03GnxU10uFFl8SdmU/l+7ajhioqqkJiXQOactDBGJmKJdE+LLAo22KGviEuiE8Vsy+adb+wYTnKGmIMW7z6znQ/+5t4wRSZigXSXCY2FM9+lrHDfxSdYGqy+sQI1fR6HNvVKdSdKnXypAuuSz14Ay7CoeLuGWQ+UBH1ff9sArzz9Du1nO1F1FXPQZMnT/jW4ZNibX/7yHG78s2Vs+/Y+VF3FMm2SChJ46Md3yL+RCAm5voXPufrFzC/ZMqqqY9nQ3Z+Bxxv6ZUkk0Yli3XV9QccxAzQebpvkaESskKdcoaMoFrMK96BrvlHbdd3H/BmbqGr6AtE2d0f49TV7Rk2WH2L5LPpbAxdcHfLGH22i9UT7ha6N/vcffO4kmXPSmXVf8QRFG30WPTWbOY+U0ny0jbhUFxnlqZLkiHGT61v4naxeTX7GOdKSGtFUH6blwLQ0th35wIScTxKdKGZ6A9dyGRb4oFGIq5bkSOLelxpInZGEuzBVnnJdJ10bRFWDP4xwu3qBwLV4pLoTHQpX53L0l6cD1tRSNIX8FdlB39PX3E/jwZaA1vSGx+DAj45LonMJZ6JD1s4RISNVnMhgWTob932c7LRKMpLr8QwmU9M8G9OamGZOkuhEseTCJBRdwTYCs5r0mSlhiEgIMZLPiMPrc+N29QXs6+gZfQMXjZ3ZprLpt+STNTed5qNtw5V13a1Rclsh2fOCf78Gu72omoqJFbBvoHNwQuOdLP1tA3jaB0gpSor5Bi0iOkgVJxIpNHeU0NwRfIhvKEmiE8V0l8aKLyxk9/cPj5qno+oKd3xrdRgjE0L4Kew/vZ6Vc18dHo9s22BaDg6euSPg1VLdiR6qpvLwT+/k6POnOfliBapDYf4Ts5j9yIwx35NanIzmVPH1X3IsXaXk9sIJjnhieXt9bPjaFqo216M6VEDhpm8sZcGT5eEOTUxhE1nFUbCYW7yV8qLdOPUBWrsK2Hf6bjp6pP15JJFEJ8qt+KOFuDPi2PP/juBpHyCzPJWb/+IGchdnhTs0EYWGnnx9b3cKa9Ne4x09BxkVPz5VTQvwGm4WzNhEoruD9u5cDp+7nfae/DHfI9Wd6KC7NBZ/fA6LPz7nql6v6iq3fXMVb399G6bXxLZAc6m4Ulws/9z8CY52Ym342hYqN9djeS1Mr79iteXv9pJckMj0WwrCHJ2YaiajinPD7Ncozj2KrvvnYGan1bBu2XO8ufuz9PTL53SkkEQnyimKwsKPlLPwI/LUTIzPyCdfa0esJSDjl8evoW0mDW0zr+k9Ut2JTWX3FJMyLYmDz56gu66XaTfms+ipcuJSo3d9mP5WD1UXkpyRDI/J3v86KomOmFSTMRfH5eijJO8wmjZ6DqamGsyZvo3dJx4M2bnE+EiiI8QUF8vjlxPiOplbvJWctCp6Pakcr7yR5s7icId1TaS6E3uy52ew/p9vCncYIeNpG0B1qMOVnJF6G/uDvEOI0JvMa1lSfDumpQckOqpqk57cEPLziesniY4QU9jQhaH6mdjrQpPobufuFf+DrnlRVZvkhDay06rZdfx+qpoWhDu8ayLVHRHJUoqDr2SuaArTVkvXNDHxJrujWq8nDU0N7HxrWQo9/ekkutvp9aRBkMHfcc5eSvP3k5LQSmt3AefrF+EzZdHgiSKJjhBTXJIjCVXrJT0tKaaGqi2c8f5wkjNE13wsK3+T6qZ52KhhjO76SHVHRCLdpbHmT5ay7Tv7MDz+J9yKpuBMcLD889H1UEFErqEHc8FM9oiEAW8i1c1zmZZ9YtTCl4oChZmnKMg8i9fnYufxh2hsLx3en5rYxLplz6GqBrpmUpB1innFW3lz92fwDAZ/YCDGZ0okOpZhceSXpzny89OYXpNZ9xez9DPzcCU5wx2aEGKCZKdVjUpyhuiaQXxcN30DqZMfVAhIdUdEokVPzSZlWhJ7/+sIfU0eClfncMMXFpJckBju0EQMGFmxUbXAh1Tr0yZ/RMLO4w/iGUyirHAvuubFsjQUxUTTbMBC13zcsujXvL7zD+j1pAOwcs4rOPRBhta+deg+NNVgSdnbbD/62KTEPdVMiUTnjT/eTNWmuuEnTft/eIxzb1XzxCv3S59/IWLUgDeB+LiegO0KFoM+dxgiCi2p7kystjOdtJ7oIKUokZxFmSiK9B+8kuK1BRSvlcYDInRGzru58/0fs6M0kzhnsFvXyZ9XatsaB8+u4+DZO8hMqeK2Jc/juGTOjqKYzCzYy8Gz69FUH2lJjVz6UaKqNgWZZyYx8qkl5hOdtlMdVL1fN7ygG4A5aNFT38fZN6uY/dDYax4IEcsuPiFrZMeMncTp8THVSvpY5U2snvcyuuYb3maYGrUtszHM6O1wNZJUd0LP9Jq8/sVN1GxrQNEUsP3r3zzyszujujOaENEmYN5Ndm6EziFVcLv6CRw/AJpqkRTfCYBtq9goEOSVphXzt+NhE32D1K9Rw8HWYHPB8PUb1O1qnPyAhAgzV00mi6rb2Ly35sKY5t3EuZykrV0XgReQ61fTPJcjFbdgmA68hhPT1GhoLWPX8Qeu+5gObYCM5DrinIGVonBKnb0Kd1Eqh7Jy2FG6m+pnutm8t4ZF1W3hDi3q7PnPo9Rsa8AYMPH1Gfj6DdrOdLLx/2wPd2hCTAkjr1H+Ks5uDmXlRPQ1qr07H1UJ7DroMxw0tpcAYNkadS3lmOboW2/D1DlXt2RS4pyKYj6FTMyJR1EDMx3NqZIkY4fFFDPyCVnKjA0x1WUtmBNVN3K6ZgVJ8W0MeBMZ8F7v77zNotJ3KS/ahWVpaKpBXessdhx7GNNyhDTm6yXVndA49vzpUSMAACyfReV7dRiDpgx3FmICRU8VZ7S+gVQqGxcwPfcojgujCExLY9CXwPmGRcOv233ifm5f2kFSfDsAimLR0lnEkfO3hiXu8dJUH9lpVdi2QnPHdCw78tKKyIsoxIpuysOV5MTwmNjWxXKhqqvM++C1LeA3mTqremg+0kpibgJ5y7JkfLgYt6GxzvvjM7ByNhLnSo+JtXKuxLQcdPaOr8XtjPwDlE/b7e+uc6HDTn7mGZaXv86uEw+FIsyQkbk743NpknORjemVRGeitRxvZ/f3D9N2upOsOWnc8MWFZJanhTssMcGCz8WJrvXcdp+4n9auAsqn7UHXBqlpnsvxyhsxzIuNr7yGmzd3f4bMlFoS3R109maP+/oULgWZp1gz/3fYtsLQiLwthx+nqaMk3KGNEvOJjqqrPPbLu3j9jzbRfqYLRQV3Whx3/ctNJGTHhzu8AJZpsfHPtnPm9SpU3Z/cJObE8+jP14ckXtuyqdpST9OhVhJy4im7d7p0n5tCkhxJw39WcqWiebXmTt+OrvtGbdM1g+Lco+w5dR9WhI2vlurO9Su+rYDTv6/ENkePo08vS5XPyglWt7uJlz+10Z9s2tBV1c3592p55KfryVuaFe7wxASJ1ipOIIWK+qVU1C+94utau6bR2jVtUqKaCG5XNzcueGFUa22AWxY9z0tbv4LPiJx1gSLr6jxBUoqSePLl++lt7McYNEkpSozYCsnRX5zm7JtVmIMm5qB/W2dVD2/88WY+8Mu7x3VsY8DgxY9soO10J75+A92ts+3b+3j0F+vJmpMeguiFiDwKFopiY9nX/yTe5fCMdXAc2iCDEZboDJHqzrW78U+XUbOtAW+vD2PARHOqqA6VdX+/Jtyhxbz3/2b3cHdUANsCw2Oy6W9388Tv7gtjZGIixEIVZ6qannMUJWj7BZiWfYKK+siZcxSZV+cJkpgbeRWcSx3+31OjPugBbNOm6WAr/W0DxGdcf5a8/4fHaTnRgTnoP77hMTA88OaXNvPU2w+PJ2wRBWbbp/je7hTWvVjFtuJkor/B8uXp2iA3zH6dopzjKIpFe3c+u0/cd13DBJo7iyjIPIV6SfuWAW8Cg77QfK7EOXuJc/bS058R0nk/Ut25Nom58Tz19sMc+80ZGg+0kFaayoIPzyIxJ/KvH5HIGDRpP9NJXKqL5MKxq8i2bdN2qiPovpZj7RMVngiT2KniTE1OfQBVDRzmqyoWDn0wDBGNbUolOtHA5zGCbldUBWOMfVfr5EsVw0nOSN11ffTU95KUL0OZYtHQU7PKb3SyNutXbCvOwF2UGvMXlNuW/Jz0pAa0Cx/GGcl13Ln8OV7d/gUGvElXePdoB8/eQU56JTo+VNXCssCyHew5eS9B2zpeA10bZM38F8lLr7hQdbI5fO42TtWE9vsj1Z2r50p2svTpeeEOI+qd/N053v+b3di2f+HurLlp3Peft5GQFfiYRVEUnEkOvD2+gH2uZBkyGCukihMbGtpmUl60C8clQ7ptFBrbImvZlphvLx1tStcXoToCvy3u9DiSChIm8MyROZRPjM9Qi87qZ7rZNXOvv8vaFEhy0pIaSEtsQhuxeJui+BdvKyvYd83H6+nP5I2dn+Vc/RI6e7Kpa5nNxn0fo7511rhjXT3vJfLSK9A0E4fuxaH7WDTzPQoyT4372JdKnb2KtLXrcBelsqN0N+s3PUv1M90sqm7DVZMZ8vOJqavhQAvv/sVOvL0+fH0+zEGTpsNtvPKpd8Z8z6KPz0aPGz3EVI/TWPKpORMdrpgEQ1Wc6me6Wb/pWQ5n5+IuSpUk5yooiklp/j7WLXuWO5b+hOLcIwRbj2eytHRNo761DJ9xcfSBz3BQUb+Irr7ssMUVjFR0IswNX1jAubeq8XQMYHhMVIeKqqus+4cbxz2vaO4HStn9vcOjuwopkFKUSFL+RCZRYrINPTWrfsY/LGBH6dSo4gxJim/HCpK865pJalLTdR2zbyCNPSdDO0/A6egnP+PsqIQMQNd8zC3eRl1reUjPN0SqO2KiHXz2BMZg4DDsjvNdtJ3uJGNWasB7Vn5pEZ72QU68cA7NoWL6LOZ9qIzln1swSVGLiSBVnPGyWbv4F2Sl1g4vgJ2eXE9+xhm2H3s0TDEpbDv6GIVZpyjJO4Rlq5yvX0x9W+R1M5ZEJ0J0nu/myPOn6ajoZvbDJbhSXTTsbyGlKIkFH55FyrRrG2oTzOJPzqVqUz3NR9v860HEaagOlXv+PTr7t4vgFlW38YN9HuJf6PYvBuqM7bVygunszQ66eJth6rR2FYYhouBcjn4sW0MjcEip2zWxi5LK3B0xkXob+4I+cFZ1lf5WT9BER9VUbv/bVaz5+hJ66vpInpYone6inMzFGb/c9AoyU+qGkxwAh+6jMPsUqVWNYWxPrVDbMpvaltlhOv/VkUQnApx9q4q3vrwF0+u/Mat8txZFVbj7326m7N7ikJ1Hd2k8+ov11O5soulQCwk58cy8ezoOt/wYxIKpXsUZqbsvi6aOYnLSKofbX1qWgmE6OBdB3WB6PWn+NQguYVnK8GraE02qO2IiTL+lgOajbZiDox84mF6L7PmX/7mKS3ERl+KayPDEBJMqTujkpJ9H17wB2xUsctIqo3Ydnskid7hhZgyabPiTbcNJzhDbsnnrK1vJmpNOaklyyM6nKArTVucybbX8YsQSqeIE2nL4cRaUbGJmwX40zaC+rZQDp+/EG6IuaaFg2xr7T69n+ew3h5/WWZaKz3Ry9PwtkxaHVHdEqC18qpyjz5/G0zYwfH3T3Ro3/OECaS4Q46SKE1qD3gRMSw9Ys8aytZB1/YxlkuiEWfORNjCDTyizDIsjz5/m5m8sn+SoRDRx1WSS5PCSGO/EytmIu6hILiiAZekcOncHh87dEe5QLquiYQl9A6nMLd5GfFwXTR3FHD9/E/2DKZMei1R3RKjEpbh48tX7OfCjE1S8U0N8ehyLPzWHGXdE7yKJYrRgDUykihN6lY0LWFj6fsB224aa5sgeNhYJJNEJM92tYVtjd87oax5joUIhRMxo6iihqWNyhqpdiVR3pibLtOiq7sWZoJOQHZqnxO60ONZ8fQlrvh45w0XF+A0NS3vuQOCSF/G/kSpOqA14E9l08EPcsuj5S6o6CqmJzbR1R86800gkiU6YZc1NJy7NRV9TYEKjOVWK1xZc9v21OxvZ+59H6K7rI395Njd8YUFIGhcIIaY2qe5MHeffq2Xjn23H129gmxbZCzO59/u3Bl3vRkxtI4dJr815MWD/jtJkqeJMAK/hvrBEwsVtTscga5f8gt9t/iqWLbfzY5F/mTBTFIWHn7uT5x9+bdRinoqmkD4zlbJ7po/53pMvVfDun+8YbhfdVd3D2TereOLl+0mdLsnOVDHbPsX3dqew7sUqthUnI7cmIlSkuhP72s508sYXN41adqDxQAsvfextPvz6A+Ne1kDEhuDNbooCXucGqeJMgBn5B1CVwO6cimKRm15Bfdv413SLVZLoRICMWan8wcEn2P3vhzn3djWaQ2X2IzNY8OFyNKcW9D2WabH5b/eMujjZpo2vz2Dnvx7k7n+5ebLCF2Ey8sKzNvNXbCueul3WxMSS6k7066nvpflYO8mFiWTNSR/efuinJzF9lzTDMW26a3tpOdpO9gL5/k51Q1Uc1/vPs/vCvBsZlja5nPoAqhp8moNDD+zIJi6SRCdC6E7tmsYy9zb0YwwGjo+1LZu6nY2hDk9EGOmyJiabVHeik2VavPONHZz+/Xk0h4ZlWmTMSuOhH99BXKqL7ppe7CANcRRVobepXxKdKezSKs6u1Ex5mBYmNc1zKMw6hUP3jdquKdakLUUQrdRwByCujyvFGfTiBJCQLYOXYpWrJpNF1W08dyCF9ZueZUfpbtxFqaStXScXHzEpUmevwl2UyqGsHHaU7qb6mW42761hUXVbuEMTQRz66UnOvF6JOWjh7fVheExajrfz9p9uA6Dopjz0uMCRA6bPkiRnCltU3cbmvTWc/+fn2VG6m0NZOZLkhFFdSzmtXdPwGQ7g4rpwh86tZdCXEOboIpskOlHKleSkdH0Rmmv0BUp3ayz73IIwRSUmQ5IjCWecDppDLjwiLFJnryJt7TrcRansKN3N+k3PUv1MN4uq24K2nBXhc+inJzE8o8f2Wz6Lqs31eHt9zPtQGXHpcaiOi7cDultn/hNlJOaM3X2t6XAr7/7lTt78yhbOvlmFZVpjvlZEj6GHadXPdLN+07PsvlDFkYdp4WWj8v6BD7Pn1L309PsXmrYslUR3Jw5tINzhRTQZuhbF7vj2GkzfVirfq0V1atimzcovLbxsAwMhhAgVmbsT+Xx9gUOcwd+9yRgwiM908+TL97H3v45Q8XYtziQHiz8xh9mPzBjzmAefPc72fzqA6bWwLZvzG2vIW5bNgz+6HVWT56fRSubiRDgF5k3fRnxcN5pqoWmDlOYfIDutijd2/gG21C6CkkQnijncOvf9x1r6Wz30tw6QMj0Jh1u+pUKIySNzdyLb9FvzOfXy+YChzgm58bgz4gBwp8dx8zM3cPMzNwzvNwYMzrxWRfPRVtJmplL+YAmuJCee9gG2/eOBUV1Cff0GDfuaqdhYw8y75EFbtLnSXBxFsUiOb8VrxOEZTA5ztFNXfuaZC0nOxd89TTNJiOsiL/Ms9a3SeS0YuSuOAfGZbuIzZV5OrBu6GB37k27WZL7BrpQMaSUtIoZUdyLTmq8tpWqTf5iaOWii6gqqQ2Pdt9eM2Tq6r8XDrx55jcEuL75+A92tsfNfDvL4b++h5Xg7qq5gDo5+j6/f4MzrVZLoRJkrVXGmZR9nxZzXUBUTRbFo785j65EPMuBNDHPkU09aYiO6FthhTdO8pCU2SaIzBkl0hIgCU6XLmqb6WFT6LjMKDqKpPprai9l3+h56+uVGORpIdSfyJObG89G3HuToL89Qt7uRtBkpLPrYbFKLx34yv/Xv99LX4sE2/FUgw2NiDpq888wOln1mXtAESVH9c0dFdLiajmppSQ2snvcyunax01dmSh23Lv4lb+3+TDjCDjlFMbFtFYj89aJ6PekYpjOgnbRpOun1pIUpqsgniY4QESz4Im2x24Dg5oW/JjutCl3zzyvITa9g/Q0/4vfbvyCdZaKIVHciizstjhs+v4AbPn91jWoqNtYMJzlDbAvq9zbzwA9uQ9ECbwo1p8a8x2eGJF4xsa52Lk75tF2oyug5XqpqkRLfSkpCE119OZMZdkilJdWzYs5rpCc1YlkqlY0L2HvqbkwrcpP16uY5LCnbgKb6htfUsSwFn+Gkpnl2mKOLXDJzSYgINts+xXMHUrByNsZ8e8/khJZRSQ6AqoKmGsws2BfGyKYep6OfRHc7inL9nbRGdmazcjZy70vNJDmSpCtbFBiroYCigB6n89CP78CV7MSZ6MCRoKO5NFZ/bQk5C+V7GylcNZlB/3ctHdUS4rqCLlJp2SpuV+9kfBkTIj6uk3XLfkpGcgOKYqNpJtNzj3DLol+FO7TLsiydDXueprlzOpalYlkqzZ3T2bD3U1i21C3GIv8yQkQ4Z5z/11SJwaFqI6UktGDZgTdYumaQkVIfhogimY2CHfIuOw5tgNXzf0deegWWrWJZOntO3kN18/xxHTczOY5OGbUWNcofKuHor85geS8muqquUHxbIaqukrs4i6d3fpCa7Q34+g0KV+USf6GxgQi/oYqNU3UE7HO9e/Ud1RraZ5CRUjfq4ROAppp09OSFPO7JMqtwD6o6uuW6rplkpdaSHN9Kd/+1J+yFWSdZNPMdEt2d9HlSOHj2Dmpb5oQq5GF9A6m8u/9jaKp/OKFpBX6PxWiS6AghIkJPfwaqEvj00DA1Orpzx3yfpvqwbRXLDlz0MNYoismi0ncpK9yLrvno7M1m78l7aekqCsnxb1r4G7LTqtFUEw0T8LFq3qv0DaTS1l0YknOIyLfm60tpPNhKx7kuLNNC1VUSstzc/s2LN8W6S6PkNvmZiCSXDnVGC7wJDjYXZyxnapczq3AvCv1omj8x8BkOTtWsZNA39hpLkS4tqWlU57IhlqWSFN92zYnOtOzjrJ730nBCmJzQzpr5v2PHMYua5nkhiflSkuBcPUl0hIhQ/iE+bdzyiyqMcY4IURST7NRqVNWguWN6RI5D7uzNoa0rn4yUWvQLF1Xb9pfrz9QtC3h9SkIzK+e+SnpSPTYK9a1l7D7xQFRfgK9k5ZxXKco5PnxBTUtq5ralP+et3U/T1Zc9rmPHx3WSlVoTcAOgqj7mTt/OliOPX/ex39HjWQZ8b7fKJ5ac4hAyTyeSORMdfOh391K3u4m2kx2kliQz7cY8WSMnggVrWKPkBnZGc8NVjwzwGW7e2PVZ5kzfTmHWKQZ9bk5Wr6amOfSVisnU2lVAdmrVcPI2RFVNOq/jc3TxzHcCql66ZrCk7J0JS3TE1ZNER4gINGqyaHEmcc706+4Jk5lSw62Lnh+eb6EoFrtOPEB10/iGI02E9w8+ydJZGyjJO4ymmrR0TmPPyXsZ8CaNep3L0c+dy59F1wdRFQCb/Mwz3LHsJ7y+83NEQweda+Vy9DE951jgxVkxmFu8jR3HHhnX8eNdPViWBpdcsFUFEtyd133c1Nmr6GQn27w7WfteG9W/kS5s0UBRFApX5lK4cuxqqgi/iW5YM+hL4ODZOzl49s6QHC8SnKm9gfJpe1AUE/VC7m6YOvWtZfRdR/eyxDE+HxPiOgGbybwexbu6yEipwzOYSGvXtEk9d6SSREeICHI1LT+vhab6WLv4Fzgdoxe9WDX3Fdq78+n1pIci7JAxLSd7Tt7PnpP3XdgS/EN6Rv4BVNW8kOT4aapFQlwXWanVtHTG3loeie4OTEsP8hTSJjWxadzH7+zNDhi3DmCaGo3txeM6tnRhEyL0psqyA6E24E3krT1Ps7RsAznp5zFMJ2drl3L0/K3Xdbz+gSQS3N0B2z2DSUxeomGzvPx1SvMPYdoqCuAZTOTd/U/RP5gySTFEJkl0hIgQE3HRKsg8jRJk3ouiWMzIO8ThitvGE/IEuvzFISWhJWCogJ9NUnx7TCY6vZ70McaVQ3v3+CcGG6aL45U3Mmf6dhy678KxFQzTycnq1eM+vqyxI0RoTLVlByZCT38Gmw49GZJjHTp3GyvmvD5qvSHDdHD43NqQHP9qFOceYUbeYTTNYGi2qqZ2cNPC37Bhz6cnLY5IJImOEBEkMd6NGp9MnFMnbe26cR/PoQ9CkERHUy2cDs+4jx8urV2FTMs+MXxDPkRRoKMnetd2uJxBXzw1zbOYnnuckes1KgrUtpaH5BxHz99Cd18mc6ZvJ87ZT0P7DI5W3BIwdHA8pLoTGsagibfHizs9DkWV4SlThVRxIk9l4yJUxWLRzPeIc/Yy4E3g8LnbqGhYMmkxlE/bhX7J9dBf7W8m3tU1pas6kujEAGPQpG5XI5ZhU7AyB2eCdOMQfo0dJSgEJjo+w0Fda1kYIgqNysYFzJ+xGU01htd5MEyd1q4COnrywxzdxHE5A5NTRYF507dR1xKKBeMUqpvnUT3BE2ilunP9TK/Jpr/dw4kXzgE2zkQnt/zFcsofnBHu0MQEkipOZKtoWEJFwxIUxcQOQwdQh+4Nut22FXTdC4NBd08JkuhEMdNnUbO9njf/eAtD97KWabHu22uYdX9JeIMTEaHPk8bpmhuG2xErij/Jae6cTkNb9K5ibpgu3tr1GRaXbaQg8zSWrXGubjFHzq8Nd2gTKjutelQ1Z0hGSh0KVsjX1ZloUt25du/91S5Ov3oec9A/jNEzOMA7z+wgPsPNtBujd20TMTap4kSPcCQ5ANXNc5hTtCNgDqdpOejpm9qfpZLoRKHBHi/v//UuTr9WiW0EPq3f+GfbyVmUScq00A03ERNn6EndD/bFs37Tf7AtL4O4EP5qHjy7jsb2GZTmH0DTfFQ1LqC6aS7R3o3F400ad6exy3FoA8THddM3kIJhuibsPNfCMB1B5+mYlo4dpd9Pqe5cvcEeL6dersAcsZAngOEx2f39w5LoxBip4oirdaJqDdNzjuF29qLrPixLxbJVdh57KOoegIWaJDpR6NVPv0vT4dagSQ6AZdqcermCFV9cNMmRiWvlqslk894a4l/wX8gOZU3EkzqFxvZSGttLQ3jM2KUoFstmvUFp/kEsW0NRLE7X3MDBs+sId3J4tnYp5UW7RzViMEydivpFhDu28ZLqzpX1tw6gOtSARAegq7Y3DBGJiSJVHHEtfEYcr+/8A0ryDpOXcY4+Typn6pbT0y+fnZLoRJm2Ux00H2sLeqEbYhkWg92+MfeL8LtYxfGwftOz7CjNIM6ZE5IGBOPldnUzr3gLuenn8QwmcaJqNfVts8ISi6JYzC7awazCPeiaj/rWmRw8dweeweQJO+eCkveZkX8YTTPR8FdPZhXuxeNN5FQIuo+Nx5GK20hKaCc/4yyWpaGqJs0d0zlwZn1Y4woVqe5cXlJ+QtDtigq5i8e5qvAU1NfcT+X7dai6SskdhcSlhL9yK1Uccb1My8nZuuWcrVse7lAiiiQ6UaazuufC6tSBw1eGONw6xbcVTl5Q4prNtk/x3IEU7n2pjn2zIudJndvZwz0r/xuHPoimWiQntJORUs/Bs7dzumblpMezet7vKMw6NVzBmJ57lNyMCn6/4/P4DPcEnNG+UDEZ/aBA133Mnb4j7ImOZWtsPfw4Ce4OUhJa6OnPiMkndlLdCU53aaz80iJ2/utBDM+Fa4ACepzOyi9JBf9aHPrZSbb+/V4UVUVR4L2/3Mn6797EzLvC15peqjhChJ4kOlEmszwN0zd2NccRr1N0cz6Fq2KzxW4sccbp4EoEiJgL2ZzibcNJzhBd87Go9F3O1S3FtCavo1+iu31UkgP+dpkOfZCZBfs5UXVjyM+pKDa6Frx7TSS14+7zpF3XCt7RRKo7wS399DwSc+LZ8x9H6GvxkLs4kzV/spSMstRwh3bdTK/JQJcXd7rrwoO8idVR0cW2b+/DHLSAi591G766lYKtObjT4iY8hpGkiiPExJFEJ8qkFCVRckchle/WYgxcfKKn6gpFNxcw7wMzmXHnNJRgrZmEuILc9POjkpwhNgrJCa109EzeZOe0pCYsS4VLmtjomkFWag0nqkJ/TttW6e7PICWhLWBfe3du6E8orujS6o75z61sXvupKV3dmfVACbMeiP7OmpZpseOfD3LopyewLX/FatVXF7PoqVC0Sh/bqVfOYxpBHhiqChVv1zDv8clrvS9VHBGMolg4dQ9ew41tT+1mAuMliU4Uuuu7N7P3P49w5Ben8fUbFN2cx03/Z5l0WYsSrppMwH8jbSS9CKSHNZ6RPANJpCa2BGzXFJMBb/D5AROl15OKEmSxU9PS6OqbuPkIe0/ew62Ln0dVDVQFLEvBsnX2n75rws4ZCTTVh6JYEdNhbqSR1Z3dwPpNz0p1Jwbs/Fd/kjM0DM8cNNn2nX2401wTukSC6TWxzSDNfCz7siMmQkmqOCI4m9lFO5hfsgVNNTAtnaPnb+Zk9WqiveFMuEiiE4U0h8rKLy2SMdlRaGSXtTUXuqwpuYnhDmvY8ao1ZKXVjJqjYpoazZ3TJ7QBQDAdPbl092WRmtQ4qspkWSpnaidusmVTxww27v0E80o2k5rYQkdPLkfP30Jnb2wOB3U5+lk55xXyMs+iAJ29Wew68eCkVu+ulszdiR2WYXHouZMX5xpdYHhMdv37oQlNdErXF3Hop4Hntm0oua0gpOdaVB1YHQakiiOCKivcy8IZm9B1/zVY00wWztiEaTk4U3tDmKOLTpLoCDEJIrnL2khNHTPYd+oulsx6GwUbVTFp7Chh+5FHwxCNwnsHPsLKua+Qn3kGgF5PGruOP0j/QOqEnrm9J58th5+Y0HNEBpvbl/6U5ITW4WQyLamJdct+wqvbv8iAN3KS8CEydyc2+PqNMasnfU0TOx8ud3EWcx+byfEXzmIMmCgKaE6NFV9aRFJ+aH7mR1Zsgs07Wp8mVRwRaH7JluEkZ4iu+5hfskUSneskiY4QE2xy1soJnXP1SznfsJCk+DYGfQlhvdn1Gm62HP4QmupDUw28E9JpberKTKkh0d05qmKmKP7x4TPz93O08pYwRnd5MncnujmTHMSlOOlvHQjYlzl74httrP3/VlL+YAln3qhCdaiUP1hC1pzQDCMOnHcT7FYrsq8DIjzinMHXw4pz9k1yJLEjJImOoih3A/+Gf9rwD23b/vYl+13AT4Fl+CcnfMi27cpQnFuISBUtVZxgLFunqy9yhmqZlmNSO75NFYnuzqDbdc0gOSGwMhJp83hk7k70UhSFm76xnHf/fMfFxjqAHqdx458tm5QY8pZlk7csO2THk3k3kyvB3UFR9nFU1aS2eTZdfaH7XobLWM1wuvsjZy5vtBl3oqMoigb8P+BOoBbYoyjKK7ZtHx/xsqeBDtu2ZyqK8gTwHeBD4z23EJEq2qo4Ymrq6MlBUQKHDxmGg9aui2txuRx9rJr7CnkZ5wDo7M1m5/GHImbeklR3otPsh2fgSnGy618P0V3XS+bsNNZ8fQm5i7PCHdo1k+5pk2tmwV6WztqAolgo2Mwr3srJ6lUcPnd7uEMblwOn13PTwt+MWlbBMHUOnL4zjFFFt1BUdFYAZ23brgBQFOV54CFgZKLzEPA3F/78W+D7iqIotm0HaXsiRPSK5iqOmHq6+nJo6igmJ61y+MJqWQpe00VFw1CzE5t1y35CYnw76vA8nkbWLXuOV7d/kUHf5HbjG4tUd6JTyW2FlETxAtdSxZl8bmcPS2dtGL3GGgazi3ZS0zwnIhupXK36tjI2HXqCRaXvkhzfTnd/OofO3U5T+4xwhxa1QpHoFAA1I/5eC1y6hPrwa2zbNhRF6QIygFFXHkVRPgt8FiArP/qe6Agx2z7FcwdSuPelOvbNkid6IvJtOfQh5hZvYWbhfjTVoK6lnINnbx8enpadVkV8XHfAPB5VNZmRf3BCFm4dD6nuiMkiVZzwKMg6TbBWy6pqMi37eFQnOgBN7TPYIIlNyERUMwLbtn8A/ACgbEGZVHtEVHLG6eDyT+CXC56IdJatcfT8Wo6eXxt0f6K7Awj8ONY1g+T4yKySSHVHTCSp4oSXPfx/wXbIWjNitFAkOnXAtBF/L7ywLdhrahVF0YEUhlZMFEIIEbE6enJRgtw7+AwHrd2RPeRIqjsi1KSKE351LeUsm/VWwHbL1qhqnAdAorud5eWvk5teiWVrnG9YwIEzd0ZMIxUxeQKbu1+7PUCZoigliqI4gSeAVy55zSvAxy/8+QPAuzI/R8QaV03m8J+NpBfDGIkQodPRk0dLZyGGefG5mGmp+Iw4qhoXhDGyq5M6exVpa9fhLkpld2qmv7rzTDeLqttG/c4KcTmumkwWVbf5qzibnmVH6W7cRamkrV0nSc4kG/AmsufEfRimjmHqmKaGYeocPX8zXX05OHUPd93wI3LTz6OqFrrmY0beIW5b8nOCl4JELBt3RefCnJsvAm/hby/9Y9u2jymK8n+BvbZtvwL8CPiZoihngXb8yZAQMWNkl7U1F7qsKbmRt9iiENdj06EnmVe8hdKCA2iqQW1LOYfO3oFhOsMd2lWT6o64XlLFiSyJ7naKco+hKhY2Cu3duew+cR9dfbkAzMg/iKb5UNWLSY2mmaQmNpGRXE9bd0G4QhdhoERqYaVsQZn93Ve+G+4whLisi13W4lm/6Vl2pWQQ59Sly5oQEazz5E481Z2s7GpDbb2Hom8ly9wdEWDkXBwr8w12pchcnHBz6h4eWPN9HPrAcCJjmBrt3fls3PdJAFbPe5GSvKMB7/UZDvaeupfzwx0lRSx56qmn9tm2vfzS7RHVjECIaCJr5QgRnaS6I65EqjiRKVi1RtdM0pMaSU+qp70nn46ePKZln0LXfAHv7+6T4apTjSQ6QoyDU3WgxieTnpaEIhdBIaKGdGYTQxWbYKSjWmRKT2oYtX7OEBuFlMQW2nvyqahfzLziraiKMZwQmaZGZ282bd35kx2yCDNJdIQQQkxZUt2ZmkZWbFQtsC+TVHEiU0dvLoXmySDJjj1crfEabt7a8/SIrmsqlQ0L2H9mPdJ+euqRREcIIcSUJtWdqSPYGjhxzsBbIXeuVHEi0bn6xcwdrtb4t1mWQndf5qhqTa8nnfcPfjRMUYpIIomOENdhaH7O3S80sq10N3F6PGlyUYwYcc5elpZtoCDrNLatcr5hAYfORVeXMDH5LlfdOamUB7xeEqDoIvNuop/XF8/7B57gjmU/Q7FNFMXfMDrR3UFSfDs9/VKFFaNJoiPENbjYZc1zYS0F6bIWaTTVx10r/ge3sw9VtQCYWbCfjJQ6Nux5Ghm6IC5nrOrOzg82Brz2E3aXVHyiQLAqjsy7iV7l0/aiKtbwQsaaaqPoA6yY8yrv7PtEWGMTkUcSHSGu0sguaykzNkiXtQg1PecYTn1gOMkB/xoKKQktZKVW09I5PYzRiWhxaXVn7eaNAa+p/s06Nj9WI/N5IphUcWJPQdbpUV3XAFQVslJrUBUTy9bCFNm1cbu6cWiD9PRnYBM4T0yEhiQ6QlyFoSTn3peaMXI2EudKlypOhEpLrsehB7YVVRSb1MRmSXTEVRtZ3TlMasB+T+Ju1m9qk/k8EUiqOLHLtsdICmwFOwoq9m5nDzct/DXpSU1YtoJpOdh5/EHqW2eFO7SYJImOEFfJqTrAlUhmchxKbmK4wxFj6O7Lwmc4ApId21bp6U8PU1Qimo19c3yx4sMzK6S6EwaLqtuCbpcqjk2iuwPL0ugfTAl3MCF1vmEBMwv2o2nm8DbTUqlvLRs7CYoYNrcv/RlJ8W2oqo0GOPBx04IXeHP3p+nuywp3gDFHEh0hREypbFjAwhnvY1m+4a48pqXiGUyisX1GeIMTMWVkxWcHUt2ZTEMVm8pvdKLrgbcy69OmbhUnI7mWGxe8QJyzD4Ce/nS2HvlgzEzUP3TuDjJS6klJaEZRbGxbpX8wid0n7g93aFeUkVxPfFxX4NA7xWBW4R72nro3TJHFLkl0hLgKs+1T7MT/VOwdPT4KiuNTl8+MY8OeT7Fizu/JSqsGW6G+tezCRVC+c5rqQ1UNfIY73KHEjEvn80h1Z2KNnHeza+beoO2hYSpWccDl6OP2pf+LQ/cOb0tJaGbdsud4eeuXo2b+yuUYppMNez5FVkoNqUlN9PSnX3iIFfmf725XT9Cqk6raJLg7Jz+gKUASHSEuY+RTw7VZv2JbXoasrxAFejwZvLP/46iKiY0SBcMZJp5DH2DlnFcoyDoD2PR60th1/AFau4rCHVpMuFx1Jxip+Fze0GdvMJXf6GR91tSt2FxOcd5hFMUatU1VQdd85GeeprZlTpgiCzWFlq4iWqLs86utqwBNvXSxUzBMnYa20jBEFPsk0RFiDIFPDafmE8JoFgtPL0PltsU/Jy2pEU31j2tPSWjjtiU/5/Wdn6NvIC3M0cWOS6s76l8Gfl5YpiUVn8sY2eFS1QIfUsjn8dgSXN3oWuCNtKJYxMf1hCEiMZLHm8SZumWU5h8YnkdqmhoD3gQq6heHN7gYJYmOEJcYWcWRp4YiFqQmNpKa1Dxq8i6AqprMmraHA2fWhymy2DSyurPPuTdg/4DXkPk8QVz87D3H+qy3htcpu5RU1cfW3FlEacH+IJ0nFVo7C8MSkxht/+m7aO8uoHzaLhz6ADXNszlRdSOG6Qp3aDFJEh0hgnjuQArrEiA1LQl3oVxURXRLdHdiWSpcUuDSVIvkBLnBnihDCc+lOk8Gn89zUikPeG20J0Cumsyrfu1QFWdX6W7iXLFdsYl3dVGSd4g4Vx+NbTP8HcNCsJZKXWs5Pf0ZJCe0Dld2DNNBY3sJ7T354z6+CAWFysYFVDYuCHcgU4IkOkIIEeM6enKGh6yNZJg6LZ3TwhDR1DbWfJ6dH2wMeO0n7K6orPgMVWd+sM/jb81/Bbf/pob1mW9cqOI4Y3qdsvyMM9y08LcoWGiayYy8Q7R35/LegafGPdzWtlXe3vsJ5hTtpDjvCJatcrZuKWdqbwhR9EJEF0l0hBAixvUNpFHTPJvC7FPomn9Ii2UpGKaTs3XLwhzd5EqKbyM9qZ6+gVRauwoJZ6emS+fzrN28MeA11b9ZF3XzeUbOsVmf+SvQrpzo7ChNnhLzbhTFZM383w3/HgI4dC8ZyQ2U5B/kXAh+H03LydHKWzhaecu4jyVEtJNER4gRhp5Cxv+mGyNnI+/o6VHQsFKIK9tx/GFm9+5gVuEedM1HfdtMDp29Ha8vPtyhTQpFsVgz70UKsk5j2SoK0DeQwjv7PsagLyFscY2s7hwmNWC/JzFwfZ5INlTFWb/p2QvVmZyrWmDZzeUWZo0dGcn1cElXNABd91GSezgkiY4Q4iJJdIS4YGSXNf9q2ukoMf50UUwdtq1youpGTlTdGO5QwqJ82i4Ksk6P6kiVHN/Gmvm/470DHw1jZH5jf86Mns8TbHgbWBRnV5CW2EZTZz717eGrVMX/xl/FOZQV+9WZ62Fa2pjfGdOSWzIhQk1+q8SUN1TFqX7Gf4GWLmtCxJ6ywj0BbXdV1SI7rQqHNoDPjAtTZJd36XyeS4e36YkGC79SjTPNQFH8q633Vsdx7PvTsHyTv37U0BC0WJ5jMx4dPXn4DNeoBT0BfIZjyg0jvVaa6sPt6sEzmIRpXXk4pBAgiY6Y4kaOJbdyNnIoXZ5CChGLgq0tAoANmmbgC+zVEFGG5vNcOrztnnsP4Mr2oWn28LaEkkHiPuZl545ZkxskU2cI2vVT2HToCW5f+jNUxRpe3LOqaT41zbGymGeo2SyY8T5zpu/EtkFR4FT1DRw6dwfhnGMnooMkOmLKc6oO1Phk0pPjZKhalIuP66Qo+ziaZlDXMovO3txwhyQmlU1qYjO65qW9Ow/LvniJq20pZ0b+ATR19PyIvoEUBrzhm6NzLS79bFIVg+nFG1FVe9R2XbeYu6CZkx0fG7HVJi/jLAWZp/EacZxvWExPf3Q0N4g1HT15/G7LVynIPI3L0U9zRzHd/VffhnuqKZ+2izlFO0c1cCgv2oPPiON41U1hjExEA0l0hBAxoTj3ECvmvIaCjaJYzCveytnapew/c3e4QwspRbHQVAPDdCBPMy9Kcrdx6+Jf4nb1YKOiYLPz+APUNM8D4PC5teRnnsalD6DrPkxTw7JVdh5/iGj9d/QPVbOD7lOViyUqBYtbFv2K7LRKHLoP01KZXbSL3Sfup7Jx4SRFK0ayLJ2a5rnhDiMqzC3ehn7JAqi65mNO8Q5JdMQVSaIjhIh6Tkc/K+a8Nmp4korBzMIDVDfPo7Ur+teKURWTJWUbKC04gKpY9A2ksPfkPTS0zwx3aGGnYHH7sp/hdnajjpiWsnreK3T3ZdHVl82gL4HXdnyekrxD5KRV0d2fwdnaZfQPpoQv8HEyLQft3fmkp9ShjsjVLEultuVid7bC7JPDSQ5woaplsWLO76ltKZcV2acwVTXISqnGRqWlcxr2ONfxmQguR3/Q7U7dgz/Rj84HFWJySKIjpqyhLmt3v9DIttLdxOnx8nEZpfIzzmLbgROvVcXH9Nwjk5LouBx95KZXYFoOGtpKQz5ZduXcV5iWfWI4mUuK7+DmRb9h496PT/kVz7PSqnDqA6OSHPAP7ZpZsI99p+8BwDBdnKldwZnaFWGIcmLsPPEg65c/i6oa6JqBz3DgM+I4ePZiM4DpOUeHk5yRLEslJ62SutbIblktJkZ+xhnWLHgBbAUUG9tW2XLocZo7i8Md2ihdfdmkJTUFbO/uy0SSHHElkuiIKUe6rMWisS52iv8iPsFmTdvFkpnvYF1ItmwUNh18gpbO6SE5vsvRd2Hu0egZ86rqY17xVrYceTwk54lWLocn6AAuVbVxu3omPZ7J1N2XxSvb/ogZeQdJTmylrSufysYFmJZz+DWm5RiexD2KokhL4ynK7ezhpoW/CWjScevi53lpy5cjqgvhvtN3sXbxL9BUA0UB2/a34t53+q5whyaiwOT3nhQijBZVt7F5bw3Vz/jXyjmUlSNJTgyoa5053L1oJMvWqWxcMKHnTktqYPHMd9A0A4fuxaF7ceqD3Lr4eTQ18Cn69YiP68a0A29IVQWSE1pDco5o1to5DS3I999nOKhvnfzOY5PNa7g5WbOa3Sce4Fz9slFJDsC5+iWYZmCF0bYVmjuKJylKEUmm5x4l+Pwum2k5JyY7nMtq7ijmnX0fo6GtlP6BJBrbZ/Du/qdobC8Nd2giCsijHDFluGoySXJ4ufelBoycjbiLiiTBiRE+w82OYw+xet7LACjY2Cicql5BW3fhhJ57Rt4hVDVYb2KbvIxz1LbMHvc5evvTRk0uH2JZCm3dU3vYGoDHm8TJ6pXMmrZ7eIiWYer0etKpbJof5ujCr7mjmBPVK5k7fSe2rWChwIWqoxWBczLExHPqHrQgn1uqYuLUB8IQ0eW1dRfy/sGPhDsMEYUk0RFTUmZy5JTlRWjUNM+jpXM607JPoKkG9a1lk9KyVdcGUZXAJ6MKjGqHOh4+M47TNTcwq3DvcPehoeEbxypvDsk5ot2hc7fT2lVI2bS9OPUBqhrncbZuGZYMzQLgSMXtnKtfSm76eXyGi/rWMll0cQpraC+lvGhXwNwtG5WGthlhikqI0JMrgBAiZgx4EzlTe8OknrOmZQ5FOccDbhhUxaKxvSRk5zl4dh19AynMmb4Dl6Of1q5CDpxZL2uhDFOoay2XifWX0T+QSkX9knCHISJAS2cRDW2l5GWcG/7s8hkOqprm09WXE+boxNVwOfooLdhPSkILrV2FnG9YJB0Ug5BER0wZs+1TfG93CutcibwjHdZEiNS3ltHYPoPc9Aocug/LUrBsjUNnb2PAmxjCMykx1zFMCBEuCluPfJCi7OOU5B3GslUqGhZT1xL7c9piQWpiE+uWPYeqmuiawbTsU8wv2cqbuz+NZzA53OFFFEl0RMwb2WVtbeav2JaXgTtXGhCIUFHYcvhx8jPPUJR9Ap/poKJ+MR1TvOWzECLSKVQ3z6P6wqK6InqsnPMKDn1wuJOirvlQFYPFMzey49ij4Q0uwkiiI2La0Fo58S/4u6zFOXNw5yZKkiNCTKG+ddaU6PAlxFSiKBbFuYeZkXcIG4VzdUuoapqPrN8iwkVTfaQlNQa0i1dVm4KsM+EJKoJJoiNikqyVI8SV2DgdHgzDJZ23hAjK5paFz5OdVjU8jyUjuY6CrNNsP/pYmGOLDknxrSS52+nqy6ZvIDXc4QCQnVrJ/BmbSIrvoL07jyMVt9LZmxvusK6abavYKARrD25Z8ll+KUl0RMxKciShar2kpyXhLpQkR4gh07KPsbz8LZwOD7atcLZ2KQfO3oktCY8Qw7LTqkYlOQAO3Udh1mnSkupleOpl6JqXmxf+iqzUGixLQ1NNalrK2XHs4bB+zhRkneTG+b8b7ojpdnaTl1HBxn0fo727IGxxXQvL1qhrmUVB1plRLcINU+dcnTQbuZQsGCqEEFNITloFq+e9gtvVi3ZhIuvMwgMsL3/zqt6va4NkpVaRFN82/Pfl5a/xgVu/wwfX/j1r5r9AnLNnIr8EISZFTtr5oC3iFcUkJ61y8gOKIstmvUF2ajW6ZuB0DKJpBoVZp5g3fWsYo7JZXv7mqO+pqvrntyyZuTGMcV273Sfvp7svA5/hxGc4MUwHLZ3TOHL+1nCHFnGkoiOEEFPI/BmbA27edM3HjLxDHDiz7rLtSedM38aCGZuwLA1VNenoyUbXTJLjW9E0/5PFouzjZKVW8/vtX4z6dVoUxaIg8zS56efwDCZxvmER/YMp4Q5LTJJBXzympaNrxqjtlqXh9cWHKarIpygWxXlHAxYk1TWDsml7OVoZnptxXfPidvYG3ZeeXD/J0YyP1xfPG7v+gMyUGpLiO+joyYmq4XeTSRIdEXNcNZls3ltD/AuNWJlv8I6eI9NGhbgg0d0ZdLtlK8Q5++j1BE90CjJPsaBks/+m78KNX3pyA4rCqAVTVdXGqQ9SlHOM8w2LQx3+pFEVgzuW/ZTUxCYcug/T1JhXso0thz5IQ/vMcIcXg2xy0yvITqtkwJtEVeN8BsOcTFQ1zmdx6bsB220UqpvnhCGi0ElNbCI5oZXuvoyQ3yArioWiWEH36Zo3pOe6FqblwLI1VAJjC+1SAJNFobWriNauonAHEtEk0RExY6gBwQ/2ebjz/R+zozSTOGcOaWvXhTs0ISJGe3cebmc3asDAZYX+gbHXX5hdtBP9kkVRNdXGDpwPi0P3kp7UENWJzsyC/aQlNQ1Xv/wVK5M183/Hi1u+hm3LyO9QURWTtUv+l4zkBnTNi2npLCp9h/cPfISWMN7EDfoS2HToCW5a8FvUC9UJw3Sw5fDjUbswo6b6uHXxL8lMqcO2FRTFpq07n/cPPIlpOUNyDsvS6erNJi2p6ZLtCk0hXET5Wtm2yuma5cyatmdUlc5nOjhWeVPY4hITSz6pRUwYquJUP9PN+k3Pcjg7F3dRqiQ5Qlzi8Lm1mJZjVIJiGA6OVNyCZY/97CvO2XfV5/AZDrr6ssYTZtgV5x0OOj9DVU3SkhrCEFHsKi3YR2ZKPQ7di6L4hzg5dB83LfwtwTpLTaamjhJe3PI1Nh96nNoLi2neuuh5bpj9Gk5Hf1hjux6Ly94mM6UGXfPh0L3omo+M5FqWzHo7pOfZfeJ+fIYD0/LfZpqmhmG6OHDmzpCe51odOncH5+oXY5j6hfktDo6dv4mK+sVhjUtMHKnoiKg3VMmJf6EbK2cjh9KliiNij6b6KMk7RGHWKQa8CZypXU5bd+E1H6erL4eN+z7B4pnvkJFch8ebxLHzN1PZuOCy76trLSPR3TE8F2ck01LRVP9wEMtSMC3HFY8X6UxrrMujjTXmPnE9ZuQdCppU6pqX1MSmsM89sG2FRTPfIzWxabgSMCP/ALnpFby24w8v+4Ag0vj/rS+dO2MyI+8we0/eF7LztHUX8PrOz1E+bTepic20dBVypvaGsA8Rs22Vfafu5dDZdcQ5e+kfTJbf5xgn310RE4ZbSSfHoeRG41hbIcamqT7uvOHHJLvb0XUflgVFOSfYf/pOztYtv+bjdfTk8d6Bj17Te05UraE47whOPOiaiWWDZTnYf/oOctMrKcw6DYpNc+d0dp+4P2qH9gw5W7uM9KSGUa2Fbds/Qb2zNzuMkcWesYcB+odG5qRV4HQM0NJZFJYb5ezUKlISWkYNd9JUizhnH9OyT15YQDQ6XNogYIiqGvirZ6Gb0do3kMb+M3eF7HihZJhOej3p4Q5DTAJJdIQQIsLNyD9Icnz78FNvVQUVH0tnbaCyccGkJBWDvgTe2Pk5Zk3bRV7GOfoHUjhZvYrWrmmcrVuBggXK5W5ao0tV03xy0yuYnnsMbLBRMS2NTQefIJQ3gwLO1i8lNbE5YA6Yz3Bx+9Kfo2s+bNt/k3608maOnb9lUuPzr0IfmCA4dC/pyfVRleg0dUwnJ+38qDl6lgXNndORn2sRiyTRETHhe7tVZLCaiFXTsk8EHdpjWRqZKbU0tpdOShyDvniOVNzGkYrbAvbZqOGeThFiCrtOPMSJqjVkp1Uz4E2gvnVmVA1Tihbn6xdRkHmavIwKFMXEsvThifJxzj6UEfff84q30dpZSFPHjHGfV1EsMpLrAP9Qq7GS9F5PGpatAaOTHZ/hoKc/Y9xxTKa9p+7lrht+hGob6JqBYepYls7ek/eGOzQhJoR8YouoNbLL2vpNz7KtOIM4PZ602avCHZoQITXoc2PbjLrhA1AUG58RF56gwigpvpWlszaQk1aFz3BypmY5x6punpBqUnd/Ft390d1YIdLZqGw5/CEykuvISvUnlX0Dyaxd/HzAz7ym+igr3DvuRCcn7Tw3LfwN6oU2yKals+XQ40G7vNW3leH1udFVH6rqz+Yty99drLIxeqo5AD39Gby6/QvMLNhHenIj7d25nK1bxqAvIdyhCTEhJNERUeniWjndrM/8FYeycnDnJpIqSY6IQWdqbqAg8+yoqo5lwaA3nrbu/DBGNvncrm7uuuFH6PogquJf7HRuyTaSEtrYcezRcIcnxqGtu4C27gIActIrsO3AoVSKAk7HwLjO43L0ceui50cNlXPgZe2SX/DSli/jM0c/PLBtlbf3fpJVc18hJ63yQqz57Dz+YFTORRv0JXCscnKH/wkRLpLoiKhyaRVnR2mGrJUjYl5zZzGHz93KwtL3sSwNRbEZ9Ll57+BHmGrj6sun7UZTDdQRX7auGRRln+TQ2S76B1PCF5wImdbOQlQ1cGFHn+mgqnHeuI49PecYKMHGWdpMyzlBRf2SgD2ewWTeO/BRNNWHotgYZmjWnJlMGcm1lBftIt7VTX1rGWdqlwckdULEGkl0RNRJciRx70sNpM5Iwl2YKlUcMSWcrF7DufolZKXU4DXctHYVMtWSHPDfrAVvca2RktAiiU6MMC0ne07eww2zX0dVTFTVxmc46O7L4nzjonEd2+noR1ONgO2qauK6wto4puUY17nDpSTvIDfMfgNV8aGqkJ7cwMzCfbyx67P4DHe4wxNiwkiiI4QQYWFTlHOcWYV70DUvVY3zOV27/LKrk/sMN/VtsyYxxsjT2ZtDVmptwNN+VTXpkXaxMeV8w2I6enKZWbCPOGc/tS3lVDfNu9AY4Po1dZQwx9wxqnU4gG1rNLWXjOvYkUhVDZaXvzlq6KuuGbidvcyetosj59eGLzghJpgkOkIIEQbLyt9gRt6h4Zut5IRWivOO8Nbup6Wz12WcqlnJjPyDoxIdw9Rp7pgu62LEoM7eXPaeCt1ClgAtnUU0tpeQm35++PfPZzioa51Fe0/szXlLTWwK2hBR00wKsk9JoiNimlxNRdQYmp/zvd0prE17jXf0nCk4cEfEgoS4DkrzD45agFDXDBLd7RTlHKeycWEYo4tsvZ503t3/FCvmvEZKYjO2pXK+YQH7T98d7tDENbHJSaukJO8QqmJR2TSf+tYyJmc4psLWw48zPfcoM/IPYdtQUb+Eqqbxzf2JVF6fe7i73KUGvdJtTcQ2SXREVBjZZW3thS5rinRZE1EqK7UmaCtkh+4jL+OsJDpX0NZdyBu7/gBVMbBt1b+GjwgZl6Of2UU7yM88w6A3npPVq0I+ZHJJ2dvMLNiHrvlQFCjIOkVty2x2HHuYyUh2bFQqGxdOid+1Xk86Xb1ZpCU1DrfHBjBMByerV4YxMiEmniQ6IqJJlzURiwbGeIpqWir9g8mTHE30ioYhfskJLSyb9SbZadUYhpMztcs4cv5W7HHOM5koTt3DPSv/G5ejf7jpQ2ZqHUcqbuFE1Y0hOUdSfCtlhXtHVTQduo/CrJNkptTS2jUtJOcRF20+9ARrl/ycJHcHlq2iqiZHKm6moa0sBEe3yUyppTDrJKalU9W4gO7+zBAcV4jxi/yrhJiyhpKc6mdkrRwRW5o6SvAZTjTVizqiGGHbGufqloYvMBFSblc362/4MbrmX/NHc3qYPX0nifEdbD/6WLjDC2rWtN04HZ5Rne10zceCGZs4U7s8JOvG5GecC7pd13zkZ54OS6KjKBYzC/ZSVrgPVTWpapzHiao1UblOTjAebxJv7PocKQlNxDn7aO/JD9FiwzY3zH6NktwjaJoP21aZM30H+0+v52zd8hAcX4jxkXq/iGhJjiRUTSUzLUmGqomYYdsq7+z/OD2eDAzTgc9wMuiLY+uRxyJoQr1Ndlol07KP43Z1hzuYqFQ+bVfQNX8Ks04R7+oKX2CXkZ9xdlSlZYhlaaQnNYTkHD7TGXQxUMvSMIzwJBY3zn+BJWXvkJrYQnJ8O3Omb2f9DT9GVQJbmV+Jrg2SENeBch3vnWhdfTk0dcwIUZID2alVFOceQdf9QxBV1ULXDJbO2oDL0ReScwgxHlLREUKIMOjpz+C1HZ8nOb4VTfPR2ZsbdN5OOCS627lj6U+HV6BXFZNTNSs4eHYdU3HtnuuVkVKHpgbe7FqWRkpiZK750z+YjGXXjUrOwH8DO+BNDMk5aprnsLz8zYDtNgqVjQtCco5rkZLQRH7m2UvaL5skxHUxLfs4VU1XF5OqGqyY/RrTc45i2Sq2rXLgzJ2cq4/dKu20nONomi9gu22r5Gee5XzD+NY8EmK8IuOqKoQQU5JCd38WHT35EZPkgM2ti3+J29WNQ/fi0L1omklZ4V4Ks06FO7io0tGTi2kFfl9V1aS7PyMMEV3ZyeqVWJcsimlZKl29WSGbd+Ez4th86HF8hhOvz4nXcGKYOjuOPRSW5C8zpQ7swAbMDt1LdlrlVR9nxezfU5RzDE0zceg+nI5BlpW/RX7GmRBGG1ksa+y5ZpfbJ8RkkYqOiEgXu6w1YmW+Ia2khZgkyQmtJLi6R80dAv9k8VmFe6htmR2ewKLQqeqVlOYfQLtkzZ/G9hL6PGlhjGxsrV1F7D15N8vK38IGVMWioyeXLYcfD+l5GttLeXHz18hJO4+i2jS1F4dtPoxnMClo5z7D1OkbSL2qYzi0AaZfSHJG0jUf80q2UB+SSf+Rp7JxIWWF+1EvqeooikVd68R8zfGuLuYWbyM7rYpeTyrHK2/8/9u79+A4r/O+49/n3V3cQVy4IEDwAhIMKYIXi1JE6mZLtE0rslJLjuOo6kxTuWPZSS+TdNrpxHFmmpn8peSPdKaetI3qulLbWKUbK47sxFYjyRYtSiJEUaQI8QJSIHEhQYIEiAtx3d339I9dkgCxIEAC2Hd38fvMYLj77uG7D3Dw7uLZc85zuDywdlGeS3KfEh3JKqqyJhKsSGgcf4aPFSLhsQxHk9uGx6p444Nn2bn576he1k3CD9N2fgcfnvpC0KHdUlv3PZy9sJ2Ksh4mYiVz/mP/diX8yIKXrb4T3X0biMULCXmxKeWXnTPazu+Y0zkKC0bwnUeI6VMVS4ryd43blaGVHG37DNsb9wHJn5mZY//R31yUxLW06ApfvP+/EQpNEPJ8KkovUVd9lvc+/hIdPdsW/Pkk9ynRkawxea8cVVkTCcaVoZWQZqF4PBHO2w0VF4JnCUqL+hmbKCWWuLHQu2+ontfe/wZmfmoBfm6MTfsuzJWh+qDDyAjnPP7hg6/xme3/l4rSyziM8YkS9rf8BmMT5XM6x/BYRdrpp75vXOrP73LZx9s/TfuF7dRHT5Hww3RduotYvHhRnutTjb9IVjFMJaRmyVGz+zb/jM6eLdpTS6ZRoiNZ4dpIzqGS5fi1r3OkWqM4IndqWcllVq84juHo7GlicKRmzv/XdyEOHP8SD279EZ4l8DxHLB5heKyKU10qF5vOxtXN3P0rP8fw8cyn/eI2mo//+pR9frJnDZakMzxaxc+av0lx4SAhL87V0SpuJyl1LsTh05/j3k2vXy9q4PtGwo9wtO3RRYo6e4yMV2SknHRt9dkpo27XhEMxSooGF230UXKXEh3JGuWRG5+cWd3CVPcRWWqaGvazvfEtPPMBx9b1b9PS9gjH2j8953N09mxhYLiGjasOUlw4xPnejZy9sB3f11vGzVbXnOCejW9Mqdi1tvZjnDMOHH8ywMjkTozOY8Pe0+d2Mjq+jK3r36akcJBL/Wv4qG03Q9o8c8GMTZRSUjQ07biZz0RsYUpmS37Ru5aISJ4oK+5je+NbU/ZB8YizrXEfnZeaGLqNSl+DwzV80PrFxQgzr2xbv29KkgPJvXLW1bXwQevjxBMFAUWW3QojwxRGRhkarc6r0a5zl+/i3OW7gg4jbx1vf4j7m35MOHzjmkskQnRd3jRlyqjINfNKdMysGtgLrAPOAk87566kaZcAjqbudjjn9DGXTLHZneQ7zRXseaWd/euWsTize0Xy2+qaExjTp3WY+ayuOcHx9ocDiCq/FRdeTXvcYRSER5Xo3KQgPMpD235IbVU7vgvhO4/3TzxBx0UtJM814dA4m9Y0s6bmBBPxIlo7dy16ktd+cStlxX1sXf82vh8i5CXo7lvPgWNPLerzSu6a74jOt4A3nHPPm9m3Uvf/IE27Uefcjnk+l+Sha2tzzv5hP7tr9rJ/3XKK11aqAIHIHZlhTYGztDvRy/xd6l/N6poT08pxJ/wwo3NcyL6UPHL3XpZXdBHy/OsVyh7Y8mOGRyvpHVwdcHQyVyFvgsd3fZeSooHrI8jRii5OdDzA0bbPLuIzGx+ffYSTnQ+wrOQyo+Plus7kluY7XvwU8FLq9kvAl+d5PllC7u7oZd/BTjq+PciBDc3JKmtKckTuWGdPEy5NsuMwOi81BRBR/vuo7bMk/AJ8/8bPPZ6IcKj1C3k1JWshlBX3Ub3s/JR9hQA8L0ZTw7sBRSV3orH+MCWFg1OmyUbCMbY0vEthZHjRnz+eKKBvqF5Jjsxqvq/Ctc657tTtC0DtDO2KzOygmb1nZl+e6WRm9s1Uu4MDfQPzDE2yVWFnlLs7ennhg5LUXjnNFDdUU7V7j5IckXkYHqvkUOsXiCfCU74+PLUnazeozHWDwzW81vwcnT1NDI8to+fKGn750W9xpntH0KFlneKCq/h+aNpxz6C0SO/5uWRV9NSUdTLXJPwQ0YquACISSW/WqWtm9jpQl+ahP5p8xznnzGz65PCkBufcOTNrBN40s6POuU9ubuScewF4AWDj9o0znUvyQHmknLKSAghFNIojsoBOn9vJ+cubWL3iJODo6tnMyHhF0GHltcGRKPtbvhp0GFmv/+oKPG/6hpqJRIgLfesyH5DcsZHxcnzfppV6NnOMTZQGFJXIdLMmOs65GTczMbOLZrbSOddtZiuBnhnOcS71b5uZ/QK4B5iW6IiIyPyNjFfQ2rkr6DBEpoglijh29mGaGt4hkhoNSPgesUQhJzoeDDg6uR2nunayrq4FjxtT13zfGJsopXdwVYCRiUw136lrrwLPpm4/C/ztzQ3MrMrMClO3o8DDwLF5Pq+IiIjkmJYzj/Lesae4PLCKoZEqTp+7l5++9zuMxzQKkEuuDK3kwLEvEYsXMBEvIJ6IMDgS5c1Dv83tbLS6kMqLe/lU48/51U0/pa76E0hTgVKWnvlWXXse+IGZfR1oB54GMLP7gN91zj0HNAF/aWY+ycTqeeecEp0l6lqVtY///SAPRX/KgYrlKiUtIrKEdPZsobNnS9BhLHmFkRGiFZ2TRmFuL0Fpv7idzp4mqsovEIsXMRjgxqjr6j5iV9NPMPPxzKex/jAX+hr55UdPE1TiJdlhXomOc64X+Hya4weB51K33wG2z+d5JD8kCxCMUvLDQd7d0ExRQS3FdWVanyMiIpJB29a/xdZ1+0n4oevrat489E8ZHru9oiW+CwdeFjwcGmdX00+mVYCrq25jdc1Jui5tDjA6CZpqX8qiu1ZlrePbgzeqrK2tVJU1ERGRDKtffootDe8QCsUpiIwTCU9QWtTPozteJhene9VWn01bzS8SjtFQ2xJARJJNlOjIoro2Ve3FDyvwa1/XXjkiIiIB2rSmeVppaM9zlBUPsKz0ckBR3Tnf90iXoPkuuXGvLG1KdGTRlUfKKShKvtiYpqqJiMgSURAZoay4j+Qy5exQEBlNe9x3HgXhsQxHM38X+9aTbh2On4jQpv2sljyluiIiIiILKBIa48Ftf8PK5W043yPhh2k+8QSdPVuDDo3OniYqy3qmrGkBMBx9QysDiurO+S7MW0eeYfeOl5MHzOHhONFxPz1X1gUamwRPiY5kxCPfbyceXEEWEZEs56ip7KCsuJ8rQ7X0X023T7fkis986gfUVHYS8hLgJQgT48GtrzIyVhH44v1TXffRWH+Y0sJBwuEYvm/4Lsz7J57Az9GpXpf6G/ibX/5bVkVbCYfHudC7geGxyqDDkiyQm7/RkhNuVFm7kKqyVq0ijyIigGcJ1tZ+zKpoKxPxQmqrz1JcMAw4zBw9VxrYd+Qf4zu9Teea0qJ+opVdhEKJKcc9L0ZTwzu8ffTpgCJLiicK+dmBb9BYf4RV0VZGxss51bWTKzk4mjNZPFFA+8VtQYchWUavoLLgrhUg6Pj2II9F9/LuhuUqQCAikhLyYuy570WWlVwmEo7hUuuobdInQSuq2tm6/m2Otu0OJMaloKRwgLKSKwwORxmbKFuw8xYXDiWrgN00NcwzKCvuX7DnmY+EX8Cprp2c6toZdCgii0qJjiwo7ZUjInJrjfWHqSi9TDiUrHxlaYa6w6E4G1YdUqKzCDwvzkPbXqF++Wl8P0TIi3P2wnaaj/8j3ALUaBq4WoPnJaYdTyQ8LvStn/f5RWTulOjIginsjFIemaCspAC/9nWK165VgiMicpOG2pbrSc6thLz4tGNV5edpXHmEUChO58Umuvs2sFR2fvcsgXM272Tk3o2vUb/8dHIxfmrUpaHuY4ZGqjnW/ul5xxlLFHHs7MNsWffO9X72fSOWKOREh94TRTJJiY6IiEgGxRKFs7bxfY9zl+4CoHrZOeqqz1Bdfo766Cd4lsDzHA21LZy/vIn9LV8hn5OdqvJudjX9hKryCzjf4+yFbXzQ+jjxND/HSGiMUCiWmoqW7mfiaKw/Mq3iWDgU4661BxYk0QFoOfMIgyNRmhreoSgyQndfIy1tjzA2Ub4g5xeRuVGiIyIikkGnOu9jRWU7kUmbNk5epxOLR4jFizh8+nM8uPUV1qw4iWcxzKZOc4uEY9RHW6mtPsPFvsYMfxeZUVI4wJ5ffYlIeCJ5IJRgXV0LZcX9vHHo2evtCiMjPLD1R9RVnwFnjIyXceDYk/T0r5tyPs8SyUpoaUTC4wsYudFxcSsdF4MvJy2ylGnDUFkwm91JvtPs8cj32zlQtizocEREstL53o20du4ikQgRixcwES9gbKKEkx27ONO9ncOn9/CTd/8lNZWdrK45STgUw/NmWssTY03N8cx/Exmycc37eDY1MQmFEiyvOE9FaU/qiOOz9/xv6qrbCHkJQqE45SX9PLrjZcqK+6b8X9+FGRievteBc3Cpf+1ifRsiEhCN6Mi8Ta6ytju6l/3rVGVNRGRmxpFPPk9r105WVHYwHivm4pX1ODf1s8fGlUemjPqk45wRTxQsZrCBqiy7OK1MM4DvPMpL+hgYXkFV+QXKS3oJef6UNp6XYNPq9zl06temHG8+/ut87t6/wrM4nudI+B6+H+ZQ62OL+r2ISOYp0ZF5UZU1EZE7Mzq+7Nb7fpib9Ry+C3Gm++4FjCq79A6sorbqLOGb96SxBP1XVwBQWjSQtkBByPMpL+2ddvzywFp+duAbbG54h8qyS/QO1HOi40FtMCmSh5ToyB27u6OXFz+s4LG3XtBeOSIiC6zt/I5pa3kgOc0qnijAzOfDU3sYGF4RUISL71TXTu5a24xnPp6XTPziiTDdvRu4OloNQN9Q3bTpbdfa9VxpSHvewZEozcefXLzARSQrKNGReSkoCkMooiRHRGSBdfY0sXbFcVZFW/FCMXw/+ZZ9rP0hro4sp7u3kfFYacBRLq7xWCmvNT/HvZv+H3XVZ4gnIpzuupeWM49ebzMyVknHxa2sWXHselLo+x6xeBGnz90bVOgikgWU6IiIiGQlY3/LV1i+7Bx1y9uYiBXTcXEr47GSoAPLqKuj1ew78swt2xw49iX6BuvYtOZ9IuFxui5t4mjbbmLx4gxFKSLZSImOiIhI1jJ6B1fTO7g66ECymsOjtet+WrvuDzoUEckiSnTkthV2Rtl3sJOqHw7yUHQvByqWo8/MRERERCSbKNGR23KtytoXfvE93t0QpaiglpW79wQdloiIiIjIFEp0ZE4mj+I8Ft3LgcqoChCIiIiISNZSoiNzstmd5D2vAq9kGR/WaBRHRERERLLb9B22RGZQUKS8WERERERygxIdERERERHJO0p0ZFaFnVFe+KCER77fTrz8laDDERERERGZleYiyYwKO6Nsdid54YNRHnvrf7B/3XKKCqqp0vocEREREclySnQkrWtV1kpSVdaO1NRSXFemKmsiIiIikhOU6Mg01/bKeeJHPezf0ExRQa1GcUREREQkp5hzLugY0jKzS0B70HEEKApcDjoImTP1V25Rf+Ue9VluUX/lFvVXblF/TdfgnKu5+WDWJjpLnZkddM7dF3QcMjfqr9yi/so96rPcov7KLeqv3KL+mjtVXRMRERERkbyjREdERERERPKOEp3s9ULQAchtUX/lFvVX7lGf5Rb1V25Rf+UW9dccaY2OiIiIiIjkHY3oiIiIiIhI3lGikyXM7LfM7GMz881sxkoaZva4mZ00s9Nm9q1Mxig3mFm1mf2DmZ1K/Vs1Q7uEmR1Ofb2a6TiXutmuFzMrNLO9qccPmNm6AMKUlDn019fM7NKka+q5IOKUJDP7npn1mFnLDI+bmf2nVH9+ZGb3ZjpGuWEO/bXbzAYmXV//IdMxyg1mtsbMfm5mx1J/H/5+mja6xmahRCd7tABfAfbN1MDMQsBfAF8EtgD/xMy2ZCY8ucm3gDeccxuBN1L30xl1zu1IfT2ZufBkjtfL14ErzrlfAf4j8KeZjVKuuY3Xt72TrqnvZjRIudmLwOO3ePyLwMbU1zeB/5KBmGRmL3Lr/gL45aTr608yEJPMLA78O+fcFuAB4F+leU3UNTYLJTpZwjl33Dl3cpZmu4DTzrk259wE8H+ApxY/OknjKeCl1O2XgC8HF4rMYC7Xy+R+/Gvg82ZmGYxRbtDrW45xzu0D+m7R5Cngf7qk94BKM1uZmejkZnPoL8kizrlu59yh1O0h4Diw6qZmusZmoUQnt6wCOifd72L6L71kRq1zrjt1+wJQO0O7IjM7aGbvmdmXMxOapMzlernexjkXBwaA5RmJTm4219e330xN0fhrM1uTmdDkDuk9K/c8aGZHzOynZrY16GAkKTWt+h7gwE0P6RqbRTjoAJYSM3sdqEvz0B855/420/HIrd2qvybfcc45M5upfGGDc+6cmTUCb5rZUefcJwsdq8gS8WPgZefcuJn9DsnRuM8FHJNIvjhE8j3rqpk9AfyI5JQoCZCZlQE/BP6Nc24w6HhyjRKdDHLO7ZnnKc4Bkz/BXJ06JovgVv1lZhfNbKVzrjs1TNwzwznOpf5tM7NfkPxERolOZszlernWpsvMwkAF0JuZ8OQms/aXc25y33wX+LMMxCV3Tu9ZOWTyH9HOub83s/9sZlHn3OUg41rKzCxCMsn5K+fcK2ma6Bqbhaau5Zb3gY1mtt7MCoBnAFXyCsarwLOp288C00bkzKzKzApTt6PAw8CxjEUoc7leJvfjV4E3nTYXC8qs/XXT3PMnSc5Zl+z1KvDPUpWhHgAGJk35lSxjZnXX1iia2S6SfyPqg5+ApPrivwPHnXN/PkMzXWOz0IhOljCz3wC+A9QAf2dmh51zv2Zm9cB3nXNPOOfiZvavgdeAEPA959zHAYa9lD0P/MDMvg60A08DWLI0+O86554DmoC/NDOf5BvG8845JToZMtP1YmZ/Ahx0zr1K8k3kf5nZaZKLdJ8JLuKlbY799Xtm9iTJakR9wNcCC1gws5eB3UDUzLqAPwYiAM65/wr8PfAEcBoYAf55MJEKzKm/vgr8CzOLA6PAM/rgJ1APA78NHDWzw6lj3wbWgq6xuTL9DouIiIiISL7R1DUREREREck7SnRERERERCTvKNEREREREZG8o0RHRERERETyjhIdERERERHJO0p0REREREQk7yjRERERERGRvKNER0RERERE8s7/B8ouK9dFmMF1AAAAAElFTkSuQmCC",
"text/plain": [
"<Figure size 1008x432 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"fig, ax = plt.subplots(figsize=(14, 6))\n",
"_ = boundary(X_test, ax)\n",
"plt.scatter(X_test[:,0], X_test[:,1], c=y_test,cmap='viridis')"
]
}
],
"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": "3.9.4"
}
},
"nbformat": 4,
"nbformat_minor": 5
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment