example/demo_attention.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Self-Attention Model Demonstration\n",
    "\n",
    "## Introduction\n",
    "\n",
    "This notebook demonstrates the usage of the NLarge library for data augmentation and text classification using a self-attention model. The primary goal is to showcase how self-attention mechanisms can be leveraged to improve the performance of Natural Language Processing (NLP) tasks, specifically sentiment analysis.\n",
    "\n",
    "The NLarge library provides various data augmentation techniques, including traditional methods like synonym replacement and random substitution, as well as advanced techniques using Large Language Models (LLMs). In this notebook, we will focus on using the self-attention mechanism to enhance the model's ability to capture contextual dependencies in the text.\n",
    "\n",
    "We will use the `TextClassifierAttentionNetwork` class, to classify text data. The model will be trained on augmented data to demonstrate the effectiveness of data augmentation in improving model performance.\n",
    "\n",
    "## Steps\n",
    "\n",
    "1. **Data Augmentation**: We will augment the training data using the NLarge library to increase the diversity and robustness of the dataset.\n",
    "2. **Model Initialization**: We will initialize the `TextClassifierMultiHeadAttentionNetwork` model with pretrained embeddings.\n",
    "3. **Training**: The model will be trained on the augmented dataset.\n",
    "4. **Evaluation**: We will evaluate the model's performance on the test dataset.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import datasets\n",
    "from datasets import Dataset, Features, Value, concatenate_datasets\n",
    "from NLarge.dataset_concat import augment_data, MODE\n",
    "from NLarge.pipeline import TextClassificationPipeline\n",
    "from NLarge.model.Attention import TextClassifierAttentionNetwork"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'text': 'the rock is destined to be the 21st century\\'s new \" conan \" and that he\\'s going to make a splash even greater than arnold schwarzenegger , jean-claud van damme or steven segal .',\n",
       " 'label': 1}"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "original_train_data, original_test_data = datasets.load_dataset(\n",
    "    \"rotten_tomatoes\", split=[\"train\", \"test\"]\n",
    ")\n",
    "\n",
    "features = Features({\"text\": Value(\"string\"), \"label\": Value(\"int64\")})\n",
    "original_train_data = Dataset.from_dict(\n",
    "    {\n",
    "        \"text\": original_train_data[\"text\"],\n",
    "        \"label\": original_train_data[\"label\"],\n",
    "    },\n",
    "    features=features,\n",
    ")\n",
    "\n",
    "original_train_data[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Skipping augmentation for text due to error: Sample larger than population or is negative\n",
      "Skipping augmentation for text due to error: Sample larger than population or is negative\n",
      "Skipping augmentation for text due to error: Sample larger than population or is negative\n",
      "Skipping augmentation for text due to error: Sample larger than population or is negative\n",
      "Skipping augmentation for text due to error: Sample larger than population or is negative\n",
      "Skipping augmentation for text due to error: Sample larger than population or is negative\n",
      "Original train size: 8530\n",
      "Train size after 100% augmentation: 17052\n"
     ]
    }
   ],
   "source": [
    "# Augment and increase size by 100%\n",
    "percentages = {\n",
    "    MODE.RANDOM.SUBSTITUTE: 0.5,  # 50% of data for random augmentation\n",
    "    MODE.SYNONYM.WORDNET: 0.5,  # 50% of data for synonym augmentation\n",
    "}\n",
    "\n",
    "augmented_data_list = augment_data(original_train_data, percentages)\n",
    "\n",
    "\n",
    "# Convert augmented data into Datasets\n",
    "augmented_dataset = Dataset.from_dict(\n",
    "    {\n",
    "        \"text\": [item[\"text\"] for item in augmented_data_list],\n",
    "        \"label\": [item[\"label\"] for item in augmented_data_list],\n",
    "    },\n",
    "    features=features,\n",
    ")\n",
    "\n",
    "# Concatenate original and augmented datasets\n",
    "augmented_train_data = concatenate_datasets(\n",
    "    [original_train_data, augmented_dataset]\n",
    ")\n",
    "\n",
    "print(f\"Original train size: {len(original_train_data)}\")\n",
    "print(f\"Train size after 100% augmentation: {len(augmented_train_data)}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We will now train the model on the augmented dataset using the `TextClassificationPipeline` class. The augmented dataset, which combines the original training data with the augmented data, will be used to initialize the pipeline. The model, `TextClassifierAttentionNetwork`, will be trained for 10 epochs on this dataset. This training process aims to leverage the increased diversity and robustness of the augmented data to improve the model's performance on the text classification task. The training process is initiated by calling the `train_model` method on the pipeline object, specifying the number of epochs."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "8aa8e5e60bc84eb1b32519bd5b2fe709",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Map:   0%|          | 0/17052 [00:00<?, ? examples/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0ac44b5c95bb47e78ad5018ee0cc92b3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Map:   0%|          | 0/1066 [00:00<?, ? examples/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "84ca13981f564f5fb5e3d079f3e27bdb",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Map:   0%|          | 0/13641 [00:00<?, ? examples/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7b9a2b87fe144b6898fe768bd465fee8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Map:   0%|          | 0/3411 [00:00<?, ? examples/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0ef9b4c642d846fea9cfb72306b498b3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Map:   0%|          | 0/1066 [00:00<?, ? examples/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "training...: 100%|██████████| 27/27 [00:02<00:00, 11.14it/s]\n",
      "evaluating...: 100%|██████████| 7/7 [00:00<00:00, 31.67it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 1/10\n",
      "Train Loss: 0.673, Train Acc: 0.508\n",
      "Valid Loss: 0.636, Valid Acc: 0.567\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "training...: 100%|██████████| 27/27 [00:02<00:00, 11.43it/s]\n",
      "evaluating...: 100%|██████████| 7/7 [00:00<00:00, 31.68it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 2/10\n",
      "Train Loss: 0.580, Train Acc: 0.711\n",
      "Valid Loss: 0.546, Valid Acc: 0.753\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "training...: 100%|██████████| 27/27 [00:02<00:00, 11.57it/s]\n",
      "evaluating...: 100%|██████████| 7/7 [00:00<00:00, 30.70it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 3/10\n",
      "Train Loss: 0.510, Train Acc: 0.801\n",
      "Valid Loss: 0.515, Valid Acc: 0.794\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "training...: 100%|██████████| 27/27 [00:02<00:00, 11.43it/s]\n",
      "evaluating...: 100%|██████████| 7/7 [00:00<00:00, 30.88it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 4/10\n",
      "Train Loss: 0.473, Train Acc: 0.843\n",
      "Valid Loss: 0.494, Valid Acc: 0.816\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "training...: 100%|██████████| 27/27 [00:02<00:00, 11.28it/s]\n",
      "evaluating...: 100%|██████████| 7/7 [00:00<00:00, 31.84it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 5/10\n",
      "Train Loss: 0.446, Train Acc: 0.871\n",
      "Valid Loss: 0.481, Valid Acc: 0.829\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "training...: 100%|██████████| 27/27 [00:02<00:00, 11.67it/s]\n",
      "evaluating...: 100%|██████████| 7/7 [00:00<00:00, 31.97it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 6/10\n",
      "Train Loss: 0.426, Train Acc: 0.891\n",
      "Valid Loss: 0.472, Valid Acc: 0.838\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "training...: 100%|██████████| 27/27 [00:02<00:00, 11.82it/s]\n",
      "evaluating...: 100%|██████████| 7/7 [00:00<00:00, 30.60it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 7/10\n",
      "Train Loss: 0.411, Train Acc: 0.906\n",
      "Valid Loss: 0.465, Valid Acc: 0.844\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "training...: 100%|██████████| 27/27 [00:02<00:00, 10.20it/s]\n",
      "evaluating...: 100%|██████████| 7/7 [00:00<00:00, 30.56it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 8/10\n",
      "Train Loss: 0.400, Train Acc: 0.917\n",
      "Valid Loss: 0.460, Valid Acc: 0.853\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "training...: 100%|██████████| 27/27 [00:02<00:00, 11.10it/s]\n",
      "evaluating...: 100%|██████████| 7/7 [00:00<00:00, 24.88it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 9/10\n",
      "Train Loss: 0.391, Train Acc: 0.926\n",
      "Valid Loss: 0.457, Valid Acc: 0.852\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "training...: 100%|██████████| 27/27 [00:02<00:00, 11.22it/s]\n",
      "evaluating...: 100%|██████████| 7/7 [00:00<00:00, 30.42it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 10/10\n",
      "Train Loss: 0.385, Train Acc: 0.932\n",
      "Valid Loss: 0.454, Valid Acc: 0.854\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "pipeline_augmented = TextClassificationPipeline(\n",
    "    augmented_data=augmented_train_data,\n",
    "    test_data=original_test_data,\n",
    "    max_length=128,\n",
    "    test_size=0.2,\n",
    "    model_class=TextClassifierAttentionNetwork,\n",
    ")\n",
    "pipeline_augmented.train_model(n_epochs=10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Plotting the training and validation loss and accuracy curves will help us understand the model's performance during training. We will use the `plot_metrics` function to visualize these curves."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA04AAAIjCAYAAAA0vUuxAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjkuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8hTgPZAAAACXBIWXMAAA9hAAAPYQGoP6dpAACaeklEQVR4nOzdd3hT1f8H8HeSJmnTNt27hdKyZ6G0CDJFtixFlsoQgZ+CosAXxYkLB4ooipMliCDKUhCEIrI3ZbdsCl100D2SJuf3R9pAaAttmjYd79fz8CQ5uTn35DRA3r3nfq5ECCFAREREREREpZJaewBERERERETVHYMTERERERHRAzA4ERERERERPQCDExERERER0QMwOBERERERET0AgxMREREREdEDMDgRERERERE9AIMTERERERHRAzA4ERERERERPQCDExHVet27d0f37t2tPQwiqibmzJkDiURi0lZQUIBZs2YhICAAUqkUQ4YMsc7gqtCyZcsgkUhw7do1aw+FqEZgcCKqAXbt2gWJRFLin4MHDxbbfv/+/ejcuTNUKhW8vb3x0ksvISsry2Sb2NhYDBgwAGq1Gs2bN8eff/5ZrJ9169bB09MT6enpZRpn0ZeRoj9yuRyBgYF46aWXkJaWZtZ7r6327t2Lfv36wc/PD7a2tqhXrx4GDhyIVatWWXtoFVbaZ/Xjjz8utm1sbCyGDx8OZ2dnqNVqDB48GFeuXDHZJj8/Hy+++CI8PDzg7++PDz74oFg/N2/ehIODA/bt21fmcV67dg3jx49HcHAwbG1t4e3tja5du+Kdd94p/5umGm/JkiWYN28ehg0bhuXLl+OVV16pcJ+rVq3CggULirXHxcVhzpw5iIyMrPA+ymLu3LnYsGFDleyLqDazsfYAiKjsXnrpJYSFhZm0NWzY0ORxZGQkevbsiWbNmmH+/Pm4efMmPvvsM1y8eBF///23cbuxY8ciNjYWn3zyCfbt24cnn3wSUVFRCAwMBADk5eVh5syZ+OCDD+Dk5FSucX777bdwcHBAdnY2IiIisHDhQhw/fhx79+41743XMmvXrsWIESMQEhKCadOmwcXFBVevXsXu3bvx448/YvTo0dYeYoX16tULY8aMMWlr27atyeOsrCz06NED6enpeP311yGXy/HFF1+gW7duiIyMhJubGwBg3rx5+Pnnn/HGG28gMzMT7733HoKDgzFq1ChjX//73/8waNAgPPzww2Ua36VLlxAWFgY7Ozs8++yzCAwMRHx8PI4fP45PPvkE7777bgVngGqanTt3ws/PD1988YXF+ly1ahXOnDmDl19+2aQ9Li4O7777LgIDAxESEmKx/ZVm7ty5GDZsWLGjaM888wxGjhwJpVJZ6WMgqg0YnIhqkC5dumDYsGH33eb111+Hi4sLdu3aBbVaDQAIDAzExIkT8c8//6B3797Izc3Fzp07sWvXLnTt2hX/93//h/3792Pbtm2YPHkyAOCzzz6Dk5MTnnvuuXKPc9iwYXB3dwcATJ48GSNHjsSaNWtw+PBhhIeHl7u/2mbOnDlo3rw5Dh48CIVCYfLcrVu3rDQqy2rcuDGefvrp+26zaNEiXLx4EYcPHzb+QqBfv35o2bIlPv/8c8ydOxcA8Ndff2HGjBmYNWsWAODGjRvYtGmTMTjt3bsXf/75J6Kioso8vi+++AJZWVmIjIxE/fr1TZ6rLT8DKp9bt27B2dnZ2sOoUjKZDDKZzNrDIKoxuFSPqIbJzMxEQUFBic9lZGRg+/btePrpp42hCQDGjBkDBwcH/PbbbwAMR5OEEHBxcQFgWFrl7OyMnJwcAIblUx9//DG+/PJLSKUV/2eiS5cuAIDLly8b21JTUzFz5ky0atUKDg4OUKvV6NevH06ePGny2qJlir/99hs+/PBD+Pv7w9bWFj179sSlS5eK7euHH35AcHAw7OzsEB4ejj179pQ4plu3bmHChAnw8vKCra0t2rRpg+XLl5tsc+3aNUgkEnz22Wf45ptvEBQUBJVKhd69e+PGjRsQQuD999+Hv78/7OzsMHjwYKSmpj5wPi5fvoywsLBioQkAPD09jffbtWuHxx9/3OT5Vq1aQSKR4NSpU8a2NWvWQCKR4Pz58wCA69ev44UXXkCTJk1gZ2cHNzc3PPnkk8XOYyg6v2H37t2YPHky3NzcoFarMWbMGNy+ffuB7+NBcnNzkZeXV+rzv//+O8LCwkyOojZt2hQ9e/Y0flaL+in6rAKAq6ur8bOq1+sxbdo0zJo1C/7+/mUe2+XLl+Hv718sNAGmP4OxY8fC3d0dWq222Ha9e/dGkyZNjI8lEgmmTp2KDRs2oGXLllAqlWjRogW2bt1a7LUnTpxAv379oFar4eDggJ49exZbdlv089m3bx+mT58ODw8P2NvbY+jQoUhKSqrQGNeuXYvmzZvDzs4OHTt2xOnTpwEA33//PRo2bAhbW1t07969zOe+WPL9lCYhIQHjx4+Hv78/lEolfHx8MHjw4GJj/Pvvv9GlSxfY29vD0dERAwYMwNmzZ0vtt+jv+b///ouzZ88al5bu2rWr1Nds3LgRAwYMgK+vL5RKJYKDg/H+++9Dp9MZt+nevTs2b96M69evG/sMDAzErl27jJ/58ePHG59btmyZ8bWHDh1C37594eTkBJVKhW7duhVbhlq0NPrSpUsYN24cnJ2d4eTkhPHjxxv/fgCGn3l2djaWL19u3Ne4ceMAlH6O06JFi9CiRQsolUr4+vpiypQpxZZbd+/eHS1btsS5c+fQo0cPqFQq+Pn54dNPPy113ohqPEFE1d6///4rAAgHBwcBQMhkMtG9e3dx5MgRk+327t0rAIg1a9YU66Nz586iXbt2xsfBwcFi5MiR4sqVK2LlypVCIpGIvXv3CiGEGD16tBg2bFi5x/nOO+8IACIpKcmkfebMmQKA+Pvvv41tR44cEcHBweK1114T33//vXjvvfeEn5+fcHJyErGxscXee9u2bUVoaKj44osvxJw5c4RKpRLh4eEm+/npp58EANGpUyfx1VdfiZdfflk4OzuLoKAg0a1bN+N2OTk5olmzZkIul4tXXnlFfPXVV6JLly4CgFiwYIFxu6tXrwoAIiQkRDRv3lzMnz9fvPnmm0KhUIiHHnpIvP7668Z9vfTSS0IikYjx48c/cJ4aN24sAgICxI0bN+673UsvvSQ8PDyMj1NSUoREIhFSqVR8/fXXxvYpU6aYbLd27VrRpk0b8fbbb4sffvhBvP7668LFxUXUr19fZGdnG7dbunSpACBatWolunTpIr766isxZcoUIZVKRdeuXYVer3/geykJAGFvby8kEokAIJo1ayZ++eUXk210Op1QKpXi+eefL/b6N998UwAQGRkZQgghJkyYIFq2bClOnTol9u/fL7y9vcUHH3wghBDihx9+EPXq1RM5OTnlGuOkSZOETCYTERER991u+/btAoD4888/Tdrj4+OFTCYT7733nsn7btOmjfDx8RHvv/++WLBggQgKChIqlUokJycbtztz5oywt7c3bvfxxx+LBg0aCKVSKQ4ePGjcrujn07ZtW/HII4+IhQsXihkzZgiZTCaGDx9u9hhbt24tAgICxMcffyw+/vhj4eTkJOrVqye+/vpr0bx5c/H5558bP+c9evR44Fxa+v2UplOnTsLJyUm8+eab4qeffhJz584VPXr0EP/9959xm59//llIJBLRt29fsXDhQvHJJ5+IwMBA4ezsLK5evWrcrujfKiGEyMrKEitWrBBNmzYV/v7+YsWKFWLFihUiISGh1LEMGTJEDB8+XMybN098++234sknnxQAxMyZM43b/PPPPyIkJES4u7sb+1y/fr1ISEgQ7733ngAgJk2aZHzu8uXLQgghIiIihEKhEB07dhSff/65+OKLL0Tr1q2FQqEQhw4dKvYe2rZtKx5//HGxaNEi8dxzzwkAYtasWcbtVqxYIZRKpejSpYtxX/v37zf5mZQ0N48++qhYuHChmDp1qpDJZCIsLExoNBrjdt26dRO+vr4iICBATJs2TSxatEg88sgjAoDYsmXLA3+eRDURgxNRDbBv3z7xxBNPiMWLF4uNGzeKjz76SLi5uQlbW1tx/Phx43Zr164VAMTu3buL9fHkk08Kb29v4+OIiAjh4uIiAAgA4uWXXzbuy87OTly7dq3c4yz6Dzc6OlokJSWJa9euiSVLlgg7Ozvh4eFh8qU9Ly9P6HQ6k9dfvXpVKJVKky96RcGpWbNmIj8/39j+5ZdfCgDi9OnTQgghNBqN8PT0FCEhISbb/fDDDwKASXBasGCBACBWrlxpbNNoNKJjx47CwcHB+IW9KDh5eHiItLQ047azZ882fknWarXG9lGjRgmFQiHy8vLuO0+LFy8WAIxfTN966y2xZ8+eYvNR9PM8d+6cEEKITZs2CaVSKQYNGiRGjBhh3K5169Zi6NChxsclhYgDBw4IAOLnn382thV9aQoNDTX5QvTpp58KAGLjxo33fR+l6dSpk1iwYIHYuHGj+Pbbb0XLli0FALFo0SLjNklJSQKAyc+6yDfffCMAiKioKCGEEDdu3BAtWrQwfla7dOkiMjMzRVpamvDw8BCrV68u9xjPnDkj7OzsjMF42rRpYsOGDSafUSEMAc/f399kvoUQYv78+UIikYgrV64Y24p+ppcuXTK2nTx5UgAQCxcuNLYNGTJEKBQK4xdlIYSIi4sTjo6OomvXrsa2op/Po48+ahJiX3nlFSGTyYyfyfKOUalUmnxR/v777wUA4e3tbfzsC3Hnc373tiWx9Pspye3btwUAMW/evFK3yczMFM7OzmLixIkm7QkJCcLJycmk/e7gVKRbt26iRYsW932vRUr6OzZ58mShUqlM/v4PGDBA1K9fv9i2R44cEQDE0qVLTdr1er1o1KiR6NOnj8kc5eTkiAYNGohevXoVew/PPvusSR9Dhw4Vbm5uJm329vZi7NixxcZxb3C6deuWUCgUonfv3ib/Hn399dcCgFiyZImxrVu3bsX+TcnPzxfe3t7iiSeeKLYvotqAwYmohrp48aKws7MTffr0Mbb9/PPPAoDJbyWLPPPMM8LJycmkLTMzUxw8eFDExMQIIQxfwEJDQ8Wbb74phBBi0aJFokmTJqJx48bi22+/feCYiv4jv/dPq1atxNGjR0t9XUFBgUhOThZJSUmidevWYsiQIcbnioLTp59+avKa48ePm3y5379/vwAgvvvuO5PtNBqNcHJyMglOvXv3Ft7e3sWCyq+//mrym/ui4PTCCy+YbLdhw4YSv8QVBbK7v0CWZuvWraJ3795CLpcb5ykoKEjs27fPuE1CQoLJe/rf//4nOnfuLBYuXCh8fX2FEIYvlFKpVMyfP7/E/Wg0GuPcOjs7GwOyEHe+NH3//fcmr8nMzBQ2NjZi8uTJD3wfZZGfny9atmwpnJ2djV84Y2JiBADxySefFNu+KFieOHHC5H2cOHFCnD171vhze+WVV0Tnzp2FEELs2bNHhIeHC39/f/Hiiy+ahOfSREdHi6efflo4OzsbfwYODg7ihx9+MNnu1VdfFXZ2diahIjQ0VDz88MMm2wEQ/fv3L7YftVotXnnlFSGE4bOuUqlKPMIyefJkIZVKRXp6uhDizs/nt99+M9lu3bp1AoA4efKkRcYYGRkpAIgpU6aYtBd9zu93VK6y3s+98vLyhEKhEAMGDBCpqaklblPUz86dO0VSUpLJn969e4uGDRsat61ocLpbRkaGSEpKEitXrhQARGRkpPG58ganon/Xli9fXuw9PPfcc0KpVBo//0Xv4fDhwyZ9zJ8/XwAwzrsQZQ9Oq1atKvGIUX5+vlCr1SaBqFu3bsLBwaHYkelBgwaJtm3bljpfRDUZz3EiqqEaNmyIwYMH499//zWuq7ezswNgKN98r7y8POPzRRwcHNChQwcEBAQAAJYuXYqEhAS89tpr2LFjB/73v//h448/xqeffooZM2bg33//LdPY/vjjD2zfvh2rVq3CQw89hFu3bhXbt16vxxdffIFGjRpBqVTC3d0dHh4eOHXqVInlz+vVq2fyuOicl6Jzca5fvw4AaNSokcl2crkcQUFBJm3Xr19Ho0aNip2/1axZM5O+Stt3UZXBonm7t70s5wf16dMH27ZtQ1paGnbv3o0pU6bg+vXreOyxx4zFCby8vNCoUSPjeVp79uxBly5d0LVrV8TFxeHKlSvYt28f9Hq98TwywHBO0Ntvv42AgACTuU1LSytxbu+dMwcHB/j4+Fjs2i4KhQJTp05FWloajh07BuDBn9W7twEMP8eQkBA0b94cUqkUUVFRWLRoEb788kukpqZiwIABGDJkCNauXYvt27fjww8/fOC4GjdujBUrViA5ORmnTp3C3LlzYWNjg0mTJmHHjh3G7caMGYPc3FysX78eABAdHY1jx47hmWeeKdbnvZ8VwPBZLfpMJCUlIScnx+S8oyLNmjWDXq/HjRs37tvnvZ/9io6xIp/nyno/91Iqlfjkk0/w999/w8vLC127dsWnn36KhIQE4zYXL14EADzyyCPw8PAw+fPPP/9YtOjH2bNnMXToUDg5OUGtVsPDw8NYDKWsl28oSdF7GDt2bLH38NNPPyE/P79Y/+bMZ2mK/u279+epUCgQFBRU7N9Gf3//YtfDuvvzTlTbsKoeUQ0WEBAAjUaD7OxsqNVq+Pj4AADi4+OLbRsfHw9fX99S+8rIyMAbb7yBzz77DPb29vj1119NytcOGzYMv/zyC3r06PHAcXXt2tVYVW/gwIFo1aoVnnrqKRw7dswYVubOnYu33noLzz77LN5//324urpCKpXi5Zdfhl6vL9ZnaZWfhBAPHE9FlbZvS4xJpVKhS5cu6NKlC9zd3fHuu+/i77//xtixYwEAnTt3RkREBHJzc3Hs2DG8/fbbaNmyJZydnbFnzx6cP38eDg4OJqW+X3zxRSxduhQvv/wyOnbsCCcnJ0gkEowcObLEua0KRV/Ki4pnuLq6QqlUlvpZBXDfz+srr7yCp59+Gu3atcOKFSvg6uqK2bNnAwBmzZqFDz/8sMwlxWUyGVq1aoVWrVqhY8eO6NGjB3755Rc8+uijAIDmzZsjNDQUK1euxJgxY7By5UooFAoMHz68xL5KUpHPaVn6tMQYq+rvmLn7efnllzFw4EBs2LAB27Ztw1tvvYWPPvoIO3fuRNu2bY2f7RUrVsDb27vY621sLPOVJy0tDd26dYNarTaWxre1tcXx48fx6quvVujvWNFr582bV2qZcgcHB5PH1fHfxqrYN5E1MDgR1WBXrlyBra2t8T/Sli1bwsbGBkePHjX5wqTRaBAZGVnil6gi7733Hho0aICnnnoKgOE6I3d/Gff19TXrYo0ODg545513MH78ePz2228YOXIkAENFtR49emDx4sUm26elpRlDV3kUVUe7ePEiHnnkEWO7VqvF1atX0aZNG5NtT506Bb1eb3LUqaicdUmV1qpC+/btAZgG3y5dumDp0qVYvXo1dDodOnXqBKlUis6dOxuDU6dOnUy+wPz+++8YO3YsPv/8c2NbXl5eqRchvnjxokkgzsrKQnx8PPr372+x91Z0UVsPDw8AgFQqRatWrXD06NFi2x46dAhBQUFwdHQssa+//voL+/fvN/52Pi4uzvhLA8DwWY2NjTVrnCX9DADDEZ3p06cjPj4eq1atwoABA0wq/ZWVh4cHVCoVoqOjiz0XFRUFqVRa7MhPWVlqjOVRme+nJMHBwZgxYwZmzJiBixcvIiQkBJ9//jlWrlyJ4OBgAIaqiEWhtzLs2rULKSkpWLduHbp27Wpsv3r1arFt7z0a86D2ovegVqst+h5K29+9iv7ti46ONjlSr9FocPXq1UqdV6KagEv1iGqAkkr1njx5Eps2bULv3r2NX/6dnJzw6KOPYuXKlcjMzDRuu2LFCmRlZeHJJ58ssf8LFy7g66+/xpdffmn8D9bLy8vkujjnz58v8be4ZfHUU0/B398fn3zyibFNJpMV+63k2rVrK/SF18PDA9999x00Go2xfdmyZcUCQ//+/ZGQkIA1a9YY2woKCrBw4UI4ODigW7duZo2hrCIiIkps37JlCwDTZTJFS/A++eQTtG7d2rh8qkuXLoiIiMDRo0dNlukBJc/twoULTUol3+2HH34wKWX97bffoqCgAP369SvnOyv5s5qZmYkFCxbA3d0doaGhxvZhw4bhyJEjJuEpOjoaO3fuLPWzqtFoMH36dLz55pvGsuFeXl64dOmSsUx/WT6re/bsKbF8d0k/AwAYNWoUJBIJpk2bhitXrjzwGlWlkclk6N27NzZu3GiyFDIxMRGrVq1C586dTS4lUB6WGmN5VOb7uVtOTk6x0vbBwcFwdHQ0Lvfs06cP1Go15s6dW+LPtiwlz8ui6JcUd/8d02g0WLRoUbFt7e3tS1y6Z29vDwDF/m0KDQ1FcHAwPvvsM2RlZRV7nbnvwd7evtRfnNzt0UcfhUKhwFdffWXy/hYvXoz09HQMGDDArP0T1RY84kRUA4wYMQJ2dnbo1KkTPD09ce7cOfzwww9QqVT4+OOPTbb98MMP0alTJ3Tr1g2TJk3CzZs38fnnn6N3797o27dvif2/8sorGDFihMnFaYcNG4bBgwfj9ddfBwD8+eef+Ouvv8wav1wux7Rp0/C///0PW7duRd++ffHYY4/hvffew/jx49GpUyecPn0av/zyS7Hzkcqzjw8++ACTJ0/GI488ghEjRuDq1atYunRpsT4nTZqE77//HuPGjcOxY8cQGBiI33//Hfv27cOCBQtKPdJhKYMHD0aDBg0wcOBABAcHIzs7Gzt27MCff/6JsLAwDBw40Lhtw4YN4e3tjejoaLz44ovG9q5du+LVV18FgGLB6bHHHsOKFSvg5OSE5s2b48CBA9ixYwfc3NxKHI9Go0HPnj0xfPhwREdHY9GiRejcuTMGDRpk3GbZsmUYP348li5darwGTEm++eYbbNiwAQMHDkS9evUQHx+PJUuWICYmBitWrDC5dtULL7yAH3/8EQMGDMDMmTMhl8sxf/58eHl5YcaMGSX2/+WXXwIApk2bZmzr378/pkyZgtGjR6NTp054//33H3jh5k8++QTHjh3D448/jtatWwMAjh8/jp9//hmurq54+eWXTbb38PBA3759sXbtWjg7O1foC+QHH3yA7du3o3PnznjhhRdgY2OD77//Hvn5+RW6Bo4lx1gelfV+7nbhwgXjZ7R58+awsbHB+vXrkZiYaDyKrVar8e233+KZZ55Bu3btMHLkSHh4eCAmJgabN2/Gww8/jK+//rrCY+nUqRNcXFwwduxYvPTSS5BIJFixYkWJy9NCQ0OxZs0aTJ8+HWFhYXBwcDD+vXd2dsZ3330HR0dH2Nvbo0OHDmjQoAF++ukn9OvXDy1atMD48ePh5+eH2NhY/Pvvv1Cr1fjzzz/LPebQ0FDs2LED8+fPh6+vLxo0aIAOHToU287DwwOzZ8/Gu+++i759+2LQoEHGfxPCwsKqJIwTVWtWKUlBROXy5ZdfivDwcOHq6ipsbGyEj4+PePrpp8XFixdL3H7Pnj2iU6dOwtbWVnh4eIgpU6aYVNu62+bNm4WDg4OIi4sr9txHH30kfH19hY+PT4nVz+5V2nWchBAiPT3dpLpdXl6emDFjhvDx8RF2dnbi4YcfFgcOHBDdunUzqYBXVFVv7dq1Jv0VVby7tyrVokWLjNeQad++vdi9e3exPoUQIjExUYwfP164u7sLhUIhWrVqVayvon3cWz2vtDEVVai69/pa9/r111/FyJEjRXBwsLCzsxO2traiefPm4o033ijx51R0jZi7r8+l0WiESqUSCoVC5Obmmmx/+/Zt43tzcHAQffr0EVFRUaJ+/fomlbWKxvvff/+JSZMmCRcXF+Hg4CCeeuopkZKSYtLnwoULBQCxdevW+763f/75R/Tq1Ut4e3sLuVwunJ2dRe/evUutzHbjxg0xbNgwoVarhYODg3jsscdK/VwnJCQIR0dHsWnTpmLP/f3336Jp06bC2dlZjBkzplhZ8Xvt27dPTJkyRbRs2VI4OTkJuVwu6tWrJ8aNG1dqVcTffvvNeO2dkqCEynRCiGLzLoShelqfPn2Eg4ODUKlUokePHsZr6xQp7fNU9Pn7999/LTLG8n7OS1JZ76dIcnKymDJlimjatKmwt7cXTk5OokOHDsUq9BX116dPH+Hk5CRsbW1FcHCwGDdunEllz4pW1du3b5946KGHhJ2dnfD19RWzZs0S27ZtK/Y+srKyxOjRo42VG++usLdx40bRvHlzYWNjU+zfshMnTojHH39cuLm5CaVSKerXry+GDx9u8veotH9vS7o2U1RUlOjatauxBH/R57GkbYUwlB9v2rSpkMvlwsvLSzz//PPi9u3bZZqvsWPHllhJkKg2kAjBM/iIiOqioqNIR44cMZ7bU5rhw4fj2rVrOHz4cBWNrvrZuHEjhgwZgt27dxc7yldd1IQxEhHVVFyqR0RE9yWEwK5du7By5UprD8WqfvzxRwQFBaFz587WHkqpasIYiYhqKgYnIiK6L4lEYtFr4NQ0q1evxqlTp7B582aTAirVSU0YIxFRTcfgREREdB+jRo2Cg4MDJkyYgBdeeMHawylRTRgjEVFNx3OciIiIiIiIHoDXcSIiIiIiInoABiciIiIiIqIHqHPnOOn1esTFxcHR0ZEnzxIRERER1WFCCGRmZsLX1xdS6f2PKdW54BQXF4eAgABrD4OIiIiIiKqJGzduwN/f/77b1Lng5OjoCMAwOWq12sqjAbRaLf755x/07t0bcrnc2sOpMThv5uG8mYfzZj7OnXk4b+bhvJmH82Y+zp15qtO8ZWRkICAgwJgR7qfOBaei5XlqtbraBCeVSgW1Wm31D05NwnkzD+fNPJw383HuzMN5Mw/nzTycN/Nx7sxTHeetLKfwsDgEERERERHRAzA4ERERERERPQCDExERERER0QPUuXOcykIIgYKCAuh0ukrfl1arhY2NDfLy8qpkf7VFdZ83mUwGGxsblrwnIiIiqiUYnO6h0WgQHx+PnJycKtmfEALe3t64ceMGv2SXQ02YN5VKBR8fHygUCmsPhYiIiIgqiMHpLnq9HlevXoVMJoOvry8UCkWlfynX6/XIysqCg4PDAy+6RXdU53kTQkCj0SApKQlXr15Fo0aNqt0YiYiIiKh8GJzuotFooNfrERAQAJVKVSX71Ov10Gg0sLW15Zfrcqju82ZnZwe5XI7r168bx0lERERENVf1+8ZZDVTHL+JU8/BzRERERFR78JsdERERERHRAzA4ERERERERPQCDE5UqMDAQCxYssPYwiIiIiIisjsGpFpBIJPf9M2fOHLP6PXLkCCZNmmSRMf7666+QyWSYMmWKRfojIiIiIqpKDE61QHx8vPHPggULoFarTdpmzpxp3Lbo4r5l4eHhYbHqgosXL8asWbPw66+/Ii8vzyJ9EhERERFVFQanBxBCIEdTUKl/cjW6EtuFEGUao7e3t/GPk5MTJBKJ8XFUVBQcHR3x999/IzQ0FEqlEnv37sXly5cxePBgeHl5wcHBAWFhYdixY4dJv/cu1ZNIJPjpp58wdOhQqFQqNGrUCJs2bXrg+K5evYr9+/fjtddeQ+PGjbFu3bpi2yxZsgQtWrSAUqmEj48Ppk6danwuLS0NkydPhpeXF2xtbdGyZUv89ddfZZobIiIiIiJL4HWcHiBXq0Pzt7dZZd/n3usDlcIyP6LXXnsNn332GYKCguDi4oIbN26gf//++PDDD6FUKvHzzz9j4MCBiI6ORr169Urt591338Wnn36KefPmYeHChXjqqadw/fp1uLq6lvqapUuXYsCAAXBycsLTTz+NxYsXY/To0cbnv/32W0yfPh0ff/wx+vXrh/T0dOzbtw+A4XpN/fr1Q2ZmJlauXIng4GCcO3eu0i9MTERERER0NwanOuK9995Dr169jI9dXV3Rpk0b4+P3338f69evx6ZNm0yO9txr3LhxGDVqFABg7ty5+Oqrr3D48GH07du3xO31ej2WLVuGhQsXAgBGjhyJGTNm4OrVq2jQoAEA4IMPPsCMGTMwbdo04+vCwsIAADt27MDhw4dx/vx5NG7cGAAQFBQEvV6PjIwMc6aCiIiIiKjcGJwewE4uw7n3+lRa/3q9HpkZmXBUOxa7YKqdXGax/bRv397kcVZWFubMmYPNmzcjPj4eBQUFyM3NRUxMzH37ad26tfG+vb091Go1bt26Ver227dvR3Z2Nvr37w8AcHd3R69evbBkyRK8//77uHXrFuLi4tCzZ88SXx8ZGQl/f39jaCIiIiKimidXo0NqjgapWRrcysjBkSQJ+urLdlpKdcHg9AASicRiy+VKotfrUaCQQaWwKRacLMne3t7k8cyZM7F9+3Z89tlnaNiwIezs7DBs2DBoNJr79iOXy00eSyQS6PX6UrdfvHgxUlNTYWdnZ2zT6/U4deoU3n33XZP2kjzoeSIiIiKqWnq9QGZeAVKy85GarTH+ScnW4PZd9+9+Lleru6cXGV7K08JTqbDKezAHg1MdtW/fPowbNw5Dhw4FYDgCde3aNYvuIyUlBRs3bsTq1avRokULY7tOp0Pnzp3xzz//oG/fvggMDERERAR69OhRrI/WrVvj5s2buHDhAo86EREREVUCrU6P23cFn3tDjyEI5eN2ttawTY4GOjOOFilkUrjaK+CikkOXmw6tjkecqAZo1KgR1q1bh4EDB0IikeCtt96675Ejc6xYsQJubm4YPnx4sWIO/fv3x+LFi9G3b1/MmTMH//d//wdPT09jIYh9+/bhxRdfRLdu3dC1a1c88cQTmD9/Pho2bIioqCgIIdCpUyeLjpeIiIiopjNUhNaVeCTIEIjykZqtLbw1tGXmle1SNfdyVNrAxV4BV3sF3ApvS/rjZq+Ei70cDkobSCQSaLVabNmyBZ6OSgu/+8rF4FRHzZ8/H88++yw6deoEd3d3vPrqqxYvtrBkyRIMHTq0xAp4TzzxBJ555hkkJydj7NixyMvLwxdffIGZM2fC3d0dw4YNM277xx9/YObMmRg1ahSys7PRsGFDzJ0716JjJSIiIqqO9HqBtFztXUHoTvAp7QhRfkH5fxkulQAuqpJCj+KucKQ0trvYy6G0sdz5+DUBg1MtM27cOIwbN874uHv37iVeDyowMBA7d+40aZsyZYrJ43uX7pXUT1paWqljOXXqVKnPDR8+HMOHDzc+njx5MiZPnlzitq6urliyZIlJG6vqERERUU2UX6ArXPKWX2w5XEnnCt3O0cCcGgpKG+k9oUcBV3slXO3lhbem4cjJTg6plJd7uR8GJyIiIiIiMwghkFcAXE/NQUa+3uToT2nnCmXlm7csTm1rc1fYuROASlsip1LIeN1LC2NwIiIiIiK6hxAC6blaxKXlIT49F3HpeUhIz0V8Wh7i0nORkJ6H+PQ85BfYAEf2lqtvmVQCF9WDzgtSwNVBAVeV4aiRXFZ51ZepbBiciIiIiKhOEUIgI68A8YVBKD69MByl5SEh405b8RLaJbOTS02Wv7kZzwMqHo7c7JVwtLXhsrgaiMGJiIiIiGqVzDwtEtLzEJeeh/i0u44WpechLs1wtChbU7ZQ5GavgLeTLXyc7ODrbAtvJ1v4OtnBx8kW7vY2OL5vF4YM7F/sWpdU+zA4EREREVGNkaMpMC6fMzladNdSuswynkfkrJIbApFTYSByNgQin8Jg5O1kC1t56ZXjtFotztStwnJ1GoMTEREREVULuRod4gvPHyrtaFFGGa85pLa1MQQgZ9sSw5G3ky1UCn4VprLjp4WIiIiIKl2eVofEjLw7R4vS7xw1KgpHt3O0ZerLQWljODLkbAcftS18nAuXzznbGo8Y2Sv5NZcsi58oIiIiIqoQTYG+MBTlIuGucHR3sYWUbE2Z+lIpZCbL5XyMy+fuHC1ytOX5RFT1GJyIiIiIqFRanSEU3b18zni0KN0QkpKz8svUl9JGWuw8oqKjRUVFF9R2Nrz+EFVLDE5k1L17d4SEhGDBggUAgMDAQLz88st4+eWXS32NRCLB+vXrMWTIkArt21L9EBERUdkJIZCSlY9rmcCW0wlIyi5+3aJbmfkQ4sF9KWykhnOH1HcVWbhnKZ2zSs5QRDUWg1MtMHDgQGi1WmzdurXYc3v27EHXrl1x8uRJtG7dulz9HjlyBPb29pYaJgBgzpw52LBhAyIjI03a4+Pj4eLiYtF9lSY3Nxd+fn6QSqWIjY2FUqmskv0SERFZg14vcCszH9dSshGTkoNrKdm4ftdtVn4BABvgzKlS+5DLJPBS3zmP6O6S3L7OhqNFbvYKhiKq1RicaoEJEybgiSeewM2bN+Hv72/y3NKlS9G+fftyhyYA8PDwsNQQH8jb27vK9vXHH3+gRYsWEEJgw4YNGDFiRJXtm4iIqDLo9ALx6bkmgehasuH2emo28rT6+77eWSEQ5O0CX2e7u5bS2Rqr0rnbK3nBVqrzGJweRAhAm1N5/ev1hv41MkAqNX1OrgLK8Jubxx57DB4eHli2bBnefPNNY3tWVhbWrl2LefPmISUlBVOnTsXu3btx+/ZtBAcH4/XXX8eoUaNK7ffepXoXL17EhAkTcPjwYQQFBeHLL78s9ppXX30V69evx82bN+Ht7Y2nnnoKb7/9NuRyOZYtW4Z3330XAIy/kVq6dCnGjRtXbKne6dOnMW3aNBw4cAAqlQpPPPEE5s+fDwcHBwDA+PHjkZycjO7du2P+/PnQaDQYOXIkFixY8MAL0C1evBhPP/00hBBYvHhxseB09uxZvPrqq9i9ezeEEAgJCcGyZcsQHBwMAFiyZAk+//xzXLp0Ca6urnjiiSfw9ddf33efREREFaXV6RF7O7fYEaPrKdm4kZoLja70cCSVAP4uKtR3UyHQzd54G+iugreDHBHbt6F//3BexJXoPhicHkSbA8z1rbTupQCcS3vy9ThA8eClcjY2NhgzZgyWLVuGN954wxhK1q5dC51Oh1GjRiErKwuhoaF49dVXoVarsXnzZjzzzDMIDg5GeHj4A/eh1+vx+OOPw8vLC4cOHUJ6enqJ5z45Ojpi2bJl8PX1xenTpzFx4kQ4Ojpi1qxZGDFiBM6cOYOtW7dix44dAAAnJ6difWRnZ6NPnz7o2LEjjhw5glu3buG5557D1KlTsWzZMuN2e/bsQUBAAP79919cunQJI0aMQEhICCZOnFjq+7h8+TIOHDiAdevWQQiBV155BdevX0f9+vUBALGxsejatSu6d++OnTt3Qq1WY9++fSgoMFwz4ttvv8X06dPx8ccfo1+/fkhPT8e+ffseOH9ERERlkafV4ebtHFxLzsH1VEMoulYYjm7ezoVOX/rJRnKZBAGuhkBUz1WFQDcV6rvbI9DNHn7OdlDYSEt8nVZbthLgRHUdg1Mt8eyzz2LevHn477//0L17dwCGozlPPPEEnJyc4OTkhJkzZxq3f/HFF7Ft2zb89ttvZQpOO3bsQFRUFLZt2wZfX0OQnDt3Lvr162ey3d1HvAIDAzFz5kysXr0as2bNgp2dHRwcHGBjY3PfpXmrVq1CXl4efv75Z+M5Vl9//TUGDhyITz75BF5eXgAAZ2dnLFy4EHK5HE2bNsWAAQMQERFx3+C0ZMkS9OvXz3g+VZ8+fbB06VLMmTMHAPDNN9/AyckJq1evNv7WrXHjxsbXf/DBB5gxYwamTZtmbAsLC3vg/BERERXJ0RQgJrUwHN0VjK6n5CAuPfe+hRiUNlLUd1Ohvpu9IRi52RuPIPk620HG5XRElYbB6UHkKsORn0qi1+uRkZkJtaMjpCUt1Sujpk2bolOnTliyZAm6d++OS5cuYc+ePXjvvfcAADqdDnPnzsVvv/2G2NhYaDQa5OfnQ6Uq2z7Onz+PgIAAY2gCgI4dOxbbbs2aNfjqq69w+fJlZGVloaCgAGq1uszvo2hfbdq0MSlM8fDDD0Ov1yM6OtoYnJo2bQqZTGbcxsfHB6dPny61X51Oh+XLl5ssMXz66acxc+ZMvP3225BKpYiMjESXLl1KXKpw69YtxMXFoWfPnuV6P0REVPdk5mlLPd8oMeP+pbvtFTJDIHIvHpA8HXmuEZG1MDg9iERSpuVyZtPrAbnOsI97g1M5TZgwAS+++CK++eYbLF26FMHBwejWrRsAYN68efjyyy+xYMECtGrVCvb29nj55Zeh0ZTtYnRlceDAATz11FN499130adPH+ORm88//9xi+7jbveFGIpFAry99ffe2bdsQGxtb7JwmnU6HiIgI9OrVC3Z2dqW+/n7PERFR3SKEQFqO9q7zjIqOHhkeP+hir2pbGzRwtzcJRkVHktwdWJ2OqDpicKpFhg8fjmnTpmHVqlX4+eef8fzzzxv/4d23bx8GDx6Mp59+GoDhSNeFCxfQvHnzMvXdrFkz3LhxA/Hx8fDx8QEAHDx40GSb/fv3o379+njjjTeMbdevXzfZRqFQQKfTPXBfy5YtQ3Z2tvGo0759+yCVStGkSZMyjbckixcvxsiRI03GBwAffvghFi9ejF69eqF169ZYvnw5tFptsWDm6OiIwMBAREREoEePHmaPg4iIagYhBJKzNCbL6Yy3ydnIyCu47+vdHRSGQOSqKnYEyVmlqKJ3QUSWwuBUizg4OGDEiBGYPXs2MjIyMG7cOONzjRo1wu+//479+/fDxcUF8+fPR2JiYpmD06OPPorGjRtj7NixmDdvHjIyMooFkEaNGiEmJgarV69GWFgYNm/ejPXr15tsExgYiKtXryIyMhL+/v5wdHQsdh2lp556Cu+88w7Gjh2LOXPmICkpCS+++CKeeeYZ4zK98kpKSsKff/6JTZs2oWXLlibPjRkzBkOHDkVqaiqmTp2KhQsXYuTIkZg9ezacnJxw8OBBhIeHo0mTJpgzZw7+7//+D56enujXrx8yMzOxb98+vPjii2aNi4iIrEuvF0jMzCt2vtG1lBzEpGQjW3P/X/Z5qZUlnm9U300FR1tWqCOqTRicapkJEyZg8eLF6N+/v8n5SG+++SauXLmCPn36QKVSYdKkSRgyZAjS09PL1K9UKsX69esxYcIEhIeHIzAwEF999RX69u1r3GbQoEF45ZVXMHXqVOTn52PAgAF46623jIUXAOCJJ57AunXr0KNHD6SlpRnLkd9NpVJh27ZtmDZtGsLCwkzKkZurqNBESecn9ezZE3Z2dli5ciVeeukl7Ny5E//73//QrVs3yGQyhISE4OGHHwYAjB07Fnl5efjiiy8wc+ZMuLu7Y9iwYWaPi4iIKp9OLxCXlmtSvvvuggz5BaUv85ZIAF8nuxLPN6rnqoKdQlbqa4modpEIcb/aLbVPRkYGnJyckJ6eXqxoQV5eHq5evYoGDRrA1ta2Ssaj1+uRkZEBtVpdvDgElaomzJs1Pk8PotVqsWXLFvTv35/X6igHzpv5OHfm4byVnxAC15Mz8evmXfAMboEbt/OMwejG7RxodaV/3ZFJJQhwsTMGo3p3BaQAVzsobWp3OOLnzXycO/NUp3m7Xza4F484ERERUY2i1elx6VYWzsVl4Fx8Bs7GpeNcXEbhOUcyICqq2GsUMikCXO0Kl9KZnm/k62wHuax6/hKOiKoPBiciIiKqtjLytDhfGJCKgtLFxCxodMWX19lIJfBQ6tEy0AtBHg53lta528NbbctrHBFRhTA4ERERkdUJIRCfnmcMR0W3Mak5JW7vqLRBM181mvuo0dxXjRa+atR3sUXEP1vRv3+I1Zf/EFHtY/Xg9M0332DevHlISEhAmzZtsHDhQoSHh5e4rVarxUcffYTly5cjNjYWTZo0wSeffGJSoICIiIiqN61OjytJ2TgXn24SlG7naEvc3tfJFs191Wju64TmPoaQ5O9iV+xaR1ptya8nIrIEqwanNWvWYPr06fjuu+/QoUMHLFiwAH369EF0dDQ8PT2Lbf/mm29i5cqV+PHHH9G0aVNs27YNQ4cOxf79+9G2bVuLjauO1cugSsLPERERkJVfgKj4DJyNu3MUKToxE5oSKtnJpBI08nQwHkVq7qNGMx81XOx5zSMisj6rBqf58+dj4sSJGD9+PADgu+++w+bNm7FkyRK89tprxbZfsWIF3njjDfTv3x8A8Pzzz2PHjh34/PPPsXLlyhL3kZ+fj/z8fOPjjIwMAIbfSpX0mykhBLKysopdW6iyFH25FkJAry+9HCqZqgnzlpWVZRxndfktaNE4qst4agrOm/k4d+apifMmhEBiZj7Ox2ca/iQYbq+XstTOXilDM29Hwx8fRzTzVqORpz2U8uIV7Mo6DzVx3qoDzpv5OHfmqU7zVp4xWK0cuUajgUqlwu+//44hQ4YY28eOHYu0tDRs3Lix2Gvc3Nzw6aefYsKECca2p59+Gnv37sW1a9dK3M+cOXPw7rvvFmtftWoVVCpVsXZHR0e4uLjA3d0dCoWi2DIAogcRQkCj0SA5ORm3b99GZmamtYdERGRROgEk5QI3syWIzZEgNhuIzZYgq6Dk/zOdFAJ+KgF/e8DPXsDfXsBVCbBWAxFZW05ODkaPHl29y5EnJydDp9PBy8vLpN3LywtRJZQRBYA+ffpg/vz56Nq1K4KDgxEREYF169ZBpyv9qt6zZ8/G9OnTjY8zMjIQEBCA3r17lzg5QgjcunXLeGSqsgkhkJeXB1tbW4a0cqgJ8+bh4YEWLVpUq/FptVps374dvXr14onT5cB5Mx/nzjzVad6y8wtwITEL5wqPIJ2Pz0B0YlaJF42VSoBgD3s081YbjiL5OKKptyPcqmipXXWat5qE82Y+zp15qtO8lec7v9WLQ5THl19+iYkTJ6Jp06aQSCQIDg7G+PHjsWTJklJfo1QqS1x2J5fLS/1B+fv7Q6fTVcnhQ61Wi927d6Nr165W/+DUJNV93uRyOWSy6nvBxPt9/ql0nDfzce7MU9XzdivTUNXubOG5SOfjMnA1JRslrU1RKWRo5qM2OR+pibcjbEtYalfV+HkzD+fNfJw781SHeSvP/q0WnNzd3SGTyZCYmGjSnpiYCG9v7xJf4+HhgQ0bNiAvLw8pKSnw9fXFa6+9hqCgIIuPTyaTVckXX5lMhoKCAtja2lr9g1OTcN6IiMyn0wtcTc42Kft9Li4DyVn5JW7v6ag0hqMWvk5o7qtGfVcVpFxrR0R1iNWCk0KhQGhoKCIiIoznOOn1ekRERGDq1Kn3fa2trS38/Pyg1Wrxxx9/YPjw4VUwYiIioponV6NDVILptZGi4jORqy2+zF0qAYI8ile183CsmoJJRETVmVWX6k2fPh1jx45F+/btER4ejgULFiA7O9tYZW/MmDHw8/PDRx99BAA4dOgQYmNjERISgtjYWMyZMwd6vR6zZs2y5tsgIiKqFpKz8o3hyFD+Ox1Xk7OhL2GpnZ1chqY+jiYhqam3GnYK6y+1IyKqjqwanEaMGIGkpCS8/fbbSEhIQEhICLZu3WosGBETEwOpVGrcPi8vD2+++SauXLkCBwcH9O/fHytWrICzs7OV3gEREVHV0+sFrqfm4Gyc6QVkb2WWvNTO3UFhvHhsc1/DBWQD3ewh41I7IqIys3pxiKlTp5a6NG/Xrl0mj7t164Zz585VwaiIiIiqhzytDtEJmSZL7c7HZyBHU3ypnUQCNHC3NzmK1NxXDU9HWyuMnIiodrF6cCIiIiIDvV7gXHwG9l9KwraLUiz8ah+ulLLUTmkjRdN7qto19XaEvZL/tRMRVQb+60pERGQler1AVEImDl5JwYErKTh8NRXpuUWXwpACyAYAuNor0OKuI0hFS+1sZNJS+yYiIsticCIiIqoier3AhVuZOHA5BQevpODQ1VSk5ZheM9BBaYPQ+s5wyE3EkO7t0TrAFZ6Oymp1MW0iorqIwYmIiKiSCCFwITELB6/cCUqp2RqTbVQKGcICXdEx2A0PBbmhpa8aQq/Dli1b0L2xB69VR0RUTTA4ERERWYgQApduZRmX3h26koqUEoJS+0BXPBTkioeC3NDKzwnye5bcafXFCz8QEZF1MTgRERGZSQiBy0nZOFB0ROlKCpKzTIOSrVyKsEBDSHooyA2t/YsHJSIiqv4YnIiIiMpICIGryUVBKRUHr6Qg6Z5rJyltpGgf6IKHGrihY7AbWvs7Q2HDoEREVNMxOBEREZVCCIFrKTmGpXeFBR3uvciswkaK0HouxnOU2gQ4QWkjs9KIiYiosjA4ERERFRJCICb17qCUioSMPJNtFDIp2tZzNgalkABn2MoZlIiIajsGJyIiqrOEELh5O9d4NOnglRTEpRcPSiH1nPFQkBs6BrmhbT0GJSKiuojBiYiI6pSbt3OMR5MOXklBbFquyfNymQQhAc7oWFjMoW09F9gpGJSIiOo6BiciIqrVYtNycfByirHy3c3bpkHJRipBm7uCUmh9BiUiIiqOwYmIiGqV+PRck3OUYlJzTJ63kUrQ2t/JsPQu2BCUVAr+d0hERPfH/ymIiKhGS8zIMzlH6VqKaVCSSSVo5XcnKLWv7wJ7Jf/7IyKi8uH/HEREVKPcysgzuY7S1eRsk+elEhiCUmHVu/b1XeBoK7fSaImIqLZgcCIiomrtVmYeDl1JNZ6jdCWpeFBq4etUWB7cFWGBrgxKRERkcQxORERUrSRn5RuX3R28kopLt7JMnpdIgBa+ajzUoHDpXaArnOwYlIiIqHIxOBERkVWlZOXj0NVUY0GHiyUEpWbeauM5SuGBrnBSMSgREVHVYnAiIqIqlZqtweGrhqNJBy6nIDoxs9g2Tb0dC5feuaFDA1c4qxRWGCkREdEdDE5ERFSpsrXA9nO3cPh6Gg5eSUFUQslB6aEgwzlKHRq4wcWeQYmIiKoXBiciIrK4G6k52HQyDptPxeF8vAziaKTJ8429HAxL74LcEN7AFW4OSusMlIiIqIwYnIiIyCKSs/Kx+VQ8NkbG4nhM2l3PSBDsYY9Owe6GpXdBrnBnUCIiohqGwYmIiMyWmafFP2cTsfFkHPZdSoZOLwAYCjp0CnbDgJZe0N04hZFDHoZczoIORERUczE4ERFRueQX6LArOgmbIuOw43wi8gv0xufa+DthUIgfBrb2gafaFlqtFlsST1lxtERERJbB4ERERA+k0wscupKCjZFx2HImHpl5BcbngjzsMbiNHwaF+KKBu70VR0lERFR5GJyIiKhEQgicjk3Hxsg4/HkyDrcy843PeattMSjEF4Pa+KKFrxoSicSKIyUiIqp8DE5ERGTiclIWNkXGYdPJOFxNzja2O9nJ0b+VNwa18UOHBq6QShmWiIio7mBwIiIiJKTn4a9TcdgYGYfTsenGdlu5FL2ae2NQG190bewOpY3MiqMkIiKyHgYnIqI6Kj1Hi7/PxGNjZBwOXk2BMBTEg0wqQddG7hgc4odezb1gr+R/FURERPzfkIioDsnV6BARlYiNkXHYFX0LWp0wPhcW6IJBIX7o39KbF6QlIiK6B4MTEVEtp9Xpse9SMjZFxmHb2QRka3TG55p6O2JwiB8GtvGBv4vKiqMkIiKq3hiciIhqISEEjsfcxsbIOGw+FY+UbI3xOX8XOwwO8cWgNn5o4u1oxVESERHVHAxORES1SHRCJjZGxmLTyTjcvJ1rbHezV+Cx1j4YFOKHdvWcWT6ciIionBiciIhquBupOfjzVBw2RcYhKiHT2G6vkKFPS28MDvHDw8FusJFJrThKIiKimo3BiYioBkrJyseW04aKeEev3za2K2RSdG/igcEhfujZzBO2cpYPJyIisgQGJyKiGiIrvwDbzyVgY2Qc9lxMhk5vqIgnkQAdg9wwOMQXfVv4wEklt/JIiYiIah8GJyKiaiy/QIf/opOw6WQcdpxPRJ5Wb3yutb8TBrXxxcA2vvBS21pxlERERLUfgxMRUTWj0wscupqCTZFx2HI6Hhl5BcbngtztMSjEF4Pa+CLIw8GKoyQiompLrwf0WkCnLbwtuOtxwV3t9z4usMDrCu7/nE4LmU6DbrdTgB6dALmHtWerzBiciIiqASEEzsZlYGNkLP48GY+EjDzjc15qJQa29sXgED+09FOzIh4RmUcIIPc2kBEHZMYbb2VpNxF+9Qxkv/0CSGUAJIY1wBIJIJHe9bjovtTwuNj9cr7G5HX3267oOZSj78JiOGXu++7XlPb+ir9GotPDKz0SkmgAEv2dkGAMDpUQSsrSH+5c3Lw6kgJwBqAtyHvAltULgxMRkRVdTc42lA+PjMOV5Gxju9rWBv1b+WBQiC86NHCDTMqwRET3odMCmQkmgQgZsUBGvGlbCV9UpQB8ACC9qgdd89kAeAgArlh5IGUhtQGkckAmN9yXyQsfl9Zelu1szOqjQEhw5Hgk2ts5W3tWyoXBiYioiiVm5OHPk3HYdDIOp27e+aaitJHi0eZeGNzGF92aeEBpw4p4RAQgL+OeIBRnGogy4oDsJJT5KIOdK6D2Nfxx9IHO3gunr91Cy1ZtYCOVGPoRAhCF51Qa7xfeClH6fePjir7mfq/HPa9/0GtECfss6TV3v+eSxln8NUKvR1pmFpxc3CG1URQGCZtyhhBzw8u9j2X377MarVYQWi1uXdIDNjXr/FwGJyKiKpCeq8XWM4by4QeupBj+PwYgk0rQuaE7Bof4oncLbzgo+c8yUZ2h1wFZt0oOQne3abLK1p9UDjj6AGqfwlvfO7dF9x19ALnpl1W9VovrWVvQom1/QM6qnOVRoNVi95Yt6N+/P6Scu1qP/0MTEVWSPK0OEedvYWNkLHZFJ0Gju1MRL7S+CwaH+KJ/Kx+4OyitOEoiqhSanFKOEhXeZsQBWYmA0JWtP6WTIRCpfQFH35LDkcodkPJC10SVhcGJiMiCCnR67Lucgo2RsfjnbCKy8u9UxGvi5WisiBfgqrLiKInIbHo9kJNSchC6uy2vjCcMSaSAg3cJQcivsK0wJCnsK/d9EdEDMTgREVWQEALHY9KwKTIWm0/HIzlLY3zOz9kOg0J8MTjEF0291VYcJRE9UEF+4VGiwiNFRffvDUl6bdn6k9vfFYjuCUJFt/aehnNRiKja499UIiIzXUjMNFTEOxmHG6m5xnZXewUGtPLB4BBftKvnAikr4hFZV1EZ7hKD0F33c1LK2KEEsPcoIQjdE4qU6mp1Qj4RVQyDExFROdy8nYM/T8ZjY2QsohIyje32Chn6tPDGwBBfdG7oDrmM5xkQVTqd1hB2spOBnBRIMhMRdGsXpBGHgexE05BUkPvg/gBApiw9CBXdOngDNorKfW9EVO0wOBERPUBmXgH2Jkiw4qfDOHo9zdgul0nQvYknBof4omdTL9gpWD6cqEI0OUBOsjEIGW9La7vnPCIbAK0AILaU/ovKcBsrz/maVpxT+wJ2LjxKREQlYnAiIipFfoEOKw/G4OudF3E7RwYgDRIJ8FADNwwO8UXflt5wVvG3zkQlEgLISwNyUgvDTvJdtyklB6KyHhW6m0RqCET27tDbuSA+Qwfvxm0hc/K/JyT5AHI7i79NIqo7GJyIiO6h1wtsPBmLz7ZdQGya4Yucp63AhO5NMLhtALydatYF+4gsQlcA5N4Vgu4++mPSlnLnvr7gwf3eS6YwlNW2dyu8dQdUbqW32TkbLvwJQKfV4uiWLejfqz9kvKYOEVkYgxMRUSEhBP67kIRPtkbjfHwGAMBLrcRLPYJhl3gKAx8OhJxfxqi20OaVcBToPoEo97Z5+1E4AirXwrBzd+hxK95m7w4oHLhUjoiqJQYnIiIAJ2+k4eO/o3DgiqGqlqOtDZ7vHozxnRrARqLHli2nrDxCovsQAsjPKAw6qfcEopLaUgBtthk7khjOASoKPMUC0T1tKjdAziO0RFQ7MDgRUZ12NTkbn22LxubT8QAAhUyKsZ3q44XuDeFibzh/SavVW3OIVBfpdVBoM4CkaECTVvJRoLsDUU4KoNM8sNtipPK7jvy4lWFZnAuvOUREdRb/9SOiOulWZh6+iriI1YdvoEAvIJEAj7f1xyu9GsHfRWXt4VFtJoQh9GTcBNJjDRdaTb9ZeGt4bJMZj376AuBMOfuW2xcGILfiS+BUJbTxOkNERGXG4EREdUpmnhY/7r6CH/dcRa5WBwB4pKknZvVtgqbeaiuPjmq8ogutGkPQTcMFVk0CUhygy79vN0VRRtg6Q3L3sjeTQFRCG6vGERFVGgYnIqoT8gt0WHUoBgt3XkJqtmFJU0iAM17r1xQPBblZeXRUY+Rn3glExjB0z2NtThk6kgAOnoDaD3DyA9T+hbd+gJM/tCov/L3nKPoNGMSCJERE1QSDExHVanq9wJ+n4vDZP9G4kWooLR7kbo9ZfZugTwtvSLhMiYpoc0sIRYVHiIoCUn76g/sBDEeCCkNQieHI0Qewuc81wLRaCAn/iyYiqk74rzIR1Vp7Libh47+jcDbOUFrc01GJlx9tjOHt/WEjk1p5dFSlCjRA5r1L5u45WpSbWra+bJ3uCkG+xY4WQe3LJXNERLUQgxMR1Tqnb6bj463nse9SYWlxpQ3+r3swxj8cCJWC/+zVOroCICuh+NGiu4otIOsWAPHgvuT2d4Wg4kvooPYFlI6V/paIiKj64TcIIqo1riVn47N/ovHXqTulxZ/pWB9TejSEq/19lkVR9aXXA9lJ961Ah8wEQOge3JdMeU8IKjxidPdyOltnVpkjIqISMTgRUY2XlJmPhTsvYtWhGGNp8aEhfnilV2MEuLK0eLUlhOE6RKWeU1R4X699cF9SG8DR9/5Hi1RuDEVERGQ2BiciqrGy8gsKS4tfQY7GcMShexMPzOrTFM19WVrc6go0QNIFeKafhOR4EpB973K6OKAg98H9SKSAg3fxo0VFj9V+hgp1UlnlvyciIqqzGJyIqMbRFOix+kgMvoq4iOQsQ2nxNv5OeLVfU3QKdrfy6Oqggnwg5RKQFAXcijLcJkUBKZchFzp0BIAr93m9vUfJgajosaM3IGNJbiIisi4GJyKqMfR6gc2n4/HZP9G4nmK4Vk4Dd3v8r08T9GvJ0uKVriAfSL54JxgVBaXUK6WeYySUjsiQusDRrymkzgHFl9CpfQEbZRW/ESIiovJjcCKiGmHfpWR8/HcUTscarqPj7qDEtEcbYWRYAOQsLW5Z2rw7R5CSooBb5w23qVdLL8KgdAI8mwIeTQCPZoZbz2YosHXHrr//Rv/+/SHlhVyJiKgGY3AiomrtTGw6PtkahT0XkwEA9goZJncLxoTODWCv5D9hFaLNA1Iumi6vSyo6gqQv+TXGgFT4p+i+o0/JhRe0ZSjsQEREVAPwWwcRVUsxKTn4fHs0NkbGAQDkMgmefqg+pvZoCDcHLu0qF20ekHwBSIoGks4bbm+dB25fLT0g2ToZjhzdHZI8mhrON+KSSCIiqoMYnIioWknJysfCnZfwy6Hr0OoMFywdEuKL6b2aoJ4bS4vflzbX9BykW1GGoHT72n0CkjPg2azYEjs4eDEgERER3YXBiYiqhez8AizeexU/7L6CrPwCAECXRu54tW9TtPRzsvLoqhltruEI0r1L7O4XkOxcTINRUVBy8GRAIiIiKgMGJyKyKq1Oj9VHbuDLHReRnJUPAGjl54RX+zZF50Z1vLS4JqeUJXbXAIiSX1MUkO5dYseAREREVCEMTkRkFUIIbDmdgHnbonCtsLR4fTcVZvZuggGtfCCV1qEv+caAVFTBrjAo3b6O0gOSa+GRo3uKNNh7MCARERFVAgYnIqpy+y8n45O/o3DyZlFpcQVe6tkII8PqQWFTi0uLa7JLWWJ3n4CkcitliZ1HlQ6diIiormNwIqIqcy4uA59sjcJ/F5IAGEqLT+wahOe6BMGhNpUW12QXHjUqPHJUFJTSYlB6QHI3PXLk0dQQlOzr+HJFIiKiaqIWfVMhourqRmoO5m+/gA2RsRACsJFK8FSHepj6SCN4ONbg0uL5WUBy9J1zj4qCUlpM6a+x97jr3KMmd5bbMSARERFVawxORFRpUrM1+HrnJaw8eB0anaHa28A2vpjRqzEC3e2tPLpy0GngnH0FkpO/AqkX75T6Ti9jQPJsWrjcrilg71Z14yYiIiKLYXAiIovL0RRgyd6r+P6/K8gsLC3+cEM3vNa3GVr515DS4revAZd2AJciYHP1P3TTZAMXStjO3rN4BTsGJCIiolqHwYmILEar0+O3ozewYMdFJGUaSou38FXjtX5N0aVRNS9moMkBru8rDEs7gJRLxqckAPJlDpAHtIXUs7npEjuVq/XGTERERFWGwYmIKkwIga1nEjBvWzSuJGcDAAJc7TCzdxMMbO1bPUuLC2GocFcUlK7tA3T5d56XyIB6DwENe0Ib2B1bj8Wg/4DHIJXLrTdmIiIishoGJyKqkINXUvDx31GIvJEGAHC1V+ClRxpidIf61a+0eF46cHW3cQke0m+YPu8UADTsCTR8FGjQFbAtXFao1QKSm1U/XiIiIqo2GJyIyCzn4zPw6dYo/BttKC2uUsjwXJcgTOzSAI621eSojF4PJJy6E5RuHAKE7s7zMiUQ+LAhKDV8FHBvzIvHEhERUYkYnIioXG7eNpQWX3/iTmnxUeH18GLPhvB0tLX28IDsZODyv4awdDkCyE4yfd6t0Z2jSvUfBhQq64yTiIiIahQGJyIqk9vZGnzz7yX8fOBOafEBrX0ws3cTNLBmaXFdARB77M65SnEnYHKRWYUD0KBbYVjqCbgEWmukREREVIMxOBHRfeVqdFiy7yq+23XZWFq8Y5AbXuvXFG0CnK0zqPRYw9GkSxHAlX8N5y7dzavVnaNKAR0AG4V1xklERES1BoMTEZWoQKfH78du4osdF5CYYag219TbEa/1a4pujT0gqcpzgQrygZgDd85VunXO9HlbZyD4EUNQCn4EUPtU3diIiIioTmBwIiITQgj8cy4Rn26NwuUkQ2lxP2c7zOzTGIPb+FVdafHUK4aQdGmHoRKeNueuJyWAf/s7RR182wJSWdWMi4iIiOokBiciMjp8NRUf/30ex2PSAAAuKjmmPtIITz9UD0qbSg4mmmzg2t475yqlXjF93sGrMCj1BIJ68MKzREREVKUYnIgI0QmZmLctCjvO3wIA2MqleK5zECZ1C4K6skqLCwHcOl94rtIO4Pp+QKe587zUBqjX8c65Sl4tWSqciIiIrIbBiagOi0vLxRfbL+CP4zehF4BMKsGIsAC83LMRPNWVUFo8Nw24suvOuUqZcabPO9e7s/yuQVdA6Wj5MRARERGZwerB6ZtvvsG8efOQkJCANm3aYOHChQgPDy91+wULFuDbb79FTEwM3N3dMWzYMHz00Uewta0G148hqiHScrT4ad8lLN1/DZoCQ2nxfi29MbNPEwR7OFhuR3o9EB9551ylm0dML0BrYwsEdrlzVMmtIY8qERERUbVk1eC0Zs0aTJ8+Hd999x06dOiABQsWoE+fPoiOjoanp2ex7VetWoXXXnsNS5YsQadOnXDhwgWMGzcOEokE8+fPt8I7IKpZ8rQ6RMRK8NYXe5CRZygtHt7AFa/1a4p29Vwss5OsJODyzjsXoM1JMX3evcmdc5XqdwLkdpbZLxEREVElsmpwmj9/PiZOnIjx48cDAL777jts3rwZS5YswWuvvVZs+/379+Phhx/G6NGjAQCBgYEYNWoUDh06VKXjJqqJbmdr8OR3B3EpSQagAE29HfFq36bo3qSCpcV1BYYjSUVFHeIjTZ9XOAJB3e6EJed6FXkbRERERFZhteCk0Whw7NgxzJ4929gmlUrx6KOP4sCBAyW+plOnTli5ciUOHz6M8PBwXLlyBVu2bMEzzzxT6n7y8/ORn59vfJyRkQEA0Gq10Gq1Fno35isaQ3UYS03CeSufPK0OE5Yfw6WkbKjlArP7N8fQdv6QSSUoKCgof4cZsZBcjoD08k5Iru2GJD/D5Gnh1Qr64J4QwY9A+IUBsrsKTNTAnxk/b+bj3JmH82Yezpt5OG/m49yZpzrNW3nGIBFCiEocS6ni4uLg5+eH/fv3o2PHjsb2WbNm4b///iv1KNJXX32FmTNnQgiBgoIC/N///R++/fbbUvczZ84cvPvuu8XaV61aBZVKVfE3QlTN6QWw9IIUp1KlsJMJTGupg085P/pSvQZuWRfgmXEKnpmnoc6LNXk+X+aAJHVL3HJsjVvqlsiXO1vuDRARERFVkpycHIwePRrp6elQq9X33dbqxSHKY9euXZg7dy4WLVqEDh064NKlS5g2bRref/99vPXWWyW+Zvbs2Zg+fbrxcUZGBgICAtC7d+8HTk5V0Gq12L59O3r16gW5vJLKPtdCnLeyEULg/S3ROJUaA7lMgu+eCkHaxaMPnjchgNtXDEeULkdAcn0fJAW5d56WSCH82kMEPQIR/Aik3m3gJZXBqwrekzXw82Y+zp15OG/m4byZh/NmPs6dearTvBWtRisLqwUnd3d3yGQyJCYmmrQnJibC29u7xNe89dZbeOaZZ/Dcc88BAFq1aoXs7GxMmjQJb7zxBqRSabHXKJVKKJXKYu1yudzqP6i7Vbfx1BSct/v7cfcVrDgYAwCYPzwEnRp5YMvFUuYtPwu4tufOuUq3r5k+7+hjrH4nCeoOiZ2FiknUIPy8mY9zZx7Om3k4b+bhvJmPc2ee6jBv5dm/1YKTQqFAaGgoIiIiMGTIEACAXq9HREQEpk6dWuJrcnJyioUjmUwGwPCbdSK6Y9PJOHy45TwA4I3+zTCwja/pOl4hgFvn7gSl6wcA/V3PS+VA/Y53rqvk2ZylwomIiKjOsupSvenTp2Ps2LFo3749wsPDsWDBAmRnZxur7I0ZMwZ+fn746KOPAAADBw7E/Pnz0bZtW+NSvbfeegsDBw40BigiAg5eScHM304CAMZ1CsRzXRoYnsjLgO/tQ5D9uRW4+i+QGW/6QpdAoGEvw5GlwC6A0oLXdCIiIiKqwawanEaMGIGkpCS8/fbbSEhIQEhICLZu3QovL8OZEjExMSZHmN58801IJBK8+eabiI2NhYeHBwYOHIgPP/zQWm+BqNq5kJiJST8fhUanR98W3njrseaQ5KUDBxfB5sA3CNNk3dnYxg5o0PVOqXC3YOsNnIiIiKgas3pxiKlTp5a6NG/Xrl0mj21sbPDOO+/gnXfeqYKREdU8iRl5GLfkMDLyChBa3wULhgZBtucz4MBCIC8dEgBZSi/YhTwBWePeQL2OgNzW2sMmIiIiqvasHpyIyDIy87QYt/QI4tLz0NxNihVN9sH2m9FA7m3DBh5NUdBlFiKuSNH/0ccg40msRERERGXG4ERUC2gK9Hh+5XFcjU/Cy6qdeFG3GbLdqYYn3RoB3V8DWgyF0OmBq1usO1giIiKiGojBiaiGE0Lgzd+PoNHVFVig3AR3fTqQB8A1COj2GtBqGCAtLJ6i01t1rEREREQ1FYMTUU1WkI+dv8zD9Cs/wVteuCTPuR7Q7VWg9UhAxr/iRERERJbAb1VENVGBBoj8Bdk7PkbPvARAAmTbesO+12ygzWjARmHtERIRERHVKgxORDWJrgA4+Suw+1MgLQb2ABKEC84GT0TP0TMBG6W1R0hERERUKzE4EdUEeh1wei2w62Pg9lUAQJJwwjcFg5HX+hl8NDwMkEisPEgiIiKi2ovBiag60+uAs+sNgSnlIgBAZ+eGr/IH4PucHghr5Iclw9pDwtBEREREVKkYnIiqI70eOL/JEJiSzhva7FyQEzYVTxxrgfM5ejT3UWPRU+0gl0mtO1YiIiKiOoDBiag6EQKI2gzs+ghIPGNos3UCOr2IvHbP4emfz+F8Shr8nO2wdHwYHG15EVsiIiKiqsDgRFQdCAFc/Af490Mg/qShTakGHnoBeOh56JROmPbLMRyPSYPa1gbLxofBS21r3TETERER1SEMTkTWJARwOQL4dy4Qe8zQpnAAOvwf0HEKoHKFEALvbTqLbWcToZBJ8eOY9mjk5WjdcRMRERHVMQxORNYgBHB1tyEw3ThoaJOrgPCJQKdpgL2bcdMf91zB8gPXAQDzR7RBhyC3knokIiIiokrE4ERU1a7tMwSm63sNj21sgfYTgM4vAw6eJptuOhmHuVuiAABvDmiGx1r7VvFgiYiIiAhgcCKqOjGHgF1zgSu7DI9lCiB0PND5FUDtU2zzA5dTMPM3w/lO4x8OxITODapwsERERER0NwYnosp285ghMF3aYXgslQPtxgBdZgBOfiW+5EJiJiatOAqNTo9+Lb3x5oDmvFYTERERkRUxOBFVlviThiV5F7YaHktkQNungC4zAZf6pb4sIT0P45YcRmZeAdrXd8EXI0IgkzI0EREREVkTgxORpSWcMVyHKeovw2OJFGgzCug6E3ANuu9LM/O0GLf0MOLS8xDkYY8fx7SHrVxWBYMmIiIiovthcCKylFtRhsB0bkNhgwRo9STQ7VXAveEDX64p0OP5lccRlZAJdwcllo8Ph4u9olKHTERERERlw+BEVFHJF4H/PgFO/w5AGNpaPA50fw3waFKmLoQQeO2PU9h7KRkqhQxLx4UhwFVVeWMmIiIionJhcCIyV8plYPc84NQaQOgNbc0GAt1nA14tytXV5/9cwLoTsZBJJfjmqXZo5e9UCQMmIiIiInMxOBGV1+3rhsAUuQoQOkNbk/6GI0w+bcrd3S+HruPrfy8BAOYObYkeTTwf8AoiIiIiqmoMTkRllX4T2P0ZcGIFoC8wtDXqbQhMfqFmdRlxPhFvbTgDAJjWsxFGhNWz1GiJiIiIyIIYnIgeJCMe2DsfOLYM0GkMbUE9gB6vAwHhZncbeSMNU1edgF4AT4b64+VHG1lmvERERERkcQxORKXJTAT2LQCOLgEK8gxtgV0Mgal+pwp1fT0lGxOWHUGuVoeujT0w9/FWvMAtERERUTXG4ER0r+xkQ2A6/BNQkGtoC3gIeOQNoEHXCnefmq3BuKVHkJKtQQtfNRY91Q5ymbTC/RIRERFR5WFwIiqSkwrsXwgc+h7QZhva/NobAlNQD8ACR4RyNTpMWH4EV5Oz4edsh6XjwuCg5F9DIiIiouqO39iIctOAg4uAA4sATaahzbct0OMNoOGjFglMAKDTC0xbfQInYtLgZCfH8mfD4Km2tUjfRERERFS5GJyo7srLAA59B+z/GshPN7R5tTKcw9Skn8UCE2C4wO27f57FP+cSobCR4qex7dHQ09Fi/RMRERFR5WJworonPws4/L1hWV7ubUObRzOgx2yg6UBAavnzjX7YfQU/H7gOiQRYMCIEYYGuFt8HEREREVUeBieqOzQ5wJGfDIUfclIMbe6NDddhaj60UgITAGyMjMVHf0cBAN4c0Bz9W/lUyn6IiIiIqPIwOFHtp80Fji4F9n4BZN8ytLkGGwJTyycAqazSdr3/cjJmrj0JAHj24QaY0LlBpe2LiIiIiCoPgxPVXgX5wPGfgT2fA5nxhjbn+kC3V4HWIwBZ5X78oxMyMXnFMWh1Av1beePNAc0qdX9EREREVHkYnKj2KdAAkSuB3Z8DGTcNbU4BQNf/ASGjAZm80oeQkJ6HcUsPIzOvAGGBLpg/PARSKS9wS0RERFRTMThR7aHTAidXA7s/BdJiDG2OvkDXGUDbZwAbZZUMIyNPi3FLDyM+PQ/BHvb4cUx72MorbzkgEREREVU+Bieq+XQFwOm1wH+fALevGtocvIAuM4B2YwF51V0rSVOgx/MrjyEqIRMejkosGx8OZ5WiyvZPRERERJWDwYlqLqGH5OwfwJ7PgJSLhjaVO9D5FSBsAiC3q9rhCIFX/ziFfZdSoFLIsHRcGAJcVVU6BiIiIiKqHAxOVCNJoregR9QbsImMNTTYuQIPTwPCJwIKe6uM6bN/orH+RCxkUgkWPdUOLf2crDIOIiIiIrI8Bieqea7thc3vY6AGIGydIOn0ItDh/wClo9WGtPLgdXzz72UAwEdDW6F7E0+rjYWIiIiILI/BiWqeA4sAAHFO7eHx3G+QO7pZdTg7ziXi7Y1nAAAvP9oIw8MCrDoeIiIiIrI8qbUHQFQut68DF/4GAJz3HQbYqq06nMgbaZj663HoBTC8vT+m9Wxk1fEQERERUeVgcKKa5egSQOihb9ANWba+Vh3KteRsTFh2BHlaPbo19sCHQ1tBIuG1moiIiIhqIwYnqjm0ucDx5QAAffvnrDqUlKx8jFt6GCnZGrT0U2PRU+0gl/GvExEREVFtVe5veoGBgXjvvfcQExNTGeMhKt2ZP4Dc24BTPYiGva02jFyNDs/9fBTXUnLg72KHJePCYK/k6YJEREREtVm5g9PLL7+MdevWISgoCL169cLq1auRn59fGWMjukMI4ND3hvthEwCpzCrD0OkFXlp9Aidi0uBkJ8ey8eHwdKy6C+wSERERkXWYFZwiIyNx+PBhNGvWDC+++CJ8fHwwdepUHD9+vDLGSATcPAIknAJsbIF2Y6wyBCEE5mw6i+3nEqGwkeKnse3R0NPBKmMhIiIioqpl9kkZ7dq1w1dffYW4uDi88847+OmnnxAWFoaQkBAsWbIEQghLjpPquqKjTa2GASpXqwzh+91XsOLgdUgkwJcjQhAWaJ1xEBEREVHVM/vEDK1Wi/Xr12Pp0qXYvn07HnroIUyYMAE3b97E66+/jh07dmDVqlWWHCvVVZkJwLkNhvthE60yhI2Rsfj47ygAwFsDmqNfKx+rjIOIiIiIrKPcwen48eNYunQpfv31V0ilUowZMwZffPEFmjZtatxm6NChCAsLs+hAqQ47thzQFwABHQDfkCrf/f7LyZi59iQA4LnODfBs5wZVPgYiIiIisq5yB6ewsDD06tUL3377LYYMGQK5XF5smwYNGmDkyJEWGSDVcQUaw7WbACB8UpXvPiohA5N/PgatTmBAax+83r9ZlY+BiIiIiKyv3MHpypUrqF+//n23sbe3x9KlS80eFJFR1J9AVgLg4AU0G1Slu45Pz8X4pUeQmV+A8EBXfP5kG0ilvMAtERERUV1U7uIQt27dwqFDh4q1Hzp0CEePHrXIoIiMDv9ouA0dD9goqmy3GXlajF96BPHpeWjo6YAfxoTCVm6dEuhEREREZH3lDk5TpkzBjRs3irXHxsZiypQpFhkUEQAg/hQQcwCQ2gCh46pst5oCPf5vxTFEJWTCw1GJZePD4KyqutBGRERERNVPuYPTuXPn0K5du2Ltbdu2xblz5ywyKCIAwOEfDLfNBwPqqqliJ4TArN9PYv/lFNgrZFg6Lgz+Lqoq2TcRERERVV/lDk5KpRKJiYnF2uPj42FjY3Z1cyJTOanA6bWG+1VYFGLetmhsiIyDjVSCb58ORUs/pyrbNxERERFVX+UOTr1798bs2bORnp5ubEtLS8Prr7+OXr16WXRwVIedWAkU5AHerQxlyKvAioPXsWjXZQDAR4+3QtfGHlWyXyIiIiKq/sp9iOizzz5D165dUb9+fbRt2xYAEBkZCS8vL6xYscLiA6Q6SK8DjhQWhQifDEgqv5Ld9nOJeGfjGQDAK482xpPtAyp9n0RERERUc5Q7OPn5+eHUqVP45ZdfcPLkSdjZ2WH8+PEYNWpUidd0Iiq3i/8AaTGAnQvQalil7+5EzG28+Otx6AUwon0AXurZsNL3SUREREQ1i1knJdnb22PSpKq/GCnVEUVFIdo+A8jtKnVX15KzMWH5UeRp9ejexAMfDG0JSRUc4SIiIiKimsXsag7nzp1DTEwMNBqNSfugQVV7kVKqZZIvApd3ApAAYRMqdVcpWfkYu/QwUrM1aOmnxjej20EuK/dpf0RERERUB5Q7OF25cgVDhw7F6dOnIZFIIIQAAONv6XU6nWVHSHVL0QVvm/QDXAIrbTe5Gh0mLD+K6yk58Hexw5JxYbBXsiokEREREZWs3L9enzZtGho0aIBbt25BpVLh7Nmz2L17N9q3b49du3ZVwhCpzsjPBCJXGe6HT6y03ej0Ai/+egKRN9LgrJJj+bPh8HS0rbT9EREREVHNV+5fsR84cAA7d+6Eu7s7pFIppFIpOnfujI8++ggvvfQSTpw4URnjpLrg5GpAkwm4NQIadK+UXQgh8M6mM9hxPhEKGyl+GtMewR4OlbIvIiIiIqo9yn3ESafTwdHREQDg7u6OuLg4AED9+vURHR1t2dFR3SHEnWV64ZMAaeWca/Tdf1ew8mAMJBLgyxEhaB/oWin7ISIiIqLapdxHnFq2bImTJ0+iQYMG6NChAz799FMoFAr88MMPCAoKqowxUl1w9T8gORpQOABtRlbKLjaciMUnW6MAAG8/1hz9WvlUyn6IiIiIqPYpd3B68803kZ2dDQB477338Nhjj6FLly5wc3PDmjVrLD5AqiOKjja1GQXYqi3e/f5Lyfjf7ycBABO7NMD4hxtYfB9EREREVHuVOzj16dPHeL9hw4aIiopCamoqXFxceP0bMk9aDBC9xXC/EopCRCVkYPKKY9DqBAa09sHsfs0svg8iIiIiqt3KdSKJVquFjY0Nzpw5Y9Lu6urK0ETmO7IYEHogqDvg0cSiXcen52LckiPIzC9AeANXfP5kG0il/KwSERERUfmUKzjJ5XLUq1eP12oiy9HmAsd/NtwPn2TRrjPytBi35AgSMvLQ0NMBPz7THrZymUX3QURERER1Q7lLl73xxht4/fXXkZqaWhnjobrmzDogNxVwCgAa97VYt5oCPSb/fAzRiZnwdFRi2fgwOKnkFuufiIiIiOqWcp/j9PXXX+PSpUvw9fVF/fr1YW9vb/L88ePHLTY4quWEAA5/b7gfNgGQWuZokF4vMOv3kzhwJQX2ChmWjg+Dv4vKIn0TERERUd1U7uA0ZMiQShgG1Uk3jwDxJwEbW6DdWIt1O++faGyIjIONVIJvnw5FC18ni/VNRERERHVTuYPTO++8UxnjoLro8A+G25bDAJVlLkS74sA1fLvrMgDg4ydao2tjD4v0S0RERER1W7nPcSKyiMxE4OwGw30LlSD/52wC3tl0FgAwo1djDAv1t0i/RERERETlPuIklUrvW3qcFfeoTI4tA/RaIKAD4BtS4e6Ox9zGS6tPQC+AUeEBmPpIwwr3SURERERUpNzBaf369SaPtVotTpw4geXLl+Pdd9+12MCoFtNpgaNLDPctUIL8anI2nlt+FHlaPXo08cD7g1vyumJEREREZFHlDk6DBw8u1jZs2DC0aNECa9aswYQJEywyMKrFzv8JZCUA9p5As0EV6iolKx/jlh5BarYGrf2d8PXodrCRcQUqEREREVmWxb5hPvTQQ4iIiLBUd1SbHf7RcNt+PGCjMLsbjQ6Y9MsJXE/JQYCrHRaPDYO9sty/CyAiIiIieiCLfMvMzc3FV199BT8/P0t0R7VZwmkgZj8gtQFCx5vdTYFOj+UXpThzOwMuKjmWjw+Hh6PSggMlIiIiIrqj3MHJxcXF5PwRIQQyMzOhUqmwcuVKiw6OaqGiEuTNBgFqH7O6EELg/S1ROHNbCqWNFD+NbY8gDwcLDpKIiIiIyFS5g9MXX3xhEpykUik8PDzQoUMHuLi4WHRwVMvkpAKn1hruV6AoxL5LKVh1+CYkEJj/ZCuE1rfMNaCIiIiIiEpT7uA0bty4ShgG1QknVgIFuYB3K6DeQ2Z38+vhGABAJy+B3s29LDU6IiIiIqJSlbs4xNKlS7F27dpi7WvXrsXy5cstMiiqhfQ64MhPhvvhkwAzy4WnZOXjn3MJAICHvfSWGh0RERER0X2VOzh99NFHcHd3L9bu6emJuXPnWmRQVAtd3A6kXQdsnYGWw8zuZv2JWGh1Aq381PCzt9zwiIiIiIjup9zBKSYmBg0aNCjWXr9+fcTExFhkUFQLFRWFaPcMoFCZ1YUQAmuO3AAADGvHCo5EREREVHXKHZw8PT1x6tSpYu0nT56Em5ubRQZFtUzyReByBAAJEPac2d0cj0nDxVtZsJVLMbC1t+XGR0RERET0AOUOTqNGjcJLL72Ef//9FzqdDjqdDjt37sS0adMwcuTIyhgj1XRF5zY17gu4BJrdzW+FR5sGtPKFo63cAgMjIiIiIiqbclfVe//993Ht2jX07NkTNjaGl+v1eowZM4bnOFFx+ZlA5CrD/fCJZneTlV+AP0/FAQBGhAVYYmRERERERGVW7iNOCoUCa9asQXR0NH755ResW7cOly9fxpIlS6BQKMwaxDfffIPAwEDY2tqiQ4cOOHz4cKnbdu/eHRKJpNifAQMGmLVvqmQnVwP5GYBbIyCoh9nd/HUyDjkaHYLc7REWyOuFEREREVHVKvcRpyKNGjVCo0aNKjyANWvWYPr06fjuu+/QoUMHLFiwAH369EF0dDQ8PT2Lbb9u3TpoNBrj45SUFLRp0wZPPvlkhcdCFiYEcPhHw/3wiYC03DndaM1RwzK9EWEBJhdgJiIiIiKqCuX+JvvEE0/gk08+Kdb+6aefmhVe5s+fj4kTJ2L8+PFo3rw5vvvuO6hUKixZsqTE7V1dXeHt7W38s337dqhUKgan6ujqbiA5GlA4AG1Gmd1NdEImTsSkwUYqwePt/C04QCIiIiKisin3Eafdu3djzpw5xdr79euHzz//vFx9aTQaHDt2DLNnzza2SaVSPProozhw4ECZ+li8eDFGjhwJe/uSL+qTn5+P/Px84+OMjAwAgFarhVarLdd4K0PRGKrDWCxNduh7SAHoWg2HXmYHmPkefz10DQDQo4kHnG2lJj+72jhvlYnzZh7Om/k4d+bhvJmH82Yezpv5OHfmqU7zVp4xSIQQojyd29nZITIyEk2aNDFpj4qKQtu2bZGbm1vmvuLi4uDn54f9+/ejY8eOxvZZs2bhv//+w6FDh+77+sOHD6NDhw44dOgQwsPDS9xmzpw5ePfdd4u1r1q1CiqVedcTogez0ySj19kZkEAgotlHyLI177pLBXrg7WMyZBdIMKmpDi1cyvVxJSIiIiIqVU5ODkaPHo309HSo1er7blvuI06tWrXCmjVr8Pbbb5u0r169Gs2bNy9vdxWyePFitGrVqtTQBACzZ8/G9OnTjY8zMjIQEBCA3r17P3ByqoJWq8X27dvRq1cvyOW1p8S29N/3IYGAPrAruj5ufjW9LacTkH3oFLzUSkwf1RUyqeH8pto6b5WN82Yezpv5OHfm4byZh/NmHs6b+Th35qlO81a0Gq0syh2c3nrrLTz++OO4fPkyHnnkEQBAREQEVq1ahd9//71cfbm7u0MmkyExMdGkPTExEd7e97/AaXZ2NlavXo333nvvvtsplUoolcpi7XK53Oo/qLtVt/FUiDYPOLECACDtMBnSCryv308YSpAPbx8AW2Xxqo21at6qEOfNPJw383HuzMN5Mw/nzTycN/Nx7sxTHeatPPsvd3GIgQMHYsOGDbh06RJeeOEFzJgxA7Gxsdi5cycaNmxYrr4UCgVCQ0MRERFhbNPr9YiIiDBZuleStWvXIj8/H08//XR53wJVtrPrgNxUwCnAcNFbM91IzcHeS8kADMGJiIiIiMhazCpHPmDAAON1kzIyMvDrr79i5syZOHbsGHQ6Xbn6mj59OsaOHYv27dsjPDwcCxYsQHZ2NsaPHw8AGDNmDPz8/PDRRx+ZvG7x4sUYMmQI3NzczHkLVFmEAA59b7gfNgGQmV3xHmuP3YQQwMMN3RDgyvPRiIiIiMh6zP5Wu3v3bixevBh//PEHfH198fjjj+Obb74pdz8jRoxAUlIS3n77bSQkJCAkJARbt26Fl5cXACAmJgbSe67/Ex0djb179+Kff/4xd/hUWW4eBeIjAZkSaDvG7G50eoG1xms31bPQ4IiIiIiIzFOu4JSQkIBly5Zh8eLFyMjIwPDhw5Gfn48NGzZUqDDE1KlTMXXq1BKf27VrV7G2Jk2aoJzFAKmqHP7BcNtqGGBv/tHAPReTEJ+eB2eVHL2be1locERERERE5inzOU4DBw5EkyZNcOrUKSxYsABxcXFYuHBhZY6NaprMRODsesP98EkV6mrNEcPRpiEhfrCVyyo6MiIiIiKiCinzEae///4bL730Ep5//nk0atSoMsdENdXx5YBeC/iHA74hZneTnJWP7ecMlRZHhLEoBBERERFZX5mPOO3duxeZmZkIDQ1Fhw4d8PXXXyM5Obkyx0Y1iU4LHF1iuF/Bo03rj8eiQC/QJsAZzXysf60tIiIiIqIyB6eHHnoIP/74I+Lj4zF58mSsXr0avr6+0Ov12L59OzIzMytznFTdRf0FZMYD9p5A88FmdyOEwOojMQCAESxBTkRERETVRLmv42Rvb49nn30We/fuxenTpzFjxgx8/PHH8PT0xKBBgypjjFQTHCosCtF+PGBT/EK1ZXXs+m1cTsqGnVyGgW18LDQ4IiIiIqKKKXdwuluTJk3w6aef4ubNm/j1118tNSaqaRJOAzH7AakNEDq+Ql0VFYV4rLUPHG15BW4iIiIiqh4qFJyKyGQyDBkyBJs2bbJEd1TTHP7RcNtsIKA2/yhRZp4Wf52KB8CiEERERERUvVgkOFEdlpMKnPrNcD98coW6+vNkPHK1OgR72CO0vosFBkdEREREZBkMTlQxkb8ABbmAVyug3kMV6mrNUcMyvZFh9SCRSCwxOiIiIiIii2BwIvPpdcCRnwz3wycCFQg75+MzcPJGGmykEgxt52ehARIRERERWQaDE5nv0g7g9jXA1hlo9WSFuioqCtGruRfcHZQVHxsRERERkQUxOJH5Dn1vuG33DKBQmd1NnlaHDZGxAFgUgoiIiIiqJwYnMk/yJeByBAAJ0H5Chbr651wi0nK08HWyRZdGHpYZHxERERGRBTE4kXmKzm1q3AdwbVChrtYciQEADGsfAJmURSGIiIiIqPphcKLyy88yVNMDDEUhKuBGag72XUqBRAI8GepvgcEREREREVkegxOV36nVQH4G4NYQCHqkQl39VliCvHNDdwS4mn+eFBERERFRZWJwovIRAjj8o+F+2ERAav5HSKcXWHv0JgAWhSAiIiKi6o3Bicrn2h4gKQqQ2wMhoyrU1e4LSUjIyIOLSo5ezb0sNEAiIiIiIstjcKLyKSpBHjIKsHWqUFerC4tCDG3rD6WNrKIjIyIiIiKqNAxOVHZpN4DoLYb7YRUrCpGUmY+I87cAcJkeEREREVV/DE5UdkeXAEIPNOgKeDatUFfrjt9EgV6gbT1nNPF2tNAAiYiIiIgqB4MTlY02Dzi+3HA/fFKFuhJCYM0RQzW9Ee15tImIiIiIqj8GJyqbs+uAnBTAKQBo3K9CXR25dhtXkrOhUsjwWBtfCw2QiIiIiKjyMDjRgwlxpyhE+2cBmU2Fuis62jSwtS8clBXri4iIiIioKjA40YPFHgPiIwGZEmg3tkJdZeRpsfl0HABgOItCEBEREVENweBED1Z0tKnVMMDerUJdbYqMQ55Wj0aeDmhXz7niYyMiIiIiqgIMTnR/WbeAs+sN98MrVoIcAH47WlgUIiwAEomkwv0REREREVUFBie6v2PLAb0W8A8DfNtWqKuzcek4dTMdcpkEj7fzt9AAiYiIiIgqH4MTlU6nNVy7CahwCXIA+K2wKETv5t5wtVdUuD8iIiIioqrC4ESli/oLyIwD7D2B5kMq1FWeVof1J2IBGJbpERERERHVJAxOVLrDPxpuQ8cBNhU7QrTtbAIy8grg52yHzg3dKz42IiIiIqIqxOBEJUs4A1zfB0hkQPvxFe5u9WHDMr0n2/tDKmVRCCIiIiKqWRicqGSHfzDcNhsIqH0r1NX1lGwcuJICiQR4sj2X6RERERFRzcPgRMXl3gZO/Wa432FyhbsrKkHepZEH/JztKtwfEREREVFVY3Ci4k78AhTkAl4tgXodK9RVgU6PtUdvAgBGsigEEREREdVQDE5kSq8HjhQWhQifCFTwIrX/XUjCrcx8uNor8GgzLwsMkIiIiIio6jE4kalL24Hb1wBbJ6DV8Ap3t7rw2k2Pt/WDwoYfNyIiIiKqmfhNlkwVFYVo+wygUFWoq1sZedgZdQsAr91ERERERDUbgxPdkXIZuLQDgAQIm1Dh7v44HgudXiC0vgsaeTlWfHxERERERFbC4ER3FF3wtlFvwDWoQl0JIbDmSAwAYARLkBMRERFRDcfgRAb5WUDkL4b7HSZVuLtDV1NxLSUH9goZBrT2qXB/RERERETWxOBEBqfWAPkZgGswEPRIhbv7rbAoxKAQX9grbSrcHxERERGRNTE4ESDEnWV64RMBacU+Fum5Wmw+HQ8AGM5lekRERERUCzA4EXBtD5B0HpDbAyGjK9zdpshY5Bfo0cTLESEBzhUfHxERERGRlTE40Z0S5G1GGq7fVEFrjhqW6Y0IC4CkghfQJSIiIiKqDhic6rq0G0DUZsP98IkV7u5MbDrOxGZAIZNiaFu/CvdHRERERFQdMDjVdUeXAEIPBHYBPJtVuLs1hUUherfwgou9osL9ERERERFVBwxOdZk2Dzi+3HC/w+QKd5en1WFDZCwAYGRYvQr3R0RERERUXTA41WVn1wM5KYDaH2jcr8Ld/X0mHpl5BfB3sUOnYDcLDJCIiIiIqHpgcKrLiopChD0LyCp+raXVhw3L9Ia3D4BUyqIQRERERFR7MDjVVTePAnHHAZkSaDe2wt1dTc7GoaupkEqAYaH+FhggEREREVH1weBUVxUdbWr5BGDvXuHufissQd61sQd8ne0q3B8RERERUXXC4FQXZd0ynN8EWKQEeYFOj9+P3QQAjAwLqHB/RERERETVDYNTXXR8OaDTAH7tAb92Fe7u3+gkJGXmw91BgUeaellggERERERE1QuDU12j0wJHlhjuW6AEOQCsORIDAHi8nT8UNvxIEREREVHtw2+5dU3UZiAzDrD3AJoPrnB3iRl52Bl1C4Chmh4RERERUW3E4FTXHP7RcBs6DrBRVri734/dhF4AYYEuaOjpUOH+iIiIiIiqIwanuiThDHB9LyCRAaHjK9ydXi+M1fR4tImIiIiIajMGp7rkSOHRpmYDASe/Cnd38GoKrqfkwEFpgwGtfSrcHxERERFRdcXgVFfk3gZO/Wa4Hz7JIl3+dsRwtGlQiC9UChuL9ElEREREVB0xONUVkasAbQ7g2QKo36nC3aXnaLHlTAIAYASX6RERERFRLcfgVBfo9XeKQnSYBEgkFe5yQ2QsNAV6NPV2RGt/pwr3R0RERERUnTE41QWXdgC3rwK2TkCrJyvcnRACqwuX6Y0MC4DEAkGMiIiIiKg6Y3CqCw7/YLht+wygsK9wd2diM3A+PgMKGymGtK14kQkiIiIiouqOwam2S7kMXNoOQAKETbBIl6uPxAAA+rbwhrNKYZE+iYiIiIiqMwan2u7IT4bbRr0B16AKd5er0WFTZBwAwzI9IiIiIqK6gMGpNsvPAk78YrhvoRLkW07HIzO/AAGudngoyM0ifRIRERERVXcMTrXZ6d+A/HTANRgIfsQiXa4pLAoxon0ApFIWhSAiIiKiuoHBqbYSAjhUWBQifCIgrfiP+kpSFg5fS4VUAgwL5TI9IiIiIqo7GJxqq2t7gaTzgNweaDPKIl2uOWo42tS9iSe8nWwt0icRERERUU3A4FRbFZUgbzMCsHOucHdanR5/HLsJABjBohBEREREVMcwONVG6TeBqM2G+2ETLdLlzqhbSM7SwN1BiUeaelqkTyIiIiKimoLBqTY6ugQQOiCwC+DV3CJdFhWFeCLUD3IZPzZEREREVLfwG3Bto80Dji0z3LdQCfKE9Dzsir4FwFBNj4iIiIiormFwqm3ObQByUgC1H9Ckv0W6/P3YDegFEN7AFUEeDhbpk4iIiIioJmFwqm0OfW+4bf8sILOpcHd6vTBW0+PRJiIiIiKqqxicapObx4C444BMAYSOs0iXB66k4EZqLhyVNujfyscifRIRERER1TQMTrVJUQnylk8A9u4W6bKoKMTgtr6wU8gs0icRERERUU3D4FRbZCUBZ9cZ7odbpgT57WwNtp5JAACMaF/PIn0SEREREdVEDE61xfFlgE4D+LUH/EIt0uWGyFhodHo091GjpZ/aIn0SEREREdVEDE61ga4AOLLEcN9CJciFEMZleiPDAyCRSCzSLxERERFRTcTgVBtEbwYy4wCVO9BiiEW6PHUzHVEJmVDYSDG4jZ9F+iQiIiIiqqkYnGqDQ4VFIULHATZKi3S5uvBoU/+W3nBSyS3SJxERERFRTcXgVNMlngWu7wUkMsO1mywgR1OAP0/GAQBGhLEoBBERERERg1NNd/hHw22zxwAnyyyp23wqHln5BajvpsJDQa4W6ZOIiIiIqCZjcKrJctOAU2sM9y1UFAK4c+2m4e1ZFIKIiIiICGBwqtkifwG0OYBnC6D+wxbp8tKtLBy9fhsyqQTDQv0t0icRERERUU3H4FRT6fV3lumFTwQsdGTot6OGo009mnjAS21rkT6JiIiIiGo6qwenb775BoGBgbC1tUWHDh1w+PDh+26flpaGKVOmwMfHB0qlEo0bN8aWLVuqaLTVyOUI4PZVQOkEtB5ukS41BXr8cewmABaFICIiIiK6m401d75mzRpMnz4d3333HTp06IAFCxagT58+iI6OhqenZ7HtNRoNevXqBU9PT/z+++/w8/PD9evX4ezsXPWDt7ZD3xtu2z4NKOwt0uXOqESkZGvg6ahEjyYeFumTiIiIiKg2sGpwmj9/PiZOnIjx48cDAL777jts3rwZS5YswWuvvVZs+yVLliA1NRX79++HXG64tlBgYGBVDrl6SLkMXNoOQAKETbBYt0XXbnoi1B82MqsfjCQiIiIiqjasFpw0Gg2OHTuG2bNnG9ukUikeffRRHDhwoMTXbNq0CR07dsSUKVOwceNGeHh4YPTo0Xj11Vchk8lKfE1+fj7y8/ONjzMyMgAAWq0WWq3Wgu/IPEVjKM9YpId+gAyAPrgndOp6gAXeR3x6Hv67kAQAeDzEu1rMzf2YM2/EeTMX5818nDvzcN7Mw3kzD+fNfJw781SneSvPGCRCCFGJYylVXFwc/Pz8sH//fnTs2NHYPmvWLPz33384dOhQsdc0bdoU165dw1NPPYUXXngBly5dwgsvvICXXnoJ77zzTon7mTNnDt59991i7atWrYJKpbLcG6oiMl0++pydBrkuBweCZuCWUxuL9LvtpgRbbsjQUK3Hiy30FumTiIiIiKg6y8nJwejRo5Geng61Wn3fba26VK+89Ho9PD098cMPP0AmkyE0NBSxsbGYN29eqcFp9uzZmD59uvFxRkYGAgIC0Lt37wdOTlXQarXYvn07evXqZVx+eD/S48sgO5UD4dIA7Ue+CkgqvqROrxf49Is9APLwf71ao3+Ib4X7rGzlnTcy4LyZh/NmPs6deThv5uG8mYfzZj7OnXmq07wVrUYrC6sFJ3d3d8hkMiQmJpq0JyYmwtvbu8TX+Pj4QC6XmyzLa9asGRISEqDRaKBQKIq9RqlUQqlUFmuXy+VW/0HdrUzjEQI4tgQAIAmfBLmi+Psyx56LSYhNy4OjrQ0eC/GHXF7yssfqqLr9HGsKzpt5OG/m49yZh/NmHs6beThv5uPcmac6zFt59m+1CgAKhQKhoaGIiIgwtun1ekRERJgs3bvbww8/jEuXLkGvv7OU7MKFC/Dx8SkxNNU61/cBt84BchUQMtpi3a4pLAoxtK0fbGtQaCIiIiIiqipWLZ02ffp0/Pjjj1i+fDnOnz+P559/HtnZ2cYqe2PGjDEpHvH8888jNTUV06ZNw4ULF7B582bMnTsXU6ZMsdZbqFpFJchbjwDsnC3SZWq2Bv+cNRz1G94+wCJ9EhERERHVNlY9x2nEiBFISkrC22+/jYSEBISEhGDr1q3w8vICAMTExEAqvZPtAgICsG3bNrzyyito3bo1/Pz8MG3aNLz66qvWegtVJ/0mELXZcD98ksW6XX8iFhqdHi391Gjp52SxfomIiIiIahOrF4eYOnUqpk6dWuJzu3btKtbWsWNHHDx4sJJHVQ0dXQoIHRDYBfBqbpEuhRD4rXCZ3oiwehbpk4iIiIioNuJVTmuCgnzg2DLD/fCJFus28kYaohMzobSRYlCb6l9Jj4iIiIjIWhicaoKz64GcZEDtBzQZYLFui4pCDGjlAyc7VoIhIiIiIioNg1NNcPgHw237ZwGZZVZXZucX4M+TcQCAEWEsCkFEREREdD8MTtXdzWNA7DFApgDajbVYt5tPxSNbo0MDd3uEN3C1WL9ERERERLURg1N1V3S0qcXjgIOHxbpdfSQGgKEEuUQisVi/RERERES1EYNTdZaVBJxdZ7jfwXIlyC8mZuJ4TBpkUgmeCPWzWL9ERERERLUVg1N1dnw5oNMAfqGGPxZSVBTikaae8HS0tVi/RERERES1FYNTdaUrAI4uMdy34AVvNQV6rDsRCwAYyaIQRERERERlwuBUXUVvBjJiAZU70GKoxbrdcT4RqdkaeKmV6NbYcudMERERERHVZgxO1dXhHw23oeMAG6XFul1duExvWKg/bGT88RMRERERlQW/OVdHieeAa3sAiQxoP95i3d68nYM9F5MAGKrpERERERFR2TA4VUdHCo82NR0AOPlbrNvfj92EEECnYDfUd7O3WL9ERERERLUdg1N1k5sGnFxtuN9hssW61ekF1h69CQAYwaIQRERERETlwuBU3USuArQ5gGdzoP7DFut276VkxKblwslOjj4tvC3WLxERERFRXcDgVJ3o9XeW6YVPBCQSi3X9W2FRiKFt/WArl1msXyIiIiKiuoDBqTq5HAGkXgGUTkCr4RbrNiUrH/+cSwDAohBEREREROZgcKpODv9guG37NKB0sFi360/EQqsTaO3vhOa+aov1S0RERERUVzA4VRepV4CL2w33wyZYrFshBNYULtNjUQgiIiIiIvMwOFUT0uNLAQigYS/ALdhi/R6PScPFW1mwlUsxsI2vxfolIiIiIqpLGJyqAZkuH9KTqwwPwidZtO81R2IAAANa+UJtK7do30REREREdQWDUzXgf3s/JHnpgEsDoOGjFus3K78Af52KBwCMDOcyPSIiIiIiczE4WZsQaJC0w3A/fCIgtdyP5K+TccjR6BDkYY/29V0s1i8RERERUV3D4GRlkpj9cMq7ASFXASFPWbTv1UVFIdoHQGLBa0IREREREdU1DE5WJj26GACgb/kkYOdssX6jEzIReSMNNlIJHm/nb7F+iYiIiIjqIgYna0qPhSR6MwBA395yJcgBGEuQ92zmCQ9HpUX7JiIiIiKqaxicrOnYUkiEDskOTQHP5hbrNr9Ah3UnbgIARobVs1i/RERERER1FYOTNTl4QTj64oq75SrpAcD2c4lIy9HCW22Lro09LNo3EREREVFdxOBkTeETUTD1OOKd21u026Jlek+294dMyqIQREREREQVxeBkbVIbQGK5H8ON1BzsuZgMABjentduIiIiIiKyBAanWmbtMcO5TZ0buiPAVWXl0RARERER1Q4MTrWITi+w9qhhmd7wMB5tIiIiIiKyFAanWmT3xSTE/3979x4cVX33cfxzEnIn4WruFwQUAhIkCWBA7VQQTJFHLGKwaRug6jgNCM1oC7QWKAhaW6AjGMFBnBGRixVEHzFirFARJASDARFBLSABAgq5YULInucPJI8pyuKO8Dsneb9mzgx7djf57DcZJp895/y2olZtQwM0pEeU6TgAAABAs0FxakZWfbMoxJ194hQc4G84DQAAANB8UJyaiRPVddrw0TFJUhan6QEAAAA/KopTM/Hyji901mOrd0JbdY+OMB0HAAAAaFYoTs2AbduNn900mqNNAAAAwI+O4tQMFB84qU+P1ygkwF+3p8SYjgMAAAA0OxSnZmDFN0ebbk+JUXhwgOE0AAAAQPNDcXK5qtp6/e+HRyRJo/txmh4AAABwOVCcXO7VnUf0dX2DulwVptTEdqbjAAAAAM0SxcnlVhYdlCSN7psoy7IMpwEAAACaJ4qTi+05UqmdX1QowN/SnalxpuMAAAAAzRbFycXOL0E+ODlKHVsHGU4DAAAANF8UJ5eqrW/Qmg8OS5Ky+OwmAAAA4LKiOLnUmx8dU8XX9YptE6ybrrnKdBwAAACgWaM4udT5RSHuSk+Qvx+LQgAAAACXE8XJhQ5+eVqb938py5JGpcWbjgMAAAA0exQnF1pdfG5RiBu7dlRC+1DDaQAAAIDmj+LkMmcbPFq9/QtJLAoBAAAAXCkUJ5fZtO+4jlbWql1ogG7tEWU6DgAAANAiUJxc5vxnN/08NV5BrfwNpwEAAABaBoqTi5RX1apwT7kkTtMDAAAAriSKk4u8vOOwznps9Ulsq2ujwk3HAQAAAFoMipNL2LatVd+cpjeao00AAADAFUVxcomi/5zUZydqFBror2EpsabjAAAAAC0KxcklVhQdlCQNT4lV66BWhtMAAAAALQvFyQUqa+v1eukRSVJWP07TAwAAAK40ipMLrCspU229R9dEtlafhLam4wAAAAAtDsXJBc5/dlNW3wRZlmU4DQAAANDyUJwcbndZhUoPVyjA39LPU+NNxwEAAABaJIqTw51fgnxIj2i1Dws0nAYAAABomShODlZb36A1HxyWdO40PQAAAABmUJwcrGD3UVXWnlVc2xDd2LWj6TgAAABAi0VxcrAV286dpjcqPV5+fiwKAQAAAJhCcXKoA1/WaMtnX8qypFHpnKYHAAAAmERxcqhV288dbbr5mqsU1zbEcBoAAACgZaM4OdDZBo9Wb/9CEotCAAAAAE5AcXKgd/YeV3lVndqHBWpwcpTpOAAAAECLR3FyoJXfnKY3MjVOga34EQEAAACm8Ve5w5RX1urtj8slcZoeAAAA4BQUJ4d5accXavDYSktqp66R4abjAAAAABDFyVFs29aqonOn6XG0CQAAAHAOipODvP/5V/rPl6cVFuivYb1iTMcBAAAA8A2Kk4Os/OZo0/9cH6uwoFaG0wAAAAA4j+LkEJVf1+v10iOSpKy+iYbTAAAAAPg2ipNDvPrhEdWd9ahbVLh6x7cxHQcAAADAt1CcHGJV8WFJ5xaFsCzLcBoAAAAA30ZxcoBD1dJHR6oU6O+nO/vEmY4DAAAA4L9QnBxga/m5H8OQnlFqFxZoOA0AAACA/0ZxMqy2vkHFJ86dmjeaRSEAAAAAR6I4GVaw+5i+brAU3zZYA7p0MB0HAAAAwHegOBl2flGIkalx8vNjUQgAAADAiShOBn1+okbb/nNSlmyNTGVRCAAAAMCpKE4Grdp+SJKU3NZWTJtgw2kAAAAAfB+Kk0GdOoSqc8cw3RBpm44CAAAA4CIoTgZl9U3UGw8OUK/2FCcAAADAyShOhlmWJdaEAAAAAJyN4gQAAAAAXjiiOC1cuFCdOnVScHCw+vfvr23btn3vY5977jlZltVkCw5mYQUAAAAAl4/x4rRy5Url5eVp2rRp2rFjh3r37q2hQ4eqvLz8e58TERGhI0eONG4HDhy4gokBAAAAtDTGi9PcuXN13333aezYserRo4eefvpphYaG6tlnn/3e51iWpejo6MYtKirqCiYGAAAA0NK0MvnNz5w5o+LiYk2ZMqVxn5+fnwYPHqwtW7Z87/Oqq6uVlJQkj8ej1NRUzZ49Wz179vzOx9bV1amurq7xdmVlpSSpvr5e9fX1P9Ir8d35DE7I4ibMzTfMzTfMzXfMzjfMzTfMzTfMzXfMzjdOmtsPyWDZtm1sLeyysjLFxcXpvffeU0ZGRuP+3//+99q4caPef//9C56zZcsW7du3TykpKaqoqNDf/vY3bdq0Sbt371Z8fPwFj58+fbpmzJhxwf7ly5crNDT0x31BAAAAAFzj9OnT+sUvfqGKigpFRERc9LFGjzj5IiMjo0nJGjBggJKTk7Vo0SLNnDnzgsdPmTJFeXl5jbcrKyuVkJCgIUOGeB3OlVBfX68NGzbo1ltvVUBAgOk4rsHcfMPcfMPcfMfsfMPcfMPcfMPcfMfsfOOkuZ0/G+1SGC1OHTt2lL+/v44dO9Zk/7FjxxQdHX1JXyMgIEB9+vTR/v37v/P+oKAgBQUFfefzTP+gvs1pedyCufmGufmGufmO2fmGufmGufmGufmO2fnGCXP7Id/f6OIQgYGBSktLU2FhYeM+j8ejwsLCJkeVLqahoUGlpaWKiYm5XDEBAAAAtHDGT9XLy8tTTk6O0tPT1a9fP82fP181NTUaO3asJOnXv/614uLiNGfOHEnSX/7yF91www3q2rWrTp06pSeeeEIHDhzQvffea/JlAAAAAGjGjBenrKwsHT9+XH/+85919OhRXX/99XrjjTcalxg/ePCg/Pz+/8DYyZMndd999+no0aNq166d0tLS9N5776lHjx6mXgIAAACAZs54cZKk8ePHa/z48d953zvvvNPk9rx58zRv3rwrkAoAAAAAzjH+AbgAAAAA4HQUJwAAAADwguIEAAAAAF5QnAAAAADAC4oTAAAAAHjhiFX1riTbtiVJlZWVhpOcU19fr9OnT6uystL4Jye7CXPzDXPzDXPzHbPzDXPzDXPzDXPzHbPzjZPmdr4TnO8IF9PiilNVVZUkKSEhwXASAAAAAE5QVVWlNm3aXPQxln0p9aoZ8Xg8KisrU3h4uCzLMh1HlZWVSkhI0KFDhxQREWE6jmswN98wN98wN98xO98wN98wN98wN98xO984aW62bauqqkqxsbHy87v4VUwt7oiTn5+f4uPjTce4QEREhPFfHDdibr5hbr5hbr5jdr5hbr5hbr5hbr5jdr5xyty8HWk6j8UhAAAAAMALihMAAAAAeEFxMiwoKEjTpk1TUFCQ6Siuwtx8w9x8w9x8x+x8w9x8w9x8w9x8x+x849a5tbjFIQAAAADgh+KIEwAAAAB4QXECAAAAAC8oTgAAAADgBcUJAAAAALygOBm0cOFCderUScHBwerfv7+2bdtmOpLjbdq0ScOHD1dsbKwsy9LatWtNR3KFOXPmqG/fvgoPD1dkZKRGjBihvXv3mo7lePn5+UpJSWn8gL6MjAytX7/edCzXeeyxx2RZliZNmmQ6iuNNnz5dlmU12bp37246liscPnxYv/zlL9WhQweFhISoV69e2r59u+lYjtapU6cLft8sy1Jubq7paI7W0NCgRx55RFdffbVCQkLUpUsXzZw5U6y35l1VVZUmTZqkpKQkhYSEaMCAASoqKjId65JRnAxZuXKl8vLyNG3aNO3YsUO9e/fW0KFDVV5ebjqao9XU1Kh3795auHCh6SiusnHjRuXm5mrr1q3asGGD6uvrNWTIENXU1JiO5mjx8fF67LHHVFxcrO3bt+uWW27RHXfcod27d5uO5hpFRUVatGiRUlJSTEdxjZ49e+rIkSON27vvvms6kuOdPHlSAwcOVEBAgNavX6+PPvpIf//739WuXTvT0RytqKioye/ahg0bJEmjRo0ynMzZHn/8ceXn52vBggXas2ePHn/8cf31r3/Vk08+aTqa4917773asGGDnn/+eZWWlmrIkCEaPHiwDh8+bDraJWE5ckP69++vvn37asGCBZIkj8ejhIQETZgwQZMnTzaczh0sy9KaNWs0YsQI01Fc5/jx44qMjNTGjRt18803m47jKu3bt9cTTzyh3/zmN6ajOF51dbVSU1P11FNPadasWbr++us1f/5807Ecbfr06Vq7dq1KSkpMR3GVyZMna/Pmzfr3v/9tOoqrTZo0Sa+99pr27dsny7JMx3Gs22+/XVFRUVqyZEnjvpEjRyokJETLli0zmMzZvv76a4WHh+uVV17RsGHDGvenpaUpMzNTs2bNMpju0nDEyYAzZ86ouLhYgwcPbtzn5+enwYMHa8uWLQaToaWoqKiQdK4E4NI0NDRoxYoVqqmpUUZGhuk4rpCbm6thw4Y1+b8O3u3bt0+xsbHq3LmzsrOzdfDgQdORHG/dunVKT0/XqFGjFBkZqT59+uiZZ54xHctVzpw5o2XLlmncuHGUJi8GDBigwsJCffLJJ5KknTt36t1331VmZqbhZM529uxZNTQ0KDg4uMn+kJAQ1xxZb2U6QEt04sQJNTQ0KCoqqsn+qKgoffzxx4ZSoaXweDyaNGmSBg4cqOuuu850HMcrLS1VRkaGamtr1bp1a61Zs0Y9evQwHcvxVqxYoR07drjq3HUn6N+/v5577jl169ZNR44c0YwZM3TTTTdp165dCg8PNx3PsT777DPl5+crLy9PU6dOVVFRkR588EEFBgYqJyfHdDxXWLt2rU6dOqUxY8aYjuJ4kydPVmVlpbp37y5/f381NDTo0UcfVXZ2tulojhYeHq6MjAzNnDlTycnJioqK0osvvqgtW7aoa9eupuNdEooT0MLk5uZq165drnl3x7Ru3bqppKREFRUVeumll5STk6ONGzdSni7i0KFDmjhxojZs2HDBO4u4uG+/Y52SkqL+/fsrKSlJq1at4vTQi/B4PEpPT9fs2bMlSX369NGuXbv09NNPU5wu0ZIlS5SZmanY2FjTURxv1apVeuGFF7R8+XL17NlTJSUlmjRpkmJjY/l98+L555/XuHHjFBcXJ39/f6Wmpuqee+5RcXGx6WiXhOJkQMeOHeXv769jx4412X/s2DFFR0cbSoWWYPz48Xrttde0adMmxcfHm47jCoGBgY3vhKWlpamoqEj/+Mc/tGjRIsPJnKu4uFjl5eVKTU1t3NfQ0KBNmzZpwYIFqqurk7+/v8GE7tG2bVtde+212r9/v+kojhYTE3PBmxnJycn65z//aSiRuxw4cEBvvfWWXn75ZdNRXOHhhx/W5MmTNXr0aElSr169dODAAc2ZM4fi5EWXLl20ceNG1dTUqLKyUjExMcrKylLnzp1NR7skXONkQGBgoNLS0lRYWNi4z+PxqLCwkGsncFnYtq3x48drzZo1evvtt3X11VebjuRaHo9HdXV1pmM42qBBg1RaWqqSkpLGLT09XdnZ2SopKaE0/QDV1dX69NNPFRMTYzqKow0cOPCCj1j45JNPlJSUZCiRuyxdulSRkZFNLtjH9zt9+rT8/Jr+Ce3v7y+Px2MokfuEhYUpJiZGJ0+eVEFBge644w7TkS4JR5wMycvLU05OjtLT09WvXz/Nnz9fNTU1Gjt2rOlojlZdXd3kndfPP/9cJSUlat++vRITEw0mc7bc3FwtX75cr7zyisLDw3X06FFJUps2bRQSEmI4nXNNmTJFmZmZSkxMVFVVlZYvX6533nlHBQUFpqM5Wnh4+AXXz4WFhalDhw5cV+fFQw89pOHDhyspKUllZWWaNm2a/P39dc8995iO5mi/+93vNGDAAM2ePVt33323tm3bpsWLF2vx4sWmozmex+PR0qVLlZOTo1at+LPwUgwfPlyPPvqoEhMT1bNnT33wwQeaO3euxo0bZzqa4xUUFMi2bXXr1k379+/Xww8/rO7du7vn718bxjz55JN2YmKiHRgYaPfr18/eunWr6UiO969//cuWdMGWk5NjOpqjfdfMJNlLly41Hc3Rxo0bZyclJdmBgYH2VVddZQ8aNMh+8803TcdypZ/85Cf2xIkTTcdwvKysLDsmJsYODAy04+Li7KysLHv//v2mY7nCq6++al933XV2UFCQ3b17d3vx4sWmI7lCQUGBLcneu3ev6SiuUVlZaU+cONFOTEy0g4OD7c6dO9t//OMf7bq6OtPRHG/lypV2586d7cDAQDs6OtrOzc21T506ZTrWJeNznAAAAADAC65xAgAAAAAvKE4AAAAA4AXFCQAAAAC8oDgBAAAAgBcUJwAAAADwguIEAAAAAF5QnAAAAADAC4oTAAAAAHhBcQIA4AewLEtr1641HQMAcIVRnAAArjFmzBhZlnXBdtttt5mOBgBo5lqZDgAAwA9x2223aenSpU32BQUFGUoDAGgpOOIEAHCVoKAgRUdHN9natWsn6dxpdPn5+crMzFRISIg6d+6sl156qcnzS0tLdcsttygkJEQdOnTQ/fffr+rq6iaPefbZZ9WzZ08FBQUpJiZG48ePb3L/iRMndOeddyo0NFTXXHON1q1bd3lfNADAOIoTAKBZeeSRRzRy5Ejt3LlT2dnZGj16tPbs2SNJqqmp0dChQ9WuXTsVFRVp9erVeuutt5oUo/z8fOXm5ur+++9XaWmp1q1bp65duzb5HjNmzNDdd9+tDz/8UD/72c+UnZ2tr7766oq+TgDAlWXZtm2bDgEAwKUYM2aMli1bpuDg4Cb7p06dqqlTp8qyLD3wwAPKz89vvO+GG25QamqqnnrqKT3zzDP6wx/+oEOHDiksLEyS9Prrr2v48OEqKytTVFSU4uLiNHbsWM2aNes7M1iWpT/96U+aOXOmpHNlrHXr1lq/fj3XWgFAM8Y1TgAAV/npT3/apBhJUvv27Rv/nZGR0eS+jIwMlZSUSJL27Nmj3r17N5YmSRo4cKA8Ho/27t0ry7JUVlamQYMGXTRDSkpK47/DwsIUERGh8vJyX18SAMAFKE4AAFcJCwu74NS5H0tISMglPS4gIKDJbcuy5PF4LkckAIBDcI0TAKBZ2bp16wW3k5OTJUnJycnauXOnampqGu/fvHmz/Pz81K1bN4WHh6tTp04qLCy8opkBAM7HEScAgKvU1dXp6NGjTfa1atVKHTt2lCStXr1a6enpuvHGG/XCCy9o27ZtWrJkiSQpOztb06ZNU05OjqZPn67jx49rwoQJ+tWvfqWoqChJ0vTp0/XAAw8oMjJSmZmZqqqq0ubNmzVhwoQr+0IBAI5CcQIAuMobb7yhmJiYJvu6deumjz/+WNK5Fe9WrFih3/72t4qJidGLL76oHj16SJJCQ0NVUFCgiRMnqm/fvgoNDdXIkSM1d+7cxq+Vk5Oj2tpazZs3Tw899JA6duyou+6668q9QACAI7GqHgCg2bAsS2vWrNGIESNMRwEANDNc4wQAAAAAXlCcAAAAAMALrnECADQbnH0OALhcOOIEAAAAAF5QnAAAAADAC4oTAAAAAHhBcQIAAAAALyhOAAAAAOAFxQkAAAAAvKA4AQAAAIAXFCcAAAAA8OL/AEhQ8ZrA4S/9AAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 1000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "pipeline_augmented.plot_acc(title=\"50% Random Swap, 50% Synonym on Self-Attention\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": ".venv",
   "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": "3.12.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}