hs_detection.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Hate Speech Detection\n",
    "This notebook performs hate speech detection on a dataset of tweets. The pipeline is as follows:\n",
    "- Load & preprocess data\n",
    "- Finetune a pretrained RoBERTa model for hate speech classification\n",
    "- Evaluate the model on the test set"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Setup"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/envs/aiahs/lib/python3.11/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
      "  from .autonotebook import tqdm as notebook_tqdm\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from tqdm import tqdm\n",
    "import torch\n",
    "from transformers import AutoTokenizer, AutoModel\n",
    "from pytorch_lightning import Trainer\n",
    "from src.data_utils import *\n",
    "from src.models import HateSpeechClassifier\n",
    "from src.train import train\n",
    "from src.utils import *\n",
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Config"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# set torch matmul precision\n",
    "# torch.set_float32_matmul_precision('medium')"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train & eval model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Set params & create dataloaders"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "# %%script false --no-raise-error\n",
    "# clear GPU memory\n",
    "torch.cuda.empty_cache()\n",
    "\n",
    "# create data loaders\n",
    "dataset = 'dynabench' # dataset, either 'ucb' or 'dynabench'\n",
    "data_path = f'data/hs_{dataset}_clean.csv' # path to data\n",
    "bsz = 128\n",
    "max_len = 256\n",
    "num_workers = 12\n",
    "lm_name = 'facebook/roberta-hate-speech-dynabench-r4-target' # lm on HF hub \n",
    "tokenizer = AutoTokenizer.from_pretrained(lm_name)  # use tokenizer from HF hub\n",
    "resample = True  # resample to balance classes\n",
    "\n",
    "# create data loaders (use resampling to balance classes)\n",
    "train_loader, val_loader, test_loader = create_dataloaders(data_path, tokenizer, max_len, bsz, num_workers, resample=resample)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%script false --no-raise-error\n",
    "# create the model\n",
    "model = HateSpeechClassifier(lm_name, pos_weight=1.0)\n",
    "\n",
    "# training args\n",
    "args = {'num_epochs': 20, 'precision': '32', 'patience': 3, 'ckpt_dir': f'model_ckpts/{dataset}/{lm_name}','ckpt_name': 'ckpt_best', 'resume_ckpt': None, 'tb_logdir': f'{dataset}/{lm_name}'}\n",
    "\n",
    "# train\n",
    "model, trainer = train(model, train_loader, val_loader, args)\n",
    "\n",
    "# test\n",
    "trainer.test(dataloaders=test_loader)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Evaluate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Some weights of the model checkpoint at facebook/roberta-hate-speech-dynabench-r4-target were not used when initializing RobertaModel: ['classifier.out_proj.bias', 'classifier.dense.weight', 'classifier.dense.bias', 'classifier.out_proj.weight']\n",
      "- This IS expected if you are initializing RobertaModel from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model).\n",
      "- This IS NOT expected if you are initializing RobertaModel from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).\n",
      "Some weights of RobertaModel were not initialized from the model checkpoint at facebook/roberta-hate-speech-dynabench-r4-target and are newly initialized: ['roberta.pooler.dense.bias', 'roberta.pooler.dense.weight']\n",
      "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n",
      "GPU available: True (cuda), used: True\n",
      "TPU available: False, using: 0 TPU cores\n",
      "IPU available: False, using: 0 IPUs\n",
      "HPU available: False, using: 0 HPUs\n",
      "LOCAL_RANK: 0 - CUDA_VISIBLE_DEVICES: [0]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predicting DataLoader 0: 100%|██████████| 33/33 [00:12<00:00,  2.72it/s]\n",
      "counts in test_labels: {0: 1852, 1: 2268}\n",
      "counts in test_preds: {0: 1703, 1: 2417}\n",
      "confusion matrix: \n",
      "[[1068  784]\n",
      " [ 635 1633]]\n"
     ]
    }
   ],
   "source": [
    "# get test predictions\n",
    "# run after training\n",
    "# test_out = trainer.predict(dataloaders=test_loader)\n",
    "\n",
    "# or load from checkpoint\n",
    "ckpt_path = f'model_ckpts/{dataset}/{lm_name}/ckpt_best.ckpt'\n",
    "model = HateSpeechClassifier.load_from_checkpoint(ckpt_path)\n",
    "trainer = Trainer()\n",
    "test_out = trainer.predict(model, dataloaders=test_loader)\n",
    "\n",
    "# stack batch preds and labels\n",
    "labels = torch.cat([x[0] for x in test_out]).cpu().numpy()\n",
    "preds_lm = torch.cat([x[1] for x in test_out]).cpu().numpy()\n",
    "# get value counts of 0's & 1's in test_labels, test_preds\n",
    "unique_labels, counts_labels = np.unique(labels, return_counts=True)\n",
    "unique_preds, counts_preds = np.unique(preds_lm, return_counts=True)\n",
    "\n",
    "# print the counts\n",
    "print(f\"counts in test_labels: {dict(zip(unique_labels, counts_labels))}\")\n",
    "print(f\"counts in test_preds: {dict(zip(unique_preds, counts_preds))}\")\n",
    "\n",
    "# print confusion matrix\n",
    "from sklearn.metrics import confusion_matrix\n",
    "print(f\"confusion matrix: \\n{confusion_matrix(labels, preds_lm)}\")\n",
    "\n",
    "# print some TNs, FPs, FNs, TPs\n",
    "tns = test_loader.dataset.df[(labels==0) & (preds_lm==0)][['text']]\n",
    "fps = test_loader.dataset.df[(labels==0) & (preds_lm==1)][['text']]\n",
    "fns = test_loader.dataset.df[(labels==1) & (preds_lm==0)][['text']]\n",
    "tps = test_loader.dataset.df[(labels==1) & (preds_lm==1)][['text']]\n",
    "# print(f'TNs: \\n{tns.sample(2)}\\nFPs: \\n{fps.sample(2)}\\nFNs: \\n{fns.sample(2)}\\nTPs: \\n{tps.sample(2)}')\n",
    "\n",
    "# create dataframe w samples of each\n",
    "num_samples = 20\n",
    "df_tn_samples = tns.sample(num_samples).assign(label=0, pred=0)\n",
    "df_fp_samples = fps.sample(num_samples).assign(label=0, pred=1)\n",
    "df_fn_samples = fns.sample(num_samples).assign(label=1, pred=0)\n",
    "df_tp_samples = tps.sample(num_samples).assign(label=1, pred=1)\n",
    "# concat samples into single df\n",
    "df_samples = pd.concat([df_tn_samples, df_fp_samples, df_fn_samples, df_tp_samples]).reset_index(drop=True)\n",
    "# save to csv\n",
    "samples_path = f'outputs/{dataset}/{lm_name}/conf_samples.tsv'\n",
    "os.makedirs(os.path.dirname(samples_path), exist_ok=True)\n",
    "df_samples.to_csv(samples_path, sep='\\t', index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## EOOH"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Viz EOOH annotations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "hate speech gt labels in original data: class\n",
      "0    0.942299\n",
      "1    0.057701\n",
      "Name: proportion, dtype: float64\n",
      "hate speech gt labels in train set: class\n",
      "0    0.940886\n",
      "1    0.059114\n",
      "Name: proportion, dtype: float64\n",
      "hate speech gt labels in test set: class\n",
      "0    0.94998\n",
      "1    0.05002\n",
      "Name: proportion, dtype: float64\n"
     ]
    }
   ],
   "source": [
    "# load eooh annotations for dataset\n",
    "pd.set_option('display.max_colwidth', 30)\n",
    "df_hs = pd.read_csv('data/hs_davidson2017.csv')\n",
    "df_hs['class'] = df_hs['class'].apply(lambda x: 1 if x == 0 else 0) # convert to 0/1 labels\n",
    "# get value counts for each class\n",
    "print(f'hate speech gt labels in original data: {df_hs[\"class\"].value_counts(normalize=True)}')\n",
    "print(f'hate speech gt labels in train set: {df_hs[df_hs[\"split\"] == \"train\"][\"class\"].value_counts(normalize=True)}')\n",
    "print(f'hate speech gt labels in test set: {df_hs[df_hs[\"split\"] == \"test\"][\"class\"].value_counts(normalize=True)}')\n",
    "\n",
    "df_hs_eooh = pd.read_csv('data/hs_davidson2017_eooh_analyzed.csv')\n",
    "# df_hs_eooh.head()\n",
    "# print(df_hs_eooh.head())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAk4AAAPeCAYAAAD6SJUCAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8pXeV/AAAACXBIWXMAAA9hAAAPYQGoP6dpAAEAAElEQVR4nOzdd3xTVf8H8E+SJt17QimjLW2BFlp2oYAgiDJUQAFZoj6AIuLP8SCijwwVUEER4RFEBFniYAgK+Iiy924LpVAKnXTTvZP7+yNtILKSNjej/bxfL16am3vv+d6TNPnmnHPPkQiCIICIiIiIHkpq6gCIiIiILAUTJyIiIiIdMXEiIiIi0hETJyIiIiIdMXEiIiIi0hETJyIiIiIdMXEiIiIi0hETJyIiIiIdMXEiIiIi0hETJyIT+eqrrxAcHGyUssaPH4/x48drHp84cQLBwcHYs2ePUcqfOXMm+vXrZ5Sy6qqkpATvvfceevbsieDgYHz88cemDgkAEBwcjK+++kqvY2pf3xMnTogUFVHjZWXqAIgagq1bt+Ldd9/VPFYoFHB2dkZwcDD69OmD4cOHw8HBod7lZGZm4qeffkL//v3Rpk2bep/PkMw5Nl2sXLkS27Ztw9SpU+Hn54eAgIC79vnqq6+wbNmyh56ra9euWL9+vRhh1tnOnTuRm5uLiRMnGvS8JSUlWL16Nf73v/8hNTUV1tbW8PHxQZcuXTBp0iR4e3sbtDwiU2PiRGRA06dPR7NmzVBdXY2cnBycPHkS8+fPx9q1a/Hf//4XISEhmn1feeUVTJ48Wa/zZ2VlYdmyZfD19dUrOVm9erVe5dTFg2L78MMPYe7LYh4/fhwdOnTAtGnT7rvPgAED0Lx5c83j0tJSzJkzBwMGDMCAAQM02z08PAwWV3R0NGQymV7HdOnSBdHR0ZDL5Zptv/32G65evWrQxKmqqgrjxo1DYmIinn76aYwbNw6lpaW4evUqfvvtNwwYMICJEzU4TJyIDKh3794ICwvTPJ4yZQqOHTuGl19+GVOnTsWuXbtgY2MDALCysoKVlbh/gmVlZbC1tYVCoRC1nIe58wvcXOXm5iIwMPCB+4SEhGglv3l5eZgzZw6Cg4Px1FNPiRKXtbW13sdIpdI6HaevvXv34tKlS1i0aBGGDh2q9VxFRQWqqqpEj6FWaWkp7OzsjFYeNV4c40QkssjISEydOhVpaWnYsWOHZvu9xjgdOXIEzz33HDp37oyIiAgMHDgQn3/+OQD1uJVnnnkGAPDuu+8iODgYwcHB2Lp1KwD1OKYhQ4YgNjYWY8eORYcOHTTH/nOMUy2VSoXPP/8cPXv2RHh4OF5++WXcvHlTa59+/fph5syZdx175zkfFtu9xjiVlpZi4cKF6NOnD0JDQzFw4ECsXr36rpap4OBgzJs3D3v37sWQIUMQGhqKwYMH4+DBgw+qdo3c3FzMmjULPXr0QFhYGJ588kls27ZN83zteKDU1FTs379fE3tqaqpO57+XY8eOYcyYMQgPD0fnzp3xyiuv4Nq1a5rnt2zZguDgYPzyyy9ax61YsQLBwcE4cOCA1vX/c4xTZmYmZs2ahaioKISGhqJfv36YPXs2Kisrta6pdozT+PHjsX//fqSlpWmur1+/figpKUF4eDg++uiju64hIyMDbdq0wcqVK+97nSkpKQCAjh073vWctbX1Xd3T165dw+uvv47u3bujffv2GDhwIL744gutfS5duoR//etf6NixIyIiIvD888/j/PnzWvts3boVwcHBOHnyJObMmYPIyEj06dNH8/yBAwc09R8REYHJkyfj6tWrWufIzs7Gu+++i969eyM0NBRRUVF45ZVX6vW6U+PAFiciI3jqqafw+eef4/Dhwxg5cuQ997l69SqmTJmC4OBgTJ8+HQqFAklJSTh79iwAICAgANOnT8fSpUsxatQodOrUCYD2l1Z+fj4mTZqEwYMH48knn4S7u/sD4/r6668hkUgwadIk5Obm4vvvv8fEiRPx66+/alrGdKFLbHcSBAGvvPKKJuFq06YNDh06hE8//VSTFNzpzJkz+N///ocxY8bA3t4e69evx/Tp07Fv3z64urreN67y8nKMHz8eycnJGDt2LJo1a4Y9e/Zg5syZKCwsxPPPP4+AgAB8+umnWLBgAXx8fPDCCy8AANzc3HS+/jsdPXoUkyZNQrNmzTBt2jSUl5djw4YNeO6557B161Y0a9YMI0aMwJ9//omFCxeiZ8+eaNKkCeLj47Fs2TI888wzWknAP2VmZuKZZ55BUVERRo4cCX9/f2RmZuKPP/5AeXn5PVsXX375ZRQVFSEjI0MzFs/e3h729vbo378/du/ejXfffVerS/C3336DIAh3tSTdqWnTpgCA7du3Y+rUqZBIJPfd9/Llyxg7diysrKwwatQo+Pr6Ijk5GX///TfeeOMNAOq/gbFjx8Le3h7/+te/YGVlhR9//BHjx4/Hhg0b0KFDB61zzp07F25ubnj11VdRWlqqiWXmzJmIiorC22+/jbKyMvzwww8YM2YMtm3bhmbNmgEAXnvtNSQkJGDcuHHw9fVFXl4ejhw5gps3b2r2IbongYjqbcuWLUJQUJAQHR193306deokPP3005rHS5cuFYKCgjSP16xZIwQFBQm5ubn3PUd0dLQQFBQkbNmy5a7nxo0bJwQFBQk//PDDPZ8bN26c5vHx48eFoKAgoVevXkJRUZFm+65du4SgoCDh+++/12zr27ev8M477zz0nA+K7Z133hH69u2refznn38KQUFBwn//+1+t/V577TUhODhYSEpK0mwLCgoS2rVrp7UtLi5OCAoKEtavX39XWXdau3atEBQUJPz666+abZWVlcKoUaOE8PBwrWvv27evMHny5Aee759yc3OFoKAgYenSpZptTz31lBAZGSncunVLK96QkBBhxowZmm1ZWVlC165dhRdeeEGoqKgQnn76aeGRRx7Riqn2+u88/4wZM4SQkJB7vtdUKpUgCLdf3+PHj2uemzx5stZrUOvQoUNCUFCQcODAAa3tQ4cO1Xp976WsrEwYOHCgEBQUJPTt21eYOXOm8PPPPws5OTl37Tt27FghIiJCSEtLu2fMgiAIU6dOFdq1ayckJydrtmVmZgoRERHC2LFjNdtq/96ee+45obq6WrO9uLhY6Ny5s/D+++9rlZGdnS106tRJs72goEAICgoSvv322wdeH9G9sKuOyEjs7OxQUlJy3+ednJwAAH/99RdUKlWdylAoFBg+fLjO+z/99NNa3SmPP/44PD09tbqKxHDw4EHIZLK7ug9ffPFFCIJwVzdcjx49tAZlh4SEwMHBQdNV9KByPD09MWTIEM02uVyO8ePHo7S0FKdOnTLA1dyWlZWFuLg4DBs2DC4uLlrx9ujRQ6tePT098cEHH+DIkSMYO3Ys4uLiMH/+/AfefalSqbB371707dtXayxdrQe1+NxPjx494OXlhZ07d2q2XblyBfHx8XjyyScfeKyNjQ1+/vlnvPTSSwDUXWjvvfceoqKi8OGHH2q6DvPy8nDq1CmMGDFC00r1z5iVSiWOHDmC/v37w8/PT/O8l5cXhgwZgjNnzqC4uFjr2JEjR2q1kh09ehSFhYUYPHgw8vLyNP+kUik6dOig6bq0sbGBXC7HyZMnUVBQoG+VUSPHxInISEpLS2Fvb3/f5wcNGoSOHTvi/fffR48ePfDGG29g165deiVR3t7eeg0Eb9GihdZjiUSCFi1aIC0tTedz1EVaWhq8vLzuShJqpwD4Z/lNmjS56xzOzs4oLCx8aDktWrSAVKr9UVdbTnp6ut6xP0jt+Vq1anXXcwEBAbh165amSwkABg8ejEceeQTR0dEYOXIkIiMjH3j+vLw8FBcXo3Xr1gaLWSqVYujQodi7dy/KysoAqKcusLa2xuOPP/7Q4x0dHTFjxgz8/fff+Pvvv/Hxxx+jVatW2LBhA5YvXw7g9liooKCg+54nLy8PZWVl9607lUp11/i7f3ap3bhxAwDw/PPPIzIyUuvf4cOHkZubC0D9A+Ptt9/GwYMH0bNnT4wdOxarVq1Cdnb2Q6+XiGOciIwgIyMDRUVFWq0m/2RjY4ONGzfixIkT2L9/Pw4dOoRdu3bhxx9/xHfffafTLen6jEuqL6VSqfdt8nV1v3IEM5/i4GFu3bqF2NhYAEBCQgJUKtVdSZ4xPP3001i9erVmAP5vv/2GRx55BI6Ojnqdx9fXF8888wwGDBiA/v37Y+fOnZrxS2L4552Dte+HTz/9FJ6ennftf+f7aOLEiejXrx/27t2Lw4cP48svv8Q333yD77//Hm3bthUtZrJ8bHEiMoJff/0VABAVFfXA/aRSKSIjI/Huu+9i165deOONN3D8+HFNF0NdumIeJCkpSeuxIAhISkqCr6+vZtv9Wnb+2VqjT2y+vr7Iysq6q+slMTFR87wh+Pr6Iikp6a5Wu9py/tltVF+157t+/fpdzyUmJsLV1VXrlvl58+ahpKQEb731Fs6cOYPvv//+ged3c3ODg4PDXXeI6eJBr09QUBDatm2LnTt34vTp00hPT6/X9ArOzs7w8/PTtODUdr1duXLlvse4ubnB1tb2vnUnlUrv2fJ4p9py3N3d0aNHj7v+devWTWv/5s2b48UXX8R3332H3377DVVVVfjuu+/0ulZqfJg4EYns2LFj+O9//4tmzZo9cMxIfn7+XdtqJ5KsHStia2sLAA/totLV9u3btZKXPXv2IDs7G71799Zs8/Pzw4ULFzQxAMC+ffvu6jbRJ7bevXtDqVRi48aNWtvXrl0LiUSiVX599O7dG9nZ2di1a5dmW3V1NdavXw87Ozt06dLFIOXU8vLyQps2bbB9+3aterhy5QqOHDmidbfcnj17sGvXLrz11luYPHkyBg8ejCVLltwzcagllUrRv39/7Nu3DzExMXc9/6AWOFtbWxQVFd33+aeeegpHjhzB999/DxcXF51eg8uXLyMvL++u7Wlpabh27Zqm283NzQ1dunTBli1b7kq4a2OWyWTo2bMn/vrrL60pAXJycvDbb7+hU6dOD519v1evXnBwcMDKlSvvOYdUbaxlZWWoqKjQeq558+awt7fXep8T3Qu76ogM6ODBg0hMTIRSqUROTg5OnDiBI0eOoGnTpvj6668fOCnh8uXLcfr0afTp0we+vr7Izc3Fpk2b4OPjo7m9v3nz5nBycsLmzZthb28POzs7tG/fXmswrT6cnZ0xZswYDB8+XDMdQYsWLbSmTHj22Wfxxx9/4F//+heeeOIJJCcnY+fOnXd1O+oTW79+/dCtWzd88cUXmrmFjhw5gr/++gvPP//8A7s09TFq1Cj8+OOPmDlzJi5evAhfX1/88ccfOHv2LGbNmmWQZXD+acaMGZg0aRJGjRqFZ555RjMdgaOjo2ZW8tzcXMyZMwfdunXDuHHjAAD/+c9/cOLECbz77rvYtGnTfbvs3nzzTRw5cgTjx4/HyJEjERAQgOzsbOzZswebNm3S3GTwT+3atcOuXbuwYMEChIWFwc7OTmturSFDhuCzzz7Dn3/+ieeee06nSUuPHDmCr776Cv369UOHDh1gZ2eH1NRUbNmyBZWVlXjttdc0+77//vt47rnnMGzYMIwaNQrNmjVDWloa9u/fr2mR/b//+z8cPXoUY8aMwZgxYyCTyfDjjz+isrIS//73vx8aj4ODA+bMmYMZM2Zg+PDhGDRoENzc3JCeno4DBw6gY8eO+OCDD3Djxg1MnDgRjz/+OAIDAyGTybB3717k5ORg8ODBDy2HGjcmTkQGtHTpUgDqO7dcXFwQFBSEWbNm6bRWXb9+/ZCWloYtW7bg1q1bcHV1RdeuXfHaa69pxprI5XIsXLgQn3/+OebMmYPq6mosWLCgzonTyy+/jPj4eHzzzTcoKSlBZGQkZs+erWk9AtS/4mfOnIk1a9Zg/vz5CA0NxYoVK/DJJ59onUuf2KRSKb7++mssXboUu3btwtatW+Hr64sZM2bgxRdfrNO13IuNjQ3Wr1+PRYsWYdu2bSguLkarVq2wYMECve4+1EePHj3w7bffYunSpVi6dCmsrKzQpUsX/Pvf/9bUxZw5c1BZWYkFCxZoutBcXV0xb948TJ06FatXr8akSZPueX5vb2/89NNP+PLLL7Fz504UFxfD29sbvXv3fuAYtzFjxiAuLg5bt27F2rVr4evrq5U4eXh4oGfPnjhw4IDO3XSPPfYYSkpKcOTIERw/fhwFBQVwcnJC+/bt8cILL6B79+6afUNCQjRx//DDD6ioqEDTpk3xxBNPaPZp3bo1Nm7ciMWLF2PlypUQBAHt27fHZ599dtccTvczdOhQeHl54ZtvvsHq1atRWVkJb29vdO7cWfOa+/j4YPDgwTh27Bh27NgBmUwGf39/LFmyBAMHDtSpHGq8JIKlj64kIiKDePXVV3HlyhX8+eefpg6FyGxxjBMRESErK0uv1iaixopddUREjVhKSgrOnj2LX375RbMcChHdH1uciIgasVOnTmHGjBlITU3FwoUL7zn/ERHdxjFORERERDpiixMRERGRjpg4EREREemIg8MNRKVSobq6GlKp1ODLYhAREZF4BEGASqWClZXVQ9eLZOJkINXV1fdcAoGIiIgsQ1hYGBQKxQP3YeJkILUZalhYmMFXjFcqlYiJiRHl3HQb69k4WM/Gw7o2DtazcYhZz7XnflhrE8DEyWBqu+dkMplofzhinptuYz0bB+vZeFjXxsF6Ng4x61mXoTYcHE5ERESkIyZORERERDpi4kRERESkIyZORERERDpi4kRERESkIyZORERERDpi4kRERESkIyZORERERDoy68Rp06ZNGDp0KDp27IiOHTti1KhROHDggOb58ePHIzg4WOvfBx98oHWO9PR0TJ48GR06dEBkZCQ++eQTVFdXa+1z4sQJDBs2DKGhoRgwYAC2bt1qlOsjIiIiy2LWM4f7+Pjg7bffRosWLSAIArZv345XX30V27ZtQ+vWrQEAI0eOxPTp0zXH2Nraav5fqVRiypQp8PDwwObNm5GVlYV33nkHcrkcb775JgAgJSUFU6ZMwejRo7Fo0SIcO3YM77//Pjw9PdGrVy/jXjARERGZNbNOnPr166f1+I033sAPP/yA8+fPaxInGxsbeHp63vP4w4cPIyEhAWvWrIGHhwfatGmD119/HYsWLcK0adOgUCiwefNmNGvWDDNnzgQABAQE4MyZM1i7di0TJyIiItJi1l11d1Iqlfj9999RWlqKiIgIzfadO3eiW7duGDJkCBYvXoyysjLNc+fPn0dQUBA8PDw026KiolBcXIyEhATNPpGRkVplRUVF4fz58+JeEBEREVkcs25xAoD4+HiMHj0aFRUVsLOzw/LlyxEYGAgAGDJkCJo2bQovLy/Ex8dj0aJFuH79OpYtWwYAyMnJ0UqaAGgeZ2dnP3Cf4uJilJeXw8bGRq94lUplna5Tl3OKcW66jfVsHKxn42FdGwfr2TjErGd9zmn2iVOrVq2wfft2FBUV4Y8//sA777yDDRs2IDAwEKNGjdLsFxwcDE9PT0ycOBHJyclo3ry5SeKNiYmxyHPTbaxn42A9Gw/r2jhYz8Zh6no2+8RJoVCgRYsWAIDQ0FDExMRg3bp1mDdv3l37dujQAQCQlJSE5s2bw8PDA9HR0Vr75OTkAIBmXJSHh4dm2537ODg46N3aBABhYWGQyWR6H/cgSqUSMTExopybbmM9Gwfr2XhY18bBejYOMeu59ty6MPvE6Z9UKhUqKyvv+VxcXByA20lReHg4VqxYgdzcXLi7uwMAjh49CgcHB013X3h4OA4ePKh1nqNHjyI8PLxO8clkMtH+cMQ8N93GejYO1rPxsK6Ng/VsHKauZ7MeHL548WKcOnUKqampiI+Px+LFi3Hy5EkMHToUycnJWL58OWJjY5Gamoq//voL77zzDrp06YKQkBAA6kHegYGBmDFjBi5fvoxDhw5hyZIlGDt2LBQKBQBg9OjRSElJwaeffopr165h48aN2L17NyZOnGjCKyei+1EJgkmPJ6LGzaxbnHJzc/HOO+8gKysLjo6OCA4OxurVq9GzZ0/cvHkTx44dw7p161BaWoomTZrgsccew9SpUzXHy2QyrFixAnPmzMGoUaNga2uLYcOGac375Ofnh5UrV2LBggVYt24dfHx88NFHH3EqAiIzJZVIsDvmJvJK7t3y/CBu9go8EdZEhKiIqLEw68Rp/vz5932uSZMm2LBhw0PP4evri1WrVj1wn27dumH79u36hkdEJpJXUomsogpTh0FEjZBZd9URERERmRMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6YuJEREREpCMmTkREREQ6MuvEadOmTRg6dCg6duyIjh07YtSoUThw4IDm+YqKCsydOxfdunVDREQEXnvtNeTk5GidIz09HZMnT0aHDh0QGRmJTz75BNXV1Vr7nDhxAsOGDUNoaCgGDBiArVu3GuX6iIiIyLKYdeLk4+ODt99+G1u3bsWWLVvQvXt3vPrqq7h69SoAYP78+di3bx+WLFmC9evXIysrC9OmTdMcr1QqMWXKFFRVVWHz5s1YuHAhtm3bhqVLl2r2SUlJwZQpU9CtWzf8+uuveP755/H+++/j0KFDRr9eIiIiMm9mnTj169cPffr0QcuWLdGqVSu88cYbsLOzw/nz51FUVIQtW7Zg5syZiIyMRGhoKObPn49z587h/PnzAIDDhw8jISEBn332Gdq0aYM+ffrg9ddfx8aNG1FZWQkA2Lx5M5o1a4aZM2ciICAA48aNw8CBA7F27VrTXTgRERGZJbNOnO6kVCrx+++/o7S0FBEREYiNjUVVVRV69Oih2ScgIABNmzbVJE7nz59HUFAQPDw8NPtERUWhuLgYCQkJmn0iIyO1yoqKitKcg4iIiKiWlakDeJj4+HiMHj0aFRUVsLOzw/LlyxEYGIi4uDjI5XI4OTlp7e/u7o7s7GwAQE5OjlbSBEDz+GH7FBcXo7y8HDY2NnrFq1Qq9dpfn3OKcW66jfVsHPWtZ5lMBkFQQRBUeh9be0xjeY35njYO1rNxiFnP+pzT7BOnVq1aYfv27SgqKsIff/yBd955Bxs2bDB1WPcVExNjkeem21jPxlGXera1tUXbtm2Rm5uLjLxivY+XVToAaIX4+HiUlZXpfbyl4nvaOFjPxmHqejb7xEmhUKBFixYAgNDQUMTExGDdunV44oknUFVVhcLCQq1Wp9zcXHh6egJQtxxFR0drna/2rrs79/nnnXg5OTlwcHDQu7UJAMLCwiCTyfQ+7kGUSiViYmJEOTfdxno2DkPUs7u7O5QKB/2Pc7QGAAQHB9epXEvD97RxsJ6NQ8x6rj23Lsw+cfonlUqFyspKhIaGQi6X49ixYxg4cCAAIDExEenp6QgPDwcAhIeHY8WKFcjNzYW7uzsA4OjRo3BwcEBgYKBmn4MHD2qVcfToUc059CWTyUT7wxHz3HQb69k46lPPEokUEon+QzRrj2lsry/f08bBejYOU9ezWQ8OX7x4MU6dOoXU1FTEx8dj8eLFOHnyJIYOHQpHR0eMGDECCxcuxPHjxxEbG4tZs2YhIiJCk/RERUUhMDAQM2bMwOXLl3Ho0CEsWbIEY8eOhUKhAACMHj0aKSkp+PTTT3Ht2jVs3LgRu3fvxsSJE0134URERGSWzLrFKTc3F++88w6ysrLg6OiI4OBgrF69Gj179gQAzJo1C1KpFNOnT0dlZSWioqIwe/ZszfEymQwrVqzAnDlzMGrUKNja2mLYsGGYPn26Zh8/Pz+sXLkSCxYswLp16+Dj44OPPvoIvXr1Mvr1EhERkXkz68Rp/vz5D3ze2toas2fP1kqW/snX1xerVq164Hm6deuG7du31yVEIiIiakTMuquOiIiIyJwwcSIiIiLSERMnIiIiIh0xcSIiIiLSERMnIiIiIh0xcSIiIiLSERMnIiIiIh0xcSIiIiLSERMnIiIiIh0xcSIiIiLSERMnIiIiIh0xcSIiIiLSERMnIiIiIh0xcSIiIiLSERMnIiIiIh0xcSIiIiLSERMnIiIiIh0xcSIiIiLSERMnIiIiIh0xcSIiIiLSERMnIiIiIh0xcSIiIiLSERMnIiIiIh0xcSIiIiLSERMnIiIiIh0xcSIiIiLSERMnIiIiIh0xcSIiIiLSERMnIiIiIh0xcSIiIiLSERMnIiIiIh0xcSIiIiLSERMnIiIiIh0xcSIiIiLSERMnIiIiIh0xcSIiIiLSERMnIiIiIh0xcSIiIiLSERMnIiIiIh0xcSIiIiLSERMnIiIiIh0xcSIyEJUgmORYIiIyHitTB0DUUEglEuyOuYm8kkq9jnOzV+CJsCYiRUVERIbExInIgPJKKpFVVGHqMIiISCRm3VW3cuVKjBgxAhEREYiMjMTUqVORmJiotc/48eMRHBys9e+DDz7Q2ic9PR2TJ09Ghw4dEBkZiU8++QTV1dVa+5w4cQLDhg1DaGgoBgwYgK1bt4p+fURERGRZzLrF6eTJkxg7dizCwsKgVCrx+eef46WXXsLvv/8OOzs7zX4jR47E9OnTNY9tbW01/69UKjFlyhR4eHhg8+bNyMrKwjvvvAO5XI4333wTAJCSkoIpU6Zg9OjRWLRoEY4dO4b3338fnp6e6NWrl/EumIiIiMyaWSdOq1ev1nq8cOFCREZG4uLFi+jSpYtmu42NDTw9Pe95jsOHDyMhIQFr1qyBh4cH2rRpg9dffx2LFi3CtGnToFAosHnzZjRr1gwzZ84EAAQEBODMmTNYu3YtEyciIiLSMOvE6Z+KiooAAM7Ozlrbd+7ciR07dsDT0xN9+/bF1KlTNa1O58+fR1BQEDw8PDT7R0VFYc6cOUhISEDbtm1x/vx5REZGap0zKioK8+fP1ztGpVKp9zG6nlOMc9Nt9a1nmUwGQVBBEFR6HVe7f2N5fU1VzwDrmsTBejYOMetZn3NaTOKkUqkwf/58dOzYEUFBQZrtQ4YMQdOmTeHl5YX4+HgsWrQI169fx7JlywAAOTk5WkkTAM3j7OzsB+5TXFyM8vJy2NjY6BxnTExMna7P1Oem2+pSz7a2tmjbti1yc3ORkVes17GySgcArRAfH4+ysjK9y7ZUxq5ngHVN4mI9G4ep69liEqe5c+fi6tWr2LRpk9b2UaNGaf4/ODgYnp6emDhxIpKTk9G8eXNjh4mwsDDIZDKDnlOpVCImJkaUc9Nthqhnd3d3KBUO+h3jaA1A/f5tDExVzwDrmsTBejYOMeu59ty6sIjEad68edi/fz82bNgAHx+fB+7boUMHAEBSUhKaN28ODw8PREdHa+2Tk5MDAJpxUR4eHpptd+7j4OCgV2sToO5GEOsPR8xz0231qWeJRAqJRL+bVWv3b2yvrbHrufa42rIbE352GAfr2ThMXc9mPR2BIAiYN28e/vzzT3z//ffw8/N76DFxcXEAbidF4eHhuHLlCnJzczX7HD16FA4ODggMDNTsc/z4ca3zHD16FOHh4Qa6EiIiImoIzDpxmjt3Lnbs2IHFixfD3t4e2dnZyM7ORnl5OQAgOTkZy5cvR2xsLFJTU/HXX3/hnXfeQZcuXRASEgJAPcg7MDAQM2bMwOXLl3Ho0CEsWbIEY8eOhUKhAACMHj0aKSkp+PTTT3Ht2jVs3LgRu3fvxsSJE0116URERGSGzLqr7ocffgCgnuTyTgsWLMDw4cMhl8tx7NgxrFu3DqWlpWjSpAkee+wxTJ06VbOvTCbDihUrMGfOHIwaNQq2trYYNmyY1rxPfn5+WLlyJRYsWIB169bBx8cHH330EaciICIiIi1mnTjFx8c/8PkmTZpgw4YNDz2Pr68vVq1a9cB9unXrhu3bt+sTHhERETUyZt1VR0RERGROmDgRERER6YiJExEREZGOmDgRERER6YiJExEREZGOmDgRERER6YiJExEREZGORJnH6ebNm5BIJJp15aKjo7Fz504EBgZqLcpLREREZElEaXF66623NGu/ZWdn44UXXkBMTAy++OILLFu2TIwiiYiIiEQnSuJ09epVtG/fHgCwe/dutG7dGps3b8aiRYuwbds2MYokIiIiEp0oiVN1dbVmAd2jR4+iX79+AAB/f39kZ2eLUSQRERGR6ERJnAIDA7F582acPn0aR48eRe/evQEAWVlZcHFxEaNIIiIiItGJkji9/fbb+PHHHzF+/HgMHjwYISEhAIC///5b04VHREREZGlEuauuW7duOH78OIqLi+Hs7KzZPnLkSNjZ2YlRZIMnl8tNHQIREVGjJ0qL04QJE1BSUqKVNAGAi4sL/u///k+MIhu8tu3aQSaT1elYlSAYOBoiIqLGSZQWp5MnT6Kqququ7RUVFThz5owYRTZ4VjIZdkWn41bp3fX6IG72CjwR1kSkqIiIiBoXgyZOly9f1vx/QkKC1h10KpUKhw4dgre3tyGLbFTySiqQXaxf4kRERESGY9DE6emnn4ZEIoFEIsHzzz9/1/M2NjZ4//33DVkkERERkdEYNHH666+/IAgC+vfvj59//hlubm6a5+RyOdzd3es8ToeIiIjI1AyaOPn6+gLQ7rIjIiIiaihEuasOALZv347Ro0cjKioKaWlpAIC1a9di7969YhVJREREJCpREqdNmzZh4cKF6NOnD4qKiqBSqQAATk5O+P7778UokoiIiEh0oiROGzZswEcffYRXXnkFUuntIkJDQ3HlyhUxiiQiIiISnSiJU2pqKtq0aXPXdoVCgbKyMjGKJCIiIhKdKIlTs2bNEBcXd9f2Q4cOISAgQIwiiYiIiEQnyszhL7zwAubNm4fKykoAQHR0NH777Td88803+Oijj8QokoiIiEh0oiROzz77LKytrbFkyRKUlZXhrbfegpeXF2bNmoXBgweLUSQRERGR6ERJnADgySefxJNPPomysjKUlpbC3d1drKKIiIiIjEK0eZyqq6tx9OhR/Prrr7C2tgYAZGZmoqSkRKwiiYiIiEQlSotTWloa/vWvf+HmzZuorKxEz5494eDggFWrVqGyshLz5s0To1giIiIiUYnS4vTxxx8jNDQUJ0+e1LQ2AcCAAQNw/PhxMYokIiIiEp0oLU5nzpzBDz/8AIVCobXd19cXmZmZYhRJREREJDpRWpxUKpVmmZU7ZWRkwN7eXowiiYiIiEQnSuLUs2fPu9akKykpwVdffYU+ffqIUSQRERGR6ERJnGbOnImzZ89i0KBBqKysxNtvv41+/fohMzMTb7/9thhFEhEREYlOlDFOPj4++PXXX/H7778jPj4epaWleOaZZzB06FDY2NiIUSQRERGR6ESbANPKygpPPfWUWKcnIiIiMjpREqdHHnkEXbt2RZcuXdC9e3f4+fmJUQwRERHVkUoQIJVIjH6spRMlcXrjjTdw+vRpfPvtt/jPf/4Db29vdOnSBV27dkXXrl3RsmVLMYolIiIiHUklEuyOuYm8kkq9jnOzV+CJsCYiRWX+REmcnnrqKU03XVZWFk6dOoV9+/Zh7ty5UKlUiIuLE6NYIiIi0kNeSSWyiipMHYZFEW2MU1lZGc6cOYOTJ0/ixIkTuHTpElq3bo2uXbuKVSQRERGRqERJnEaPHo1Lly4hICAAXbt2xaRJk9ClSxc4OzuLURwRERGRUYgyj1NiYiJsbW3h7+8Pf39/BAQEMGkiIiIiiydKi9OJEycQHx+PkydP4vDhw/jiiy8gl8vRtWtXdOvWDSNHjhSjWCIiIiJRidLiJJFIEBISggkTJmDp0qVYtWoVevbsiT179mD27Nk6n2flypUYMWIEIiIiEBkZialTpyIxMVFrn4qKCsydOxfdunVDREQEXnvtNeTk5Gjtk56ejsmTJ6NDhw6IjIzEJ598gurqaq19Tpw4gWHDhiE0NBQDBgzA1q1b614BRERE1CAZNHFatmwZysrKcPHiRaxZswYvv/wyunXrhtGjRyM+Ph7jxo3DV199pfP5Tp48ibFjx+Knn37CmjVrUF1djZdeegmlpaWafebPn499+/ZhyZIlWL9+PbKysjBt2jTN80qlElOmTEFVVRU2b96MhQsXYtu2bVi6dKlmn5SUFEyZMgXdunXDr7/+iueffx7vv/8+Dh06ZJiKISIiogbBoF11y5cvx3PPPYdnn30Wbdq0QdeuXTFy5Eh06dIFjo6Oep9v9erVWo8XLlyIyMhIXLx4EV26dEFRURG2bNmCRYsWITIyEoA6kRo0aBDOnz+P8PBwHD58GAkJCVizZg08PDzQpk0bvP7661i0aBGmTZsGhUKBzZs3o1mzZpg5cyYAICAgAGfOnMHatWvRq1ev+lcMERERNQgGTZwEQQCgbilycHAw5KkBAEVFRQCgGWgeGxuLqqoq9OjRQ7NPQEAAmjZtqkmczp8/j6CgIHh4eGj2iYqKwpw5c5CQkIC2bdvi/PnzmsTrzn3mz5+vd4xKpbIul/ZAKpUKMpkMKgEQBJVex9buL0ZcDU1tHdW1rmQyGQRBxdfoIUxVzwDrmsRhqfVsaZ9ZYtazPuc0+OBwiUQiStKkUqkwf/58dOzYEUFBQQCAnJwcyOVyODk5ae3r7u6O7OxszT53Jk0ANI8ftk9xcTHKy8v1Wpg4JiZGvwvTga2tLdq2bYtbebnIyCvW61hZpQOAVoiPj0dZWZnBY2uI6vIa1r5Gubl8jXRl7HoGWNckLkuqZ0v+zDJ1PRs8cRo4cCAkD1m/5uTJk3qfd+7cubh69So2bdpU19CMIiwsDDKZzKDnVKnU2b2rmzuUCv2SUndHawBAcHCwQWNqiJRKJWJiYur1Grq78zV6GFPVM8C6JnFYcj1b0meWmPVce25dGDxxeu211+o0nulB5s2bh/3792PDhg3w8fHRbPfw8EBVVRUKCwu1Wp1yc3Ph6emp2Sc6OlrrfLV33d25zz/vxMvJyYGDg4NerU2AuulTrD8cqQSQSPQbz1+7v6X9MZtSfV5DiUTK10hHxq7n2uNqy25MxPxcotsssZ4t8TPL1PVs8MRp8ODBcHd3N8i5BEHAhx9+iD///BPr16+Hn5+f1vOhoaGQy+U4duwYBg4cCEA9+WZ6ejrCw8MBAOHh4VixYgVyc3M1cR09ehQODg4IDAzU7HPw4EGtcx89elRzDiIiIiLAwNMRPKyLTl9z587Fjh07sHjxYtjb2yM7OxvZ2dkoLy8HADg6OmLEiBFYuHAhjh8/jtjYWMyaNQsRERGapCcqKgqBgYGYMWMGLl++jEOHDmHJkiUYO3YsFAoFAPUSMSkpKfj0009x7do1bNy4Ebt378bEiRMNej1ERERk2US5q85QfvjhBwDA+PHjtbYvWLAAw4cPBwDMmjULUqkU06dPR2VlJaKiorQm2ZTJZFixYgXmzJmDUaNGwdbWFsOGDcP06dM1+/j5+WHlypVYsGAB1q1bBx8fH3z00UecioCIiIi0GDRxunz5siFPh/j4+IfuY21tjdmzZz9wRnJfX1+sWrXqgefp1q0btm/frm+IRERE1IiIsuQKERERUUPExImIiIhIR0yciIiIiHRksMRp2LBhKCgoAHB7sV8iIiKihsRgidO1a9c0ydLy5ctRWlpqqFMTERERmQWD3VXXpk0bvPvuu+jUqRMEQcDq1athZ2d3z32nTZtmqGKJiIiIjMZgidOCBQvw1VdfYd++fZBIJDh06NA9p0SXSCRMnIiIiMgiGSxx8vf3xxdffAEACAkJwdq1aw229AoRkSGoBAFVShXkMt4XQ0R1Y/C16gDDT4RJRFRfiTnFWHUoC/N+u4QW7vbo1doD7z7RBgorJlFEpDtREicASE5Oxvfff49r164BAAIDAzFhwgQ0b95crCKJiO7pfEo+Dl7JRu2iUAlZxUjIKkZucSWWjAqHVGrYdTaJqOES5afWoUOHMGjQIERHRyM4OBjBwcG4cOECBg8ejCNHjohRJBHRPV1ML8CBmqSpUwtXHJrRF4ue7QArqQQ7LqRjwe44U4dIRBZElBanxYsXY+LEiXj77be1ti9atAiLFi1Cz549xSiWiEhLlVKFY9dyAQCdW7ji6fCm8HOzg5+bHaQS4M2fLmDVoevo4OeCIe2bmjhaIrIEorQ4Xbt2Dc8888xd20eMGIGEhAQxiiQiusuFlHyUVCrhZGOF7v7ukEhud8kN79gMUx8JAAB8/r8rqFaqTBUmEVkQURInNzc3xMXd3fwdFxfHO+2IyCgqqpQ4nXQLANDd3x2ye4xjmto3EC52ciTmlODX8+nGDpGILJAoXXXPPvssPvjgA6SkpKBjx44AgLNnz2LVqlWYOHGiGEUSEWk5k3wLFdUquNsrEOzjeM99HKytMKV3AD7ZcxlL/76KJ8ObcqoCInogURKnV199FQ4ODvjuu+/w+eefAwC8vLwwbdo0TJgwQYwiiYg0VCoBsWmFAIBu/m6QSu5/19yEyBb49lAiknJLse1sGkZ28TNWmERkgURJnCQSCSZOnIiJEyeiuLgYAODg4CBGUUREd0nNL0NZlRI2cin8PR782WNvbYXJvf2xYPdlrD16g4kTET2Q6G3SDg4OTJqIyKjiM4oAAIFeDvcc2/RPIzv7QSGT4tLNQlxMLxA7PCKyYOzMJ6IGpVqlQkK2uqU72PveY5v+ydVegf5tvQAAW86kiRYbEVk+Jk5E1KAk5ZaisloFe2sZmrrY6nzcM52aAQB+PZ+GKk5NQET3wcSJiBqUKzXddEFejg8cFP5PvVt7wsPBGrklldgfny1WeERk4QyeOFVVVeH555/HjRs3DH1qIqIHqlaqkJhTAgAIus8UBPdjJZNiWIR69vBfzqQYPDYiahgMnjjJ5XLEx8cb+rRERA+VXlCOapUAe4UM3o7Weh8/oqa7bt/lbBSVVxk6PCJqAETpqnvyySfxyy+/iHFqIqL7SskrBQD4udlpLa+iqxAfJ/h72qNSqcKBK+yuI6K7iTKPk1KpxA8//ICjR48iNDQUtrbaAzTfffddMYolokYu5ZY6cWruZlfncwxo642VBxLxv4uZXPiXiO4iSuJ05coVtG3bFgBw/fp1refq8iuQiOhhyquUyCysAAD4udY9cXqsrQ9WHkjEvstZqKxWQWHFe2iI6DZREqf169eLcVoiovtKvVUGAHC1k8PBpu4fbRF+LvBwsEZOcQVOXM9Fr9aehgqRiBoAUX9KJSUl4dChQygvLwcACIIgZnFE1Igl59W/mw4ApFIJBtRMhvm/i5n1jouIGhZREqdbt27h+eefx8CBAzF58mRkZ6sHWc6aNQsLFy4Uo0giauRqxzf51TNxAtTddQDw56VMqFT8wUdEt4mSOC1YsABWVlbYv38/bGxsNNsHDRqEQ4cOiVEkETViheVVyC+tggRAM1fdZwu/n8gAd9grZMgoLEcs164jojuIkjgdOXIE//73v+Hj46O1vWXLlkhPTxejSCJqxNJrxjd5OVnD2kpW7/PZyGXoGegBADjAWcSJ6A6iJE6lpaVaLU218vPzoVAoxCiSiBqxmwXqcZRNnevf2lSrT7B6UPjBq0yciOg2URKnzp07Y/v27VrbVCoVvv32W3Tr1k2MIomoEcsoVCdOPs53/2Crq941d9OdTc5HIWcRJ6IaokxH8O9//xsTJ05EbGwsqqqq8NlnnyEhIQEFBQX44YcfxCiSiBqpKqUK2cXq+ZuaGDBx8nOzg7+nPRKzS3A0IQePhzaBShD0Wjj4TvU5lojMhyiJU1BQEP744w9s2LAB9vb2KC0txYABAzB27Fh4eXmJUSQRNVJZhRUQBMDeWgYHa8N+pPUJ8kRidgkOXMnG46FNIJVIsDvmJvJKKvU6j5u9Ak+ENTFobERkGqIkTgDg6OiIV155RazTExEBAG4WqAeG+zjZGHxlgj5Bnlhz5AYOxGdDEARIJBLklVQiq6jCoOUQkeUQLXEqKCjAL7/8gmvXrgEAAgMDMXz4cLi4uIhVJBE1QrXjm5oYcGB4re7+7rC2kiK9oBzXsosR6OVo8DKIyLKIMjj81KlT6NevH9avX4/CwkIUFhZi/fr1ePTRR3Hq1CkxiiSiRkgQBM0ddYYcGF7LRi5D11ZuAID9nJaAiCBS4jRv3jwMGjQIf/31F5YtW4Zly5Zh7969GDx4MObNmydGkUTUCBWVV6O0UgmpBPB2tBaljNq7644k5IhyfiKyLKIkTklJSXjhhRcgk92eiE4mk2HixIlISkoSo0giaoRqW5s8HKxhJRNn6c2o1uqJME9cz0NltUqUMojIcojySdO2bVskJibetT0xMREhISFiFElEjdDt8U2G76arFeztCA8HBUorlTifki9aOURkGQw2OPzy5cua/58wYQI+/vhjJCUloUOHDgCACxcuYOPGjXj77bcNVSQRNXJZNYmTt5N4iZNUKkGPAA/suJCOw1ezRS2LiMyfwRKnp59+GhKJBIJweyXxzz777K793nrrLQwaNMhQxRJRIyUIgmbiS0+RxjfVigqsSZwScjCiYzNRyyIi82awxOmvv/4y1KmIiB4qv7QKVUoBVlIJ3OzEXQOzZ804pwupBRjMiSyJGjWDJU6+vr6GOpWWU6dOYfXq1YiNjUV2djaWL1+O/v37a56fOXMmtm3bpnVMVFQUVq9erXmcn5+PDz/8EPv27YNUKsVjjz2G9957D/b29pp9Ll++jHnz5iEmJgZubm4YN24cJk2aJMo10cNxaQt6mNpJKD0crCGVivt6+7rYwt/DHok5JbieUwJ3B3FbuIjIfIk2AWZmZibOnDmDvLw8qFTad6JMmDBB5/OUlpYiODgYI0aMwLRp0+65T69evbBgwQLNY4VC+9fn22+/jezsbKxZswZVVVWYNWsWPvjgAyxevBgAUFxcjJdeegmRkZGYO3curly5glmzZsHJyQmjRo3SOVYyHC5tQQ+TVaQe3+QlcjddrZ6BHkjMKUFCVjETJ6JGTJTEaevWrfjggw8gl8vh6uqq9ZxEItErcerTpw/69OnzwH0UCgU8PT3v+dy1a9dw6NAh/PLLLwgLCwMAvP/++5g8eTJmzJgBb29v7NixA1VVVZg/fz4UCgVat26NuLg4rFmzhomTCXFpC3qQ2veGp5PxEqf1x5NwLbsE3fzdjVImEZkfURKnL7/8Eq+++iqmTJkCqVScuVXudPLkSURGRsLJyQndu3fH//3f/2kStnPnzsHJyUmTNAFAjx49IJVKER0djQEDBuD8+fPo3LmzVktVVFQUVq1ahYKCAjg7O+sci1KpNNyF1VCpVJDJZFAJgCDoN49M7f5ixCUmmUwGQVAZ9Xprj6lrXZkiZktkiHpWqZTIrk2cHBQ613l96rprSxdIJUB2cQUKyyrhaKP7x6epXuP61jXpxlLr2dI+s8SsZ33OKUriVF5ejsGDBxslaerVqxcGDBiAZs2aISUlBZ9//jkmTZqEH3/8ETKZDDk5OXBzc9M6xsrKCs7OzsjOVi+hkJOTg2bNtO+U8fDw0DynT+IUExNTzyu6m62tLdq2bYtbebnIyCvW61hZpQOAVoiPj0dZWZnBYxND7fXm5prmeuvyGpo6ZktUn3q+np6NimoVpBKgqjgPGSW6jXGqT13b2toirJkLLqTk4+KNm2jpItf5WFO/xmJ8LtHdLKmeLfkzy9T1LEriNGLECOzZsweTJ08W4/RaBg8erPn/4OBgBAcHo3///ppWKGMLCwvTmjHdEGrHiLm6uUOpcNDrWPea8R/BwcEGjckY3N2Ne71KpRIxMTH1eg2NHbMlMkQ9l0nUcyl5OFjDt4nuY9rqW9e9UhJwISUfBUoFfHy8jVZuXRmirunhLLmeLekzS8x6rj23LkRJnN566y1MmTIFhw4dQlBQEKystIt59913xSgWAODn5wdXV1ckJSUhMjISHh4eyMvL09qnuroaBQUFmnFRHh4eyMnRXoeq9nFty5OuZDKZaH84UgkgkejXile7v6X9MQPq2E1xvfV5DU0VsyWqTz2nF6i76bwcrfWq7/rWdc9ADyzbl4CUW2UAJJDoePemqV9jMT+X6DZLrGdL/MwydT2LkjitXLkShw8fRqtWre56TtcPmrrKyMhAfn6+JimKiIhAYWEhYmNjERoaCgA4fvw4VCoV2rdvDwAIDw/HkiVLUFVVBblc3fx+9OhRtGrVSq9uOiIyjvR8dfeA2BNf/lPHFi6QyyQorVQit6QSHry7jqjRESVxWrNmDebPn4/hw4fX+1wlJSVITk7WPE5NTUVcXBycnZ3h7OyMZcuWYeDAgfDw8EBKSgo+++wztGjRAr169QIABAQEoFevXvjPf/6DuXPnoqqqCh9++CEGDx4Mb291U/vQoUOxfPlyvPfee5g0aRKuXr2KdevWidoyRkR1V7u4r5ejcZc/sbaSoaW7Pa5mFSMlr5SJE1EjJEripFAo0LFjR4OcKzY2Vmv6gtr5moYNG4Y5c+bgypUr2L59O4qKiuDl5YWePXvi9ddf17pDbtGiRfjwww/x/PPPaybAfP/99zXPOzo6YvXq1Zg3bx6GDx8OV1dXTJ06lVMREJmh7KIKFFdUAwDcHcSdMfxeAjwdcDWrGMl5pYho7vrwA4ioQRElcZowYQI2bNiglZzUVbdu3RAfH3/f5++cIfx+XFxcNJNd3k9ISAg2bdqkd3xEZFzxGUUAABdbOeQy8e/c/adAL3vgIpCWXwalSoBM5FnLici8iJI4RUdH4/jx49i3bx9at2591+DwZcuWiVEsETUClzMKAcBk3WTeTjawlctQVqVERkE5fF1tTRIHEZmGKImTk5MTHnvsMTFOTUSN3OWaFicPE3TTAerlgPzcbHElsxjJt0qZOBE1MqIkTneuG0dEZEiaFicj31F3Jz83O1zJVA8Qj+TyK0SNivEHCBAR1VG1UoUrmepZjt3tTdPiBADNXe0AABmF5aiotqxlNoiofkRpcerXr98D52v666+/xCiWiBq4G7klqKxWQSGTwtlW9yVPDM3JVg4XWznyy6qQdqsM/p76zbxMRJZLlMTp+eef13pcXV2NS5cu4fDhw3jppZfEKJKIGoG4m+rxTd5O1qJPpvswfm52yE8rQHJeKRMnokbEKIlTrY0bNyI2NlaMIomoEagd3+TjbNyJL++luZsdYtIKkJLXeBZmJiIjj3Hq3bs3/vjjD2MWSUQNyOWaFicfJ9MnTs1q7qbLK61EcXm1iaMhImMxauK0Z88euLi4GLNIImpAaqciMIfEyUYug7eT+s6+pLwSE0dDRMYiSlfd008/rTX+QBAE5OTkIC8vD7NnzxajSCJq4ArKqpBWs7ivt5MNiipM38rTwt0emYUVSMotRbumXBCcqDEQJXHq37+/1mOJRAI3Nzd07doVAQEBYhRJRA1c7VIrvi62sFXIzCJxauluh5PX85CUVwqVSoCUy68QNXiiJE7Tpk0T47RE1IjVDgwP8XE0cSS3eTvZwEYuRXmVCje5/ApRo8AJMInIItRORRDSxHwSJ6lEghZu9gDUc0wRUcNn0BankJCQh86tIpFIcOnSJUMWS0SNwO0WJycUllWZOJrbWrrbIT6zCEm5pegZaOpoiEhsBk2cli1bdt/nzp8/j/Xr10OlUhmySCJqBFQqAVdqxjiF+Dji5PU8E0d0W3N39fIr2cUVKK6ohoO1KCMgiMhMGPQv/J+DwgEgMTERixcvxr59+zB06FBMnz7dkEUSUSOQeqsMJZVKKGRStPKwN6vEyU5hBW8n65q760p4dx1RAyfaT6PMzEx89dVX2L59O6KiorB9+3YEBQWJVRyR2VOqBJRVKVGlVMHFVm7yJUMsSVxNN11rbwdYycxvaGbLmmkJrucwcSJq6AyeOBUVFWHFihXYsGED2rRpg7Vr16Jz586GLobIYmQVluPItVwk55VqtjnbyhHm64x2TZ1MGJnlqJ0xPMTHPOvL39MeJ67nISm3FNVKlVkmd0RkGAZNnFatWoVvv/0WHh4eWLx48T277ojMmVwuN9i5qpUq7IvPxqWbhZptEgkghQQFZVU4nJCD6NR8vBTVymBlNlS1A8PbmNEddXfydLCGo40VisqruegvUQNn0MRp8eLFsLGxQfPmzbF9+3Zs3779nvs9aBA5kSm1bdcOMpms3ueprFZhZ3Q6Um+pZ7oO9nFEt1ZucLGVo0op4EpmEU7dyENheTW+OZiIfiHeCGvGLp77uZxh3i1OEokE/h72uJBagMScEiZORA2YQROnfy61QmRprGQy7IpOx61S/W53b+luh56tPQEAFdVKbD+XjozCcshlEgxt3xR+bnaafRVWEoT6OqOVhz1+vZCO7KIKjPn2OHZN76W1H6mVVlZr5kgypzmc/snf00GdOGWXQBUiQMrPQqIGyaCJ08KFCw15OiKTyCupQHaxfomTq50CAKASBOyJzUBGYTlsrKR4KsL3vgvS2ltbYURHX/wefRMpt8rw+uZz+GlKJMfH/MOVzGIIAuDhYA0PB2tTh3Nfvi62sLaSoqxKiYyCcjR14SziRA0RP6GJDOjvy1m4kVsKmVSCpx+QNNWytpJhVGc/OFpb4WxyPr76O8FIkVqOyzfNb6mVe5FJJWjpoZ5F/Fp2sYmjISKxMHEiMpA/L2ViX3w2AODREC94PyRpquVqr8BHw0IBAF/9fRXRqflihWiRasc3mevA8DsFaBKnEgiCYOJoiEgMTJyIDKCovArvbIkGAHRo5ow2TfQbxPxUuC+GdmgKlQDM3xXHL907XLp5e6kVc9fSwx5WUvVdk1lFFaYOh4hEwMSJqJ4EQcCmk8nIK6mEj5MNolp71Ok8M58IgcJKiuOJedh/JdvAUVomQRA0XXX6JqOmIJdJ4V/T6nQls8jE0RCRGJg4EdVTfGYRLqQWwEoqwYiOvrCS1u3PytfFFhN7tAQAfLL7MpQqtjrdLChHYXk1rKQSBHjZmzocnQTVjMVSD2rna0jU0DBxIqqHskolDtSMa5r+aOt630k19ZEAONpY4XJGEbafSzNEiBYtrqa1KcDTAdZW9Z9fyxhauNlBIZOiuKIa6QXlpg6HiAyMiRNRPRxOyEF5tQq+LrZ45ZGAep/PxU6hOc/XB65B1chbnTQTX1rAwPBaVjKppnWM3XVEDQ8TJ6I6Ss8v0wxcHt3FD3IDzb80rnsLOFhbISGrGAca+VinOAsa33SnIG91onc1s7jRJ79EDQ0TJ6I6UKoE/H05CwDQrqkTAgy4xIaTjRyju/gBAL49nGiw81qi20utWE6LEwD4udrBVi5DWZUSN/JKTB0OERkQEyeiOriQko/ckkrYymXoGVi3u+geZGLPlpBJJTiSkIuL6QUGP78lKK9SIrFmIklLa3GSSSWa7sVL6YUP2ZuILAkTJyI9FZVX4fj1XABAz0B32MoNP2i5masdngj1AQCsPnTd4Oe3BFczi6ESADd7BbwczXeplftpV5PsJeaUoKhcvyV8iMh8MXEi0tPBKzmoUgpo4myDtiK2hPyrlz8A4Lfom8grqRStHHMVl3F7qRVLXDzc3cEaPk42EATgXEq+qcMhIgNh4kSkh+s5JUjILoZEAvQL8RL1Cz3czwXtmjqhUqnCtkY4NUGcBc0Yfj/tmqpjP5N0i3M6ETUQTJyIdFRZrcK+ePWA8Ag/F3g4iN99NLprcwDA5pPJje6L9/JNy1mj7n6CvB0hl0mQU1yJk9fzTB0OERkAEyciHR2/noui8mo42Vihu7+7Ucp8KrwpbORSXM0qxtnkW0Yp0xwIgoDLGZY5FcGdFFZSzdQEa47eqPN5VI0saSYyZ1amDoDIEmQWluN8cj4AoG+Il8HmbHoYJxs5Boc1xZazqfjhZAo6tXAzSrmmlllYgVulVZBJJQj0MtxUD6YQ7ueCi+mF+N/FDKw/dgNSPbt33ewVeCKsiUjREZG+2OJE9BCqmjmbBABB3g5o6W7cNdOe66qe0+m36HQUNpK7s2oHhvt72MNGhLsWjcnDwRohPo5QCcCe2AxkFVXo9a8x3hhAZM6YOBE9xPnUfGQVVcDaSorerT2NXn6nFq4I8LRHeZUKf8RmGL18U6gd3xRiwd10d+oX7AUAOJ10CxXVShNHQ0T1wcSJ6AEKy6pw7Jp6zqZerT1gb2383m2JRIKnw30BAL+eTzd6+aZw+Y6pCBqCtk2d4O9pj4pqFS5yQkwii8bEieg+BEHA3/FZqFYJ8HWxFXXOpod5qiZxOnotB1mF5SaLw1hur1HXMBInqUSCl6JaAQDOJt9CtVJl4oiIqK6YOBHdR0xaAZJySyGTSvCoyHM2PUxzdztENHeBSgB2Rt80WRzGUFGtxLVs9fpulnxH3T8906kZnG3lKKlQIpatTkQWy+wTp1OnTuHll19GVFQUgoODsXfvXq3nBUHAl19+iaioKLRv3x4TJ07EjRs3tPbJz8/HW2+9hY4dO6Jz586YNWsWSkq0F968fPkyxowZg7CwMPTp0werVq0S+9LIjOWVVOLQ1RwAQM8Ad7jaK0wcEe7ormvYk2EmZBVDqRLgbCuHj5ONqcMxGGsrGR4JUo+RO3Ujj61ORBbK7BOn0tJSBAcHY/bs2fd8ftWqVVi/fj3mzJmDn376Cba2tnjppZdQUVGh2eftt99GQkIC1qxZgxUrVuD06dP44IMPNM8XFxfjpZdeQtOmTbF161bMmDEDy5Ytw48//ij69ZH5qVaqsOdiBqpVApq72SHcz8XUIQEABrdvAplUgujUAs3itw1RXO3AcAtdauVBOrZwgaONFUorlYhJa5yLNxNZOrNPnPr06YM33ngDAwYMuOs5QRCwbt06vPLKK+jfvz9CQkLw6aefIisrS9Myde3aNRw6dAgfffQROnTogM6dO+P999/H77//jszMTADAjh07UFVVhfnz56N169YYPHgwxo8fjzVr1hj1Wsk8/BaTgeyiCtjIpXisrbfZfHl7OFijV2sPAMD2BjxI/PJNy5/48n6spFJ0bamei+vUDd5hR2SJzD5xepDU1FRkZ2ejR48emm2Ojo7o0KEDzp07BwA4d+4cnJycEBYWptmnR48ekEqliI6OBgCcP38enTt3hkJxuzsmKioK169fR0EBfxU2JptOJON0Uj4A4PF2Pia5i+5BngpvCgDYcT6twS7BcjnD8pdaeZA2TZzgYidHWZUSp240ntngiRoK8/pW0FN2djYAwN1de/kLd3d35OSox6fk5OTAzU17tmUrKys4Oztrjs/JyUGzZs209vHw8NA85+zsrHNMSqXhf0GqVCrIZDKoBEAQ9BsXUbu/GHGJSSaTQRBURr3e0zfyMHtHLAAg0t8Nzd1s9ShfpSlfzJgfDfaErVyGG7mlOJeUhw5m0o2oj9rrvNf1CoKASzUtTkFeDvfcp67vDfX56/f3UPeyb78/pBIBvQLdsTM6A+eSb6FdE0e42MlFiflBdU2GY6n1bIrP2foQs571OadFJ07mKCYmxuDntLW1Rdu2bXErLxcZefqNbZFVOgBohfj4eJSVlRk8NjHUXm9urvGuN7WwGu/vy0OVUkCguzWaWZcjI0P3ySZ9bFQAfJGfX4CM7HxRY+7kI8fhFCW+3RuNlyIstzvrXn8ruaVK5JVUQioByjITcT5Hu5u0Pu8NoH5/D/Up+5/vDxtBgI+9DBklSuy9mIYov/sPgjfE37AYn0t0N0uq54e9n0sqVYjLrUJRhQr2cimcrCUIcJVDLpOY/HvF1PVs0YmTp6f6DpXc3Fx4eXlptufm5iIkJASAuuUoL097VfLq6moUFBRojvfw8NC0UNWqfVzb8qSrsLAwyGSGXSJCpVJn965u7lAq9Fu3y93RGgAQHBxs0JiMwd3dONebUVCOaSuPo6hShXA/FwwJ89F7aRMXF8ea/zqjUqbfnWD6xjzRJguH15/FiZvV+GJCe1gZad08Q1EqlYiJibnn38qflzIBZCPI2xHdOkXc9xx1eW8Ahvl7qEvZ93p/POpUiU0nU5BepESZlSNaedx7KZ/6xPyguibDseR6/uf7WaUScDQxF+dTCqCqGQ2QXdNimloCDAlrAnd39f7G/l4Rs55rz60Li06cmjVrBk9PTxw7dgxt2rQBoL5D7sKFC3juuecAABERESgsLERsbCxCQ0MBAMePH4dKpUL79u0BAOHh4ViyZAmqqqogl6ubzI8ePYpWrVrp1U0HqJs+xfrDkUoAiUS/L8na/S3tjxlQxy729WYWlmPCmlO4WVAOfw97fDexC3ZFp6GoQt9kRKopX+yY+4R4w9VOjtySSpy4kY/eQcZfBsYQ7vW3cilD/cs31Nf5gfVRl3quPa627LqqW9l3vz88HGwQ4eeCs8n52BefA19XO1hb3R2XIWIW83OJbrPEer7zPSkIAvZdydbMbt/MxRYhTRxRUqFEdFo+8kqq8OPpVIzv3gKA6b5XTF3PZv9TtaSkBHFxcYiLiwOgHhAeFxeH9PR0SCQSTJgwAV9//TX++usvxMfHY8aMGfDy8kL//v0BAAEBAejVqxf+85//IDo6GmfOnMGHH36IwYMHw9vbGwAwdOhQyOVyvPfee7h69Sp27dqFdevW4YUXXjDZdZP40vLLMHLlMVzLLkFTZxusmdgJbmYwX9PDyGVSDG7fBACwvYHN6RRbc4t+mK9+P1gsVXd/dzjbylFcUY2jCbmmDocauVM3buFieiEkUN8cM7yjL9o1dUbXVm54rktz+DjZoKJahc2nUnCrES8+bfaJU2xsLJ5++mk8/fTTAIAFCxbg6aefxtKlSwEAkyZNwrhx4/DBBx/gmWeeQWlpKb799ltYW1trzrFo0SL4+/vj+eefx+TJk9GxY0fMmzdP87yjoyNWr16N1NRUDB8+HAsXLsTUqVMxatQoo14rGU/czUKMXHEMSbml8HOzxY9TItHM1c7UYemsdgmW/13MRFlldZ3PozKzO/NqE6dQX8sdu6UPuUyKR0PUwwyi0wqQdssyxiFSw3MtuxjHEtXJe58gTwT/Yx41e2srjOjoCzc7BYorqjF7x0VThWpyZt9V161bN8THx9/3eYlEgtdffx2vv/76ffdxcXHB4sWLH1hOSEgINm3aVOc4yXLsic3Amz+dR2mlEv4e9tjwr25o6mJrUXfEdGruiibONrhZUI4DV3IgCALy9PwF6GavwBNhTUSKUH9ZheXIKqqAVNIw53C6Hz83O7Rr6oSL6YX441IGxnZtDmu5ZXX3kGWrrFZhf7z6LvNwP5f73q1rJZNiQDtv/Hw6BTsupOOJUB+z+gwxFrNvcSIyFEEQsPSvq3h5wxmUVioRFeiBrVN7oKmLralD05tUKsGQmu66ndHpyCupRFZRhV7/9E20xBabrm5tCvB0gJ3C7H/TGVTv1p5wtpWjqLwaf1/OMugcXbXjNonu5+SNPBRXVMPJxgo9A9wfuK+Pkw16t1aPq/zPrxdRXmU5PzgNhYkTNQpllUpM23QOn/95BQAwsUdLrH2hC1zszH9M0/0M7aCeDPOvuMwGMQN1bJp6QGpoIxnfdCeFlRSPt/OBRAJcySrWLDtjCG3btavzQFpz68olw8sqKse5ZPVErH2CPXW6S7dviCeaudoip7gCP51OETtEs9O4ftZRo5SeX4ZJ607jYnoh5DIJ5j0Viue6Njd1WPUW5uuMFu52SMotxeWMIjRxtryWszvFaMY3Nb7ECQB8nG3QvZU7jiXmYl98FjwdreHpaP3wAx/CSibDruh03CrVb4oNc+vKJXHsjs2ASgBaedjD30O3aTaspFJM6e2P//x6ESsPJOK5rs0ht7BpUeqj8VwpNUpnkm7hyWVHcDG9EG72Cmz8V/cGkTQB6vF9Q9urW52iUy1/aaCLtYlT08YzvumfOrd0RQt3O1SrBPwWnY4yA3WD5JXo141rjl25ZHixaQW4klkMCYDerfWbs/DZzn7wcLBGWn4Zfm3Aa2feCxMnC3DqRh6W70vAz2fScOhqNoor6n4XVWPy8+kUPPfNceQUVyDExxG/vtoTXVu5PfxAC1LbXXc1s9iixxrkFlcgvaAcANCukbY4AYBUIsHj7XzgbCtHYXk19sRmsLuMRPPf/QkAgCAfR72HLdjIZXgpqhUA4Ov9CVCpGs/7lImTmYtNK8DoVSfx2R/xuJBagLPJ+Vh79AYOJ+RA2YjeqPpQqQR8/Psl/PuXaFQqVRjYzhtbXukBPzfLmW5AV8E+jgj2doRSEHAtW/8lSMxFbYuZv6c9HMxsYWVjs5HLMDisCaykEiTnldbMpk5kWAlZxdgdq15WqnML1zqdY1z35nC0scK17BIcvJptyPDMGhMnMxfg6YBhEU0xLMIX/UM80cTZBkqVgDNJt7A3LtOgd980BNVKFV7bfA6rDl0HAEzvF4ivx3aCfQP+Mh7aQT0O5Uqm5SZO51LyAahvhSbA09Ea/duoJ+g9eDUHu2Numjgiami+3n8NggC0aeIID4e6jaVztJFjRMdmAICfz6QaMjyzxsTJzNkqZFj0THt8MSocjwR74tlOzfBEqPrum8sZRTh+Pe/hJ2kkKqtV+P5YEn6Pvgm5TIIvR4fjzceCIZVKHn6wBRtSM84p5VYpSusxGaYp1d7VE9G8br98G6JgH0d0bO4CAHjr5wuIu1lo2oCowcgsLMevNasO9Knnkk3PdFInTn9ezER+aeMYF8fEycJIJBIEeTuiX7B6tuGT1/MQn2G4W5ctVWW1Cr+eT0NiTgnsFTKsmdhVM7t2Q9fSwx6+LrYQBHXzu6VRqQScr2lximCLk5aeAR4I8LRHaaUSL649hczCclOHRA3ADyeTUa0S0KWlK/zquWJCqK8z2jRxQqVShR0XGscgcSZOFirU1xmdavqlD1zJbhDz+NRVVc0fbHpBOWyspNg0qTui9LxDxNK1rxlQHZ9peUl0Yk4JisqrYSOXItjH0dThmBWpVILnujRHgKc9bhaU44U1p3hzCNVLlVKFH04mAwDGR7Y0yDmfrWl1+vl04+iuY+JkwSL93eFqJ0dZlRLHExtnl51KJeD3mJtIyy+DQibFxJ4t77tcQENWu7Zben45isr1m6/H1Gq76dr7ujSquWB0ZauQYe0LXeHhoMClm4V4bdNZVCtVpg6LLNSflzKRWVgBDwdrPN7OxyDnfCq8KaykEsSkFeByRsPvUuanlAWTSSWa/ukLqfnILa4wcUTGJQgC9l/JRlJuKaykEjwV3rTezc6WysVOgabONgAsb5B4bTddeM14Hrqbn5sdvn2+C2zkUuyLz8bsHRd5YwjVyfpjSQCA57r6QWFlmBTA3cEaj7ZRDx/ZdjbNIOc0Z0ycLFwLd3sEeNpDENR33zQmZ5PzNbNNPx7qY5FrzhlSiI+61cnSfvGdS84HwPFNDxPu54IvR0dAIgE2nkjGyoOJpg6JLExCVhGOJeZCKoHBJwKunVNud2xGg0/qmTg1AL1ae0IqAZLzSpFR0DgGjybnleJIgjpR7N3aAwGeui0V0JC19naAVALkFFcix0JaH0srqzXjstji9HAD2/ngP4PbAgAW7r6MH08lmzgisiQ/1YxBerSNt8F/aPYN9oK1lRTJeaW41MDvAGXi1AA428o1g2pP3Wj4Y52KK9QzKgsA2jZx4tw/NWzkMrTysAegnqrCEsSkFkCpEuDjZGPxa+0Zy4tRrTCljz8A4N2tMdjFOZ5IB1VKFbbWdKON7Oxn8PPbW1uhd83QkT01E2s2VEycGojOLdRLiSTmlDTosU4qlYDdsTdRVqWEh4MCfYM9IZE07Hma9BHsrU6g4zOKLKK5/GxNNx2TX/3MfDwEz3VtDpUAvL75HA5caTyzNlPd7I/PRk5xBTwcFHgkuH5zN93PE6Hqwea7mTiRJXCzVyDAU93acDrplomjEc/ppFtIzy+HQibFoLAmsOJdWFpaedhDIZOiuKIaafllpg7noU5czwWgXtyWdCeRSPDR06EY0r4JqpQCXl5/BmeSGn5rM9Xdz6dTAADDInxFu3v10TbekMskSMgqRkKWZbR61wW/dRqQLi3VrU7xmUUWd0u6LrKKyjVftH2DPeGq56KUjYGVTIrW3urxXnE3zfuDq1qpwukb6iS/u7+7iaOxPDKpBJ+PVK8oUFalxMQ1pxBbc7ME0Z1yiivw9+UsAMCzInTT1XK2laNHgHoOvd0xDbfViYlTA+LtZKOZQTo2rWENzqtWqfC/i5lQCUCApz0nSnyANjV3113NKkJltfnO93PpZhGKK6rhaGOFNk2cTB2ORVJYSfH12E7o0tIVReXVGLPqOKJT800dFpmZ7efSUK0S0KGZM4K8xf3srO2u++MSEyeyEB2aqWeQjk1XD7ptKE5ez0NuSSVs5TL0C/HiuKYHaOpiA2dbOaqUglkvwXKy5kaGLi3dIGvg6wmKyVYhw5oXuqJzC1cUlldj7LcnNJOKEgmCgF9qFuB9RsTWplr9auZzik0rRFYDXSKIiVMD4+/pAHuFDKWVSrP+0tRHbnEFztSM2+oX4gU7hZWJIzJvEokEbZuqW3DM+bbgkzULVHdr5WbiSCyfg7UV1r7YVdPyNPbbEzh0lQPGSZ3AXM4ogsJKiidrFgQXk5ejjeYH/L74LNHLMwUmTg2MTCpBaM26ZdFp+aYNxgAEQcDfl7OgEgB/D3vNAHh6sDY+jpAASMsvM8sVy5WCgFMc32RQDtZWWPtCV/Rq7aFZFLixLLpK9/fzGfWg8IHtfOBsJzdKmX1D1K1OteOqGhomTg1QqK8zJBL1umWWPiHmxfRCpBeUQy6ToA+nHtCZo40czd3Uy8+Y4yDx5IJqFJZXw8HaCu2acnyTodhbW+Hb5ztjcM3ddm/8FI0le69YxNQUZHjlVUr8el6dPNcuxGsM/WoSp8NXcxrkAvRMnBogB2srzUzaJy14QszSymrN7ODd/d3hZGOcX0sNxZ3ddSozG+92MVvdCtaphSunlDAwaysZlo6OwL+iWgEAluy9ih9Pp5n1jQIkjr1xmSgoq0ITZxv0DPQwWrmhTZ3h6WiNkkolTl1veOPt+InVQIXWfGleSMlHWaVlZvx/xWWhvFoFd3sFwpu5mDoci+PvaQ9buQzFFdW4nlti6nC0XMxSJ07d/Dm+SQwyqQTvD2mLBcNCIZdJEJteiE0nkxvsYF26t59rllgZ0bGZUW/AkEol6FszyWZD7K5j4tRANXezg5ONFcqrVRa5JMPljEKcqBk83CfIE1LedaU3K6lU0+oUnWo+8/tUVqsQU5M49Qww3q/gxmhk52b4YVJ3ONtaoaCsCj+dTsWJ67kN6o5burf0/DIcrLlB4BkjdtPV6qcZ55Rp9LLFxsSpgZJIJGjXVD1IfLOFLQQqCALm7LgIAUCglwP8asbqkP7Cam4USM4rNZtB4meSb6GsWoCbvUITH4mnc0s3vPpIAAI87aEUBBxPzMMPJ5OReqvU1KGRiLacSYUgqO9abelh/Jtqolp7Qi6T4EZuKW7kmFeLd30xcWrA2jZ1glQCnLpxy6Kmv//zUiaOJ+bBSipBLyP2yzdEzrZytHBXJ57mMinqgSvqcWt9gjzYkmgkdgoZBoc1wcB23rCVy5BbUoktZ9Ow/XwaMtl91+CoVAJ+rpm7aVQX8eduuhcHayvNGqoHG9jUGEycGjAHayvNLLGbT6aYOBrdVClVWLj7MgCgZ6AHnGw5ILy+2te06ly8WYBqpekHCO+PV3+IPhIkzkKjdG8SiQQhPk4YH9kC7X2dIZUASbml2HwqBb+cScXVrCJ24TUQJ67nITmvFA7WVngitInJ4uhd8zd+IJ6JE1mQLjWLp245m2oRt4X+cDIZiTklcLdXoHdrtjYZQksPezjaWKG8SoXLGaZteUzLL8PVrGJIAUQFcv4mU7CVy9A3xAvju7dAiI8jpBL167IrJgOrDiXir7hMJGYXo7zK/D8v6N5+qlnQd2iHprBVyEwWR5+axOlYYq5FfP/oiolTA9fayxE+Tja4VVqF/10070F6heVVWLL3KgDg/wYEwUZuuj/4hkQqkSDczwUAcDb5lknn9NlfM5NwkLscLlyk2aRc7BQY2M4HL/Rohc4tXGGvkKGiWoXY9ELsjL6Jj3fF4dkVR7H0r6s4k5TH6QwsRGF5leaGoJGdjT8o/E5tmjjC09EapZVKnLnRcKYlYOLUwMmkEs0fj7kPEv96/zXklVQiwNMeo03UL99QtWvqBIVMilulVbiRa7pBwfsuq5vsOzaxNlkMpM3Bxgo9Az3wYlQrDIvwRZivM5xt5VAJ6vGRn/95BSO+Pob2c//A2G+P48u9V3HsWu5DW6RU9UjQ63NsY7fzQjoqqlUI8nbQ/GAyFYlEgt6ta7rrrjSc7jou+tUIjOzih6/2JeBIQi6SckvQwt38li1Jyy/D6sPXAQDvPtEGck6KaFDWVjKE+jrhbHI+zibfQisT3GVTVqnE0WvqgeFMnMyPVCJBczc7zYzzVlIJ3B2scehqNk5cz0NeSSWOJOTiSEIuAEAhk6KDnzN6tfbEwHY+CPJ20JrZXyqRYHfMTeSV6Hc3p5u9Ak+EmW5cjqX76ZS6m25kZz+zWGmhT7AntpxNxYEr2Xh3UBtTh2MQTJwagWaudujV2hMHr2Tjx1MpmPF4iKlDusuiP+JRWa1Cd383PFqzujYZVgc/F5xLyUfqrTJkFZbDy9G4yctflzNRWqlEM1dbtHTmR4+5c7NXYEy35hjTrTkEQUBCVjGOX8/Dyet5OJGYi6yiCpy6cUvTKuXvYY8RnZpheEdfNHG2BQDklVQiq6jCxFfSeFzOKMSF1AJYSSV4OsLX1OEAAHoFekAiAS5nFCGzsBzeTjamDqne+LO+kXiupuvr5zOpqDKDO6vuFJNagG3n0gAA7w1qaxa/khoiJxu55i5LUyzFs6Nmzayh7ZvwNbYwEokErb0dMb57C3z1XAROzHoU+99+BAuGh+HREC8orKRIzCnBZ3/Eo+fCvzFt01lEp+abOuxGp3am8P5tvOHhYB6tuq72CrSvWfmhoXTXMXFqJB5t4w0PBwWyiyrMagp8QRDw4e+XAABPhzdFWDNOiCimri3V86pcyy5Ben6Z0cotKKvSTEMwtD27YSydRCJBSw97PNe1OVZP7IKz/xmAz55pj66t3KASgN+ib+LJZUew9ugNZOvZ4mSnkHF8VB1UVqs0P0BHdjHtoPB/6lNzh/TBBpI4sb28kVBYSTGiUzOsPJCIzSeTMbCdj6lDAgD871ImTl7Pg7WVFP82wy7EhsbNXoFgb0fEZxbh7/gso9X5H7EZqFSqB6wG+zjifIZRiiUjcbC2wrOd/fBsZz/E3SzEqoOJ2HEhHVezinE1qxghPo6ICvSAvfXDv3KsrWQcH1UHf17KRF5JJbwcrTUDss1Fn2BPLP07AYeu5kCpEoy6bp4Y2OLUiIzu0hyAurnUmK0N91NZrcKCXXEAgEm9/OHrYmviiBqHrq3UrU5xN4sQm2acNex2XFB30z0Vbh7jLkg8bZo44fNR4fjrrT5oX9OCfDmjCOuOJeFCSr7O02HUjo/S55++iVZDsu7YDQDqQeFWZnZzTYdmLnC0Ua+X2BC6cM2rdklUrTzs0d1f3ZReO0GaKa0/noQbuaXwcLDGy48EmDqcRqO21QkAPv0jXvTysorKNXfTDW3fVPTyyDy0cLfHqM5+GN3FD16O1qhUqrD/Sja2nktDUXmVqcNrUGoXRZdJJRjbvbmpw7mLlUyKXjXddQ1hnBMTp0bmua7qP6qfTqWYdHmF/NJKLP1LPdnl248FwUGHJnwynO7+bpBJJDh4JRv74sUd8/bjyRSoBKBjcxc0d+eCzY2Nt5MNRnXxwyNBnrCSSpB6qwwbTyQj3sSz2Dck644lAQAea+utuaPR3NTOIs7EiSzOwHY+cLGTI72gXPRB4g8apLn0rwQUlFUhxMcRz3bmZJfG5mKnQGSAesmTj367JNqdlpXVKqw7rv5Qf75HS1HKIPMnlUjQwc8FY7o1h7eTNSqqVdhzMQN7YjNQwaVd6qWgrArbzqoHhU+IbGnaYB6gdt26Cyn5yC+17C5V/sxvZGzkMozq4oeVBxKx+nAiBrT1Fq2s+w3wzCmuwNqj6skuI/3dsfmk9ozmLd3t0NPMBjc2RH2DPRF3sxDXskuw4XgSXujZyuBl/B6TjuyiCng7WZt0sVEyD652CjzbyQ+nbuTh5I08xGcW4WZBGZ4IbQIfZ8uf38cUfj6dgrIqJYK8HdDd383U4dxXE2dbBHk74EpmMQ4n5GCIBXfbs8WpEXo+siVkUgmOJ+bhYrq4g4PvNcBzx4V0qAR1guRkK7/r+YKyalFjIjUbuQxvPhYEAPj8f1dws8CwNwwIgqCZDX5CZEsorG5/3MjlcoOWRZZDJpWgu787Rnbyg7OtHIXl1fj5TIrJ11GsL1O8p6uUKqw5cgOA+m/M3OdH0yy/Em/Z3XVMnBqhpi62GFRzy27tF5uxJOWW4Fp2CSQSoBdblUxudJfmiGjugqKKaszcEmPQL65TN24hNq0Q1lZSjOmqPWC1bbt2kMm4iHNj5uNsg+e6+qG1lwNUAnDoag5+i76JkgrT/XCq6xxQMpkMbdu1M3A0D/dbdDrS8svg4aDAM52MN3dTXefa6hOs/sw/eDXbopNki++q++qrr7Bs2TKtba1atcKePXsAABUVFVi4cCF27dqFyspKREVFYfbs2fDw8NDsn56ejjlz5uDEiROws7PD008/jbfeegtWVhZfPff1UlQr7LyQjp0X0jHz8RB4GWEa/GqVSjMJYodmLnCzV4heJj2YTCrBZ8+0x6Clh3HgSjZ+PpOKkQYYcyYIAhbV3LE3vGMzuP7jtbaSybArOh23SvW7u4rduA2LtZUMT4T6wDetAIeu5CAxpwQLdl9GgLeDSeKp6/xRrnZyDGrfFEql8cZrCYKAlQcSAQATe7SEjdx4P0TqOtdWlVIFuUyCzMIKxGcWIcTHScQoxdMgMoPWrVtjzZo1msd3/pKdP38+Dhw4gCVLlsDR0REffvghpk2bhs2bNwMAlEolpkyZAg8PD2zevBlZWVl45513IJfL8eabbxr9Wowl3M8FnVq44kzSLaw+ch3vPiH+4otnk/KRX1YFO4XMrPviG5tAL0e8OSAIC3dfxoc7L6FrSze0rOciwL/H3MTJG3mwkUvxWr/Ae+6TV1KB7GL9EidXOybbDY1EIkGHZi5o4myDXTEZyCutxMgVx9C/jfddCwcbQ13W1xME4y9jtT8+G5czimCvkGF895ZGLx+oW1218rDHlcxiHIjPttjEqUF01clkMnh6emr+ubmpv5SLioqwZcsWzJw5E5GRkQgNDcX8+fNx7tw5nD9/HgBw+PBhJCQk4LPPPkObNm3Qp08fvP7669i4cSMqKy175P/DvNJHPXfSuqNJyCkWdyHOgrIqnKpZH61Xaw9YW7GbxpxM6uWPzi1cUVRRjcnrT6O4Ht0lZZVKzP9dPbHpK30C0ZQTm5IOvBzVXXedmruiWiVgz8UM7Iy+iTLedXcXQRDw3/0JAIAx3ZrD2c5yxgy29lLPIXfwquWOc2oQiVNSUhKioqLw6KOP4q233kJ6unqW4tjYWFRVVaFHjx6afQMCAtC0aVNN4nT+/HkEBQVpdd1FRUWhuLgYCQkJRr0OY3u0jRfaN3NGWZUS3xxMFK0cQRDwV1wmqlUCmrnaaiZfJPMhk0rw37Ed4eVojSuZxXjrp/NQ1XGer6/3JyC9oBy+LraY0sffwJFSQ2ZtJcOLPVvi42GhsJJKcD2nBJtOJCPNDFY6MCcHr+bg1I1bUFhJ8VKUZf2NtfZSd8Oeun4LpZWWeSOQxXfVtW/fHgsWLECrVq2QnZ2N5cuXY+zYsdi5cydycnIgl8vh5KTdHOju7o7sbHW2m5OTo5U0AdA8rt1HH2L0catUKshkMqgE/ZuEa/e/X1zT+wXiX+vOYN2xG3ipZwuDrqgtk8kgCCrEphUg5VYZrKQS9Av2BCA8ZGCgShO7oa/3QepTz6aKGbhdz/Ut191ejv+OicCYb0/gj4uZ+ODXWMwe0gZSPdaV2h+fjWX71D84Zj4eDLn07usyVT3XHgcYv64t8T1tqrqSSASM7dYC6bdKsfFkCvJLq/DLmVRE+Dkj0t/tgcuJmCrm2t8YKpX4XXYqlYBPd18GAIzv1hyeDnKLej97OCjQzNUWqbfKcPRqNvqGeOl8bO11ivE9q885LT5x6tOnj+b/Q0JC0KFDB/Tt2xe7d++GjY3x5wWJiYkx+DltbW3Rtm1b3MrLRUZesV7HyiodALRCfHw8ysru/tXmIgho7SbH1bwqzPvlBF4MN0yfc23M19OycPCqermNUE85ygtzkVH44GN9bFQAfJGfX4CM7Hy9yn3Y9eoSc13q2dQx5+Ya5r0hATClkxOWnSzAhhPJyMjKxpROTpDqMM4ktbAa7/6VC5UA9G9liybVN3H+Hqv5mqqegdvXnJSUhPLycr2OtbGxgb+/f53q2tTvDzE+O3Qptz51ZStUoK+fHOcyBdzIr8a5lAJczSxERx9r+Djcu6vfZH9Lbupyr169qne5+jqSUoaLNwthayVBlHuppvdEX4Z4jer0fnZzQJ8gT2w8kYwtxy7DtTxdr+MBcb5n9WHxidM/OTk5oWXLlkhOTkaPHj1QVVWFwsJCrVan3NxceHqq78zx8PBAdHS01jlyctRf9LX76CMsLMzgt1nX/opxdXOHUqHf3SbujuoWpODg4Pvu875jDp5fcxp/XCvDq09EGKwrrVqpwl/XS1GtApo42yCqbVOdvoBdXBxr/uuMSpl+ya8u13s/9alnU8WsOYe74d4b4eFAi+ZpmLElBnuvl0GpcMDC4aEPbI28nFGERXvPorRaQJeWrvhqYheteZvuZKp6BgA/dzuoBAH+/nXv3qhLXVvie9pU78s768pPZgM/X+BGTgn+upyN4kolDiaXI9DLHj0D3OFsqz22x1Qxu9b8bbRu3RpSqXgjYKqUKrz192EAwJQ+Aejd7d43XujDFO/nECdbbDyRjEu3gPDwcJ2PVSqViImJEeV7tvbcumhwiVNJSQlSUlLg6emJ0NBQyOVyHDt2DAMHDgQAJCYmIj09XfNihYeHY8WKFcjNzYW7u3oJiqNHj8LBwQGBgfq/KWUymWjz00glgESi3x9l7f4PiqlPsDcGtvPGHxczMXvHJfw4OVKv7pn7WfRHPG7klkIuk2BAW2/IpLrWizpmiUQqyvU+tPQ61LOpYzZ0uc90bg5ruRXe+ukC9sVn44mlRzDz8RAM7dAUtorb+1cpVfjxVAo+/O0SKqpV8HOzxYpxnWBr/fDBqsauZwCwkcvrfMt57VQIdSvb8t7Tpnpf3quuWnk6YryrHY4n5uFCSj4SskqQmF2C9r4u6NzSFfY1a12aKubaj0upVCrq/GTfHr6BG7mlcLdX4F+9AwxSlinezz0C3GEllSAptxSp+eVo4a7fXbxifs/qwuITp08++QR9+/ZF06ZNkZWVha+++gpSqRRDhgyBo6MjRowYgYULF8LZ2RkODg746KOPEBERoUmcoqKiEBgYiBkzZuDf//43srOzsWTJEowdOxYKReO59Xn20HY4VDPgcMvZ1HqvH3fgSjaW19z18WiIN28jt0BDOzRFa28HvPHjBcTdLMSMLdH48LdL6B3sCRdbOUoqqrEvPhsFZeopBfoGe2LxyHCLmJ+rLrdR8z1sWtZWMvQJ8kTbJk44nJCD5LxSnE/NR0x6Adr4OKJjc1d4ORpujKa5Sb1ViiV71QujvzuojUUvjO5oI0fHFq44eT0PB69kY3xk/aY/MTbLrfkaGRkZePPNN5Gfnw83Nzd06tQJP/30k2ZKglmzZkEqlWL69OlaE2DWkslkWLFiBebMmYNRo0bB1tYWw4YNw/Tp0011SSbR1MUWrz/aGgt2X8bHu+IQGeCOZq51W8n+amYRpv9wDoIAdG3phmAf3kVnqUJ8nPDrqz3x7eFEbDqRjNRbZfg9+qbWPh4O1ni5jz9e7NnKIC2VRA/i6WiNYRG+SM4rxbFrucgoLEdseiFi0wvRyt0OtgoZHg3xtqhb9B9GEATM/vUiyqqU6NbKDSM6+po6pHrrE+SJk9fzcOBKNsab8eLE92LxidMXX3zxwOetra0xe/ZsrWTpn3x9fbFq1SpDh2ZxXoxqhd+ibyImrQAvbziDX17uofdstDcLyjDhu5MoKKtCRHMXDArz0Xt2aDIvCisppj4SiJd7B+D49VzEphWgtFIJlQD0DHBH55ZukDFhIiNr7mYHP1dbpOeX40zyLdzIKcH13FK8+dMFWEkl6Obvhsfa+qB/W2/4WvhcYr/H3MRfl7Mgl0nw8bBQs1+TThd9gjzx2R/xOHotF5XVqvuOiTRHFp84keHIZVJ8Pa4jnlx2BLFphZi1LQaLn+2g8x9pRkE5Jqw+iZsF5QjwtMd3z3fBrpibDz+QLIJUKkGPAA/0CPB4+M5ERiCRSODragtfV1sUlVchKbcUKbdKcSWzGEcScnEkIRezd1yEv6c9urRwQ5dWbuja0g1+brYWk3wk55bi3S3qQcsv9wlAoFfDaMFv28QJHg4K5BRX4nRSnkV9rjBxIi3NXO2w7LkIjFt9AlvPpsHaSoYPn2r3wLlTAOBiegFeWnsaGYXl8Hayxvcvdr1rfTIiIrE42sjRL8QLY7u3wI2cEvx5KRP/u5SB00m3kJitHkz+4+kUAIC3kzU6t1QnUd393dHay8Esu5krqpWY9sNZFFVUo1MLV0x/tLWpQzIYqVSC3q09sfVcGg5eyWHiRJatR6AHPno6DO9tj8EPJ5ORnl+GJaPC75kIVSlV2Hg8CZ/+EY/SSiUCPO2xZmLXOo+PIiJttSvR6zKVxz+Z8s4jU2rpYY9Jvf0xqbc/8ksrcfrGLZxKysOp63mISStAZmEFfo++qRmv52onR7dW7ujm74ZbNXdbmrpFShAEzN15CdGpBXC2lWPpcxGQP+QHrKXpHaROnA5cycbMJ0JMHY7OmDjRPY3p1hweDgpM33wOB65kI+qTvzGuewsMDPWBj5MNbpVW4kRiHjaeSMK17BIAQI8Ad3w9rtNdc6sQUd3VdSV6AGjhboeo1vrPR9eQuNgp0L+tN/q39QagXkvxQmo+Tl3Pw8kbeTh94xZulVZhz8UM7LmonrDV2VaOVh728PewR1MX24eO4atNbuuaqN4rMf78zyvYdCIZEgmw+NkOFj9O6156tfaARALE3SxERkE5fJyNP2l1XTBxovt6rJ0PfpoSiXe2xCDuZiFWHkzEynusaedur8AbA4IwuovfQ7v0iKhu6jaFQt0/4uvT0mXObBUydPd3R3d/9bx9ldUqxKQV4HhiLo4n5uLYtVwUlFXhfEo+zqfkQ2ElRYCHPYJ8HOHnanfPJMpGrk5ud0Wn630zjJu9Ak+ENdHa9s3Ba/jqb/V0Lh8+FapJ+hoadwdrhPu54FxyPvbGZWJc9xamDkknTJzogdo3c8Gu6VHYF5+FtUeTcC2rGJmF5bCRy9CphSt6BrpjdNfmcLJhKxNRQ1Kflq7ayUItgcJKik4tXNGphSte7RuI7w4n4kxSPq7nlOB6TgnKqpSIyyhCXEYRbORSBHo5INjbEU1dbO9KKvNKKpBdXPe7iKuVKnz0exzWHr0BAPj3wGCLSSbqakBbb5xLzsf/LjFxogZEIpGgX4g3+oWof/Woala0NMfBlERkWI1tslBrKxkCvRwQ6OUAQRCQXlCOq5lFuJJZjLIqJWLTChGbVgh7axlaezki2NsRQV71H9OZll+GGb9cwJGEXADqpGnqIwH1Pq+5e6ytDz7dE49j13JQWF5lET/CmTiR3pgwEVFjIJFI4OtiC18XW/Ru7YnU/DJcySxCQlYxSiqUmu68vXGZuHizGFIIkEqleg0sr6hSYtnfV7FsXwLKq1SwU8jwxahwDGznI+KVmY9ALwf4e9ojMbsEB+KzMbRDU1OH9FBMnIiIiB5CKpWguZsdmrvZ4ZFgTyTnqueLSswpRm5JJf67/xoA9cDyZq7qZMvT0Rqudoq7xkWVVFQjs7AciTkluJpZjEqlejHmrq3c8PHToWhtoIXWLcVjbX2w4sA1/O9SJhMnIjJf9R382xAHDhPpwkoqhb+nA/w9HVClVKGiqhqp+eX4Ky4LBWVVKCirwsX0QgCABOrB49ZWUigFAZXVKlRUq7TO5+9hj9f7t8aTHZqafBoEUxjQ1hsrDlzD/stZFjGLOBMnokaqPoN/73UnEFFjJJdJEdrUFf8Z6ovVBxNwIa0IKbdKkVlQjpySSlRWq1BWpURZlVLrODd7BZo426CHvztmDW7TKBOmWhF+LvBwsEZOcQWOJeaiT5B531jAxImokavL4F8iupu1XIZWHvZo5WEPQD2JZWmlOmmqqFJBJpVAYSWFvbUM1lbqOZ+8HK0bddIEqLtBB7bzxsYTyfg9Ot3sEyfzbg8jIiKyUBKJBPbWVvBwsIavqy18nG3gZq/QJE1025D26rFNe2IzUPmPrkxzw8Spgasdx1JX9TmWiIhIF11bucHL0RqF5dU4dDXb1OE8ELvqGjiOYyEiargayk0aMqkEg8KaYO3RG9h5IR2PtjHf2dKZODUSHMdCRNTwNKTZ3Yd2aIq1R2/gz0uZKK9SwkZunl2aTJyIiIgsWEOZ3b1jcxf4utgiLb8M+y5nmW2PB8c4ERERkclJJBIM6aBOlrafTzNxNPfHxImIiIjMwvCIZgCAv+KykFNsnsNLmDgRERGRWQj2cUQHPxdUqwRsO2uerU5MnIiIiMhsjOysbnX68XQKBDOcEoeJExEREZmNoR2awkYuRUJWMc4m55s6nLswcSIiIiKz4WQjx6CaO+p+OpVi4mjuxsSJiIiIzMqozn4AgB0X0lFQWmXiaLQxcSIiIiKz0rWVG0J8HFFWpcQPp5JNHY4WJk5EpLf6rIEok5nnbMBEZD4kEgleimoFAFh75AaqlOaz8C9nDicivdVnDcQW7naIMrOlHohMpfZHSENYb87Qngxvik/2xCOjsBy7Ym5iSJiPqUMCwMSJiOqhbks98GOHqFZ9foSY43pzhmRtJcPzkS2w+M8rWHUoEYNDzWPhX36CERERmVhDWW/O0MZ2b4Fl+xIQm1aIwwm5cDR1QOAYJyIiIjJTbvYKjO3WAgCw+M8rZjEhJhMnIiIiMltT+wbATiFDTFohTqSZfv06Jk5ERERktjwcrDV32P0QWwSlyrStTkyciIiIyKxN6u0PZ1s5UouU2BefbdJYmDgREVGDUJ/5xci8OdnI8cGQNvCyk8HHydqksfCuOiIiahB4a3/D9nR4U7REFkJ9nU0aBxMnIiJqUHhrP4mJXXVEREREOmLiRERERKQjJk5EREREOmLiRERERKQjJk5EREREOmLiRERERKQjJk5EREREOmLiRERERKQjJk7/sHHjRvTr1w9hYWF49tlnER0dbeqQiIiIyEwwcbrDrl27sGDBArz66qvYtm0bQkJC8NJLLyE3N9fUoREREZEZYOJ0hzVr1mDkyJEYMWIEAgMDMXfuXNjY2GDLli2mDo10VJ9FPmUymYGjISKihoZr1dWorKzExYsXMWXKFM02qVSKHj164Ny5cyaMzHRqkxCpRGLqUHRWn0U+W7jbIcoEi3xaYj0TEZmCXC43dQhMnGrdunULSqUS7u7uWtvd3d2RmJj40OOFmlaOyspKg7dcqFQqAICbrRWk0K81xclGCqVSCTdbGSSCfi+3p4McgkqF49dzUVRepdexXo42CG3mUqdy6xNz7bFQKSERlHodK6iU6nLtWM9iHutoLTNJPdf3eEs81lR13diOZT3rzsdJgarq6jr/UAwOCRHle1apVH9fCDr0WEgEXfZqBDIzM9G7d29s3rwZERERmu2ffvopTp06hZ9//vmBx1dWViImJkbsMImIiEgkYWFhUCgUD9yHLU41XF1dIZPJ7hoInpubCw8Pj4ceb2VlhbCwMEilUkjY5UJERGQxBEGASqWCldXD0yImTjUUCgXatWuHY8eOoX///gDUXWTHjh3DuHHjHnq8VCp9aJZKRERElo2J0x1eeOEFvPPOOwgNDUX79u3x/fffo6ysDMOHDzd1aERERGQGmDjdYdCgQcjLy8PSpUuRnZ2NNm3a4Ntvv9Wpq46IiIgaPg4OJyIiItIRJ8AkIiIi0hETJyIiIiIdMXEiIiIi0hETJyIiIiIdMXEyExs3bkS/fv0QFhaGZ599FtHR0Q/cf/fu3Xj88ccRFhaGoUOH4sCBA0aK1LLpU88//fQTxowZgy5duqBLly6YOHHiQ18XUtP3/Vzr999/R3BwMKZOnSpyhA2HvnVdWFiIuXPnIioqCqGhoRg4cCA/P3Sgbz2vXbsWAwcORPv27dGnTx/Mnz8fFRUVRorWMp06dQovv/wyoqKiEBwcjL179z70mBMnTmDYsGEIDQ3FgAEDsHXrVvEDFcjkfv/9d6Fdu3bCL7/8Ily9elV4//33hc6dOws5OTn33P/MmTNCmzZthFWrVgkJCQnCF198IbRr106Ij483cuSWRd96fvPNN4UNGzYIly5dEhISEoSZM2cKnTp1EjIyMowcuWXRt55rpaSkCL169RLGjBkjvPLKK0aK1rLpW9cVFRXC8OHDhUmTJgmnT58WUlJShBMnTghxcXFGjtyy6FvPO3bsEEJDQ4UdO3YIKSkpwqFDh4SePXsK8+fPN3LklmX//v3C559/Lvzvf/8TgoKChD///POB+ycnJwsdOnQQFixYICQkJAjr168X2rRpIxw8eFDUOJk4mYFnnnlGmDt3ruaxUqkUoqKihJUrV95z/9dff12YPHmy1rZnn31W+M9//iNqnJZO33r+p+rqaiEiIkLYtm2bSBE2DHWp5+rqamHUqFHCTz/9JLzzzjtMnHSkb11v2rRJePTRR4XKykpjhdgg6FvPc+fOFSZMmKC1bcGCBcLo0aNFjbMh0SVx+vTTT4XBgwdrbfu///s/4cUXXxQzNIFddSZWWVmJixcvokePHpptUqkUPXr0wLlz5+55zPnz5xEZGam1LSoqCufPnxczVItWl3r+p7KyMlRXV8PZ2VmsMC1eXet5+fLlcHd3x7PPPmuMMBuEutT133//jfDwcMybNw89evTAkCFDsGLFCs3K8HS3utRzREQELl68qOnOS0lJwYEDB9CnTx+jxNxYmOq7kDOHm9itW7egVCrh7u6utd3d3R2JiYn3PCYnJ+eu2czd3d2Rk5MjWpyWri71/E+LFi2Cl5eX1gcoaatLPZ8+fRq//PILtm/fboQIG4661HVKSgqOHz+OoUOH4ptvvkFycjLmzp2L6upqTJs2zRhhW5y61PPQoUNx69YtjBkzBoIgoLq6GqNHj8bLL79sjJAbjXt9F3p4eKC4uBjl5eWwsbERpVy2OBHp4JtvvsGuXbuwbNkyWFtbmzqcBqO4uBgzZszAhx9+CDc3N1OH0+AJggB3d3d8+OGHCA0NxaBBg/Dyyy9j8+bNpg6tQTlx4gRWrlyJ2bNnY+vWrVi2bBkOHDiA5cuXmzo0MgC2OJmYq6srZDIZcnNztbbn5ubed408Dw+Pu1qXHrQ/1a2ea61evRrffPMN1qxZg5CQEDHDtHj61nNKSgrS0tLwyiuvaLapVCoAQNu2bbFnzx40b95c3KAtVF3e056enrCysoJMJtNs8/f3R3Z2NiorK6FQKESN2RLVpZ6//PJLPPnkk5qu5+DgYJSWluKDDz7AK6+8AqmUbRaGcK/vwpycHDg4OIjW2gSwxcnkFAoF2rVrh2PHjmm2qVQqHDt2DBEREfc8Jjw8HMePH9fadvToUYSHh4sZqkWrSz0DwKpVq/Df//4X3377LcLCwowRqkXTt579/f2xc+dObN++XfOvX79+6NatG7Zv3w4fHx9jhm9R6vKe7tixI5KTkzXJKQDcuHEDnp6eTJruoy71XF5efldyVJusClwe1mBM9V3IxMkMvPDCC/jpp5+wbds2XLt2DXPmzEFZWRmGDx8OAJgxYwYWL16s2X/ChAk4dOgQvvvuO1y7dg1fffUVYmNjMW7cOFNdgkXQt56/+eYbfPnll5g/fz58fX2RnZ2N7OxslJSUmOoSLII+9WxtbY2goCCtf05OTrC3t0dQUBC/zB9C3/f0c889h/z8fHz88ce4fv069u/fj5UrV2Ls2LGmugSLoG899+3bFz/88AN+//13pKSk4MiRI/jyyy/Rt29frdY+0lZSUoK4uDjExcUBAFJTUxEXF4f09HQAwOLFizFjxgzN/qNHj0ZKSgo+/fRTXLt2DRs3bsTu3bsxceJEUeNkV50ZGDRoEPLy8rB06VJkZ2ejTZs2+PbbbzXNwDdv3tT69dKxY0csWrQIS5Ysweeff46WLVti+fLlCAoKMtUlWAR963nz5s2oqqrC9OnTtc4zbdo0vPbaa0aN3ZLoW89Ud/rWdZMmTbB69WosWLAATz75JLy9vTFhwgRMmjTJVJdgEfSt51deeQUSiQRLlixBZmYm3Nzc0LdvX7zxxhumugSLEBsbiwkTJmgeL1iwAAAwbNgwLFy4ENnZ2bh586bmeT8/P6xcuRILFizAunXr4OPjg48++gi9evUSNU6JwHZDIiIiIp3wZx8RERGRjpg4EREREemIiRMRERGRjpg4EREREemIiRMRERGRjpg4EREREemIiRMRERGRjpg4EREREemIiRMRNQgnTpxAcHAwCgsLddp//Pjx+Pjjj0WOiogaGs4cTkRGExwc/MDn67OcTWVlJQoKCuDh4QGJRPLQ/fPz82FlZQUHBwcAQL9+/TBhwoR6r3OVl5eHL7/8EgcOHEBOTg6cnZ0REhKCqVOnolOnTvU6NxGZHteqIyKjOXz4sOb/d+3ahaVLl2LPnj2abXZ2dnU+t0KhgKenp877u7i41LmsB3nttddQVVWFhQsXws/PD7m5uTh27Bjy8/NFKQ9QJ41cEJnIONhVR0RG4+npqfnn6OgIiUSieezu7o41a9agd+/eCA0NxVNPPYWDBw8CAARBwMSJE/HSSy+htpE8Pz8fvXv3xpdffgng3l11Z86cwfjx49GhQwd06dIFL730EgoKCgBod9WNHz8eaWlpWLBgAYKDgxEcHIzS0lJ07NhRK7EDgL179yI8PBzFxcV3XV9hYSFOnz6Nt99+G927d4evry/at2+PKVOm4NFHH9Xa74MPPkCPHj0QFhaGIUOGYN++fZrn//jjDwwePBihoaHo168fvvvuO61y+vXrh+XLl2PGjBno2LEjPvjgAwDA6dOnMWbMGLRv3x59+vTBRx99hNLS0rq9WER0T0yciMgsrFu3DmvWrME777yDHTt2ICoqClOnTsWNGzcgkUjwySefICYmBuvWrQMAzJ49G97e3nj11Vfveb64uDhMnDgRAQEB+PHHH7Fp0yb07dsXSqXyrn2/+uor+Pj4YPr06Th8+DAOHz4MOzs7DB48GFu3btXad8uWLRg4cKCmi+9OdnZ2sLOzw969e1FZWXnPuFQqFSZNmoSzZ8/is88+w65du/DWW29BKlV/HMfGxuL//u//MGjQIOzcuRPTpk3Dl19+eVcc3333HUJCQrB9+3ZMnToVycnJmDRpEh577DHs2LEDX3zxBc6cOYMPP/zw4ZVPRLoTiIhMYMuWLUKnTp00j6OiooSvv/5aa58RI0YIc+bM0TzetWuXEBYWJixatEgIDw8Xrl+/rnnu+PHjQlBQkFBQUCAIgiC8+eabwujRo+9b/rhx44SPPvpI87hv377CmjVrtPa5cOGC0KZNGyEzM1MQBEHIyckR2rZtK5w4ceK+592zZ4/QpUsXISwsTBg1apSwePFiIS4uTvP8oUOHhJCQECExMfGex7/55pvCCy+8oLXtk08+EQYNGqQV69SpU7X2mTVrlvCf//xHa9upU6eEkJAQoby8/L7xEpF+2OJERCZXXFyMrKwsdOzYUWt7x44dce3aNc3jJ554AgMGDMA333yDGTNmoGXLlvc9Z1xcHCIjI+sVV/v27REYGIjt27cDAHbs2IGmTZuiS5cu9z1m4MCBOHToEL7++mv06tULJ0+exPDhwzUtRnFxcfDx8UGrVq3ueXxiYuI96yEpKUmrtSw0NFRrn8uXL2Pr1q2IiIjQ/PvXv/4FlUqF1NTUulw+Ed0DB4cTkcUoKytDbGwsZDIZkpKSHrivjY2NQcp89tlnsXHjRkyePBlbt27F8OHDH3rXnrW1NXr27ImePXvi1VdfxXvvvYevvvoKw4cPN1hctra2Wo9LS0sxevRojB8//q59mzRpYpAyiYhjnIjIDDg4OMDLywtnz57V2n727FkEBgZqHi9cuBBSqRSrVq3C+vXrcezYsfueMzg4+IHP/5NcLodKpbpr+5NPPon09HSsW7cOCQkJGDZsmM7nrBUYGKgZpB0cHIyMjAxcv379nvv6+/vfsx5atmwJmUx23zLatm2LhIQEtGjR4q5/vOOOyHCYOBGRWXjppZewatUq7Nq1C4mJiVi0aBEuX76MCRMmAAD279+PLVu2YNGiRejZsydeeuklzJw5U3OX3D9NnjwZMTExmDNnDi5fvoxr165h06ZNyMvLu+f+vr6+OHXqFDIzM7X2cXZ2xoABA/Dpp5+iZ8+e8PHxue813Lp1CxMmTMCvv/6Ky5cvIyUlBbt378a3336ruauua9eu6Ny5M6ZPn44jR44gJSUFBw4c0NxB+OKLL+LYsWNYvnw5rl+/jm3btmHjxo148cUXH1h/kyZNwrlz5zBv3jzExcXhxo0b2Lt3L+bNm/fA44hIP+yqIyKzMGHCBBQXF2PhwoXIy8tDQEAA/vvf/6Jly5bIy8vDe++9h9deew3t2rUDoJ4v6fDhw5g9ezaWLFly1/latWqF7777Dp//f3v3HdbU9f8B/A0IdYCoOFr3JC6mW1Fq60atgIoDEffWuhEn7r33KO5VV63W1aqtVrRqXVRAAVHcDNk7Ob8//OV+iaAmlgzs+/U8fSr3nuR+cpObvHPuuSfLl6Nbt24oWLAgbG1t0bFjx1y3P3r0aMyYMQOtWrVCRkYGQkJCpHVdu3bFiRMn4O7u/sHHUKRIEdjZ2WHHjh148uQJsrKy8OWXX6Jbt24YOnSo1G7NmjVYtGgRxo0bh9TUVFSqVAnjx48HANSpUwcrV67E6tWrsWHDBpQqVQqjR4+Gm5vbB7dds2ZN7Nq1CytXrkSvXr0AABUqVECHDh0+eDsi0gxnDici+ohjx45hwYIFuHTpEk97Ef3HsceJiOg9UlNTERUVhS1btqBHjx4MTUTEMU5ERO+zdetWtG/fHiVLlsTgwYP1XQ4RGQCeqiMiIiJSE3uciIiIiNTE4ERERESkJgYnIiIiIjUxOBERERGpicGJiIiISE0MTkRERERqYnAiIiIiUhODExEREZGaGJyIiIiI1MTgRERERKQmBiciIiIiNTE4EREREamJwYmIiIhITQxORERERGpicCIi0oKIiAj0798f9erVg0wmw6+//qrvkgzatWvXIJPJcO3aNX2XQvRBBfRdANF/3cOHD7Fp0yZcu3YNb968QbFixdCoUSMMHToUNWrU0Hd59Il8fHzw9OlTjB07FhYWFqhbt+4H2yclJWH79u04e/YsIiMjIZfLUbFiRTg7O8PLywtlypTRaPu///477t69i1GjRv2bh0FE7zASQgh9F0H0X3X27FmMGzcOxYoVg7u7O8qXL49nz57h0KFDiIuLw4oVK9C6dWt9l0kaSktLg52dHYYOHYqxY8d+tH1kZCS8vb3x4sULtGvXDvXq1YOpqSlCQkJw4sQJFCtWDGfOnNGohtmzZ2PPnj0ICQn51IehUwqFApmZmTA1NYWxMU+GkOFijxORnjx58gSTJk1ChQoVsGfPHpQoUUJa5+Xlhd69e2PSpEk4fvw4KlSooMdKc0pNTUWhQoVyLM/KyoJCoYCZmZkeqjIcsbGxAICiRYt+tG1WVhZGjhyJmJgY7Ny5E/Xr11dZP3bsWGzZskUrdRqC9PR0KSx98cUX+i6H6KMY64n0ZOvWrUhNTcWcOXNUQhMAlChRArNnz0ZKSkqOD81Xr17B19cXTk5OqFu3Lr755hvMnDkTGRkZUpuEhATMnz8f33zzDerWrYsWLVpg0qRJ0gf6kSNHIJPJ8PTpU5X7zm2cSZ8+fdCxY0cEBgaid+/esLOzw/Lly/H06VPIZDJs27YN27dvR6tWrWBjY4OwsDAAQFhYGEaPHo2GDRvCxsYGbm5u+O2331S2p6zj5s2bWLBgARo3bgx7e3uMGDFCqjW733//HZ6ennBwcICjoyPc3d3x888/q7S5c+cOBgwYgHr16sHOzg6enp64efOmSpukpCTMmzdP2j9NmjRBv3798M8//3zwOQOA+/fvY+DAgXB0dISDgwP69u2L27dvS+vXrFmDli1bAgAWL14MmUyGb7755r33d/bsWQQHB2Po0KE5QhMAmJubq/Ra3bhxA6NHj8bXX3+NunXrwtnZGfPnz0daWprUxsfHB3v27AEAyGQy6T8lhUKB7du3w8XFBTY2NmjatClmzJiB+Ph4lW0rFAqsWbMGTk5OsLOzQ58+fRAaGopvvvkGPj4+Km0jIyOl59vOzg7du3fHxYsXVdooX18nT57EihUr0Lx5c9jZ2SEpKem9Y5y0/XwSaYo9TkR6cuHCBZQrVy7XD0sAaNCgAcqVK4fff/9dWvbq1St07doViYmJ6N69O6pWrYpXr17hzJkzSEtLg5mZGZKTk9G7d2+EhYXB3d0dtWvXxps3b3D+/Hm8evUqR0hTR1xcHAYNGgQXFxd07twZVlZW0rojR44gPT0d3bt3h5mZGSwtLfHw4UP07NkTZcqUwaBBg1C4cGGcOnUKI0aMwJo1a3Kcfpw7dy6KFi2KkSNH4tmzZ9ixYwdmz56NlStXqmzH19cXNWrUwJAhQ2BhYYGgoCBcunQJnTp1AgAEBARg0KBBqFu3LkaOHAkjIyMcOXIEffv2xd69e2FrawsAmDlzJs6cOQNPT09Uq1YNcXFxuHnzJsLCwlCnTp337oeHDx+id+/eKFKkCAYOHIgCBQrgwIED6NOnD3bv3g07Ozu0bt0aFhYWWLBgATp27IgWLVqgSJEi771PZZj87rvv1HouTp8+jbS0NPTs2RPFihXD3bt3sXv3brx8+RKrV68GAHh4eOD169f4888/sXjx4hz3MWPGDBw9ehRubm7o06cPnj59ij179uD+/fvYt28fTE1NAQDLli3D1q1b0bJlSzRv3hzBwcEYMGAA0tPTVe4vOjoaPXr0QGpqKvr06YPixYvj6NGjGDZsGFavXp3j+V6/fj1MTU0xYMAAZGRkSNt7l7afT6JPIohI5xISEoS1tbUYNmzYB9sNHTpUWFtbi8TERCGEEJMmTRI1a9YUd+/ezdFWoVAIIYRYtWqVsLa2FmfPnn1vm8OHDwtra2sRGRmpsv7q1avC2tpaXL16VVrm6ekprK2txb59+1TaRkZGCmtra+Ho6ChiYmJU1vXt21d07NhRpKenq2zbw8NDtGnTRlqmrMPb21uqTQgh5s+fL2rVqiUSEhKEEG/3l4ODg+jWrZtIS0vL9TEpFArRpk0b0b9/f5X7Sk1NFd98843o16+ftKxevXrCz88vx/75mOHDh4s6deqIJ0+eSMtevXolHBwcRO/evXPsm61bt370Prt06SLq1aundg2pqak5lm3atEnIZDLx7NkzaZmfn5+wtrbO0fb69evC2tpaHD9+XGX5H3/8obI8KipK1K5dWwwfPlyl3Zo1a4S1tbWYPHmytGzevHnC2tpaXL9+XVqWlJQkvvnmG9GyZUshl8uFEP97fX377bc5Hse7rz1dPJ9En4Kn6oj0IDk5GQA+2BORfX1ycjIUCgV+/fVXtGzZEjY2NjnaGhkZAXh76qdmzZq5DipXttGUmZkZ3Nzccl3Xpk0blV6suLg4XL16Fe3bt0dSUhJiY2MRGxuLN2/ewMnJCREREXj16pXKfXTv3l2ltvr160Mul+PZs2cAgD///BPJyckYPHhwjnEwytsFBQUhIiICnTp1wps3b6TtpqSkoEmTJrh+/ToUCgWAt2OP7ty5k6OOD5HL5fjzzz/RqlUrlTFnpUuXRseOHXHz5k0kJSWpfX9KSUlJH30dZFewYEHp3ykpKYiNjYWDgwOEELh///5Hb3/69GlYWFigWbNm0j6KjY1FnTp1ULhwYelUWUBAALKystCrVy+V23t6eua4z99//x22trYqvadFihSBh4cHnj17htDQUJX2Xbp0UXkcudH280n0qXiqjkgPsgeiD8kesGJjY5GUlPTRKQqePHmCNm3a5E2h/69MmTLvHfBdvnz5HNsXQmDVqlVYtWpVrreJiYlRuby+bNmyKuuVg6oTEhKk+wTwwcceEREBAJg8efJ72yQmJsLS0hITJkyAj48Pvv76a9SpUwfOzs7o0qXLBwfhx8bGIjU1FVWqVMmxrlq1alAoFHjx4oXGU0iYm5sjMjJS7fbPnz/H6tWrcf78+RxjktQJbo8fP0ZiYiKaNGmS6/qYmBhpOwBQsWJFlfXFihWDpaVljprs7Oxy3FfVqlWl9dbW1tLyd18zudH280n0qRiciPTAwsICpUqV+uil4iEhIShTpgzMzc1VBv/+W+/reVJ+g3/Xh3oH3l2nvI/+/fujefPmud7m3Q/j911+LjSYLUXZdtKkSahVq1aubQoXLgwA6NChA+rXr49z587hzz//xLZt27BlyxasWbMGzs7Oam8zL1StWhX379/Hixcv8NVXX32wrVwuR79+/RAfH4+BAweiatWqKFy4MF69egUfH5/3Pn/ZKRQKWFlZYenSpbmu/5QxcJr6WG8TkH+fT/r8MTgR6UnLli1x8OBB3LhxI9cB4jdu3MCzZ8/g4eEB4O0Hmrm5OR4+fPjB+61YseJH2yh7dBITE1WWK0+N/RvKb/mmpqZo2rTpv74/4H9B6+HDh6hUqdIHt2tubq7WdkuXLo3evXujd+/eiImJgaurKzZu3PjeD9oSJUqgUKFCePToUY514eHhMDY2/mjwyU3Lli1x4sQJHD9+HEOGDPlg2wcPHiAiIgKLFi1Cly5dpOV//vlnjrbvC8cVK1ZEQEAAHB0dPxhglL2AT548Uem5efPmTY6errJly753v2S/L01o+/kk+lQc40SkJwMGDEDBggUxc+ZMvHnzRmVdXFwcZs6ciUKFCmHgwIEA3vbKtGrVChcuXMC9e/dy3J/yG3qbNm0QHByMc+fOvbeNMohcv35dWieXy3Hw4MF//bisrKzQsGFDHDhwAK9fv86xPrdpBj7GyckJRYoUwaZNm3Jc0aV8THXr1kXFihXxww8/5HoKVLlduVyeIzBaWVmhdOnSKlM6vMvExATNmjXDb7/9pjKNQ3R0NE6cOIF69erB3Nxc48fWtm1bWFtbY+PGjbh161aO9UlJSVixYgWA//XMZe+JE0Jg586dOW6nnGdLebpTqX379pDL5Vi/fn2O22RlZUntmzRpggIFCmDfvn0qbZTTHGTn7OyMu3fvqtSfkpKCgwcPoly5cqhevXruD/4DtP18En0q9jgR6UnlypWxcOFCTJw4EZ06dULXrl1VZg5/8+YNli9frnJaa9y4cfjzzz/Rp08fdO/eHdWqVUNUVBROnz6NvXv3omjRohgwYADOnDmDMWPGwN3dHXXq1EF8fDzOnz8PPz8/1KxZEzVq1IC9vT2WL1+O+Ph4WFpa4pdffkFWVlaePLaZM2eiV69e6NSpE7p3744KFSogOjoat2/fxsuXL3H8+HGN7s/c3BxTpkzBtGnT0LVrV3Ts2BFFixZFcHAw0tLSsGjRIhgbG2Pu3LkYNGgQOnbsCDc3N5QpUwavXr3CtWvXYG5ujo0bNyI5ORnOzs5o27YtatasicKFC+PKlSu4d+9ejrmJ3vX999/jypUr6NWrF3r16gUTExMcOHAAGRkZmDhx4iftK1NTU6xduxb9+vWDp6cn2rVrB0dHR5iamuLhw4c4ceIEihYtirFjx6Jq1aqoWLEiFi1ahFevXsHc3BxnzpzJEY4ASJfhz507F05OTjAxMYGLiwsaNmwIDw8PbNq0CUFBQWjWrBlMTU0RERGB06dPY+rUqWjXrh1KliwJLy8v/PDDDxg6dCiaN2+OkJAQ/PHHHyhevLhKj9bgwYNx8uRJDBo0CH369IGlpSWOHTuGp0+fYs2aNZ80E7gunk+iT8HgRKRH7du3R9WqVbF582bpZ1aUv1U3ZMgQlQG1wNtB2gcPHsSqVavw888/IykpCWXKlEGLFi2k0y5FihTBnj17sGbNGpw7dw5Hjx6FlZUVmjRpojIge+nSpZgxYwY2b96MokWLomvXrmjUqBH69ev3rx9X9erVcfjwYaxduxZHjx5FXFwcSpQogdq1a2PEiBGfdJ/dunWDlZUVNm/ejPXr16NAgQKoWrUqvL29pTaNGjXCgQMHsH79euzevRspKSkoVaoUbG1tpVOeBQsWRM+ePfHnn3/i7NmzEEKgYsWKUtj7kBo1amDPnj1YtmwZNm3aBCEEbG1tsWTJklwHR6urUqVKOHbsGLZv345z587ht99+g0KhQKVKldCtWzf06dMHwNuQtXHjRsydOxebNm3CF198gdatW6N379455oFq06YN+vTpg5MnT+L48eMQQsDFxQXA259jqVu3Lvbv348VK1bAxMQE5cqVQ+fOneHo6Cjdx4QJE1CwYEH8+OOPCAgIgL29PbZt24ZevXqpXCxQsmRJ7N+/H0uWLMHu3buRnp4OmUyGjRs34uuvv/7k/aLt55PoU/C36oiISG0JCQlo0KABvv/+ewwbNkzf5RDpHMc4ERFRrnK7knPHjh0AgIYNG+q6HCKDwFN1RESUq19++QVHjx5FixYtULhwYfz99984ceIEnJycUK9ePX2XR6QXDE5ERJQrmUwGExMTbN26FcnJybCysoKXlxe+//57fZdGpDcc40RERESkJo5xIiIiIlITgxMRERGRmjjGKY8oFApkZWXB2Nj4k3+BnoiIiHRPCAGFQoECBQp8dMJWBqc8kpWVlevPYBAREVH+YGNjozK5a24YnPKIMqHa2NjAxMREz9UQERGRuuRyOe7du6fWzwMxOOUR5ek5ExMTBiciIqJ8SJ2hNhwcTkRERKQmBiciIiIiNTE4EREREamJwYmIiIhITQxORERERGpicCIiIiJSE4MTERERkZoYnIiIiIjUxOBEREREpCYGJyIiIiI1MTgRERERqYnBiYiIiEhNDE5EREREatJrcLp+/TqGDh0KJycnyGQy/Prrr9K6zMxMLFmyBJ06dYK9vT2cnJwwadIkvHr1SuU+4uLiMH78eDg6OqJ+/frw9fVFcnKySpvg4GD06tULNjY2cHZ2xpYtW3LUcurUKbRr1w42Njbo1KkTfv/9d+08aCIiIsq39BqcUlJSIJPJMHPmzBzr0tLScP/+fQwbNgxHjhzB2rVr8ejRIwwbNkyl3YQJExAaGgp/f39s3LgRN27cwIwZM6T1SUlJGDBgAMqWLYsjR45g0qRJWLt2LQ4cOCC1+fvvvzF+/Hh07doVx44dw7fffosRI0bgwYMH2nvwRERElP8IA2FtbS3OnTv3wTZ37twR1tbW4tmzZ0IIIUJDQ4W1tbW4e/eu1Ob3338XMplMvHz5UgghxJ49e0SDBg1Eenq61GbJkiWibdu20t9jxowRgwcPVtlWt27dxPTp09WuPysrS9y4cUNkZWWpfRsiIiLSP00+wwvoO7hpIikpCUZGRihatCgA4NatWyhatChsbGykNk2bNoWxsTHu3r2L1q1b4/bt26hfvz7MzMykNk5OTtiyZQvi4+NhaWmJ27dvw9vbW2VbTk5OKqcO1SWXyz/twdEHGRsBRsYmeq1BKORQCL2WQEREWqDJZ3e+CU7p6elYunQpXFxcYG5uDgCIjo5GiRIlVNoVKFAAlpaWiIqKktqUL19epU3JkiWldZaWloiOjpaWKVlZWSE6OlrjOu/du6fxbejDChUqhNq1a+PuxcVIiovUSw3mxSrA9utJCLl/H6mpqXqpgYiI9C9fBKfMzEyMGTMGQgj4+fnpu5wPsrGxgYmJfntGPldJcZFIjAnTaw0ymUyv2yciorwnl8vV7vgw+OCUmZmJ77//Hs+fP8eOHTuk3ibgbc9RbGysSvusrCzEx8ejVKlSUpt3e46Ufyt7mXJrExMTk6MXSh0mJiYMTp8xPrdERP9tBj2PkzI0PX78GNu3b0fx4sVV1js4OCAhIQGBgYHSsqtXr0KhUMDW1hYAYG9vjxs3biAzM1Nqc+XKFVSpUgWWlpZSm6tXr6rc95UrV2Bvb6+lR0ZERET5kV6DU3JyMoKCghAUFAQAePr0KYKCgvD8+XNkZmZi9OjRCAwMxNKlSyGXyxEVFYWoqChkZGQAAKpVq4bmzZtj+vTpuHv3Lm7evIk5c+bAxcUFZcqUAQB06tQJpqammDp1Kh4+fIhffvkFO3fuRL9+/aQ6vLy8cOnSJfzwww8ICwvDmjVrEBgYCE9PT93vFCIiIjJYRkIIvV0ndO3aNXh5eeVY7urqipEjR+Lbb7/N9XY7d+5Eo0aNALydAHPOnDk4f/48jI2N0aZNG0ybNg1FihSR2gcHB2P27Nm4d+8eihcvDk9PTwwePFjlPk+dOoWVK1fi2bNnqFy5MiZOnAhnZ2e1H4tcLsft27dhb2/P0zlacuXYKL2NcbKwqoamXdboZdtERKRdmnyG6zU4fU4YnLSPwYmIiLRBk89wgx7jRERERGRIGJyIiIiI1MTgRERERKQmBiciIiIiNTE4EREREamJwYmIiIhITQxORERERGpicCIiIiJSE4MTERERkZoYnIiIiIjUxOBEREREpCYGJyIiIiI1MTgRERERqYnBiYiIiEhNDE5EREREamJwIiIiIlITgxMRERGRmhiciIiIiNTE4ERERESkJgYnIiIiIjUxOBERERGpicGJiIiISE0MTkRERERqYnAiIiIiUhODExEREZGaGJyIiIiI1MTgRERERKQmBiciIiIiNTE4EREREamJwYmIiIhITQxORERERGpicCIiIiJSE4MTERERkZoYnIiIiIjUxOBEREREpCYGJyIiIiI1MTgRERERqYnBiYiIiEhNDE5EREREamJwIiIiIlITgxMRERGRmhiciIiIiNTE4ERERESkJgYnIiIiIjUxOBERERGpicGJiIiISE0MTkRERERqYnAiIiIiUhODExEREZGaGJyIiIiI1MTgRERERKQmBiciIiIiNek1OF2/fh1Dhw6Fk5MTZDIZfv31V5X1QgisWrUKTk5OsLW1hbe3NyIiIlTaxMXFYfz48XB0dET9+vXh6+uL5ORklTbBwcHo1asXbGxs4OzsjC1btuSo5dSpU2jXrh1sbGzQqVMn/P7773n+eImIiCh/02twSklJgUwmw8yZM3Ndv2XLFuzatQuzZs3CwYMHUahQIQwYMADp6elSmwkTJiA0NBT+/v7YuHEjbty4gRkzZkjrk5KSMGDAAJQtWxZHjhzBpEmTsHbtWhw4cEBq8/fff2P8+PHo2rUrjh07hm+//RYjRozAgwcPtPfgiYiIKN/Ra3BydnbG2LFj0bp16xzrhBDYuXMnhg0bhlatWqFmzZpYvHgxXr9+LfVMhYWF4dKlS5g7dy7s7OxQv359TJs2DSdPnsSrV68AAMePH0dmZibmz5+PGjVqwMXFBX369IG/v7+0rZ07d6J58+YYOHAgqlWrhu+//x61a9fG7t27dbMjiIiIKF8w2DFOT58+RVRUFJo2bSots7CwgJ2dHW7dugUAuHXrFooWLQobGxupTdOmTWFsbIy7d+8CAG7fvo369evDzMxMauPk5IRHjx4hPj5eatOkSROV7Ts5OeH27dvaenhERESUDxXQdwHvExUVBQCwsrJSWW5lZYXo6GgAQHR0NEqUKKGyvkCBArC0tJRuHx0djfLly6u0KVmypLTO0tIS0dHR0rLctqMJuVyu8W3o40xMTPRdAgA+v0REnyNN3tsNNjjlV/fu3dN3CZ+dQoUKoXbt2vouAwAQEhKC1NRUfZdBRER6YrDBqVSpUgCAmJgYlC5dWloeExODmjVrAnjbcxQbG6tyu6ysLMTHx0u3L1myZI6eI+Xfyl6m3NrExMTk6IVSh42NjcH0jlDek8lk+i6BiIjymFwuV7vjw2CDU/ny5VGqVCkEBASgVq1aAN5eIXfnzh307NkTAODg4ICEhAQEBgaibt26AICrV69CoVDA1tYWAGBvb4+VK1ciMzMTpqamAIArV66gSpUqsLS0lNpcvXoV3t7e0vavXLkCe3t7jes2MTFhcPqM8bklIvpv0+vg8OTkZAQFBSEoKAjA2wHhQUFBeP78OYyMjODl5YUNGzbgt99+Q0hICCZNmoTSpUujVatWAIBq1aqhefPmmD59Ou7evYubN29izpw5cHFxQZkyZQAAnTp1gqmpKaZOnYqHDx/il19+wc6dO9GvXz+pDi8vL1y6dAk//PADwsLCsGbNGgQGBsLT01P3O4WIiIgMlpEQQuhr49euXYOXl1eO5a6urli4cCGEEFi9ejUOHjyIhIQE1KtXDzNnzkSVKlWktnFxcZgzZw7Onz8PY2NjtGnTBtOmTUORIkWkNsHBwZg9ezbu3buH4sWLw9PTE4MHD1bZ5qlTp7By5Uo8e/YMlStXxsSJE+Hs7Kz2Y5HL5bh9+zbs7e3ZK6ElV46NQmJMmF62bWFVDU27rNHLtomISLs0+QzXa3D6nDA4aR+DExERaYMmn+EGO48TERERkaFhcCIiIiJSE4MTERERkZoYnIiIiIjUxOBEREREpCYGJyIiIiI1MTgRERERqYnBiYiIiEhNDE5EREREamJwIiIiIlITgxMRERGRmhiciIiIiNTE4ERERESkJgYnIiIiIjUxOBERERGpicGJiIiISE0MTkRERERqYnAiIiIiUhODExEREZGaGJy0SCgU+i7BIGogIiL6XBTQdwGfMyNjYyT8ehnyNwl62b5J8aIo2spJL9smIiL6HDE4aZn8TQKyomP1XQYRERHlAZ6qIyIiIlITgxMRERGRmhicyCAGkBtCDURERB/DMU4EI2NjRJ9ZhczYp3rZvmmJ8ijZdoxetk1ERKQJBicCAGTGPkVm1CN9l0FERGTQeKqOKA8pFHJ9l2AQNRARfa7Y40SUh4yNTfDb5YWIi4/Uy/aLWVbAt04+etk2EdF/gcbB6Z9//kGBAgUgk8kAAL/++iuOHDmC6tWrY+TIkTAzM8vzIonyk7j4SETHhuq7DCIi0gKNT9XNmDEDERERAIDIyEiMGzcOhQoVwunTp7FkyZK8ro+IiIjIYGgcnCIiIlCrVi0AwKlTp9CgQQMsW7YMCxYswNmzZ/O8QCIiIiJDoXFwEkJA8f9z7gQEBKBFixYAgK+++gpv3rzJ2+qIiIiIDIjGwalu3brYsGEDjh07huvXr+Prr78GADx9+hQlS5bM6/qIiIiIDIbGwcnX1xf379/HnDlzMHToUFSqVAkAcObMGTg4OOR5gURERESGQuOr6mrWrImff/45x/JJkybB2JjTQhEREdHn65OSTkJCAn788UcsW7YMcXFxAIDQ0FDExsbmZW1EREREBkXjHqfg4GB4e3ujaNGiePbsGbp3745ixYrh7NmzePHiBRYvXqyNOomIiIj0TuMep4ULF8LNzQ1nz55VmezS2dkZN27cyNPiiIiIiAyJxsHp3r176NGjR47lZcqUQVRUVJ4URURERGSINA5OZmZmSEpKyrE8IiICJUqUyJOiiIiIiAyRxsHpm2++wbp165CZmSkte/78OZYuXYo2bdrkaXFEREREhkTj4OTj44OUlBQ0bdoU6enp6NOnD9q0aYMiRYpg7Nix2qiRiIiIyCBofFWdhYUF/P39cfPmTQQHByMlJQV16tRB06ZNtVEfERERkcHQODgp1atXD/Xq1cvLWoiIiIgMmsan6ubOnYudO3fmWL57927MmzcvT4oiIiIiMkQaB6czZ87A0dExx3IHBwecOXMmT4oiIiIiMkQaB6e4uDhYWFjkWG5ubo43b97kSVFEREREhkjj4FSpUiVcunQpx/I//vgDFSpUyJOiiIiIiAyRxoPDvb29MWfOHMTGxqJx48YAgICAAPj7+8PX1zfPCyQiIiIyFBoHp65duyIjIwMbN27E+vXrAQDlypXDrFmz0KVLl7yuj4iIiMhgfNJ0BL169UKvXr0QGxuLL774AkWKFMnruoiIiIgMjsZjnLIrUaKEVkOTXC7HypUr8c0338DW1hatWrXCunXrIISQ2gghsGrVKjg5OcHW1hbe3t6IiIhQuZ+4uDiMHz8ejo6OqF+/Pnx9fZGcnKzSJjg4GL169YKNjQ2cnZ2xZcsWrT0uIiIiyp807nGKjo7GokWLEBAQgNjYWJUQAwBBQUF5VtyWLVuwb98+LFq0CNWrV0dgYCCmTJkCCwsLeHl5SW127dqFhQsXonz58li1ahUGDBiAX375BV988QUAYMKECYiKioK/vz8yMzPh6+uLGTNmYNmyZQCApKQkDBgwAE2aNIGfnx8ePHgAX19fFC1aFB4eHnn2eIiIiCh/0zg4+fj44MWLFxg+fDhKly6tjZokt27dwrfffouvv/4aAFC+fHmcPHkSd+/eBfC2t2nnzp0YNmwYWrVqBQBYvHgxmjZtil9//RUuLi4ICwvDpUuXcOjQIdjY2AAApk2bhsGDB2PSpEkoU6YMjh8/jszMTMyfPx9mZmaoUaMGgoKC4O/vz+BEREREEo2D082bN7F3717UqlVLG/WocHBwwMGDB/Ho0SNUqVIFwcHBuHnzJnx8fAAAT58+RVRUlMrv5FlYWMDOzg63bt2Ci4sLbt26haJFi0qhCQCaNm0KY2Nj3L17F61bt8bt27dRv359mJmZSW2cnJywZcsWxMfHw9LSUuuPlYiIiAyfxsHpq6++ynF6TlsGDx6MpKQktG/fHiYmJpDL5Rg7diw6d+4MAIiKigIAWFlZqdzOysoK0dHRAN6eWixRooTK+gIFCsDS0lK6fXR0NMqXL6/SpmTJktI6TYKTXC6X/m1iYqL27bQpe025yQ915ocagfxTJxER/Y8m75kaBydfX18sW7YMfn5+OcJGXjt16hR+/vlnLFu2DNWrV0dQUBAWLFiA0qVLw9XVVavb/lT37t0DABQqVAi1a9fWczVvhYSEIDU1Ndd1+aHO/FAjkH/qJCKiT6dxcBo7dixSU1PRunVrFCxYEKampirr//rrrzwrbvHixRg8eDBcXFwAADKZDM+fP8emTZvg6uqKUqVKAQBiYmJUxlvFxMSgZs2aAN72HMXGxqrcb1ZWFuLj46XblyxZUuqhUlL+rex5UpeNjY3B9DooyWQyfZeglvxQZ36oEcg/dRIRGQK5XC51fHzMJ/U46UpaWhqMjIxUlpmYmEinCsuXL49SpUohICBAGnOVlJSEO3fuoGfPngDejpNKSEhAYGAg6tatCwC4evUqFAoFbG1tAQD29vZYuXIlMjMzpSB45coVVKlSRePxTSYmJgYXnAytnvfJD3XmhxqB/FMnEVF+o3Fw0uUpspYtW2Ljxo0oW7asdKrO398f7u7uAAAjIyN4eXlhw4YNqFSpkjQdQenSpaWr7KpVq4bmzZtj+vTp8PPzQ2ZmJubMmQMXFxeUKVMGANCpUyesW7cOU6dOxaBBg/Dw4UPs3LkTU6ZM0dljJSIiIsP3STOHP3nyBIcPH0ZkZCSmTp0KKysr/P777yhbtixq1KiRZ8VNmzYNq1atgp+fn3Q6zsPDAyNGjJDaDBo0CKmpqZgxYwYSEhJQr149bN26VZrDCQCWLl2KOXPmoG/fvjA2NkabNm0wbdo0ab2FhQW2bduG2bNnw83NDcWLF8fw4cM5FQERERGp0Dg4/fXXXxg0aBAcHR1x/fp1jB07FlZWVggJCcHhw4exevXqPCvO3NwcU6dOxdSpU9/bxsjICGPGjMGYMWPe26ZYsWLSZJfvU7NmTezdu/eTayUiIqLPn8Y/ubJs2TJ8//338Pf3VxkY3rhxY9y+fTsvayMiIiIyKBoHpwcPHkjjh7IrUaIE3rx5kydFERERERkijYOThYWFNHFkdkFBQdJgayIiIqLPkcbBycXFBUuXLkVUVBSMjIygUChw8+ZNLFq0CF26dNFCiURERESGQePgNHbsWFStWhVff/01UlJS4OLiAk9PTzg4OGDYsGHaqJGIiIjIIGh0VZ0QAtHR0Zg2bRpGjBiBBw8eIDk5GbVr10blypW1VCIRERGRYdA4OLVp0wYnTpxA5cqV8dVXX2mrLiIiIiKDo9GpOmNjY1SqVAlxcXFaKoeIiIjIcGk8xmn8+PFYvHgxHjx4oI16iIiIiAyWxjOHT548Gampqfjuu+9gamqKggULqqz/66+/8qw4IiIiIkOicXDy9fXVRh1EREREBk+j4JSZmYm//voLw4cPR4UKFbRVExEREZFB0miMk6mpKc6ePautWoiIiIgMmsaDw1u1aoXffvtNG7UQERERGTSNxzhVqlQJ69atw99//406deqgUKFCKuu9vLzyrDgiIiIiQ6JxcDp06BAsLCwQGBiIwMBAlXVGRkYMTkRERPTZ0jg4nT9/Xht1EBERERk8jcc4EREREf1XadzjNGXKlA+uX7BgwScXQ0RERGTINA5OCQkJKn9nZWXh4cOHSEhIQOPGjfOsMCIiIiJDo3FwWrduXY5lCoUCs2bN4qSYRERE9FnLkzFOxsbG8Pb2xo4dO/Li7oiIiIgMUp4NDo+MjERWVlZe3R0RERGRwdH4VN27g7+FEIiKisLFixfh6uqaZ4URERERGRqNg9P9+/dV/jY2NkaJEiXg4+MDd3f3PCuMiIiIyNBoHJx27dqljTqIiIiIDJ7GY5wiIyMRERGRY3lERASePn2aFzURERERGSSNg9OUKVNw69atHMvv3Lnz0ckxiYiIiPIzjYPT/fv34ejomGO5vb09goKC8qQoIiIiIkOkcXAyMjJCcnJyjuWJiYmQy+V5UhQRERGRIdI4ODVo0ACbNm1SCUlyuRybN29GvXr18rQ4IiIiIkOi8VV1EyZMQO/evdGuXTvUr18fAHDjxg0kJSVx5nAiIiL6rGnc41S9enUcP34c7du3R0xMDJKTk/Hdd9/h1KlTsLa21kaNRERERAZB4x4nAChTpgzGjRuX17UQERERGTSNe5wOHz6MU6dO5Vh+6tQpHD16NE+KIiIiIjJEGgenzZs3o3jx4jmWW1lZYePGjXlSFBEREZEh0jg4PX/+HOXLl8+xvGzZsnjx4kWeFEVERERkiDQOTlZWVggJCcmxPDg4GMWKFcuLmoiIiIgMksaDw11cXDBv3jwUKVIEDRo0AAD89ddfmD9/PlxcXPK8QCIiIiJDoXFwGjNmDJ49ewZvb28UKPD25gqFAt999x3Gjh2b5wUSERERGQqNg5OZmRlWrlyJR48eITg4GAULFoS1tTXKlSunjfqIiIiIDMYnzeMEAJaWlmjUqBFKlCiRl/UQERERGSyNglNCQgJWrFiBX375BQkJCQCAokWLwsXFBd9//z2KFi2qlSKJiIiIDIHawSkuLg4eHh54/fo1OnXqhKpVqwIAwsLCcPToUQQEBGD//v2wtLTUWrFERERE+qR2cFq3bh3MzMxw7tw5lCxZUmXd6NGj0b9/f6xbtw6+vr55XiQRERGRIVB7HqfffvsNkyZNyhGaAKBUqVKYOHEifv311zwtjoiIiMiQqB2cXr9+jRo1arx3vbW1NaKiovKkKCIiIiJDpHZwKl68OJ4+ffre9U+fPuX4JiIiIvqsqR2cnJycsHLlSmRkZORYl5GRgVWrVqF58+Z5WhwRERGRIVF7cPiYMWPg7u6Otm3bolevXqhatSqEEAgPD8fevXuRkZGBxYsXa7NWIiIiIr1SOzh9+eWX2L9/P/z8/LB8+XIIIQAARkZGaNq0KWbMmIGvvvpKa4USERER6ZtGE2BWqFABW7duRXx8PB4/fgwAqFixIooVK6aN2oiIiIgMitpjnLKztLSEra0tbG1ttR6aXr16hQkTJqBRo0awtbVFp06dcO/ePWm9EAKrVq2Ck5MTbG1t4e3tjYiICJX7iIuLw/jx4+Ho6Ij69evD19cXycnJKm2Cg4PRq1cv2NjYwNnZGVu2bNHq4yIiIqL855OCk67Ex8ejZ8+eMDU1xZYtW3Dy5ElMnjxZ5eq9LVu2YNeuXZg1axYOHjyIQoUKYcCAAUhPT5faTJgwAaGhofD398fGjRtx48YNzJgxQ1qflJSEAQMGoGzZsjhy5AgmTZqEtWvX4sCBAzp9vERERGTYPvlHfnVhy5Yt+PLLL7FgwQJpWYUKFaR/CyGwc+dODBs2DK1atQIALF68GE2bNsWvv/4KFxcXhIWF4dKlSzh06BBsbGwAANOmTcPgwYMxadIklClTBsePH0dmZibmz58PMzMz1KhRA0FBQfD394eHh4duHzQREREZLIPucTp//jzq1q2L0aNHo0mTJujSpQsOHjworX/69CmioqLQtGlTaZmFhQXs7Oxw69YtAMCtW7dQtGhRKTQBQNOmTWFsbIy7d+8CAG7fvo369evDzMxMauPk5IRHjx4hPj5e2w+TiIiI8gm1epxcXV2xfft2WFpaYu3atRgwYAAKFSqk7doQGRmJffv2oV+/fhg6dCju3buHuXPnwtTUFK6urtJM5VZWViq3s7KyQnR0NAAgOjoaJUqUUFlfoEABWFpaSrePjo5G+fLlVdoof1omOjpao4k95XK59G8TExO1b6dN2WvKTX6oMz/UCOSfOomI6H80ec9UKziFhYUhNTUVlpaWWLduHXr27KmT4CSEQN26dTFu3DgAQO3atfHw4UPs378frq6uWt/+p1AOXC9UqBBq166t52reCgkJQWpqaq7r8kOd+aFGIP/USUREn06t4FSrVi1MmTIF9erVgxAC27ZtQ+HChXNtO3LkyDwrrlSpUqhWrZrKsqpVq+LMmTPSegCIiYlB6dKlpTYxMTGoWbMmgLc9R7GxsSr3kZWVhfj4eOn2JUuWlHqolJR/5/ajxh9iY2NjML0OSjKZTN8lqCU/1JkfagTyT51ERIZALperXLH/IWoFpwULFmDNmjW4cOECjIyMcOnSpVzDgZGRUZ4GJ0dHRzx69EhlWUREBMqVKwcAKF++PEqVKoWAgADUqlULwNsr5O7cuYOePXsCABwcHJCQkIDAwEDUrVsXAHD16lUoFArY2toCAOzt7bFy5UpkZmbC1NQUAHDlyhVUqVJF49/fMzExMbjgZGj1vE9+qDM/1AjknzqJiPIbtYJT1apVsWLFCgBAzZo1sX379hzjirShb9++6NmzJzZu3Ij27dvj7t27OHjwIGbPng3gbVDz8vLChg0bUKlSJZQvXx6rVq1C6dKlpavsqlWrhubNm2P69Onw8/NDZmYm5syZAxcXF5QpUwYA0KlTJ6xbtw5Tp07FoEGD8PDhQ+zcuRNTpkzR+mMkIiKi/EPj6QiCg4O1UUeubG1tsXbtWixfvhzr1q1D+fLl4evri86dO0ttBg0ahNTUVMyYMQMJCQmoV68etm7dii+++EJqs3TpUsyZMwd9+/aFsbEx2rRpg2nTpknrLSwssG3bNsyePRtubm4oXrw4hg8fzqkIiIiISMUnzeP05MkT7NixA2FhYQCA6tWrw8vLCxUrVszT4gCgZcuWaNmy5XvXGxkZYcyYMRgzZsx72xQrVgzLli374HZq1qyJvXv3fnKdRERE9PnTeB6nS5cuoUOHDrh79y5kMhlkMhnu3LkDFxcX/Pnnn9qokYiIiMggaNzjtGzZMnh7e2PChAkqy5cuXYqlS5eiWbNmeVYcERERkSHRuMcpLCwMXbt2zbHc3d0doaGheVIUERERkSHSODiVKFECQUFBOZYHBQXp5Eo7IiIiIn3R+FRdt27dMGPGDERGRsLR0REA8Pfff2PLli3w9vbO6/qIiIiIDIbGwWnEiBEwNzfHDz/8gOXLlwMASpcujZEjR8LLyyvPCyQiIiIyFBoHJyMjI3h7e8Pb2xtJSUkAAHNz8zwvjIiIiMjQfNI8TkoMTERERPRfovHgcCIiIqL/KgYnIiIiIjUxOBERERGpSaPglJmZib59+yIiIkJL5RAREREZLo2Ck6mpKUJCQrRVCxEREZFB0/hUXefOnXHo0CFt1EJERERk0DSejkAul2Pfvn24cuUK6tati0KFCqmsnzJlSp4VR0RERGRINA5ODx48QO3atQEAjx49UllnZGSUN1URERERGSCNg9OuXbu0UQcRERGRwfvk6QgeP36MS5cuIS0tDQAghMizooiIiIgMkcY9Tm/evMH333+Pa9euwcjICGfPnkWFChXg6+sLS0tL+Pj4aKNOIiIiIr3TuMdpwYIFKFCgAC5evIiCBQtKyzt06IBLly7laXFEREREhkTjHqc///wT27Ztw5dffqmyvHLlynj+/HmeFUZERERkaDTucUpJSVHpaVKKi4uDmZlZnhRFREREZIg0Dk7169fHsWPHVJYpFAps3boVjRo1yqu6iIiIiAyOxqfqJk6cCG9vbwQGBiIzMxNLlixBaGgo4uPjsW/fPm3USERERGQQNA5O1tbWOHPmDHbv3o0iRYogJSUFrVu3Ru/evVG6dGlt1EhERERkEDQOTgBgYWGBYcOG5XUtRERERAbtk4JTfHw8Dh06hLCwMABA9erV4ebmhmLFiuVlbUREREQGRePB4devX8c333yDXbt2ISEhAQkJCdi1axe+/fZbXL9+XRs1EhERERkEjXucZs+ejQ4dOmDWrFkwMTEBAMjlcvj5+WH27Nn4+eef87xIIiIiIkOgcY/T48eP0a9fPyk0AYCJiQm8vb3x+PHjPC2OiP6bFEKh7xIMogYiMjwa9zjVrl0b4eHhqFq1qsry8PBw1KxZM88KI6L/LmMjYyy5fgKRiTF62X4FCytMbNBRL9smIsOmVnAKDg6W/u3l5YV58+bh8ePHsLOzAwDcuXMHe/bswYQJE7RTJRH950QmxiAs/rW+yyAiUqFWcOrSpQuMjIwghJCWLVmyJEe78ePHo0OHDnlXHREREZEBUSs4/fbbb9qug4iIiMjgqRWcypUrp+06iIiIiAzeJ02A+erVK9y8eROxsbFQKFSvPPHy8sqTwoiIiIgMjcbB6ciRI5gxYwZMTU1RvHhxlXVGRkYMTkRERPTZ0jg4rVq1CiNGjMCQIUNgbKzxNFBERERE+ZbGySctLQ0uLi4MTURERPSfo3H6cXd3x+nTp7VRCxEREZFB0/hU3fjx4zFkyBBcunQJ1tbWKFBA9S6mTJmSZ8URERERGRKNg9OmTZtw+fJlVKlSJcc6IyOjPCmKiIiIyBBpHJz8/f0xf/58uLm5aaMeIiIiIoOl8RgnMzMzODo6aqMWIiIiIoOmcXDy8vLC7t27tVELERERkUHT+FTd3bt3cfXqVVy4cAE1atTIMTh87dq1eVYcERERkSHRODgVLVoUbdq00UYtRERERAZN4+C0YMECbdRBREREZPA4/TcRERGRmjTucfrmm28+OF/Tb7/99q8KIiIiIjJUGgenvn37qvydlZWF+/fv4/LlyxgwYECeFUZERERkaP51cFLas2cPAgMD/3VBRERERIYqz8Y4tWjRAmfOnMmruyMiIiIyOHkWnE6fPo1ixYrl1d3lavPmzZDJZJg3b560LD09HX5+fmjUqBEcHBwwatQoREdHq9zu+fPnGDx4MOzs7NCkSRMsWrQIWVlZKm2uXbsGV1dX1K1bF61bt8aRI0e0+liIiIgo/9H4VF2XLl1UBocLIRAdHY3Y2FjMnDkzT4vL7u7du9i/fz9kMpnK8vnz5+P333/HypUrYWFhgTlz5mDkyJHYv38/AEAul2PIkCEoWbIk9u/fj9evX2Py5MkwNTXFuHHjAACRkZEYMmQIevTogaVLlyIgIADTpk1DqVKl0Lx5c609JiIiIspfNA5OrVq1UvnbyMgIJUqUQMOGDVGtWrU8Kyy75ORkTJw4EXPnzsWGDRuk5YmJiTh8+DCWLl2KJk2aAHgbpDp06IDbt2/D3t4ely9fRmhoKPz9/VGyZEnUqlULY8aMwdKlSzFy5EiYmZlh//79KF++PHx8fAAA1apVw82bN7F9+3YGJyIiIpJoHJxGjhypjTo+aPbs2XB2dkbTpk1VglNgYCAyMzPRtGlTaVm1atVQtmxZKTjdvn0b1tbWKFmypNTGyckJs2bNQmhoKGrXro3bt29LwSt7m/nz52tcq1wul/5tYmKi8e21IXtNuckPdeaHGoH8U6eh434kIl3S5FjXODjp2smTJ3H//n0cOnQox7ro6GiYmpqiaNGiKsutrKwQFRUltckemgBIf3+sTVJSEtLS0lCwYEG167137x4AoFChQqhdu7bat9OmkJAQpKam5rouP9SZH2oE8k+dho77kYgMmdrBqWbNmh+c+BJ4e9ru/v37/7oopRcvXmDevHn44Ycf8MUXX+TZ/WqTjY2NwXxbVnp3XJihyg915ocagfxTp6HjfiT6b5DL5VLHx8eoHZzWrl373nW3b9/Grl27oFAo1L07tfzzzz+IiYmBm5ubtEwul+P69evYs2cPtm3bhszMTCQkJKj0OsXExKBUqVIA3vYc3b17V+V+lVfdZW/z7pV40dHRMDc316i3CXh7isHQgpOh1fM++aHO/FAjkH/qNHTcj0T0LrWD07uDwgEgPDwcy5Ytw4ULF9CpUyeMHj06T4tr3Lgxfv75Z5VlU6ZMQdWqVTFo0CB89dVXMDU1RUBAANq2bSvV9Pz5c9jb2wMA7O3tsXHjRsTExMDKygoAcOXKFZibm6N69epSmz/++ENlO1euXJHug4iIiAj4xDFOr169wpo1a3Ds2DE4OTnh2LFjsLa2zuvaYG5unuN+CxcujGLFiknL3d3dsXDhQlhaWsLc3Bxz586Fg4ODFHqcnJxQvXp1TJo0CRMnTkRUVBRWrlyJ3r17w8zMDADQo0cP7NmzB4sXL4a7uzuuXr2KU6dOYdOmTXn+mIiIiCj/0ig4JSYmYuPGjdi9ezdq1aqF7du3o379+tqqTS2+vr4wNjbG6NGjkZGRAScnJ5X5pExMTLBx40bMmjULHh4eKFSoEFxdXVV6xypUqIBNmzZhwYIF2LlzJ7788kvMnTuXUxEQERGRCrWD05YtW7B161aULFkSy5Yty/XUnS7s2rVL5e8vvvgCM2fO/ODkm+XKlcOWLVs+eL+NGjXCsWPH8qJEIiIi+kypHZyWLVuGggULomLFijh27Nh7Q8aHBpETERER5WdqB6d3f2qFiIiI6L9G7eC0cOFCbdZBREREZPCM9V0AERERUX7B4ERERESkJgYnIiIiIjUxOBERERGpicGJiIiISE0MTkRERERqYnAiIiIiUhODExEREZGaGJyIiIiI1MTgRERERKQmBiciItI7hRD6LsEgaiDDp/Zv1REREWmLsZERdl6PwcvETL1s/0sLU3g1sNLLtil/YXAiIiKD8DIxE0/j9ROciNTFU3VEREREamJwIiIiIlITgxMRERGRmhiciIiIiNTE4ERERESkJgYnIiIiIjUxOBERERGpicGJiIiISE0MTkRERERqYnAiIiIiUhODExEREZGaGJyIiIiI1MTgRERERKQmBiciIiIiNTE4EREREamJwYmIiIhITQxORERERGpicCIiIiJSE4MTERERkZoYnIiIiIjUxOBEREREpCYGJyIiIiI1MTgRERERqYnBiYiIiEhNDE5EREREamJwIiIiIlITgxMRERGRmhiciIiIiNTE4ERERESkJgYnIiIiIjUxOBERERGpicGJiIiISE0MTkRERGoQQui7BIOo4b+ugL4LICIiyg+MjIxw52oikhKy9LJ986IFYNfYQi/bpv9hcCIi+owphICxkdF/voa8kpSQhYQ3cn2XQXrE4ERE9BkzNjLC8muBeJqYopftl7cojHGN6upl20TaYNDBadOmTTh79izCw8NRsGBBODg4YMKECahatarUJj09HQsXLsQvv/yCjIwMODk5YebMmShZsqTU5vnz55g1axauXbuGwoULo0uXLhg/fjwKFPjfw7927RoWLlyIhw8f4quvvsKwYcPg5uam08dLRKQNTxNTEB6XqO8yiD4LBj04/K+//kLv3r1x8OBB+Pv7IysrCwMGDEBKyv++Oc2fPx8XLlzAypUrsWvXLrx+/RojR46U1svlcgwZMgSZmZnYv38/Fi5ciKNHj2L16tVSm8jISAwZMgSNGjXCTz/9hL59+2LatGm4dOmSTh8vERERGTaD7nHatm2byt8LFy5EkyZN8M8//6BBgwZITEzE4cOHsXTpUjRp0gTA2yDVoUMH3L59G/b29rh8+TJCQ0Ph7++PkiVLolatWhgzZgyWLl2KkSNHwszMDPv370f58uXh4+MDAKhWrRpu3ryJ7du3o3nz5jp/3ERERGSYDDo4vSsx8W1Xs6WlJQAgMDAQmZmZaNq0qdSmWrVqKFu2rBScbt++DWtra5VTd05OTpg1axZCQ0NRu3Zt3L59Wwpe2dvMnz9f4xrl8v8NGjQxMdH49tqQvabc5Ic680ONQP6p09BxP+ad/LIv80Od+aFG+jSa7NN8E5wUCgXmz58PR0dHWFtbAwCio6NhamqKokWLqrS1srJCVFSU1CZ7aAIg/f2xNklJSUhLS0PBggXVrvPevXsAgEKFCqF27doaPELtCQkJQWpqaq7r8kOd+aFGIP/Uaei4H/NOftmX+aHO/FAj6Ua+CU5+fn54+PAh9u7dq+9SPsjGxsZgvpUoyWQyfZeglvxQZ36oEcg/dRo67se8k1/2ZX6oMz/UmN/I5XKp4+Nj8kVwmj17Ni5evIjdu3fjyy+/lJaXLFkSmZmZSEhIUOl1iomJQalSpaQ2d+/eVbm/6OhoAFBpo1yWvY25ublGvU3A265cQwtOhlbP++SHOvNDjUD+qdPQcT/mnfyyL/NDnfmhxs+ZQV9VJ4TA7Nmzce7cOezYsQMVKlRQWV+3bl2YmpoiICBAWhYeHo7nz5/D3t4eAGBvb48HDx4gJiZGanPlyhWYm5ujevXqUpurV6+q3PeVK1ek+yAiIiICDDw4+fn54fjx41i2bBmKFCmCqKgoREVFIS0tDQBgYWEBd3d3LFy4EFevXkVgYCB8fX3h4OAghR4nJydUr14dkyZNQnBwMC5duoSVK1eid+/eMDMzAwD06NEDkZGRWLx4McLCwrBnzx6cOnUK3t7eenrkREREZIgM+lTdvn37AAB9+vRRWb5gwQJpckpfX18YGxtj9OjRKhNgKpmYmGDjxo2YNWsWPDw8UKhQIbi6umL06NFSmwoVKmDTpk1YsGABdu7ciS+//BJz587lVARERESkwqCDU0hIyEfbfPHFF5g5c6ZKWHpXuXLlsGXLlg/eT6NGjXDs2DFNSyQiIqL/EIM+VUdERERkSBiciIiIiNTE4ERERESkJgYnIiIiIjUxOBERERGpicGJiIiISE0MTkRERERqYnAiIiIiUhODExEREZGaGJyIiIiI1MTgRERERKQmBiciIiIiNTE4EREREamJwYmIiIhITQxORERERGpicCIiIiJSE4MTERERkZoYnIiIiIjUxOBEREREpCYGJyIiIiI1MTgRERERqYnBiYiIiEhNDE5EREREamJwIiIiIlITgxMRERGRmhiciIiIiNTE4ERE9IkUQui7BIOogei/pIC+CyAiyq+MjYyw9NqfiEyM18v2K1hYYkKjZnrZNtF/FYMTEdG/EJkYj7C4N/oug4h0hKfqiIiIiNTE4ERERESkJgYnIiIiIjUxOBERERGpicGJiIiISE0MTkRERERqYnAiIiIiUhODExEREZGaGJyIiIiI1MTgRERERKQmBiciIiIiNTE4EREREamJwYmIiIhITQxORERERGpicCIiIiJSE4MTERERkZoYnIj+YxRCru8SDKIGos+RUAh9l2AQNWhTAX0XQES6ZWxkgk3XF+FFYqRetv+VRQUMaTBZL9sm+twZGRsh6lQcMmOz9LJ90xIFUKp9Mb1sW1cYnIj+g14kRuJxfKi+yyAiLciMzUJGlH6C038BT9URERERqYnBiYiIiEhNDE5EREREamJwIiIiIlITg9M79uzZg2+++QY2Njbo1q0b7t69q++SiIiIPiuGMGXBp9bAq+qy+eWXX7BgwQL4+fnBzs4OO3bswIABA3D69GlYWVnpuzwiIqLPgpGxEeJPP0RWbKpetl+gRCFYtqvxabfN41ryNX9/f3Tv3h3u7u4AAD8/P1y8eBGHDx/G4MGD9VwdERHR5yMrNhVZUcn6LkNjDE7/LyMjA//88w+GDBkiLTM2NkbTpk1x69atj95eCCHdj4mJCQC8/X/xojA2MtJO0R9TzAJyuRxy+YdnaTYxMYGJVSUojPXzcjApXu6jdZqYmKBIscqAkanuCsumiOXHawTe1lnMsjKM9FSnZVH19mU5i8ooAP3UWMZCvRorW5SEqZ5GE5SzKKH2813ZwhKm0M8xXs6iqFr7spJFIb292ZezKKT2vixrYQITKHRUmaoyFibqvQ8VNQL09J5exMJIrRpNrIxRwFg/x45JcWO1n29jq4IwNtbPKTvj4gVV6lT+X/lZ/iFGQp1W/wGvXr1CixYtsH//fjg4OEjLFy9ejOvXr+PHH3/84O0zMjJw7949bZdJREREWmJjYwMzM7MPtmGPUx4pUKAAbGxsYGxsDCN99TARERGRxoQQUCgUKFDg47GIwen/FS9eHCYmJoiJiVFZHhMTg5IlS3709sbGxh9NqURERJS/cTqC/2dmZoY6deogICBAWqZQKBAQEKBy6o6IiIj+u9jjlE2/fv0wefJk1K1bF7a2ttixYwdSU1Ph5uam79KIiIjIADA4ZdOhQwfExsZi9erViIqKQq1atbB161a1TtURERHR549X1RERERGpiWOciIiIiNTE4ERERESkJgYnIiIiIjUxOBERERGpicGJtOrVq1d48uSJvsvI954+fYrg4GB9l/HZSU3Vzy+zE5H2aPuaNwanfE6h0M8PYqrj/v37aNeuHR49eqTvUj4qJSXloz9KqS9BQUFwd3fHw4cP9V3KZ+XRo0cYNGgQYmJitP5GS/qVlpam7xJIh7T9fDM45WPh4eHYuXMn3rx5o+9ScggODkbv3r3h4eEBZ2dnfZfzQeHh4RgyZAiio6P1XUoOwcHB6NmzJ9zd3dGpUyd9l/NZ+e2335CWlgYrKyv+vuRn7J9//kHLli3x8uVLfZdCOhAYGAgnJye8ePFCa9vgBJj5VFxcHDp27AiFQoHU1FT07t0bRYsW1XdZAICwsDB4eXnBw8MDPj4+kMvlMDEx0XdZ7/X3338jMzMTZcqU0XcpKsLDw9G3b194eXlh3LhxUCgUMDbmd528kpaWhi+++ALA2659hqfPT3BwMPr06YPu3bvjyy+/1Hc5H2Tox3d+OEaCg4PRt29fuLm54auvvtLadgz3WaIPKlasGNq3b4+yZcti1apV2LBhA5KSkvRdFoKDg9GtWzdkZmbi5cuXSEpKgomJicGeBgPe/pBzVlaWvstQERwcDHd3d8THxyMoKEh6UzXUU7OGWte73q3T1NQUAPT6+szIyJD+nV9OGb558wZRUVH6LuODgoKC4OHhgT59+sDHx0daHhcXp7+isrlx4wZ++OEH7Nu3D69fvzbY4/vNmzeIi4uDkZGRQb8+w8LC0LNnT/Ts2RNTp07Vaq0MTvmQ8k2+cePGaN68OZYsWYIdO3Zg7dq1eg1PgYGB6NWrF3r16oUFCxbgxYsXmDhxokGGp/T0dJV/FypUSI/VqAoKCkKPHj3g5eWFn376CWFhYejXr5/BhafIyEicOHECAGBsbGzQb6rA23pHjhyJV69eAQCSk5NRuHBhAP+rX9ePITIyEvPmzcO9e/cAwOA/nAAgPj4eXl5e8Pf3l/aloQkLC0PXrl0xatQojB07Vlq+detWTJ8+Xe8XBZw7dw6DBg3C8ePH4e/vD09PTzx+/Nigjm8AiI2NxZgxY7Bx40bExsYa7OszODgYvXr1ghACTZs2hVwu12qtDE75kPK019dff41z585BoVBg69at2LlzJ9atW6eX8JScnAxvb2+4ublhwoQJaNWqFXr27InY2FhMmjTJoMLT8+fPMWvWLPz+++8A3vZCFC9eXFovhEBWVpZ00OnyjUy5H3v37o2xY8dCJpNh+fLliIyMRL9+/SCEMIg314yMDOzfvx/z5s3DsWPHAOSPD/2rV69i8uTJiI2NRVxcHMzNzQG8DU5GRkYqpyIyMzO1Xs+rV6/w888/Y8+ePQgKCgJg+PvR0tISbdq0wenTp/Hjjz8aXHhKT0/Hvn37IJfL8e2330rLN2/ejI0bN6Jnz556/aKUkpKCv//+GzNmzMDhw4exbNkyVK5cGd26dTO48FSiRAlYW1vj5s2b2LVrl0GGp/v376NHjx5wdXVFvXr1sHLlSly4cAFZWVlaq5W/VZdPhIWF4eLFi7Czs0P9+vWl5YcOHcLFixexatUqnDlzBuPHj4e3tzdGjBghfSho24sXL/DkyRNUrVoVpUqVQlZWFgoUKIDMzEycPHkS+/btg5WVFRYvXgxzc3O9j3m6du0aZs2ahRo1amDw4ME4ceIEkpKSMHfu3FzbZ2ZmSqd0tOnZs2cwMTGBmZkZSpQooTKm4M6dOxg7diwqVKiA7du3w8jISO9jIu7fv48jR47g8uXLGDx4MNzc3ADkHAthCGMjlPvq8ePH8Pb2hrW1NSwsLBAcHIyGDRsiLS0NlpaWkMvlSEpKQlZWFqpUqYJBgwahQAHtDAVV7pdr165hypQpqFevHvr3749atWqp1Kz8d2xsrN5/cDz7sbBp0ybs2bMH3bt3R7du3XKMEczMzMSNGzdgbW0NKysrndQXHh6OV69eoUiRIti6dSv+/vtvHD16FL/++itWrlyJZcuWwcnJSSe15Obu3bsYO3YsypUrBx8fH9SuXRsA8PjxY8ybNw+3b9/Gjz/+iEqVKun9+M6+/eXLl+Py5ctwdnZGnz59UKJECZX1mZmZuH79OmrVqqXyJVTbnj17ho4dO6JHjx6YPHkykpKSMGzYMKSnp2PIkCH4+uuvYWJikufvQQxO+UBcXBwaN24MALC1tUWRIkUwatQoVKxYEWlpaRg4cCAWLFgABwcH/PLLL/Dx8YGbmxsmTpyIIkWKaLW29PR0TJ8+HQ8ePMDYsWOlK+iU4SkrKwsnTpwwiPCU/eC5cuUKVq1aherVqyMoKAhJSUnSB6jyQBNCIDk5GWXLlsWUKVO0Wm9aWhqmTp2K27dvY8WKFbC1tYVcLpd6QoDcw5OuQ8mzZ8/g7++PadOmAQBCQkJw4MABXLlyJdfwlJmZiS1btqBq1apo166dzur8kIiICAwdOhQRERGwt7dHsWLFkJKSgsKFC8PIyAjp6ekoWLAgRo8ejZo1a2q1FuV+unr1Knx9feHo6IgBAwZI4Ql427u3YMECmJiYYNKkSTAzM9NqTbnJyMiAqakpYmNjUaJECek1t2HDBuzbty9HeMrIyMDs2bNx8eJF/PTTTzoJTsHBwejSpQumTp2KPn36ICQkBKtWrUJAQAAUCgX27duH2rVrqxwza9euRZ06ddCyZUut1we8Hde0atUq3LlzB0eOHEH16tVVQv3ChQtx4cIFnD17FhUrVtRJTR/ysfAkhEBmZibmzJmDv/76C3v37tXJc618Dm/duiWFJ+Vnjk7Ck6B8YcmSJUImk4nly5eLESNGiIEDB4ru3buLv/76S4waNUr0799fJCcnCyGE+Omnn0TDhg1FdHS0TmoLCAgQo0aNEp6enuLChQvS8szMTOn/R48eFT179hSenp4iKSlJJ3Vl9+TJE7FlyxYxefJkERsbK4QQ4tKlS6J79+7C3t5eNGvWTEybNk0MHDhQDBkyRIwbN04MHz5cDBkyRAQHB+ukxosXL4pRo0YJV1dXcefOHSGEEHK5XCgUCqnN7du3RatWrYSrq6vKcl1QKBTiwIEDwtnZWUyfPl1aHhwcLPz8/ETbtm3FoUOHpOVpaWli1qxZws7OTjx8+FCntQohxNOnT8W+ffvE9OnTxcyZM8XFixfFs2fPhBBCREREiA4dOoiePXuKuLi4HLeVy+U6q1P5PAYEBIiWLVuK8ePHi/v37wshhMjKyhJz5swRMplMBAYG6qym7EJDQ8XEiRPFd999J+rXry+8vb3F5s2bpfXr168XzZs3F2vWrBEvX74UQggxe/ZsYW9vL+7evauTGkNCQoStra1YvXq1yvKgoCAxceJE0aBBAxEaGiqEeLtPhRBizZo1Ot+vCoVC3LhxQ3Tr1k20atVKxMTESMuFECIsLEyMGTNGhIWF6aym3Cj3kRCqx8KyZcuEq6urWLlypVS7n5+fsLOz09lz/SHp6elCCCESExOFp6en6Natm/j111+lx5NX75kMTgbs5cuX0geoEELMnz9f1KtXT5w8eVLcvn1b7Nq1S7Rt21a0adNGNGzYULx48UJqm5iYqPX6sh9Q169fF8OHDxd9+vRRCU/KF2xmZqY4cOCA6Nevn0qduhAcHCxat24tZs6cKebNmydSUlKkdX/99Zfw8PAQI0eOVNnXuvLkyROxa9cu6e8rV66I4cOHiy5dukj1KBQKlQP+xo0bolOnTuLp06c6q1P5ekpISBC7d+8WnTt3FlOmTJHWh4SE5AhPc+bMEXZ2dnr5wA8KChItWrQQgwYNEm5ubqJz585CJpOJgQMHiuvXrwshhHj06JFo3ry58PT0FFFRUSr7WRuh9OXLl+LevXu5rsstPN25c0fMnTtX2Nrain/++SfP61FHcHCwaNCggZg+fbrYunWrOHjwoOjWrZuwt7cXo0ePltpt2LBBNG/eXKxevVpMmjRJ2Nra6ux5f/jwoWjYsKHw8vKSlmVkZEj/DgoKEsOHDxfNmjUTQUFBQgghli9fLurWrauTGqOiosSbN2+kUCmXy8WNGzdEz549RYcOHaQvuMrXQPbadenp06di3rx50t8fC08rVqwQPj4+ws7OTqevzydPngh/f38xbtw4MWHCBPHzzz+LR48eSevfDU89evQQv/zyi8rj+bcYnAxUenq6cHd3F97e3uLvv/+Wls+ZM0fY2tqKgwcPCiGEePbsmTh37pw4e/asTmtTyi08eXp6ij/++CNHm8zMTJ0EuuwePXokGjduLJYsWaJy4GT/9+XLl6XwdO3aNZXba7NXJzMzU6xZs0Y0btxYbNu2TVquTnhKS0vTWl3vunfvnmjQoIH0LTgxMVHs3LnzveHJxcVFuLu76y00PX78WDRt2lQsX75cpKSkSPttx44dwtnZWfTo0UP6dvzo0SPRunVr4erqKn2D1obk5GQxatQo4eHh8d6ArjxOAgICRJs2bUTjxo2Fvb293nqaXr9+LVxcXMTy5ctVlr98+VIsW7ZM1KlTR0ydOlVavnXrViGTyYS9vb3OPkiDgoKEvb29aNmypfD09JTeF4X4X4+3EG8D4PDhw0XLli3F5MmTha2t7XtDbF767bffRPfu3UW7du2Eq6ur+Omnn4QQ/+t56tmzp+jUqZN4/fq11mv5kPf1Jr/vPXPFihXC0dFRODg46DQ0BQUFCScnJzFs2DDRu3dv0atXLyGTyUTv3r3FxYsXpXbK8JmYmCi6dOkivL298/RMB4OTAfv7779F+/btxahRo8TNmzel5fPmzRO1a9cWBw8eVPl2ootTN0+ePBF9+vRRefPPHp5u3LghhgwZIoYOHapyekbXp5WEeHug+/n5iXHjxonU1NQc67PX9Oeff4pevXqJfv36qexrbXvy5IlYunSpaNu2rcrpj9zC07un7XQhKChIODo6ivnz5wsh/rfPEhMTxY4dO3KEpwcPHggfHx/h7OwsnW7StTVr1ohRo0aJzMzMHN8yDx48KBwcHMTChQul121oaKhOevBOnDghBg0aJAYOHChu3bqVaxvl/r1586bo1KmT1EOiD9evXxeurq7iyZMn0n5U/j86OlpMnz5dNG/eXOXLxsGDB6VTYtoWGBgo6tSpIzZs2CBevHghJk+eLDw8PMSBAwekNtnDU0hIiBg4cKDOwuj58+eFvb29+OGHH0RAQIBYsGCBkMlkUn0KhULcvHlTdOzYUXTv3l0vx7cQufcm+/r6Suvf1/O0d+9enT3XQrx9r2zWrJlYvny5ypf3n376SRq+cPXqVWm5su6kpKQ8P7YZnAyU8gV6584d0bp1azFq1Chx48YNaf38+fNFnTp1xOHDh1VOPWlbSkqKaNSokejatavKN43sB9Tly5dFixYtxMmTJ3VW1/u4urqKZcuW5bpOWbPyADt//rzo37+/Tk4lZt9fz549E+vXrxdff/212L17t7RcGZ66du2q8tzrSlBQkLC1tc3R46DsmXlfz1NoaKiIiorSaa3Z9e/fX4wZM0ZlWfb9PWvWLNGwYUOVHiZtnR55/vy52L9/v/T32bNnRf/+/VXCU/YPy/T0dLF//37x4sULvZ2yUdqxY4dwdHR87/oHDx6IOnXqqAQVXcnIyBDTpk0TCxYskJaFhoZ+NDwFBweLV69eab2+58+fi759+4odO3YIId720rVs2VJ89913QiaTSce5XC4Xt2/fFpGRkVqvKTcf6k3OLTylp6eLBQsWiDVr1ui81q1bt4rhw4eL9PT0HGOWzpw5Ixo0aCB8fX1FVlaWtDwvT89lx3mcDIjINm+Qci4PW1tbLF26FCEhIdixYwfu3r0LAJgyZQq8vLzg6+uLs2fP6qQ+uVyOQoUK4cKFC0hISMC0adNw//59AG/nwVHO0dSsWTNUqFABAQEBOqkrN5mZmUhOTsarV69QrFgxADlnZVbu47lz5yI+Ph4tW7bE2rVrtfrTDElJSUhLS4OxsbE0T1DZsmWRlJSEmJgY+Pv7Y+vWrQCAJk2awNPTE4ULF8bKlSuRnp6us/lTlBMIjhgxQmUCwQ0bNmD48OFISkqCubk5XF1d0bVrVwQHB2PMmDEAgGrVquntsvmMjAyYmJhIV88o97GxsbE0O3zr1q2hUChUfrtMG9NNxMXFYdu2bdi5cyf27t0rbbtHjx5QKBRYt24dbt++LdWakZGBefPmYenSpTqbAuNDypQpA4VCgRs3buQ6r1CNGjVQvnx5vH79GoBuZz03NTXF5MmTpRnBFQoFqlWrhiFDhqBy5co4cuQIDh48CADS1b0AIJPJULp0aa3XZ2JiAkdHR7Rv3x6vX79Gv3794OTkhB07dqB9+/aYM2cOduzYAWNjY9jZ2aF8+fJar+ldyp8ncXV1RdWqVSGEkI5pd3d3BAYGYurUqdLjSU1NxcKFC7Fnzx60atVK5/XeunULGRkZMDMzy3GFc5s2beDl5YWTJ0/i5cuX0jGlrSuhGZwMxNOnT+Hl5YWkpCTpAz17eFqyZAnu3buHXbt2ISEhAQAwadIkDB06FHXr1tVJjSYmJsjKykKhQoVw9OhRJCcnS+FJOb1AVlYW0tLSUKRIEVhbW+ukruyUk/GZmprC1NQUNWvWxLlz5xAZGSkdTNnf4CMjIxEUFCS9+RcsWFBrtb1+/RoDBw7Ezz//jOTkZOmDcfPmzTh06BBmz54NFxcXHDp0SCU8jRgxAkuWLMEXX3yhk6kHhBD46aefkJWVhWbNmknLN2/ejO3bt0tzhCnfaN3c3NChQwdERUVJ+1HXlB/sZmZmqFWrFs6dO4eIiAiYmppKgV75JpqamooSJUpo9bLpsLAwzJgxA61atUK9evVw/Phx7N69G0Du4QkAFi5cKM0kXaFCBa3Vpq7atWvD2NgYR44cUZkMVLmvo6KiYG5uDplMBgA6eW2Ghobi+PHjiIyMzDFPnRACVapUUQlPP/74IwBobS6ud7158wYpKSkoXbo0Bg8ejFKlSmHPnj0oX748JkyYAEtLS1SoUAFlypTB+vXrERcXp5fJJIODg+Hh4QFPT09MmTIFwNvnLzY2Vjqmu3btKoWnrKwsrF69GkeOHMH+/fu1Pk1HduL/JyQ2NTWV9pXy9WhkZKTyhd3ExEQ3E7JqpR+LNBYaGiqaNWsm3NzcpEFsytMLyv/fuHFD1KpVS6enwBITE6XxQcruT2XXd3JysmjTpo1wdXUVly9fFunp6SI9PV2sXr1atGzZUkREROisTiHensvu0qWL6Nu3r7Rs165d0jQOyqtaslu1apXw8vLK9ZJ0bRg8eLDo2LGjOHbsmBBCiG3btokGDRqIS5cuCSGEiIyMFEuWLBEuLi566Q6PjIwUCQkJIiUlRUyYMEHY29uLR48eiV27domGDRtKdb4rJSVFZ/swO+VYh+yn3R49eiRatWol2rZtK00/kN3cuXNF//79tXqhwqFDh0S3bt2keqZMmSK6d++ucgWl8rTd4MGDxZAhQ/Q2mF4IIWJjY0VoaKg4ffq0CA0NlY6VHTt2iJo1a4q5c+fmGDy/fPly0bp1a51dJZuYmCiaNm0qmjZtKqZNmyYmTJggXr58Kb1fZj8tEx4eLnx9fUX79u3F0aNHdVLfuXPnRI8ePUSbNm3E6tWrpedy+PDhYvz48VK7efPmicOHD+v8Qhml0NBQUadOHbFp0yaV5evXrxceHh5SXcrTdq6urqJp06Y6vVIyN9u3bxcymUw6xa0cE6b8XLpy5Ypo06aNePLkidZrYXDSs+xjL8LCwkTHjh1F586dpRfvuy+O3r17v3fMTl579eqV8PDwEAcPHvxgeOrevbvo1KmTaNu2rRg2bJho0aKFXgYGp6eni59++km0bt1aDB8+XFo+a9YsIZPJhJ+fn3TQBQUFiXnz5okGDRpofZ6myMhIsXv3bilIjh49Wnz33XdizJgxokGDBjmu5Hv69KmYPXu2cHd3F2/evNHZgNGMjAzh6ekpnJycRHx8vEhPTxfff/+9kMlkwsbGRuUKP6U1a9aoXMmkS2FhYcLX11e4uroKZ2dnMWzYMHH48GEhhBAnT54UzZs3F87OzuLMmTPiwYMH4v79+2LRokXC3t5e64OuN27cKFxdXaUP88ePH+cans6dOydcXV1Fw4YN9TblQEhIiOjWrZto27atqFOnjrCxsREdOnQQN27ckL4I1a5dW3Tv3l0sXrxYbN68WUyaNEnUq1dP5zXPnTtXeHp6iqtXr4oBAwaIbt26iXHjxonr16/nGBMWEhIiZs2apZPxQ4GBgaJevXpi7dq1Yu7cucLV1VWMHDlSBAYGih9//FHUqVNHrFy5Uvj4+IhGjRqpXD6vSwqFQixbtizH/FWbNm0SDRs2lK6GVh7jSUlJYsuWLcLd3V2vFyoI8fYLiJubm3Bycsr1dbdgwQLh6ekpEhIStF4Lg5MeKa+omjx5snQpZUhIiOjUqZNKeFJKS0sT3t7eYu/evTqrcfDgwaJTp07i6NGjOcKT8o0qPT1d/Pzzz2L9+vXixx9/1Enif5eyprS0NHHq1CnRsmVLMWzYMGn9okWLhKOjo6hTp45o1qyZaNeunejcubPWA15wcLBo06aNGDFihDh37py0fNy4cUImk4nFixfn6FkU4u3gUl1NYJpdSEiIcHNzEx07dhRxcXEiJSVF+Pn5iTp16kiX8CvrVE4gqI8PfOX8QlOnThWrV68WO3bsEO3atRMNGzYUfn5+Qoi3E4p6enoKmUwm6tatKzp27Cjc3Ny09gGQfYqItWvXCm9vbyHE//aXMjx5eHiohKc//vhDp3NyZffw4UPh6OgoFi1aJO7cuSNSUlLE7t27Rffu3UXdunXF77//LoQQ4sKFC8LDw0N8++23wtXVVfj4+Oh0UlPle821a9fE6NGjpd6vP/74Q6xevVrIZDIxefJkadC18otd9quvtOXx48di3bp1Yv369dKyCxcuiD59+ojhw4eLkydPii1btoiOHTuKPn366O1q0/zSm/yhnriTJ0+K9u3biwYNGogDBw6IO3fuiFu3bomFCxfq5AuREoOTngQFBYmWLVuKGTNmiIMHD6rMMRESEiI6duwoOnXqJCIjI0VmZqaQy+VixYoVolWrVlr/BpVbD4mLi4tKeHr3irSgoCC9XEqrfIPM3iuXPTxl73n666+/xMmTJ8WmTZtEQECA1udOCQ0NFQ0aNBBLly7N9TTh+PHjRbt27XLdr7qm3HdyuVyEhoYKDw8P4ebmJuLj40VqaqoYO3assLOzk67uW7Fihc4mEHzXq1evRLt27XK92m/69Omifv36Kutu3LghLl68KIKDg6VZ4/Pay5cvxejRo8Xly5eFEEKsXr1afP/990KIt8eI8nkNDw+XwlP26Sf0ITk5WQwcOFDMmTNHCKHak3jnzh3Rr18/YW9vLx48eCCEeBtCUlNTRXp6usqVatqk/LBXSkpKEp07dxYzZsyQlvn4+IjGjRuLGTNmSF+Kdu7cmWPuM21ITEwUbm5uokmTJmLp0qUq686fPy+8vLzE6NGjpeNG+esOupZfepNjYmJEo0aNVK5EFUL1fVH5CwvKL0QuLi7Cw8NDpz1iDE568PjxY9GsWTOxePFilRfqu6ftPDw8RKNGjYSXl5cYPny4aNGihda/3b+vh2TUqFGiQ4cO4ujRoyrTH6Snp4spU6YINzc3nXSRZvf06VOxfPlyKeC9G55Onz4tWrVqJXx8fHRal3L7o0ePlno/lDIyMkRkZKQU2qZPny7atGkjfvrpJ51OK5G9zuy1KS1cuFDIZDLRpUsXERcXJ9LT08W4ceNEgwYNxLhx44SdnZ1OJhDMzZUrV4SHh4d4/vy5FNyVtUdHR4sRI0aIli1b6uyncoR423vs4eEhBg4cKAIDA8WyZcvExIkTc22bnJwshg0bJoYMGaKXcWFKiYmJokOHDuL06dNCiLfHT/b3oCtXrojmzZuLefPmCYVCobVLu98nIyND9OnTRzRr1kzEx8er1NWjRw/x7Nkz4ePjI5o1ayZCQ0OFQqEQz58/Fz4+PjodX/nPP/+INm3aiB49ekghU+nChQviu+++E+PGjdNJ79eH5IfeZIVCIRYtWiTq1q0rnXZXevdLZVBQkLh69aoIDQ1VeX3oAoOTHixfvlwMGjRIrcGBmzZtEkuXLhX+/v5aPwX2sR6SsWPHirZt24pjx45JPSSzZ88WdnZ2Ov+5krS0NHHw4EHRsmVLsWDBAqkXLnt4SkpKErt37xYdO3Z874SD2pKZmSl69eqV45TMvHnzhKOjo2jRooUYOXKkEEKIadOmiaZNm+p83itlL0lAQIDK8s2bN4uGDRuKgwcPCldXV+mNNj09XYwZM0bY2NjodZDoli1bRMOGDXN8kCvfWB8+fJjrG6+2RUREiP79+4uRI0cKV1dX4erqKiZNmiR8fHzElClTxPjx48WECRPEtGnTxJAhQ3T+00NKyoAeGhoqZDKZuHLlihDif70N2b/MDR8+XPTo0UP3Rf6/kJAQ4e7uLjp06CCFzIiICOHp6Sm++eYb0apVq1x7S3QtKChIfPfdd2L69Ok5wtOlS5f0dipWiPzRm/zy5Utx8uRJcfLkSXHt2jXpooTsx/C78529u591idMR6JhCocDNmzdRpkyZHJfTKtcDby+ZBoDBgwdj/Pjx8Pb21uolyunp6Vi9ejU6duyI8ePHS79ynpmZiadPnyIqKgrLly9Hw4YNsX79epw6dQozZszAoUOHsHfvXtja2mqttnfdu3cPrq6uaNu2Lfr27Ytr165hx44d0pQDRkZGUCgUKFKkCFq3bo2XL18iIiJCZ/UBQFpaGmJjYxESEoLw8HBs2rQJ8+bNw8uXLzFmzBiMGjUK//zzD9auXYs5c+agefPmsLGx0WmNGRkZePXqFX744QfcvHkTwNspB7Zu3YoVK1agW7duWLJkCQoUKIC+ffsiJSUF8+fPx/nz51GnTh2d1ppdiRIlIJfLER4eDuB/00sof8W9bNmyKFWqFN68eaPTuipVqoSpU6ciLS0Njx49wrNnz1CwYEG8fv0ar169QkZGBpKSkvDixQtMmDBBq/OFvU9gYCA6d+6Mp0+fonjx4vjqq69w9uxZJCUlSdMJKI8fAChVqhQKFSqk8zqVz2n16tWxaNEiWFhYoF+/foiPj0elSpXQpk0bPHv2DNOnT5fee3QxHcL71KxZE/Pnz0dgYCB27NiB0NBQaZ2TkxPKlSun85rS09MBvN0vmZmZMDY2RrVq1eDg4IB//vkHffv2RXp6OhYuXIhvv/0Ww4YNw/jx47F9+3bs27dPZ8d4cHAwevfujTVr1mDixInw8/NDoUKFMHnyZPj6+uLYsWPS4wDevm/Nnj0b3333HRISEvQynQN7nHRE+e1YLpeL7t27i5kzZ6osf9e8efNUButp+9uUpj0kMplM579TJMTbb3YODg5i9uzZ0rJt27aJzp07i7lz56r0PGVlZYmEhATh5eUlDXLVpStXrojatWuLli1bCgcHB7Fv3z7pFEJGRobo37+/GDdunM7ryu7Ro0diwIABYtiwYWLatGmicePGOQaJhoaGipYtWwoPDw+9jMF68eKF+Omnn8SBAwdEbGysePbsmbC3t5eOISHeHlfK2l6+fCnc3d3F+fPndV6rEG97RAYPHiz69eun09OFH6M8dubOnSst8/X1FXZ2duLcuXO5nkoaM2aMWLRokRBCNz06Hzt1rOz9TEpKEl5eXuKHH37QWW3q+Oeff0TXrl3F2LFjdfpzJO/KL73JQUFBws7OTixevFi8fPlSnD9/XvTt21e4urqKoKAgsWLFCiGTyaQpJdLT04Wfn59wcHCQTi/qA4OTDkRGRor9+/dLT/TQoUNFmzZtcszXpBQRESFGjx6t08FuiYmJol27dmLatGkiLCxMbNy4UbRt21aMGjVK7NixQ/z444+iZcuW0txCfn5+IiQkRGf1CfG/g0w58Df7m+XBgwdF586dhZ+fnxRO5HK5WLlypWjVqpXeTos8f/5c3Lt3L8ccOHK5XIwePVqsWLFCJ4NYPyQ8PFz069dP2NraqvzYcPbXZXh4uF6ulnzw4IHo3LmzmDBhgsqYwA0bNohatWqJ+fPn59h3y5cvF23bts31dLOuhIeHi/79+4v+/fuL69evq6zTx3P9vmMnJiZGDB48WDRp0kQcOXJE+qmcpKQksXLlStGoUSMRHh6ukxrV+bBXjtGJj48XixcvFi1bttTbBRXvc+fOHeHp6amTn3Z5H+WYu0GDBkmn4JRTDvz5559CiLdfiLp06SK+++478ebNG5GcnKzTn0p6/vy5aNSokRg9erTK8v379wt7e3sRHh4uMjMzxcqVK4VMJhOHDx8WS5cu1ft8UkIwOGld9sHWv/32mxDi7Y/31qtXT4wYMUKlrfLNbNWqVaJ37946vxzdkHtIlAfZu79Btm3bNunDYOfOncLDw0N06dJF+Pj4iLFjx4rmzZvr7fLf90lPTxcrVqwQTk5OepvP5V2PHz+WfkMt+we9Pj+UHjx4IBo0aCBWrFihMh7w8uXLYteuXdKbaO/evcX69evF9u3bxZQpU4Sjo6Pe5kTK7tGjR2LIkCGie/fu4vbt23qr433Hjr+/v1i4cKEIDQ0V/fv3FzVr1hRff/216NWrl/D09BTNmzfX6QeUuh/2nTt3Fr169RL//POPcHd3F8+fP9dZjerK3nOmL4bemxwZGSnc3d3F0KFDVd5zLl++LBo2bCj11iYnJ0uD1d+df0pfGJy06H2DrVNTU8WWLVuEra2t9I00Li5O3Lx5U8ydO1c4OjrqbbIxQ+0hyX6QZX9TdXR0lN5UhXg7EHP16tVi0KBBYvny5dKPVxqKY8eOiTlz5oimTZsaxId7dso32v79++vlR4Wze/Pmjejdu7fKKVkh3j7nMplMDBkyRBw4cECcP39euLm5iRYtWohOnTqJCRMm6HXQ6LtCQ0PFqFGjcp3BXFfed+w4ODioTL564MABsXjxYuHj4yP27dunlx5GTT7s3d3ddX4lb35jyL3JQqi+54SGhoqkpCTRuHFjsXjxYpV2CQkJYteuXXo9/ZmdkRD6GFn1+UtPT8ekSZNgZWWFGTNmSMszMjKQkpKC8PBwhIaGYt++fQgKCkKBAgVQsWJFFC5cGHPnztXpbwF9TEZGBtavX4/Dhw9j165dqFy5sl7qiIiIwNy5c2FqaoqSJUvit99+w+LFi+Hk5CT9tp+SEEKvg0VzEx4ejpkzZ8LS0hJjx45FtWrV9F1SDhEREVi4cCHevHmDKVOmwN7eXi91hIWFYejQoZgzZw4aNmwIY2Nj7Nu3D/PmzYOPjw/Onz8PMzMzDBgwAA0aNEBiYiJMTEyk3yg0JMofJtUnTY4dfXv06BHmzJmDmzdvYsyYMejfvz8AqNQZHh4OMzMzvfw4bn7z5MkT+Pn5wdjYGEOGDEH9+vUBwGCe94iICMybNw9paWkICQlBly5d4OvrCwDSb6AChlMvAA4O15YPDba2t7cX7du3F3369BHJycniypUr4siRI+L+/fs5enr0zdB6SN73DSq/iI6ONvhvyYbQS3Ls2DFRq1YtlZ7NFy9eSF36ISEh0iBSQ+phMmQfOnYMZXC1kiGeOs7PDKk3OTePHj0SXl5eomXLluKvv/6Slhva61LJQOLb5+dDl6N///336NevH549e4aNGzeiSZMmcHV1Ra1atVCiRAl9ly4JDw/HoUOH8PLlS+zcuRO1a9fWd0moUqUKZs2ahfr16yMgIAA3btyQ1ol80HlqZWUFCwsLfZfxQdWqVcPSpUtRtmxZvdVQrlw5mJiY4Ny5cwDePrdffvkl6tevD4VCAWtra3To0AHGxsYoVqyY3urMTz507BiaihUrYvr06RBCYMOGDdJ0GQbT45DPVK5cGdOmTYOpqSkWL16M27dv67skFZUrV8bs2bNRtWpVbNy4UXq+De2sgRJfhVpibm6OGTNm4MiRIxg4cCA2bdoEb29vjB8/Hn379kWXLl1QuXJlvHr1St+lvlfVqlWxcuVKLFiwwKBOK73vTdVQD7L8SN+nlsqXLw8LCwscPXoUz549U3lulR+ejx49Qrly5fQyz1B+lZ+OHUP/sM9vKleujIkTJ6JMmTIoXbq0vsvJoVKlSvnm+eYYJy178eIFYmJiULZsWZXeJIVCgbFjx6JKlSoYM2YMAMN88zJkhjIeh7TjzJkzGD9+PDp06IDBgwejevXqAICkpCSsX78ehw4dwp49e1CjRg09V5r/5KdjJywsDKtWrYKPj49ee0E/F4Yw5u5D8sPzzeCkB4Yy2PpzkB8OMvo0crkcP/74I+bMmYOKFSvCwcEBBQoUwOvXrxEYGIjNmzcbxOnj/Co/HTuG/mFPecvQn28GJx376aefcO/ePZw6dQpbtmzhG38eMPSDjP6dO3fuYOvWrXjy5AmKFCmCevXqoWvXrqhUqZK+S8v3eOwQaY7BSYfyw+XoRIYo+2XJRET6xOCkYzExMTAzMzP4K6uIDInINi+XMMA5uojov4PBiYiIiEhNnI6AiIiISE0MTkRERERqYnAiIiIiUhODExEREZGaGJyIiIiI1MTgRERERKQmBiciIiIiNTE4EREREamJwYmI8p2oqCjMmTMH3377LerWrQtnZ2cMHToUAQEBat3+yJEjqF+/vparJKLPUQF9F0BEpImnT5+iZ8+eKFq0KCZNmgRra2tkZWXh8uXL8PPzw+nTp/VdosYyMzNhamqq7zKISA3scSKifMXPzw9GRkb48ccf0bZtW1SpUgU1atRAv379cPDgQQCAv78/OnXqBHt7ezg7O2PWrFlITk4GAFy7dg1TpkxBYmIiZDIZZDIZ1qxZAwDIyMjAokWL0Lx5c9jb26Nbt264du2ayvYPHjwIZ2dn2NnZYcSIEfD398/Re7V37160atUKdevWRdu2bXHs2DGV9TKZDHv37sXQoUNhb2+PDRs2oHXr1ti2bZtKu6CgIMhkMjx+/DgvdyER/QsMTkSUb8TFxeHSpUvo3bs3ChcunGN90aJFAQBGRkaYOnUqTpw4gYULF+Lq1atYsmQJAMDBwQG+vr4wNzfH5cuXcfnyZfTv3x8AMHv2bNy6dQsrVqzA8ePH0a5dOwwcOBAREREAgJs3b2LmzJnw8vLCsWPH0LRpU2zcuFGlhnPnzmH+/Pno168ffv75Z/To0QO+vr64evWqSru1a9eidevW+Pnnn9G1a1e4u7vjyJEjKm0OHz6MBg0aoFKlSnmy/4jo32NwIqJ848mTJxBCoGrVqh9s5+3tjcaNG6N8+fJo0qQJvv/+e5w6dQoAYGZmBgsLCxgZGaFUqVIoVaoUihQpgufPn+PIkSNYtWoV6tevj4oVK2LAgAGoV6+eFGh2796NFi1aYMCAAahSpQp69+6N5s2bq2x727ZtcHV1Re/evVGlShX069cPrVu3xg8//KDSrmPHjnB3d0eFChVQtmxZuLq64tGjR7h79y6At6fvTpw4AXd397zafUSUBzjGiYjyDSGEWu2uXLmCTZs2ITw8HElJSZDL5UhPT0dqaioKFSqU620ePHgAuVyOdu3aqSzPyMhAsWLFAACPHj1Cq1atVNbb2tri4sWL0t/h4eHw8PBQaePo6IidO3eqLKtbt67K32XKlIGzszMOHToEW1tbXLhwARkZGTnqISL9YnAionyjUqVKMDIyQnh4+HvbPH36FEOGDEHPnj0xduxYWFpa4ubNm5g6dSoyMzPfG5xSUlJgYmKCw4cPw8TERGVdbqcF/63c7rNbt26YNGkSfH19ceTIEXTo0OG99RKRfvBUHRHlG8WKFYOTkxP27NmDlJSUHOsTEhLwzz//QAgBHx8f2Nvbo0qVKnj9+rVKO1NTU8jlcpVltWrVglwuR2xsLCpVqqTyX6lSpQAAVapUQWBgoMrt7t27p/J31apV8ffff6ss+/vvv1G9evWPPj5nZ2cUKlQI+/btw6VLl3iajsgAMTgRUb4yc+ZMKBQKdOvWDWfOnEFERATCwsKwc+dOeHh4oFKlSsjMzMSuXbsQGRmJY8eOYf/+/Sr3Ua5cOaSkpCAgIACxsbFITU1FlSpV0KlTJ0yaNAlnz55FZGQk7t69i02bNkmn4jw9PfH777/D398fERER2L9/P/744w8YGRlJ9z1w4EAcPXoUe/fuRUREBPz9/XHu3DlpAPqHmJiYwM3NDcuWLUOlSpXg4OCQp/uOiP49I6HuoAEiIgPx+vVrbNy4ERcvpy6KswAAAY9JREFUXsTr169RokQJ1KlTB97e3mjUqBG2b9+OrVu3IjExEfXr10enTp0wefJkXL9+XbrybubMmTh9+jTi4uIwcuRIjBo1CpmZmdiwYQOOHTuG169fo1ixYrC3t8eoUaMgk8kAvJ2OYO3atYiPj4eTkxPq1q2LPXv24PLly1J9e/fuxQ8//ICXL1+iXLlyGDZsGLp06SKtl8lkWLduXY7xUgAQGRmJVq1aYeLEiRg4cKB2dyQRaYzBiYjoX5g2bRrCw8Oxd+/ePLm/GzduwNvbGxcvXkTJkiXz5D6JKO/wVB0RkQa2bduG4OBgPH78GLt27cKxY8fg6ur6r+83IyMDL1++xJo1a9C2bVuGJiIDxavqiIg0cPfuXWzduhXJycmoUKECpk6dim7duv3r+z1x4gSmTp2KWrVqYfHixXlQKRFpA0/VEREREamJp+qIiIiI1MTgRERERKQmBiciIiIiNTE4EREREamJwYmIiIhITQxORERERGpicCIiIiJSE4MTERERkZoYnIiIiIjU9H/Co5+Jj/dppQAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 600x1000 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from collections import Counter\n",
    "\n",
    "# Set the style of seaborn for better visualization\n",
    "sns.set_style(\"whitegrid\")\n",
    "\n",
    "# Create subplots\n",
    "fig, axes = plt.subplots(2, 1, figsize=(6, 10))\n",
    "\n",
    "# Plotting the histogram for toxicity scores on the first subplot\n",
    "sns.histplot(df_hs_eooh['toxicity'], bins=30, kde=True, ax=axes[0])\n",
    "axes[0].set_title('Distribution of Toxicity Scores')\n",
    "axes[0].set_xlabel('Toxicity Score')\n",
    "axes[0].set_ylabel('Number of Tweets')\n",
    "\n",
    "# Ensure all entries in 'categories' are strings\n",
    "df_hs_eooh['categories'] = df_hs_eooh['categories'].astype(str)\n",
    "\n",
    "# Splitting the categories and counting occurrences\n",
    "categories = df_hs_eooh['categories'].str.split().sum()\n",
    "category_counts = Counter(categories)\n",
    "\n",
    "# Plotting the top categories\n",
    "sns.barplot(x=list(category_counts.keys()), y=list(category_counts.values()), ax=axes[1])\n",
    "axes[1].set_title('Occurrences of Categories')\n",
    "axes[1].set_xlabel('Category')\n",
    "axes[1].set_ylabel('Number of Occurrences')\n",
    "axes[1].tick_params(axis='x', rotation=45)\n",
    "\n",
    "# Adjust the layout\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Evaluate EOOH annotations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "counts in labels: label\n",
      "1    2268\n",
      "0    1852\n",
      "Name: count, dtype: int64 counts in preds: toxicity\n",
      "0    3411\n",
      "1     709\n",
      "Name: count, dtype: int64\n",
      "acc: 0.514\n",
      "precision: 0.688\n",
      "recall: 0.215\n",
      "f1: 0.328\n",
      "confusion matrix: \n",
      "[[1631  221]\n",
      " [1780  488]]\n"
     ]
    }
   ],
   "source": [
    "# load EOOH annots\n",
    "df_hs = pd.read_csv(f'data/hs_{dataset}_eooh_clean.csv')\n",
    "# get test split\n",
    "df_hs = df_hs[df_hs['split'] == 'test']\n",
    "\n",
    "# get labels, preds\n",
    "labels = df_hs['label']\n",
    "# preds_eooh = df_hs['hate']\n",
    "toxic_thresh = 0.5\n",
    "preds_eooh = (df_hs['toxicity'] >= toxic_thresh).astype(int)\n",
    "# print value counts in labels, preds\n",
    "print(f\"counts in labels: {labels.value_counts()} counts in preds: {preds_eooh.value_counts()}\")\n",
    "\n",
    "# compute metrics\n",
    "metrics = ['acc', 'precision', 'recall', 'f1']  # list of metrics to compute\n",
    "metrics_dict = compute_metrics(labels, preds_eooh, metrics)  \n",
    "# print metrics\n",
    "for metric in metrics:\n",
    "    print(f\"{metric}: {metrics_dict[metric]:.3f}\")\n",
    "    \n",
    "# print confusion matrix\n",
    "from sklearn.metrics import confusion_matrix\n",
    "print(f\"confusion matrix: \\n{confusion_matrix(labels, preds_eooh)}\")\n",
    "\n",
    "# print some TNs, FPs, FNs, TPs\n",
    "tns = df_hs[(labels==0) & (preds_eooh==0)][['text']]\n",
    "fps = df_hs[(labels==0) & (preds_eooh==1)][['text']]\n",
    "fns = df_hs[(labels==1) & (preds_eooh==0)][['text']]\n",
    "tps = df_hs[(labels==1) & (preds_eooh==1)][['text']]\n",
    "# print(f'TNs: \\n{tns.sample(2)}\\nFPs: \\n{fps.sample(2)}\\nFNs: \\n{fns.sample(2)}\\nTPs: \\n{tps.sample(2)}')\n",
    "\n",
    "# create dataframe w samples of each\n",
    "num_samples = 20\n",
    "df_tn_samples = tns.sample(num_samples).assign(label=0, pred=0)\n",
    "df_fp_samples = fps.sample(num_samples).assign(label=0, pred=1)\n",
    "df_fn_samples = fns.sample(num_samples).assign(label=1, pred=0)\n",
    "df_tp_samples = tps.sample(num_samples).assign(label=1, pred=1)\n",
    "# concat samples into single df\n",
    "df_samples = pd.concat([df_tn_samples, df_fp_samples, df_fn_samples, df_tp_samples]).reset_index(drop=True)\n",
    "# save to csv\n",
    "samples_path = f'outputs/{dataset}/eooh/conf_samples.tsv'\n",
    "os.makedirs(os.path.dirname(samples_path), exist_ok=True)\n",
    "df_samples.to_csv(samples_path, sep='\\t', index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Compare our model preds w/ EOOH"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sim b/w lm preds & eooh preds: 0.475\n"
     ]
    }
   ],
   "source": [
    "# compute similarity b/w lm preds and eooh preds\n",
    "sim = np.sum(preds_lm == preds_eooh) / len(preds_lm)\n",
    "print(f\"sim b/w lm preds & eooh preds: {sim:.3f}\")\n",
    "\n",
    "# get samples where our lm is correct but eooh is wrong, and vice versa\n",
    "num_samples = 20\n",
    "lm_correct_indices = (labels == preds_lm)\n",
    "lm_wrong_indices = (labels != preds_lm)\n",
    "eooh_correct_indices = (labels == preds_eooh)\n",
    "eooh_wrong_indices = (labels != preds_eooh)\n",
    "\n",
    "# TN for lm, FP for eooh\n",
    "tn_lm_condition = (labels == 0) & lm_correct_indices & eooh_wrong_indices\n",
    "tn_lm_samples = df_hs[tn_lm_condition][['text']].sample(num_samples).assign(label=0, pred_lm=0, pred_eooh=1)\n",
    "# TP for lm, FN for eooh\n",
    "tp_lm_condition = (labels == 1) & lm_correct_indices & eooh_wrong_indices\n",
    "tp_lm_samples = df_hs[tp_lm_condition][['text']].sample(num_samples).assign(label=1, pred_lm=1, pred_eooh=0)\n",
    "# TN for eooh, FP for lm\n",
    "tn_eooh_condition = (labels == 0) & lm_wrong_indices & eooh_correct_indices\n",
    "tn_eooh_samples = df_hs[tn_eooh_condition][['text']].sample(num_samples).assign(label=0, pred_lm=1, pred_eooh=0)\n",
    "# TP for eooh, FN for lm\n",
    "tp_eooh_condition = (labels == 1) & lm_wrong_indices & eooh_correct_indices\n",
    "tp_eooh_samples = df_hs[tp_eooh_condition][['text']].sample(num_samples).assign(label=1, pred_lm=0, pred_eooh=1)\n",
    "\n",
    "# concat samples into single df\n",
    "df_disagreement_samples = pd.concat([tn_lm_samples, tp_lm_samples, tn_eooh_samples, tp_eooh_samples]).reset_index(drop=True)\n",
    "# save to csv\n",
    "disagreement_samples_path = f'outputs/{dataset}/disagreement_samples.tsv'\n",
    "os.makedirs(os.path.dirname(disagreement_samples_path), exist_ok=True)\n",
    "df_disagreement_samples.to_csv(disagreement_samples_path, sep='\\t', index=False)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "aiahs",
   "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": "3.11.4"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}