notebooks/LightGBM_training.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "unexpected-index",
   "metadata": {
    "id": "unexpected-index"
   },
   "source": [
    "# Training the LightGBM Model for Forecasting Fuel Load in Tropics\n",
    "This notebooks demonstrates training a LightGBM model with hyperparameter optimization, followed by feature importance visualization using SHAP. LightGBM is a gradient boosting framework that uses tree based learning algorithms. This notebook utilizes the `deepfuel-ML/src/models/lightgbm_module.py` script for model training."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "continent-budget",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-19T05:47:51.305070Z",
     "start_time": "2021-03-19T05:47:47.998598Z"
    },
    "id": "continent-budget"
   },
   "outputs": [],
   "source": [
    "import os\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from joblib import dump, load\n",
    "import shap"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "russian-irish",
   "metadata": {
    "id": "russian-irish"
   },
   "source": [
    "### Data directory"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "growing-publication",
   "metadata": {
    "id": "growing-publication",
    "outputId": "bb0a467e-76a8-4bb0-cf71-37b8b4634d88",
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "../data/tropics\r\n",
      "├── results\r\n",
      "│   ├── tropics_output_Aug.csv\r\n",
      "│   ├── tropics_output_Dec.csv\r\n",
      "│   ├── tropics_output_July.csv\r\n",
      "│   ├── tropics_output_Nov.csv\r\n",
      "│   ├── tropics_output_Oct.csv\r\n",
      "│   └── tropics_output_Sept.csv\r\n",
      "├── tropics_test.csv\r\n",
      "├── tropics_train.csv\r\n",
      "└── tropics_val.csv\r\n",
      "\r\n",
      "1 directory, 9 files\r\n"
     ]
    }
   ],
   "source": [
    "# The training, validation and test set required for model training are placed in data/tropics/\n",
    "! tree ../data/tropics"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "reverse-comparative",
   "metadata": {
    "id": "reverse-comparative"
   },
   "source": [
    "### Input Features\n",
    "\n",
    "- Latitude\n",
    "- Longitude\n",
    "- Fire Weather Index: fwinx\n",
    "- Drought Code: drtcode\n",
    "- Fire Danegr Severity Rating: fdsrte\n",
    "- Fraction of Burnable Area: fraction_of_burnable_area\n",
    "- d2m\n",
    "- Evaporation Rate: erate\n",
    "- fg10\n",
    "- si10\n",
    "- Volumetric Soil Water Level 1: swvl1\n",
    "- 2m Temperature: t2m\n",
    "- tprate\n",
    "- Climatic Region: climatic_region\n",
    "- Slope: slor\n",
    "- Month: month\n",
    "- Fuel Load: actual_load (target variable)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "earned-herald",
   "metadata": {
    "id": "earned-herald",
    "outputId": "3cf8beb7-7835-40e0-b6f5-93ad495921f4"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "latitude,longitude,fwinx,drtcode,fdsrte,fraction_of_burnable_area,d2m,erate,fg10,si10,swvl1,t2m,tprate,climatic_region,slor,actual_load,month\r\n"
     ]
    }
   ],
   "source": [
    "# Check header of training set matches input features\n",
    "! head -n 1 ../data/tropics/tropics_train.csv"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "intended-overall",
   "metadata": {
    "id": "intended-overall"
   },
   "source": [
    "### Model Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "provincial-purse",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-19T05:48:52.164268Z",
     "start_time": "2021-03-19T05:47:51.307074Z"
    },
    "id": "provincial-purse",
    "outputId": "0836d016-90bc-45f6-b283-d939a4b9d5ef",
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Link for the created Neptune experiment--------\n",
      "https://ui.neptune.ai/shared/step-by-step-monitoring-experiments-live/e/STEP-498\n",
      "---------------------------------------\n",
      "/home/smose_roshni/miniconda3/envs/ml-fuel-test/lib/python3.8/site-packages/lightgbm/engine.py:156: UserWarning: Found `early_stopping_rounds` in params. Will use it instead of argument\n",
      "  warnings.warn(\"Found `{}` in params. Will use it instead of argument\".format(alias))\n",
      "[LightGBM] [Warning] Unknown parameter: lamda_l2\n",
      "[LightGBM] [Warning] Unknown parameter: lamda_l1\n",
      "[LightGBM] [Warning] Unknown parameter: lamda_l2\n",
      "[LightGBM] [Warning] Unknown parameter: lamda_l1\n",
      "[LightGBM] [Warning] Auto-choosing row-wise multi-threading, the overhead of testing was 0.001602 seconds.\n",
      "You can set `force_row_wise=true` to remove the overhead.\n",
      "And if memory is not enough, you can set `force_col_wise=true`.\n",
      "[LightGBM] [Info] Total Bins 3279\n",
      "[LightGBM] [Info] Number of data points in the train set: 293980, number of used features: 16\n",
      "[LightGBM] [Warning] Unknown parameter: lamda_l2\n",
      "[LightGBM] [Warning] Unknown parameter: lamda_l1\n",
      "[LightGBM] [Info] Start training from score 1838268313.128554\n",
      "[1]\ttrain's rmse: 3.71086e+09\tval's rmse: 4.98509e+09\n",
      "Training until validation scores don't improve for 20 rounds\n",
      "[2]\ttrain's rmse: 3.69872e+09\tval's rmse: 4.97686e+09\n",
      "[3]\ttrain's rmse: 3.68677e+09\tval's rmse: 4.96807e+09\n",
      "[4]\ttrain's rmse: 3.67527e+09\tval's rmse: 4.96038e+09\n",
      "[5]\ttrain's rmse: 3.66333e+09\tval's rmse: 4.95125e+09\n",
      "[6]\ttrain's rmse: 3.65021e+09\tval's rmse: 4.93877e+09\n",
      "[7]\ttrain's rmse: 3.63756e+09\tval's rmse: 4.92753e+09\n",
      "[8]\ttrain's rmse: 3.62793e+09\tval's rmse: 4.92339e+09\n",
      "[9]\ttrain's rmse: 3.61553e+09\tval's rmse: 4.91262e+09\n",
      "[10]\ttrain's rmse: 3.60311e+09\tval's rmse: 4.90133e+09\n",
      "[11]\ttrain's rmse: 3.59199e+09\tval's rmse: 4.89393e+09\n",
      "[12]\ttrain's rmse: 3.58439e+09\tval's rmse: 4.89054e+09\n",
      "[13]\ttrain's rmse: 3.57523e+09\tval's rmse: 4.88666e+09\n",
      "[14]\ttrain's rmse: 3.56669e+09\tval's rmse: 4.8825e+09\n",
      "[15]\ttrain's rmse: 3.55629e+09\tval's rmse: 4.87507e+09\n",
      "[16]\ttrain's rmse: 3.54411e+09\tval's rmse: 4.86576e+09\n",
      "[17]\ttrain's rmse: 3.53588e+09\tval's rmse: 4.86247e+09\n",
      "[18]\ttrain's rmse: 3.52409e+09\tval's rmse: 4.85267e+09\n",
      "[19]\ttrain's rmse: 3.51381e+09\tval's rmse: 4.84565e+09\n",
      "[20]\ttrain's rmse: 3.50263e+09\tval's rmse: 4.83609e+09\n",
      "[21]\ttrain's rmse: 3.49249e+09\tval's rmse: 4.82885e+09\n",
      "[22]\ttrain's rmse: 3.48227e+09\tval's rmse: 4.82291e+09\n",
      "[23]\ttrain's rmse: 3.47495e+09\tval's rmse: 4.81974e+09\n",
      "[24]\ttrain's rmse: 3.4651e+09\tval's rmse: 4.81196e+09\n",
      "[25]\ttrain's rmse: 3.45387e+09\tval's rmse: 4.80182e+09\n",
      "[26]\ttrain's rmse: 3.444e+09\tval's rmse: 4.79458e+09\n",
      "[27]\ttrain's rmse: 3.43391e+09\tval's rmse: 4.78725e+09\n",
      "[28]\ttrain's rmse: 3.42592e+09\tval's rmse: 4.7833e+09\n",
      "[29]\ttrain's rmse: 3.41527e+09\tval's rmse: 4.7736e+09\n",
      "[30]\ttrain's rmse: 3.40589e+09\tval's rmse: 4.76741e+09\n",
      "[31]\ttrain's rmse: 3.39548e+09\tval's rmse: 4.76029e+09\n",
      "[32]\ttrain's rmse: 3.38494e+09\tval's rmse: 4.74902e+09\n",
      "[33]\ttrain's rmse: 3.37617e+09\tval's rmse: 4.74268e+09\n",
      "[34]\ttrain's rmse: 3.36848e+09\tval's rmse: 4.73919e+09\n",
      "[35]\ttrain's rmse: 3.36037e+09\tval's rmse: 4.73504e+09\n",
      "[36]\ttrain's rmse: 3.35031e+09\tval's rmse: 4.72685e+09\n",
      "[37]\ttrain's rmse: 3.34248e+09\tval's rmse: 4.72332e+09\n",
      "[38]\ttrain's rmse: 3.33497e+09\tval's rmse: 4.7203e+09\n",
      "[39]\ttrain's rmse: 3.32808e+09\tval's rmse: 4.71671e+09\n",
      "[40]\ttrain's rmse: 3.31841e+09\tval's rmse: 4.70906e+09\n",
      "[41]\ttrain's rmse: 3.30925e+09\tval's rmse: 4.70222e+09\n",
      "[42]\ttrain's rmse: 3.29966e+09\tval's rmse: 4.69442e+09\n",
      "[43]\ttrain's rmse: 3.2928e+09\tval's rmse: 4.68964e+09\n",
      "[44]\ttrain's rmse: 3.28561e+09\tval's rmse: 4.68602e+09\n",
      "[45]\ttrain's rmse: 3.276e+09\tval's rmse: 4.67953e+09\n",
      "[46]\ttrain's rmse: 3.26671e+09\tval's rmse: 4.67195e+09\n",
      "[47]\ttrain's rmse: 3.25814e+09\tval's rmse: 4.66603e+09\n",
      "[48]\ttrain's rmse: 3.24866e+09\tval's rmse: 4.65632e+09\n",
      "[49]\ttrain's rmse: 3.23948e+09\tval's rmse: 4.64912e+09\n",
      "[50]\ttrain's rmse: 3.23415e+09\tval's rmse: 4.64727e+09\n",
      "[51]\ttrain's rmse: 3.22755e+09\tval's rmse: 4.64416e+09\n",
      "[52]\ttrain's rmse: 3.21825e+09\tval's rmse: 4.63847e+09\n",
      "[53]\ttrain's rmse: 3.20914e+09\tval's rmse: 4.63267e+09\n",
      "[54]\ttrain's rmse: 3.2002e+09\tval's rmse: 4.62697e+09\n",
      "[55]\ttrain's rmse: 3.19136e+09\tval's rmse: 4.62162e+09\n",
      "[56]\ttrain's rmse: 3.18596e+09\tval's rmse: 4.61923e+09\n",
      "[57]\ttrain's rmse: 3.17731e+09\tval's rmse: 4.61394e+09\n",
      "[58]\ttrain's rmse: 3.1687e+09\tval's rmse: 4.60912e+09\n",
      "[59]\ttrain's rmse: 3.16028e+09\tval's rmse: 4.60405e+09\n",
      "[60]\ttrain's rmse: 3.15191e+09\tval's rmse: 4.59963e+09\n",
      "[61]\ttrain's rmse: 3.14541e+09\tval's rmse: 4.59394e+09\n",
      "[62]\ttrain's rmse: 3.13888e+09\tval's rmse: 4.59012e+09\n",
      "[63]\ttrain's rmse: 3.1312e+09\tval's rmse: 4.58621e+09\n",
      "[64]\ttrain's rmse: 3.12284e+09\tval's rmse: 4.58023e+09\n",
      "[65]\ttrain's rmse: 3.11638e+09\tval's rmse: 4.57452e+09\n",
      "[66]\ttrain's rmse: 3.11075e+09\tval's rmse: 4.57169e+09\n",
      "[67]\ttrain's rmse: 3.10451e+09\tval's rmse: 4.56805e+09\n",
      "[68]\ttrain's rmse: 3.09745e+09\tval's rmse: 4.56347e+09\n",
      "[69]\ttrain's rmse: 3.08955e+09\tval's rmse: 4.55797e+09\n",
      "[70]\ttrain's rmse: 3.08183e+09\tval's rmse: 4.55338e+09\n",
      "[71]\ttrain's rmse: 3.07411e+09\tval's rmse: 4.54857e+09\n",
      "[72]\ttrain's rmse: 3.06717e+09\tval's rmse: 4.54538e+09\n",
      "[73]\ttrain's rmse: 3.06136e+09\tval's rmse: 4.54193e+09\n",
      "[74]\ttrain's rmse: 3.05541e+09\tval's rmse: 4.54143e+09\n",
      "[75]\ttrain's rmse: 3.04777e+09\tval's rmse: 4.53634e+09\n",
      "[76]\ttrain's rmse: 3.04083e+09\tval's rmse: 4.53269e+09\n",
      "[77]\ttrain's rmse: 3.03415e+09\tval's rmse: 4.5304e+09\n",
      "[78]\ttrain's rmse: 3.02658e+09\tval's rmse: 4.52401e+09\n",
      "[79]\ttrain's rmse: 3.01923e+09\tval's rmse: 4.521e+09\n",
      "[80]\ttrain's rmse: 3.01213e+09\tval's rmse: 4.51799e+09\n",
      "[81]\ttrain's rmse: 3.00479e+09\tval's rmse: 4.51388e+09\n",
      "[82]\ttrain's rmse: 2.9973e+09\tval's rmse: 4.50856e+09\n",
      "[83]\ttrain's rmse: 2.99197e+09\tval's rmse: 4.50636e+09\n",
      "[84]\ttrain's rmse: 2.98602e+09\tval's rmse: 4.50275e+09\n",
      "[85]\ttrain's rmse: 2.97873e+09\tval's rmse: 4.49715e+09\n",
      "[86]\ttrain's rmse: 2.97166e+09\tval's rmse: 4.49234e+09\n",
      "[87]\ttrain's rmse: 2.9646e+09\tval's rmse: 4.48978e+09\n",
      "[88]\ttrain's rmse: 2.96062e+09\tval's rmse: 4.48854e+09\n",
      "[89]\ttrain's rmse: 2.95545e+09\tval's rmse: 4.48546e+09\n",
      "[90]\ttrain's rmse: 2.94932e+09\tval's rmse: 4.48028e+09\n",
      "[91]\ttrain's rmse: 2.94323e+09\tval's rmse: 4.47881e+09\n",
      "[92]\ttrain's rmse: 2.93624e+09\tval's rmse: 4.47308e+09\n",
      "[93]\ttrain's rmse: 2.93106e+09\tval's rmse: 4.47027e+09\n",
      "[94]\ttrain's rmse: 2.92618e+09\tval's rmse: 4.46739e+09\n",
      "[95]\ttrain's rmse: 2.92123e+09\tval's rmse: 4.46489e+09\n",
      "[96]\ttrain's rmse: 2.91455e+09\tval's rmse: 4.46071e+09\n",
      "[97]\ttrain's rmse: 2.91022e+09\tval's rmse: 4.45936e+09\n",
      "[98]\ttrain's rmse: 2.90409e+09\tval's rmse: 4.45557e+09\n",
      "[99]\ttrain's rmse: 2.89755e+09\tval's rmse: 4.4517e+09\n",
      "[100]\ttrain's rmse: 2.89299e+09\tval's rmse: 4.44929e+09\n",
      "[101]\ttrain's rmse: 2.8886e+09\tval's rmse: 4.4473e+09\n",
      "[102]\ttrain's rmse: 2.88291e+09\tval's rmse: 4.44464e+09\n",
      "[103]\ttrain's rmse: 2.87847e+09\tval's rmse: 4.44283e+09\n",
      "[104]\ttrain's rmse: 2.87247e+09\tval's rmse: 4.43975e+09\n",
      "[105]\ttrain's rmse: 2.86657e+09\tval's rmse: 4.43729e+09\n",
      "[106]\ttrain's rmse: 2.86073e+09\tval's rmse: 4.4363e+09\n",
      "[107]\ttrain's rmse: 2.85494e+09\tval's rmse: 4.43474e+09\n",
      "[108]\ttrain's rmse: 2.85047e+09\tval's rmse: 4.43138e+09\n",
      "[109]\ttrain's rmse: 2.84524e+09\tval's rmse: 4.43078e+09\n",
      "[110]\ttrain's rmse: 2.83915e+09\tval's rmse: 4.42591e+09\n",
      "[111]\ttrain's rmse: 2.83349e+09\tval's rmse: 4.4242e+09\n",
      "[112]\ttrain's rmse: 2.82821e+09\tval's rmse: 4.42208e+09\n",
      "[113]\ttrain's rmse: 2.82231e+09\tval's rmse: 4.41902e+09\n",
      "[114]\ttrain's rmse: 2.81647e+09\tval's rmse: 4.41494e+09\n",
      "[115]\ttrain's rmse: 2.8113e+09\tval's rmse: 4.41285e+09\n",
      "[116]\ttrain's rmse: 2.80542e+09\tval's rmse: 4.40817e+09\n",
      "[117]\ttrain's rmse: 2.80055e+09\tval's rmse: 4.40621e+09\n",
      "[118]\ttrain's rmse: 2.79474e+09\tval's rmse: 4.4014e+09\n",
      "[119]\ttrain's rmse: 2.7897e+09\tval's rmse: 4.39947e+09\n",
      "[120]\ttrain's rmse: 2.78397e+09\tval's rmse: 4.39798e+09\n",
      "[121]\ttrain's rmse: 2.77886e+09\tval's rmse: 4.39491e+09\n",
      "[122]\ttrain's rmse: 2.77345e+09\tval's rmse: 4.39091e+09\n",
      "[123]\ttrain's rmse: 2.76865e+09\tval's rmse: 4.38768e+09\n",
      "[124]\ttrain's rmse: 2.76459e+09\tval's rmse: 4.38618e+09\n",
      "[125]\ttrain's rmse: 2.75979e+09\tval's rmse: 4.38504e+09\n",
      "[126]\ttrain's rmse: 2.75469e+09\tval's rmse: 4.38209e+09\n",
      "[127]\ttrain's rmse: 2.7521e+09\tval's rmse: 4.38132e+09\n",
      "[128]\ttrain's rmse: 2.74671e+09\tval's rmse: 4.37997e+09\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[129]\ttrain's rmse: 2.7414e+09\tval's rmse: 4.37538e+09\n",
      "[130]\ttrain's rmse: 2.73756e+09\tval's rmse: 4.37348e+09\n",
      "[131]\ttrain's rmse: 2.73404e+09\tval's rmse: 4.37231e+09\n",
      "[132]\ttrain's rmse: 2.72914e+09\tval's rmse: 4.37209e+09\n",
      "[133]\ttrain's rmse: 2.72418e+09\tval's rmse: 4.36973e+09\n",
      "[134]\ttrain's rmse: 2.71895e+09\tval's rmse: 4.3671e+09\n",
      "[135]\ttrain's rmse: 2.71407e+09\tval's rmse: 4.36484e+09\n",
      "[136]\ttrain's rmse: 2.71161e+09\tval's rmse: 4.36397e+09\n",
      "[137]\ttrain's rmse: 2.70814e+09\tval's rmse: 4.36242e+09\n",
      "[138]\ttrain's rmse: 2.70314e+09\tval's rmse: 4.36016e+09\n",
      "[139]\ttrain's rmse: 2.69851e+09\tval's rmse: 4.35818e+09\n",
      "[140]\ttrain's rmse: 2.69428e+09\tval's rmse: 4.35555e+09\n",
      "[141]\ttrain's rmse: 2.69004e+09\tval's rmse: 4.35278e+09\n",
      "[142]\ttrain's rmse: 2.68584e+09\tval's rmse: 4.35022e+09\n",
      "[143]\ttrain's rmse: 2.6809e+09\tval's rmse: 4.34768e+09\n",
      "[144]\ttrain's rmse: 2.67584e+09\tval's rmse: 4.3442e+09\n",
      "[145]\ttrain's rmse: 2.6718e+09\tval's rmse: 4.34402e+09\n",
      "[146]\ttrain's rmse: 2.66791e+09\tval's rmse: 4.3417e+09\n",
      "[147]\ttrain's rmse: 2.6631e+09\tval's rmse: 4.33949e+09\n",
      "[148]\ttrain's rmse: 2.65836e+09\tval's rmse: 4.33531e+09\n",
      "[149]\ttrain's rmse: 2.65407e+09\tval's rmse: 4.33403e+09\n",
      "[150]\ttrain's rmse: 2.65105e+09\tval's rmse: 4.33273e+09\n",
      "[151]\ttrain's rmse: 2.64667e+09\tval's rmse: 4.32897e+09\n",
      "[152]\ttrain's rmse: 2.64249e+09\tval's rmse: 4.32757e+09\n",
      "[153]\ttrain's rmse: 2.63794e+09\tval's rmse: 4.32439e+09\n",
      "[154]\ttrain's rmse: 2.63372e+09\tval's rmse: 4.32295e+09\n",
      "[155]\ttrain's rmse: 2.62927e+09\tval's rmse: 4.31984e+09\n",
      "[156]\ttrain's rmse: 2.62492e+09\tval's rmse: 4.31921e+09\n",
      "[157]\ttrain's rmse: 2.62088e+09\tval's rmse: 4.31726e+09\n",
      "[158]\ttrain's rmse: 2.61678e+09\tval's rmse: 4.31727e+09\n",
      "[159]\ttrain's rmse: 2.61414e+09\tval's rmse: 4.31675e+09\n",
      "[160]\ttrain's rmse: 2.60952e+09\tval's rmse: 4.31566e+09\n",
      "[161]\ttrain's rmse: 2.60667e+09\tval's rmse: 4.31442e+09\n",
      "[162]\ttrain's rmse: 2.60259e+09\tval's rmse: 4.31262e+09\n",
      "[163]\ttrain's rmse: 2.59854e+09\tval's rmse: 4.31052e+09\n",
      "[164]\ttrain's rmse: 2.59404e+09\tval's rmse: 4.30974e+09\n",
      "[165]\ttrain's rmse: 2.59129e+09\tval's rmse: 4.3085e+09\n",
      "[166]\ttrain's rmse: 2.58771e+09\tval's rmse: 4.30681e+09\n",
      "[167]\ttrain's rmse: 2.58332e+09\tval's rmse: 4.30622e+09\n",
      "[168]\ttrain's rmse: 2.57992e+09\tval's rmse: 4.30412e+09\n",
      "[169]\ttrain's rmse: 2.57601e+09\tval's rmse: 4.30413e+09\n",
      "[170]\ttrain's rmse: 2.5721e+09\tval's rmse: 4.3004e+09\n",
      "[171]\ttrain's rmse: 2.56856e+09\tval's rmse: 4.30006e+09\n",
      "[172]\ttrain's rmse: 2.56487e+09\tval's rmse: 4.29849e+09\n",
      "[173]\ttrain's rmse: 2.56118e+09\tval's rmse: 4.2961e+09\n",
      "[174]\ttrain's rmse: 2.55839e+09\tval's rmse: 4.29513e+09\n",
      "[175]\ttrain's rmse: 2.55489e+09\tval's rmse: 4.29394e+09\n",
      "[176]\ttrain's rmse: 2.55163e+09\tval's rmse: 4.29266e+09\n",
      "[177]\ttrain's rmse: 2.54833e+09\tval's rmse: 4.29081e+09\n",
      "[178]\ttrain's rmse: 2.54458e+09\tval's rmse: 4.28838e+09\n",
      "[179]\ttrain's rmse: 2.54106e+09\tval's rmse: 4.28654e+09\n",
      "[180]\ttrain's rmse: 2.53793e+09\tval's rmse: 4.2856e+09\n",
      "[181]\ttrain's rmse: 2.53401e+09\tval's rmse: 4.28351e+09\n",
      "[182]\ttrain's rmse: 2.53124e+09\tval's rmse: 4.28163e+09\n",
      "[183]\ttrain's rmse: 2.5277e+09\tval's rmse: 4.28011e+09\n",
      "[184]\ttrain's rmse: 2.52418e+09\tval's rmse: 4.27896e+09\n",
      "[185]\ttrain's rmse: 2.52051e+09\tval's rmse: 4.27695e+09\n",
      "[186]\ttrain's rmse: 2.51698e+09\tval's rmse: 4.27535e+09\n",
      "[187]\ttrain's rmse: 2.51352e+09\tval's rmse: 4.27541e+09\n",
      "[188]\ttrain's rmse: 2.51056e+09\tval's rmse: 4.27404e+09\n",
      "[189]\ttrain's rmse: 2.50732e+09\tval's rmse: 4.27237e+09\n",
      "[190]\ttrain's rmse: 2.50382e+09\tval's rmse: 4.27104e+09\n",
      "[191]\ttrain's rmse: 2.50116e+09\tval's rmse: 4.26952e+09\n",
      "[192]\ttrain's rmse: 2.49881e+09\tval's rmse: 4.2688e+09\n",
      "[193]\ttrain's rmse: 2.4965e+09\tval's rmse: 4.26798e+09\n",
      "[194]\ttrain's rmse: 2.49399e+09\tval's rmse: 4.26635e+09\n",
      "[195]\ttrain's rmse: 2.49075e+09\tval's rmse: 4.26517e+09\n",
      "[196]\ttrain's rmse: 2.48749e+09\tval's rmse: 4.26538e+09\n",
      "[197]\ttrain's rmse: 2.4851e+09\tval's rmse: 4.26495e+09\n",
      "[198]\ttrain's rmse: 2.48169e+09\tval's rmse: 4.26449e+09\n",
      "[199]\ttrain's rmse: 2.47931e+09\tval's rmse: 4.26405e+09\n",
      "[200]\ttrain's rmse: 2.47577e+09\tval's rmse: 4.2632e+09\n",
      "[201]\ttrain's rmse: 2.47332e+09\tval's rmse: 4.26134e+09\n",
      "[202]\ttrain's rmse: 2.47033e+09\tval's rmse: 4.26087e+09\n",
      "[203]\ttrain's rmse: 2.46701e+09\tval's rmse: 4.26008e+09\n",
      "[204]\ttrain's rmse: 2.46371e+09\tval's rmse: 4.25855e+09\n",
      "[205]\ttrain's rmse: 2.4605e+09\tval's rmse: 4.25771e+09\n",
      "[206]\ttrain's rmse: 2.45811e+09\tval's rmse: 4.2565e+09\n",
      "[207]\ttrain's rmse: 2.45521e+09\tval's rmse: 4.25577e+09\n",
      "[208]\ttrain's rmse: 2.45174e+09\tval's rmse: 4.25379e+09\n",
      "[209]\ttrain's rmse: 2.44906e+09\tval's rmse: 4.25304e+09\n",
      "[210]\ttrain's rmse: 2.44613e+09\tval's rmse: 4.25274e+09\n",
      "[211]\ttrain's rmse: 2.44412e+09\tval's rmse: 4.2523e+09\n",
      "[212]\ttrain's rmse: 2.4411e+09\tval's rmse: 4.2505e+09\n",
      "[213]\ttrain's rmse: 2.43886e+09\tval's rmse: 4.24985e+09\n",
      "[214]\ttrain's rmse: 2.43597e+09\tval's rmse: 4.24777e+09\n",
      "[215]\ttrain's rmse: 2.43308e+09\tval's rmse: 4.24673e+09\n",
      "[216]\ttrain's rmse: 2.42958e+09\tval's rmse: 4.24545e+09\n",
      "[217]\ttrain's rmse: 2.42732e+09\tval's rmse: 4.24497e+09\n",
      "[218]\ttrain's rmse: 2.42555e+09\tval's rmse: 4.24465e+09\n",
      "[219]\ttrain's rmse: 2.42342e+09\tval's rmse: 4.24417e+09\n",
      "[220]\ttrain's rmse: 2.42045e+09\tval's rmse: 4.24217e+09\n",
      "[221]\ttrain's rmse: 2.41712e+09\tval's rmse: 4.24072e+09\n",
      "[222]\ttrain's rmse: 2.41438e+09\tval's rmse: 4.23955e+09\n",
      "[223]\ttrain's rmse: 2.41107e+09\tval's rmse: 4.23717e+09\n",
      "[224]\ttrain's rmse: 2.40895e+09\tval's rmse: 4.23682e+09\n",
      "[225]\ttrain's rmse: 2.40648e+09\tval's rmse: 4.2361e+09\n",
      "[226]\ttrain's rmse: 2.40429e+09\tval's rmse: 4.23488e+09\n",
      "[227]\ttrain's rmse: 2.402e+09\tval's rmse: 4.23422e+09\n",
      "[228]\ttrain's rmse: 2.39961e+09\tval's rmse: 4.23374e+09\n",
      "[229]\ttrain's rmse: 2.39677e+09\tval's rmse: 4.23229e+09\n",
      "[230]\ttrain's rmse: 2.39426e+09\tval's rmse: 4.23187e+09\n",
      "[231]\ttrain's rmse: 2.39203e+09\tval's rmse: 4.23112e+09\n",
      "[232]\ttrain's rmse: 2.38971e+09\tval's rmse: 4.23038e+09\n",
      "[233]\ttrain's rmse: 2.38675e+09\tval's rmse: 4.22885e+09\n",
      "[234]\ttrain's rmse: 2.38403e+09\tval's rmse: 4.2269e+09\n",
      "[235]\ttrain's rmse: 2.38148e+09\tval's rmse: 4.22558e+09\n",
      "[236]\ttrain's rmse: 2.37887e+09\tval's rmse: 4.22378e+09\n",
      "[237]\ttrain's rmse: 2.37606e+09\tval's rmse: 4.2223e+09\n",
      "[238]\ttrain's rmse: 2.3731e+09\tval's rmse: 4.22065e+09\n",
      "[239]\ttrain's rmse: 2.37015e+09\tval's rmse: 4.21878e+09\n",
      "[240]\ttrain's rmse: 2.36761e+09\tval's rmse: 4.21844e+09\n",
      "[241]\ttrain's rmse: 2.36512e+09\tval's rmse: 4.21808e+09\n",
      "[242]\ttrain's rmse: 2.36273e+09\tval's rmse: 4.21702e+09\n",
      "[243]\ttrain's rmse: 2.35986e+09\tval's rmse: 4.21626e+09\n",
      "[244]\ttrain's rmse: 2.35774e+09\tval's rmse: 4.2154e+09\n",
      "[245]\ttrain's rmse: 2.35572e+09\tval's rmse: 4.21418e+09\n",
      "[246]\ttrain's rmse: 2.35289e+09\tval's rmse: 4.21259e+09\n",
      "[247]\ttrain's rmse: 2.35008e+09\tval's rmse: 4.2111e+09\n",
      "[248]\ttrain's rmse: 2.34734e+09\tval's rmse: 4.21017e+09\n",
      "[249]\ttrain's rmse: 2.34545e+09\tval's rmse: 4.21049e+09\n",
      "[250]\ttrain's rmse: 2.3439e+09\tval's rmse: 4.20976e+09\n",
      "[251]\ttrain's rmse: 2.34199e+09\tval's rmse: 4.20926e+09\n",
      "[252]\ttrain's rmse: 2.33952e+09\tval's rmse: 4.20773e+09\n",
      "[253]\ttrain's rmse: 2.33747e+09\tval's rmse: 4.20708e+09\n",
      "[254]\ttrain's rmse: 2.33508e+09\tval's rmse: 4.20435e+09\n",
      "[255]\ttrain's rmse: 2.33247e+09\tval's rmse: 4.20332e+09\n",
      "[256]\ttrain's rmse: 2.33055e+09\tval's rmse: 4.2021e+09\n",
      "[257]\ttrain's rmse: 2.3279e+09\tval's rmse: 4.20077e+09\n",
      "[258]\ttrain's rmse: 2.3253e+09\tval's rmse: 4.19974e+09\n",
      "[259]\ttrain's rmse: 2.3229e+09\tval's rmse: 4.19802e+09\n",
      "[260]\ttrain's rmse: 2.32038e+09\tval's rmse: 4.19634e+09\n",
      "[261]\ttrain's rmse: 2.31791e+09\tval's rmse: 4.19509e+09\n",
      "[262]\ttrain's rmse: 2.31547e+09\tval's rmse: 4.19357e+09\n",
      "[263]\ttrain's rmse: 2.31299e+09\tval's rmse: 4.19146e+09\n",
      "[264]\ttrain's rmse: 2.31043e+09\tval's rmse: 4.18993e+09\n",
      "[265]\ttrain's rmse: 2.30798e+09\tval's rmse: 4.189e+09\n",
      "[266]\ttrain's rmse: 2.30567e+09\tval's rmse: 4.18763e+09\n",
      "[267]\ttrain's rmse: 2.30407e+09\tval's rmse: 4.18669e+09\n",
      "[268]\ttrain's rmse: 2.30265e+09\tval's rmse: 4.1862e+09\n",
      "[269]\ttrain's rmse: 2.30068e+09\tval's rmse: 4.18586e+09\n",
      "[270]\ttrain's rmse: 2.29831e+09\tval's rmse: 4.18466e+09\n",
      "[271]\ttrain's rmse: 2.29582e+09\tval's rmse: 4.18318e+09\n",
      "[272]\ttrain's rmse: 2.29358e+09\tval's rmse: 4.18276e+09\n",
      "[273]\ttrain's rmse: 2.29102e+09\tval's rmse: 4.18042e+09\n",
      "[274]\ttrain's rmse: 2.28892e+09\tval's rmse: 4.17952e+09\n",
      "[275]\ttrain's rmse: 2.28675e+09\tval's rmse: 4.1788e+09\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[276]\ttrain's rmse: 2.28485e+09\tval's rmse: 4.17752e+09\n",
      "[277]\ttrain's rmse: 2.2831e+09\tval's rmse: 4.17689e+09\n",
      "[278]\ttrain's rmse: 2.28106e+09\tval's rmse: 4.176e+09\n",
      "[279]\ttrain's rmse: 2.2786e+09\tval's rmse: 4.17509e+09\n",
      "[280]\ttrain's rmse: 2.27627e+09\tval's rmse: 4.17425e+09\n",
      "[281]\ttrain's rmse: 2.27397e+09\tval's rmse: 4.17248e+09\n",
      "[282]\ttrain's rmse: 2.27211e+09\tval's rmse: 4.17252e+09\n",
      "[283]\ttrain's rmse: 2.26973e+09\tval's rmse: 4.17168e+09\n",
      "[284]\ttrain's rmse: 2.26756e+09\tval's rmse: 4.17046e+09\n",
      "[285]\ttrain's rmse: 2.26536e+09\tval's rmse: 4.169e+09\n",
      "[286]\ttrain's rmse: 2.2632e+09\tval's rmse: 4.16845e+09\n",
      "[287]\ttrain's rmse: 2.26103e+09\tval's rmse: 4.1679e+09\n",
      "[288]\ttrain's rmse: 2.25828e+09\tval's rmse: 4.16666e+09\n",
      "[289]\ttrain's rmse: 2.25553e+09\tval's rmse: 4.16543e+09\n",
      "[290]\ttrain's rmse: 2.25332e+09\tval's rmse: 4.16404e+09\n",
      "[291]\ttrain's rmse: 2.25152e+09\tval's rmse: 4.16418e+09\n",
      "[292]\ttrain's rmse: 2.24921e+09\tval's rmse: 4.16334e+09\n",
      "[293]\ttrain's rmse: 2.24706e+09\tval's rmse: 4.16205e+09\n",
      "[294]\ttrain's rmse: 2.24589e+09\tval's rmse: 4.16194e+09\n",
      "[295]\ttrain's rmse: 2.2438e+09\tval's rmse: 4.1605e+09\n",
      "[296]\ttrain's rmse: 2.24239e+09\tval's rmse: 4.16003e+09\n",
      "[297]\ttrain's rmse: 2.23994e+09\tval's rmse: 4.15835e+09\n",
      "[298]\ttrain's rmse: 2.23757e+09\tval's rmse: 4.15695e+09\n",
      "[299]\ttrain's rmse: 2.2358e+09\tval's rmse: 4.15626e+09\n",
      "[300]\ttrain's rmse: 2.23424e+09\tval's rmse: 4.15586e+09\n",
      "[301]\ttrain's rmse: 2.23192e+09\tval's rmse: 4.1535e+09\n",
      "[302]\ttrain's rmse: 2.22992e+09\tval's rmse: 4.15253e+09\n",
      "[303]\ttrain's rmse: 2.22779e+09\tval's rmse: 4.15183e+09\n",
      "[304]\ttrain's rmse: 2.22656e+09\tval's rmse: 4.15187e+09\n",
      "[305]\ttrain's rmse: 2.22487e+09\tval's rmse: 4.15202e+09\n",
      "[306]\ttrain's rmse: 2.22348e+09\tval's rmse: 4.15176e+09\n",
      "[307]\ttrain's rmse: 2.22085e+09\tval's rmse: 4.15065e+09\n",
      "[308]\ttrain's rmse: 2.21855e+09\tval's rmse: 4.1499e+09\n",
      "[309]\ttrain's rmse: 2.21706e+09\tval's rmse: 4.15036e+09\n",
      "[310]\ttrain's rmse: 2.21505e+09\tval's rmse: 4.14989e+09\n",
      "[311]\ttrain's rmse: 2.21267e+09\tval's rmse: 4.14823e+09\n",
      "[312]\ttrain's rmse: 2.21064e+09\tval's rmse: 4.14707e+09\n",
      "[313]\ttrain's rmse: 2.20881e+09\tval's rmse: 4.14668e+09\n",
      "[314]\ttrain's rmse: 2.20668e+09\tval's rmse: 4.14584e+09\n",
      "[315]\ttrain's rmse: 2.20461e+09\tval's rmse: 4.14525e+09\n",
      "[316]\ttrain's rmse: 2.20261e+09\tval's rmse: 4.1449e+09\n",
      "[317]\ttrain's rmse: 2.20061e+09\tval's rmse: 4.14338e+09\n",
      "[318]\ttrain's rmse: 2.19853e+09\tval's rmse: 4.14221e+09\n",
      "[319]\ttrain's rmse: 2.19725e+09\tval's rmse: 4.14182e+09\n",
      "[320]\ttrain's rmse: 2.19611e+09\tval's rmse: 4.14166e+09\n",
      "[321]\ttrain's rmse: 2.19427e+09\tval's rmse: 4.14056e+09\n",
      "[322]\ttrain's rmse: 2.19233e+09\tval's rmse: 4.14023e+09\n",
      "[323]\ttrain's rmse: 2.19048e+09\tval's rmse: 4.1398e+09\n",
      "[324]\ttrain's rmse: 2.18894e+09\tval's rmse: 4.13899e+09\n",
      "[325]\ttrain's rmse: 2.18667e+09\tval's rmse: 4.13738e+09\n",
      "[326]\ttrain's rmse: 2.18527e+09\tval's rmse: 4.1373e+09\n",
      "[327]\ttrain's rmse: 2.18385e+09\tval's rmse: 4.13668e+09\n",
      "[328]\ttrain's rmse: 2.18251e+09\tval's rmse: 4.13545e+09\n",
      "[329]\ttrain's rmse: 2.1809e+09\tval's rmse: 4.13549e+09\n",
      "[330]\ttrain's rmse: 2.17944e+09\tval's rmse: 4.13519e+09\n",
      "[331]\ttrain's rmse: 2.17719e+09\tval's rmse: 4.13373e+09\n",
      "[332]\ttrain's rmse: 2.17503e+09\tval's rmse: 4.13336e+09\n",
      "[333]\ttrain's rmse: 2.17319e+09\tval's rmse: 4.13284e+09\n",
      "[334]\ttrain's rmse: 2.17104e+09\tval's rmse: 4.13218e+09\n",
      "[335]\ttrain's rmse: 2.16887e+09\tval's rmse: 4.13099e+09\n",
      "[336]\ttrain's rmse: 2.16696e+09\tval's rmse: 4.13079e+09\n",
      "[337]\ttrain's rmse: 2.16491e+09\tval's rmse: 4.13015e+09\n",
      "[338]\ttrain's rmse: 2.16304e+09\tval's rmse: 4.12983e+09\n",
      "[339]\ttrain's rmse: 2.16167e+09\tval's rmse: 4.12955e+09\n",
      "[340]\ttrain's rmse: 2.15949e+09\tval's rmse: 4.12876e+09\n",
      "[341]\ttrain's rmse: 2.15795e+09\tval's rmse: 4.12854e+09\n",
      "[342]\ttrain's rmse: 2.15655e+09\tval's rmse: 4.12823e+09\n",
      "[343]\ttrain's rmse: 2.15436e+09\tval's rmse: 4.12715e+09\n",
      "[344]\ttrain's rmse: 2.15242e+09\tval's rmse: 4.12672e+09\n",
      "[345]\ttrain's rmse: 2.15079e+09\tval's rmse: 4.12647e+09\n",
      "[346]\ttrain's rmse: 2.1493e+09\tval's rmse: 4.12638e+09\n",
      "[347]\ttrain's rmse: 2.14798e+09\tval's rmse: 4.12569e+09\n",
      "[348]\ttrain's rmse: 2.14593e+09\tval's rmse: 4.1245e+09\n",
      "[349]\ttrain's rmse: 2.14446e+09\tval's rmse: 4.1243e+09\n",
      "[350]\ttrain's rmse: 2.143e+09\tval's rmse: 4.12363e+09\n",
      "[351]\ttrain's rmse: 2.14087e+09\tval's rmse: 4.1228e+09\n",
      "[352]\ttrain's rmse: 2.13917e+09\tval's rmse: 4.12167e+09\n",
      "[353]\ttrain's rmse: 2.13717e+09\tval's rmse: 4.12107e+09\n",
      "[354]\ttrain's rmse: 2.13524e+09\tval's rmse: 4.12115e+09\n",
      "[355]\ttrain's rmse: 2.13374e+09\tval's rmse: 4.12075e+09\n",
      "[356]\ttrain's rmse: 2.13227e+09\tval's rmse: 4.1196e+09\n",
      "[357]\ttrain's rmse: 2.13081e+09\tval's rmse: 4.11957e+09\n",
      "[358]\ttrain's rmse: 2.12952e+09\tval's rmse: 4.11865e+09\n",
      "[359]\ttrain's rmse: 2.1281e+09\tval's rmse: 4.11837e+09\n",
      "[360]\ttrain's rmse: 2.12623e+09\tval's rmse: 4.11739e+09\n",
      "[361]\ttrain's rmse: 2.12488e+09\tval's rmse: 4.11764e+09\n",
      "[362]\ttrain's rmse: 2.12315e+09\tval's rmse: 4.11673e+09\n",
      "[363]\ttrain's rmse: 2.122e+09\tval's rmse: 4.11619e+09\n",
      "[364]\ttrain's rmse: 2.12006e+09\tval's rmse: 4.11577e+09\n",
      "[365]\ttrain's rmse: 2.11845e+09\tval's rmse: 4.11575e+09\n",
      "[366]\ttrain's rmse: 2.11677e+09\tval's rmse: 4.11482e+09\n",
      "[367]\ttrain's rmse: 2.1152e+09\tval's rmse: 4.1146e+09\n",
      "[368]\ttrain's rmse: 2.11351e+09\tval's rmse: 4.11446e+09\n",
      "[369]\ttrain's rmse: 2.11161e+09\tval's rmse: 4.11361e+09\n",
      "[370]\ttrain's rmse: 2.11044e+09\tval's rmse: 4.11336e+09\n",
      "[371]\ttrain's rmse: 2.109e+09\tval's rmse: 4.11315e+09\n",
      "[372]\ttrain's rmse: 2.10757e+09\tval's rmse: 4.11301e+09\n",
      "[373]\ttrain's rmse: 2.10613e+09\tval's rmse: 4.11256e+09\n",
      "[374]\ttrain's rmse: 2.10484e+09\tval's rmse: 4.11249e+09\n",
      "[375]\ttrain's rmse: 2.10302e+09\tval's rmse: 4.11194e+09\n",
      "[376]\ttrain's rmse: 2.10117e+09\tval's rmse: 4.11096e+09\n",
      "[377]\ttrain's rmse: 2.10008e+09\tval's rmse: 4.11086e+09\n",
      "[378]\ttrain's rmse: 2.09857e+09\tval's rmse: 4.1106e+09\n",
      "[379]\ttrain's rmse: 2.09686e+09\tval's rmse: 4.10989e+09\n",
      "[380]\ttrain's rmse: 2.09505e+09\tval's rmse: 4.1091e+09\n",
      "[381]\ttrain's rmse: 2.09338e+09\tval's rmse: 4.10805e+09\n",
      "[382]\ttrain's rmse: 2.09179e+09\tval's rmse: 4.10722e+09\n",
      "[383]\ttrain's rmse: 2.09031e+09\tval's rmse: 4.10671e+09\n",
      "[384]\ttrain's rmse: 2.0893e+09\tval's rmse: 4.1063e+09\n",
      "[385]\ttrain's rmse: 2.08759e+09\tval's rmse: 4.1058e+09\n",
      "[386]\ttrain's rmse: 2.08601e+09\tval's rmse: 4.10542e+09\n",
      "[387]\ttrain's rmse: 2.08465e+09\tval's rmse: 4.10502e+09\n",
      "[388]\ttrain's rmse: 2.08294e+09\tval's rmse: 4.10402e+09\n",
      "[389]\ttrain's rmse: 2.08158e+09\tval's rmse: 4.10356e+09\n",
      "[390]\ttrain's rmse: 2.07999e+09\tval's rmse: 4.10275e+09\n",
      "[391]\ttrain's rmse: 2.07879e+09\tval's rmse: 4.10273e+09\n",
      "[392]\ttrain's rmse: 2.07717e+09\tval's rmse: 4.10212e+09\n",
      "[393]\ttrain's rmse: 2.07573e+09\tval's rmse: 4.10179e+09\n",
      "[394]\ttrain's rmse: 2.07438e+09\tval's rmse: 4.10153e+09\n",
      "[395]\ttrain's rmse: 2.07269e+09\tval's rmse: 4.10142e+09\n",
      "[396]\ttrain's rmse: 2.07139e+09\tval's rmse: 4.10065e+09\n",
      "[397]\ttrain's rmse: 2.07e+09\tval's rmse: 4.10082e+09\n",
      "[398]\ttrain's rmse: 2.06834e+09\tval's rmse: 4.1001e+09\n",
      "[399]\ttrain's rmse: 2.06712e+09\tval's rmse: 4.09984e+09\n",
      "[400]\ttrain's rmse: 2.06545e+09\tval's rmse: 4.09935e+09\n",
      "[401]\ttrain's rmse: 2.06461e+09\tval's rmse: 4.09896e+09\n",
      "[402]\ttrain's rmse: 2.06318e+09\tval's rmse: 4.09803e+09\n",
      "[403]\ttrain's rmse: 2.06169e+09\tval's rmse: 4.09743e+09\n",
      "[404]\ttrain's rmse: 2.06005e+09\tval's rmse: 4.09683e+09\n",
      "[405]\ttrain's rmse: 2.05854e+09\tval's rmse: 4.09618e+09\n",
      "[406]\ttrain's rmse: 2.05738e+09\tval's rmse: 4.0962e+09\n",
      "[407]\ttrain's rmse: 2.05616e+09\tval's rmse: 4.0965e+09\n",
      "[408]\ttrain's rmse: 2.05482e+09\tval's rmse: 4.09596e+09\n",
      "[409]\ttrain's rmse: 2.05317e+09\tval's rmse: 4.09537e+09\n",
      "[410]\ttrain's rmse: 2.05147e+09\tval's rmse: 4.095e+09\n",
      "[411]\ttrain's rmse: 2.04987e+09\tval's rmse: 4.09423e+09\n",
      "[412]\ttrain's rmse: 2.04848e+09\tval's rmse: 4.09381e+09\n",
      "[413]\ttrain's rmse: 2.04688e+09\tval's rmse: 4.0932e+09\n",
      "[414]\ttrain's rmse: 2.04575e+09\tval's rmse: 4.09347e+09\n",
      "[415]\ttrain's rmse: 2.04408e+09\tval's rmse: 4.09284e+09\n",
      "[416]\ttrain's rmse: 2.04254e+09\tval's rmse: 4.09278e+09\n",
      "[417]\ttrain's rmse: 2.04095e+09\tval's rmse: 4.09239e+09\n",
      "[418]\ttrain's rmse: 2.0397e+09\tval's rmse: 4.09236e+09\n",
      "[419]\ttrain's rmse: 2.0381e+09\tval's rmse: 4.09221e+09\n",
      "[420]\ttrain's rmse: 2.03651e+09\tval's rmse: 4.09175e+09\n",
      "[421]\ttrain's rmse: 2.03527e+09\tval's rmse: 4.09129e+09\n",
      "[422]\ttrain's rmse: 2.03415e+09\tval's rmse: 4.09134e+09\n",
      "[423]\ttrain's rmse: 2.03275e+09\tval's rmse: 4.09075e+09\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[424]\ttrain's rmse: 2.03129e+09\tval's rmse: 4.08999e+09\n",
      "[425]\ttrain's rmse: 2.02967e+09\tval's rmse: 4.08926e+09\n",
      "[426]\ttrain's rmse: 2.02841e+09\tval's rmse: 4.08922e+09\n",
      "[427]\ttrain's rmse: 2.02696e+09\tval's rmse: 4.08872e+09\n",
      "[428]\ttrain's rmse: 2.02603e+09\tval's rmse: 4.089e+09\n",
      "[429]\ttrain's rmse: 2.02505e+09\tval's rmse: 4.08852e+09\n",
      "[430]\ttrain's rmse: 2.02371e+09\tval's rmse: 4.08816e+09\n",
      "[431]\ttrain's rmse: 2.02277e+09\tval's rmse: 4.08776e+09\n",
      "[432]\ttrain's rmse: 2.02082e+09\tval's rmse: 4.08697e+09\n",
      "[433]\ttrain's rmse: 2.01982e+09\tval's rmse: 4.08714e+09\n",
      "[434]\ttrain's rmse: 2.01885e+09\tval's rmse: 4.08635e+09\n",
      "[435]\ttrain's rmse: 2.01797e+09\tval's rmse: 4.08637e+09\n",
      "[436]\ttrain's rmse: 2.01634e+09\tval's rmse: 4.08549e+09\n",
      "[437]\ttrain's rmse: 2.01521e+09\tval's rmse: 4.08521e+09\n",
      "[438]\ttrain's rmse: 2.01365e+09\tval's rmse: 4.08412e+09\n",
      "[439]\ttrain's rmse: 2.01278e+09\tval's rmse: 4.08393e+09\n",
      "[440]\ttrain's rmse: 2.01151e+09\tval's rmse: 4.0841e+09\n",
      "[441]\ttrain's rmse: 2.00971e+09\tval's rmse: 4.08348e+09\n",
      "[442]\ttrain's rmse: 2.0078e+09\tval's rmse: 4.08261e+09\n",
      "[443]\ttrain's rmse: 2.00603e+09\tval's rmse: 4.08188e+09\n",
      "[444]\ttrain's rmse: 2.00504e+09\tval's rmse: 4.08134e+09\n",
      "[445]\ttrain's rmse: 2.00335e+09\tval's rmse: 4.08101e+09\n",
      "[446]\ttrain's rmse: 2.002e+09\tval's rmse: 4.08083e+09\n",
      "[447]\ttrain's rmse: 2.00102e+09\tval's rmse: 4.08035e+09\n",
      "[448]\ttrain's rmse: 1.99935e+09\tval's rmse: 4.07997e+09\n",
      "[449]\ttrain's rmse: 1.99794e+09\tval's rmse: 4.07985e+09\n",
      "[450]\ttrain's rmse: 1.99711e+09\tval's rmse: 4.07982e+09\n",
      "[451]\ttrain's rmse: 1.99544e+09\tval's rmse: 4.07941e+09\n",
      "[452]\ttrain's rmse: 1.99404e+09\tval's rmse: 4.07984e+09\n",
      "[453]\ttrain's rmse: 1.99231e+09\tval's rmse: 4.07899e+09\n",
      "[454]\ttrain's rmse: 1.99103e+09\tval's rmse: 4.07842e+09\n",
      "[455]\ttrain's rmse: 1.98987e+09\tval's rmse: 4.07822e+09\n",
      "[456]\ttrain's rmse: 1.98847e+09\tval's rmse: 4.07803e+09\n",
      "[457]\ttrain's rmse: 1.98737e+09\tval's rmse: 4.07796e+09\n",
      "[458]\ttrain's rmse: 1.98615e+09\tval's rmse: 4.07719e+09\n",
      "[459]\ttrain's rmse: 1.98443e+09\tval's rmse: 4.07711e+09\n",
      "[460]\ttrain's rmse: 1.98292e+09\tval's rmse: 4.0766e+09\n",
      "[461]\ttrain's rmse: 1.98182e+09\tval's rmse: 4.07627e+09\n",
      "[462]\ttrain's rmse: 1.98086e+09\tval's rmse: 4.07584e+09\n",
      "[463]\ttrain's rmse: 1.97915e+09\tval's rmse: 4.07519e+09\n",
      "[464]\ttrain's rmse: 1.97737e+09\tval's rmse: 4.07473e+09\n",
      "[465]\ttrain's rmse: 1.97638e+09\tval's rmse: 4.07461e+09\n",
      "[466]\ttrain's rmse: 1.97511e+09\tval's rmse: 4.07455e+09\n",
      "[467]\ttrain's rmse: 1.97395e+09\tval's rmse: 4.07463e+09\n",
      "[468]\ttrain's rmse: 1.97306e+09\tval's rmse: 4.07431e+09\n",
      "[469]\ttrain's rmse: 1.97165e+09\tval's rmse: 4.07392e+09\n",
      "[470]\ttrain's rmse: 1.97044e+09\tval's rmse: 4.07379e+09\n",
      "[471]\ttrain's rmse: 1.96911e+09\tval's rmse: 4.0736e+09\n",
      "[472]\ttrain's rmse: 1.96816e+09\tval's rmse: 4.07357e+09\n",
      "[473]\ttrain's rmse: 1.96689e+09\tval's rmse: 4.07329e+09\n",
      "[474]\ttrain's rmse: 1.96544e+09\tval's rmse: 4.07285e+09\n",
      "[475]\ttrain's rmse: 1.96436e+09\tval's rmse: 4.07283e+09\n",
      "[476]\ttrain's rmse: 1.96326e+09\tval's rmse: 4.07223e+09\n",
      "[477]\ttrain's rmse: 1.96193e+09\tval's rmse: 4.07218e+09\n",
      "[478]\ttrain's rmse: 1.96029e+09\tval's rmse: 4.07134e+09\n",
      "[479]\ttrain's rmse: 1.95863e+09\tval's rmse: 4.07055e+09\n",
      "[480]\ttrain's rmse: 1.95704e+09\tval's rmse: 4.06964e+09\n",
      "[481]\ttrain's rmse: 1.95543e+09\tval's rmse: 4.06936e+09\n",
      "[482]\ttrain's rmse: 1.95399e+09\tval's rmse: 4.06895e+09\n",
      "[483]\ttrain's rmse: 1.9533e+09\tval's rmse: 4.06901e+09\n",
      "[484]\ttrain's rmse: 1.95213e+09\tval's rmse: 4.06862e+09\n",
      "[485]\ttrain's rmse: 1.95113e+09\tval's rmse: 4.06874e+09\n",
      "[486]\ttrain's rmse: 1.94985e+09\tval's rmse: 4.06836e+09\n",
      "[487]\ttrain's rmse: 1.94844e+09\tval's rmse: 4.06861e+09\n",
      "[488]\ttrain's rmse: 1.94717e+09\tval's rmse: 4.06835e+09\n",
      "[489]\ttrain's rmse: 1.9461e+09\tval's rmse: 4.06791e+09\n",
      "[490]\ttrain's rmse: 1.94475e+09\tval's rmse: 4.06765e+09\n",
      "[491]\ttrain's rmse: 1.94336e+09\tval's rmse: 4.06761e+09\n",
      "[492]\ttrain's rmse: 1.94197e+09\tval's rmse: 4.06735e+09\n",
      "[493]\ttrain's rmse: 1.94114e+09\tval's rmse: 4.0671e+09\n",
      "[494]\ttrain's rmse: 1.94044e+09\tval's rmse: 4.06697e+09\n",
      "[495]\ttrain's rmse: 1.93903e+09\tval's rmse: 4.06658e+09\n",
      "[496]\ttrain's rmse: 1.9383e+09\tval's rmse: 4.06638e+09\n",
      "[497]\ttrain's rmse: 1.93677e+09\tval's rmse: 4.06615e+09\n",
      "[498]\ttrain's rmse: 1.93579e+09\tval's rmse: 4.06611e+09\n",
      "[499]\ttrain's rmse: 1.93463e+09\tval's rmse: 4.06556e+09\n",
      "[500]\ttrain's rmse: 1.93344e+09\tval's rmse: 4.06479e+09\n",
      "[501]\ttrain's rmse: 1.93263e+09\tval's rmse: 4.06446e+09\n",
      "[502]\ttrain's rmse: 1.93147e+09\tval's rmse: 4.06462e+09\n",
      "[503]\ttrain's rmse: 1.93072e+09\tval's rmse: 4.06449e+09\n",
      "[504]\ttrain's rmse: 1.92924e+09\tval's rmse: 4.0638e+09\n",
      "[505]\ttrain's rmse: 1.92847e+09\tval's rmse: 4.06362e+09\n",
      "[506]\ttrain's rmse: 1.9271e+09\tval's rmse: 4.06321e+09\n",
      "[507]\ttrain's rmse: 1.92608e+09\tval's rmse: 4.0634e+09\n",
      "[508]\ttrain's rmse: 1.92472e+09\tval's rmse: 4.06305e+09\n",
      "[509]\ttrain's rmse: 1.92391e+09\tval's rmse: 4.06288e+09\n",
      "[510]\ttrain's rmse: 1.92314e+09\tval's rmse: 4.0632e+09\n",
      "[511]\ttrain's rmse: 1.92242e+09\tval's rmse: 4.06312e+09\n",
      "[512]\ttrain's rmse: 1.92123e+09\tval's rmse: 4.06258e+09\n",
      "[513]\ttrain's rmse: 1.92036e+09\tval's rmse: 4.06253e+09\n",
      "[514]\ttrain's rmse: 1.91934e+09\tval's rmse: 4.06237e+09\n",
      "[515]\ttrain's rmse: 1.91844e+09\tval's rmse: 4.06234e+09\n",
      "[516]\ttrain's rmse: 1.91734e+09\tval's rmse: 4.06225e+09\n",
      "[517]\ttrain's rmse: 1.91618e+09\tval's rmse: 4.06193e+09\n",
      "[518]\ttrain's rmse: 1.91538e+09\tval's rmse: 4.06157e+09\n",
      "[519]\ttrain's rmse: 1.91437e+09\tval's rmse: 4.06122e+09\n",
      "[520]\ttrain's rmse: 1.91329e+09\tval's rmse: 4.06061e+09\n",
      "[521]\ttrain's rmse: 1.91201e+09\tval's rmse: 4.06036e+09\n",
      "[522]\ttrain's rmse: 1.91137e+09\tval's rmse: 4.06028e+09\n",
      "[523]\ttrain's rmse: 1.91049e+09\tval's rmse: 4.06039e+09\n",
      "[524]\ttrain's rmse: 1.90966e+09\tval's rmse: 4.06031e+09\n",
      "[525]\ttrain's rmse: 1.90872e+09\tval's rmse: 4.06031e+09\n",
      "[526]\ttrain's rmse: 1.90736e+09\tval's rmse: 4.06006e+09\n",
      "[527]\ttrain's rmse: 1.90631e+09\tval's rmse: 4.0597e+09\n",
      "[528]\ttrain's rmse: 1.90545e+09\tval's rmse: 4.0596e+09\n",
      "[529]\ttrain's rmse: 1.90462e+09\tval's rmse: 4.05947e+09\n",
      "[530]\ttrain's rmse: 1.90384e+09\tval's rmse: 4.05947e+09\n",
      "[531]\ttrain's rmse: 1.90312e+09\tval's rmse: 4.05944e+09\n",
      "[532]\ttrain's rmse: 1.90214e+09\tval's rmse: 4.05897e+09\n",
      "[533]\ttrain's rmse: 1.90135e+09\tval's rmse: 4.05854e+09\n",
      "[534]\ttrain's rmse: 1.90032e+09\tval's rmse: 4.05785e+09\n",
      "[535]\ttrain's rmse: 1.89951e+09\tval's rmse: 4.05783e+09\n",
      "[536]\ttrain's rmse: 1.89828e+09\tval's rmse: 4.05761e+09\n",
      "[537]\ttrain's rmse: 1.89758e+09\tval's rmse: 4.05735e+09\n",
      "[538]\ttrain's rmse: 1.89634e+09\tval's rmse: 4.05664e+09\n",
      "[539]\ttrain's rmse: 1.8956e+09\tval's rmse: 4.05664e+09\n",
      "[540]\ttrain's rmse: 1.89489e+09\tval's rmse: 4.05644e+09\n",
      "[541]\ttrain's rmse: 1.89432e+09\tval's rmse: 4.05631e+09\n",
      "[542]\ttrain's rmse: 1.8938e+09\tval's rmse: 4.05629e+09\n",
      "[543]\ttrain's rmse: 1.89292e+09\tval's rmse: 4.05618e+09\n",
      "[544]\ttrain's rmse: 1.89217e+09\tval's rmse: 4.05587e+09\n",
      "[545]\ttrain's rmse: 1.89142e+09\tval's rmse: 4.05556e+09\n",
      "[546]\ttrain's rmse: 1.89055e+09\tval's rmse: 4.05555e+09\n",
      "[547]\ttrain's rmse: 1.88962e+09\tval's rmse: 4.05514e+09\n",
      "[548]\ttrain's rmse: 1.88895e+09\tval's rmse: 4.0551e+09\n",
      "[549]\ttrain's rmse: 1.88828e+09\tval's rmse: 4.0549e+09\n",
      "[550]\ttrain's rmse: 1.88739e+09\tval's rmse: 4.05479e+09\n",
      "[551]\ttrain's rmse: 1.88665e+09\tval's rmse: 4.05476e+09\n",
      "[552]\ttrain's rmse: 1.88577e+09\tval's rmse: 4.05438e+09\n",
      "[553]\ttrain's rmse: 1.88482e+09\tval's rmse: 4.05432e+09\n",
      "[554]\ttrain's rmse: 1.88407e+09\tval's rmse: 4.05421e+09\n",
      "[555]\ttrain's rmse: 1.88324e+09\tval's rmse: 4.05414e+09\n",
      "[556]\ttrain's rmse: 1.88229e+09\tval's rmse: 4.05347e+09\n",
      "[557]\ttrain's rmse: 1.88143e+09\tval's rmse: 4.05343e+09\n",
      "[558]\ttrain's rmse: 1.88065e+09\tval's rmse: 4.05336e+09\n",
      "[559]\ttrain's rmse: 1.88e+09\tval's rmse: 4.05329e+09\n",
      "[560]\ttrain's rmse: 1.87925e+09\tval's rmse: 4.05324e+09\n",
      "[561]\ttrain's rmse: 1.87855e+09\tval's rmse: 4.05325e+09\n",
      "[562]\ttrain's rmse: 1.8777e+09\tval's rmse: 4.05279e+09\n",
      "[563]\ttrain's rmse: 1.8769e+09\tval's rmse: 4.05265e+09\n",
      "[564]\ttrain's rmse: 1.87614e+09\tval's rmse: 4.0527e+09\n",
      "[565]\ttrain's rmse: 1.87546e+09\tval's rmse: 4.05258e+09\n",
      "[566]\ttrain's rmse: 1.87454e+09\tval's rmse: 4.05192e+09\n",
      "[567]\ttrain's rmse: 1.87365e+09\tval's rmse: 4.05121e+09\n",
      "[568]\ttrain's rmse: 1.87295e+09\tval's rmse: 4.05101e+09\n",
      "[569]\ttrain's rmse: 1.87227e+09\tval's rmse: 4.05085e+09\n",
      "[570]\ttrain's rmse: 1.8715e+09\tval's rmse: 4.05088e+09\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[571]\ttrain's rmse: 1.87078e+09\tval's rmse: 4.05086e+09\n",
      "[572]\ttrain's rmse: 1.87012e+09\tval's rmse: 4.05076e+09\n",
      "[573]\ttrain's rmse: 1.86944e+09\tval's rmse: 4.05069e+09\n",
      "[574]\ttrain's rmse: 1.86872e+09\tval's rmse: 4.0507e+09\n",
      "[575]\ttrain's rmse: 1.86797e+09\tval's rmse: 4.05071e+09\n",
      "[576]\ttrain's rmse: 1.86724e+09\tval's rmse: 4.05057e+09\n",
      "[577]\ttrain's rmse: 1.86651e+09\tval's rmse: 4.05032e+09\n",
      "[578]\ttrain's rmse: 1.86585e+09\tval's rmse: 4.05032e+09\n",
      "[579]\ttrain's rmse: 1.86516e+09\tval's rmse: 4.05012e+09\n",
      "[580]\ttrain's rmse: 1.86442e+09\tval's rmse: 4.05013e+09\n",
      "[581]\ttrain's rmse: 1.86373e+09\tval's rmse: 4.04978e+09\n",
      "[582]\ttrain's rmse: 1.86306e+09\tval's rmse: 4.04967e+09\n",
      "[583]\ttrain's rmse: 1.86237e+09\tval's rmse: 4.04954e+09\n",
      "[584]\ttrain's rmse: 1.86151e+09\tval's rmse: 4.04907e+09\n",
      "[585]\ttrain's rmse: 1.86082e+09\tval's rmse: 4.04914e+09\n",
      "[586]\ttrain's rmse: 1.8602e+09\tval's rmse: 4.04918e+09\n",
      "[587]\ttrain's rmse: 1.85953e+09\tval's rmse: 4.04921e+09\n",
      "[588]\ttrain's rmse: 1.85905e+09\tval's rmse: 4.04921e+09\n",
      "[589]\ttrain's rmse: 1.85846e+09\tval's rmse: 4.0491e+09\n",
      "[590]\ttrain's rmse: 1.85776e+09\tval's rmse: 4.04908e+09\n",
      "[591]\ttrain's rmse: 1.85707e+09\tval's rmse: 4.04911e+09\n",
      "[592]\ttrain's rmse: 1.85637e+09\tval's rmse: 4.04892e+09\n",
      "[593]\ttrain's rmse: 1.85554e+09\tval's rmse: 4.04858e+09\n",
      "[594]\ttrain's rmse: 1.8548e+09\tval's rmse: 4.04851e+09\n",
      "[595]\ttrain's rmse: 1.85411e+09\tval's rmse: 4.04852e+09\n",
      "[596]\ttrain's rmse: 1.85343e+09\tval's rmse: 4.04841e+09\n",
      "[597]\ttrain's rmse: 1.8528e+09\tval's rmse: 4.04851e+09\n",
      "[598]\ttrain's rmse: 1.85205e+09\tval's rmse: 4.04791e+09\n",
      "[599]\ttrain's rmse: 1.85145e+09\tval's rmse: 4.04799e+09\n",
      "[600]\ttrain's rmse: 1.85085e+09\tval's rmse: 4.04808e+09\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[600]\ttrain's rmse: 1.85085e+09\tval's rmse: 4.04808e+09\n",
      "RMSE  : 4048078792.399872\n",
      "-----------------------------------------------------------------\n",
      "Training results \n",
      "\n",
      "Training error:  1850847335.2996273\n",
      "Validation error:  4048078792.399872\n",
      "Test error:  2529971560.3976746\n",
      "Model file save at ['/home/smose_roshni/roshni/deepfuel-ML/src/pre-trained_models/LightGBM.joblib']\n"
     ]
    }
   ],
   "source": [
    "!python  '../src/train.py'  --model_name 'LightGBM' --data_path '../data/tropics'  --exp_name 'LightGBM_exp'"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "public-delay",
   "metadata": {
    "id": "public-delay"
   },
   "source": [
    "Training logs can be viewed live at the following link shown above: \n",
    "`https://ui.neptune.ai/shared/step-by-step-monitoring-experiments-live/e/STEP-xxx`"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "entertaining-kenya",
   "metadata": {
    "id": "entertaining-kenya"
   },
   "source": [
    "### Loading the trained model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "hidden-accounting",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-19T05:48:52.587460Z",
     "start_time": "2021-03-19T05:48:52.166467Z"
    },
    "id": "hidden-accounting"
   },
   "outputs": [],
   "source": [
    "model = load('../src/pre-trained_models/LightGBM.joblib')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "advanced-minneapolis",
   "metadata": {
    "id": "advanced-minneapolis"
   },
   "source": [
    "### Feature importance using SHAP\n",
    "[SHAP](https://github.com/slundberg/shap) (SHapley Additive exPlanations) is a game theoretic approach to explain the output of any machine learning model. It connects optimal credit allocation with local explanations using the classic Shapley values from game theory and their related extensions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "married-address",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-19T05:48:52.646189Z",
     "start_time": "2021-03-19T05:48:52.589254Z"
    },
    "id": "married-address"
   },
   "outputs": [],
   "source": [
    "tropic_val = pd.read_csv('../data/tropics/tropics_val.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "surrounded-tourism",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-19T06:41:35.476349Z",
     "start_time": "2021-03-19T06:16:40.251320Z"
    },
    "id": "surrounded-tourism",
    "outputId": "9b87c7f4-665c-4c90-f081-3d6beb7f467f"
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAngAAAHhCAYAAAD9FsMsAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAADyfElEQVR4nOzddZjc1r3w8a+kYVrmXXsNa8ZYtuPEdshhaLBpmzZtU77l3jKmzDdt32LaNCklTZs0DA057MSRGde0zDA7OzwjeP/QeMF2bMcxxOvzeZ61BUdHR5qR9NM5RxrJsiwEQRAEQRCEsUM+0QUQBEEQBEEQji4R4AmCIAiCIIwxIsATBEEQBEEYY0SAJwiCIAiCMMaIAE8QBEEQBGGMEQGeIAiCIAjCGCMCPEEQBEEQhEOQJKlRkqRZ+0zTJEk6W5Kk70iSdP1h5HGzJEk/O3alHOY4HisRBEEQBEEYqyzL+uaJLsO+RA2eIAiCIAjCmyBJ0h2SJH0iN5wnSdK9kiRtlyTpaUmS/rpPrV2VJEmP5uY/IkmS71iUSdTgCcLBiZ96OQk89NBDAFx++eUnuCSCIBxj0rHJ9er9z/XWfw60rnskSUqNGJ9ygDTfBMKWZU2TJKkQWAPcO2K+CiwEIsB/gRuAPx5hyV+XCPAEQRAEQRAOz7WWZW3eOyJJknaANOcAnwSwLKtfkqT795n/X8uyBnLLvwpMOhYFFU20giAIgiAIx8/IGkCDY1TZJgI8QRAEQRBOcdIB/o7Ys8CNAJIk5QNve3NlOzIiwBMEQRAE4RR3VAO87wClkiRtB+4DNOz+dseV6IMnCIIgCIJwCJZl1R5gmpobfHbE5DjwTsuyUpIkhYAXgVtz6W/eZ/lR40eTCPAEQRAEQTjFHdWHcwuAxyRJUgAPcKdlWU8dzRUcDhHgCYIgCIIgHCWWZXUDC050OUSAJwiCIAjCKe7YvF7vRBIPWQiCIAiCIIwxIsATBEEQBEEYY0QTrSAIgiAIpzjRRCsIgiAIgiC8xYkaPEEQBEEQTnFjrwZPBHiCILwlNfxqNU/cs5sdeSXUGRZzLqvkjP+Z+YbzsSwLwwLZtECSkJWxdyIXBOHNGnvnBcmyrBNdBkF4KxMHyFESqe/lgQf6ST++C6UtTUx2E4wlaC8rIFrkp66phWA8SXdhHtvGVVEY6aenuIKZO5oJJOzf5p4YayDm8ZE9bSJX3n/RUN7/+O2D7LkriC+expPJkPa4MRWJmmg3nV4/L0yZQtrtZlJXH3PiYW56/qLXK6YgCG9txyYSk965/7neuuukjvpEgCcIBycOkKNg0/vu5tboVCScBCJR8gbjLGpZz/SBHbQVlfH3hVeTdbgo7B8kkhfAm0xjyibv0e4jlIqxsWwmHXI55dEuLul4gluWXE1Z2OA/c2fjzRqMj2QpjMbwJ1KYLicAgXScs5pfYfFnv0JvIAiWxVWvbefiTTsJzivhHXctO8F7RRCEI3CMArx3HSDAu/OkDvBEE60gCEdVvCVO47NddNX3E0rEyZ9Vwr86qpGDLop7Bqjb2YpbiXF250tIQHlbD73BIh6ccyF9RXm0e108P308P3n8L8zu3g5AdbSdu2dczfmdK7lVPZsfnX8xUY+XpTtbqW7vocfvojhqIlkSzlSWjMdJzOnj31PP5aZnN/HQ/Dq2VZVw38JpBFIZTtveRTKcwVvgOrE7SxAE4RgRAZ4gCG/aU3e28uBtbVR39OMzdWRFRpYASSL4724848oZDFjUNnbQU5pHRSY+6jY8Lzk4NPzohAoyikJJfGBommKZpBwWn37bTTw6bQEmCjc9u5Z3rV1NxuVic1UdSDIZv5tgOI5DN8k4JfLSFnnxDB9+Zi1fesd5ZJwOXqur5oyGNv55+oNYUorSheWUXFXHgitqcDjEiwUE4dR0UlfWHZAI8ARBOGyDSYOVa+M0rO7D/6t1WLqFO2ORdSlM9HuIB330FHkJxBJDp0vdKzOxoY2BggCGIjOY58NMV9MSqqRmsJ2E08PKujOwLJBNA12yg6xfnn4xK3ZvxqdnaA5V8YMLz2V3SQkAX/3P80zp6KfBX43XSoM0fHLeU1nI9D1deMw0MWcQAI9u4NYNMk4HVZEYKZ+XrMuNknXTuSZK6/r1NH/xOfJSMcKuEGV5GabdejEv/3YP6Y4YeJxMvqyahR+cclz3tyAIx4sI8ARBOMV07orz5KdX09qVoDs/n66iQip7BphsWoxvjw+9TDNSmCI9mKApVIHD1DFk+/SScnkYH28k+GQWJOioKMKQFe6ZcSV5qQgpJ0TlAKFYjJm7t+Hua+OXZy+j31HM3ya8A4+h015eNBTcSabF5I7+ofJlLAdZWcJp2l1onppZS4YsH1r3CE9XnYUEpBwKV7+2nUA6y4RIDCzIS0U5p1HDk02ztmQqjaEKorKft+15CQMZ7RILw+llWqyVneOqefn2DJFvP4NEmjZvCUXJGG4jQ9ThY86Pz6D2uilIMsiyjJE1kRUJSR57Fw1BEE4O4iGLMUZV1WeBpzRN+95xWNdXgSWapl1+lPP9E+DQNO19RzPfI3RKHiD/+EMLrf/YzotVFXT5/Sxu6+WGV1ZTNRBm0OvlhXkTkfuyjN+VGlom6VeI5jvJOhU2TS6lKm4AUB5t5R0bH+AVeTkxKY9auYPn5swm5XKwqFVjSYvGLcvfR1zPIxBLUdYXwZAMLFnBytXmZZ0O/njhbF6tHQ/AD+98kopwHICKRC/3q3PYVlVBV9DHtooiLl63jdcmV/ONB19BBlzJDM6MDoDuVEj5PVy8+0XGDfYAYEgyj9YupttbyKLOLTT4axjwhXClddrLCqifXAOAU8/yg4d/TTCRQrYUdgSrWV84mbjbh+52IOsGhYlBnIZOgRGn15dHzOXFn0rgQicvkyDl85ApCuEvc6PMq6GoSKKgNo/iZRXEetJ4gi4C5b7j8jkLwknomNw1WdJ79jvXS9bfTuo7NFGDJxwxTdN+MHL8eAaXwpHLJHSeu62J3RmZ3TWFJPcMEljZiieawhdLURBNEjKSDFRXkRfL8MTsyczpjlCfV8H2ggoWt+7m6nWP8XjZhaS9Mq6UCRZkXfa50KmbbCkq5uXxXhY0dtInFSJjUm620yU5yDcSXLB+LbvzS8ngJ6F46AyU0BMsQnc6mNDcydTdregjml2xLK5bt4OpnYOs2LqFc7fvosOfT2t+Ic1FJVT3xbljSRlJlxNZN9hTXMTpDV3E/B5enVDBijU7cOaycmQN5FSGhEsZyl6xTHSHfTp8tXIWptNOnfK5CEWiTN/ewLaptWQdTrqCheTFO9gVrOTVEvu9fA7doCQ5QHeoABOJ89rW47QMLOCRaWfQWlqJJ5KgV9axLNAHZKY2NBN4aSebQtXUDPYQSUcJu700BCsIZRIESLKjsJo57btoLS2lP5iPI2tQVCzj6E+QGsiAQyF/YoCqG2cx8x2T6N4cZvdjrUwoNTCri/BNzCd/QnBoO03dQHYMb7cgCGOXCPAEYQwwTYsHt+n8+7UUmedbMZIGVtJgbms3SYdC3OXAZVgkgj7yJBlFkenyunm1OUl7fj7Ta7JctrqedXVVnNbeh6/VYE5zN3OauwFoyA+wIpMFYGNZDcFEAf54hqxXwXRIOJMG9eXFrJtWzozWHmJuF+9fuQ5/rtZsY94srEHP0K33mrJaom4vkmViyEtYsHEXXSURsi4n7YX5hIu9mLpMKJbEkmDXxCo6SwpZtHsX52/bCUB1LEzGqdDtD1EQSaDLdm2fKcsM+r2c9+pWAM7Z2YprxL25KUu4dIN2ZxlT6UBCAiwWdmzjiXGLsfYJLFFk8qNxZmxvQPfI6FmF7QXjeKHqNDyJ9FDSgtQg4xNdlA3EcFp27aUEVEd6aCkoh1xzrWKaLOqopzrWQ1zxsjk0ifU1dSBJOJMpnGmdHk8+sayXc5rWoVsuKhvqeXTGYtJuB3v6FJ6fMpfiWJIFTW3oO6NEv/Ii62/WiDh9OJIZNjokKlJr6fYWoGBSHW4jLfvQZQchI0F98TicmSSKZeDAYmH7ThyGzqbKiewqqSSQSJCfiJKS3GSdTmTTxJ9MURPvpjdUQNLlRkfGK+mYWfBnErj0DHtKyqkY7EGXPHQHCgkkomQdLhI+L+MGuwiZKVodBUwa6KAgE6cnUIBRGqQwYOHsjeFSLJpchRR39iBbBuFp43BdPYeioEn1KxuIdmcxIkmSuGgJK1T3dFA0zsPAwmmkHS66FB/xrb0U1wVRdBPn5la821spSQ9ARQHKBdNwVQeI7Bgk/lILBd4srgofbf0y7pZefJEYmaCf7NI6ijbtwUpkiPt8tEcdhJJxSvQYsgxkdXA7waFAyEOmO4EV8uJ/5zzMjkHoj0NHP2zrwiwKIc2rxuoahC2dEHAiLxyPMqkY3jabhi+sJNqToer0Ygo8JpLbgfL2+Sjn2P09Ld1A/8nT6L1R2oOlOCIxypUYkuJAXliLfN4UzKfqkaryaffkE97Uh6O+g/K5BYTIIE8vR55bRXJdO83/3kXozCriBflgwaQlhUfUjcDc1YP5WjPyklrk2qLDWsZKZjAe3oJUGkQ5a/IbXqfwxokAbwxTVXUO8AtgPhAG/gz8UNM0Q1XVWqABuBH4ClADrALeq2laR275cuCPwHKgC/gx8CdggqZpjaqq3gws1TRthaqqvwaWAUtUVf0y0KZp2lRVVe8AdE3TPjiiXI3A1zVN+3tu/Cbga0AJ8AD2dVEfkX4c8H/AUuwm04eA/9U0LXo099fJ7L3/TvH39fYuk8qqsRQZh2Ewp7WHec09mIpEuCiPjvwgliRx55QamoqC4JJBklg3oYLqvihnNHZiKAryiK4bpYNxPJmhj4OMJZOU/UPjhiKxat5EGisDnLW1iabiPIqS6aHgDqDTU0pXQRlN4ypYsd3CAqb0dDIp1Uy+HuWxusXoLgcyFtX9YdZOm8KkPXs4u30zk8K9dIYL+fGF76ClpJCMJ4M3ZRJ3BGjMs/vlWRIYIy5Uxj4XrWTIi5FQkLDQJQlPKkOfOx8JC7CQAMXMBWWmZQd2ABZYDoWsIpMXjSMPWrxaOJesxwmSjKHIKIYJWHT6CklkExSaaSyG25Ha84rBspANE4C6SBt1kXYAvEaWZNA19JBI1uNGyZpIQNzt4bXiqcSdXiSgoH+Q2mQXlbE+XE4FOa0jIyED5zZvwG3q9HhDPDZpEVmXmya/HyWTxZnKEnXl0e8OAdBvBshPDTLo8pNRZG7Y/MxQbWPJng3M6t/DhIF2wh4/99StGG4md7hIKy6W9mzhselL0CUZVyqD5II+X4gQcZZ07GB9xVQ6CuzPJe1w48xkSLu97CydwPy2bZzduZmCZAyAmkgvGxO1rC6sBfKYHm5C7V879LmN62jjpS19zO7egIxBCDBwEURBdgcpSEdRWqDgpT08Uq2SKggQ9oeYt24LautO+wPMBfDOvgaym1tJ4MQFuDBwkwCgEgd7f5rdHY7CnV1YSMSdLh6sm0M65KJOb6G8qwswc59t0h5qk3AhQV8M/QdPABYOMrnvFkjRJEZjX64cEtJgAh7YiAHEf/UyVXoWCTAbdmLkvnT671/E+bULcX7vMlKn/x/mmlZeqZxGzD3IRQ2rkLGwkMjghqAHomnWVdaxetwMe52GiefhBi7c8xpl6Sj87Cru+0MXEZcfHt5Gyu9FdzmZdVEZl3xlKm+EsaGN1Jm/gHgGQh68r34OeVrZQZexDJPUit9ivtwAgOunb8P5+XPf0HqPvZO6NfaAxDsBxihVVfOAJ4GVQDlwKXAT8Ll9kl6PHcBVAX7gOyPm/QPIYAd/S4H3vN76NE37BPAC8F1N0wKaph3WWUNV1WXAb4CPAoW5Ml8/Yr4HeAbYCkwAZgDVwC8PJ/9TgWVZ3LVhOJjaG5rpbgdT2/uwcq8riQe8AHR43DRV5IFbGQosPvvwK1yyfheDhUHiIR9Jr/1+OF2WGAz6mNvRjwUYsozTtDBHnAudaZOG8gBv03ZQHE2yoKGTM7Y1jQiSLCbGGoiHvCRDPp6ccxr5VpaaWB8Feh9rx9XxyrS5dJYWk3Lb611ZVsCLs6pwKT3g6GNcpIEvPHUrX3/iV4xL1VMo7cbraCbmcdNTmE/C5+XdL2/CqRv4Uxmuf3ULkmkHVIMeJxGPi4zfTTTgZdXkKgxZIu7ysrJ6FhYSCcXNayXTiDsVfn7RYm6+/Ex2F+cPn/IlCVNRMBUZZAlZN0GSyHjdpLwudEUh6vLT5i/hmQkzeLx6IbsClWwtnIhuOQkMRlF0A5eRpTQ5MOrzc+rDnx2WHWyasoTpVIh7/CBLWJJE1O2nrq+d+rKJpJ1uOxAFpva34jbtPEqSg1Qkhh9AMZwOkCAjO4emmbKC7nDgz6aZ2N8xqrbRVCwmDNjBp0s3hoI7AEuWKElFaC4sR1ccyKY5tH8sSaYxv4Lqvv6h4A4g43OTdQ8HsOsqpxFMJUZtf22sa2i4OVDKaBIzB5pxMFxGOTdcnI6yt7FZwaIiHSbst4PYyb3tQ8vv/d/EgZv00BQZc8QlfXg7pRHheae/iLTD/k5WRXvZezNwoLTD0xgK7vau+UDpAHx6aijtyBIAGH9/DcswMde0AhaN+eVMGmhF3hs4YgEmRO1a5F1FVUM5WIqMoSg05ZWBYdL91012cAcgSThytfHbn+nZr0yHYjy02Q7uAAZTGI9tPeQyVnN4KLgD0P+59iCphaNFBHhj16XYwdn3NE1La5q2DbsG7oP7pPu2pmm9mqYNAncCKoCqqtXAucAXNE0b1DStG/juMSjnjcA9mqY9qWmarmnaX4HVI+ZfBkiapn1T07Skpmlh4BvADaqqHvPORNFo9C0/LEkSM4avqUMkw6S1KIRkAZaFP2q/uiQwEBn1WhGAyZ39dvAiSSBLtNWW0V1WQNrtYsXaXZT0RuioKGL3ZPsiYigSuiKRcUi44waT+kdXppYNJobXIUl0eCoYCAbxJ1M4LIvVM6bhdnfjJMzDc87CkmUsWWYgFOSpCVU05AdJGlmm9nbg09MoxKke7OSvi67my5d/kQdnrcCvJ1g/cwp7JlSzbepEFne28fvbHuL//v4E09v7kJNpOt1Ofnn2fH56wULunz0RybJY0tyF7nVjSRKN/vH8aeal3FV3Pn3eAlbOmMDu8kL6Qj4enD956DItWSbTw3tw5AIp2TSRdR3ZMpEkadSl25Blwu4gq8umsSW/ikHLCVkLCYuLWzSqEv1D+fa6g3gG0zjTWWTdwJlKg2XZQfnQBzm8H3XZQSBtB0hZtwsLSDg9Q0ktIKG4h8Zlw+4fmfB4kC074A1l46RdHkzZIuEYTguQlYcbdXx6ilBq+HOVTIuU7CA/YU+zZHlEGAOmYr/mxp0ZbrZWssao/F1GlrDHP2pa2BUYGi5IxxjNosNbMGo9e0OsjKyMmAa9rjxcuh249PtCI+bkyo+JwchlRpZ/3/719nhhKoqcu1Ho9hccMOW+huuEh8trHSC4A8goTvYGdPvmK82tIpaII5UHAYnC5CB93rx9Sihh5Wqri+OR4YVNCywoTNqflX9uIU4jOzxbsfdD8STv0LTDPefIp1WPKKSEPL/60Oeo8iBWxXBfUGN22UHTH2z4WNn7OY38O9mJJtqxqwZo0jRt5Hljd276SB0jhuPA3qNw7+1g84j5TUe1hLZqQNtnWsOI4QnAOFVVB/ZJY2HXTLYdgzINCQaDJ8XwEx/085XH0zy2NcvULe04LZOpnf00lObTG/Qy4PfQWJqPIkks3jFA6UCM7vwAWBaLm7oJ5wco6xskEIkTy7MvwHGfm6LugaF1lHf2s7amlJTbiTudxZJAyVi0TirGDHnZXZLPpJ4BMrKMO6vbAWNOq6eaeNCHL2tfZBxGlrJ0OyYSsmWw91TU4/PwyJRaO02uyXR9xXj+tmA5RUkHe0qnAfDIrPOQdIu0ezhAifo9jGvroM9VTFqSeGLBJB6cN5Up3WGSTgfV4SgePdcIpsiYikxbfoD2ohDzW3vsYESWqQoP8LXHnsZhmjQHS7lw10YmDrZRluinMt7LyioVl5FlWZtGrzufxtJqwg4fDt1u5pSHW3dJON2kQz4ky8KdSBJ3evEaWSxkog4P/61WQZJQ0lkyeV5qwn0s7NxJQ6iMjWWTchlZuf8s0g4Hc9t2kXC62VFSRV24BV8yTpcnD1OS2FlcTsSl5BrwIOpxM1BSxLzd9fTmFaMrDgzLiTcVt2sci0qYONDJxEgnJhIxxUNT0RTGRzrp94aoHuikOV9BNg2CqQTj4t3IEkzv2EOvLx/dAMmyGD/Yjdxj8vCcxVy8+SVemjCXlOykNNyD5ZDpKCqlMBljWtceVtdMZ0K4k7JoPzHFQ0uoiAnJTjzpDBWxXrbnV1GR6EeyLDoKS2hbOg+/dwoTX1tPpjeJFTdIKi52BCqpivZSKieJVJVS4negECeVsmiuHYdeHKCgq5dAIoZPT2DIDozJVXhqA/TtSeBo6UMyLRSfgwHLhZLU8WbTZCUH8YIQhdEIBXqcC1rW0OgrpTA5iIk0IhzL1aE5AN3u++mYVAgZEyOdQY5EsZJZTBTI90EqCyndDh4KvMiVebhqS9j9ShgsC3fIQZUjiWRZKG+bjfNbF+MOejHXfoH0e/7G2b3dbC0eT0OhRbUxgOx0oJw5GeXquZgPbubcinyKSmrperkTZ2s/NVMkpi6bjDR7OcUfPoPL/7CW7Q+24ptRRGJSJUiw6B3Dl4PDPudcMhP+9T6MZ3ehXDAN5ew6hlMceFnJ68L33KfI/voFpNIgzv89542vV3jDRIA3drUA41VVlUYEeRNz0w/H3sBpHLBnxPDBmAeYFgWK946oquoARrbDtAG1+yxTC+zKDTcBOzRNm3nIEp/CyoMyt1/nBbxA6KBpu1YXMPnLr/LfQDF9wSCTInGenjqOKS3dzG/qxt8fo6soxGBxHllnHw7Dbo7JuJ0U9EXIOBV6Kgro87vwSQqmJLGxtIBrt7bTOKkKwyEjGSYT61txZQwsWSLjlMkoMoGUgS+TIep2k5I9eMwU79Lu5d/zr0CywB9pxz+xnJpIH1fs2sC2kirO+ti3GXR7mdAfZUZfjPK4XUP03+lnUtbXZ7+yxDDxpaGhvBZPLEXWqbBodwfjwwlqw/bdf3JkvzzL4u9nzOLlqePAsoiu3cHSPe0s29PGF556ghld9sMlMdcuxmWG7zeK0hGKUgOc1b6BvGyccbFuKhN9bK6YRFp2Mrm7ibWFU8kqTiwJ5FxNyuT+Zha3biKueOhz++j35bOhbCKGpYBlkfW6kHWT1rxS0oqL6T2NzO6qJ+72URRNYkoSDQVlvFQ+i8J4GF80xqzYDsKePDLBQhyyA8f/nku1blDzs2ewUhkaJ0wmcc4savOyGNPHU2VZuH0y7tPqmLC8nFClF9M0sbLLycSyKB6FfL+LibltLQWmvc53aO4hvo8H65+h7jM+fZ/xCSOGq0alv5K99U35QMWIdCH2v3M9mMp9xsv3GS8YMTwx93eszDrEfLkiD+9Tn8ALnPl6iVbYn9TpAB+cdMAk5R9dQPlHFxxZIffhuG4+juvmv6Fl5LpS3L+85qis/9g4+Wvs9iUCvLHrEewHLL6qqupPsc+bXwL+cDgLa5rWmnvtyY9UVf0A4AG+fojFOoF9H49aA/xEVdUJQDt2Hz/niPl/Ax7PPYzxHPAOYDHDAd7DwPdz79z7f0AM+/y8SNO0+w5nW4TRyhaV8u5nLuTd+80ZvtR+/0u7aN2TpmN6NWdvawXAkKG4J0wgmcbf0kORQ+F3l55OR76fL//rZdrqSpEt+zTpMC28SQPFtBusLGBiayfnbN2GJ5ultbiQXnMy7TUKCbeH87et5IymNQB869k7GXR7+eqln2Dl9BUMuu3LekNhkKb8AFfs7KAkluC1yiLe/do2sKzcgw7YzcGl+cTcFhdu38omaTjUCGSySIbddy6lyHZwl1vm2Sk1LN3Tjscwmdg33IfNl8nQ7C9jXNzuJ3br4jNp9tcwr38XoWycLl8hT45bQEWkl3l9TRRl4nR4wzSF7CaoklSYsMuiIViBUpxhwkAnSbeHdUXjSSouqpK9SJJMxvTgL/ehdySonO6i6vfvYtaFoy/UCw/3A/7fRcChAwewX8qMW8brFpcC4VQnAjzhJKFpWkRV1QuAW4DPAxHgduynUQ/Xu7Cfom3Ffor2FuAcIP066W8Bbs81p7blat3+gf2AxlrsJuAfMqJZVdO051RV/ST207lFwIPA3SPmJ1RVPTe33HbsJuT2XBoR4B0jX/vxcJz+uW8F2bUzw/LtzZT2pUg7nbgzWdxZg7Pq2/BaEp59ehA5M3ouuLMphsWCnbvxZLNkFIUdFRWsmjqFiN9HR2E+P3nwR6OWX/TJ71NfWsW89g58mTQJl90Ua8oSzniM8u5OzhgID6Uf2YV9wOfm92fNY3xnFEwTcq9PyUskSMkODFnGZVr40xniuYc6iuL2C5sty6K+rJq5bY0A7C4qZXN5DRc2vkJRfJAlm7sZ57N4skrFVBQsye7r2F1difcdc4j39jFhTimyNkDxghLmfeGqA+7fQ9WACYIgvFnilyyEw6aq6oXYrzHx7tO3byw7VbbzsOx+rBXtc6/R6XTSVVGCZUH+YILytn5aplRg5YIpHZi2pRV/3H5lSCzfwfiBfqoHI2ypqWbr+BF9f/qjLOt6iZl99QCkFQf1JTW87/qP0ppXyB/v+T1Xv+fzmLKMJ6vz6ZVruX/eZP541x1sKJ2N7nYhGQaKbhJ1O7nlgkX0Brz88u//RTEtHIpFwuPFG0kwob8bU7ZozS+iKT+Ph9RpeLI6V2zcSSiZRc7qFKQGcRomdZMc7GiFXtmHN5skXRBE/cQ0Zl41HiOVwV/qP8AeEgThGDsmVW2GdNN+53rF+vNJXa0navCE16Wq6jzsfnWbsJt4vwfcfQoFd8I+Jl1czaSL7afoklmLbU0ZsvfV03hvmMrd7ZiKCyVjURBO0Jfvo35OBT4rxcLtG1g/fjrRSB8xj2dUng7DZKN3LrFCH/MHNuM2TOZ0tvDnf93Kde/5Imc27OIdm3fRnJfPpN4Ij86ZTH15EZurK5nQ0kprsBTJsjBkib+cPY/eoJerX9uGM9dsG0jGSbncyIZJc6iI5T0a/X4fJUkXH1j1MhM7+5HKfKx44koceV5k93APgtetaQs5X2+OIAjCW4II8ISDKcBuoq3AbuJ9DPjfE1oi4S3D65Q4bbIbvjCHxV+YMzT95T/uYfWtDaScMgWkyAS99HqKkQ2TxxYtQDJNigcieNIZvPEUnkQaU1LY6a9jYf9m9t6gV0YGUTui3Ft3GS7cTI4kwekiYFj402kWNzSw3TcZQ5LIehwYTjfv0razsq6GxuI8GkoDeDM6sZSb0GASCfDrCcbHW3ls6jwuuCifKR9ZhlKVf0L2nyAIbyUndWXdAYkmWkE4OHGAHAX/WPoYNEXpKC8mXBjCH09SEOmnIBJDx36fmyVLTBvciRpejy4r/GfGxawZNwMllaG3rBAj996zir7dTG3vwpId/Ld2KhMG4hQPpgjGU+gOmUfm1/Ho7InIpsV7Xt7EKxMr+MH9DxNIp1nQv5ms08Wac87hkoeuPrE7RRCEI3FMIjFd+sB+53qHddtJHfWJAE8QDk4cIEeBZVo88oEX2LMmTNjroGIgg1M3cVkZuookGkLjUAyTsN/NhEgcD1kG8/JwZ+PcqN1He6CELVVTKYn2M7ulgaTHy6SdX8UXsh+SuOuz99HxhEJzSQEvTh/HGU2dlAzG2V5SQF1HH20lXt696TmK4ik2FE/j2leuwZ3nPkSpBUF4CxIB3mESAZ4gHJw4QI6R9LYuLMNCmVSM0zu6t0g2a/LX9zxNsqkXf3oQXzpN943n8ZHPTMPpkpH2+SWOhx56CIDLL7981PSB3jTt3VlWP9lJwYu7mbS0lBmfmIesnNTnbUE4lR2jAO+DBwjw/nRSnyhEgCcIBycOkJPA6wV4giCMOSLAO0ziIQtBEARBEE5pY+G3Z/clAjxBEI6rrTsTBNv78IZT+JZW4yv2HnohQRCEY0oEeIIgCEfEtCyWf2AnX3nkeSb29iOZMVqCAVZNmcCyuR4mfmgu/tOHf+442p1gz5cepubuZ8jK+XhvuZbQh/b9JVNBEAThQESAJwjCMbflI/dzR72Hj+/pwWUYeE0DHS/twQI+d9VleDJZHr/8D6SW1eELOij527NIlosSdOql2egoJL+6ntj315EuDTLxz5exeFbgRG+WIAhjhGiiFQRBeIO2+76LK+nhg0i8UFfDA/OmMbelk+k9vUzp6mNceICd5SVc+vEP8+z//R+V0V5SBPASJUAf5VYjMg4yvT6SvXlsixfx5Z/sZnlnG2c37aCufQ8vnHYpuFyo86M4UMgvcOLwOWj5zmpee6qXaE0ee/JDWPXdFHWmKA8nyDhlBvL8dPqcVIQHwS3RUZBPRbiVLVXTiToVQpEu5ibiuCcUM/mCGs68aQKyPPYuBIIgjD0iwBME4ajrG9T5f9e+wJr8Ij6QX8OMZB9xl5MfXLicHzz4NB9//xUYskxhNE5DaSFy1uB7/34KTzTAAB6q2YSbFABp8skSQCGLTIZfLDmLZ8dN5vnqiUR7/Lw7luX6LWv5x5Qz+M+5T1ISizOhr4mgFOeBxRcyGCqnyfDxL2852QW1BLM617V1E0xmCEZizNzTQsbtYs3MOlIeN+GSMiyHg6xDYUNhIdNe3UTN0420vNTO//yrhfkrarjkwgJq5hac4L0sCMLRM/Zu3ESAJwjCUaUbJn8/61HuOltlR1kxn3nmZXwkkE0HC1raeWT+VD75xCpuWLWRv50+h+9efDZn7GpixdY9AJg4CFNFGbu5c8YyujyVvH3zRkKZNAmnxN1z5tnpZJn/LKrj6vX1NLqDTO/oZnxPjKTDRY9nHAWDjXhiBjGfSaPbSVaxfwkj6nSwJ+jjbTtaWbxhN6YssX56DSmP/eLjrMtJRJZ5qCgPS5JYc8HpTB+MsTAaZ09RATt3w/bP76S8r4cdlUUEUmm+/a1a8peNPyH7WxCEN0800QqCIIxgWdbQS4dNy+IDV6/mroVzyL7zIqr7IlyycTtT+7tRsPDrGa7buIlNFZXcsGojALokAxB3uUblGyfEF867gZ8vuRiAv85fwLO3/YGwI0AokWbQ5wbL4oZVWwkRpygpk0l4STpzv05hQJgSlq3dTp4cwztnGi+VlwIgWRbVAzEWb9iNBCimRV1jF+vmTB5af6fLgbX3ZcqSxHnrdvOyOoWIy0lpJsvGidXcsHI1y9a1kAzAfz62kW0F4whGdPrzioi7oX9yJSvyDYz13RT3tTOl2s3Mn52De5Ko+RME4dgTAZ4gCG/Yb9bpfOIp+72g52zdzXf+8zSBpI510RLSLicAzSUFLNrdjDLiZepz2jr4/sVnc8GmXZRHY8xs7wRg3bgKbjn3dD70wlrcWZ1JtPDY5HcNLbehopLugIfPXXIxv/jb0xSk0oSSaTyGQRw/ZFzgHL4D92azyICbDC5T593rN5NWFNaWVbCivpnqdBpLkpByZTMliZrObjJOF2V9A/RXlEPIfojDYZg0leTzSqkdmLVlnZzX3ovLMEkj44rJSIlS3rWlBSdJttY2kfY6eC2vmM1hB+UJmVlb2qnYWM+aV9bwgzOuY0brAM0FfjJBiemRbi64spJZRRbZxRMpqwlixHQceaODXkEQhDdCBHinMFVVnwWe0jTte0e4/BbgO5qm3X2Uy6UDKzRNe/Zo5iscOaMvSXNLjM/+vJ0pDT3cctkycNhNnjsqSvjJxcvoyQuQleVRy704eTwbK0uZ094NQF4mwyeff4Uv3nA+9916Bxc39fPqL37Fj849l0kDPXiyWVI4UTBZ2rybrSUVAEzrbWdCfBtnt05AloOUxJLA8M+MKKY5tE4LC9k0sYB75k3lpvXrkIAPrllHH7tRck0xHXl5uJI6MY+Ln1+4kHl6hmAqykXNT3Ljxh7umXgu/62ZQnlvnMcWTBnKf9Dp4PrnXiGNgokMWMxu6qCpMo8VPa8xq9Eum+wN8fLk0/HoaZYlV6Fg8r05l/DI1MmsK4vy8/ueI+nzcP+Zc1m9xkGj1036+TRycpBzdzbwP6vW4TVSbCouQ1K8hP0uXIZBKJbEkTZwZVN01PiQLQe+YjfR2hqWf2AcdSvK9vspN0EQTj0iwBMOSVXVWqABqNE0rXXvdE3TZh4qjXD8JbIWLVGLUi9EM6CbFroFbgVMU8IESv0SGd0iMpDlxad76LhnG/GoQUnfIDsrSkj6PbhiaQY9XmZ1dHLTa6+yq6KcLw7I6A4HP7virKH1zW9u5yF1hj1iWfafJIFl0RkKcfHHbqLlGz9ib+i3ZE8zeZk4LtMAYHK4j9/dez8WMlG8ZPHTSzG/fPwe6vo6SDstPrTuWZyWSXlfhkA6w4DXRZk5QFE6QRonzqSHECavVE1kcUMzLstk5ZTx3HzJWewqy+cbK5+mPBMhi0KcAP1BP4FsmqbqIh5cMoeOYIDyaJwbVt/LlJ4GWpnCkvo+ltSvos/vZ09lEbtL8wGYEO6lRmnm5+9+B+e9vAXFSFEaSbBtUhFXtieH9osv1kFMlhnwO1AwaQkW8tsFKwC45d6nmdodBqB4MMofL1nG1a/tYMmmJhxZg2IrjsNQyOInT3LQ7fRQ2xRBtgxMRcKUZHTZT0mThcvIouxJUbdpLesfnsm9NSV05fnw62Eu37Mafwp+tvhiNlSVUZrK8OVnHqMkGqXVUYYr6yDtUOgr9BDSB3G4DPKiBt5klkhdCT5nFiWaRlEUDNOLlTUIpPuJnVZCmiBWRiEwPUSZL4VrdgmkwTM4gDS5FKkogOR0wBNrSHTpUFkME0pwhtx48iz2vNSJf7yP4ngGqyCIZ0Yh1su7yKzcjek0cE4tg/J8Yk0JvJfPwZccANPEQiEb18n4Q7itJA43pO5dj+HzgsuFHE3jmhjCMbkYs7QAI5bFEQkjNXRDwAvTqyGcgKoCqC4Cjws6w9ATgT2d4PVgIhGPKHhr/ThqC8G0IJEG3YApVdDRD0Ev5u5ujCw4ThuHlM5COIYR05EnFCE5FQjHADCcLjAtlJD7DR/P2d4UStCJ7FZeP5FlQc8gFAVAOUi6Q9EN6I9BScg+hk8hog+eIAhvaY/uMXnb/Sa6uc8My7Kru3JVXgE9Q1pRMGQZUy7HuaSYLz/6NOft2M1VG+v5pzqHL1x3CQCV4Qi3n3Ea17y6nYbTC8mPJu1sLAuHYdAd8g+vR5LAyAV5WCzZ08S81k5Wj6/m9CY77n926kScVma4aICBTBYnJgpBkoQpY8AqZn59hgKllUACmpVJzKg3eaXaz6YphXztmRcB8JIhj0HWFZYxr383fYECXLrJnNZeFu1p55JNDWQzQbpwY+ZO4oXROACz9rRRX1HM3csXUF8YYttF72TVX75DVB/uJ1eQSPDZZ9Yyu7cNvAn+Z+3z6A43WaeDR8+ay3avh0ktXXzyidfo8OdREY8A8PDkuWwI+Njkn8T0aadz9p5GztrezMuTKpnUMzCUf3l4EBwKlZEEoWQGCQsPxtD8UDRJNOOiqyxEbVcvzqzJoMuNpMg8MG8iL08ox3I6KEkv5YrGV3nn1gdJN7oZn2rBQ5Y7ZpzJP+faNZDbLYvURZfxrfteJJTKMqA4efy8ufQUhVBy/Sl9iTSX/Hct6U4Hs/u7kbGI4aaCHRTTTYoCXI81kEWhkfHEnvYSp5ci6nETR8LCAnqoxUs/Og7Wy6ejSzEc5m7cloFCmqgcxGvFKbZeJoufCCEqaMOLRJgQm5jFIu6gCAPrG/a+MJEwcOIiS4JC+slnHLvxAl2Mp5sJAJTSQAHtuMgyuk4ZQMF+YtL+5lFVCG39o76P7dTRRzUSBuPYSj69w4vn+2AggYULGZAxSfsCuJwmUiSJgkTG4UP+7AU4fv4AlgWd0kT6KafiF8sp/uTc/Ur0enZ85CU6bq3Hke9i1iPnk3dG2f6J4ilY8R14ZSdMrYRnb4byI+jn2dQDZ98MjT1w7ix49Cvgdr7xfIS3DBHgCQCoqno7sALIB1qA72madmdu9obc//WqqlrAjzVN+66qqo3A1zVN+/tB0ljAMk3TXsyt52zsZmFHbjwI/Bq4HIgC3zxA2a4EvgFMAjpyZfvHUdz8MeMLzx0guAM78DKH+8LFFCco0tBdetbhoCY8SCiVBWQenzncJNlekMfyjl7Oa9hJzZoIM7/9Wci9C05XFFZPHP71iYJYgrxogsayIpAkNkyo4if3/ZeSwRh9bh+3LlO5/UyV87fv5Kr1m5EwyeLCwEkaNxYKCacDX9bAQqEgkaQ0K9PHRHqMEgZlN/ctmMrE+IgLLmAiM6u9l3ZvARlMihJ2Tdr/u/spvLlgMoMLD0kyjK5FCSvK0AMVG8pq+O+EOczfGSWGfZHsLMpnwOXnutW7mKZs5JnyJbgSCu9+/EWeXDCDhN/HlM4BnGmJTy7/AKVSN62BIq5d1crHHnuA+upSnqtZRvX2Ym6+7xW2VRWxqq6GZTua6Q36eXrOZAxzb1Bs1ySkceBGByxSfot/X7oIU5Ep643wjkdfRXfIbKos4m+Lpw1thxJJ8flXHxy1VwC2FVYOTbEkiWCigx0VRcxv6GCw0EdPcR7SiIdlEj43LTUlzNzeQqsvn9JEHJcUpdpqRceFK5evE4NCwmTw4GcQDwn23kFIQBGtKOi8Ki9Fl+z+hLqs4DQMwI1kWSQlPw1MZqa1kQAxyIVjBUSZwHaUXKC7t25FxkImC0A+/UMBO0APw9/DXmooo4n9SSNyyw2PCO4AdFz0UZ37LBS6GT86wBtIAPKINcu4E4O5stvld+kJrJ8/CKZdL1RuNdJPGZ1feJGiT8w5rCb0+LYBOm6tt8s0kKHp2+uZ898L909498t2cAdQ3w6/fxJufvsh89/PLx+1gzuAZzbDQ2vg2tPfeD4nrbFXg7f/zY1wqnoRmIcd4H0HuENV97a7sfeWc6qmaQFN0757gOUPJ82B/AKoA2YAc4C3Yd9iA6Cq6vnAbcBngELgvcCvVVVdfpj5vynRaPSkGi44zBYgeW8z6gierD40PK+lY2g4L5Hiqg1bmdrViza+kqzzwPeF+fE4T/3qt/QF/HYAKEkk3C42Vlfg0w2+f+HZLG1oZNMPbuGdr22gzR9Cxw0oODDoz3Nx042X864PXc1zs0uR3EkenjOJjCTR7QvgleP4rDSlgwnunD+bF2tr0CWJtBOSuHFYJv0BN4o1HOEq1uiml7DPxZ+WzKI/4AOgsyiP+tKi4Y2wLFb7FnLb2StYNbuOl2fX8fdzFjHocqJgMSCXcXbbFuaHdzCxo5crX1zHl55YyQ2vrcVA4kNPbWd1aBpzdkSZ1dGLbMH0lm6uXL9raBXT2/pYOWcqP7r6XL5+wyU8M2cqYbeTZxZMxpIALNI4yALjaOHVOZMwFftU3VWcR2dxHrJlEfGO/rB96cyo8Qx2UHX9jtX4M/Y7Bauj/czubUXOZrEAXzyNU8/u91mGokksJFIuBxIWci6okxh996AP1RHsf3E0UbBgKEjbu3/31p0N9Z8cmj86D51DNzXqI+ooXLn3JgI4SR9y2dcjYyCNKLOD/ffPgVmjx1zDZTNy5ZRCzqHg7lDHshJw2Ddhe/MLSAdMn3TvcxnP9x9W/vsOp737HNf5viPK51gPHysW0n5/JztRgycAoGnabSNG/6mq6ueBs4Gtx2qdqqrKwA3ApZqmdeamfQm4akSyTwO/1DTthdz4alVV/w7cCDx/rMq2VzAYPKmGb79Y4cr7DfYMgM8JiSykDAALSbYjZ1mCWfkmhQ1dvBooxJBk0g6FQa+LAZ+H/ESaj698GYdh0FRUwJ7iQpqK7NqsKV19yLqJ6chdVCyGrsuXbtlGzOUm6hx++jOQSjOnuZMMCkuamjh7dwMAZ+1pYE9+IY9MH8+i3W34Mjq3LVJZNcmuhfn0lVfy/M/u4B1rtjPoClGQsC/Ybga56YUNZBwynXluYnlZpg00kUTmiyuu44+nL+Ws7U389J5nkQFDsSgxOolJhciOFMlEMR9bv5qMx8mTS+ZhZp0sbesl5XbQnB/CKcFrC6fitCzWjTfo87gxZZn6ixYz8+4GQmG7ZtBLiiBRrKjBPOMFDGQ2sZBeq5wf3fMyeSQYqSjTR5IQYAcMhqLQWJw/VHNoyjKRoJe7Lz6Ndz6h4dKzVNGFE53ySHgoH0U38Gfi9BcHOHtPM4/OHMeO0gJKYgkkw83LVXWc0baTPneA/138Lq7cuh1HRuLe+39PxqVQlIzx5cXv5JzXNtLr8eJJ6lz19Es0T8kj6fJimQ7y2g1KewZYP72Ky7auI+lwE9Xz6aGYEnpJImPiIo2HPgrJIpHCS5IAbmJIWGRx0cwMPPQz1dxEVnYSJ4Bs2eGdhygmfgJWlGqriU4mYCJRQTsWEmHy6aCWUjpwkhn6qllI6HiQMIhQhITJ3pBxHJvpZBJgUUYDWZyjgjNpRC5DTbQlQVhcB89uhpj9HZMxGMcWuqlFQac61Ap4IZEBRYbl06FzAKs9Cn0xwCKxeB7+Sg/GfzdjpU1S1dV4//Qu+ObdGPEsXXotHncelb8cvi891LHsqQkw9baltPxkE+4aP1P/39IDpve+6yzY0GLXuJ0xBf7ngsPKf99h99euhcY+WNsA1y+BFXOOKJ9jPSwcPhHgCXsDrZuB64Fy7DOgHyg5xqsuAdxA44hpDfukmQCco6rq50ZMU4AXEPZTVyCx5f2Hc1g7YUSTVkdvhtapp7NFX8bkoEV+noOKBqjd2M6CpzZx39xprK2uYFJvP59+8nluuWA5IIEFMzrauXrDJi7buJVfnrnU7oOXNECGP/z5PioHYoDEpJ7hprDN5WVc8aH3cuGmXZyxzQ4AP/GMxtNTJ9FQUkDG4SDpclCQhEBmuGZKAubpW/jjk1uRczU8O5lKm1zONc90Mmhu5+4zpvGBGy/iB/c+T0E8xbqy6Ty3ZB7dXg+tHhf58SQfe/pFlqzaQRwfuiJTEI3RXF1Cd0EeaYdCxOUk4nRg7n0qWJJ4deJ4Zq5pHiqLhUyVYb+cWcFkItvppRwL2F2QT3U8jsM0aKks4qrmu2lhMhnclNHGJa+W89fzVXRH7rPK1Wztri6hqbSAYDJBWdiNjsz09W10ObeR9TmYs7uBQStIRUsEV9Dgnn/eR0soxNaiSp5QJ/O7qefxTNEMvJbBgkgXL0+eScVAAm3CQvzJQSoHM3xx+0ZC492EpxcR6wrjHhfigj09KF0NJE+vJf/GKchLlnBBmUy2T8UqCeLxKVimQbYlgrciD088DR6ZSp8HSZb2a3J0AZNHjC85xLdx5K8K28279p99fzc83d7Xtn1PTh6g9hDrORwSdjNG/mGk22tvL9S9ZRvanhWzUICaIyxL+XvrKH9v3SEKIsFP3mP/vRk+N/zjU28uj5PayV9jty8R4AkA7wQ+CFwAbNU0zVRVVWP4G3+gXl37er00MYbPfwCVI4Z7gQz2eXl3blrtPss3AXdomvbTwyiDcIQqil1UXDmehSOmffp04J1l8MP5fB6AqTzxyiDWbQ0UxRP05d4Tt7Wqkq2VFXz/gvO5+YEnqO0L0xX08+nnVnF6Q9tQfgXRFBsqy5nV0cV9s2cSdzpxZ/Wh+hSXYTKrrYuGkgI+sOo1KiP2E4h9Hi8+M4M/o+MmSjGNRBmHmbuctkulQ+t4+yo7wNNlmeK43WRX1pVg5q4Gnlu2GFOSiHi9/OGcM7n5nidJYPHU8unsHF9CQa6J2mValCRSTE6meK2seCjI+/niM6gc7OOCXVvIIpH0x5kYG96+DG5MCZ4/YxrdBVmm1/dTkLbYNrGC6elaZnXtAKDRV0l1qpX3P5Zl7ZRaevP9KLKOU4dQf4zN5QVcvraLRm8+jaUluJIm57W1s2FaBQNKiFI9zmszqig1kwwaLkI3zefjn57BRxQFmIrbdfR63ij5w8MSMvKkYpxgVw8LgvCWJgI8ASAE6EAPIKuq+j7sPnUP5+b3YAdwdcDrvQLl9dKsAd6rqupK7OBuqCZO0zRDVdU7gW+rqroZSAI/2iffX2D3B3wFeBn7Jnk2IGmaph3JxgpH7oLTQ1xw+lyuadX58C09JCSJqMvJuPAAHaE8/rZwHs//+o/IlmX34cKd6+hrUR2NcOekeeT1Zbl31gwwLe5aOAvZMPna4y9hSRY/eegxvvXfZ6lL7MREQcfFptKJTIz0Up3ZhZMUEhYZ2UQxLSQkXGRJ5x6ciPjcjA8Psnx786hyn7t9Lb9ZtnhoPONw5LrYW2ysKaY+4OOc8ODQfF82i9cwmNfTT0vAxysFefQ6FN5++bV8Z20hX3v27xTHIEI+MhZ93iA/Ua/iv9OnMCuZREEmWdRJRyCP+vJifhG4ni//9z4SsodfqecTUCzGzw9ybibJtOky0z80E9k5MjA7Z799v2jE8CUH+GzexMsxBOGUNxb63O1LBHgCwF+Ac4FdQAL4GyOaQDVNS6qq+g3gLlVVPcBPNU37/sgMDpLmE8CfgX7s/nx3YAdte30a+A2wHRjEfor2yhH5PqGq6oeAnwJTsYPILRzgaVvh+Fla7WDrzytGTCnCsiz+cOET/HL5Et7/6hryU2kCJIa6nrcHQ/xhxSIG85105IWGlnx0zmR++Ph/Uawsr1RMpDrRiicZJWxVkMSP2tyLlywK+tArOJqdJbwwezwrNrUSc7pxpC37AYU8he8+9AKbKktpKg4xsTdMPoMU6TE++OoL/GnxMtxZnXe+tI7G8mJ6PCFaHQ6aPS52e91MSKbJj0cIpXR0p4tQNktxIklvSaFdWElic940/jH+OrZOK6PbWYSpyJQMNBHxu7lq6xZC84o59711LJ53DVgm2ztNppRPoKN/HjNLHPxJHnsXEkEQ3noky7IOnUoQTl3iADkCf3j/CxSt3E3V4AC14TC6JHPTjW+nKBHjngVzMdLm0J69cPsO/vrPfxOXXEz87pcA+PCzq/nsU6uH8vMTw0UaN1HcpFjnmcM1H7mWf975NwYylSBJ+I0Y09I72JU3jscnTyEYl3n3jpcpSNlNvVnF4HeLltObX0VBxiTh9wLwQsDPqiq7mVcxdF79y6/JuP3cu+A8sopCb9bg0XEVRDxurlm3i0t2tLNm7jj80Ti7J5TywRsLuWjFiCdxBUE4lo7JHVJC+tR+53qf9auT+m5M1OAJgnDUfeT2Zbzw60L+8Ug/83a0kmdk6A4G6CgIYigyeCTQTa5Yt5mfP/gYJjKKJTG3uYMN4yrIT45+5YeFjIVCBh8Z/FSn+vnfR1bx8bddzWmtrXxo9SoSjgK+pl7BMzMmsbixmTmRHnqdBlFXKZYEnUoB03cnWFeXJlFk1yBWdPXxwQ31tF16FlFZ4sLmZp6eofLYxDrcus45jU28q2EjP3i2k5ji4JW65dzYfBnvV8QbpgRhLBmLTbSiBk8QDk4cIG/Sqto/UNjWz+/PWcRdi+bTlW8HV95Mhsd+fhuT+u1fftCRGHA5eXxeHcWDcZZu7wRkZAy8pJAAXZLp9vlwJA2mPHQJyaebaNW9rHHHKZ2tc8V73jZq3YmEjjscQ6oIkWmKkjEkIs2D3P/9nSSTUFihMPeaKjI7e1nZ6WTytCDWK010NmfoLS5h/NQAV99USWBcALdXFr/xKggn3jE5COPSp/c71/utX57UB7wI8ATh4MQBchSkd4fRvvAcgae38ctzlxPxejlnRz3Ld+2iIAIZnGRRyKy8gZlylN4fPI33ybW4TYN2dwlSJg/Lktg1rYber57DO95dNSrYeuihhwC4/PLLT9QmCoJwfByjAO8zBwjwfnFSB3iiiVYQhGPOPamAM/9zJfeeneGbjzxJ2uHgLwvmcsvSs/nebQuZWOJCkoebPcuWTwXThJ2d5Gclkv0WZfOKmHAEP9YuCIJwKGOxiVYEeIIgHDfXPDv8G5k/OFRiWYaplRQAR/DT6YIgCKc0EeAJgiAIgnBKG4s1eOJRMEEQBEEQhDFG1OAJgvCW8eANK/E9tYtgIkr9hFoiPi+vleVxmbePt/9TPEAhCIJwuESAJwjCCffIFY9T88QmSpwOVk6vIxKaQSjXwDA5ZvB8ykHHlQ9y0afmU1XjIlBXdoJLLAjCWDIWm2hFgCcIwgl139S/cdauZ8m64IrrvkafPwiAP5slLxXnE9p/GS+bzO9qwHzxKW6vPY26/mYWte9CUSfgfubmE7sBgiAIb0EiwBME4bjLdMZ4+Xsb2fF0K+OTKeJyDYYJhYnUUIAXdzqJO/P55cILeOmOLw8tWx4PE8gkcZomvLSJqO+dzPDlse03K/Zbz51rEtzwjBNZN5AAQ5a4/0qZt011Hq9NFQThpCBq8ARBEN6UF4J/Ju1Oo/tkts2Yx+X/fRglN++Lzz7Dh667flT6mMs7NNycX83G8pmUxnt5rqqGf884k8nhDq7YvorCb22B91wHwM3PpPnhKhNDkcmPxykcyCIB/XkernzQw/K8OP95j48i79g7qQuCIIAI8ARBOE66btvExi+/ymCJg/7gOPrzQowPD4x6lF8xDBY2N9Hj89NYXAyWxayeLiLuIKYkc9e8a9EVu/bttdIi+n0hVvtCVA/2cu6u9bRd+BO+c/G7GFy1G2mOiuFwE4oZQ/fmBZEklbFOXtCrmfm9Ti7esobo+BrqLp/KZ5e4KfWLgE8QTkVjsQ+e+KkyQTg4cYAcBtM0kSSJ+J4wxmCazt+uI1XfS4PlJ1ufIiG7KHUOEg0EafF5mN3ZwcyOdlJOB8FMBgvYU1bBsxMnc8f8ubh1nWBW58urHmFW1w7aQhV4swken37R8EqzYTAjrC+rw59NsaugjDNbt/GFVfdSkIzh+PE/QZKo6hikMjLAbx/9HdN6W/Fl03z9onfww/OuxpdO49QtsrKE00jjzWYoHJ9P0Ehi7uxmujHITZeVcdb5407YvhUEYZRjEokNSF/c71yfb/3kpI76RIAnvKWpqnoHoGua9sETVARxgIxgWRZ7vvA02+7ag+VwkPVKxNIOljS2ECAFWGQkmf/Mncv5u3dTGh2kPVhIhyOP+QONPDh9Bq+OG8fNTz5BV34hBbEowXSKF6fOZFv1eABiisVzlaVctnsLX33hL0Pr3l40mWemnE3YV8Ck3t1cv+E/SEBacfG7xe8h7QrQ6/MTSLbRUuDi9sXn4dBNFN3kFw/+hQ+veXIor4TTif/7fwdJ2ncD8WbSZBQFw+EEy0LWDWTT5EMTkvz2XUXHYS8LgnAQIsA7TKKJVnjLUFX1WeApTdO+d6LLMtYk0gbRpEV0V4TtLWme63LQ3pQksWeADk+AMxt2UxeOEPb5SGHSEcxj0O3mU6+8jGRa3LpwMX35+Zzd2oHTNJln6FR1deEzsqRw0Ug5DiXLxnm17C4ppqq/h9LoIACV0X7MEOworeD8nTu5cEc9m2tqWTtpKrJpsmL9avaUlg+VNWBI/K/2PIaeGLUN5YP9vP+1v9FUMI5x4Zahs7zbyDAp3E5VpJWvX3QZj02/lPH93cimiUM3KeuJYyijz9N9vsD+wR2AJJF0e8CymNnRzDee/Dcfue6jRLx+ftfm5Hc/TuMydCwDCqIpDMUi4vcjAU49jSSB7nCQdjhBlinOpilIx9kdzAdMKr0K75slM8eXwpAgmOdhZTOs6YKgCxaU2XcUacNibavFlDKZRaUWZT44vUaiOwGVAYm0IdGfNPE5ocRnN3I3DZiU+CUCbhl5n22zLAvpQNsrCELO2Ds+RIAnCMdAR9jga3dF6Yla/M8FPlr6DB54LcXkcgWnIrG1VeeieW6umWjy5G8aQYILPzGBiimBoTyaegy+/s8o29p0YkkLyTDJ5k5CM/vC1EWiNAUDrCsptBeQAAsCmQynd/VhSBI7g14U08QEOkN5yJbJBfXbaSgs4ptPPUlFdJB7Z83msRmzeH7qDIpicb648klKYzFeq67hxfETuGfuXPoDdrmm9/byqeeesZ9gzfGRZirNbKqeQDg/n8KsTiKQT78/SGE8CkD1YD8MDu+fOc2NrJ04BVOW+fZFl1KaSjNpMAZASpbZXjadokgradmD20yRlZzIhoXXTDOtZ+eofa3LCu2hcuZ2bOLbT/yLx6bPp6mw1J7ngMGgzp9nn8+ypq3M6G6jPVjARR/46nAGlrV/sCdJbKkYx0ev/QgRr9+eZIGlKGQUBSyLbo8LV8YACbIuB1nLMbQslsWShnoe//MPMWSZL1z6bm5bdC7prgEiT6xi+qtP0RYs4KIPf33Uuh9qHCoAWPBEx3DZimKDRFwedJcLLHN4OcsYUXALMIfLMZJlIVkWN61+mjvnLyPpcrMfaeifAy5bMJBiIOTBdLz+jyCFomnyBlNYLpmBEj+6IpEfS1OazDBhvJMnk16cCvz5Qpmrp8h8d5XJv+pN1DKJ350v43GMXv8zzSZffM7E74Tfn68wvWjsXYgF4VgQTbTC61JVtRH4E3AesBBoAG4AZgLfBUqAfwMf1TRNV1V1DvALYD4QBv4M/FDTNENV1drc8jcCXwFqgFXAezVN61BV9dfAxwAdyAJtmqZNzTXRKkAKuA6IA9/RNO0Px3r7c47oAPnIrRGe3pQBwKGAbhw43ZWRbgra7cgnv8LNx/922tC8G341wKs7swdcTrIsrt3VREDXeaC2mn6vZ2ieWzd4184GAHo8bh6eULPPFll868knuHBHPQB3zZvP/1u6bGj2L++/j4WtLfbwmUu5e75dJtk0eeLW3+PT9QOW6bnps9lRObyumvZmztizC8Wy8GfSo9KmHQ7+etYFALxUXszuvBAz+geY0dXFxL5eegpLMUhT1dXMjO4WxiWa8JhpAunE0F2phR2KNOVVsaN4PEXxTnaWlPKti27i6k0NyJbFfbNq6XJ58KR1OssCjA/HGHQ76Q142c/ec+E+gZEnk8aZhmjANSKoOkBQuI/Pr3yAm5+8hwWf+RH1pVWj5hXFB6nr7eSV8VMOmse+JNPEkt/kL0yaJp954RF+cdYR/DKImQssX2fbnVmDmvbBoRBx0O+ip9iPZFmMb42gmBZtZQFSHidOGR64UuKS/wwfYt9fKvPV04e3Tzctin5tMGgfSiwsh9XvFvUSp7hjEuGHpS/td64vsH58Ut9NiN+iFQ7lvcD/AAXABuA+4BxgLjAbuAK4XlXVPOBJYCVQDlwK3AR8bp/8rgeWA1WAH/gOgKZpnwBeAL6raVpA07SpI5a5FngIKAQ+CfxaVdXxR31LDyAajR7RcCQ+XMP1esEdwGB6+JySiAwHc9FodFQe+7IkiYxiH77KPjdpuiwNRaX+7AGCMUlCHxEk9Pt8o2b3jRiPjajlMWXZbnocWQ4g4nIBMK29BUcu+Iu4nOyuGsfdS8/lrqXn0lpQSMzhIi05yeBga2gcJb2dSMlBGoMBzttZz3tXPc+nVz7J5RvXcf7GNZQMtPC98y7D9CQoT4bJTydQgPunLOajl3yM36qXYCLRHgjwmfPfyduv/RJfOfd9XLGlEX9Wx6sbXLW5AQyD7iI/73+tnpteq+eTL21mVnvvfrulIBkHScKdzQxNO3fnJs7fuZG8WBpfcvjzkQ4j7E+4XGwur9kvuAPo84dYVzXh0Jns46jcjssyivn6362DOkhwByCbo59FlE27xJYkYeVmKIY9LWtCc//owL8vac/bexxlDYhmhre6JzFc7iM9NsXwyT18rFhI+/2d7ESAJxzKrZqmbdM0LQvcCUwEvqZpWlzTtGbgWUDFDugywPc0TUtrmrYN+DGw78MR39Y0rVfTtMFcfuphlOEZTdMe1DTN1DTtP8AAMO8obNshBYPBIxr+5MV+/G77BPHOMz2cMcUOjIpDEuX59mG3YKKD911fjCSDJMN5H64dlc9nLvPjGllZYVlDtUy1kSiF6Qz9bhcDbteoMlfFE3tba9lUXDBcM5X7f1ZHB3fOm8/G8nLCXi+pXJMjgNMwmNzbM5TXaW2tQ8HA8t27KEglh+YNerw8Ofs07ps9B4CyyABvf+U5LlmziqruDlwjopG8RIJBPUS9NZ7t1KL0y6zYsI0Pvvwiv7r7Aa7ZuIlFzS1D6av7e0g4i5jT249zRI3huvKJfOH89/HUxHnccvrb+Mfss7hz1nLSDnsfyIZEWhoOAgLpFIMFHopSaWrDdhOwYsHi5uFtBLt/X9hnN0OnnS4WN+6wP7v1L3HrvbdSHuunvCdOSW+c4r74YQVa98xewq6iMvKS8f3m1fW0c/G2tYeRyzDZMJjb1viGljmQuS27eWjmAQ67Q7XmWBb5A0nkg9yxpF0KUb/9WRiyRDjfrlkORtM4DAs54CDus4+Fj8+TuHGul2XV9rLjQ/DJ0+xjY+9x5HVKfOdM+y2JLgV+tHz4gDjSY1MMn9zDwuETdd3CoXSMGE4AhqZpPftMC2I3uTZpmjbyKrE7N/318ovnln0jZXgjy50wZ05zser7RSTSFsUhGcuy6B40yffJSBKE4yYlQRlZlpi/oggkCW9w9OG4Yrab1T8sIp2F/phJUUBia2OK5+qzXDi5kKamAL0eD5cWS3QlJDrCBpG4Tkt3gMhptZTJFu4emBJL0ZlxIMcSeLM6Z+7eRWEmw8+XnkVjQR5l0RjlAwPE3W4MReGH55zHhK4OauJJymKDfG7lSmZ3d1ITDpOVJJyWRdLp5NHTFtMZDKHl+7l+w3q8uo4vncafTnPRti3ct2gphqLgTacJplMMEhi1fXu/KDPCHTgwGVmn1JFfRDxUzIKefv562jVMWPlb8tIJthdVY0nD96VNeSXUDPaOyvOybU+ztVLFkiQuqH+Kr1x6JoMeSCsybsNeS8IpjWpm9WUyjLSkeQev1k5hU/k4Prj6GZ65/ZvccvoV/G3+WbQVh3L9HS1kw2BaVwthr4+eQD66w4lHz6I272RXSSXvvfYjGE4XDj1LXjJOxO1FVxwMKi6eGz8FyTLsmi1TQjINvJk0smkS8wXs5lBZRsKirq+bYCbJppqJFDogpus4slksp0J1NEJxNsH2wjISuKkJwNQiiXgWfA5IGLAnDMmsybRshMUraoi4XJxnwp6OJIWpOKn8IOdNdpPvh0099q551wzoSoDPCV0xiKctXLKXyjyJ+n57X2cMSGVh6TjY2Q+6KTExz08642VuuUwkK+N1WEiWF9l0UxaSaRy086wI2J/js9crdMah2AsuZf9ak68vkfnoXAmXAiH3yV+rIrxVjb3vlgjwhKOlBRivqqo0IsibmJt+uI6w3eityeeW8OUuSJIkUZanDM0bOewNvf7PZgU8MgEPFAXti+HS2X6WzrbnqfOOtGTTAPjA686vBuYMjVmWRVtrkjxJ5+lfb2X7fzsoNCzkTJpgXyefW9+ER9exgCwSLiwKEnGWb17Lg/MWUJyMEXV7KE2HieMlhYs8YiQDLp6bMoeM4sBh6CzaVU9LfhFpXWbnlOHmS6/l4M/zr8R0hlhdWkp+MsaAN0AgneSq7a9QGg+zpayGtOLESYasQ+fTL/4OgAdnLAAg4XLyF3UK//vc88zqbOB3i985qqmxobCUpXu20ZpfxBVbXuOJyfZO/tXSi1EMneUN2/GaUT6rxnnnOV4KytzgdABOoG6f/efiwBXMI/v9lR9g/r5pRtr3PskB7O136ePwlewzHuRI7pWWH+C1gKdX7j+tbNSY/R2eVDA6jSxJVI6O/fdT7Bt7F19BONZEgCccLY9gP2DxVVVVfwpMAL4EvJGHITqByUe/aMKbIUkS1TV2EHHlj06HH42e//zmJI98bzV1mxoZMCx84Th5AynKegf4xFNPsLW8jKenzaSuu4vycB/+VJqwL8SjCxaRHdGn79HTFvNEdSVtIT+XNrZSlM6AZTGtp5HWPJ0fnHcG7qxJeVeM2nAnn1/1ANN77del/OX+XwHw5YvfxcPTL+HeuYuRLIu/LVhuZ25ZeLKDfOol+716U3vPoq1g+J12liyztbSKQCLGn+edQcyfZ9egWRb/XrScnz101X6vHhEEYewYC33u9iUCPOGo0DQtoqrqBcAtwOeBCHA78H9vIJtbgNtVVR3Afop25lEvqHDULZ/lZfk/zwLOOuD8XBcrfnrBs3j7Y+RFdGIe16jgDiCrKDTn27VJj9ZWMzk8wHefuov5PRsBWNq4hXM+djPeUIZ31ke4aM9wH7YBj49bll7Kabt38MykPHaWVPDClGlDtXRuXSftcBJzuQlk0jx0+49Z9MkfsLuojBmdzViyzGXXTOI7ywoPsAUHeJ2IIAhjylgM8MRrUgTh4MQBcpT94NxnscImFHpGNZPuCgXoMRVufHoTvnSWjVOK+dCef1KSHH6BnucHf8cXN/jMyxpfev73uA37ydasJPNq9QSWNv8UgP939b/55uzzGQj4kS2L83Zu4vKtGv+dOpeLt6/HtGDxZ5ey6O2zju/GC4LwZh2TSKxX+tp+5/pi6/snddQnnqIVBOG4+uozZ/O+u04jY5lYlgWmSSKb5oWqct737CYKYyk8WYNFWzpZVTFtaLnXqifhyMB7XnuOz714G3GHi4TDhQlYlsnADxcPpf3kf67jgRntvGvDa/gTMZrzCvnc5TfyzMSZ9H7mSj656qMiuBMEYUwTTbSCIBx3VdPy+fYzSxnoTHL3WQ+zZupksCyUEffQFhI/OfM61lVNwKtnSCgunr712yxq34kJKCikZAd9C2ey6dNzsEKjm1KXXz+D5dfbw7pZgJkxcHm8vMUfwBYEQTgqRIAnCMIJk1/u5cNbr+b3df/G3xVFm1XFknXNOLIG2yblc0ZXBy354wkMxJmU6CRjuWibVEv1nR/FO70ab9BHAbD+oYcOuh6HLIFHnO4EQTiwsdgHT5zxBEE4oSRF4WN73kGiJ87T0//B1nFVJIIertv8HPn1EZpdNaybWM3/bLvhRBdVEIQxSgR4giAIx4ivxM/lvR/mcqAvarJh20I2P7mH2UtK+J9z9/+5L0EQBOH1iQBPEIS3nKKgzLmLQrBo3okuiiAIp4SxV4MnnqIVBEEQBEEYY0QNniAIJ62v3zvIb55LM+BcitM0Gf9QC2W1bl78aumJLpogCCeRsfjCU1GDJwjCSal9IM0tL+kMuOz71Kzbxa6CfAL1Mb79wMCJLZwgCCcVC2m/v5OdqMETBOGkdM0fokiKAr7c++9MwILVJYXseDRGdSrDB64XNXmCIJyaRIAnCMJJaUufSdLrG54gwdz2Hm5ctxXFtPh3eALP397MXx5XT1whBUE4KYyFGrt9iSZaQRBOOpO/048rYWFK2J1nch1ort9Yj1c3cJkmF+5s5PlJ1SeymIIgCCeMCPAEQTiplH6pj46kg76i0H49ow15+C7cksCfyfKXKXfxyyUPHOdSCoJwcpEO8HdyEwGeIAgnlR7FRcLpAMNEtiywhqO8v86bRtjjJulysqO2mlmpDLLTgW/A5P8tuP/EFVoQhLe0sfiQhWRZY/HhYOFkoaqqBSzTNO3FE12W1yEOkBNkT3+WgGThNEHyyBT9XwZLVqgdiFMaS1EaSWB63VjAK8VB+j0uSOosicSoymSH8glGBzlz2wYKI0lcCwq46N9XobiUE7dhgiC8Gcck8mqXvrPfub7S+uZJHeWJhywEQTghVu7J8tcNBndswz5lmxYFqQyDbhfBrMm07ihdPhdKJs3slh4m15TQWJJPQ0GAdr+HFVkTcovOjCToSmRw6SYRhzIU4JnAXTMmcvucOpbtbuf6bau4c8atKKkgeDJ0B/2kSoN84AuTiWeh7NxJeN0i+BOEU81YvJMXAZ5w0lJV1alpWvbQKYVDSesWv3guTVfMxO+ReKnZZE0nmJZEqazTHNaxkJGcEi6PA19aJ6nIpN0OdAscWBiyjJLMoEsykizhdElkFAV0k8ntfVy3fgcZReaOs+bRl+cnkMkyuzvCPEVhY1k+pksinEsflWXWl4ZIuRxAgN2lhbzzpU3sqCoCwHBImAz3McnKMjsLA+SFEyyJRDFkCbdh8lRlESmHHbCtrSpmZmI+3uI4kiIR9Xgp6ulA3bGL1rdtxpfQaZYsYoEo5fEUMbeb1kA1hivAi3XV3LjuIcoSEXRZRjFSRDz5BNJZ+lwh2vNd1ET7CKTTFKXjWMhYgISEgZs4xcgYSPIADgwUUyaLC4UMHtJYGGQlBYdlYuWWloC1xeN4aMJpnNtcT68/wLbCcj6x5kV8VhwZAwsXEmDgQCaBjEWWIABOYux9d4wE4HJizZyEhQNpfT2SlQUJJLugIEmYSKQdThyYOLI6yJAliGnIpDwSncFiJIfC1BlOzG1dWANpzEmVOCvdmKt3YaQhs2guvhlBrKc2Y0Sy4HGhl+SR6c7gdqXxyBmsvACdRjXe+u0EUz3IsoRUGoCEnd664xNYLzVi/O0VrHgKhyuB3B0G3cCycp+6SwYLrKwJkgRuJ9L4Invv7exFUhSk6xdCnpdsWxZ9bTvOpgYkBcwPXICrSIHH1sBlKtx8PbzrFnhgNWQNkCUsnx+rvBipqR0pnQaXA+78DGxsxlq5Deo7ocADtUVQ34lVU4I0uxqpptDOoz8GsmQPN/XA81shzwfXngHdAxBNwrIZ8KlLwaGAYcBvHofGbrjmdKxH1sDq3TB1HHz2EqTJZQc+eO9+EVbVw6UL4Px5R++ksKcTfvs4lITgM5eD23n08haOC9FEKxwXqqp+CvgsUAwMAn/RNO2r+zbRqqp6DfBNoBZoBG7WNO2+3Lz3AV8H/gB8GohomjbzGBf9lDhAPvKvOLe+krFHFAmKfWDkZloWRNL2xarAbV9MYSgoIGuAItv/J+2LI2BHXwEXmBbfvOc5xvUP8vNLTmdbdQkAldEkV9W3AfDc+BI2l+bZ+VkWpE0wLUjrdl4eJ/ObO1lXVzlU5sruGFOjSTKyxIaCADG3EzmWpiKV4uzWTha0dXPLkrm0FIUAmNLRx/ndEbqCvqFtsIDgYIzPPPokRakIhfSxh0oGKAZgHDvRKWD7OJmLm19Fyn0dLGTARAKeGj+DFU1b99une+OmQarJ5IIumSwF7EICdLw4hr5eFpAetXynL8TcG25mzV3fpToWBuCPc85icVsjc/qaRqzDAUhYZElRiok7t640XrqH8jNxY+LPfXQJFFIH+iqMYuAkTiUBWpCxa0xX1i6jarCNKf17hvaFlJs3crv30nHhIDNqXhu1SFhU0nTA5Qw8gGdojszg0L4/HBYOwAkkyVCITiC33Tpe2ofSDZXztAmwtmFUWSyCSCSR0PfJWwJ82MFz8s23F37ucvj5++Gbd8F3/21PcyigG7n1yVBRATt/iuT3jF723lVw7U/tYUWGV34E6uQ3WyJIZaDu49DaZ49/+Hz4w8fefL5HxzFpNm2VvrvfF6za+sZJ3UQrHrIQjjlVVacAPwIu0zQtCMwEHjxAujOAfwBfBoqArwJ3qaq6eESyWqASqAMWHtuSQzQaPSWGX2sxhsZxyMNBHAyHuA7pwNP30s3h4A4YuuZbEEzZF/id5YVDs7v3vqAYSDqUEYGjZOcTS0NKh5SOayBBIJ2rrLUs0E26fC5WlhfwUnk+Mbf90IXpddJWGOIfc6awprKEie39LN/ezIotjVxV30rU40I2hwsuAZ0FeaycOY0sLrIoQ8EdQA/lSBjM7G7dL8DYO1Yb6eVAhoMV14hd4swFh3awNzr16GvJ9sIKvEZ2KLgDOK2rkapY76il7JLYwabJcC3LyGE7lTJiOYPDJZMZCu4AKmKddAbLR6QwR6Xf94qojNjOvfNi5BEkPCrd6OWUfea80UuVyfA+Gd7/Fg4slP339rbWA+Qhse+2jS6peXQiDW23fTy+tmt4mj7y8zGhY4D4jrahKUPH78hlDJPUqm37pzmC4diu1uHgLlfGN5vn0RoWDp8I8ITjQSfXVUpV1YCmaQOapr1ygHTvA+7VNO0xTdN0TdMeAe4DbhqRJgt8WdO0pKZpiWNd8GAweEoMXzd3RDCQMZAMc/g6tvf/rAlG7oI3suZfyo07ZbvWba8R1+j71GkYksTsluEapbL4cA3SyNebYFl2c5UxnJdpWrwypTIX3FlgguFyICsSyDLIMv6szvRwlOKkXRPWWJSP5FI4Z2cTVzXsQvcoFA8MMrW1HaeetevMJAlLknDpOi7SODBwjqhJ85BCl5z8cvHF6NLIQGFY2O0/4PS9w24GhqY5iSJjYgEG3hGpTfaNmOd3N6OYBi9VDNfI3D1tMS2B4aY6ewkZck3CDoYPCQfJUeWQyA6NWTgPsz7MwsCFngsWLWB34QQm9o+s7VIOuN176XhG1VMCFNNJN1Wvu5ykGCOm23vrjVEABQsJZcQ+kUkjYey/ty870MuwrVxN4P7T7T9l1JOWR1zVf+0S+3i8dsnwtLwRL/BGgQUT8M8aPzRl6Pi9YqHddAyQ78dz6cL90xzBcGD6eDh9yqgyvtk8j9bwsTP2XpMi+uAJx5ymaXtUVb0B+BjwJ1VVNwLf0TTtiX2S1gBr9pm2GzhtxHiHpmlphKPqKyu8LBznoDtqUeSXWNdp8s8NWdK6xNIKkwc26sRNCW8ySVJWcKdMkEwybgdZScYwQZLsHmeGKSEDeZhEUxJZ06KhKI+fXXw6gy4HZHVQFDoDHp6cUErSodCS57cDOtOCRMa+WroVSNsXdt3vAqeToniaQUUhq9j3pqYsgQW+rM5HtuzBpxsYwFPVJXT7PSwKRyhNpUm4XdT29GDIDrzZOC9Veen3lVOUMcmLxSgf7KDfL9PlLKEw1UgyXULc7USrqKAilaC5JJ9fzb2Aq/ZsxGOkkIwUPd5iXLqb6t5B9viL8JoZCpOJXLgj5ep3ZBwkcdONhEXCqdNvBfHokMGJIXkIWQNIZIk4gvjMDA4zi4JJMJPk+bt/yD8nLWdALiTjlVjWuIsZfW1kcGNhYuG0G0hlN85CE0dvGJkUlsOJrCfQnR4kyUCSQJpcjnLVmZiWBLc/iRWXIOBG6onY/awcCnpap6+sFH86jb9/ADnoJWCZGNFiYgEnXa4CFgW6KfnUZZjPbMVsHEBaOgV5UiHGfavJDFpY1y7DOzWA+ch6zN4UilcmNa2W7KZu3CUy7mScovEluFN+Ii/kEWxrQPY7kBZOgrYweFzI//g05ppmzNtewhpMYOVZSNouiCcxZQeWYSKXBJEkMFsHsDxupOIg0tI6+xUXj25GcshIP7oeVzyFEjHIrm1HevAVsp7xSL99H063Bf95xQ5eVsyFvzwDP74PEmkkpwOqS2HaOCytHmlbM0wuhye+hfTKDqx1TfDKLphQjFVXBqt2w+zxWLWFSNVFkM7CQAK8LkhmIBKHvz8PVYXw/nOhuQdiKZg/Ec6eZR+EH1gBU6vsPngXz4cXtmFtaIJJlXClapdpX2dMg7U/gzW7YflMqD1KP82nKPD0t+H+V6E0z94/Y9xYeC3KvkQfPOG4UlXVBXwU+DF2M2ycXB88VVVvBXyapr17RPo7gaimaR/Z2wdP07Sj0MnksIkD5Bi78p9pHt6h27V4hgkuB75UFh3IoIAMiiIRiqf55KMat1+o0hLK1X5ZFtWxFLUDMc5u6QKgKejjbzNqh5p8P/bCOqb0DvDqlAlM62hm/p4NPFU7l8vTHSz5x2VIITf+8sCJ2XhBEN6oYxKJtUjf3+9cX2N97aSO+kQNnnDMqao6FZgAPA8kgQh7Ow6N9hfgKVVV/wY8BVwAXA2cfdwKKxx397/DDbj3merZL920j/fzgDqZ/FiCaT1hVlcWMTGaojyZwZPVh56q7fJ7RvUVbCkIMa17gCWvbuMTe96GLF/AFcdygwRBOOmMxTt50QdPOB5c2E/GdgADwKeAazRNG/UYn6ZpLwHvBX4GhIGfAO9+nf56wilm+2+qWH9bLVmHzHkNrXxM20ZZ0n54I+V0sKM4n90hH1O7w3iy9pOPDsNkeneYic09LPrubGRZnPIEQTg1iCZaQTg4cYC8xUz/vyjLntvFpEiM9VXlRHx2c22328maohDjB6Isae4kKcnUdUdYuLuBKX84i3lX1JzgkguCcBQck2bTJukH+53rx1tfFU20giAIx8u2zwVZfe10rvpxmGILBj1ukKDZZzfr9rtdXPX8Vjy6wfbxJby9492HyFEQhFPdWHzIQgR4giCcdBaN8zCp3EmmPctGj9N+MWyu313U66Y7z0c64MGodB0iJ0EQhLFJBHiCIJyU6mocaJ06smlhWtZQgCdZFi/OrOKuB09DksbeXbkgCMfC2DtXiB7HgiCclL51mZ+NeX7cloWsm2DaL2de0tLBN74/UQR3giCc0kSAJwjCSWlcsZNH3uEga1r2+/B6+3nvS2v5yY0hZs4pONHFEwThJGId4O9kJ5poBUE4aV0yx0v2p14eeughAC6//PITXCJBEE5GY/EhC1GDJwiCIAiCMMaIGjxBEMYO0wTl2uHx7L/AIU5zgiAcnKjBEwRBeCsbGdwBON8OEz58YsoiCIJwAolbW0EQxgbzdbpFN/Ye33IIgnDSETV4giAIb1XRxOvPa+48fuUQBOGkMxafohUBniAIY4Ije5BTslM0VgiCcGoRAZ4gCGOCnud5/ZmVoh+eIAgHIx3g7+QmAjxBEMYEKZ09eIJbHzs+BREEQXgLEAGeIAhjguV2HjzBR/4IK9cen8IIgnBSsZD2+zvZiQDvGFBV9Q5VVf80YnyLqqrXH+cyLFNVdeB4rvNwqap6g6qqG050OYSxRdLNQ3eMPvd7IF0NjR3Ho0iCIJwkxmKAJ1nWWHhW5K1FVdU7AF3TtA8ep/XdDCzVNG3F8VjfKeYtdYBYlsUzzRZOWWJ5zfAJKNudIP5qN945RbjHB/dbrvfv9Qw+0ox/SSmp9X34Ti8l/+qJ9PxiE8n6MIkNvSTCGZJBDyHDQE5lQFEwY2mIGnYmEuBT8EwvQO9MkMRBe0UBzkiCkt4ozlgadMveYzKQWywLJLxusopMj89FJOjDQKKzJA8pk2Ug6MVjShhYPD27FhcS17+wGVcmy66qYrpCPkJIXPrqNvIH44TiqaHtsveAhYMExbRQQeuxOS07ZCjwQ8aAfD94nVAYhKAHWvrgs5fDB8+H/ii8tB2mV9vpVtXDrHEwoexYlEoQTkXH5BCvl/5vv3P9VOtzJ3WUJx4tEw6bqqoSoGiapp/ospyqPvhfkz9vts9Dn1sg8fNzFDKtMbYvvAe9M4HsdzDl+avwnVYytEzT/zxL3++2AjDwz50A9N2+nZaPPQ/mcN5OwNlrB08jJg+zgLhBSusl4XHxyDlzSHrdOCoMrn10lR3c7WWMzjcvmQagIJZE6Y4AkGno5JeXLWbl7Al841/P8akPXowpy/zf7Y9T19HPjZ+6inDAiz+Z4YGf3E1BLPk6e0VCx0+ECipoPcw9+QbpJvRE7eHIAV7H8qHfwdo98Ng6aOwGlwNCPugdBK8LVn4HFk85NmUTBOFNe0vdyR8lIsB7E1RVDQA3A1cDJUAL8JEDpGsEvq5p2t9VVT0beAq4EfguUA7cC3wC+DlwLTAIfFbTtP/klp8L/AqYCSjAK8AnNE3bnWv6/Sogq6oay61yDjAOeErTNEcuDwn4EPBJYDwQAX6sadqvD7GNFvAZ4D259Z+jqupG4DvANUAesDpXnl25ZYLAb4DLgCjwDeDPwApN055VVfV9uf0xOZfeB/wwtx+9wIvApzRNa87NfxZYA9QCFwDdwOc0TXvgYGUfayzL4i9bhk9Dt2+x+Pk5MPhYM3qnHXSYcZ3wv3eNCvDCf9954AwPGMUdns6iPJJeNwC6Q8GhG4dYwqaMGM4qMv8+YyZnb27g+Zm1mLLdY2Rc7yAvTB9POOAFIO51EUykDpDbaDqOE9uocsdKSGbs4YxuB3dgT7v7JRHgCYJwXIk+eG/ObcBi4DwgBFwBHE7nHgU4G5gNTAcuwg7a7geKsIOdP+cCH7BvLm4GqrCDnBjwdwBN0+4GfgA8q2laIPe35wDr/Gguj48B+cB84NXD3M4PANcDAWAd8EdgGnA6doD6KvCwqqp7e7n/EpiYSzMbuJTR1/Z93ZLL63Ts4LMXeEhV1ZHLvBc7AM4Dfg38ZcT+OWai0ehbZliSJKYWDEdl0/PtoMozo2B0o8Uk36hlXRP2b7J9s4LxJIzo3rE3OHsjFMuisj/KtpoS6tqHf23iniUzmNAVHpW2tSh0yPxO+MtJJ5SCY8RXVhrxocwad8K/P2JYDI+F4WNF9METhqiqWgp0AbM0Tduyz7w7GNEH7wA1eCuBUk3TenLz/wX4NU27NDfuA+LAPE3T9nsYQVXVWcCm3DKJA/XB21tTOKIGbyvwG03TfvMGt9MC3qtp2l9z48VADzB+RA2bDISxA7lVQAK4WNO0Z3LzJwG7gHP2rcHLLZsALtc07clc+gDQD5yladqqXA3eFk3TPp6b78cOcg+4f46yt9QB0hq1+P4rJk4Zvn66TKnfPgmF791N5MFG/GeUU/KRmaOWMaIZdpx9P+kdEeSQEzOexTUuSOGNU+j5f5vJdiSwsqMfUDiccK25sog91aUURmJM2dWGxzhwlaAJWBIYssz2iiIsWcKjG7QWhlhbW8buyiIW7mpDsuCx0yYTDng5b1MDM5u60OqqmNnczYyWHma19CAx/IEMn34tZNLUUE8RPYe9L980CTuIsyyoLYX6X8NTG+3aurm1dsB3/2pYOBk+fvHxK5cgjG3HJPLaKv1iv3P9DOszJ3WUJ5poj1xt7v8dR7CssTe4y0lgN8sCkAvaAIIwFCD9FLu2MMjwNa4EaHoD5T2SsgI0jhiekPt/Y66MezmBGmAn4NqnXAcrYwngBhr2TtA0Laaqancuv1W5yR0j5sdH7p9TSXVQ4nfn718ZWnDNJAqumXTAZZSgi+lr3n7AeeWfP+2Iy3IkS57+unPGDw2Fkzrf+2Mhd28OUB1Ns7qymLY8P2G3k4kdfZTF04x6IYoi0XJbLV2pci746D2Hf/b/9U3w8cuOYCsO4uLT7L+9rlx8dPMXBEE4TCLAO3KNuf/rgK3HeF2/B9qBOZqm9Y2owdt7LTuc3lSN2GV98gjWPzL/vcFa3T5BKgC5ZtUM9hV7d27yuIPk3QOksQPQvX34AkApdp9G4RRT4HXw809V8/P95ix83WXaHnoIPW4cXnBn/efICycIwpj0lmqqOUpEgHeENE3rVlX1HuC3uSbHJuDAVShvXgi7Zmwg10T6nX3mdwLjVFV1aZqWeZ08fgN8VVXVddh95gqBCZqmvfZGCpLb7juxt/szmqa1qaqaD5wDPJmrfbsTuFlV1U1ACvj+QfIzVVX9K/DdXDPyAHZfu+3YD28IwmGxlEM0LPuAuAjuBEE4NYiHLN6cm4D1wHPYT4s+gP3QwdH2WWAZdjPuC8DD+8z/N3ZtV6eqqgOqqk5gf7/Ffnjjtlw+azlYlcjBfQioB55VVTWKXZt4HcM3QZ8GmrGbhDdj1xpa2DV1B/JZQANeyy1XAVyhadrhPZopCADmIe7Be+88PuUQBOGkIx6yEIQjoKrqVOwauSpN09pPdHneIHGAnAQeeughnO1hLvro/QdOcMPp8PcvHtcyCYJwTByTyGuT9Kv9zvWzrU+d1FGeaKIVjjpVVScy/PqUYuzXoDx/EgZ3wknE9B7kt2i/eM3xK4ggCMJbgAjwTnGqqj6G3fy7H03TAkeYrQe4FfvBiQTwPHazriAcM9LBbuxnHajXgiAIgm0sNtWIAO8Up2naUX9Bl6ZpW4FZRztfQTgYvcD/+jOP4EXMgiAIJzNx1hMEYez4ypWjxwvd4rUogiAc0lh8yELU4AmCMHb84Eb7TxAE4Q0YCwHdvkQNniAIgiAIwhgjAjxBEMaOHW1Y53wd68JvQyxxoksjCMJJwjrA38lONNEKgjAmrLjidixux35NloUVfDfS16+B795woosmCIJw3IkaPEEQxgQPdmg3snu09b17T2CJBEE4WYiHLARBEN6y7Jq7kWMmx+i194IgjDFj70whavAEQRgj9u81IwFIVx/3kgiCIJxoIsATBGHMGron/8/LJ7IYgiC8xYkmWkEQhBNsj/szbBo/he2lFWwvC/KNxx6l2HeQX7EA+OadcPUZx6eAgiAIbwEiwBME4aTQePlvST/fSF2mi4k7O7hiJ2TwE6EcX9LEREHBOPDCW9ohmQKv5/gWWhCEk8JYeC3KvkSAJwjCW5LeNcCL33qZVzYM0lhWyLxuhWszYWRMwH5a1k2MEnaRJI8EBQTpfd38LN+7kN53Ftz+6eO1CYIgnCTGQpPsvg4Z4KmqWgjcBZwO7NI0bcGxLpSqql8FlmiadvmxXtdhlEUF/gxMAG7TNO0zB0lrAcs0TXvxOBXvkFRVPRt4StO01/2sVVVtBL6uadrfj1e5hLemrGHfxypZg6wiE89aSDJggUOW8Jk6/ZEMva0xirwQqCulsyFKoidKpZKhrTmBZUjEJQd61yAdDjd16TjhjESkMUx9Y4xMzGBSTxhMWDVuIgta2zitsY0+Tz6/P+cMpEyGV8dXYkkSkncuTdeUAfCVJx4m4synODU4osT2adnHAGncB902CbDueA7ueC63pP2U7d46PxdguSSiNRV0VlUwKdyDMnccEbcfuaaIdCRGUUhBnjbu/7N333F2VOXjxz9n5va923ezm03vCQmhDU16BwEbIiKoqFi/NtSf+lVERVREEXtXEEH8KqKAFOm9HloggfSySXazyda7u7fOnN8fc7dk0zYhm7t387xfr5tMOXPuM7fMPvecOTPkljcTOHIWuRkNJJ5ahikroyJuk8OQKyshms3S+9J6ggdPJlASQk2qItORJjilGtPSiamIk+vOELSyKBWAoI3bncUuD5MJhQjZYKUyEA6QbE8SCllYnofX2oMpjWGFA1jl0R3vbDIN0a1fj55el5ht8AyYrCFQGgTAS+XwjEHlPOzSnb+G2+P25rCCCpRCBXZ+WrfxDF7Www7b0JuG2I6fzySzqGhwt+MRQviG04L3CSAOVGutc3s7AMdxHsFPQK7qW6a1/t7efp434XvAvVrrLxc6ECFG0l8We1x6n8dH/vMC73l4MavHlfOJz51NJugfJr5x7+1848G7qQAy0SrumH0AX3/bubSUlXP4ijZ+eNMDVPUkCZHj+QNm8K8jD+XZSZVsKY1CENR0j98//yeW1U3i4ne+G8vzuOtnv+PADW0AVHqtvHXFC3zo3AvBHkgUznj2DQ7tfp3v3X8LAIlgKbaB7mCUccnW/nIhkrvcRzVk2sI/CPZ1z6iMQa1MklkZYzEzGPfqasazBgOUMXCF+75tbKAyv62nFJdc+BluOeQAXvvBl5i/ZR0pwqwKTGZGbjk5FaU5XM3k1HquP+I9PDftUMp7uvnQI/8h3q1QeHz3pGP5zdFHUp3t5fE/fos5bc1EUPQdqm08XAJstKdiHVZDw2OfhfCgJMjz4KKfwN+egOl1cP83aa4dx9e/uJgv33A3vXaYlWV1GGDW6fXU19lsuvYF6mgjiIt39gJq//O+Xb6OAKnGbl4+7T56l3YSVlkqQmkm3ngG5e+Zvd3ym59u4clLHifblWV++SYOWPoUvGU23Ps1KB1IVo3n0f3+f5D56ytY0yopu//D2DOqhxWTEHtqLHbRDmcU7XTg9e0ld47j7A8/r6YDiwrxxGP99R3r+1dsPv2gR3VLggseWowycP+hM/qTu/GdHVzx4N35CwlDbbKDpfX1tJSVA3DpQy9R1ZMCFFmCPDVnDsvGlfnJXZ6xLb569nv40UmnAeBZFpNbe/rX266iNVS2TVxnrHmeb95/a/98aTbB/806B119KG9Uzeb1mlm4ykK9iUO0GvRoZjpePqFqYRo5gv2JYV+ZodMAD81cwC2HHMvxKxczf8s6ACKkiZseEqqcEpNkSmo9K2um8Ny0QwHoLInz4PyjAMWaiip+/ZajMEqxJVRC2M3ln8Me9FwKmyw17mY2LAFufXrrHXnoVT+5A1i1Ca7+Fz99Oscltz5KSTbL2ngNRvmtbSvua2bztS9SRg/BfDumdddr5JZsGtZrtu7axfQu7QQgbYKk0ormzz6yw/KLvvMK2c4sGFjcUUfKCsNTy+CGrbfJPbqazF9fAcBb3U7yezuuUwixYzttwXMc507gzPz0e4EXgGOADwHfBmqBUsdxPgd8EpgAtAM343f5uflta4GrgdOACmAFcCHwGeA44GjHcb4KbNBaz3Ec51vAsVrrU/PbVwPXAafnQ/svcJnWui2/fg3wO+AU4EhgDfAxrfUur43gOM5C4CfAIfnY/wR8X2vtOo7Tgf/D/Q+O4/wGeIfW+oFdV+n8ApgBaOCjWusV+RWPMKS1cnC3bn6/jwdeBN4PvOg4zv8BlwM/A74MlAB/Bz416PW9Hjg1/9o2Aldprf86JKgPAlcCpcAdwKe11t072IEFwLXAoUAS//28Qmud3cWOn4Lf4jkbyAEPAp/VWrcM2v+XganAyfmyVzuO81Hgc8AkYBXwFa31ffltDsrv+3z8BpNn8rGv3Fkse0sikaC0tHS/mI4FIRuwcZXCNoZIduA3XSYQIGtZBD0vv0QR6J+GVHDwocTg2grb2zbhCmWz1PT00JJ/7payEspT6XyNhrcte4Ovnn4m4XSOnnCI+u4Ozlv2AmF3YPCES4CzVjzDi5Pmc/ekUwBYWj2Tdyz9zzbPtyesQQM1FB4KbyelB0SzGQA6IiUYBpK/HAEs3P5lIXfrr1Ew58+H3JzfDWv5v7tzlr2TZ1NYeCSVR18KnUgkKB3a5RkLYXtpkvn3xzKDLgRtDEopPDPwO98AvW6GvjR7Z58ZO7b1nw+FwUQHYt6m/KB1ynhYJv+65mPuLx8LbVVvdtDTFPo7ItOFnx4p3hg8B2+nLXj5c+BuBv6stY4D38T/I/tW/ISoLl90PXAWfjL0duDDwKUAjuNY+AlFBXB4/v9LgITW+tPA48B3tNZxrfWcHYRyM35PyLz8owb4y5AyHwY+C5QD9wN/3sW+4zhOX9mHgXrg7Hw9X8jvfwWwDrg0H9+ukjuAjwHvBsYBi4E7HMfZ2ZF6qOOBJvxk57z8sin4r/UM/NfwfOC9g7Z5AjgY/7W9ErjBcZwDBq23gXOBhfiv32zgx9t7csdxxgGPArfhJ+xH4yfm/zuM2NPAp/ET/wOBBuCnQ8p8GD9hKwd+lk/uvgJchP8efx24zXGcmfnyBvhWPpapQDewz84VHHxQGevTfz3bonpajFvefyTp+hIOXruJcNpPPlpL4nz8/PfTHomRsoO8XDeFyu4eZrY0E0+m+MsJC1g8sQbPghBpTnr1NRZuaGVmSwfhdAY8j3AyzZcevZc/3nI9JyxfxhlLFlObXU2YBDZZQqRYOr4GuyzAJ194gavufJBXf/9tJne1AUFcQqQoJ8EkqrKtrK6q79+HVZVTdvsAvb2bixtgAkuJ0UmIXiaxBAuXrLJIWvYOtzHAW9Ys5fL7b6W1pJT7ZxxEigitVjU5bEpMN5sj1bxafgA1Xe2c8dojVPZ0Mr15A2e89hQ2aSYm2vj+Pfcx2e3hqNQmoqUh3IBNDkgTIoeNhyJDjI2hBma8vZrohSds/T6+ZS5c8R6YWA1nHAzfeA9fPSnOfz57FsvG1zLeShCxPELKcNBn5tLw65NIVVXRo6JkAyHsL55E2YGThvWZmfKVA6k+ZxKh2jCllVA2r5zJN521w/KHXu1QeVAVsUkxDl/YQ6ihFC45AT54wlblg0dOIvqtU7AmlRM8fRbl3z1zWPHI9P4xLYZPGbPtr+zBHMe5AchprS/Nn7D/MDBFa71uJ9v8CJistX6P4zhHAE8CNVrrzu2UfYRtW7W+Rb4Fz3GcBmADMFtrvTy/fg7wBtCgtW7Kt+D9Umv9w/z6+cBrQMX2nnPQ87wP+EE+VpNf9nHgC33J5u4MQMi3xl2qtf5jfj6G3yp4ktb6qWG24L1faz1j0PpL8BOiykEtdv8A1mutL9tBHBr4k9b6V4Pes5l9rV6O45wK/AeIaa29wfvoOM6XgLdqrU8eVN95wA+01jOHPtcuXo9z8nGMy88/AqzSWn94UJnXgGu01jcOWnYn8Ozg12nQugXAq0CJ1rp3d+LZQ2Px1IyCMZ5HcmMHrYEoPU+uZP3KblY/u4HgK80smjyNlngpU9qTTGvbQiZiM6GrkXNXvIiHTS/VZIkRopM4bdw+53RWVU0BoDNs8e1Hfrbz5x4y7QKqoYLkOccR++Tx2AdM9gd22BZYcg14IUapEWlqe1r9bptj/dHmY0XdrLcnl0nx8LsB+zmOcyF+q9f0fJ0h/K408FtdWnaWaO1C38/J1YOWrRy0rik/3TRofd+JPaXAzp53ErC2L7kbVPekHZQfjjV9E1rrXsdxNgMTd2P7tdtZ1tKX3OX14O9bXwvpt4AL8FshDX43bu1O6l0DhPFbQluGlJsGHJPvnu6j8FsBd8pxnMPwu10PAmL57eJDiq3ZzvP90nGcwX+dA/itwjiOMwP4IX7XeykDf6dr2f5rJUYxZVnEJlYRAzhvAXN3Ub61J8cth/+GXFZx2KZm5iU2ABkU8NblD/HS+AWsrajh7sMP3apbdCgDpIDoomvhwGn9Aywg/0Xqi2/Pd00IUcTG4i/5PUnwzOCEyHGcSfhdZu8C7tFaZ/IteE6+yBpgnOM4ZVrrrm1qY5cnuPQlk1Pxz90DP5EcvG5PNQJTHMdRg/Zp+pusd2rfRL4Fr5Z8sgIk8JOvvvUN29l+eCf8DLgQvzv8dGBJvkVOs+3fqikMJMZT8btTt3fRsLX4rYxn72YcAH8DbgXO11p35Vvw7hxSZuj+rQW+qbX+xw7q/A2wEViotW4d1IInf4v3A9UlAS5a8mkAWv7vZTre+wc64pVUZhJUZRLM2/IG3z/hY3z5oQd3Wk83UGpu2wcRCyHE6LA3LnQcx/8xvBnIOo5zFP4Agdfz6zX+oIE/OI7zafykYgGwRWu9EWgGdtj1p7Xe6DjOfcC1+YECCn8AwD1a66YdbTdMd+EPsPia4zg/xG9N+grw2zdR52X5rsgN+ANLVgHP5te9AFzgOM6P8RsUvvsmnqdPGf6Ahs2Ale/SPQi/C3aw7zuOcykQwW/x+4vWenvJ5I3AFx3H+TDwVyCDnxDO1lrfO4xYOoGE4ziTga8OI/7rgG85jrMceCUf32H4n4838nUuBzocx6nBP8dQ7IfGXXAwXPALesu+SiZTQjs25FL8+ZZfkyrf+e/v0lnj902QQoiiNBYvdPymTzTRWr+OP/jidqAD/4/6LYPWe/gn+CfxR1B24I9U7eu6uw5/5GmH4ziLd/A0F+O3fi3FP/euA/jAXoi9E7/l61RgE/7o3BvZwQCEYfoD/gCFzfiJ1tsHda9eh5/4rsR/Le56E8/T58/4CeQK/KTyAPyBK4O5+ed6Ff81XEV+IMlQWutm4CTgHfitr+3AvxhoNd2Zj+G3JibwX4MdtcoNfr7fA9cA1+efax3wDaDvEiqX4Y+07srv194ZKimK1qSuq6k311Fpfk1l7nrKvRt57sb37LC8Abjt/+2z+IQQYjTY5SALIfZz8gUpAnfeeSfnvO36bX6DZ4Hg9HGw8jeFCEsIsfeNSFPbE+oP2xzrjzWXFnWzntyLVggxZgVBkjshxC6NxV/yYz7By3f7TtnOqrVa6/m7Wdc9+N2F28hfJ3BMcxznInZ8fuLHtdY378t4hBBCCLF90kUrxM7JF6QI3Hnnncy9+n5mPbV+6xX5y6IIIcaMEek2fUz9cZtj/fHmI9JFK4QQhfbGV09j1rnnFjoMIYQYFSTBE0IIIcR+bSxeJkUSPCGEEELs18biuTiS4AkhisrjN67kivs9emMR3rJ8PZM6unhyyhRWVJXx1BX38Z2nTiIQDe66IiGEGMPkjtpCiKLy9fs9FrZ18vY31jKpu5dQ1uMdLyznyn88wVNzp/LZ44de51sIIXbOoLZ5FDtpwRNCFJUFrZ3UZrIE3RwEFJnSMGvHlbNeBfjAQ4v43WkHFjpEIYQoOGnBE0IUlbJ0BmUMtjdwK+VQJsuUVIbbD5lBbyjKQ2W/I72pu4BRCiGKidnOo9hJC54Qomj0dmfZFAgQ9DzCSmHlr+OZs216whE+99CLxL1eIibHw9Nv4bTmD2CXhgsctRBitBsLXbJDSYInhBj1jDFcfsITlGczzLBt7JxHZzSEFfIHU7THYlS2dhPLZqggTYgc4d4cS0+6lQP0RQWOXggh9j3pohVCjHq/Pv9xaru7iaYyVPSmKM1kqEimQSmeqK3iwcoyVFMnSSK0EccAQVx6lnQUOnQhRBEYi4MsJMETI85xnImO4xjHcaYWOhZRfK790Wo61qQJprMoNXDQDbguz1ZV8NC4GpZUlPHtc47gkTkTSBImh8ImR1Wyi0XqR2Q6kwXcAyGE2Peki1bsU47jfAD4BDAPcIHngS9rrV8taGBiVDHGcMU3VrBoSS+zUhnqgzZ2EnKeB5b/uzQTCrIlFBrYSCl+f8pCTli6gTZKCeBRQYIIOTZVXE2YNDYpDBYtVGNQ1LOOMBlykTCxWz+CSRlCZx0ItoUKh3YQ3TbBguvC5k7Y1A7T66EsPgKvihBipHi7LlJ0JMET+1op8E3gKSAHXAHc5zjODK11b0Ej249sas5wy41byGQM735vNdNnRna5zfp1af5x02bSbWl6NiaxjOH440toe6WNTSpMZzSG250jHFKkPejI2SS7c4SSGRqjIeKeIRMKkQOemlBJs7HAwIxMhpM3ttJtW7xcUUp1spdTl6+np7SEA3I56je1Ec1myQZslOuRikUwSuEFbA7d0saLVWV+d4qlMMEAQXJYWHhYtFHOZJqx8YAANhYrmUqSGAAuNgt4FZPK4Z3zM1xi9BLAJkOQHlR+PJ0ivd3XJEcYlwgh0ig8IDPs98Cg6KaOHBHCJIjRxvbG7hlsPAyBwX+CwgE4bAaMK4fHlkBpFL77PrjxEXhmmV+NrWBiDfzg/RALwzf/BhUlcPXF8MPb4Y0N8LHT4JKThx1zoRhjSH3jPrIPriB46kwiV55O9rbXSP34CawpFcR+8XasKv89TX3zXjLXPYadSgIKEw1hn7MAO9OLuWsRdrIHgyFJJR42xraxFjRgjSvBblxPdOnrYDzAIxeOk0rHsTCYQIjQ5adjXtkA/32VXAosL0uoJI17xpGo+xdDyMYKZVDZNL1VU1Fbugh1tKGCCqtMweZ2QAEGQxBFAMii4ja5mlqSddOIfO44gg8+D7c+i2uCeJMbsA8ah71mPUyogl98BIOFd+FvMM+vwppZifW3T8OshsK9QWOAsYq/S3YoZcxYGAwsRhPHceqB3wEnAJuAa4DfA9O01muGlI0ASeAwrfWLjuNcAlwO/BL4IlAO/Bb4fr7O04CNwKVa6yf2we6MyS/Id77RyJpVftJSVmbz419N3ar7c3suv2w1Lc1ZSlIZPNtvRQtms9S0d9I4fhzhbA7LGAzQGfWTMPC7UtO2RU++tc1V8O9xFZicIeR5fHBjC8H8q7wpGODo11eQjUZwI/7o10AmS11jC+loKP8IY2wbgNLuJC+Ux/njMQcRMPD1fz3NW5esHHT2jGESTXQTx8WihARrmAj5EgEyHMHzAOSwyVDSv22AXkKkgN58ord9/ppYvs6eYZ+500MtPYzrn69kJUFSO3yOXdZrKfC2E2coAOEgJPLd1JNqoHGLP60UvHodzJ88zKgLI33Ti/S+///656M/exu9X7wbsi4AoUsOI379+bhvbKJn3g9RZLFxcRlohbXIEqYbC0MP5eSI5N+7gTOVKliD1Z9IG1KU4hEgR2n/sjAJsv3vN9gkCZJB9b9DWVLEyBInThsKUKSxSOIRxCKbrwkMERSp/i1TVOBaUUq81v6YssQI0DXw/l94LC4xzC3P9cdkH1SBevnaPX15i82IZGL/tW/c5stzhvuBgmZ9SqnTgPcC44wx5yqlHKDMGPPQcLaXc/DESLgZv/t1MnA8cMlOyp4C9ALLBy2bAlQA04Fjgc8A9wA/BCqB24Dr93LM25VIJMbkdEd7dtByF9fd9bZdnS4Ygzfol65rWeTyyZ4a9GPRDEoWPaVw1cChxjb0JyJBz/QndwDRnEssm8MNDtxqLBcK4gZsDJAOBgmmMljZHB4wfn0zh2zczG/+dB93XHMbJyxpZDCFRzcl5AhgsOimjFqa+9fX0bJV6cFHc4PNcKh86d3lDulA8XiTt1fbXnIHkMkNJHcAXYMayo2Bls5R8Znc2bRpHpgH8Na29yd3ANn17f7ubOwE+v6wDf3brPoTdbODP31DE3m/nBpSQm21zGAP+eQoPGwU3laltv6/rwZvyJYulpcbEpO79Z40d5BrbN2qBM3+fo+W92skp0eKUds+Ckkp9Rng1/h/G4/PL04CVw23DknwxF7lOM4E4GTgS1rrTq11M/DtHZSdjZ+ofVFrPfgbnAS+rbXOaK1fAV4BntdaP6O1doGbgJmO45SP6M4ApaWlY3L6be+qpi8HO+cdlQQCapfbnnteNcpSWH3ZoDFMDGWIJVNEM2my+VY1BUSy2f4y0WyWUDaLyicg3UEbE/dbVnoCNq+VlwD+L4LF5aUsHldFOD3QJRrIZHGBxw6cw/Mzp5Aoi2NnPMavbqG02+WopU3UppNsrImTjAYwmPyfWJcIGTJbJU6KKaxmPi9wIM8whRWD1rmD/vwaAqQw+fG427vwad+8n5hZ2ymxczHaUPh/zP3Wwu4d1uDu6FAdHXSNv+MPgMB2ktIPngSfPdufDgbgS2/3u2oBTl0Ixx0wKj6TO5sOvf9QrJnVAFizagj/vxMInrfAL1ASIv61UwCwT5iJml6F63eq5rvNgdIQVgTc/HsZohcGtdQBqLBNlnD+PQf/nU3k5/3PvD0ljolGsPu74g1BOnEjoUE1ZYjQjotNLv/Z8wj1J4sDtQdQ2P3JpodFmgq80w/BVMX7l7mEccsr/I2iIfjqOwh+420Q6EtjM6irLthn70Whp/cjnwdONcZczcCH9Q1gznArkHPwxN42Mf//2kHLVg8t5DjOAcD9wI+01r8ZsrpFaz34nNdeoGnIPPjn83W+uXD3TyecXM7Bh5aQzRpqaofXcnTa2ZU4R8fBQKojSy7tMumAUjo39GKFbXozCgsP21Z4HqSzhmzGZcVznazs8PBKAswsVTRMiXDWcx2si3s0d8CCBo/WF7M0JQ21HR10BAM0BQJ+K2DAptTA4kPn0VYWpyqZojMeZ9Yb6ynvSVPR6yeCU1q7qIik6YpGaA9Gmd21EYWiixI8DKF8IlVGGxY2MXJ0EyaXT+os0hiyGCwUCkUKQ4psvvXEEECRQzGQyin6jrp+d2DfnMlPqa3KDMwrwDRUowIBynNtmNpqrPoavN5SrKOmwYQqzPImvGXNEI1AOEjg0hNhVQukM1Aag4OmQl05VJXCq2uhPAZzJ8KWLnh9PZRHIev5idyMej+Az58DJWEYVwGfeSu0dML0OrCH11JZSFZdKWWvXoa3tgNrSgUqEiT+j4vwVraiqmNYlf75d8q2iC//Ku5jqzDRAKq1B+pLsWfXQc7FW74Js2g9Ac8jPqUa9+E1WO85GDsWRjWUQks3rNoIT74OC6dgzZtM8MlVBNe3Y58wC+uE2dCbxXthHV46g7VsI+qEOag5DXiPr0RNq8JKZbHSaSrGVWDa0/DaGqzJVTC7Hq5/FO+FVaj5kzDOLLzWBJZxUfMboLqckgxYM6tRyQ9hXlwD5TGC4RBqajWs2wxVcagq9VPFTT+G1xpRs+uhvrKg789YMArPwSsF+rok+n4XBNmNE30lwRN724b8/1OAlfnpqYMLOI5zKHAv8B2t9c/3XWhisPKK3f/6V1blk8HqgaSwfIL/x3VHv7Gnzizh1CHL5h82vMbXTNbw5VOfYlzTFnozGepcCOQ8LMNWXcXgdwUDpOwQ7VYpAc/Fw086Q4eWEZscgUVpNq8qw6aX6IQYgUe/jTWjpr+O3b3nxY5SI2sH0336It/R9mon67Zx5OyB6ZoyOO6A7ZebVjcwXV7iP4qIigSx59QOzCuFPbNm23KWReDEmdutwz58Ghw+rX8+ePqCrQuURmBGDZy2EPDfu9C8IYMX4mHsE2b578/p8wfqPmXrhhUFMAFYUD+w7P+du6Pd2/pzEgujjp2zddfszPFb118Vh+Pn7bA+sXvM6OvPfAz4KvDdQcs+Czw83AokwRN7ldZ6veM4jwDXOI7zISCKP1IWAMdxjgH+g39plN8XJkpRLEJBxU8ePYYr3vYcU5MpbCtAb0mUltoKxrW00xqPUtXdS9a26M5f1iTo5gh4Bg+bTMBjzr/PpeLsGQXeEyGE2C2fAe5USn0UKFVKLQUSwDnDrUASPDES3oc/araRgVG0x+XXXYU/MvY6x3GuG7TNWVrrx/dplKJofPj7s/npV9cQ6U1hWdBaV05XZQwTsAllsxy2eBXjuhO4yiKeSaMAV1nYcyskuRNC7JKxR1cXrTGmSSl1OHAE/oDFRuA5Y8ywL9knl0kRYufkCzJKLH6smd9etY76ZBIVsP1z9PLdsuF0hiNfWUZJt3/JCReLLIb5L51P6cH1O69YCFFMRiQTu6vkpm2O9Wf3XDy6sr7dNPp6nYUQYjvmH1/Pz+47gmDO3WZdMhLmP85CloyrY015NWksKs6ZIMmdEGJYPEtt8ygkpVSjUmrd9h7DrUO6aIUQRcVzDcozhDI5UrEQRinSwSBpz+X6Yxfy8Qde5i3mU4UOUwhRREbhIIuLh8yPBz4H/G24FUiCJ4QoKslwkNJUmoBr8EyEVDRMKJVlUVU5zfEYy8dXFDpEIYR4U4wxjw5dppR6BP8KFD8dTh2jL2cVQoidCOaypG0bO5ejYf0mDnl+GZF1m3lk9kQauro49rBd31dXCCEGM5ba5jEKpYFpuyyVJy14Qoii8vWnT+Gua95A39lEOhalojTEIifOt2JP8rW/noodWbDrSoQQYhRTSl05ZFEMeCv+bTuHRRI8IUTROfvLczn7y3P75++8804A7MjovyuDEGL0KfS9Z7dj0pD5HuDHwF+GW4EkeEIIIYTYr422LlljzIfebB2S4Akhipoxhgd+M56wsnjqNy9St7mFzz93ZqHDEkKI3aKUOnk45YwxDw2nnCR4Qoii9pUznyGigpiwjasUrZVV/OL4B/j0Y0PvgCuEENvnjY4GvD8Oo4wBpg+nMknwhBBF61NnPcOWcIQpykLl72qRjkbYkswUODIhhNg9xphhj5AdDrlMihCiKH3tvZruaAnzO1N41sChzAA5b9u7XQghxI4UyWVSdoskeEKIovPiK51syIQ5eE0T09ZuYPryRnA9MAZlDFUmx49OvbfQYQohioRR2z4KSSlVppT6sVLqBaXU2j25VZkkeEKIovPzby2DbJa6LZ3YnqGiq4faLR2Q76bdVFPDEa8sL2yQQgix534FHApcCVQBnwHWAdcNtwJJ8IQQRWXj0nbiHRkOXbVx6xXG+P8rBZ7HlopyvnniY/s+QCFE0TFKbfMosNOB84wxtwNu/v8LgPcPtwJJ8IQQRaFxSRdfe/dTXHnZCryKMlZPm0h7WQkG6I2Faa2tBPxz8CzPY0XDVAgG+dLJD9Jy1+uFDF0IIXaXBXTmp7uVUuVAEzBzuBXIKFoxLI7jPAI8oLW+aoSfxwDHaa2fGMnnEcXhm5ctYvXqFJloGQtXbWRWSzupcJAXD55DTaqVinAH52z+L62xSpYccCkAnlJYrj/IIpbOEE8aPnQLnHrlX0iXlPKh60+jbkpJIXdLCDHKjJLLpAz2CnAC8CDwOH6XbTewbLgVSIIn9gnHcYJa62yh4xgL0imPh+/rwPMMJ55ewWsv97CxMcNhR5XS2+Py+qu9zJobZf5BfhLT0Z7j/v+0s3FDmnkLonS05Xj0gU6yg64kovAvGDz4GGfIn2yMQuXXKQUuCleBbUAZs1VXhmJQHcafRykwkMt3e1jGYBuDwe9SdZVFxlJsjIVpDwYoNVk+89TtTG3fTGDuW4hWTSNiLB44bC4bSiNM2dLJAW0JIMgbdbNZW13PlngVadsmmO+mtTz//7VVZVQk0xy/bjNPzTycimSGL3+1kWgqzYTWTqY3b2RK+yaemnMgm6qrsF0XywMLQ7irl5xnCLoeKhAgmMtgLItXp9bTUhZjQmcPtd0pbKArHGRWVwfxzb20lMXIBSzq27oIVoY56mvzidXHWPFygikHxKEiyNf+1EqyI8eBboqDPjSFC4+MDuu9X/lMG2883krGWMw+upL5x1Xt7sdHCLEdo3DU7Eeh/3D6OeB7QAXwgeFWoEzfeStC7IDjOL8APgnkgCywAXgaCAIe8HZgM/AdrfUN+W0uAS4Hfov/4ezUWs93HOd7wHuBccAm4Oda65/kt3kFWAgk8/X+TWt9qeM4MfwTTc8DyoHngE9rrVeM9L4Do+4L8vNrNvDqSz0A1NYF2bzJz5tDIUU2azD5nOpzX53AnPkxvvmlNbQ053Prvr3pO2wM2rv+8zVMX+IFpu/yI8ZgGYPlp2y4yt82kH8yA/mETfXXY/DfRDtfPmsP3Cc2lHOx8tt6g+JZXhrjC0/cyrmvP4aH4ppTPsWamkmsK4ty76z6/mTy5FWbcBpb6AmHyQUGfqcGczkCnkdVZxe5gE3A9bD7jnHGsKmyAmNZKM+jsrsbC5jY2syiKX6vh+V5/fHHupMcuGgVqWCA9vIo8VTKj3F8NYGyEqx8ta5lUdvUwcylTQDkbEUyHsDOJ5mb6yrY1FBNzg5ggOsPm0FzaQyAhU1tnL5sA4dcvZD3zd/5fXRXPdPG37+6mGQk3D+Y5J1fmMbBp9XsdDshxpgRycRumfj3bY71F65/T8GyPqWUbYx5U9d7knPwxC5prT+N30T8Ha11XGs9J7/qPcB/8Uf4fBz4teM4bxm06VSgAZgFHJ5ftgQ4FijF/4Xyfcdxzsg/z0H5Mqfnn+fS/PzvgbnAUUA98CzwH8dxgnt7X4dKJBKjbnrlsmT/fF9yB5DJGAblMqxanqJpY+dActdHbTu71aJ88rBVe54aKNU3pdRA2b4yOzoamiFr+rpDzJB4JvammNi1BYD1FePZUNlAwM3RHg1t1VLYGg2RCQRw7a2TongySWVPDyZgkw6FhuzowPbGsvoTw42VtQNFBhXvjUdxAzbRbI7O+EAL2+TNHf3JHfitmKVdya3q6EvuACLJDHbWP06nAlZ/cgfQWF5CNOfyzOocsPP3fcNrnXiW2mo/Vr7StsPyMi3TY3F6pIy2y6QAzUqpXymljt3TCiTBE2/GM1rrm7TWOa31A8A/gUsGrc8CX9VaJ7XWvQD58hu11kZr/RBwF3DKjp7AcZwa4H3Ap7TWm7TWGeDbwHjgyJHZrQGlpaWjbvrAQwbOH5s8Ldw/XRK3CORTXtuGAxbGaJhQzpTpA2WArVrt8g1xWzdT5rNEhWGrjDFfqq9lzl9kttpucD2Dpy2zdV19CZAaUrApGuauOceQtQLE0z3Ynsd5L9/FO157lkjWT4Jsz2Naew8zl2+krKuHrlCA1RUxkrYiksuRDAbpiMXIBAL0hMN+MmkMZHNbtUja+fP0qrs7+6cHx1zSncTOufSGAlR29vQvf31iLWnbGngtlEVHVUn/tq5lkbMHDq298TAm5r8xcQxTO7v7181q7SIRDnLmPD/Z3Nn7Pu2IKgKD3xMFB7yldoflZVqmx+L0fuR0/HPu/qqUWq2U+r5S6sDdqUDOwRNvxprtzB86aL5Ja50eXMBxnM/it9xNxP/7HgX+upPn6Lt1yyLHcQYvDwKTdjviMeCST9Sz4KASXNdwxDFlrFqeZOP6DAceUkI66bHsjSQzZkWYNDUCwBcvn8RTj3aycX2aAxaW0JNwufPWVhJdLpbt5wtu1k/a1KAUR/UlcMr0Zz6e8sv5Xa8mn+gZ3HzrnW08XPBb8/Ln7fnlFXbOhfw5eB4A/kWJc8oipRStkSCeMTw5aS4ffPfXmNmxhbTxWDRuAR949g7OXvwI153wHnoi46jqTPDwwbPIeS53HzCBbMAmmsnx/pcyRF3T38qVDgZIlFQSTGeobdpM5ZZ2uspLCGRzeJ7H+NbNzNrQSNJVrKuvI+B6WCaf+KZTvDSrgaBnyMQixLt7yIaCJKrKWFYep6Y3RX1nEmUplk2vJb6gHLM+wQY7gmcpJicSlNRHeec35xMsDbHq1W4mzSnhc9VBvvXbFrqaM8ypMBz/+QM4dtrOu2cBJi4s5wO/PpgVz7aTNRYznHImz98v//AJsdeNgsuibMUY8xLwEvBlpdQJwIXAQ0qpJmPMwuHUIQmeGC5vO8umbmd+/Y62cRznGOAH+C12z2qtXcdxbmXrnrGh50Gszf8/S2u9eTdjHpNsW3HksWX987PnxZg9b6Dbr2HS1i12kajFyWdWbrXsuFMqRjTGvaVxWQc//GQXXzv9f6hLpYm5ESo62lhaWw9Ac8giG/CTo2QowOrqMhZs6sDrO1gbj143y6yuTbRXhJh3Si0f/sqsbZ5nh03Ie9G4yQPdvD//asMe1VE/p5T6OZLUCbGfeQN4Hf9Cx9sewHZAEjwxXM1se/2doxzHuRD4O/5w7vOAU3dSRxng4g/IMI7jnA2cBfxjyPPMAp4A0Fq3OI7zV+BXjuN8Xmu9wXGcCuAk4H6tdTdizJo0u4KfPXhM//xjf1nJf69vhXzP5OGvN/LC9HGAP0BiXCKJ5Rk8C+xsjoDrUd3RzufvPJFYWWh7TyGEEKPuMilKqQr8v6nvwz///D78BpI7hluHJHhiuK4DrnccpwN/FO3z+IndW/FHyrYC/6O1fnIndfwXuBF/FKwBbgf+NaTM14ErHcf5MfB3rfXH8bt0vwY84jhOPdCBP+jjvr2yZ6JoHP/+GUw7uYHvfXIpjbXVOKs28ek7nueNiVU4K5vJ1ldS39oCGchFbZpLIvzk0ZMKHbYQYpQbbV20wEbgKfxTmM4zxnTsbgVymRSxRxzHuQHIDRrpOlbJF2QU+vQFj5KlnpKeNPFkio7KOMFsjvrmVmrdJt7/4m28Wj2TgzdfXehQhRB714hkYjdOvXWbY/0H1ry7kJdJGW+MaXozdcgoWiFE0fn052bSEQzSWxKmo6oUlCIbClKSytGwPMfaeAOPn3FWocMUQhSJ0XaZlDeb3IEkeEKIIjT3LRM4d24Xrrf1dUBtz6Oiq5dXqo7iMzedUKDohBCi8OQcPLFHtNaXFDoGsX+7+PKDufdtLxILhLA9Q21rF/OXrgMgnuktcHRCiGLijb5z8N40acETQhStm+44lHHNjURTvRy2dBmZsE26EjZNkHu0CiGGb7R10e4NkuAJIYra1586i3GdG9lYV0q2wmXlnAlc/Py7Ch2WEELsMeX7qFLqIaXUovyy45VS7xluHdJFK4QoetO/FgACnHHuRwsdihCiCI3Cy6RcCZwG/AT4TX7ZevxLlv19OBVIC54QQgghxOhyCXCOMeZvDFyuazUwfbgVSIInhCh61r/bCb27idvL/kzrsxsLHY4QosgYpbZ5FJgN9N2pqS/Biw9atkuS4Akhitry7z1N6o4AvdVl9NbEufUjL9K2eEuhwxJCFJFROMjiHuDHSqkw+OfkAd8B7hxuBZLgCSGK2utXvk5dRwKTdbGSWSo6e3n0+HsKHZYQQrwZlwH1QCdQjt9yNwX4ynArkEEWQoiide1BN+EtiBHIlDG5qYtIMkcyGiCQSRc6NCFEETFW4Zvs+iilbODdwPuAMvzErtEY07w79UiCJ4QoOsYYPnXq/fz1jLfSFYnyt5//m/EdCQzQkwyhApLgCSGKkzHGVUr92BjzJyAFtOxJPdJFK4QoOqdc/DpPz5xBVzhCWSLJpI4ECv+AFs1myabj9GzpKXSYQogiMQoHWdyplDr3zVQgLXhCiKITzLosrqvnvc8vYfamNgYfinNBi2zU4pHaP3O2+VTBYhRCFI/R1EWbFwFuVUo9DTQyMJIWY8wHhlOBJHhCiKLyh1+twGldR8nLOU5atR7L9dhcEaW2I0l3JEhHWZj6RBcBPBpvWMykS+YXOmQhhNhdr+Ufe0wSPLEVx3Em4v9amKa1XrO/xiBGn85Elu9+/kXmPLeceFkpNSUpWqvKKE2laVHQWhYlns4B0FxazoxcC40fekASPCHErhW+S3Yrxphvv9k6JMETu8VxnKn4V9OepLVeX+BwxAha35TlqZeTrO70uHtFjsZNOdych1GKC4+I8IV3lPLMrRt49b8ttBAgEQoQD4DdliKT9VhTW0XYNVQmU9R395AzitaSKEHAsy0Uilg6TSjnoox/3amEZZHLubTFI6TjUUoyWSpTWTZWlDGzvZPDV63jiamT+OWJxwAwob2LL/1Xk4hFmda2mWwwRCiXY0JHJ7arCGBxV9n1tFTG6SmLkgnYmIBNMJPFtiBnABSlPUkiAUWuJEKyJkamM00aC2MUKKieGieb9YiWBjjsPZOonRXnjfubaV+X5LD3TKRqapwtK7upnRknWhEq5NsmhBgDlFIn72idMeah4dQhCZ4YNsdxgoWOQewbLy9J8Y3rtmAMuPgJWTRg80Y0iqcUP1pkyNz4HJUBC5Si3GR4eVodM5s2c1BzO65t0VpdDZZiZkcXSimMMWwKh1hXXsr47l5mtLZjG4MBsrZFNhCgwoAVVOgpdbw0sQaAY5Y2cuZraxi/pYO6RBuPnXpUf5wbKsvoiIVJhyJYnseElk7qEgnCrgtABI/6RJLq3gy6ajIhpTCuRzYaQSXTBDDkQgEsY6ESWVRvLxnXAksRwgMFrm2zcbMBFKFNSe656g0wBssYUIqVj2wmELPJJT1ilUHe+zuHignRff+mCSH22Cg8B++PQ+ZrgRD+/WiHdbsySfD2c47j1AO/A04ANgHXDFp3AxAEssDbgP/Dvy4PwFLHcQzwA631dxzHqQWuxr85cgWwArhQa73UcZxq/Bskn57f9r/AZVrrtl3FMCiWjwKfAyYBq4CvaK3v2zuvghjqtv8mMPlTem0gaAxlroeXX2+A+myWdDACQFBBaSZDzHPBUnTGoiQjYaZt3oLKd30opZjc0cX6ynIi2Rx2/gkUYBuD53mAImtZLKmr6I/l2ZkNfOihFwnlPJ6aP5+jVq/j1YnjAWho72L8pjaqO5NM2NKFbaUI5ba9OIBnWf1dMAoI5HIYS6E8g51zCaeyAKQjQbAGyhnA9jxs18UNBHBtG9vzth5hpxS5XheUorc9y/JHWjj8oilv7g0QQuxTo2DU7FaMMdMGz+evjXc5kBhuHXKZFHEzfiPNZOB4/BscD3Y+/i1TaoEvAgfll8/RWsfzyZ0F3IGf2B2e//8SBj6INwOVwLz8owb4y3BjyCd3XwEuytfzdeA2x3Fm7uE+D1sikdgvp6dNHGisNYCnFDml+pMkoxQ50z+oC4whZQfoDfi/GaOZLMrz6A6HYVC5ZNBfnwpu/dvS4NdtANt41PSk+teN7+j2W8uAzeUVVLlRvv2f+/jok8/zmXufwzLQXhZlzfhy4tkU7qDDmso/Aq7LoGjxlMLy/CWepXDzSZ3tegxlyCeIgOrbl8H7DlvVHakZ+ENR6PdRpmV6rE3vr4wxLvBd4MvD3UaZIQcqsf9wHGcCfnPvTK31yvyy04D7gGnAt4DJWuuTB20zlSHn4DmOcwTwJFCjte4c8hwNwAZgttZ6eX7ZHOANoAH/R8YOY9Bar3Ec5zXgGq31jYPqvRN4Vmt91V59Uba1X35BPM9w7R/b0K+laE4rEsqi0bZIBWxySlGdzfGhmS7BO1Zi5zyaYjG2xGPEOxLM7uiiOxLimSkNZKNRDtvQRHUyRQ7F63U1ELCJZ3PYnkd5MkXIdfEsi6xlYQxkLNhUVsKKyhICOZcD17cyo7mNhs3tPHPwHCIWBHIuGJeZS9f1J53hXIoTVi+ihYlEcPv3pdsOsWFcOS0VMbKxiP9L3fWw3RxYNrlQkGhPktotCRSG5ppycqEAuYB/nqCnFOloBGUMgVyOWEWQurlx1ut2TM5QOSXKwe+eRKNuZ/LhVRz0zgmFetuE2B+MSFPbLw++Z5tj/f+8fNaoatZTSp0F/NEY0zCc8tJFu3+bmP9/7aBlq4eUWTOMeqYCLUOTu7xJ26l35aB1fV+gncUwDfil4zg/G7QsgJ8YihFgWYr/99HqXRf8/Phh1DZx10V24MVXOnn4niRPPwVV9VXkAhZrykpIBEPEkylO634My0As14trLGrUGlaY6QRJYWHIAEf0fhQ7ZO9xDMN18Lv2fD+FEGIwpdRW174DYvjXxvuf4dYhCd7+bUP+/ykMJF1Th5QZ2me1bR+WnwSOcxynTGvdNWRd46B6V+Snpw9a1/eXd2cxrAW+qbX+x3aeW4xhhx5UzqEHlW+17Iz3LOK1mZPoyJbwPW89B7esAaArUMo9tedQuTlFwkRJRmwi2cw+Se6EEMVtFA6yuHjIfA+wzBgz9G/sDkmCtx/TWq93HOcR4BrHcT4ERIErdrHZZvwkbxYDLWgaeBH4g+M4nwa2AAuALVrrjY7j3Adc6zjOB/Fb7K4F7tFaNwEMI4brgG85jrMceAX/V8xh+frf2NP9F8Xpv39fyFnvfY17J08iO+jE6LJcgpCVIhcKkAxaGDxOT368gJEKIYrFaBtkARxujPnR0IVKqS8YY348nApkkIV4HxDGb017HLhxZ4W11kngG8AtjuN0OI7zda21B5wLJIGXgQ7gT0A8v9nF+AMuluKfe9cBDL7Vyk5j0Fr/Hn9k7fVAO7AuH4NctmU/VZpLUx6E9ZGBLuKuQJweO4ad8yhNJjni78cWMEIhhHhTdtTYcvlwK5BBFkLsnHxBRqFHHtjMP366lDbiXPGfO/DCOdZEpxJMWlSke+kKWpya+UShwxRC7H0j0tT2c+e/2xzrP6PP2OfNeoMucHwncA5b7+904BvGmGFdh0m6aIUQRefEU2t54KalzHxtNWtKJxLtNtSmU5SQpIsYybAc2oQQRanvAscR/J6wPgZoBj4z3IrkKCiEKEpX3XAsxvP4+7hb6GgIUdPaTbcbYeP4SiZs2rDrCoQQIm+0nIPXd4FjpdSNxpgP7Kr8zkiCJ4QoWsqyaPA6YUuQpTMbyASCNGxqpeaU+kKHJoQoIqNtFO2bTe5AEjwhRJE7ru1TPGH9nAWvr8MyhqxJc+BdlxU6LCGE2GNKqTL8mw2cgH/3p/4M1BgzeTh1SIInhCh67bdPBeDcc88tbCBCiKI0WrpoB/kV/lXirwRuwr8axf8D/jncCiTBE0IIIYQYXU4H5hljWpVSrjHmdqWUxh9de91wKpDr4AkhxoRVy2waN/QUOgwhRBEySm3zKDAL6Lv9Z7dSqhxoAmYOtwJpwRNCFLXzLlrElE3V1CXT/O2fz7IiWsJvHzyy0GEJIYrIKEjohnoF//y7B/FvAPAroBtYNtwKJMETQhStp15opTEa5+HjptMRC3P06ibeumgVxhjU6DtgCyHEcH2UgYEVnwO+B1Sw9V2gdkq6aIUQReuyX/fQWhknEQnxltWbiGY8NpbH+Mm7Hix0aEKIIjLaumiNMauMMSvz0y3GmEuNMRcYY5YMtw5pwRNCFK2G3jRtlVEuen45h61eD0BnNEJvMlXgyIQQYs8pvwviUuBCoMYYs1ApdTxQb4z5+3DqkBY8IUTRKvNcTl+2gZmbtvQvK0+mCNp2AaMSQhSb0daCh395lI8AvwP6rnu3HvjKcCuQBE8IUbQqUhmmNrXiDboftwfUt/bS0dxbuMCEEEVlFCZ4lwDnGGP+hn8fWoDVwPThViAJnhCiKH3zssWklWILivFNW0jaNmnLYsGK9RzQ2MTN599X6BCFEGJP2fijZmEgwYsPWrZLkuCJUcdxnBscx/lDoeMQo9NLr7Zz/vlLWL45zJLqSlpqqwh4htotnUxsaqU20U0yHOLlSA2fO+LOQocrhCgCxlLbPArsbuDHSqkw9J+T9x38Cx0PiwyyEKOe4zifBS4CDgQ2aq2HfaFHUURcF1ZswFu8jvQhM2hrzdDRa/P3FR7PPNhLZTpHNhgmZinKDfRYFrU5l466ah6PhKjo6OKwNRtYMqWBrK34wgOPc5Mzi+MveIXLnnoMG0N7vIIpHY3EeruITCzFOmo28XcsoGpmOcHaMgIhGyyFCgQAQzbrYdsKCw8VCOJmXexwAON5oPxfyCoQwBj/B7ZxXZRto5QauEyLMVD47h4hRHH5AvBn/IsdB/Fb7u5jNy6TIgmeKAYbgWuAucCHChyL2NuSaaj/CHT1YvCTphBwxXlXcm99LVsCAc7f0szsniTNdTUkwyF6YlE8yyJgDDnPo728lLKeJFnPUJFq5bSWhwmR5bLFqzhtywpeKZ3HjA09lKUtVo2bT+fkUias2Yx9h0v7k6vorItxwcu3MqGrE4OFwaMtGOa+htNwsZnXtZSVsRnkVJCOmlLcgIUbsIkmUmA8QipFPJGkJxCjLN1DhijRughnHtlG9A93Q0Ml3PG/sHBqYV9rIcR2jYJz7gBQStUbY5qNMV3AO5VS44ApQKMxpnl36pIET+y2fIvaZUAN0IX/KyMORLTWH8uXeQyYorWekp//MnAi/g2TNwJHaa1fHlTnI8DDWutvD30+rfWt+TKXjNQ+iQL63f3QlQQGruppA69XlLElEODATa0cu66JRQtm49k2Yc9gZ3MkwiFQyt/GGOavasQycEDXG4S8LAA1yVaquqZx0JrW/rrTwRCBni5sz291q2xN0F5TxqMzTuB9L92FAhQ2myMN5FQQgCXl8whmXHrLIngBCwUEci6JyhJqN7bRXRbHhGws15Al6q/fsJnor+/yn3TtZvjKX+Ceb4z86ymE2G2jJcHDv1NF2aD53xhj3rUnFck5eGK3OI4zG7gaOEdrXQrMB+4AHgBOzZeJA4cAKl8e4DTgAa11W778JYPqnA4cC9ywb/Zi+BKJhEyP9HRo+78zg54HQDSXwyiFZw0crlT+nOMuS/FMeQnPlsXA+OVTdqS/nAEmtrUx+NAdzOTwbGurMp6lCLrZ7caxcwrTd/6zAdTA2dCmP0pfTg3MFfw1l2mZLtLp/cDQTPPEPa6o79wRIYYjn4wtBj4I3K217s4vLwNagTnAPOCLwHJgEfAHoB2/1W6R4zhnATcCDVrrrOM4VwJHa61Py9d1A5DTWl865LkvAS7fx+fgyRdkpGVzMOXj0NTe/2LnlM2n3vVNHq+tplUpPvHCYgKRME3jx2GAnkCAZMDi7poKUgH/mndvWd/EZY88hzIup7Q8QTzXTYpKGsum0GhqCSfAtRVtDUE21NVRt7GdcDbNprpqIuFe3rb4bipSacDCxdARDHP/hFNQnmF2YjnLS2aTMwG/izZo4wZsYokk4BFQWSo7O+m1o8SyaTJ2CWXTSznlmASR3/wHJlbD378EsxsK9CILMWaMSFPb909+cptj/f8+dMw+b9ZTSnUZY8oGzbcZY6r2pC7pohW7RWu9ynGci4BPAn9wHGcRcKXW+j7HcZ7Hb8WbB9wPrMAfHPEGkABezVdzH5ABznUc51/4yeJX9+2eiFEjGICNfwTyR+5cjmAgwO8Bz/Poak/y3s9MIuoaErEQS8aV4wUs7KxLSg1c0HjxuBqWLJjB4avf4EXjQCpHqtRiXTxEL0EWNjcSSyaI9dZyyowctd85nLIF9UQnVOUHQby7vy4LqAXeNyjMw/Z0/646f0+3FELsfwJKqZMYSGSHzmOMeWhYFY1AcGKM01rfBtzmOE4I+ARwu+M41Qx0087DHwyxGv8q3MuAB7XWJr+96zjOjfjdtJ345xv8a1/vhxilAgOHJcuyqKgu4d6/zu5flkq5fOiil2msGUdPSQmdIf88uQVdCTpjcR6Y79CVybJw7auUur18YvFH9vkuCCGKyyg6B68F+NOg+dYh84ZhXuxYEjyxWxzHmQNMAx4DkvgJmsG/gcADwOfxW+de1Fp7juOsBj6OPyhjsBuA14AIcIvWeoc3D3UcJ4D/WQ3in9cXAdjZNmLsikRsbvnnYVzznWUsSZQSdF2mJXqZkMrSGQygUmkemt7AGWUbefdfP1jocIUQYtiMMVP3Vl0yyELsrhBwBdAEdACfBc7LJ1tP43+mHtJae/nyD+C30D0wuBKt9VLgOfzBF4N/nWzP5fjJ5O/wf7kk8w+xH/vyN2Zz4oYtfPjF11mQymDbNr2RCMmSEj738LOcftVbCh2iEKJIjMJblb1pMshCiJ2TL8godv7Fy5m5voUVUxpAKWxjUMbwjoef5ILG9xc6PCHE3jcimddVpz2zzbH+8vuPKuosT1rwhBBFy/IML06tpycYoCcYoCsYoKq1DT15QqFDE0KIgpIETwhRtDIZlzXjKgduBaYUT08azwHVbmEDE0IUFaO2fRQ7SfCEEEXr17+cTCiX22rZgpY2Lrn91AJFJIQQo4MkeEKIolVfV8KCtk2klIeLS8PmVlSiBzUGTpAWQuw7Y3GQhVwmRQhR1G654UDuvPNO3JTHO85/e6HDEUIUobGQ0A0lLXhCiDHBjsjhTAgh+kgLnhBCCCH2a94YbMGTBE8IUbRyyTQ3HPx/vDZlCo2VpaiP/5JzNv5PocMSQoiCkz4NIUTReqXqWlSolNmb2mlo6eWemQez5rK7Ch2WEKLIGNQ2j2InCZ4Qomi90TCJ3miUbCzOpFyOEgy/fiJS6LCEEEVGRtEKIcQoooyLs2IN9e0dADw3cwaXn3ocAee/fFefUdjghBCigKQFTwhRlP55yv9R0ZXuT+4ADmhsJBMM0BkuIdeeLFxwQoiiMhZb8CTBE0IUpccz5Uxr7SJtD3REdEZjnL5hE8FYmN8femcBoxNCiMKSLlohRFE6pGk1Bos36uoozabI2jZrq6uJuS6J2grU+mihQxRCFImx0GI3lLTgCSGKjpfJcebKlcRIUtfezeqaWtbU1lKRSHLGw69Q3dZFSU+Oz5/0BKlUttDhCiFGOaO2fRQ7acETI8pxnEeAB7TWV+XnPwB8ApgHuMDzwJe11q8WLEhRdP457ffUVk/ksNbVeBlDfFUXr9ZPoqQ7R1VXD2c8uggrZahp7eLAL0/lkHWtKNelan6cX35/FtYY/LUuhBCDSYIn9rVS4JvAU0AOuAK4z3GcGVrr3oJGJvaetS3wpT9DMgNXvQ+zcAruMVdiPbcY48H6eCXOZd/n84/9l/OeXga5IKFgF6+Pm8zszW24yiZAgopkmhhtdMdC3DX1FEzKxlMWDbkwAbLkCFHldlPlJqjd0MmS6okkAmEimSw/Peco5ja2cPUdD/Kbox1mdntEX+7iLe9fxcqKOMevaWTh5i4emDWZY1atp6Grm3DOoycW4uXpVTw2dQq9wQARV9EYj0DKJeh6vP+ZJfRUh5jSlaYkm2NlWSlTNrYTslwykRCLJtazJWiTjgRYmEgTsiDW3Usi7bGyvprqA8o4sqOd8HONTJ4V5dTvHESsxr+0y5LfvMHSPy5HhSwO+dpCppw9qcBvpBD7h7F4JwtljCl0DGKMchznF8An8RO5LLBBaz1nSJkIkAQO01q/6DjOJcDlwC+BLwLlwG+B7wO/A04DNgKXaq2f2Ae7IV+QPfGW/4Wnl/rT4yvJvvNEgr/6V//qUy69nHE93Vx3y3/YzEQAFC4NbIL8BUY9IE4TEbrZHK7mjon+ZU9mtLcwu60FgDBJbDxsMlj5t6pZVXHf5Nl86aIz6Q1F+NSjz7Fk5mTmdqYByCjFP8bXkAgG+Ob9zxDL5ajt8UfcGuV/WD978ZmkA/nfv8pAZ6r/kxCw4YrHX6YEz98GUJkssaRff0d5nPvmzqAmm6PM9ct02TarSksGXp6lqzmwsQmAySfXc8ZvjmbjI8088v7H+st4tuLtj51F6eT4m3gjhBhzRiQT++rbXt7mWH/1HQcXddYn5+CJEaO1/jTwOPAdrXV8aHKXdwrQCywftGwKUAFMB44FPgPcA/wQqARuA64fucgHJBIJmd6DaW9tS/+0ae6AZRsZrCNawuSONjKE+5epQf/2sfHPn+u1By5eXN3b01czHgoP+pM7gArTzQFrE7z2k68zo7WZJ6ZPYVLHQONwyBjGZfx6U8EA8fTAOXrKQCoYHEjuwM80Bx36cyhsNbBAAWrQr/9gNkc85xL2Bspk7a0PtcnQoJG/jd3+Pm7s2aoMriG5KQWMjvdUpmV6NEyPFLlMihB7keM4s/ETtS9qrQd/g5PAt7XWGa31K8ArwPNa62e01i5wEzDTcZzykY6xtLRUpvdg2vrquyB/gFSfPwf7e+/BKP9wY4B3L3qamw8+mlxJEgsXgDDdZNVAq5iLRzc1GGBCspmSnJ8AbYn5LWFBMtgY/PQuh00PihydlJIlhOkt5eMvPMj8jVuY0TiQcGaUYlM4yOGNzcza3M7T0xrI5WPNBWxKM1lmtnb6sRuDCigI2/3bH964mQ2hSH/Ol7UsNof8RNVTipaaShaXRHm9JNpfpqYnSWV3b3+dFX1JpYJDPub/7pl01kRi46P9+199aBU1B1ftk/dLpmW6WKZHylhM8OQcPFEQjuMcANwP/Ehr/Zshq1u01t6g+V6gacg8+OfzdY5clGKPfeZsOPswSGXhgElYgNnwB9wbHsPd3M1lSZeDcq9wzZcu5ZN3/YcVPTbPTlzI0vHjOXJlI2tKyqlNJzj5jdfoDE9iRXUN1ekUXi88OnkKi8fXcsHiFwAI0EMJG1FAliDrWQBACd1Mb+6msmktmyrLUV3d2ErRaluc//oqZrW0smhaHYmyKK9OGsf4bJrusgilXpaPvvQkL9Y30BUpYUJXK49MmUFvMMyx3VtYOCHLpVfN47F/bmZFY4aqkMdE4xFMxVleUcb8CUHeMzdITyhAsCNA0rVomBvj8fvbCEyPcOZb4ljZGlrf6GTyjCjxhhgA4cow5zx6Fpt1K1bYovbQaqyA/AYXQuwZSfDESPOGLnAc51DgXvyu25/v+5DEPjG9fqtZNb4C+3/fRl9b2FvzD741l4OA8/pLHjFoq3dut+p1H/4brzTXc2hrEyG6+zt2g2SptlvwAoZl48azOjaLFIrLXzhl7+wTtf1T5/+/it3acu6ckq3m62qrtykTiAYYf1zdHkUmhNhzY3GQhSR4YqQ1AzP7ZhzHOQb4D/6lUX5fsKhEUZv8p/eSCX0bmzRm0JkmHoqWCTbfPu2DxBOdzGtp4YcPn1TASIUQojAkwRMj7TrgesdxOoANQAv+yNjrHMe5blC5s7TWjxcgPlGkUmHFlmwtdWTJUAHksMjQUl5OKhRk1sZWfvi4JHdCiF0bCxc2HkoukyLEzskXZJS6a9LPWbC+nRRxqllLzGrj9fqJfP30j+B5AY6qSHDlTw8sdJhCiL1rRFKxL7zrtW2O9T++bUFRp33SgieEKEpnrPkfmgNX4AFNTMPzZrI4OJ4z9TIiHWk+0bj98/eEEGJ/IAmeEKIoBWyLOB2kiWGIonA5Yu0KtljluL+WrlkhxPCNxUEWMgZfCFG0eolTRQs2KRRpeokR9zZxwsfmFjo0IYQoKGnBE0IUrRzgEaSWzeSwyGFoMNcWOiwhRJEZCxc2Hkpa8IQQRWuyuZrcebNIEqGHkCR3Qog9IneyEEKIUabu1o9x5513AjCtwLEIIcRoIQmeEEIIIfZrXvE32G1DumiFEEXtwTNvp/3TPSz7Ktxe+qtChyOEEKOCJHhCiKL1zLefoPmNDCZqEwukeGX2VG4N/aLQYQkhisxYPAdPEjwhRNFae/0aIqaXkoRHMlxDQ8Lw0kGzcV230KEJIYqIh9rmUewkwRNCFK3a9BZCxmbV9HoiXoblsxuY0Z7m5aueL3RoQghRUDLIQghRtLJWCetqa3jrMy8SzuVIBYM8O3MW+p61HPbNowodnhCiSIyFLtmhJMETQhSl1773BMsbJjBhSwcYQ28wSCyb5eC1a1gbixc6PCGEKChJ8IQQRemV3zayce5MrKyhuWQWoVSGho5OZrS2Ma83VejwhBBFRC6TIoQQo0TQMkxp3oQbCJCLhEjHIjSVl5NTFslwkG8d8I9ChyiEKBKeUts8ip0keGLUcBznEsdxVhQ6DjH6/XrCXzikcR3GKCzXA8CzLQKuS9YoEnaICV2Gj5/yYIEjFUKIwpAuWrFPOI4T1FpnCx2HKCxjDBv+voaOJ5oxLzUTX9GIVx6lta4e40JprcJ9aQNdXRnWVIzDBUrdBFWpdv56yDFUdyeZtrmDcjtIlxvngFebcS3F6pk1lPcmmN3cQpQMbirIqVte5uiuMNVfPoQQYII2B61sprSjl/ZSm08uuoeFq9rZZKZQd2QM+9UNhKws7eMmUHL8JKb84hSablnDxt8tJbM2QShuEz96HJGJcSpPGU/VyQ08s9HwvWdcFq3LkWzL0pPx6LEUBG3CBio9lwCGrGfYFI/6L4JnsBRcNNdwycEhTp5hF/ItEUIwNgdZKGNMoWMQRcpxnBhwJXAeUA48B3xaa73CcZxHgJeBqcDJwPeAm4A/AIcBIWAR8Hmt9QuO4xwNPJxf3pt/inO01o84jrMAuBY4FEgCNwNX7KOEUb4ge9HS7y9i2fdeBeDA7tWofCdCLyFShCgnCUAqEOC16gY8y0KR478Hz+KolRtAKcYnOihv74VkeKDikEt9pit/5SqDh+Kp6RO5aNXDvP+CS/j7QQ7YEHE9Fqzv4IWaOPfe/CNOX/kaK5jLGmYACoXHPBbzcmgh5fWK7vUe5Gu1jIdt8vMKIn85lWOb6jF918vqTkNnuj+kWE2U3nDQn3EN5IzfZ9L3h8T1IJXjrovDvHWO/NYWYphGJBP7yEUrtjnW//HmmUWd9UkXrXgzfg/MBY4C6oFngf84jpP/q8aHgZ/hJ38/w/+8/QqYki//InBbvnXvaeATwCqtdTz/eMRxnHHAo8BtwATgaOA04H/3xQ4mEgmZ3ovTmx7a6E8Ygxp0nI6QJcxAvh7J5Yi4/vz6ykpmtHT0r6vpTVCWS26VeYdNblBtCoOivSQCQNbKt5AZSAVtbEsRMoqHpx4AgOtnXfkiFhkihEmT2LT1xZL9C5+q/rpW3bWerX4fZwaVt9RAcgfb/5NkKbDgv0sHBoSMhvdIpmV6NE+PFE9t+yh20oIn9ojjODXAZmCK1npdfpkFtANnA1fhJ2sf3kkdpUAXMF9rvcRxnEuAy7XWMweV+RLwVq31yYOWnQf8YHC5ESRfkL1o2TWvsvQ7i4ChLXjhfAue33ibDARZXD2+vwXvwYUzcVZv7G/Bm9/UyJpgA1krgGeBGzXMad0C+G+YhyITVDw/v5yPvvtijGWBDSHP45DGDnR1nHtv/iGnrlq84xa8CRbda1121IIXu/k0jtlQN3DF+54MdAwka7HqKL2RXbfg3fP+MGfOlhY8IYZpRFKvSy5euc2x/oabZhR1midHFbGnpuX/X+Q4zuDlQWBSfnrN4BX5pPDHwIlABeDlV9Xu4nmOcRynY9AyBciJS0Vo9pcPJD6zjI4nmul+KU7J0nWYiijdE8ZjcpAeZ5HV6+ltS5ILK1zLEPMyvO+Fx7j9wCOIpXJ0l0Roj0SIZAzPTprAwqYWAN6wFDM3b8kf/RWhrCHcDiXJFLZnsII2B67eTO3mToJtQToCUVbY09moGph8QoCAbiRk52itn8+8ExuYct1JbPx7/hy8NQnC8QClx9QRaiih8pTxVB5fzzNNhqufdXllbY7uXI7uiE2P8s/By2Vc6lNZghhyQFM04mefOY+ABR9aYLhoYYQTpslHWQix90mCJ/bU2vz/s7TWm4eudBzn4wwkcH2+D4wHjtRaNw1qwev7lTS0fN/zPKC1PnvvhC0KreFdU2h415Stlk3fTrmh96E4cdD0byfcyObJDTw+azLZoM28llbaS2KsnFjHmS8tAeDFhjruP34KiavKB21ZOmh6AQBDm4EnDp6+eAYTL56xw305fLzin+8I4B9KIzssJ4QY3cbCZVGGknPwxB7RWrcAfwV+5TjOBADHcSocx3mn4zg7uo1AGf4AivZ8mR8MWd8MjHMcp2zQshv9qp0PO44TcRzHchxnuuM4Z+7dPRLF5OMbPkBJVzcHbWqh3vLoGF9NJOKxsq6ar7z/bfzt6NmsKivlzzceV+hQhRCiICTBE2/GR4GlwCOO4ySAV4Hz2fF5a1cA44BW/BG0TwGDz2R/GLgfWO04TofjOCdorZuBk4B34Hf5tgP/YvuNPmI/olyXozY1oZR/GGsvq2Rqe4LjFi/l2JUruPz1txc4QiFEsRiLFzqWQRZC7Jx8QUapp775DGv/sYY19flTPvPHsikbmnm1Msz3nz23gNEJIUbIiGReF39g9TbH+ptunFbUWZ604AkhitJbvn0U0bYUde2b+pM7BUS9FNM72wsbnBCiqIzFy6TIIAshRNEqSabZ3KMod7voKSuhoifBweuW0/vbdxc6NCFEEfFGpmGwoCTBE0IUrYDJMH/TOlLhIPWr2qnvaaeJeg64ZEGhQxNCiIKSLlohRNGKXXUUnaEYlT1JrIxFU7CGNSGFGgMnSAsh9h2j1DaPYicJnhCiaB352cMpOaGBNjtGT9BmeWU5Z6e/WOiwhBCi4KSLVghR1I6/9WzuvPNOeoF3nCsjZ4UQu28sDKoYShI8IYQQQuzXxsJ174aSLlohRNHrfjlB50e7+Xfl9YUORQghRgVpwRNCFLXlD6yl9e9l3HfmAjzLoumAO/jkkrcVOiwhRBEZi5dJkRY8IURRu+9SzWMHziUTDJKzbf5z5MHcFvllocMSQoiCkhY8IURRK82kiaQyHLZkHZZneGHuJILRkkKHJYQoIu7Ya8CTBE8IUdwaNrdz4X811b09BHE58vVVGNsrdFhCiCIigyyEEGKU6YhEiGVSlJIiQpaaXDeobe4bLoQQ+xVJ8IQQRS0RLSUTs/vnFRBxc4ULSAhRdDy17aPYSYInhChav/v7Rl4/YBJLp47HzXex5CxF1rZoeXlDgaMTQojCkXPwxLA5jlMF3AIcBazQWh82ZP3lwKla6xMLEJ7YD736x9donzSdRHUp8UQvlT29BEyOqs4kD590Lxe0f6TQIQohisBYvEyKJHhid3wCiAPVWuu93gfmOM4jwANa66v2dt2isIxnSLzURqg2TM+abjY/00L1QVV4K7bg3f0K69tDdPVmqch0k+nwcLMeyyvHkbMtyroy5Gyb8t4e7GSWXCRHONxFhUpy7/s+ycpx47CM4dmGat7z3Mu888VFLK5owAThpzP/QzoaJppLEVRQu2UzE1NdrKmsJqsCWK5FXXs7KgfKtqg4MI5JGFLTKqlqiJFb30vde6dRcng9xoOS+ZWFfimFECPAHYODLCTBE7tjOvD63k7uHMcJaq2ze7NOMXoYz/DKOx5iy53rwQLbc7ExtJJhIc/TGB/P+tI5nNryNCVuigxBHi09kpwJgavIKEX1piSRjIUiwuqqKp4+dDrHPf8Ga2pqMErhKsVLE6tZHDqK6c3NHLhlE/8+7EhmLNlEpifLi0fOBEuxobuO6ude4uj1q7lvzoGUZNKUp9O02FWQM6R1FlDYb2wi5+UAi7Z7NgL+uX2TvnYQU797eCFfTiGEGBZJ8MSwOI5zJ3Bmfvq9wLXAc8APgcnAI8CKQeUVcBXwIaAUaAWu1Vr/3HGcE4EH8uu+DdQ6jvNn4DjgaMdxvgps0FrPydf1UeBzwCRgFfAVrfV9I7zLYi/pXdrpJ3cAHnhY2OQop51Sunij9AQm9zZR4qYACJGlszwM+V/UqVgAyxhUvgvl9RkNHLFoDTO6NxDO5cjZ+QEWBjKBAP8+YA7H3vsAbjjI0oUTCfVmwfK37YnHWDGugZldzdQnOqntSbDFqgBAeUD+OVxlk1OGgDH5Zf6o3PXXviYJnhBj0FgYVDGUDLIQw6K1Phe4Gfiz1joO3AjcBnwPqAB+Bnx00CanAR8EjtRalwJHAE8MWm8DbwUOAeq01p8GHge+o7WOD0nuvgJcBFQCXwducxxn5gjt6lYSiYRMv8npdDSHFRv8W9JPltJEMEBJrpeeQIzBQtmBRmLlbX3Jk4quXlAQzea48eYbaOhoh6wLiQwAcza3srGykmQoRC5gE+tJD6rLoySVxDYevaEwvcEQIeM3HpshB3hltr3USmjyQJyj4bWVaZne36bF8CmznYOYENvjOM4NQE5rfanjOF8HztRaHzdo/c3ABK31iflWuluBi4FHtNapQeVOBB4Gpmit1w1a/ghDzsFzHOc14Bqt9Y2Dlt0JPLuPztWTL8he0PZQE2uvXUygNEDb0y1kW1OEAtDQvYoobSwun0d1rpOGZAspN876YD3LJtfhGYvSRJZIOktJKoPxoLMsxIbJZUxY183pbU8Q8VzunzqDO+YeRNQt4ajVG3hjxjRcy6KyuZOcp1g/rZaaji5mNDZxxNplYHk8OWkewWyOOc0bSZooaQLYxsMykA3ZKFcRyriogKLypHqs0jBTv+cQm1NR6JdTiP3ZiLS1nfjxpm2O9Y/8dnxRt+tJF63YUxOBNUOWrQYmAGitH3Ec52vA5cDfHcd5Bvia1lrny3pA4zCeZxrwS8dxfjZoWQBY/yZiF/tY1cnjqTp5/A7XTxgyPw2/v35Hmj78W9IvPsfSmlrKsp0saF/Py9ljQFksmzaF2qYW4t3dvG/JuwiFg9utY/Zu74UQYqySW5UJMWADcMaQZVMHz2itfwf8znGcGPAt/C7dyfnVRms99BfT9u4vtRb4ptb6H282YDF2jP/Tx+FPH+cfc/6EG43QWjeD8nxXrm0MyyfWc8CiZTtM7oQQYqyTBE/sqb8BVziOcyHwD+BE4B2ABnAc5wggjD8QIw0kAHcXdTYDQ8+tuw74luM4y4FXgAhwGLBFa/3G3tgRUbz+39IP80j1T9ChGpLlZSj8XwkliSQfWf7uQocnhCgSci9aIfK01iuAdwNXAB3AZcAfBhWJAz8FtuCPoD0duGAX1V4HOI7jdDiOszj/PL8HrgGuB9qBdcA3AGmaEQA0xco57YXFZDIuTdEIbdi89clXCUTkIyKE2H/JIAshdk6+IKPcDZNv4KTG9ayjDg8bhUcFrSw0Xyl0aEKIvW9EmtqO+tSmbY71z/yqrqib9aSLVghR1Co7c7QTp55WeonQTZiALXm5EGL49vqtmUYBSfCEEEWtq6yEFlsRbS9B4ZGssgm324UOSwghCkoSPCFEUXMzaV6eNZWy3l48oDsa49P3X1josIQQRWQs3otWBlkIIYraJZsuoWFjC22lZaRCEcavaSRWXlLosIQQoqCkBU8IUfTKfxXnQLo499xzCx2KEKII5cZeA54keEIIIYTYv+VGZnBuQUkXrRCi6CXWZeh8ugu57JMQQvikBU8IUbTWvdTCypNuZVwqTMYK8p+f3UQ6kubdWy4tdGhCiCKSHXsNeJLgCSGK12PveoBqL0KmOkpnWYyy1l5MVi6RIoQQkuAJIYpWazxOiWfx2pypAKgJHvMWrSloTEKI4pOVy6QIIcTo8fzMqXTHov3zxrKwQwUMSAhRlLLbeRQ7SfCEEEUrG7QpTySwPA+AUCbLhLbWAkclhBCFJ120QoiiVZ71iCZTnPjMqyTiMao7EtiuV+iwhBBFple6aIUQYvRQBjaXl/DkofPYMK6K5ZPrebmulh8d9tdChyaEEAUlLXhir3Icpwq4BTgKWKG1PqzAIYkx6sbbNmErj9VTJ5MOh+kuiRFKZvDcNKvGWaTXbCE8tabQYQohikBy7DXgSYIn9rpPAHGgWmud21Vhx3E+C1wEHAhs1FrP3E6ZDwDfBMYDrwKf0lq/sFejFqNWb2+aRMrluUdbufaeTkrWdbKlqpqyUIRjNneQLh1031mlmL6mi4dmTOH6t/+DZjvGvI5GTnnqMqrrYqgx2A0jhHjzMmPwThaS4Im9bTrw+nCSu7yNwDXAXOBDQ1c6jnMs8GvgncCjwOeAux3HmaW17to7Ie+ftrS7fPeXW1jfnOOsE0r48PkV/etu/ncndz6QoL42wP/+Tw3PvtDLL27vJmkUC6aH+MzbY/zommZyKY9EKEhtbYCeLpe3HB7jk5dUYVmK2/7TyQN/2cCSeClYFk2hIBlL0ZDOMCGdpVfB1K4ElckUkZ40Bz+3msa6ch45dh7YNomAzUslMTpLwvTGQtheDRc3ruPKBx8ka1ncfcBCbj94IYev3cj4ji7cgE0go5iwYQv33XgzmYDNgi/8iO5oDPuXGU5YsoKDWzrwLMXb3riHdDDC3XNPpTzVw7lL7ubRGdP54Qlv47Y//pyZnY1kLQsUNFZU86UzP0zD+gx/OHYhnm1x4QKLm84L0doD77w5yaJmj9PnBLl7o03SAzAEPUO2OwdBi1NnW4wLwq1veGRsC3qyWFmXk6bb3Pb+GD97Oss1j2ewUORyhuOm2fzjfRHiYcXKVo93/SXJmnaPLx0f4hunhHf53n7rwQxXPWegN4PVnSESVNzxsVJOnBkc1mfj5iUen3vYoyQAfz3H5pgJY++PnxBjnZJb+4i9xXGcO4Ez87Np4FrgOeCHwGTgEWA5cIjW+sQh214CXD60Bc9xnD8Dltb6/fl5BawFvqG1/vNI7csgY/YL8rMb2rjvid7++Wu/Vsuc6WFWN2b43Lc29S8/+tAoj72coisw8HvwgECGbLt/IYG0bZGxBy4u/OVP1zCpIchlX24kjWJzLEK3ZdERGChzcKKH6mSSSZ2J/mXTlm8i6ykePu4AABpDQZaWRNlcHQfAcl1WX/VDIq4LQHM8zkfecwGXPP9q/8nEpR0pDmhcxyE8jcLl+ye9g6+99X0AlKQzfOue5/yCxsUEbHK2n/BU9bbx8Wev5yfHnsXnn7jHLwL9v+nvnX0Qv1l4Hm2hMI/PngTAvy8I8sjKHD95Mn9BhfIQ2ENOa+7NQs5AyAbXQMSGnAtdmf4in35LiF8874Jntvq0/eDMEF8+IcT5Nye59dWB30tLvhBj3rgdX8x5ZavHzF+kIWgRaunuX95QZbP6ioodbtcn4xpKf+aS8V9mFtTAq5dIW4AYNUbk14a6rG2bY725rqqof9nIIAux12itzwVuBv6stY4DNwK3Ad8BKoDrgI/sZrUHAf3dsVprA7yUXz7iEonEmJ32hhzOPC+/fMggVNczDP0duLMfhp4H3d09/lFYAWbbLHl7WxvAswaOp55Sfh355zIocoMSyXQgQFk2t9VBLBOyKS/Z0P8XoD0a26q+rZ9v0HPlp61B+zW4tGUMRimsQS9ad29ym9dql/Kvx2CpdGa7RXtTaT+2Ic+R6B5Iyrf3/g6EuPUTeWZ4n42uRPdW73ffoOTR8rmV6f17WgyfJHhiJF0IPKu1vkVrndNaPwjcvpt1lAKdQ5Z1AGV7Ib5dP3lp6Zidft/bypg6IUggAOeeHGfezDClpaXMmBLinWeWEgzApPEBPnJBJZe+s5So8VAY5k4K8NlPjYOQwgBJ26K80iZgw7FHxjji0CizZ1bwjndWUd/bi+3mqMzmCHseyhgmpDJEPUNLOEJbJIIHhHoyjG9sRaVzeDkXA1RmcwRcj7KeNMrzCHoeXzz7LNaXlbG6spLfHnscwVCYXuUnfT2hAHryeP4160i6rChJO8Q9sw6jJJ0ils5wyquryAEZS3Hyykc4ceWj5BSEs0lOW/4Qfzz8JK479q0sK5+Mi0XSCpK2A6ysGsfPjzyXCZs6eGLGBDCGd8xRvPfgEr56QpDDJliEbHjrRI9QX1ZojH+5lpyBgMWx0xXvmmthZVwIWX6LHnDMFJtrzinl6ycECQWhJAxBC06cbvOFE/yWy++dGWZOrUU4AF87KcQR03f+/s6qsfjKUTZKKdx4CAMEAorfXlAyrM9GTUUpvz3NIh6Euhj86lSrIJ9PmZbp7U2PGKW2fRQ56aIVe5XjODcAOa31pY7j/BqI93Wv5tdfBRy7G120LwM3aK1/MmjZ7cBKrfUXRmg3BpMvyD5ijMHLN1d5nodSit7eDIvWJulMGK6/s4u1azJEbQimXXpLSyizbFaXhGiqiZMKBcnZFp+840k+/MpjLK+2SRKiJNlDyTfO48h3TcaOhiiNBXBTGQKhACpgY+dbBY3noSz5zSvEKDcyXbRfaN+2i/bHlUWd5cmJFWIkbQBOH7Js8m7W8QpwaN9M/hy8g/G7fsUYopTqT7b6/i8rC3DsgX4369lv2fqSJz//3us88YJLNFxFd3Rg4MGcpg5ymVLefe97Cc6u3+5zBYPbHvokuRNCjCWS4ImR9DfgG47jvAf4J3A88A7gxb4CjuME8D+HQUA5jhMB0Fqn8kV+D9ybH2zxOPBZIAL8ax/tgxilPvO1eWQPvYsTXljJNy8+kS0VJZz84mqmr+uitdLbYXInhBDbGANdskNJgidGjNZ6heM45wM/AP6EP4r2L8DgbtjL8a9x1yeZ/1/l63jCcZxP4Sd6fdfBe6tcIkUABNIedYke/nL1v8gEAkQzOXpjhnOb90XvvRBCjF5yDp7YpxzHuQVIaK0/VuhYhkm+IKPYx095ksPWbGHWqrb+Zc3VES7ccmEBoxJCjKCROQfvSx3bnoP3o4qibtaTFjwxohzHeRvwBNAFnA2cB5xR0KDEmBFEsXRCDdNXt2Pnf6y+PLUBSe+EEPs7SfDESDsev3s2AqwDPqG1friwIYmxIqAUQWX47nlHc9LidayoryRTsuOLAAshxPYVdWPddkmCJ0aU1vpLwJcKHYcYm+K5HOPbu9gcj3LrSfOZsrmTS+56ATik0KEJIYrJ2MvvJMETQhSvoz4yiUW/Ws1hjc0c+/paqjd0k4kkd72hEEKMcTLIQoidky/IKGeM4UeX30P4lS186t8XEQhIF60QY9jIDLL4cue2gyyuKS/qdj1pwRNCFDWlFHOPcuGoSknuhBAiTxI8IYQQQuznirqxbrskwRNCFLX/q/oLVek0YHFT+d+4eON7Cx2SEKLYjL38Drn5ohCiaD1z7cuEAop146tYN76CqPH44dw7Ch2WEEIUnLTgCSGK1rKrF5Mpj9FVUYoyBuV51PZmCh2WEKLojL0mPEnwhBBFKwM0TaonGY8CkIpGqGxqLmxQQggxCkiCJ4QoXpbpT+48oLc0RjBZWtiYhBDFZ+w14EmCJ4QoXpGeLL3ATc4sWmNhFja18fae3kKHJYQoNmMwwZNBFkKIorVk4UwenNVAa0kElGLR+CremFhNNpMrdGhCCFFQ0oIn9jnHcbqB07TWTxc6FlHc1k6qI+AZlGf42t+e4OSX19BRFubh8f/k9NYrCh2eEKJojL0mPEnwxD6ntY73TTuOEwVuBA4GZgBXaK2vGlzecZwY8AvgXflF/wQ+rbWWm47up4wxXHH23UzoLuP4FxdjGUMk45IJWVR3pmirmljoEIUQoqAkwROFZoCngF8B399BmZ8Cc4E5+fL/Bn4MfHIfxCd2IusaLvp3jrtXG1DQ46qtfwi7BmUM2IpwzgXPkLIDWBiMMVhdaSJZl1BAccSWTiKuR08wRDZo0xIJ0mJZBFyPGT1pVgUCuLZFfSpNWClU2UwOb91INJPzn1LBpvoYU9cmmJjcxNOTvs4D44/mLWvW8NTk6WTtEBWZVua/kSSSy7BxWjXpEovq3g7aTBmesqiojzL9xDo2JQ0r792Icj0COQ8TsciqAFbYZvY5Ezjp6/MxnuGlXy6ldWknM8+dxPSzJhToXRBCvGljrwEPZYzcS12MDMdxPgtcBtQAXcCftdZfcxzHAMdprZ8YUv4R4IHBLXj5Fr424Byt9YP5ZacAdwJVWuvUCO+GfEF24rL7cvxEuxDIn85r7eQoaYCs55/5G7AIbOnBSvnnyp3ftJn6TJaeYJC2WLR/k8cmVNLQkaQTRcJSTHI9vypjeC0c5N4/3Mr66qr+8jlbMat5A0cmnyJIlpsOOZam7DwmtHXhBizaY2HC3TnKIopVU+oB/GvnNW4m4LrYrkc07dIbC5KMhfJxGzzLwth2/24cfdkcAtkcz12z2K/DgnfedhLVB1S86ddUCLFTI5KKqa/3bHOsN98tKeq0TwZZiBHhOM5s4Gr8xKwUmA/syS0G5gAR4IVBy14EosDsNxvnriQSCZneyfQrzVlQ+WPgrg6Ffevz5VXO619VkfMTvdyQBDGecwl5hpRSxAb9GFVKEc+5VCbTBF0X8BOvnrISls2bRIAsAAuaG9lYVYrKb1rZm6a5Ok5PLNxfl7EsvHyC6uVjswfFhlIYNRCXAtpWdbNlecdAHR50re/d5vWRaZmW6b07LYZPWvDEiHAcZzqwGPggcLfWunvQut1pwTsOeAywtNYmv8wC3O3VMQLkC7IT969yOeNvOYyt/MRNMZDwAfQdX5TyX0nX8/8PKKyeDHZHCgUc2ZngiI4EWUuxsbQUC+gJ2jw4vpLanjT1vRlWBG2m5TwUYOdcmjMZfnHbo0zubScRCdNYV8PK6ZOo7G5lZpvmjGWv8JUzLmLu64ag52EsxUtT6jjx+dWEShQvLZyBZ1nEe7op3dSFMoZQ1iOMob00Qs62/N3xPDwDbjjk56gBxbv/cjSkXe6+5AlyvS4VM0t5+99PIBQP7tPXX4j9kLTgDZOcgydGhNZ6leM4F+GfJ/cHx3EWAVdqre/bzar6frqVAx2DpsHv9hUFdNp0m2WfUPzfEpcJZYq/LzU8s9EQDBiqLI/Dyg3pnMcLjS6R3gyZALQQIZTNYiwbFVIoY2gpjXBPaZhQKkttMs266nJaQjZTm7cQT2YJBYPMS2RZWl3OvNZOZnR2c9aGzTTWlJLqCYFSjGtvJbYqyS9OOpx3vrqF/z1hIVWbobsqTXO4nKpkN8dsWIN11iSszUlOqWjDDbo0XDCFjW+UkKqrYNbhlVROL0WVBnnmh6+RWNdDw7xySmaUsXFRGxWzyplz9kRK6yIAvOe/p9G1roea+RUEY3I4FUKMHnJEEiNGa30bcJvjOCHgE8DtjuNU72Y1S4EUcCjwUH7ZIUASWLa3YhV7bmaVxdeP9bs4L1m4t2vf3seltn/qFzP/CUqxoGsF05MboB0SJWlSmQjff/adw36WKdtZdvL3D9tq/oALp21TpqQuSklddJvlQogiU9RtddsnCZ4YEY7jzAGm4XevJoFO/M45bztlw/hfLwsIOI4TAVytdVZrnXQc5ybgSsdxXstvciVw4z4YYCFGufquNMZSTEoN3H/27YtfYcWvLi5gVEKI4jP2MjwZZCFGSgi4AmjC71r9LHDeDpKypfhJ4HHAN/PTvx+0/vP4rXV9j6X4o3PFfm5TJAzhAEkV6V/WFK/gwA8eWsCohBCi8GSQhRA7J1+QUeynE/7Bgu4EtV1d1LIZMFx9yhn89IETCh2aEGJkjMwgi2/0bjvI4juxom7Wky5aIUTR6ggYqhMJXGyaGY9rwcJlzbveUAghxjhJ8IQQRWt6Vw9B42HyDa0Bz1CSzhY4KiFE0Snqtrrtk3PwhBBFKxwwWAEXhUFhSJUGKUlJgieE2F1qO4/iJi14QoiidX7Lh7iz7kbKSZENBGm1Y4RO390r8QghxNgjLXhCiKKllOKMte9lkxujJxWg9Jx63vnH0wodlhCi2Iy9BjxpwRNCFLdwNEz0hgoA3nruiQWNRQghRgtpwRNCCCGEGGMkwRNCFL3AxWupuWA1d1f9qtChCCGK0RjsopUETwhR1P5d+Vsm51K0jaugJBTg7urfFDokIYQoODkHTwhR1CLAi3WTKenN0FxRRzydLHRIQohio8ZAk90Q0oInhChqvbZNrCuHcW3KW3tJWsFChySEEAUnCZ4QoqgF0uH+X98KCCezbHpxQ2GDEkKIApMETwhR1DoiYbIBRSZokbUVOcui2yt0VEKIoiKDLIQQYvR4+ppneX1WHV2xMP9w5vKPw+eSqFT8+ZOvFDo0IYQoKBlkIfaY4zhVwC3AUcAKrfVhw9zuHuBhrfU1IxmfGPseu2ED8XCcH512BMvHVQGwvK6C//fEwwWOTAhRXMZAk90QkuCJN+MTQByo1lrnhruR1vqskQtJ7E9yQfjHkXNYWVPRv2xlTSVYbuGCEkKIUUASPPFmTAde353kToideXlxkptu6yDoGeZPsHj2wS1kU4buSIiNQZtoOsOimipCGOI5Q61zCB9+9jWeXjCX3kCAp+oqmLqplVXVM/j5vNtpqiilsSxOdW+KNRVxSjyXc19djTGGN2Y0EHENkVSKgAIVDaLiASJlQY66cCLz3zax0C+HEGJfGXsNeJLgiT3jOM6dwJn56fcBaeBsrfVTjuNMB1YC39FaX5EvswT4ltb6747jPAI8oLW+ynGcqcBq4APA/wKTgKeBD2qtmxzHGQ+8DHxRa31Tvq4/4ieXp2qtpalmjHh5SYorfrzFnzGGja/3Mq4zxeqGOoKuy6mr1vHLg+bghoI47d3U9PRQ05HgtVnTUcriucnVbCmN0FMa4rDWDkwkSm13D9VdCf48axoXvrKMgxubmNLeyZMHzcFEoySBXDBARWcC050h15OlO2G447o1ZLtzHPy+qYV8SYQQYo/JIAuxR7TW5wI3A3/WWseAh4FT86tPA1b0zTuOMwGYAzy4kyovAI4HJgAlwJX552kCLgJ+6TjOPMdxPgCcDVwoyd3YohcNukCxUoSyLh0lUVCKsmSKsOvSHI8xMZnBMgalLLbEohgUKpeiuTQCQDIc5IlZE8mGghj8g9z4ZIq2cIiJ7V0ANNVW9j9VNjRw3TzbGCzXxSjF0oda9sVuCyHEiJAET+wtDzCQ4J0KXA3McxynPD//ita6dSfbf1trvUVr3QX8FXD6VmitHwCuA24HfgG8T2vdPAL7sI1EIiHT+2j6wLmR/nmMIR20KUumAOiOhMlZFuWpDB3BACiFq6CpupIJW1oZl+hFGdO/edgzlHT3ogAPaIpGqE6n2RwvAaC2rau/bCA7cIaBpxTGtlDGMO2o6oK/JjIt0zK99fSIGYOXSVFm0EFRiN3hOM4NQE5rfanjOLOB14Bq/C7XhcDv84/zgCat9Zfz2z3Ctl20k7TW6/PrLwEu11rPHPRc9cBa4Bmt9Qn7ZAd98gXZhx56spu/39FJwDPMroElz3eAZ+iKhGmxLGKZFM+Pq6EuZ6jOZpnY0004a5je1s4LE2p4fvI4KnsznP7GOqau3Mi6ceVsLIlR252isTKOpeC9z78OwMtzphBUEO5NETAGghYmEiRYGsB5ez1HfmRGYV8MIcT2jEjqpa5Kb3OsN5eHizrNk3PwxF6htV7mOE4z8HmgWWu90XGcB/C7a08BLtnTuh3HsYAbgf8ARzuO82Gt9Z/efNRitDn5mDgnHxMftGTnAx0eqLiWjRVT6K4o5XDX44Ql60jaAV5pCPOD5e/YwVb+1XzevlciFkKI0Um6aMXe9ADwJeD+/PyDwMVAFfD4m6j3cvy/9B8ALgR+4jjOgjdRnxgjmu0ScsEwE1dvYe4rjUxeuRlj28zevLOzAYQQYgiltn0UOUnwxN70AFDGQIL3KpACntJaJ3e41U44jnMyftJ4vta6R2v9KHAN8A/HcUr2QsyiiE375DySEZvu0ggGaK8qoXlcBeHObKFDE0KIgpJz8ITYOfmCjHLPl36bZ+e9hXgySSDr8uBB8/jC/0Q58PgphQ5NCLH3jcw5eN/LbHsO3tdCRd2MJ+fgCSGKWktJNac/v6h/fnN5KZGS+QWMSAghCk+6aIUQRa03ULbVfKetmHFIXYGiEUKI0UESPCFEUatMpNhSEgVgdXUFDZu7sCw5tAkh9m9yFBRCFLXmWIxAr6KbKNWtaZScNSmE2F1j8ELHkuAJIYraxU0X89zM8axoqObxBVO55PW3FTokIUTRGXsZngyyEEIUvSnX2kCMy859a6FDEUKIUUESPCGEEELs34q/wW4b0kUrhBBCCDHGSIInhBBCCDHGSBetEEIIIfZv0kUrhBBCCCFGO0nwhBBCCCHGGEnwhBBCCCHGGDkHTwghhBD7NzkHTwghhBBCjHaS4AkhhBBCjDGS4AkhhBBi/6bUto9tiqg1SqkFBYhuj0iCJ4QQQggxxkiCJ4QQQoj9m9rOYzibKfUBpdSrSqlFSql/KaXG5Zc/rZQ6PD/9K6XU4vx0QCm1RSlVMhK7MZiMohViJ5RS/wVqCh3H9gQCgZpcLrel0HGMFvJ6bE1ej23Ja7K1In097jXGnLm3KzVfCuz2ONp8d+3VwGHGmCal1HeAnwMXAA8CpwDPA8cCSaXUeGAq8Loxpmdvxb4jkuAJsRMjcSDZWxzH0Vrr/9/enUdJWZ15HP/+IAZF3IaICUFZVFRGxeDNpkk00WgUzIQwOROMAdQZMaNmOTOYGEx09DgoLpklZtCJCi5DPERNYgQVjTiaxMSrgGEcFwgtgiLghhgHAjzzx72lbzd0d3Xb3W931fM5p05Xve9b933urequp+/yVig7ju7C26Mxb49teZs05u3xrn0amGtmL+bH1wCL8/37gamSbgFeBh4kJXxDgV91RXA+ROucc84517F+A4wCRpOSvUqP3jH5fqfzBM8555xzru0eAE6U9P78+O+A+QBmthF4HPgOcB/wCHAkcGi+3+l8iNa5nuvasgPoZrw9GvP22Ja3SWPeHm13n6TNhcfnAfMlGfBHYHJh3/3Ah4FHzWyLpKXAcjPb1BWBysy64jzOOeecc66L+BCtc84551yN8QTPOeecc67G+Bw853qwEMLVpFVZG4ENwDdijLHcqLpeCGE4MAvoT7okwYQY47PlRlWOEEJ/4CZgX2AT8CwwOca4ttTAuoEQwgXAhcAhMcYlJYdTmhDCjsAPgGOB/wN+G2M8o9yoXEfzHjznerZ5pA+rkcA04NaS4ynLDODqGONw4GrS9ajqlQHTY4wHxBgPAZaRLsZa10IIo4CPAc+VHUs3MJ2U2A3P75HvlRyP6wS+yMK5GpF7bl4Adooxbi07nq4SQhgAPAP0jzFuCSH0JvXi7e+9VhBCGAd8LcZ4bNmxlCWE0AdYAIzPP8fUaw9eCKEfsBIYFGPcUHY8rvN4D55zteNs4K56Su6yvYFVMcYtAPnnC3l7XQsh9AK+Bvyi7FhKdhFwc4yxoexAuoF9Sf8AXRBCiCGEBSGET5QdlOt4PgfPuW4shPA4sE8zu/eqJDUhhC8DJwOf6qrYXI/w76S5mT8sO5CyhBA+DgTSBWcd9AaGAQtjjFNCCB8F7gwh7BdjXF9ybK4DeYLnXDcWYxzV2jEhhLHAJcAxMcaXOj+qbud54IMhhN6FIdqBeXvdCiFcAewPnFSHvbpFRwEHActDCACDgHtCCKfGGO8tNbJyrAA2A7MBYoy/CyGsA4YDdbdAq5b5EK1zPVgIYQxwFXB8vQ4/xRjXAItI86vIPxfW8/y7EMI/A4cDX4gxbiw7njLFGC+NMQ6MMQ6JMQ4hzT87vk6TO2KM60hfsfVZeHsF+gBgaZlxuY7niyyc68FCCGtJl8IoJjPHxBhfLimkUoQQDiRdJmUP4FXSZVKeLjeqcoQQ/hJYQlp48lbevDzGOLa8qLqPEEIDdbzIAiCEMAy4nnRZoT8DU2OM88qNynU0T/Ccc84552qMD9E655xzztUYT/Ccc84552qMJ3jOOeecczXGEzznnHPOuRrj18FzzjnnXKcKIVwPjAHWxBgPbuXYwaRVvnsCrwCnxBhXdn6UtcV78JxzNUfSEEkmaVAnn+dMSTcVHs+TdG5nntNtn6SlkiZVeWyXvD+6gqQ+ue4Hlh1LK2YCn6vy2CuAG2OMh5K+Zm5aZwVVyzzBc66OSRomaY6k1ZI2SHpe0h2S3pv3T5K0zQVQW9j+lfzBecF29i2QtDGf53VJCyWN65yadT5JO5M+fC6sbDOzE8xsemlBtSK/Nv69o12gM9pa0tGSNhe3mdlG4PJ867ZijP9N6o17Wwhh3xDC3SGEx0IID+XrWQKMAH6V7z8A/FUXhlozPMFzrr7NBV4EDgB2AT4O3AOoneVNJv0RP11S7+3sv9jM+pEusDobuFXS8Haeq2ynAH8ws2VlB+Lq3mzgM5L2KzuQNroWOCfGeDjwj8CP8vbFwBfz/bHALiGE/iXE16N5gudcnZLUn5TYzTCz1y1ZaWYzcq9AW8s7CPgkMBH4AHBCc8ea2WbSH/PewCHbKessSYuabBsqaYukIfnxDbnH8Q1JT0o6uYXYLpR0X5NtCySdX3h8sKR7JK2VtELSNEk7tFDlLwDzmyuzMAw4Mcf3pqS5kvaQdKmkNbnn9KzC8yfl4bZvS3oxH3NlMY7W6i3pUEl353q8Uqm3pMX5kHtzL+qPm2mrvpL+NZ9jnaSfSdqnSR2vlHRbjmGZpGZ7WAp1+paklfk5V0jqn8tYL+mpYm+XpPdI+r6kP0p6VdL9kg4u7N9B0lWFNvz2ds77SUkP5zZYJukfJFX9j4ukcZIW597mxZLGFvZt04MtaWalTZtra0kNuV4P5+1R0oe3V0ZhW4OkUyQNBOYBvfNzN0iaCGBm64FHgc9XW7+yhRD6AUcAc0IIi4BrSH83ICV7R4UQFpK+S3gVsKWMOHsyT/Ccq1Nm9jLwP8CPJU2QNKItH4DbcQbwhJn9ktQzOLm5A5WGgM8ifU3S4u0c8l/AgZIOK2ybBCwws4b8+GHgMGB30lDpTEkj2hO4pAHAg8DtwAdJPZmfBc5r4WmjgCerKH4c8AlgH2AI8DtgGTAQOBX4l2ICBQzOxw7LcZwETCnsb7bekj6Q6/FgPtf7gUsBzGxkfv5xZtbPzP62mXh/AHws3wYD64A71bhHdiJwJbAb8ENglqS+LbTB4BzvsNwW55CSlctJXy93O3BD4fgpwATgxFyHh4D5knbN+79DmrB/BDA013Vw5cm5Pebm8vcERgNnA19tIca3SToCuCWfpz/wXWC2pI9W8/xW2vpM4BvAXwA/BeYW6tVSmS+Q/mnaksvsZ2azCof8gfSe7Cl6Aa/FGA8r3A4CiDG+EGP8YozxQ8DUvO21EmPtkTzBc66+HQ0sAL4JLAJekvS9JoneUEmvFW+8M5QCgKQdSR/IlQ/p64ATtO0k9qn5+StJ82rGmdk2c/nM7FXg56QEiBzPRNLKusox15nZy2a2xcx+AjyR69MeE4DFZnaNmW0ys1Wkid0TWnjOHsD6Ksq+2MxeyQn1L4E/m9l/mtlmM5tH+u7cDxWO3wpMMbO38vDvdFJyC7Ra768CS81smpm9mevSqOeyJZJ6kdr5fDNbZWZvkt4bBwEfKRx6q5n9xsy2kobZdgP2b6Hot4B/yvEsJiX1j5rZI2a2BbgZ2E/Sbvn4U4HLzOyp3Jt8EakHZ3TePyHvX2pmb5F6fIrfu/n3wBwz+3lup6dIiWhLr2fRJOA2M5uXX6e7gDuA06p8fkuuM7PHzGwTcBmpbcZ0QLnrSUljjxBjXA8sDyF8CSCEoBDCyHz/fSGESn5yHoXfe1c9T/Ccq2Nmts7Mvmtmo0g9LOcC3ycnVtlyM9u9eCN9gBZ9CehH+qCG1HuyFmjaS3RJLmOAmR1hZne2EN4NwMl5ePIzOb7bISUiki6S9HQeQnsNGEnqrWmPocCRTZLY60m9R815FWi154U0x7HiT00eV7btUni8xsz+VHjcAAyCquo9BHimipiasyfQB1he2WBmG4A1wN6F414s7H8z3y3Woak1ORmsaNoOlfpWyti7SQxbSe1QiWFQflyMYU2hvKHA+Cav5wW8MwTYmkbnz5bRuA3aq6Fyx9KXwa8gv77v0q40WcTQnYQQZgO/BQ4IIawMIZwOfAU4PYSwmDSaUBnqPxp4OoTwDLAXcEkJIfd4fh085xwAOamYKekc0hBgW5xBmk+3pND5tztpscXFuZemreYDG0lDlGOBn+TeGoDxpOTxOOBJM9sqKdL84pA3gJ2bbBtYuP8ccJ+ZjaZ6C0mr/X7RhudUY4CkvoUkbwipxxNar3cD8NctlG0t7IOUlG/M51wKIKkfMAB4vk21eHeezzGQY+iVH1diWNVk/840Tu6fA643s7Non0bnz4YVzt/c+2lF4XFzbf12ublneh/eeX3fAN5X2P8eUttXFJPkpg4m9RB3SzHG8c3s2ubSKTHGn5KGr9274D14ztUppcn+05QWF+yQJ7aPI31QPNSGckaQ5lWNJSWGldtHSD1gJ7YnvpwU3gh8nbSirjhMsyuwmZSQ9JJ0GqknqzmPAaMkHZ7reTapl6fiRiBIOk3SjrmnbJiklq7b9TPg2DZXrHW9gMsk7SRpGGn4sTLXqrV63wwcoLRIo6+k90oqxriaFoZSc0/ZjcDFkgbmeXVXAk8Bv++g+lVjJnCupOF5vuZUUofEXXn/TcAUSftK2ok0jF38PPsR8GVJJxXe2yMkHVXl+WcB4yQdL6m3pBNI78HKFIRFpER8TH6vjAU+1aSM5tr6NEmjcs/0FKBvoV6PAccoLSjqQ+q5Ki70WU1aZFF87yJpF9LvW0f/s+F6ME/wnKtfm0i9A7eThnbWAucDXzezOW0oZzLwuJndaWarC7cngDm0sNiiCjeQVtEtN7NigjGLtFhhKak3ZwQtJKVmtgC4CribNDS4F/Drwv7VwKdJK2MbSMOvd5B6bZpzEzAyJ2Ed6TlSj85yUh3vJiUw0Eq980T8o0kLRFaSEoLiAo2pwEVKK1Ovaeb83wIiaVXmCtKw5ufb2QvbXpeTLv1xL/ASaYj+uLxaFNL8yHuAR0jttILUbgCY2RLSvLZvkl7vNaSksaohfDP7NWku4hWk98J04BQzeyTvX0ZaKHEt6Xfnc8BtTYpprq2vBf4tl/s3wGgzez3vu4WUpD1OGhJeQXqdK3E9A/wH8Ps89FxZNDIeeMDMnq2mfq4+KE0BcM4511aSzgSONLOqVmdWUd4k0gKHnnY9M1cFSQ2k1/fm1o5tQ5l9gCWkJPx/O6pc1/P5HDznnGsnM5sBzCg7Dle/8irjllYwuzrlQ7TOOeecczXGh2idc84552qM9+A555xzztUYT/Ccc84552qMJ3jOOeecczXGEzznnHPOuRrjCZ5zzjnnXI35f9Aa7BP/rJxfAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 576x568.8 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "shap_values = shap.Explainer(model).shap_values(tropic_val.drop(['actual_load'],axis=1))\n",
    "shap.summary_plot(shap_values, tropic_val.drop(['actual_load'],axis=1))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2dd6478a",
   "metadata": {},
   "source": [
    "The y-axis indicates the variable name, in order of importance from top to bottom.\n",
    "On the x-axis (Impact on model output), the horizontal location shows whether the effect of that value is associated with a higher or lower prediction. Gradient colour indicates feature value."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d0324300",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "LightGBM_training.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.8.8"
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}