tuning_experiments.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "c45e2dd6-e02e-4117-963c-a6e4df39352b",
   "metadata": {},
   "source": [
    "# Tuning experiments for the Movielens dataset\n",
    "\n",
    "Feel free to re-run this and change hyperparameters as you see fit.\n",
    "\n",
    "This notebook should also be a good place to load the data only once and then train multiple times, generate charts, etc.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "4d1dc3c1-c8d8-4e91-bbc3-bf22424ab31e",
   "metadata": {},
   "outputs": [],
   "source": [
    "from data_loader import load_data, train_test_split, MovielensDataset\n",
    "import numpy as np\n",
    "import torch\n",
    "from torch import nn\n",
    "from torch.optim import Adam\n",
    "import argparse\n",
    "import yaml\n",
    "from torch.utils.data import DataLoader\n",
    "from tqdm.notebook import tqdm\n",
    "from sklearn.metrics import mean_squared_error\n",
    "\n",
    "from gmf import GMF\n",
    "from ncf_mlp import NCF_MLP\n",
    "from neural_mf import NEURAL_MF\n",
    "\n",
    "from train import train_gmf, train_joint_nerual_mf, train_mlp, train_neural_mf, train_loop, test_loop\n",
    "\n",
    "from matplotlib import pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "eb785f3c-40a4-44ef-8e17-99a64cf0bc77",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Starting hyperparameters\n",
    "learning_rate = 0.05\n",
    "weight_decay = 0.0000001\n",
    "epochs = 2\n",
    "batch_size = 1024\n",
    "latent_dims = 16\n",
    "alpha = 0.5\n",
    "loss = nn.MSELoss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "1e5df12f-f52c-4d67-b87c-fad5bccdbc10",
   "metadata": {},
   "outputs": [],
   "source": [
    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "3ce424df-270a-438a-a667-6100406c869f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "   user_id  movie_id  rating  timestamp\n",
      "0        0         0     0.6  881250949\n",
      "1        1         1     0.6  891717742\n",
      "2        2         2     0.2  878887116\n",
      "3        3         3     0.4  880606923\n",
      "4        4         4     0.2  886397596\n"
     ]
    }
   ],
   "source": [
    "# Load the data\n",
    "dataset = '100k'\n",
    "data, num_users, num_items = load_data(dataset, rescale_data=True)\n",
    "print(data.head())\n",
    "train, test = train_test_split(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "4709cca1-1aee-4846-9f2e-4e0778e15cd0",
   "metadata": {},
   "outputs": [],
   "source": [
    "train = MovielensDataset(users=train['user_id'], movies=train['movie_id'], ratings = train['rating'])\n",
    "test = MovielensDataset(users=test['user_id'], movies=test['movie_id'], ratings = test['rating'])\n",
    "\n",
    "train_dataloader = DataLoader(train, batch_size=batch_size, shuffle=True, num_workers=4)\n",
    "test_dataloader = DataLoader(test, batch_size=batch_size, shuffle=True, num_workers= 4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "16e4d2cf",
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_learning_curve(param_history, curves, title='Training Curve', save=False):\n",
    "    \"\"\"Plot learning curves for a given model's tuning history\n",
    "\n",
    "    curves -- a dictionary from a curve label to its historical data. The length of the curve history should be equal\n",
    "              to the length of param_history\n",
    "    \"\"\"\n",
    "    fig, ax = plt.subplots(figsize=(8, 6))\n",
    "    curve_labels = curves.keys()\n",
    "    for curve_label in curve_labels:\n",
    "        curve_data = curves[curve_label]\n",
    "        ax.plot(param_history, curve_data, label=curve_label)\n",
    "\n",
    "    fig.suptitle(title)\n",
    "    fig.legend()\n",
    "    if save:\n",
    "        t = title.lower().replace(\" \", \"-\").replace(\".\",\"\")\n",
    "        fig.savefig(f\"images/{t}.png\")\n",
    "    fig.show()\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5b7dd4c7-857a-42fb-89f0-69611ed4292b",
   "metadata": {
    "jp-MarkdownHeadingCollapsed": true
   },
   "source": [
    "## Tune Hyperparameters for Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "d2007a37-f6c6-4ed4-b09a-11a45ada42a2",
   "metadata": {},
   "outputs": [],
   "source": [
    "def tune_lr(model_type='gmf', verbose=False, gmf_model=None, mlp_model=None):\n",
    "    \"\"\"Tune Learning Rate for a model type\"\"\"\n",
    "    learning_rates = np.array([0.001, 0.01, 0.05, 0.07, 0.1, 0.2])\n",
    "    rmse_history = []\n",
    "    test_rmse_history = []\n",
    "    best_model = None\n",
    "    best_lr = 0.01\n",
    "    best_RMSE = np.inf\n",
    "    for lr in learning_rates:\n",
    "        if model_type == 'gmf':\n",
    "            if verbose:\n",
    "                print('TRAINING GMF')\n",
    "            train_model = train_gmf\n",
    "            model = train_model(train_dataloader, test_dataloader, num_users, num_items, epochs, latent_dims,\n",
    "                                lr, # Param being tuned\n",
    "                                'adam', criterion=loss, device=device, weight_decay=weight_decay, rescale_data=True)\n",
    "        elif model_type == 'mlp':\n",
    "            if verbose:\n",
    "                print('TRAINING MLP')\n",
    "            train_model = train_mlp\n",
    "            model = train_model(train_dataloader, test_dataloader, num_users, num_items, epochs, latent_dims,\n",
    "                                lr, # Param being tuned\n",
    "                                'adam', criterion=loss, device=device, weight_decay=weight_decay, rescale_data=True)\n",
    "        else:\n",
    "            if verbose:\n",
    "                print('TRAINING NMF')\n",
    "            train_model = train_neural_mf\n",
    "            model = train_model(train_dataloader, test_dataloader, gmf_model, mlp_model, num_users, num_items, epochs, latent_dims,\n",
    "                                lr, # Param being tuned\n",
    "                                'adam', criterion=loss, device=device, alpha=alpha, weight_decay=weight_decay, rescale_data=True)\n",
    "\n",
    "        rmse = test_loop(train_dataloader, model, loss, device, rescale_data=True)\n",
    "        rmse_history.append(rmse)\n",
    "        test_rmse = test_loop(test_dataloader, model, loss, device, rescale_data=True)\n",
    "        test_rmse_history.append(test_rmse)\n",
    "\n",
    "        # Select the best model based on TRAIN RMSE, not TEST RMSE\n",
    "        if rmse < best_RMSE:\n",
    "            best_lr = lr\n",
    "            best_RMSE = rmse\n",
    "            best_model = model\n",
    "\n",
    "    rmse_history = np.array(rmse_history)\n",
    "    test_rmse_history = np.array(test_rmse_history)\n",
    "    test_rmse = test_loop(test_dataloader, best_model, loss, device, rescale_data=True)\n",
    "\n",
    "    if verbose:\n",
    "        print()\n",
    "        print('Finished tuning model', model_type)\n",
    "        print('Best Learning Rate:', best_lr)\n",
    "        print('Best Train RMSE:', best_RMSE)\n",
    "        print('Best Test RMSE:', test_rmse)\n",
    "\n",
    "    curve_data = {\n",
    "        'Train RMSE': rmse_history,\n",
    "        'Test RMSE': test_rmse_history,\n",
    "    }\n",
    "\n",
    "    return (best_lr, best_model, learning_rates, curve_data)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "1fd4ed52-ae40-47e9-9ee6-ff61c7db424a",
   "metadata": {},
   "outputs": [],
   "source": [
    "def tune_dims(model_type='gmf', verbose=False, gmf_model=None, mlp_model=None):\n",
    "    \"\"\"Tune Embedding Dimensions for a model type\"\"\"\n",
    "    dims = [4, 8, 16, 32]\n",
    "    rmse_history = []\n",
    "    test_rmse_history = []\n",
    "    best_model = None\n",
    "    best_dim = 0.01\n",
    "    best_RMSE = np.inf\n",
    "    for dim in dims:\n",
    "        if model_type == 'gmf':\n",
    "            if verbose:\n",
    "                print('TRAINING GMF')\n",
    "            train_model = train_gmf\n",
    "            model = train_model(train_dataloader, test_dataloader, num_users, num_items, epochs,\n",
    "                                dim, # Param being tuned \n",
    "                                learning_rate, 'adam', criterion=loss, device=device, weight_decay=weight_decay, rescale_data=True)\n",
    "        elif model_type == 'mlp':\n",
    "            if verbose:\n",
    "                print('TRAINING MLP')\n",
    "            train_model = train_mlp\n",
    "            model = train_model(train_dataloader, test_dataloader, num_users, num_items, epochs,\n",
    "                                dim, # Param being tuned \n",
    "                                learning_rate, 'adam', criterion=loss, device=device, weight_decay=weight_decay, rescale_data=True)\n",
    "        else:\n",
    "            if verbose:\n",
    "                print('TRAINING NMF')\n",
    "            train_model = train_neural_mf\n",
    "            model = train_model(train_dataloader, test_dataloader, gmf_model, mlp_model, num_users, num_items, epochs,\n",
    "                                dim, # Param being tuned \n",
    "                                learning_rate, 'adam', criterion=loss, device=device, alpha=alpha, weight_decay=weight_decay, rescale_data=True)\n",
    "\n",
    "        rmse = test_loop(train_dataloader, model, loss, device, rescale_data=True)\n",
    "        rmse_history.append(rmse)\n",
    "        test_rmse = test_loop(test_dataloader, model, loss, device, rescale_data=True)\n",
    "        test_rmse_history.append(test_rmse)\n",
    "\n",
    "        # Select the best model based on TRAIN RMSE, not TEST RMSE\n",
    "        if rmse < best_RMSE:\n",
    "            best_dim = dim\n",
    "            best_RMSE = rmse\n",
    "            best_model = model\n",
    "\n",
    "    rmse_history = np.array(rmse_history)\n",
    "    test_rmse_history = np.array(test_rmse_history)\n",
    "    test_rmse = test_loop(test_dataloader, best_model, loss, device, rescale_data=True)\n",
    "\n",
    "    if verbose:\n",
    "        print()\n",
    "        print('Finished tuning model', model_type)\n",
    "        print('Best Embedding Dim:', best_dim)\n",
    "        print('Best Train RMSE:', best_RMSE)\n",
    "        print('Best Test RMSE:', test_rmse)\n",
    "\n",
    "    curve_data = {\n",
    "        'Train RMSE': rmse_history,\n",
    "        'Test RMSE': test_rmse_history,\n",
    "    }\n",
    "\n",
    "    return (best_dim, best_model, dims, curve_data)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "id": "b673bd21-50b7-4526-8327-2206ec041712",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1\n",
      "------------------------\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.616572  [ 1024/15945812]\n",
      "loss: 0.042813  [1025024/15945812]\n",
      "loss: 0.042631  [2049024/15945812]\n",
      "loss: 0.042374  [3073024/15945812]\n",
      "loss: 0.041493  [4097024/15945812]\n",
      "loss: 0.042831  [5121024/15945812]\n",
      "loss: 0.043302  [6145024/15945812]\n",
      "loss: 0.040166  [7169024/15945812]\n",
      "loss: 0.042919  [8193024/15945812]\n",
      "loss: 0.045461  [9217024/15945812]\n",
      "loss: 0.042116  [10241024/15945812]\n",
      "loss: 0.036905  [11265024/15945812]\n",
      "loss: 0.032219  [12289024/15945812]\n",
      "loss: 0.031665  [13313024/15945812]\n",
      "loss: 0.030532  [14337024/15945812]\n",
      "loss: 0.029799  [15361024/15945812]\n",
      "Test MSE 0.810639369336749\n",
      "Test RMSE 0.9003551351198864\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.031191  [ 1024/15945812]\n",
      "loss: 0.030186  [1025024/15945812]\n",
      "loss: 0.035146  [2049024/15945812]\n",
      "loss: 0.029244  [3073024/15945812]\n",
      "loss: 0.025950  [4097024/15945812]\n",
      "loss: 0.024058  [5121024/15945812]\n",
      "loss: 0.028180  [6145024/15945812]\n",
      "loss: 0.025063  [7169024/15945812]\n",
      "loss: 0.029553  [8193024/15945812]\n",
      "loss: 0.025564  [9217024/15945812]\n",
      "loss: 0.024492  [10241024/15945812]\n",
      "loss: 0.023264  [11265024/15945812]\n",
      "loss: 0.027715  [12289024/15945812]\n",
      "loss: 0.028063  [13313024/15945812]\n",
      "loss: 0.025449  [14337024/15945812]\n",
      "loss: 0.025630  [15361024/15945812]\n",
      "Test MSE 0.7254787446116683\n",
      "Test RMSE 0.8517504004176741\n",
      "Test MSE 0.6117628504112673\n",
      "Test RMSE 0.7821527027449738\n",
      "Test MSE 0.7254787446116675\n",
      "Test RMSE 0.8517504004176737\n",
      "Epoch 1\n",
      "------------------------\n",
      "loss: 1.237849  [ 1024/15945812]\n",
      "loss: 0.044624  [1025024/15945812]\n",
      "loss: 0.045698  [2049024/15945812]\n",
      "loss: 0.041967  [3073024/15945812]\n",
      "loss: 0.042927  [4097024/15945812]\n",
      "loss: 0.035275  [5121024/15945812]\n",
      "loss: 0.035182  [6145024/15945812]\n",
      "loss: 0.032323  [7169024/15945812]\n",
      "loss: 0.031493  [8193024/15945812]\n",
      "loss: 0.030996  [9217024/15945812]\n",
      "loss: 0.029785  [10241024/15945812]\n",
      "loss: 0.033683  [11265024/15945812]\n",
      "loss: 0.028916  [12289024/15945812]\n",
      "loss: 0.029458  [13313024/15945812]\n",
      "loss: 0.027743  [14337024/15945812]\n",
      "loss: 0.029174  [15361024/15945812]\n",
      "Test MSE 0.8154931483799089\n",
      "Test RMSE 0.9030465925852934\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.031353  [ 1024/15945812]\n",
      "loss: 0.028242  [1025024/15945812]\n",
      "loss: 0.029579  [2049024/15945812]\n",
      "loss: 0.028773  [3073024/15945812]\n",
      "loss: 0.029898  [4097024/15945812]\n",
      "loss: 0.028712  [5121024/15945812]\n",
      "loss: 0.028019  [6145024/15945812]\n",
      "loss: 0.030076  [7169024/15945812]\n",
      "loss: 0.028788  [8193024/15945812]\n",
      "loss: 0.029974  [9217024/15945812]\n",
      "loss: 0.030598  [10241024/15945812]\n",
      "loss: 0.027138  [11265024/15945812]\n",
      "loss: 0.031333  [12289024/15945812]\n",
      "loss: 0.028811  [13313024/15945812]\n",
      "loss: 0.026781  [14337024/15945812]\n",
      "loss: 0.030035  [15361024/15945812]\n",
      "Test MSE 0.8144820786843522\n",
      "Test RMSE 0.90248660859004\n",
      "Test MSE 0.7190097952504886\n",
      "Test RMSE 0.8479444529274831\n",
      "Test MSE 0.8144820786843519\n",
      "Test RMSE 0.9024866085900399\n",
      "Epoch 1\n",
      "------------------------\n",
      "loss: 0.941630  [ 1024/15945812]\n",
      "loss: 0.039735  [1025024/15945812]\n",
      "loss: 0.046250  [2049024/15945812]\n",
      "loss: 0.042319  [3073024/15945812]\n",
      "loss: 0.047827  [4097024/15945812]\n",
      "loss: 0.046706  [5121024/15945812]\n",
      "loss: 0.044775  [6145024/15945812]\n",
      "loss: 0.045740  [7169024/15945812]\n",
      "loss: 0.044647  [8193024/15945812]\n",
      "loss: 0.043586  [9217024/15945812]\n",
      "loss: 0.042244  [10241024/15945812]\n",
      "loss: 0.044278  [11265024/15945812]\n",
      "loss: 0.044735  [12289024/15945812]\n",
      "loss: 0.043797  [13313024/15945812]\n",
      "loss: 0.044379  [14337024/15945812]\n",
      "loss: 0.048902  [15361024/15945812]\n",
      "Test MSE 1.0909597708259973\n",
      "Test RMSE 1.0444901966155533\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.042378  [ 1024/15945812]\n",
      "loss: 0.044628  [1025024/15945812]\n",
      "loss: 0.044835  [2049024/15945812]\n",
      "loss: 0.044899  [3073024/15945812]\n",
      "loss: 0.047098  [4097024/15945812]\n",
      "loss: 0.043482  [5121024/15945812]\n",
      "loss: 0.044013  [6145024/15945812]\n",
      "loss: 0.043742  [7169024/15945812]\n",
      "loss: 0.047529  [8193024/15945812]\n",
      "loss: 0.044552  [9217024/15945812]\n",
      "loss: 0.042531  [10241024/15945812]\n",
      "loss: 0.046297  [11265024/15945812]\n",
      "loss: 0.048002  [12289024/15945812]\n",
      "loss: 0.046109  [13313024/15945812]\n",
      "loss: 0.049526  [14337024/15945812]\n",
      "loss: 0.044779  [15361024/15945812]\n",
      "Test MSE 1.1004733835404976\n",
      "Test RMSE 1.0490345006435668\n",
      "Test MSE 1.1428705764665428\n",
      "Test RMSE 1.0690512506267147\n",
      "Test MSE 1.100473383540498\n",
      "Test RMSE 1.049034500643567\n",
      "Epoch 1\n",
      "------------------------\n",
      "loss: 0.644085  [ 1024/15945812]\n",
      "loss: 0.045673  [1025024/15945812]\n",
      "loss: 0.043694  [2049024/15945812]\n",
      "loss: 0.046016  [3073024/15945812]\n",
      "loss: 0.043027  [4097024/15945812]\n",
      "loss: 0.044833  [5121024/15945812]\n",
      "loss: 0.046092  [6145024/15945812]\n",
      "loss: 0.041812  [7169024/15945812]\n",
      "loss: 0.044514  [8193024/15945812]\n",
      "loss: 0.044354  [9217024/15945812]\n",
      "loss: 0.050106  [10241024/15945812]\n",
      "loss: 0.044922  [11265024/15945812]\n",
      "loss: 0.045146  [12289024/15945812]\n",
      "loss: 0.042662  [13313024/15945812]\n",
      "loss: 0.045327  [14337024/15945812]\n",
      "loss: 0.043629  [15361024/15945812]\n",
      "Test MSE 1.0929208912387496\n",
      "Test RMSE 1.045428568214371\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.044798  [ 1024/15945812]\n",
      "loss: 0.044093  [1025024/15945812]\n",
      "loss: 0.047391  [2049024/15945812]\n",
      "loss: 0.048437  [3073024/15945812]\n",
      "loss: 0.049623  [4097024/15945812]\n",
      "loss: 0.044805  [5121024/15945812]\n",
      "loss: 0.042060  [6145024/15945812]\n",
      "loss: 0.045815  [7169024/15945812]\n",
      "loss: 0.041526  [8193024/15945812]\n",
      "loss: 0.046877  [9217024/15945812]\n",
      "loss: 0.046326  [10241024/15945812]\n",
      "loss: 0.048386  [11265024/15945812]\n",
      "loss: 0.046513  [12289024/15945812]\n",
      "loss: 0.048327  [13313024/15945812]\n",
      "loss: 0.044897  [14337024/15945812]\n",
      "loss: 0.045607  [15361024/15945812]\n",
      "Test MSE 1.1011671816954498\n",
      "Test RMSE 1.0493651326852107\n",
      "Test MSE 1.1440778726466236\n",
      "Test RMSE 1.0696157593484792\n",
      "Test MSE 1.1011671816954507\n",
      "Test RMSE 1.0493651326852111\n",
      "Epoch 1\n",
      "------------------------\n",
      "loss: 1.224757  [ 1024/15945812]\n",
      "loss: 0.043672  [1025024/15945812]\n",
      "loss: 0.049623  [2049024/15945812]\n",
      "loss: 0.043237  [3073024/15945812]\n",
      "loss: 0.046901  [4097024/15945812]\n",
      "loss: 0.043243  [5121024/15945812]\n",
      "loss: 0.041382  [6145024/15945812]\n",
      "loss: 0.049463  [7169024/15945812]\n",
      "loss: 0.050312  [8193024/15945812]\n",
      "loss: 0.049186  [9217024/15945812]\n",
      "loss: 0.046622  [10241024/15945812]\n",
      "loss: 0.044973  [11265024/15945812]\n",
      "loss: 0.044021  [12289024/15945812]\n",
      "loss: 0.047850  [13313024/15945812]\n",
      "loss: 0.047781  [14337024/15945812]\n",
      "loss: 0.045866  [15361024/15945812]\n",
      "Test MSE 1.1168009121212616\n",
      "Test RMSE 1.0567880166434807\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.047913  [ 1024/15945812]\n",
      "loss: 0.046462  [1025024/15945812]\n",
      "loss: 0.049269  [2049024/15945812]\n",
      "loss: 0.049125  [3073024/15945812]\n",
      "loss: 0.044510  [4097024/15945812]\n",
      "loss: 0.046789  [5121024/15945812]\n",
      "loss: 0.046247  [6145024/15945812]\n",
      "loss: 0.046420  [7169024/15945812]\n",
      "loss: 0.047909  [8193024/15945812]\n",
      "loss: 0.042944  [9217024/15945812]\n",
      "loss: 0.040802  [10241024/15945812]\n",
      "loss: 0.046514  [11265024/15945812]\n",
      "loss: 0.048351  [12289024/15945812]\n",
      "loss: 0.050781  [13313024/15945812]\n",
      "loss: 0.043856  [14337024/15945812]\n",
      "loss: 0.047234  [15361024/15945812]\n",
      "Test MSE 1.0995341182776037\n",
      "Test RMSE 1.0485867242520304\n",
      "Test MSE 1.140818675069916\n",
      "Test RMSE 1.0680911361255256\n",
      "Test MSE 1.0995341182776024\n",
      "Test RMSE 1.0485867242520297\n",
      "Epoch 1\n",
      "------------------------\n",
      "loss: 1.024333  [ 1024/15945812]\n",
      "loss: 0.045065  [1025024/15945812]\n",
      "loss: 0.045331  [2049024/15945812]\n",
      "loss: 0.046499  [3073024/15945812]\n",
      "loss: 34.065971  [4097024/15945812]\n",
      "loss: 0.046565  [5121024/15945812]\n",
      "loss: 0.046814  [6145024/15945812]\n",
      "loss: 0.043600  [7169024/15945812]\n",
      "loss: 0.046914  [8193024/15945812]\n",
      "loss: 0.046173  [9217024/15945812]\n",
      "loss: 0.041588  [10241024/15945812]\n",
      "loss: 0.040936  [11265024/15945812]\n",
      "loss: 0.044671  [12289024/15945812]\n",
      "loss: 0.043282  [13313024/15945812]\n",
      "loss: 0.046179  [14337024/15945812]\n",
      "loss: 0.052974  [15361024/15945812]\n",
      "Test MSE 1.1641767114694914\n",
      "Test RMSE 1.0789702087960962\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.045858  [ 1024/15945812]\n",
      "loss: 0.048209  [1025024/15945812]\n",
      "loss: 0.043677  [2049024/15945812]\n",
      "loss: 0.046963  [3073024/15945812]\n",
      "loss: 0.044019  [4097024/15945812]\n",
      "loss: 0.044383  [5121024/15945812]\n",
      "loss: 0.044723  [6145024/15945812]\n",
      "loss: 0.045312  [7169024/15945812]\n",
      "loss: 0.043864  [8193024/15945812]\n",
      "loss: 0.043442  [9217024/15945812]\n",
      "loss: 0.042747  [10241024/15945812]\n",
      "loss: 0.843056  [11265024/15945812]\n",
      "loss: 0.046970  [12289024/15945812]\n",
      "loss: 0.043385  [13313024/15945812]\n",
      "loss: 0.043152  [14337024/15945812]\n",
      "loss: 0.043383  [15361024/15945812]\n",
      "Test MSE 1.0822535532237383\n",
      "Test RMSE 1.0403141608301496\n",
      "Test MSE 1.0996293829339783\n",
      "Test RMSE 1.04863214853159\n",
      "Test MSE 1.0822535532237378\n",
      "Test RMSE 1.0403141608301494\n",
      "Test MSE 0.725478744611668\n",
      "Test RMSE 0.851750400417674\n",
      "Best LR for GMF on 20m 0.001\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "gmf_lr, gmf_model, gmf_lr_history, gmf_lr_curves = tune_lr('gmf')\n",
    "title = f\"Train and Test RMSE vs. Learning Rate for GMF on {dataset}\"\n",
    "plot_learning_curve(gmf_lr_history, gmf_lr_curves, title, True)\n",
    "print(f\"Best LR for GMF on {dataset}\", gmf_lr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "c58e328d-9ba1-4fe8-ae02-febd490cb0d8",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1\n",
      "------------------------\n",
      "loss: 1.014542  [ 1024/15945812]\n",
      "loss: 0.043406  [1025024/15945812]\n",
      "loss: 0.044753  [2049024/15945812]\n",
      "loss: 0.041095  [3073024/15945812]\n",
      "loss: 0.037922  [4097024/15945812]\n",
      "loss: 0.035939  [5121024/15945812]\n",
      "loss: 0.035836  [6145024/15945812]\n",
      "loss: 0.039137  [7169024/15945812]\n",
      "loss: 0.035927  [8193024/15945812]\n",
      "loss: 0.033935  [9217024/15945812]\n",
      "loss: 0.035272  [10241024/15945812]\n",
      "loss: 0.034355  [11265024/15945812]\n",
      "loss: 0.034866  [12289024/15945812]\n",
      "loss: 0.038258  [13313024/15945812]\n",
      "loss: 0.038570  [14337024/15945812]\n",
      "loss: 0.034981  [15361024/15945812]\n",
      "Test MSE 0.9545923253419828\n",
      "Test RMSE 0.9770324075188002\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.037074  [ 1024/15945812]\n",
      "loss: 0.037191  [1025024/15945812]\n",
      "loss: 0.039258  [2049024/15945812]\n",
      "loss: 0.034808  [3073024/15945812]\n",
      "loss: 0.036466  [4097024/15945812]\n",
      "loss: 0.036784  [5121024/15945812]\n",
      "loss: 0.035780  [6145024/15945812]\n",
      "loss: 0.038118  [7169024/15945812]\n",
      "loss: 0.035050  [8193024/15945812]\n",
      "loss: 0.038735  [9217024/15945812]\n",
      "loss: 0.035181  [10241024/15945812]\n",
      "loss: 0.034731  [11265024/15945812]\n",
      "loss: 0.035973  [12289024/15945812]\n",
      "loss: 0.037528  [13313024/15945812]\n",
      "loss: 0.034760  [14337024/15945812]\n",
      "loss: 0.033699  [15361024/15945812]\n",
      "Test MSE 0.9461810562898298\n",
      "Test RMSE 0.972718384883225\n",
      "Test MSE 0.8882487399223679\n",
      "Test RMSE 0.9424694901811771\n",
      "Test MSE 0.9461810562898285\n",
      "Test RMSE 0.9727183848832243\n",
      "Epoch 1\n",
      "------------------------\n",
      "loss: 0.589815  [ 1024/15945812]\n",
      "loss: 0.041139  [1025024/15945812]\n",
      "loss: 0.047749  [2049024/15945812]\n",
      "loss: 0.044005  [3073024/15945812]\n",
      "loss: 0.045670  [4097024/15945812]\n",
      "loss: 0.047092  [5121024/15945812]\n",
      "loss: 0.046693  [6145024/15945812]\n",
      "loss: 0.044792  [7169024/15945812]\n",
      "loss: 0.045671  [8193024/15945812]\n",
      "loss: 0.045590  [9217024/15945812]\n",
      "loss: 0.042790  [10241024/15945812]\n",
      "loss: 0.043536  [11265024/15945812]\n",
      "loss: 0.043649  [12289024/15945812]\n",
      "loss: 0.041918  [13313024/15945812]\n",
      "loss: 0.046763  [14337024/15945812]\n",
      "loss: 0.043877  [15361024/15945812]\n",
      "Test MSE 1.0954324802530757\n",
      "Test RMSE 1.046629103480825\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.044177  [ 1024/15945812]\n",
      "loss: 0.047908  [1025024/15945812]\n",
      "loss: 0.042754  [2049024/15945812]\n",
      "loss: 0.046072  [3073024/15945812]\n",
      "loss: 0.046179  [4097024/15945812]\n",
      "loss: 0.047842  [5121024/15945812]\n",
      "loss: 0.045358  [6145024/15945812]\n",
      "loss: 0.044106  [7169024/15945812]\n",
      "loss: 0.046194  [8193024/15945812]\n",
      "loss: 0.043727  [9217024/15945812]\n",
      "loss: 0.045251  [10241024/15945812]\n",
      "loss: 0.049528  [11265024/15945812]\n",
      "loss: 0.043755  [12289024/15945812]\n",
      "loss: 0.043252  [13313024/15945812]\n",
      "loss: 0.043894  [14337024/15945812]\n",
      "loss: 0.047906  [15361024/15945812]\n",
      "Test MSE 1.086824834334673\n",
      "Test RMSE 1.0425089133118588\n",
      "Test MSE 1.1213926623500359\n",
      "Test RMSE 1.0589582911286146\n",
      "Test MSE 1.0868248343346723\n",
      "Test RMSE 1.0425089133118586\n",
      "Epoch 1\n",
      "------------------------\n",
      "loss: 0.696522  [ 1024/15945812]\n",
      "loss: 0.043812  [1025024/15945812]\n",
      "loss: 0.047602  [2049024/15945812]\n",
      "loss: 0.035985  [3073024/15945812]\n",
      "loss: 0.039083  [4097024/15945812]\n",
      "loss: 0.038413  [5121024/15945812]\n",
      "loss: 0.038325  [6145024/15945812]\n",
      "loss: 0.038546  [7169024/15945812]\n",
      "loss: 0.038636  [8193024/15945812]\n",
      "loss: 0.038343  [9217024/15945812]\n",
      "loss: 0.034690  [10241024/15945812]\n",
      "loss: 0.034137  [11265024/15945812]\n",
      "loss: 0.038042  [12289024/15945812]\n",
      "loss: 0.033453  [13313024/15945812]\n",
      "loss: 0.037787  [14337024/15945812]\n",
      "loss: 0.039694  [15361024/15945812]\n",
      "Test MSE 0.9771758338209126\n",
      "Test RMSE 0.988522045187113\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.035228  [ 1024/15945812]\n",
      "loss: 0.039748  [1025024/15945812]\n",
      "loss: 0.041330  [2049024/15945812]\n",
      "loss: 0.037799  [3073024/15945812]\n",
      "loss: 0.038643  [4097024/15945812]\n",
      "loss: 0.038317  [5121024/15945812]\n",
      "loss: 0.040005  [6145024/15945812]\n",
      "loss: 0.036174  [7169024/15945812]\n",
      "loss: 0.037951  [8193024/15945812]\n",
      "loss: 0.035730  [9217024/15945812]\n",
      "loss: 0.035661  [10241024/15945812]\n",
      "loss: 0.039569  [11265024/15945812]\n",
      "loss: 0.037397  [12289024/15945812]\n",
      "loss: 0.038027  [13313024/15945812]\n",
      "loss: 0.035911  [14337024/15945812]\n",
      "loss: 0.035773  [15361024/15945812]\n",
      "Test MSE 0.9996047315366395\n",
      "Test RMSE 0.9998023462348143\n",
      "Test MSE 0.9592027031639422\n",
      "Test RMSE 0.9793889437623554\n",
      "Test MSE 0.9996047315366389\n",
      "Test RMSE 0.999802346234814\n",
      "Epoch 1\n",
      "------------------------\n",
      "loss: 0.660290  [ 1024/15945812]\n",
      "loss: 0.046838  [1025024/15945812]\n",
      "loss: 0.045892  [2049024/15945812]\n",
      "loss: 0.048335  [3073024/15945812]\n",
      "loss: 0.046281  [4097024/15945812]\n",
      "loss: 0.042943  [5121024/15945812]\n",
      "loss: 0.047131  [6145024/15945812]\n",
      "loss: 0.038614  [7169024/15945812]\n",
      "loss: 0.045964  [8193024/15945812]\n",
      "loss: 0.044373  [9217024/15945812]\n",
      "loss: 0.044997  [10241024/15945812]\n",
      "loss: 0.049581  [11265024/15945812]\n",
      "loss: 0.044913  [12289024/15945812]\n",
      "loss: 0.046786  [13313024/15945812]\n",
      "loss: 0.046924  [14337024/15945812]\n",
      "loss: 0.044798  [15361024/15945812]\n",
      "Test MSE 1.1052752278011686\n",
      "Test RMSE 1.0513207064455492\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.044726  [ 1024/15945812]\n",
      "loss: 0.043676  [1025024/15945812]\n",
      "loss: 0.042191  [2049024/15945812]\n",
      "loss: 0.046335  [3073024/15945812]\n",
      "loss: 0.049085  [4097024/15945812]\n",
      "loss: 0.048185  [5121024/15945812]\n",
      "loss: 0.044115  [6145024/15945812]\n",
      "loss: 0.043353  [7169024/15945812]\n",
      "loss: 0.046486  [8193024/15945812]\n",
      "loss: 0.041572  [9217024/15945812]\n",
      "loss: 0.043697  [10241024/15945812]\n",
      "loss: 0.040463  [11265024/15945812]\n",
      "loss: 0.047886  [12289024/15945812]\n",
      "loss: 0.043676  [13313024/15945812]\n",
      "loss: 0.047385  [14337024/15945812]\n",
      "loss: 0.045200  [15361024/15945812]\n",
      "Test MSE 1.10642224981879\n",
      "Test RMSE 1.0518660797928556\n",
      "Test MSE 1.1443345905859335\n",
      "Test RMSE 1.0697357573653101\n",
      "Test MSE 1.1064222498187903\n",
      "Test RMSE 1.0518660797928556\n",
      "Test MSE 0.9461810562898286\n",
      "Test RMSE 0.9727183848832244\n",
      "Best Embedding Dimension for GMF on 20m 4\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "gmf_dim, gmf_model, gmf_dim_history, gmf_dim_curves = tune_dims('gmf')\n",
    "title = f\"Train and Test RMSE vs. Embedding Dim for GMF on {dataset}\"\n",
    "plot_learning_curve(gmf_dim_history, gmf_dim_curves, title, True)\n",
    "print(f\"Best Embedding Dimension for GMF on {dataset}\", gmf_dim)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "0bebb721",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1\n",
      "-------------------------------\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.225525  [ 1024/15945812]\n",
      "loss: 0.038606  [1025024/15945812]\n",
      "loss: 0.033102  [2049024/15945812]\n",
      "loss: 0.035791  [3073024/15945812]\n",
      "loss: 0.034167  [4097024/15945812]\n",
      "loss: 0.030997  [5121024/15945812]\n",
      "loss: 0.032170  [6145024/15945812]\n",
      "loss: 0.026501  [7169024/15945812]\n",
      "loss: 0.030211  [8193024/15945812]\n",
      "loss: 0.027894  [9217024/15945812]\n",
      "loss: 0.029796  [10241024/15945812]\n",
      "loss: 0.031103  [11265024/15945812]\n",
      "loss: 0.029677  [12289024/15945812]\n",
      "loss: 0.029730  [13313024/15945812]\n",
      "loss: 0.031286  [14337024/15945812]\n",
      "loss: 0.029343  [15361024/15945812]\n",
      "Test MSE 0.7981708636553706\n",
      "Test RMSE 0.8934040875524192\n",
      "Epoch 2\n",
      "-------------------------------\n",
      "loss: 0.030569  [ 1024/15945812]\n",
      "loss: 0.028693  [1025024/15945812]\n",
      "loss: 0.028079  [2049024/15945812]\n",
      "loss: 0.029555  [3073024/15945812]\n",
      "loss: 0.030745  [4097024/15945812]\n",
      "loss: 0.031634  [5121024/15945812]\n",
      "loss: 0.029537  [6145024/15945812]\n",
      "loss: 0.030461  [7169024/15945812]\n",
      "loss: 0.029776  [8193024/15945812]\n",
      "loss: 0.027817  [9217024/15945812]\n",
      "loss: 0.027935  [10241024/15945812]\n",
      "loss: 0.029249  [11265024/15945812]\n",
      "loss: 0.030462  [12289024/15945812]\n",
      "loss: 0.029034  [13313024/15945812]\n",
      "loss: 0.029806  [14337024/15945812]\n",
      "loss: 0.032440  [15361024/15945812]\n",
      "Test MSE 0.7689685786286002\n",
      "Test RMSE 0.8769085349274462\n",
      "Test MSE 0.7008428459319935\n",
      "Test RMSE 0.8371635717898823\n",
      "Test MSE 0.7689685786286008\n",
      "Test RMSE 0.8769085349274466\n",
      "Epoch 1\n",
      "-------------------------------\n",
      "loss: 0.264007  [ 1024/15945812]\n",
      "loss: 0.028456  [1025024/15945812]\n",
      "loss: 0.032617  [2049024/15945812]\n",
      "loss: 0.029240  [3073024/15945812]\n",
      "loss: 0.030066  [4097024/15945812]\n",
      "loss: 0.031704  [5121024/15945812]\n",
      "loss: 0.029736  [6145024/15945812]\n",
      "loss: 0.029424  [7169024/15945812]\n",
      "loss: 0.030793  [8193024/15945812]\n",
      "loss: 0.030502  [9217024/15945812]\n",
      "loss: 0.029919  [10241024/15945812]\n",
      "loss: 0.029813  [11265024/15945812]\n",
      "loss: 0.030650  [12289024/15945812]\n",
      "loss: 0.029286  [13313024/15945812]\n",
      "loss: 0.027722  [14337024/15945812]\n",
      "loss: 0.030588  [15361024/15945812]\n",
      "Test MSE 0.7830942391248716\n",
      "Test RMSE 0.8849261207156627\n",
      "Epoch 2\n",
      "-------------------------------\n",
      "loss: 0.029576  [ 1024/15945812]\n",
      "loss: 0.029666  [1025024/15945812]\n",
      "loss: 0.030862  [2049024/15945812]\n",
      "loss: 0.027259  [3073024/15945812]\n",
      "loss: 0.030898  [4097024/15945812]\n",
      "loss: 0.029693  [5121024/15945812]\n",
      "loss: 0.029872  [6145024/15945812]\n",
      "loss: 0.027510  [7169024/15945812]\n",
      "loss: 0.027751  [8193024/15945812]\n",
      "loss: 0.026586  [9217024/15945812]\n",
      "loss: 0.028743  [10241024/15945812]\n",
      "loss: 0.028397  [11265024/15945812]\n",
      "loss: 0.026636  [12289024/15945812]\n",
      "loss: 0.025722  [13313024/15945812]\n",
      "loss: 0.031213  [14337024/15945812]\n",
      "loss: 0.028968  [15361024/15945812]\n",
      "Test MSE 0.7780669365194816\n",
      "Test RMSE 0.8820810260511681\n",
      "Test MSE 0.7118382126597376\n",
      "Test RMSE 0.8437050507492162\n",
      "Test MSE 0.7780669365194817\n",
      "Test RMSE 0.8820810260511682\n",
      "Epoch 1\n",
      "-------------------------------\n",
      "loss: 0.786755  [ 1024/15945812]\n",
      "loss: 0.036483  [1025024/15945812]\n",
      "loss: 0.032404  [2049024/15945812]\n",
      "loss: 0.030797  [3073024/15945812]\n",
      "loss: 0.034005  [4097024/15945812]\n",
      "loss: 0.034351  [5121024/15945812]\n",
      "loss: 0.030534  [6145024/15945812]\n",
      "loss: 0.032041  [7169024/15945812]\n",
      "loss: 0.035921  [8193024/15945812]\n",
      "loss: 0.031892  [9217024/15945812]\n",
      "loss: 0.031657  [10241024/15945812]\n",
      "loss: 0.034681  [11265024/15945812]\n",
      "loss: 0.033664  [12289024/15945812]\n",
      "loss: 0.035305  [13313024/15945812]\n",
      "loss: 0.030972  [14337024/15945812]\n",
      "loss: 0.032147  [15361024/15945812]\n",
      "Test MSE 0.9721243490464528\n",
      "Test RMSE 0.9859636651755748\n",
      "Epoch 2\n",
      "-------------------------------\n",
      "loss: 0.033399  [ 1024/15945812]\n",
      "loss: 0.033839  [1025024/15945812]\n",
      "loss: 0.032742  [2049024/15945812]\n",
      "loss: 0.034208  [3073024/15945812]\n",
      "loss: 0.038321  [4097024/15945812]\n",
      "loss: 0.043805  [5121024/15945812]\n",
      "loss: 0.045806  [6145024/15945812]\n",
      "loss: 0.043441  [7169024/15945812]\n",
      "loss: 0.044489  [8193024/15945812]\n",
      "loss: 0.044079  [9217024/15945812]\n",
      "loss: 0.044027  [10241024/15945812]\n",
      "loss: 0.042299  [11265024/15945812]\n",
      "loss: 0.046650  [12289024/15945812]\n",
      "loss: 0.042054  [13313024/15945812]\n",
      "loss: 0.043341  [14337024/15945812]\n",
      "loss: 0.045406  [15361024/15945812]\n",
      "Test MSE 1.0915994817302401\n",
      "Test RMSE 1.0447963829044586\n",
      "Test MSE 1.1277348606741668\n",
      "Test RMSE 1.0619486148934736\n",
      "Test MSE 1.0915994817302401\n",
      "Test RMSE 1.0447963829044586\n",
      "Epoch 1\n",
      "-------------------------------\n",
      "loss: 1.004216  [ 1024/15945812]\n",
      "loss: 0.036515  [1025024/15945812]\n",
      "loss: 0.031266  [2049024/15945812]\n",
      "loss: 0.036015  [3073024/15945812]\n",
      "loss: 0.035148  [4097024/15945812]\n",
      "loss: 0.031895  [5121024/15945812]\n",
      "loss: 0.032677  [6145024/15945812]\n",
      "loss: 0.036134  [7169024/15945812]\n",
      "loss: 0.032996  [8193024/15945812]\n",
      "loss: 0.048047  [9217024/15945812]\n",
      "loss: 0.045763  [10241024/15945812]\n",
      "loss: 0.041941  [11265024/15945812]\n",
      "loss: 0.047227  [12289024/15945812]\n",
      "loss: 0.045224  [13313024/15945812]\n",
      "loss: 0.046055  [14337024/15945812]\n",
      "loss: 0.042563  [15361024/15945812]\n",
      "Test MSE 1.0810321052385758\n",
      "Test RMSE 1.0397269378248193\n",
      "Epoch 2\n",
      "-------------------------------\n",
      "loss: 0.042623  [ 1024/15945812]\n",
      "loss: 0.044811  [1025024/15945812]\n",
      "loss: 0.044377  [2049024/15945812]\n",
      "loss: 0.043721  [3073024/15945812]\n",
      "loss: 0.043086  [4097024/15945812]\n",
      "loss: 0.042538  [5121024/15945812]\n",
      "loss: 0.046497  [6145024/15945812]\n",
      "loss: 0.048187  [7169024/15945812]\n",
      "loss: 0.042305  [8193024/15945812]\n",
      "loss: 0.044528  [9217024/15945812]\n",
      "loss: 0.044493  [10241024/15945812]\n",
      "loss: 0.046569  [11265024/15945812]\n",
      "loss: 0.043833  [12289024/15945812]\n",
      "loss: 0.043073  [13313024/15945812]\n",
      "loss: 0.046154  [14337024/15945812]\n",
      "loss: 0.041998  [15361024/15945812]\n",
      "Test MSE 1.0867004489546221\n",
      "Test RMSE 1.0424492548582986\n",
      "Test MSE 1.122129476417165\n",
      "Test RMSE 1.059306129698665\n",
      "Test MSE 1.0867004489546221\n",
      "Test RMSE 1.0424492548582986\n",
      "Epoch 1\n",
      "-------------------------------\n",
      "loss: 0.627603  [ 1024/15945812]\n",
      "loss: 0.041344  [1025024/15945812]\n",
      "loss: 0.044966  [2049024/15945812]\n",
      "loss: 0.043358  [3073024/15945812]\n",
      "loss: 0.043199  [4097024/15945812]\n",
      "loss: 0.044163  [5121024/15945812]\n",
      "loss: 0.048451  [6145024/15945812]\n",
      "loss: 0.042610  [7169024/15945812]\n",
      "loss: 0.045257  [8193024/15945812]\n",
      "loss: 0.042490  [9217024/15945812]\n",
      "loss: 0.047148  [10241024/15945812]\n",
      "loss: 0.048609  [11265024/15945812]\n",
      "loss: 0.046932  [12289024/15945812]\n",
      "loss: 0.046396  [13313024/15945812]\n",
      "loss: 0.045447  [14337024/15945812]\n",
      "loss: 0.040946  [15361024/15945812]\n",
      "Test MSE 1.0835461930166346\n",
      "Test RMSE 1.0409352491949893\n",
      "Epoch 2\n",
      "-------------------------------\n",
      "loss: 0.041288  [ 1024/15945812]\n",
      "loss: 0.045062  [1025024/15945812]\n",
      "loss: 0.045688  [2049024/15945812]\n",
      "loss: 0.049265  [3073024/15945812]\n",
      "loss: 0.041483  [4097024/15945812]\n",
      "loss: 0.043662  [5121024/15945812]\n",
      "loss: 0.043503  [6145024/15945812]\n",
      "loss: 0.045316  [7169024/15945812]\n",
      "loss: 0.046128  [8193024/15945812]\n",
      "loss: 0.042231  [9217024/15945812]\n",
      "loss: 0.044331  [10241024/15945812]\n",
      "loss: 0.045753  [11265024/15945812]\n",
      "loss: 0.044714  [12289024/15945812]\n",
      "loss: 0.043757  [13313024/15945812]\n",
      "loss: 0.045491  [14337024/15945812]\n",
      "loss: 0.043639  [15361024/15945812]\n",
      "Test MSE 1.0914978926809307\n",
      "Test RMSE 1.0447477650997539\n",
      "Test MSE 1.1219174520520694\n",
      "Test RMSE 1.0592060479680379\n",
      "Test MSE 1.0914978926809307\n",
      "Test RMSE 1.0447477650997539\n",
      "Epoch 1\n",
      "-------------------------------\n",
      "loss: 0.152716  [ 1024/15945812]\n",
      "loss: 0.047438  [1025024/15945812]\n",
      "loss: 0.043324  [2049024/15945812]\n",
      "loss: 0.042948  [3073024/15945812]\n",
      "loss: 0.044399  [4097024/15945812]\n",
      "loss: 0.048060  [5121024/15945812]\n",
      "loss: 0.045052  [6145024/15945812]\n",
      "loss: 0.046294  [7169024/15945812]\n",
      "loss: 0.045975  [8193024/15945812]\n",
      "loss: 0.045904  [9217024/15945812]\n",
      "loss: 0.047065  [10241024/15945812]\n",
      "loss: 0.042924  [11265024/15945812]\n",
      "loss: 0.047510  [12289024/15945812]\n",
      "loss: 0.043058  [13313024/15945812]\n",
      "loss: 0.043864  [14337024/15945812]\n",
      "loss: 0.045309  [15361024/15945812]\n",
      "Test MSE 1.1040483209734655\n",
      "Test RMSE 1.0507370370237576\n",
      "Epoch 2\n",
      "-------------------------------\n",
      "loss: 0.046060  [ 1024/15945812]\n",
      "loss: 0.040876  [1025024/15945812]\n",
      "loss: 0.048496  [2049024/15945812]\n",
      "loss: 0.040639  [3073024/15945812]\n",
      "loss: 0.045140  [4097024/15945812]\n",
      "loss: 0.047548  [5121024/15945812]\n",
      "loss: 0.043327  [6145024/15945812]\n",
      "loss: 0.047192  [7169024/15945812]\n",
      "loss: 0.043648  [8193024/15945812]\n",
      "loss: 0.043046  [9217024/15945812]\n",
      "loss: 0.046079  [10241024/15945812]\n",
      "loss: 0.043630  [11265024/15945812]\n",
      "loss: 0.043298  [12289024/15945812]\n",
      "loss: 0.042327  [13313024/15945812]\n",
      "loss: 0.048187  [14337024/15945812]\n",
      "loss: 0.040935  [15361024/15945812]\n",
      "Test MSE 1.0990022852490535\n",
      "Test RMSE 1.0483330984229458\n",
      "Test MSE 1.1285902413994686\n",
      "Test RMSE 1.0623512796619905\n",
      "Test MSE 1.0990022852490535\n",
      "Test RMSE 1.0483330984229458\n",
      "Test MSE 0.7689685786286001\n",
      "Test RMSE 0.8769085349274462\n",
      "Best Learning Rate for MLP on 20m 0.001\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "mlp_lr, mlp_model, mlp_lr_history, mlp_lr_curves = tune_lr('mlp')\n",
    "title = f\"Train and Test RMSE vs. Learning Rate for MLP on {dataset}\"\n",
    "plot_learning_curve(mlp_lr_history, mlp_lr_curves, title, True)\n",
    "print(f\"Best Learning Rate for MLP on {dataset}\", mlp_lr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "7154a65d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1\n",
      "-------------------------------\n",
      "loss: 1.116874  [ 1024/15945812]\n",
      "loss: 0.030330  [1025024/15945812]\n",
      "loss: 0.033413  [2049024/15945812]\n",
      "loss: 0.033834  [3073024/15945812]\n",
      "loss: 0.031662  [4097024/15945812]\n",
      "loss: 0.027759  [5121024/15945812]\n",
      "loss: 0.033184  [6145024/15945812]\n",
      "loss: 0.032450  [7169024/15945812]\n",
      "loss: 0.032362  [8193024/15945812]\n",
      "loss: 0.033127  [9217024/15945812]\n",
      "loss: 0.031617  [10241024/15945812]\n",
      "loss: 0.028507  [11265024/15945812]\n",
      "loss: 0.031532  [12289024/15945812]\n",
      "loss: 0.035373  [13313024/15945812]\n",
      "loss: 0.033189  [14337024/15945812]\n",
      "loss: 0.035661  [15361024/15945812]\n",
      "Test MSE 0.874727843387357\n",
      "Test RMSE 0.9352688615512423\n",
      "Epoch 2\n",
      "-------------------------------\n",
      "loss: 0.033736  [ 1024/15945812]\n",
      "loss: 0.030650  [1025024/15945812]\n",
      "loss: 0.031578  [2049024/15945812]\n",
      "loss: 0.030214  [3073024/15945812]\n",
      "loss: 0.033802  [4097024/15945812]\n",
      "loss: 0.034733  [5121024/15945812]\n",
      "loss: 0.032736  [6145024/15945812]\n",
      "loss: 0.034739  [7169024/15945812]\n",
      "loss: 0.032931  [8193024/15945812]\n",
      "loss: 0.030799  [9217024/15945812]\n",
      "loss: 0.032368  [10241024/15945812]\n",
      "loss: 0.031412  [11265024/15945812]\n",
      "loss: 0.030414  [12289024/15945812]\n",
      "loss: 0.031283  [13313024/15945812]\n",
      "loss: 0.031691  [14337024/15945812]\n",
      "loss: 0.031756  [15361024/15945812]\n",
      "Test MSE 0.8755877756182953\n",
      "Test RMSE 0.9357284732326442\n",
      "Test MSE 0.8133578577365944\n",
      "Test RMSE 0.9018635471824962\n",
      "Test MSE 0.8755877756182944\n",
      "Test RMSE 0.9357284732326436\n",
      "Epoch 1\n",
      "-------------------------------\n",
      "loss: 0.227223  [ 1024/15945812]\n",
      "loss: 0.033222  [1025024/15945812]\n",
      "loss: 0.033017  [2049024/15945812]\n",
      "loss: 0.036950  [3073024/15945812]\n",
      "loss: 0.034623  [4097024/15945812]\n",
      "loss: 0.032878  [5121024/15945812]\n",
      "loss: 0.032183  [6145024/15945812]\n",
      "loss: 0.032107  [7169024/15945812]\n",
      "loss: 0.031829  [8193024/15945812]\n",
      "loss: 0.030447  [9217024/15945812]\n",
      "loss: 0.030520  [10241024/15945812]\n",
      "loss: 0.033805  [11265024/15945812]\n",
      "loss: 0.035411  [12289024/15945812]\n",
      "loss: 0.032333  [13313024/15945812]\n",
      "loss: 0.035981  [14337024/15945812]\n",
      "loss: 0.032355  [15361024/15945812]\n",
      "Test MSE 0.8560684534557936\n",
      "Test RMSE 0.9252396735202147\n",
      "Epoch 2\n",
      "-------------------------------\n",
      "loss: 0.034360  [ 1024/15945812]\n",
      "loss: 0.032527  [1025024/15945812]\n",
      "loss: 0.029927  [2049024/15945812]\n",
      "loss: 0.032326  [3073024/15945812]\n",
      "loss: 0.031798  [4097024/15945812]\n",
      "loss: 0.034315  [5121024/15945812]\n",
      "loss: 0.034203  [6145024/15945812]\n",
      "loss: 0.033749  [7169024/15945812]\n",
      "loss: 0.033231  [8193024/15945812]\n",
      "loss: 0.029428  [9217024/15945812]\n",
      "loss: 0.035758  [10241024/15945812]\n",
      "loss: 0.032297  [11265024/15945812]\n",
      "loss: 0.031099  [12289024/15945812]\n",
      "loss: 0.031320  [13313024/15945812]\n",
      "loss: 0.032265  [14337024/15945812]\n",
      "loss: 0.032867  [15361024/15945812]\n",
      "Test MSE 0.883846617587213\n",
      "Test RMSE 0.9401311704157101\n",
      "Test MSE 0.8188353303834263\n",
      "Test RMSE 0.9048952040890847\n",
      "Test MSE 0.8838466175872124\n",
      "Test RMSE 0.9401311704157098\n",
      "Epoch 1\n",
      "-------------------------------\n",
      "loss: 0.664243  [ 1024/15945812]\n",
      "loss: 0.036052  [1025024/15945812]\n",
      "loss: 0.033712  [2049024/15945812]\n",
      "loss: 0.034229  [3073024/15945812]\n",
      "loss: 0.032689  [4097024/15945812]\n",
      "loss: 0.033012  [5121024/15945812]\n",
      "loss: 0.032520  [6145024/15945812]\n",
      "loss: 0.030392  [7169024/15945812]\n",
      "loss: 0.036043  [8193024/15945812]\n",
      "loss: 0.033315  [9217024/15945812]\n",
      "loss: 0.035389  [10241024/15945812]\n",
      "loss: 0.032488  [11265024/15945812]\n",
      "loss: 0.034544  [12289024/15945812]\n",
      "loss: 0.030209  [13313024/15945812]\n",
      "loss: 0.032513  [14337024/15945812]\n",
      "loss: 0.032388  [15361024/15945812]\n",
      "Test MSE 0.8702894810638862\n",
      "Test RMSE 0.9328930705412524\n",
      "Epoch 2\n",
      "-------------------------------\n",
      "loss: 0.031535  [ 1024/15945812]\n",
      "loss: 0.032476  [1025024/15945812]\n",
      "loss: 0.034028  [2049024/15945812]\n",
      "loss: 0.031301  [3073024/15945812]\n",
      "loss: 0.033796  [4097024/15945812]\n",
      "loss: 0.030614  [5121024/15945812]\n",
      "loss: 0.031333  [6145024/15945812]\n",
      "loss: 0.031703  [7169024/15945812]\n",
      "loss: 0.033744  [8193024/15945812]\n",
      "loss: 0.032588  [9217024/15945812]\n",
      "loss: 0.036537  [10241024/15945812]\n",
      "loss: 0.033271  [11265024/15945812]\n",
      "loss: 0.034552  [12289024/15945812]\n",
      "loss: 0.033291  [13313024/15945812]\n",
      "loss: 0.034406  [14337024/15945812]\n",
      "loss: 0.033335  [15361024/15945812]\n",
      "Test MSE 0.923314793400363\n",
      "Test RMSE 0.9608927064976417\n",
      "Test MSE 0.8897118666573609\n",
      "Test RMSE 0.9432453904776641\n",
      "Test MSE 0.9233147934003629\n",
      "Test RMSE 0.9608927064976417\n",
      "Epoch 1\n",
      "-------------------------------\n",
      "loss: 0.215092  [ 1024/15945812]\n",
      "loss: 0.037314  [1025024/15945812]\n",
      "loss: 0.031606  [2049024/15945812]\n",
      "loss: 0.034724  [3073024/15945812]\n",
      "loss: 0.030935  [4097024/15945812]\n",
      "loss: 0.035546  [5121024/15945812]\n",
      "loss: 0.029930  [6145024/15945812]\n",
      "loss: 0.032659  [7169024/15945812]\n",
      "loss: 0.034350  [8193024/15945812]\n",
      "loss: 0.032697  [9217024/15945812]\n",
      "loss: 0.030159  [10241024/15945812]\n",
      "loss: 0.034538  [11265024/15945812]\n",
      "loss: 0.033615  [12289024/15945812]\n",
      "loss: 0.032549  [13313024/15945812]\n",
      "loss: 0.045394  [14337024/15945812]\n",
      "loss: 0.033994  [15361024/15945812]\n",
      "Test MSE 0.8775322162660647\n",
      "Test RMSE 0.936766895372624\n",
      "Epoch 2\n",
      "-------------------------------\n",
      "loss: 0.034715  [ 1024/15945812]\n",
      "loss: 0.032361  [1025024/15945812]\n",
      "loss: 0.031312  [2049024/15945812]\n",
      "loss: 0.031993  [3073024/15945812]\n",
      "loss: 0.032743  [4097024/15945812]\n",
      "loss: 0.034746  [5121024/15945812]\n",
      "loss: 0.036902  [6145024/15945812]\n",
      "loss: 0.032262  [7169024/15945812]\n",
      "loss: 0.034053  [8193024/15945812]\n",
      "loss: 0.034145  [9217024/15945812]\n",
      "loss: 0.030388  [10241024/15945812]\n",
      "loss: 0.032995  [11265024/15945812]\n",
      "loss: 0.031192  [12289024/15945812]\n",
      "loss: 0.028654  [13313024/15945812]\n",
      "loss: 0.036673  [14337024/15945812]\n",
      "loss: 0.033381  [15361024/15945812]\n",
      "Test MSE 0.8606497915717374\n",
      "Test RMSE 0.9277121275329634\n",
      "Test MSE 0.809693996300587\n",
      "Test RMSE 0.8998299818857932\n",
      "Test MSE 0.8606497915717364\n",
      "Test RMSE 0.9277121275329628\n",
      "Test MSE 0.8606497915717356\n",
      "Test RMSE 0.9277121275329625\n",
      "Best Embedding Dimension for MLP on 20m 32\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAvEAAAI6CAYAAACuFxeWAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8fJSN1AAAACXBIWXMAAA9hAAAPYQGoP6dpAACotUlEQVR4nOzddXiV9f/H8edZb7AARtdoRol0h5Q0It0dKooYiCClgor5VWlGCAIKiKSUIN2gdDejBmzAWJ7798f9Yzo3kI3BvXg9rutcet/nPud+nXHifT7nEzbDMAxERERERCTFcLA6gIiIiIiIJIyKeBERERGRFEZFvIiIiIhICqMiXkREREQkhXGyOoCIiIiI/De73U5ERITVMeQZcXZ2xtHR8aHXq4gXERERSeYiIiI4c+YMdrvd6ijyDPn4+JAtWzZsNluc61TEi4iIiCRjhmEQGBiIo6MjuXPnxsFBvaFTO8MwCA0N5dq1awBkz549zjEq4kVERESSsaioKEJDQ8mRIwceHh5Wx5FnxN3dHYBr166RJUuWOF1r9FVOREREJBmLjo4GwMXFxeIk8qw9+NIWGRkZ5zoV8SIiIiIpQHz9oiV1e9S/uYp4EREREZEURkW8iIiIiEgKoyJeRERERFKEWrVqMXDgQKtjJAsq4kVEREQkSdlstkdeunXrlqj7XbRoER9++OETZevWrVtMDicnJ/LkyUP//v25detWrOP8/Pyw2WzMmzcvzn0UL14cm83GjBkzYvbt27ePJk2akCVLFtzc3PDz86Nt27bcuHEDgLNnzz7077F9+/YEPw4V8anIf71gHlw2bNjwROcZOXJkih1cM2PGDGw2G2fPno33+ke9wP59edh9JMTly5cZOXIk+/fvf6zjN2zYECuDo6MjmTNnpmnTpuzevTvO8Q/eqDw9Pbl7926c68+dO4eDgwM2m42RI0fGuu7IkSN07tyZ/Pnz4+bmhq+vL2XKlOG1114jJCQkzjkedkmOatWq9dC8fn5+SXaeB8+nzz//PMnu82EePDce5/Vdq1YtatWqFbP9IOc/P4yelX+/5pydncmUKRPly5fnzTff5NChQ3Fuk5DHmhT5GjduTMaMGbHZbE+9BfBB0fDPf59/mjVrVrzv5Q/elx8UC/GJ7/0ja9astG7dmiNHjiTxI0mcwMBAhg0bRuXKlfH19cXLy4uyZcsyefLkmBla/unu3bsMHDiQHDly4ObmRunSpeMtuOTZCwwMjLl8/fXXeHl5xdr3zTffxDo+vtlX4pMxY0Y8PT2fON+LL75IYGAgZ8+eZerUqSxdupRXXnklznG5c+dm+vTpsfZt376dK1eukC5duph9165do27duvj6+rJq1SqOHDlCQEAA2bNnJzQ0NNbt165dG+tvERgYSNmyZRP8GDRPfCqybdu2WNsffvgh69ev5/fff4+1v1ixYk90nl69evHiiy8+0X0kV9mzZ4/zd3zllVcIDg5mzpw5cY59UpcvX2bUqFH4+flRunTpx77dmDFjqF27NpGRkezbt49Ro0ZRs2ZN9u/fT6FChWId6+zsTFRUFPPnz6dnz56xrps+fTqenp6xinIwWxOqVq2Kv78/w4cPx8/Pjxs3bvDnn38yb9483n77bby8vGKOd3d3j/M8S+7y588f598UwNXV1YI01nrwvC9QoIBlGQYMGECHDh2w2+3cvn2bffv2ERAQwLfffsvYsWN55513Yo4tU6YM27Zte+L3ssfx5ptvsmPHDgICAsiWLVuSvO7/i6enJxs3buTUqVNx/k0CAgLw8vKK85pNiAfvHxEREezevZvRo0ezbt06Dhw4QM6cOZ80/hPZs2cPs2bNokuXLnzwwQc4OzuzcuVK+vfvz/bt2wkICIh1fMuWLdm1axeffPIJhQsX5scff6R9+/bY7XY6dOhg0aN4+gzD4H5k3C81z4K7s+NjNdBky5Yt5v+9vb2x2Wwx+86ePUv27NmZP38+48ePZ/v27UyYMIFmzZrx2muvsWnTJm7evEmBAgV4//33ad++fcx91apVi9KlS/P1118D5hffPn36cPLkSX7++WcyZMjAsGHD6NOnzyPzubq6xuTJlSsXbdu2jbcho2PHjnz11VdcuHCB3LlzA+brsGPHjsyaNSvmuK1btxISEsLUqVNxcjLL63z58vHCCy/Euc9MmTLF+vsklor4VKRSpUqxtjNnzoyDg0Oc/f8WGhqaoMUjcuXKRa5cuRKVMblzdXWN8/fy8vIiIiLiP/+Oz1KhQoVi8lSvXh0fHx+6du3K7NmzGTVqVKxjXVxcaNq0KQEBAbGKeMMwmDFjBm3btmXKlCmxbvP111/j4ODAhg0bYrV4tGrVig8//BDDMGId/zjPs+TG3d09xWV+WuJ73j9refLkiZWhUaNGDBo0iJYtW/Luu+9SokQJGjZsCJivyWeV9+DBg1SoUIEWLVokyf1FR0cTFRX1yC+L1apV48CBAwQEBPDxxx/H7D916hQbN26kV69ecV6zCfHP948aNWrg4+NDz549mTFjBkOHDk30/SaFqlWrcurUKZydnWP21atXj4iICL7//ntGjRoVU0itWLGCNWvWxBTuALVr1+bcuXO88847tG3bNs7iOKnF/choig1fZcm5D49ugIdL0pSPgwcP5osvvmD69Om4uroSFhZG2bJlGTx4MF5eXixfvjzmF+GKFSs+9H6++OILPvzwQ95//30WLFhA//79qVGjBkWLFn2sHKdPn+a3336L9bx7IGvWrDRo0ICZM2cybNgwQkNDmT9/Pn/88UesIj5btmxERUXxyy+/0KpVq2fyS7S606QxtWrVokSJEmzcuJEqVarg4eFBjx49AJg/fz7169cne/bsuLu74+/vz3vvvce9e/di3Ud83Wn8/Pxo0qQJv/32G2XKlMHd3Z2iRYvGaTV5mFGjRlGxYkUyZsyIl5cXZcqUYdq0aXGKxYScZ/v27VStWhU3Nzdy5MjBkCFDHvvnuv8SEhLC22+/Tb58+XBxcSFnzpwMHDgwzt/q559/pmLFinh7e+Ph4UH+/Plj/t4bNmygfPnyAHTv3j3mJ+5/d2t5HOXKlQPg6tWr8V7fo0cPtm7dyrFjx2L2rV27lnPnztG9e/c4xwcFBeHl5UX69Onjvb+kenN6/vnnqV69epz90dHR5MyZk5YtW8bsmzBhAs899xzp06fH09OTokWL8v777ydJjod50P3q999/p3fv3mTKlAkvLy+6dOnCvXv3uHLlCm3atMHHx4fs2bPz9ttvx/scs9vtfPzxx+TJkwc3NzfKlSvHunXr4hx34sQJOnToQJYsWXB1dcXf35/vv/8+znFHjx7lxRdfxMPDA19fX/r168edO3fiHGcYBp999hl58+bFzc2NMmXKsHLlyjjHxded5sHr/NChQ7Rv3x5vb2+yZs1Kjx49CA4OjnX727dv07NnTzJmzEj69Olp3Lgxp0+fTvTz+QF3d3emTZuGs7Mz48aNi9kfX3eabt26kT59eo4ePUqDBg1Ily4d2bNn55NPPgHM94Nq1aqRLl06ChcuzMyZMx957gfnOHnyJCtXrozTje78+fN06tQp1r/VF198gd1uj7mPB3/Xzz77jI8++oh8+fLh6urK+vXrH3luBwcHunTpwsyZM2PdX0BAALlz56Zu3bqP+yd8LA8K+nPnzj3yuLCwMIYMGRLrfe/VV1/l9u3bsY57ks+DDBkyxFtIVahQAYCLFy/G7Pvll19Inz49rVu3jnVs9+7duXz5Mjt27IiTadmyZTz//PMxn3HLli0DzNe6v78/6dKlo0KFCvF2T5SkN3DgQFq2bEm+fPnIkSMHOXPm5O2336Z06dLkz5+fAQMG0KBBA37++edH3k+jRo145ZVXKFiwIIMHD8bX1/c/u9stW7aM9OnT4+7uToECBTh8+DCDBw+O99gePXowY8YMDMNgwYIFFChQIM6v55UqVeL999+nQ4cO+Pr60rBhQ8aNGxfv53KVKlVInz59rEt83cX+i1ri06DAwEA6derEu+++y5gxY3BwML/LnThxgkaNGjFw4EDSpUvH0aNH+fTTT9m5c+djdZX4888/eeutt3jvvffImjUrU6dOpWfPnhQsWJAaNWo88rZnz56lb9++5MmTBzA/cAcMGMClS5cYPnx4gs9z+PBh6tSpg5+fHzNmzMDDw4Px48fz448/JuZPFktoaCg1a9bk4sWLvP/++5QqVYpDhw4xfPhwDhw4wNq1a7HZbGzbto22bdvStm1bRo4ciZubG+fOnYv5W5YpU4bp06fTvXt3hg0bRuPGjQES9SvHmTNnAChcuHC819etW5e8efMSEBDAp59+CsC0adOoUaNGnO43AJUrV2b58uV07NiRvn37UqFChZjlnx8mKioqzj4HB4eY51d8unfvzhtvvMGJEydi5Vi9ejWXL1+O+YIxb948XnnlFQYMGMDnn3+Og4MDJ0+e5PDhw4/M9F8eN3OvXr1o2bIl8+bNY9++fbz//vtERUVx7NgxWrZsSZ8+fVi7di2ffvopOXLkYNCgQbFu/91335E3b16+/vpr7HY7n332GQ0bNuSPP/6gcuXKgPmcrVKlCnny5OGLL74gW7ZsrFq1itdff50bN24wYsQIwPyiVrNmTZydnRk/fjxZs2Zlzpw5vPbaa3Eey6hRoxg1ahQ9e/akVatWXLhwgd69exMdHU2RIkUe62/08ssv07ZtW3r27MmBAwcYMmQIQExBZrfbY8ZkjBw5MqarS1J1ucuRIwdly5Zl69atREVFxfxMHZ/IyEhatmxJv379eOedd/jxxx8ZMmQIISEhLFy4kMGDB5MrVy6+/fZbunXrRokSJR7aD/XB43jppZcoUKBAzLiG7Nmzc/36dapUqUJERAQffvghfn5+LFu2jLfffptTp04xfvz4WPf1v//9j8KFC/P555/j5eUV72vu33r06MHYsWNZtWoVDRs2JDo6mpkzZ9KzZ89HvqYS4+TJk4D56+3DGIZBixYtWLduHUOGDKF69er89ddfjBgxgm3btrFt27ZYvy48yedBfH7//XecnJxivccdPHgQf3//OM+JUqVKxVxfpUqVWJmGDBnC0KFD8fb2ZtSoUbRs2ZIhQ4awbt06xowZg81mY/DgwTRp0oQzZ8785/ueVdydHTk8uoFl504qDxqgHoiOjuaTTz5h/vz5XLp0ifDwcMLDw2P1PY/Pg39zIKbbzrVr1x55m9q1azNhwgRCQ0OZOnUqx48fZ8CAAfEe27hxY/r27cvGjRsJCAiIaYz7t48//phBgwbx+++/s337diZOnMiYMWPYuHEjJUuWjDlu/vz5+Pv7x7pton41MiTV6tq1q5EuXbpY+2rWrGkAxrp16x55W7vdbkRGRhp//PGHARh//vlnzHUjRoww/v3UyZs3r+Hm5macO3cuZt/9+/eNjBkzGn379k1Q7ujoaCMyMtIYPXq0kSlTJsNutyf4PG3btjXc3d2NK1euxOyLiooyihYtagDGmTNnHjtPzZo1jeLFi8dsjx071nBwcDB27doV67gFCxYYgLFixQrDMAzj888/NwDj9u3bD73vXbt2GYAxffr0x8qyfv16AzDmz59vREZGGqGhocaWLVuMIkWKGMWKFTNu3boV6/h/PgdGjBhhZMuWzYiMjDSCgoIMV1dXY8aMGcb169cNwBgxYkTM7cLCwowWLVoYgAEYjo6OxvPPP28MHTrUuHbtWpxzPDju35c6deo88vHcuHHDcHFxMd5///1Y+9u0aWNkzZrViIyMNAzDMF577TXDx8fnsf5Gj+PB6yC+S8+ePWOOmz59ugEYAwYMiHX7B3+bL7/8Mtb+0qVLG2XKlInZPnPmjAEYOXLkMO7fvx+zPyQkxMiYMaNRt27dmH0NGjQwcuXKZQQHB8e6z9dee81wc3Mzbt68aRiGYQwePNiw2WzG/v37Yx1Xr149AzDWr19vGIZh3Lp1y3BzczNeeumlWMdt2bLFAIyaNWvGyfnP5+GD1/lnn30W6/avvPKK4ebmFvO6XL58uQEYEyZMiHXc2LFj4zyv4vPg3OPGjXvoMW3btjUA4+rVq4Zh/P06ePBYDePv5+HChQtj9kVGRhqZM2c2AGPv3r0x+4OCggxHR0dj0KBBj8xmGOZ7TuPGjWPte++99wzA2LFjR6z9/fv3N2w2m3Hs2LFYj61AgQJGRETEf57r3+erWbOm0apVK8MwzL+zzWYzzpw5Y/z8889xHv+Df6/r168/9L7je//YuHGjUbBgQcPR0THW+/y//fbbb/E+H+bPn28AxuTJk2M9hqT6PDAMw1i1apXh4OBgvPnmm7H2FypUyGjQoEGc4y9fvmwAxpgxY2Jlcnd3Ny5evBizb//+/QZgZM+e3bh3717M/sWLFxuAsWTJkgRnfVru379vHD58ONb7SEoyffp0w9vbO2b7wWtj3759sY779NNPjUyZMhk//PCDsX//fuPEiRNG48aNjebNm8ccU7NmTeONN96I2c6bN6/x1Vdfxbqf55577pHvPV27do11n4ZhGLVq1TKGDRsWa98/7/vtt982atasGev92Nvb+5Gf3+Hh4UaxYsWMLl26PPJxP8qj/u3VnSYNypAhQ7wDLU6fPk2HDh3Ili0bjo6OODs7U7NmTYDHmrmgdOnSMS3pAG5ubhQuXPg/f6IFs5Wlbt26eHt7x5x7+PDhBAUFxfk2/TjnWb9+PXXq1CFr1qwx+xwdHWnbtu1/Zvkvy5Yto0SJEpQuXZqoqKiYS4MGDWL9xP+gq0ybNm346aefuHTp0hOf+4G2bdvi7OyMh4cHVatWJSQkhOXLl+Pj4/PQ23Tv3p2rV6+ycuVK5syZg4uLS5yfoR9wdXXll19+4fDhw3z11Ve0a9eO69ev8/HHH+Pv7x+rWw6Y3R527doV5/LvFsl/y5QpE02bNo3VbeDWrVv8+uuvdOnSJaaFrUKFCty+fZv27dvz66+/PnIGjsdVoECBeDN/8MEHcY5t0qRJrO0HLSgPfj355/74nu8tW7bEzc0tZtvT05OmTZuyceNGoqOjCQsLY926dbz00kt4eHjEel41atSIsLCwmOnH1q9fT/HixXnuuedinePfg/i2bdtGWFgYHTt2jLW/SpUq5M2b97/+PDGaNWsWa7tUqVKEhYXFvC7/+OMPwHye/9M/B6I9KeNf3eoexmaz0ahRo5htJycnChYsSPbs2Xn++edj9mfMmJEsWbI81ntTfH7//XeKFSsW08XjgW7dumEYRpxfLps1axZvF5H/0qNHD5YsWUJQUBDTpk2jdu3aSTJ70j/fP2rUqEF0dDQLFiyI1Zr5bw8e07+nBWzdujXp0qWL0z3sST4P/mnv3r20adOGSpUqMXbs2DjXP6pr37+vK126dKyBuw9ex7Vq1Yo1LuzB/sQ+PyTxNm3aRPPmzenUqRPPPfcc+fPn58SJE8/k3CNGjODzzz/n8uXL8V7fo0cP/vjjD5o3b06GDBke6z5dXFwoUKBAnK62SUXdadKg+GZXuHv3LtWrV8fNzY2PPvqIwoUL4+HhwYULF2jZsiX379//z/vNlClTnH2urq7/edudO3dSv359atWqxZQpU8iVKxcuLi4sXryYjz/+OM7tH+c8QUFB8Y78TorR4FevXuXkyZMP/VB+UGDWqFGDxYsX87///Y8uXboQHh5O8eLFGTp06BMXOJ9++ikvvPACoaGhrF69mrFjx9KiRQt27Njx0AFzefPmpU6dOgQEBHD27FnatWuHh4dHnKmv/snf3z/mA80wDL7++msGDRrEBx98wE8//RRznIODQ5yfRR9Xjx49WLhwIWvWrKFBgwbMnTuX8PDwWMVC586diYqKYsqUKbz88svY7XbKly/PRx99RL169RJ13gd90x9HxowZY227uLg8dH9YWFic2z/suRgREcHdu3e5e/cuUVFRfPvtt3z77bfxZnjwvAoKCiJfvnz/eY6goKBHnvtx/fv19uD59eD1FhQUhJOTU5y/xT+/QD+pc+fO4erqGucc/+bh4RHryxKY/ybx3e5h/1aPIygoKN5iOkeOHDHX/1NiZ7Rp1aoVAwYM4KuvvmLp0qVJNgXog/cPR0dHfH19YwaKPsqDf+d/d7l50HXh3485sZ8H/7Rv3z7q1atHoUKFWLFiRZz3tkyZMsU5L8DNmzeBuK/PhLyOgUQ/PyTxChYsyMKFC9m6dSsZMmTgyy+/5MqVK3G6njwNtWrVonjx4owZM4bvvvsuzvX+/v7cuHHjoROBLFu2jHnz5tGuXTsKFy6MYRgsXbqUFStWxJmiMigoiCtXrsTa5+PjE+f967+oiE+D4mu5+P3337l8+TIbNmyIaX0H4gxYehrmzZuHs7Mzy5Yti/UEXrx4caLvM1OmTHFeIEC8+xLK19cXd3f3hw7S8vX1jfn/5s2b07x5c8LDw9m+fTtjx46lQ4cO+Pn5xfSFToz8+fPHFKA1atTA3d2dYcOG8e233/L2228/9HY9evSgU6dO2O12JkyYkKBz2mw23nzzTUaPHs3BgwcTnf3fGjRoQI4cOZg+fToNGjRg+vTpVKxYMc70gd27d6d79+7cu3ePjRs3MmLECJo0acLx48cT1LJshYc9F11cXEifPj3Ozs44OjrSuXNnXn311Xjv40Hh/rjP7QdF1MOOTar58DNlykRUVBQ3b96MVQwlxWsN4NKlS+zZs4eaNWs+sj/8s5QpUyYCAwPj7H/QgvfP9wBI/EBwDw8P2rVrx9ixY/Hy8oo10PtJ/PP943E9+He+fv16rELeMAyuXLkS88tjUtm3b1/MWJ7Vq1fj7e0d55iSJUsyd+7cOGMlDhw4AECJEiWSNJM8fR988AFnzpyhQYMGeHh40KdPH1q0aBFnMP3TMmjQILp3787gwYPj/XIb35fTB4oVK4aHhwdvvfUWFy5cwNXVlUKFCjF16lQ6d+4c69j4BqfPnTuXdu3aJShv8nhHFMs9+JD5d0vHpEmTnsm5nZycYg3quH//Pj/88EOi77N27dosWbKEq1evxrQIRkdHM3/+/CfO26RJE8aMGUOmTJnibRGNj6urKzVr1sTHx4dVq1axb98+KleuHKdVM7HeffddZsyYwSeffELfvn0fuhDGSy+9xEsvvYS3t/cjp+gLDAyMt/Xw8uXLhISEJGpRiod5ULx+/fXXbNq0id27dz/yeZcuXToaNmxIREQELVq04NChQ8m+iF+0aBHjxo2L+ZJ6584dli5dSvXq1XF0dMTDw4PatWuzb98+SpUqFdMSGJ/atWvz2Wef8eeff8bqUvPvQduVKlXCzc2NOXPm8PLLL8fs37p1K+fOnUuyIr5mzZp89tlnzJ8/n/79+8fsT4oFd+7fv0+vXr2Iiori3XfffeL7Syp16tRh7Nix7N27lzJlysTsf7AQU+3atZPsXP37948ZzJzQVrqkVKdOHT777DNmz57Nm2++GbN/4cKF3Lt3jzp16iTZufbv30/dunXJlSsXa9aseWjXhZdeeokpU6awcOHCWF0lZ86cSY4cOR45JaE8W926dYv166qfn1+83eQyZsz4nw14/551Jr6FF/9rAcWH/arVoUOHWF0T/2tRx382dObPn5/Jkyc/8viHPe7EUhEvgNlPNkOGDPTr148RI0bg7OzMnDlz+PPPP5/6uRs3bsyXX35Jhw4d6NOnD0FBQXz++edPtOjOsGHDWLJkCS+88ALDhw/Hw8OD77//Pkn6pQ0cOJCFCxdSo0YN3nzzTUqVKoXdbuf8+fOsXr2at956i4oVKzJ8+HAuXrxInTp1yJUrF7dv3+abb76JNdagQIECuLu7M2fOHPz9/UmfPj05cuSI+Vn+cTk7OzNmzBjatGnDN998w7Bhw+I9zs3NjQULFvzn/fXp04fbt2/z8ssvU6JECRwdHTl69ChfffUVDg4OcabhstvtD10y+vnnn//Pf8sePXrw6aef0qFDB9zd3eOMXejduzfu7u5UrVqV7Nmzc+XKFcaOHYu3t3dMC+C5c+coUKAAXbt2Zdq0af/5GO/fv//QzEk9B7mjoyP16tVj0KBB2O12Pv30U0JCQmLN6f/NN99QrVo1qlevTv/+/fHz8+POnTucPHmSpUuXxvRJHjhwIAEBATRu3JiPPvooZnaao0ePxjpnhgwZePvtt/noo4/o1asXrVu35sKFC4wcOTJJupU98OKLL1K1alXeeuutmC9427Zti5k/+XFnUjl//jzbt2/HbrcTHBwcs9jTuXPn+OKLL6hfv36SZX5Sb775JrNmzaJx48aMHj2avHnzsnz5csaPH0///v0fOktUYpQuXTpBv0ouXbo03i/xrVq1eqIc9erVo0GDBgwePJiQkBCqVq0aMzvN888/H6elMbGOHTsW00r58ccfc+LEiVh9ogsUKBDzS0DDhg2pV68e/fv3JyQkhIIFCzJ37lx+++03Zs+enWrniBd5QEW8AOZPRMuXL+ett96iU6dOpEuXjubNmzN//vxYLU1PwwsvvBAz9WHTpk3JmTMnvXv3JkuWLHFWGH1cJUqUYO3atbz11lt07dqVDBky0LlzZ15++eX/XMXtv6RLl45NmzbxySefMHny5JhpyPLkyUPdunVjWjgrVqzI7t27GTx4MNevX8fHx4dy5crx+++/U7x4ccD8uTwgIIBRo0ZRv359IiMjGTFiRKLm1m7dujUVK1bkyy+/ZMCAAfH+/Py4BgwYwPz585kyZQqXLl3i3r17ZM6cmcqVKzNr1qw4Re79+/cf2j3oxIkTFCxY8JHnK1y4MFWqVGHr1q107NgxTvbq1aszY8YMfvrpJ27duoWvry/VqlVj1qxZMR/ohmEQHR392HPtnj59+qGZIyMjk7TrxmuvvUZYWBivv/46165do3jx4ixfvpyqVavGHFOsWDH27t3Lhx9+yLBhw7h27Ro+Pj4UKlQo1mDNbNmy8ccff/DGG2/Qv39/PDw8eOmll/juu+9o3rx5rPOOHj2adOnSMX78eH744QeKFi3KxIkTY6ZLTAoODg4sXbqUt956i08++YSIiAiqVq3K7NmzqVSp0iMHW//Tg/EAjo6OeHl5kT9/fpo2bUrv3r2fycqsCZE5c2a2bt3KkCFDYqawzJ8/P5999lmc6UWftYdNffekrX82m43FixczcuRIpk+fzscff4yvry+dO3dmzJgxSbbS8bZt22L6uTdt2jTO9dOnT4/Vorto0SKGDh3K8OHDuXnzJkWLFk1UtwSRlMhmJGW7voiICGb3no4dO7Jly5ZYc3WLSMKFhYVx5swZ8uXLZ2m3Knn2HvVvr5Z4ERF5InPnzuXSpUuULFkSBwcHtm/fzrhx46hRo4YKeBGRp0RFvIiIPBFPT0/mzZvHRx99xL1798iePTvdunXjo48+sjqaiEiqpSJeRESeSJMmTeIsiCUiIk+XVmwVEREREUlhVMSLiIiIiKQwKuJFRERERFIYFfEiIiIikqRsNtsjL/+c7z+h/Pz8+Prrrx/ruAfnc3d3p2jRoowbNy7Wuglnz56NWTn+0qVLsW4fGBiIk5MTNpst1uqtCxcupGLFinh7e+Pp6Unx4sV56623Yq6fMWNGvI85qacH1cBWEREREUlSgYGBMf8/f/58hg8fzrFjx2L2ubu7P5Mco0ePpnfv3oSFhbF27Vr69++Pl5cXffv2jXVcjhw5mDVrFkOGDInZN3PmTHLmzMn58+dj9q1du5Z27doxZswYmjVrhs1m4/Dhw6xbty7W/Xl5ecV6vGB+sUlKaokXERERkSSVLVu2mIu3tzc2my3Wvo0bN1K2bFnc3NzInz8/o0aNIioqKub2I0eOJE+ePLi6upIjRw5ef/11AGrVqsW5c+d48803Y1q4H8XT05Ns2bLh5+dHr169KFWqFKtXr45zXNeuXZk+fXqsfTNmzKBr166x9i1btoxq1arxzjvvUKRIEQoXLkyLFi349ttvYx3378ebLVs2smbNmqC/4X9RES8iIiKSkhgGRNyz5vKPriiJtWrVKjp16sTrr7/O4cOHmTRpEjNmzODjjz8GYMGCBXz11VdMmjSJEydOsHjxYkqWLAnAokWLyJUrF6NHjyYwMDBWi/+j/2QGGzZs4MiRIzg7O8e5vlmzZty6dYvNmzcDsHnzZm7evEnTpk1jHZctWzYOHTrEwYMHn+RPkCTUnUZEREQkJYkMhTE5rDn3+5fBJd0T3cXHH3/Me++9F9PKnT9/fj788EPeffddRowYwfnz58mWLRt169bF2dmZPHnyUKFCBQAyZsyIo6NjTAv7fxk8eDDDhg0jIiKCyMhI3NzcYlr1/8nZ2ZlOnToREBBAtWrVCAgIoFOnTnEK/gEDBrBp0yZKlixJ3rx5qVSpEvXr16djx464urrGHBccHEz69Olj3bZKlSrx/gqQWCriRUREROSZ2bNnD7t27YppeQeIjo4mLCyM0NBQWrduzddff03+/Pl58cUXadSoEU2bNsXJKeFl6zvvvEO3bt24fv06Q4cO5YUXXqBKlSrxHtuzZ08qV67MmDFj+Pnnn9m2bVusLj4A6dKlY/ny5Zw6dYr169ezfft23nrrLb755hu2bduGh4cHYHbj2bt3b6zbJvU4ABXxIiIiIimJs4fZIm7VuZ+Q3W5n1KhRtGzZMs51bm5u5M6dm2PHjrFmzRrWrl3LK6+8wrhx4/jjjz/i7QrzKL6+vhQsWJCCBQuycOFCChYsSKVKlahbt26cY0uUKEHRokVp3749/v7+lChRgv3798d7vwUKFKBAgQL06tWLoUOHUrhwYebPn0/37t0BcHBwoGDBggnKmlAq4kVERERSEpvtibu0WKlMmTIcO3bskUWuu7s7zZo1o1mzZrz66qsULVqUAwcOUKZMGVxcXIiOjk7weTNkyMCAAQN4++232bdvX7yDYnv06MErr7zChAkTHvt+/fz88PDw4N69ewnO9CRUxIuIiIjIMzN8+HCaNGlC7ty5ad26NQ4ODvz1118cOHCAjz76iBkzZhAdHU3FihXx8PDghx9+wN3dnbx58wJm0bxx40batWuHq6srvr6+j33uV199lU8//ZSFCxfSqlWrONf37t2b1q1b4+PjE+/tR44cSWhoKI0aNSJv3rzcvn2b//3vf0RGRlKvXr2Y4wzD4MqVK3FunyVLFhwckmZeGc1OIyIiIiLPTIMGDVi2bBlr1qyhfPnyVKpUiS+//DKmSPfx8WHKlClUrVqVUqVKsW7dOpYuXUqmTJkAc+73s2fPUqBAATJnzpygc2fOnJnOnTszcuRI7HZ7nOudnJzw9fV9aP/7mjVrcvr0abp06ULRokVp2LAhV65cYfXq1RQpUiTmuJCQELJnzx7ncu3atQTlfRSbYSTBXEEiIiIi8lSEhYVx5swZ8uXLl+Srfkry9qh/e7XEi4iIiIikMCriRURERERSGBXxIiIiIiIpjIp4EREREZEURkW8iIiISAqguUjSnkf9m6uIFxEREUnGHB0dAYiIiLA4iTxroaGhAPGuVKvFnkRERESSMScnJzw8PLh+/TrOzs5JtliQJF+GYRAaGsq1a9fw8fGJ+SL3T5onXkRERCSZi4iI4MyZM/EuUCSpl4+PD9myZcNms8W5TkW8iIiISApgt9vVpSYNcXZ2jrcF/gEV8SIiIiIiKYw6VYmIiIiIpDAq4kVEREREUhgV8SIiIiIiKYyKeBERERGRFEZFvIiIiIhICqMiXkREREQkhVERLyIiIiKSwqiIFxERERFJYVTEi4iIiIikMCriRURERERSGBXxIiIiIiIpjIp4EREREZEURkW8iIiIiEgKoyJeRERERCSFUREvIiIiIpLCqIgXEREREUlhVMSLiIiIiKQwKuJFRERERFIYJ6sDJBW73c7ly5fx9PTEZrNZHUdERERE/sUwDO7cuUOOHDlwcFBb8pNINUX85cuXyZ07t9UxREREROQ/XLhwgVy5clkdI0VLNUW8p6cnYD4pvLy8LE4jIiIiIv8WEhJC7ty5Y+o2SbxUU8Q/6ELj5eWlIl5EREQkGVPX5yenzkgiIiIiIimMingRERERkRRGRbyIiIiISAqjIl5EREREJIVRES8iIiIiksKoiBcRERERSWFUxIuIiIiIpDAq4kVEREREUhgV8SIiIiIiKYyKeBERERGRFEZFvIiIiIhICqMiXkREREQkhVERLyIiIiKSwqiIFxERERFJYVTEi4iIiIikMCriRURERERSGBXxIiIiIiIpjIp4EUmbTv0O8zqa/xUREUlhVMSLSNqz9weY3QqOLoMfXoLfhkBkmNWpREREHpuKeBFJOwwDNnwKS14DIxqylTT3bx8PU16Aq4etzSciIvKYVMSLSNoQHQVL34ANY8ztaoOg7yZoPx88fOHaIZhcC7ZPNIt9ERGRZCxRRfz48ePJly8fbm5ulC1blk2bNj3y+O+//x5/f3/c3d0pUqQIs2bNinPM7du3efXVV8mePTtubm74+/uzYsWKxMQTEYkt4h7M6wB7Z4LNARp/AXVHgM0GRV6EV7ZBofoQHQ6/DYbZL8OdK1anFhEReSinhN5g/vz5DBw4kPHjx1O1alUmTZpEw4YNOXz4MHny5Ilz/IQJExgyZAhTpkyhfPny7Ny5k969e5MhQwaaNm0KQEREBPXq1SNLliwsWLCAXLlyceHCBTw9PZ/8EYpI2nb3OvzYBi7vBSc3aBUARRvHPiZ9FujwE+yaCquHwal1MKEKNPs27rEiIiLJgM0wEva7ccWKFSlTpgwTJkyI2efv70+LFi0YO3ZsnOOrVKlC1apVGTduXMy+gQMHsnv3bjZv3gzAxIkTGTduHEePHsXZ2fmxcoSHhxMeHh6zHRISQu7cuQkODsbLyyshD0lEUqugU2ar+q0z4J4ROsyH3BUefZtrR2FRL7hywNwu2x0afAwu6Z5+XhGRVC4kJARvb2/Va0kgQd1pIiIi2LNnD/Xr14+1v379+mzdujXe24SHh+Pm5hZrn7u7Ozt37iQyMhKAJUuWULlyZV599VWyZs1KiRIlGDNmDNHR0Q/NMnbsWLy9vWMuuXPnTshDEZHU7uJumFbPLOB98kLPNf9dwANkKQq91kGVAeb2nukwqQZc3vd084qIiCRAgor4GzduEB0dTdasWWPtz5o1K1euxN9/tEGDBkydOpU9e/ZgGAa7d+8mICCAyMhIbty4AcDp06dZsGAB0dHRrFixgmHDhvHFF1/w8ccfPzTLkCFDCA4OjrlcuHAhIQ9FRFKzY7/BjCYQGgTZS0OvteBb8PFv7+QK9T+CLkvAMwcEnYSpdWHTF2B/eOOCiIjIs5Koga02my3WtmEYcfY98MEHH9CwYUMqVaqEs7MzzZs3p1u3bgA4OjoCYLfbyZIlC5MnT6Zs2bK0a9eOoUOHxuqy82+urq54eXnFuoiIsHs6zGsPUfehYD3ottzs854Y+WtC/y1QrDnYo2DdaJjZFG6r0UBERKyVoCLe19cXR0fHOK3u165di9M6/4C7uzsBAQGEhoZy9uxZzp8/j5+fH56envj6+gKQPXt2ChcuHFPUg9nP/sqVK0RERCT0MYlIWmQY8PtHsGwgGHYo3QnazwXX9E92vx4ZofVMaD4eXNLDuS0woSocWJAksUVERBIjQUW8i4sLZcuWZc2aNbH2r1mzhipVqjzyts7OzuTKlQtHR0fmzZtHkyZNcHAwT1+1alVOnjyJ3W6POf748eNkz54dFxeXhEQUkbQoOhIWvwIb/38Afc33oPl34Ph4A+X/k80Gz3eEfpsgV3kID4aFPWFRHwgLTppziIiIJECCu9MMGjSIqVOnEhAQwJEjR3jzzTc5f/48/fr1A8y+6l26dIk5/vjx48yePZsTJ06wc+dO2rVrx8GDBxkzZkzMMf379ycoKIg33niD48ePs3z5csaMGcOrr76aBA9RRFK18DvwY1v480ewOULT/0HtIWbhndQy5ofuv5lfEmwO8Nd8mFANzm1L+nOJiIg8QoLniW/bti1BQUGMHj2awMBASpQowYoVK8ibNy8AgYGBnD9/Pub46OhovvjiC44dO4azszO1a9dm69at+Pn5xRyTO3duVq9ezZtvvkmpUqXImTMnb7zxBoMHD37yRygiqdedq/Bjawj8E5w9oPUMKNzg6Z7T0cn8klDgBVjUG26fgxmNoPpbUHNw0rX+i4iIPEKC54lPrjTvqEgac+MEzG4Jt8+Dhy90/Alyln22GcJCYOW78OdccztnWWg5BTIVeLY5RERSCNVrSSdRs9OIiFjq/A5zDvjb580uLr3WPPsCHsDNC16aCK2mg5s3XNoDE6vD3lnmQFsREZGnREW8iKQsR5bCrGZw/5ZZuPdcYxbyVirREvpvBb/qEHkPlgyAnzpD6E1rc4mISKqlIl5EUo6dU2B+Z4gKg8IvQtelkM7X6lQm71zQ5VeoOwocnM0vGxOqwKn1VicTEZFUSEW8iCR/djusGQEr3gYMKNsd2s4Bl3RWJ4vNwRGqDTRXiM1UCO4Ewg8tYNVQiAq3Op2IiKQiKuJFJHmLioBf+sKWr83tF4ZBk6/MWWKSqxyloe9GKNfD3N72HUx5Aa4dsTSWiIikHiriRST5CguGOa3gwE/g4GSumlrjnaczB3xSc/Ewv2y0n2fOnnP1IEyuBTsmadCriIg8MRXxIpI8hQTC9EZw5g9wSQ8d5purpqY0RRqag14L1jX78q98F+a0Nue4FxERSSQV8SKS/Fw7ClPrmq3X6bJAt+VmEZxSeWaFjgug4ThwcoOTa8xBr8dWWp1MRERSKBXxIpK8nN0CAfUh5KI5OLTXGrOPeUpns0HFPtBnA2QtAaE3YG47WPYmRIRanU5ERFIYFfEiknwc+sWczSUsGHJXhJ6rIYOf1amSVhZ/6P07VH7N3N4dAJNqwOX9lsYSEZGURUW8iCQP28bDz90hOgKKNjHnXPfIaHWqp8PJFRp8DJ0Xg2d2CDphdh/a/BXYo61OJyIiKYCKeBGxlt1uzqO+aghgQPne0GYWOLtbnezpK1DbHPTq3xTskbB2JMxsBsEXrU4mIiLJnIp4EbFOVDgs7GnOow5QdyQ0GmcumpRWeGSENj9As+/AOR2c22wOej240OpkIiKSjKmIFxFr3L8NP7SEQ4vAwRlaToFqb6aMOeCTms0GZTpDv02Qs6w5JmBBD/ilH4SFWJ1ORESSIRXxIvLsBV+EgBfNVmcXT+i0AEq1sTqV9TIVgB6roMa7YHOAP+fCxGpwfofVyUREJJlRES8iz9bVQzC1Hlw/Yg7q7LES8teyOlXy4egMLwyF7ivBJw/cPgfTX4T1YyA6yup0IiKSTKiIF5Fn58xGswX+zmXIXBR6roFsJa1OlTzlqQT9NkOpdmDY4Y9PIaAB3DxtdTIREUkGVMSLyLNxYIHZBz48BPJUgR6/gU9uq1Mlb27e0HISvDwNXL3h0m6YWB32zQbDsDqdiIhYSEW8iDxdhgFbvjFnobFHQrEW0PkXcM9gdbKUo2Qr6L8F8laDiLvw66vwUxcIvWl1MhERsYiKeBF5euzRsHIwrBlubld6BVpNB2c3a3OlRD65oesSqDMCHJzgyBKYUBVO/2F1MhERsYCKeBF5OiLvw89dYeckc7v+x/DiWHDQ206iOThC9UHQay1kKmiOLZjVDFYPM+fcFxGRNEOfpiKS9EJvwqwWcGQpOLpAqwCo8prVqVKPHM9D341Qtru5vfVbmFoHrh21NpeIiDwzKuJFJGndOmfOonJhuzkYs9MiKPGy1alSH5d00PRraPcjeGSCKwdgck3YOUWDXkVE0gAV8SKSdAL/gmn14MZx8MppzkCTr7rVqVK3oo2h/zYoUAeiwmDF2/BjG7h7zepkIiLyFKmIF5Gkcep3mN4Q7l6FLMXNOeCzFrM6VdrgmRU6LoAXPwVHVzixGsZXhmO/WZ1MRESeEhXxIvLk9s+FOa3N6Q/9qpursHrntDpV2uLgAJX6QZ8NkLUEhN6AuW1h2SCICLU6nYiIJDEV8SKSeIYBGz+Hxf3AHgUlWkGnheYiRWKNrMWg1zqo9Kq5vXua2Vc+8E9rc4mISJJSES8iiWOPhuVvwe8fmttVXoeWU8DJ1dpcYs7D/+IYc1Gt9NnMMQpT6piLbtntVqcTEZEkoCJeRBIuIhTmdzJbebFBw8+g/oeaAz65KfAC9N8KRZuYq+WuGW7OKx980epkIiLyhPSJKyIJcy/ILASPrTAHUbaZCRX7Wp1KHiZdJmg7G5r+D5w94OwmmFAFDv1idTIREXkCKuJF5PHdPGNOIXlxF7j5QJdfoVhzq1PJf7HZoGxX6LcZcpSBsGD4uRv80h/C71idTkREEkFFvIg8nkt7zQL+5inwzgM9V0PeylankoTIVMD8d6v+Ntgc4M8fYWI1uLDT6mQiIpJAKuJF5L+dWAMzmsC965CtJPRaA5mLWJ1KEsPRGep8AN2Wm1/Gbp2FgBdh/ViIjrI6nYiIPCYV8SLyaHt/gB/bQuQ9yF8buq0Az2xWp5InlbcK9N8MJduAEQ1/fGIu1nXzjNXJRETkMaiIF5H4GQZs+ASWvGYWec+1hw4/gZuX1ckkqbh5w8tToOVUcPWCizvN7jX7fzT//UVEJNlSES8icUVHwdLXYcNYc7v6W9BiAji5WJtLno5SraH/FshTxVx1d3F/c+Br6E2rk4mIyEOoiBeR2CLuwbz2sHeWOfix8ZdQZ7g5w4mkXj55oNsyeOEDcHCCw4thQlU4s9HqZCIiEg8V8SLyt7vXYUZjOLEanNzN+cXL97Q6lTwrDo5Q423ouQYyFoA7l2FmM1j9AURFWJ1ORET+QUW8iJiCTsG0unB5H7hnhK5LoWhjq1OJFXKWgX6boExXwICt/4OpdeD6MauTiYjI/1MRLyJwcbc5B/yts+CT12yJzV3e6lRiJZd00Ox/0HaO+aXuyl8wqQbsnKJBryIiyYCKeJG07thKcw740CDIXhp6rQXfglankuTCvwn03woFXoCoMFjxtjnl6N3rVicTEUnTVMSLpGW7A2BeB4i6DwXrmQsApc9idSpJbryyQ8eF8OIn4OgKJ1bBhMpwfLXVyURE0iwV8SJpkWHAug9h2Ztg2OH5TtB+LrimtzqZJFcODlCpP/RZD1mKmav3/tgalr8NkfetTicikuaoiBdJa6IjYfErsOlzc7vme9DsO3B0tjaXpAxZi0Pv9VCxv7m9awpMqgmBf1mbS0QkjVERL5KWhN+BH9vAnz+CzRGa/g9qD9Ec8JIwzm7Q8BPotAjSZ4Ubx2DKC7Dlf2C3W51ORCRNUBEvklbcuQLTG8Gp38HZA9rPg7JdrU4lKVnBOtB/GxRpDPZIWPMB/NAcgi9ZnUxEJNVTES+SFlw/DlPrmdMEeviaK3MWrm91KkkN0mWCdnOg6Tfml8MzG2FCFTi02OpkIiKpmop4kdTu/HYIqA/B5yFjfui1BnKWtTqVpCY2G5TtBn03QY7nIew2/NwVFr9qduESEZEkpyJeJDU7shRmNYf7tyBnOXMRp4z5rU4lqZVvQfM5Vv0twAb7Z8PE6nBhl9XJRERSHRXxIqnVjskwv7O5QE/hhtB1KaTztTqVpHaOzlBnuLnmgHduuHUGAhrAhk8hOsrqdCIiqYaKeJHUxm6HNSNg5TuAAWW7Q9vZ4OJhdTJJS/yqQr/NULI1GNGwYQzMaAQ3z1idTEQkVVARL5KaREXAL31hy9fm9gvDoMlX4OhkaSxJo9x94OWp0HIKuHrBhR1m95r9c80Fx0REJNFUxIukFmHBMOdlOPATODhBiwlQ4x3NAS/WK9XGbJXPUxki7sDifrCguzlWQ0REEkVFvEhqEHLZnAP+zEZwSQ8dfoLSHaxOJfK3DHnNfvIvDDMXGjv0C0yoCmc2WZ1MRCRFUhEvktJdO2LOAX/1oLl6Zrfl5iI8IsmNg6P569CDWZJCLsHMpuYYjqgIq9OJiKQoKuJFUrKzm82ZP0IuQqZCZnGUo7TVqUQeLVdZc075Ml0AwxzDMa2uuSiZiIg8FhXxIinVwUXww0tmX/jcFaHnarPLgkhK4Joemn1rzpzkngEC/4RJNWDXNA16FRF5DIkq4sePH0++fPlwc3OjbNmybNr06D6N33//Pf7+/ri7u1OkSBFmzZoV6/oZM2Zgs9niXMLCwhITTyT12zYeFvSA6Ago2gS6/AoeGa1OJZJw/k2h/zbIXxui7sPyQTC3Pdy9bnUyEZFkLcFF/Pz58xk4cCBDhw5l3759VK9enYYNG3L+/Pl4j58wYQJDhgxh5MiRHDp0iFGjRvHqq6+ydOnSWMd5eXkRGBgY6+Lm5pa4RyWSWtnt8Nv7sGoIYECFPtBmFji7W51MJPG8skOnRdBgLDi6wPGVMKEKnFhjdTIRkWTLZhgJ+92yYsWKlClThgkTJsTs8/f3p0WLFowdOzbO8VWqVKFq1aqMGzcuZt/AgQPZvXs3mzdvBsyW+IEDB3L79u1EPgwICQnB29ub4OBgvLy8En0/IslWVLg5B/yhX8ztuqOg6huaQlJSlysHYWEvuH7E3K7QB+qN1hdVkVRC9VrSSVBLfEREBHv27KF+/fqx9tevX5+tW7fGe5vw8PA4Leru7u7s3LmTyMjImH13794lb9685MqViyZNmrBv375HZgkPDyckJCTWRSTVun8LfmhpFvAOzubiOdUGqoCX1CdbCeizHir2M7d3TobJteDKAUtjiYgkNwkq4m/cuEF0dDRZs2aNtT9r1qxcuXIl3ts0aNCAqVOnsmfPHgzDYPfu3QQEBBAZGcmNGzcAKFq0KDNmzGDJkiXMnTsXNzc3qlatyokTJx6aZezYsXh7e8dccufOnZCHIpJyBF+EgIZwbrO56mWnBebiOSKplbM7NPwUOi6EdFng+lGY8gJs/c7sUiYiIokb2Gr7V+ufYRhx9j3wwQcf0LBhQypVqoSzszPNmzenW7duADg6OgJQqVIlOnXqxHPPPUf16tX56aefKFy4MN9+++1DMwwZMoTg4OCYy4ULFxLzUESStysHYWpds2uBZ3bovhLy17I6lcizUaguvLINijQyB3GvHgqzXzIXNxMRSeMSVMT7+vri6OgYp9X92rVrcVrnH3B3dycgIIDQ0FDOnj3L+fPn8fPzw9PTE19f3/hDOThQvnz5R7bEu7q64uXlFesikqqc/gOmN4Q7gZC5qDkHfLYSVqcSebbS+UK7H6HJV+DkDqc3mINeDy+xOpmIiKUSVMS7uLhQtmxZ1qyJPWPAmjVrqFKlyiNv6+zsTK5cuXB0dGTevHk0adIEB4f4T28YBvv37yd79uwJiSeSevz1M8x+GcJDIG9V6PEb+KjLmKRRNhuU6wF9N0L258wxIj91hl9fhfC7VqcTEbGEU0JvMGjQIDp37ky5cuWoXLkykydP5vz58/TrZw5CGjJkCJcuXYqZC/748ePs3LmTihUrcuvWLb788ksOHjzIzJkzY+5z1KhRVKpUiUKFChESEsL//vc/9u/fz/fff59ED1MkhTAM2PINrB1hbhd/CVpMBGdNtypC5sLQcy1sGAObv4Z9s+HcVmg51VwFVkQkDUlwEd+2bVuCgoIYPXo0gYGBlChRghUrVpA3r7lSZGBgYKw546Ojo/niiy84duwYzs7O1K5dm61bt+Ln5xdzzO3bt+nTpw9XrlzB29ub559/no0bN1KhQoUnf4QiKYU9Gn4bAjsnmduVXoX6H8FDfrESSZOcXKDuSChYFxb1hZunYVo9qPUeVBsEjgn+WBMRSZESPE98cqV5RyVFi7wPi3rDkf9fBK3BGKj8qrWZRJK7+7dg2SA4tMjczl0JWk6CDH6WxhKRh1O9lnTUxCditdCbMKuFWcA7ukCr6SrgRR6HewZoFQAvTQIXT7iwHSZUgz/nm13TRERSMRXxIla6dQ6m1TeLD1dv6PwLlGhpdSqRlMNmg+faQf/NZkt8xB34pQ8s7An3b1udTkTkqVERL2KVwD/NvrxBJ8ArJ/RcBX7VrE4lkjJl8INuy6H2ULA5wsGFMLEanN1sdTIRkadCRbyIFU6ug+mN4O5VyFIceq2FLP5WpxJJ2RydoOa70HM1ZMgHwRdgRhNYOxKiIqxOJyKSpFTEizxr+3+EH9tAxF3wqw49VoJXDqtTiaQeucpBv03wfCfAgM1fmb963Xj4AoIiIimNiniRZ8UwYOPnsLg/2KOgRCvotBDcvK1OJpL6uHpC8++hzSxw84HA/TCpBuyerkGvIpIqqIgXeRaio2D5IPj9Q3O76hvQcgo4uVqbSyS1K9YcXtkG+WpCZCgsGwjzOsK9G1YnExF5IiriRZ62iFCY3wl2BwA2aDgO6o3WIk4iz4pXDui82Fw8zdEFji2HCVXgxFqrk4mIJJqqCJGn6d4NmNkUjq8ER1fzp/2KfaxOJZL2ODhAlQHQ+3fIXNQcVD7nZVg5GCLDrE4nIpJgKuJFnpabp8054C/tNvvkdl0CxZpZnUokbctWEvpsgAr//2V6x0SYUhuuHLQ0lohIQqmIF3kaLu0xC/ibp8A7D/RcA3kqWZ1KRACc3aHROOjwM6TLDNcOm4X8tu/Bbrc6nYjIY1ERL5LUjq8256a+d91s9eu1BjIXtjqViPxb4frQfxsUfhGiI2DV+zC7JYQEWp1MROQ/qYgXSUp7Z8HcduYsGPlrQ7cV4JnN6lQi8jDpM0P7edD4S3Byh9PrzUGvR5ZanUxE5JFUxIskBcOA9WNhyQAwouG59tDxZ3DzsjqZiPwXmw3K94S+GyFbKbh/05xRaskACL9rdToRkXipiBd5UtFR5of9H5+Y29XfhhYTwNHZ2lwikjCZC0OvdeY6DtjMX9YmVTfHuIiIJDMq4kWeRPhdmNce9v0ANgfzJ/k6H5gteyKS8ji5mOs4dF0CXjn/nmVq4ziwR1udTkQkhop4kcS6ew1mNoETq82+tG3nmD/Ji0jKl68G9N8CxV8CexT8/hHMaAy3zlmdTEQEUBEvkjg3TsK0enB5H3hkgq5LoWgjq1OJSFJyzwCtpkOLieCSHs5vg4nV4K+frE4mIqIiXiTBLuwyC/hbZyGDnzkHfO7yVqcSkafBZoPS7aHfZshVAcJDYFFvWNgL7t+2Op2IpGEq4kUS4ugKmNnUnL0ix/NmAZ+pgNWpRORpy5gPuq+EWu+DzREO/Gy2yp/banUyEUmjVMSLPK5d02B+R4i6D4XqQ9dlkD6L1alE5FlxdIJag6HHKvNXuOALZj/5daMhOtLqdCKSxqiIF/kvhgHrPoTlg8Cww/Odod1ccE1vdTIRsULu8mb3mtIdzfeETV+YXexunLQ6mYikISriRR4lOhIWvwKbPje3aw2BZt+aLXIikna5ekKL8dB6Jrj5mIPcJ1WHPTPML/4iIk+ZiniRhwm/A3Naw58/mn1gm30Ltd7THPAi8rfiLaD/VnNKyshQWPqGudrrvSCrk4lIKqciXiQ+d67A9EZwej04e0D7eVCmi9WpRCQ58s4JnX+Feh+CgzMcXQYTKsPJdVYnE5FUTEW8yL9dPw5T68GVv8DDF7otg8L1rU4lIsmZgwNUfR16rwPfInD3KsxuCb8Ngcgwq9OJSCqkIl7kn85vh4D6EHweMhaAXmsgZ1mrU4lISpH9OeizAcr3Nre3j4cpL8DVQ5bGEpHUR0W8yAOHl8DMZnD/FuQsBz1XQ8b8VqcSkZTGxQMafw4dfoJ0meHaIZhcG7ZPALvd6nQikkqoiBcB2DEZfuoC0eFQuCF0XQrpfK1OJSIpWeEG5qDXQvXN95bf3oM5L5tjbkREnpCKeEnb7HZYMxxWvgMYULY7tJ1ttqSJiDyp9FnMFvlGn4OTG5z6HcZXhqPLrU4mIimcinhJu6LC4Zc+sOUbc/uFD6DJV5oDXkSSls0GFXpD342QrSTcvwnzOsCS1yHintXpRCSFUhEvaVNYMMxpBQd+BgcnaDEBarytOeBF5OnJXAR6rYMqrwM22DsTJtWAS3utTiYiKZCKeEl7Qi6bc8Cf2Qgu6c2fukt3sDqViKQFTq5Q/0Po8it45oCgkzCtHmz6AuzRVqcTkRRERbykLdeOwNS6cPUgpM8K3VdAwTpWpxKRtCZ/Tei/BYq1AHsUrBsNM5vC7fNWJxORFEJFvKQdZzdDQAMIuQSZCkHPNeacziIiVvDICK1nmN35XNLDuS0woRocWGB1MhFJAVTES9pwcBH88JLZFz53RXMO+Ax5rU4lImmdzWZ25+u3CXKVh/BgWNgTFvY2369ERB5CRbykftu+hwXdIToCijYx+6J6ZLQ6lYjI3zLmh+6/Qc33wOYAB34yW+XPbbM6mYgkUyriJfWy2+G3IbDqfXO7Qh9oMwuc3a3NJSISH0cnqD0EeqwCn7wQfB5mNIJ1H0J0pNXpRCSZUREvqVNkGCzsAdvHm9t1R0HDz8DB0dpckmxERNnZdOI6YZGaEUSSmdwVoN9meK4DGHbY9DlMqw9Bp6xOJiLJiIp4SX3u34LZLeHQL+DgDC2nQrWBmgNeYhn00346T9tJp6k7CL6vVk5JZty84KUJ0Go6uHnD5b0wsTrsnQWGYXU6EUkGVMRL6nL7AgS8aM7y4OoFnRZCqdZWp5JkZuPx6yz7KxCA3edu0W7ydq7dCbM4lUg8SrSE/lvBrzpE3oMlA2B+Jwi9aXUyEbGYinhJPa4cNBdNuX4UPLND95XmXMwi/xAWGc3wXw8C0KhkNnzTu3IkMITWE7dx4WaoxelE4uGdC7osgXqjzV8Xjy6DCVXg1O9WJxMRC6mIl9Th9B8wvSHcCYTMRc054LOVsDqVJEOT/jjN2aBQsnq58unLpVjQrzK5MrhzLiiUVhO3cvzqHasjisTl4ABV34Bea811Lu4EmtPm/va+OQZIRNIcFfGS8v31M8x+GcJDIG816PEb+OS2OpUkQ2dv3OP7DScB+KBJMTzdnPHzTceCflUonDU9V0PCaTNpG/vO37I4qchD5CgNfTdCuZ7m9vbvYWodczVqEUlTVMRLymUYsPlrWNQL7JFQ/CXovAjcM1idTJIhwzAYvuQQEVF2qhfypXHJ7DHXZfN246e+lSmd24fboZF0nLqDzSduWJhW5BFcPKDJl9B+Pnj4wtWDMKkmbJ+oQa8iaYiKeEmZ7NGw8l1YO8LcrvQqvBwATq7W5pJk67eDV9h4/Doujg6MalYc279mK/LxcGFOr4pUK+hLaEQ0PWbsYuWBQIvSijyGIi/CK9ugYD2IDoffBsOcVnDnqtXJROQZUBEvKU/kffipC+ycDNigwRh4cYzZZ1QkHnfDoxi19DAA/WoVIH/m9PEel87ViWndytGwRDYiou28+uNe5u86/yyjiiRM+izQ8WdoOA6c3ODkWphQGY6usDqZiDxlqnokZQm9CbOam7MzOLpAqwCo/KrVqSSZ+2btca6EhJEnowev1CrwyGNdnRz5rkMZ2pbLjd2AwQsPMOkPLbIjyZjNBhX7QJ8NkLUkhAbBvPawdCBE3LM6nYg8JSriJeW4dc5ctfDCDnPxk86/mHMoizzC0SshBGw5C8Co5sVxc/7vVXsdHWx88nJJ+tbMD8DYlUf5ZOVRDPU3luQsiz/0XgdVBpjbe6abfeUv77M2l4g8FSriJWUI/NOcAz7oBHjlgh6rwK+a1akkmbPbDYb9cpBou0HDEtmoXSTLY9/WZrMxpKE/7zUsCsDEP07x/i8HiLarkJdkzMkV6n8EnReb62UEnYCpdWHTl+ZYIhFJNVTES/J3ci1MbwR3r0KW4tBrjdniJPIfFuy9yO5zt/BwceSDJsUSdR/9ahbgk5YlcbDB3J0XeH3uPsKjVAxJMlegtrnSq38zsEfBulEws5m5qrWIpAoq4iV52/8j/NgWIu5CvhrQYyV45bA6laQAt+5FMHaFOXf2wLqFyOHjnuj7alchD991KIOzo43lBwLpNXM398KjkiqqyNPhkRHazIJm34FzOji3GSZUhQMLrE4mIklARbwkT4YBG8fB4v5mK1LJ1tBxodkXXuQxfLbqGLdCIymS1ZPuVfM98f01KpmdgG7l8XBxZNOJG3SatoPboRFJkFTkKbLZoExn6LcJcpaD8GBY2BMW9YWwEKvTicgTUBEvyU90FCx7E37/yNyuOhBemgxOLpbGkpRj7/lbzN1pTg350UslcHZMmre66oUyM7tXRbzdndl3/jZtJ23naoiWvJcUIFMBczXrGu+CzQH+mgcTq8L57VYnE5FEUhEvyUtEKMzvZM6qgM2c+7jeKM0BL48tKtrOsF8OAtCqbC7K+2VM0vsvkycDP/WtTBZPV45dvUOriVs5F6Rp/CQFcHSGF4ZC95Xgkwdun4fpDeH3jyE60up0IpJAqowk+bh3A2Y2heMrzUVL2swy5z4WSYAftp/jcGAI3u7ODPn/mWWSWpFsnizsX4W8mTy4cPM+rSZu40iguiZICpGnEvTbAs+1B8MOGz+DgBchSOshiKQkKuIlebh52pxC8tJucM8AXX6FYs2sTiUpzLWQML5YfRyAwS8WJVN616d2rtwZPfi5X2WKZvPk+p1w2k7axu6zN5/a+USSlJsXvDTRXDDP1dt8751YHfb+YI5JEpFkT0W8WO/SHphazyzkvfNAj9VmS5FIAn24/Ah3w6N4LrcP7crnfurny+Lpxvw+lSmXNwMhYVF0mraDDceuPfXziiSZEi9D/y2QtxpE3oMlr8FPXczVsUUkWUtUET9+/Hjy5cuHm5sbZcuWZdOmTY88/vvvv8ff3x93d3eKFCnCrFmzHnrsvHnzsNlstGjRIjHRJKU5vhpmNIHQG5CtlDkHfObCVqeSFGjziRss/fMyDjb4uEUJHBxsz+S83h7O/NCzIrWKZCYs0k6vmbtZ8uflZ3JukSThkxu6LoG6I8HBCY4sgQlV4NR6q5OJyCMkuIifP38+AwcOZOjQoezbt4/q1avTsGFDzp8/H+/xEyZMYMiQIYwcOZJDhw4xatQoXn31VZYuXRrn2HPnzvH2229TvXr1hD8SSXn2zoK57SAyFAq8AN1XgGc2q1NJChQeFc3wX83BrF0q+1Ei57OditTdxZHJncvR9LkcRNkN3pi3j9nbzz3TDCJPxMERqr0JvdZCpkJwJxB+aAGrhkJUuNXpRCQeNsNIWOe3ihUrUqZMGSZMmBCzz9/fnxYtWjB27Ng4x1epUoWqVasybty4mH0DBw5k9+7dbN68OWZfdHQ0NWvWpHv37mzatInbt2+zePHix84VEhKCt7c3wcHBeHl5JeQhybMUFQEnVsO+2eYAVjAHVzX71pw5QSQRvl13gi/WHCezpyvr3qqJl5s1z6Vou8GIJQeZvd1s1HinQRFeqVUAm+3Z/CogkiQi7sHqYbA7wNzOWhJenqKVsiVJqF5LOglqiY+IiGDPnj3Ur18/1v769euzdevWeG8THh6Om5tbrH3u7u7s3LmTyMi/p7QaPXo0mTNnpmfPno+VJTw8nJCQkFgXScYC/4KV78GXRWF+x78L+OpvQ4sJKuAl0c4HhfLd+pMADGvsb1kBD+DoYOPD5iUY8EJBAMatOsbHy4+QwLYSEWu5pIMmX0G7ueCRCa4egMm1YMdkDXoVSUYSVMTfuHGD6OhosmbNGmt/1qxZuXLlSry3adCgAVOnTmXPnj0YhsHu3bsJCAggMjKSGzduALBlyxamTZvGlClTHjvL2LFj8fb2jrnkzv30B7FJAt27AdsnwMRqMKk67JgAoUGQPitUeR1e2QF1PjBXFBRJBMMwW77Do+xULZiJZs/lsDoSNpuNt+oXYVhjs9Vy6uYzvLvgL6Ki7RYnE0mgoo2g/zYoWBeiwmDlOzCnNdy5anUyEQGcEnOjf/80bBjGQ38u/uCDD7hy5QqVKlXCMAyyZs1Kt27d+Oyzz3B0dOTOnTt06tSJKVOm4Ovr+9gZhgwZwqBBg2K2Q0JCVMgnB9GRcGIN7J8Dx1eB/f9/bXF0gSKNoHRHs/+7Y6KeeiKxrD58lfXHruPsaGN08xLJqttKr+r58XZ3ZvDCv/h5z0WC70fyv/bP4+bsaHU0kcfnmRU6LoCdk2H1B3ByjTnotfn3UORFq9OJpGkJqqR8fX1xdHSM0+p+7dq1OK3zD7i7uxMQEMCkSZO4evUq2bNnZ/LkyXh6euLr68tff/3F2bNnadq0acxt7HazxcrJyYljx45RoECBOPfr6uqKq+vTmwNaEujqIdg3B/6ab84080CO583CvcTL4JG0K2dK2nYvPIpRSw4B0LdGAQpkTm9xorhal8uNl7szA37cx+rDV+kxYxeTu5Qjvau+xEoKYrNBxb7gVx0W9YarB2FuWyjXE+p/BC4eVicUSZMSNbC1bNmyjB8/PmZfsWLFaN68ebwDW+NTs2ZNcubMyY8//khYWBgnT56Mdf2wYcO4c+cO33zzDYULF8bFxeU/71MDJSwQehMO/Gy2ugf++ff+dJmhVFso3QGyFrcun6RqY1ceYdIfp8mVwZ01b9bE3SX5tnBvPXWD3jN3cy8imlK5vJnRvQIZ0/33+5pIshMZBr9/CNu+M7d9C0PLKZCjtKWxJOVQvZZ0EtwcNGjQIDp37ky5cuWoXLkykydP5vz58/Tr1w8wu7lcunQpZi7448ePs3PnTipWrMitW7f48ssvOXjwIDNnzgTAzc2NEiVKxDqHj48PQJz9kgxER8HJtWbhfmzl391lHJzNn1ZLd4KCdTRQVZ6q41fvMG3TGQBGNSuerAt4gCoFfJnbpxJdA3by18VgWk/cyg89K5LDx93qaCIJ4+wGDT423+d/6Q83jsPUuvDCMKgywJyqUkSeiQQX8W3btiUoKIjRo0cTGBhIiRIlWLFiBXnz5gUgMDAw1pzx0dHRfPHFFxw7dgxnZ2dq167N1q1b8fPzS7IHIc/AtSNm4f7nfLj3jxUps5WC5ztBiVaQLpN1+STNMAyDYYsPEmU3qF8sK3X84+/Kl9yUyuXDz/0q03naTk5dv0fridv4oWcF8ifDbkAi/6nAC/DKNlgyAI4ug7UjzAaelyaCdy6r04mkCQnuTpNc6eeZpyD0JhxcaBbvl/f9vd/D9/+7y7SHbCWtyydp0sI9F3nr5z9xd3ZkzaAa5MqQsvrjXrp9n85Td3D6xj0ypXNhZo8Kz3xxKpEkYxiw7wdYOdhcuM/NG5p8DSVaWp1MkinVa0lHRbzEFh0Fp37//+4yKyA6wtzv4ASFXzT7uReqr+4yYong0Ehe+GIDQfciGPxiUfrXijvoPSW4cTecrgE7OXQ5BE9XJ6Z2LUfF/PolS1KwoFOwsBdc3mtuP9ceGn4Gbvo8lthUryUdFfFiun7s7+4yd/8x+1DWEubsMqXaQLrHnwJU5GkY+ssB5uw4T6Es6Vn+enVcnBK01EWyEhIWSa+Zu9l55iauTg6M71gmxXQNEolXdCT88Sls+gIMO/jkNQe95qlodTJJRlSvJR0V8WnZ/VtwcJFZvF/a8/d+j0xQso3Z6p69lHX5RP5h/4XbvDR+C4YB8/pUolIqaLkOi4zm1Tl7WXf0Go4ONr5o/Rwtns9pdSyRJ3NuGyzqA8HnweYANd6FGu9ofRABVK8lJRXxaY09Gk6vN+d0P7ocosPN/TZHKNzg/7vLNAAnTX8nyUe03aD595s5eCmEls/n5Mu2pa2OlGQio+28u+Avftl3CYCRTYvRrWo+i1OJPKGwYFjxjrl2CECu8tByMmTMb20usZzqtaSjr8VpxY0T/99dZh7cCfx7f5Zif3eXSZ/FunwijzBnxzkOXgrBy82JIY38rY6TpJwdHfii9XN4uzszY+tZRi49zO37kbxRp1CyWoFWJEHcvM2ivVB9WDYILu6CidXNfvKlO5gLSInIE1ERn5qFBf9/d5kf4eLOv/e7Z4CSrc3iPftzejOVZO3anTDG/XYMgHdeLEpmz9S3UrODg40RTYuRwcOFr9Ye5+u1J7gdGsnwJsVwcNDrU1Kwkq0gdwX4pR+c2wK/vgInVpkz2GgVb5EnoiI+tbFHw5k/zML9yFKICjP32xyhUD2zBaTwi+CU+gohSZ3GLD/CnfAoSuXypkOFPFbHeWpsNhtv1C2Ej4czI5YcYsbWswTfj+SzVqVwdky5A3hF8MkDXZfClq9h/Rg4/Ctc2GXOKZ+/ptXpRFIsFfGpRdCpv7vLhFz6e3/mon93l/HMZl0+kUTYeuoGi/dfxmaDj1qUwDENtEp3reKHt7szb/38J7/su0TI/Ui+71gGN2ethCkpmIMjVH8L8teGRb0h6CTMag5VXoMXPlDDkkgiaGBrShYWAod+MVvdL2z/e7+b99/dZXI8r+4ykiJFRNlp+M1GTl2/R5fKeRndvITVkZ6pdUeu8sqcvYRH2amQLyNTu5bDy03rM0gqEHEPVr0Pe2aY29lKwsvTIHMRS2PJs5Em67WnREV8SmO3w9mNZuF+eAlE3Tf32xygQB14viMUbgjObtbmFHlC368/ybhVx/BN78K6t2rh7Z72Ctgdp4PoNXM3d8KjKJ7Di5k9KuCbXi2WkkocXQ6/vgb3b4KTG9T/CMr3UsNTKpdm6rVnQEV8SnHzNOyfC3/OheALf+/3Lfz/3WXagld26/KJJKELN0Op99UfhEXa+artc7z0fC6rI1nm4KVgugbsJOheBPl90zGrZwVyZfCwOpZI0rhzBRa/AqfWmduFGkDz7zRbWiqW6uu1Z0hFfHIWfsccALRvDpzf+vd+V28o+bJZvOcsq1YLSXV6zdzN2iNXqZQ/I3N7V0rzUy2evn6XztN2cun2fbJ7u/FDzwoUzOJpdSyRpGG3w87JsGa4uXZJuszQ/Htz7RJJdVJlvWYRFfHJjd0O5zb/f3eZXyEy9P+vsEGBF8zZZYo2Bmd3S2OKPC1rDl+l96zdODnY+G1gdRWr/y8w+D6dpu7g1PV7ZPBwZmaPCpTK5WN1LJGkc/UQLOwN1w6Z2+V7Qb0PwUW/PKUmqaZeSwZUxCcXt87+f3eZH+H2+b/3Zyr4d3cZby3HLqlbaEQU9b7cyKXb9+lfqwCDXyxqdaRk5ea9CLpN38lfF4NJ5+LIlK7lqFLA1+pYIkknMgzWjYLt481t3yLw8lTIXsraXJJkUny9loyoiLdS+F04ssTsLnNu89/7Xb2gREuzeM9VXt1lJM347LejjN9wipw+7qwZVAMPF82C+293w6PoM2s3W08F4eLkwLftn6dBcU0fK6nMyXWwuD/cvQoOzlDnA6g8ABy0ZkJKlyLrtWRKRfyzZhhwbqs5p/uhxRB57/+vsEH+WmbhXrSxfj6UNOfktTs0/GYTkdEGU7qUo16xrFZHSrbCIqN5Y94+Vh26ioMNPn25FK3L5bY6lkjSuhcES1+Ho8vM7Xw1oMVE/SqdwqWYei0FUBH/rNw6Zy7E9OePZteZBzLmN/u5P9cevNPuDBySthmGQfsp29l++iZ1/bMwtWt5qyMle1HRdoYsOsDPey4CMKyxP72q57c4lUgSMwzYOwt+e88cI+bmA02/huIvWZ1MEinZ12spiH6rfpoi7sGRpWar+5mNf+938YTiLeD5TpC7orrLSJr36/7LbD99EzdnB0Y0LW51nBTBydGBz1qVwsfDmSmbzvDR8iPcDo3krfqF0/xsPpKK2GxQtivkrQqLesHlffBzNzixBhp+Cq4a+C5pl4r4pGYYcH77391lIu78fV2+mmZ3Gf8m4JLOsogiyUnw/Ug+Wn4EgAEvFCJ3RnUle1w2m433G/nj4+HCuFXH+G79SW6FRjC6eQkcHVTISyriWxB6roENY2HTl+Zn7Lkt0HIK5K5gdToRS6iITyq3L/zdXebm6b/3Z/AzC/fn2oFPHsviiSRXX6w+xo274RTInI7e6g6SYDabjVdrF8Tb3ZkPfj3InB3nCb4fyZdtSuPipEGAkoo4OkOd4VCwLizqY3ZNDXgRar4L1d8GR5U0kraoT/yTiAg1B9zsnwOn/wD+/0/pnM7sr/d8R8hTWd1lRB7iwMVgmn2/GcOAH3tVpEpBTZf4JJb+eZk35+8nym5Qs3BmJnYqi7uLo9WxRJLe/duw4m048LO5nasCtJwMGfNZGkv+m/rEJx0V8Ylx8zRs/hoO/QLhIX/v96v+/91lmoJr+qebQSSFi7YbtBy/hT8vBtOidA6+bve81ZFShQ3HrtFv9h7CIu2UzZuBgK7l8fZwtjqWyNPx10+w/C3zs9jFExp9Zk4UocazZEtFfNLRb62JER0Je2eabxo+eaDWEHjjT+i2DEq3VwEv8hh+3HmePy8G4+nqxPuN/a2Ok2rUKpKF2T0r4uXmxJ5zt2g7eRvX7oRZHUvk6SjVBvptNn/1jrhjzi3/cze4f8vqZCJPnVriE2vDJ+Zo+bxVtfiESAJdvxPOC19s4E5YFKOaFadrFT+rI6U6RwJD6BKwk+t3wsmbyYPZPStq0LCkXvZo2PyVOfDVHgVeOeGliebc8pKsqCU+6aj6TKxa70G+6irgRRJh7Moj3AmLokROLzpVymt1nFTJP7sXC/pVJndGd84FhfLyhK0cu3Lnv28okhI5OEKNt6HnashYAEIuwcxmsGY4REVYnU7kqVAFKiLP1PbTQSzaewmbDT5qUVJTIT5FeTOlY0G/KhTJ6sm1O+G0mbSNvefVzUBSsZxloe9GKNMFMGDLNzC1Dlw/ZnUykSSnIl5EnpmIKDsfLD4IQIcKeSid28faQGlAVi835vetxPN5fAi+H0nHKTvYdOK61bFEnh7X9NDsW2g7G9wzwpW/YFJN2DXVXMtFJJVQES8iz0zAljOcuHaXTOlceLdBUavjpBk+Hi7M6VWR6oV8uR8ZTY8Zu1hxINDqWCJPl39T6L8V8teGqPvmLDZz28FdfYmV1EFFvIg8E5du3+ebtScAeL+Rv6Y9fMY8XJyY2rUcjUtmJzLa4LUf9zJ353mrY4k8XV7ZodMiaDAWHF3g+G8woTIcX211MpEnpiJeRJ6JUUsOcT8ymgp+GWlZJqfVcdIkVydH/tf+edpXyI3dgCGLDjBhwymrY4k8XQ4OUPkV6L0eshSDe9fhx9aw/G2IvG91OpFEUxEvIk/duiNXWX34Kk4ONj5sUQKbFmKxjKODjTEvlaR/rQIAfPrbUcauPEIqmW1Y5OGylTAL+Yr9ze1dU2ByLQj8y9JYIomlIl5Enqr7EdGMWHIIgJ7V8lEkm6fFicRmszH4xaIMaWiOS5j0x2mGLDpAtF2FvKRyzm7Q8BPotBDSZ4XrR83Za7Z+C3a71elEEkRFvIg8Vd+vP8nFW/fJ4e3G63UKWR1H/qFvzQJ8+nJJHGwwb9cFBszdS3hUtNWxRJ6+gnXNQa9FGkF0BKweBj+0gJDLVicTeWwq4kXkqTl1/S6TNpp9roc3LU46VyeLE8m/tS2fh/Edy+Di6MCKA1foNXM398KjrI4l8vSl84V2P0KTr8HZA878AeMrw+FfrU4m8lhUxIvIU2EYBsN/PUhktEHtIplpUDyr1ZHkIV4skZ2AbuXxcHFk04kbdJy6g1v3tMqlpAE2G5Trbi4Qlb00hN2Gn7rAr69C+F2r04k8kop4EXkqlv4VyJaTQbg6OTCqmQazJnfVCvkyp1dFfDyc2X/hNm0mbeNKcJjVsUSeDd9C0HMNVBsE2GDfbJhYDS7utjqZyEOpiBeRJBcSFsmHyw4D8FrtguTJ5GFxInkcz+fJwE99K5PVy5UT1+7SauJWzt64Z3UskWfDyQXqjoBuy8ArF9w6A9Pqw4ZPIVpdzCT5UREvIknuy9XHuX4nnHy+6ehTM7/VcSQBCmf1ZEG/Kvhl8uDirfu0mriNw5dDrI4l8uz4VYP+W6DEy2BEw4YxMKMR3DprdTKRWFTEi0iSOngpmFnbzgLwYfMSuDo5WhtIEix3Rg9+7lcF/+xe3LgbTtvJ29h99qbVsUSeHXcfeHkavDQZXDzhwg6YUA3+nAdaU0GSCRXxIpJk7HaDYYsPYjeg6XM5qFbI1+pIkkiZPV2Z16cS5fJm4E5YFJ2m7WD9sWtWxxJ5dmw2eK4t9N8MuStBxB34pS8s6AH3b1mdTkRFvIgknXm7LrD/wm3SuzoxrLG/1XHkCXm7O/NDz4rUKpKZsEg7vWfu5tf9l6yOJfJsZfCDbsuh9jCwOcKhRWar/JlNVieTNE5FvIgkiaC74Xz621EABtUrTFYvN4sTSVJwd3FkSpdyNC+dgyi7wcD5+/lh+zmrY4k8W45OUPMd6LkaMuSDkIswsymsHQlRmo5VrKEiXkSSxCcrjxJ8P5Ji2b3oUjmv1XEkCTk7OvBVm9J0qZwXw4APFh/k23UnMNQ3WNKaXOWg32Z4vjNgwOavYFpduH7c6mSSBqmIF5EntuvsTX7ecxGAj14qgZOj3lpSGwcHG6OaFef1FwoC8MWa43y47Ah2uwp5SWNc00Pz76DND+CeAQL/hEk1YNc0DXqVZ0qftCLyRCKj7Qz75SAA7SvkpkyeDBYnkqfFZrMxqH4RPmhSDICALWd4Z8FfREXbLU4mYoFizaD/VshfC6Luw/JBMLc93LthdTJJI1TEi8gTmbHlLMeu3iFjOhfebVDU6jjyDPSslo8vWj+Ho4ONhXsv0n/OXsIio62OJfLseeWATr9A/Y/B0QWOr4TxleHEWquTSRqgIl5EEu3y7ft8tdbsC/pew6JkSOdicSJ5Vl4um4uJncri4uTAmsNX6T59F3fCIq2OJfLsOThAldeg9++QuSjcuwZzXoYV70LkfavTSSqmIl5EEu3DZYcJjYimXN4MtCqTy+o48ozVK5aVmd0rkN7ViW2ng+gwZQdBd8OtjiVijWwloc8GqNDX3N45CSbXhisHLI0lqZeKeBFJlPXHrrHy4BUcHWx82KIEDg42qyOJBSoXyMTc3pXImM6FA5eCaTNpG5dvq/VR0ihnd2j0GXRcAOmywPUjMOUF2PY92DV2RJKWingRSbCwyGhG/HoIgB5V/fDP7mVxIrFSyVze/NS3Mjm83Th1/R6tJmzl1PW7VscSsU6heuag18INIToCVr0Ps1+CkECrk0kqoiJeRBJs/IZTnL8ZSjYvN96oW9jqOJIMFMySnp/7VyF/5nRcDg6jzcRtHLwUbHUsEeukzwzt50LjL8HJHU5vgAmV4fASq5NJKqEiXkQS5MyNe0zccAqA4U2Lkd7VyeJEklzk9HHn576VKZHTi6B7EbSbvJ3tp4OsjiViHZsNyveEvhsh+3Nw/xb81Bl+fQ3C9WuVPBkV8SLy2AzDYPivB4mItlOzcGYalshmdSRJZjKld2Vu70pUzJeRu+FRdAnYyZrDV62OJWKtzIWh51qoOhCwwb4fYFJ1uLjH6mSSgqmIF5HHtvxAIJtO3MDFyYFRzYpjs2kwq8Tl6ebMzB4VqOuflYgoO/1m72HR3otWxxKxlpML1BsFXZeCV064eRqm1YM/xoFd6yxIwqmIF5HHcicsktFLDwPwSq0C+PmmsziRJGduzo5M7FSGlmVyEm03GPTTnwRsPmN1LBHr5asO/bdA8ZZgRMP6j2B6I7h1zupkksKoiBeRx/L12hNcuxOOXyYP+tUsYHUcSQGcHB34vNVzdK/qB8DoZYf5cs1xDMOwNpiI1dwzQKsAeGkSuHjChe0wsRr89ZPVySQFUREvIv/p8OUQZmw9C8Do5iVwc3a0NpCkGA4ONoY3KcageuYsRv9bd4KRSw5ht6uQlzTOZoPn2kH/zZC7IoSHwKLesKAn3L9tdTpJAVTEi8gj2e0GwxYfINpu0LhkdmoUzmx1JElhbDYbr9cpxOjmxQGYue0cg37aT2S0Fr8RIYMfdFsBtd4HmyMcXGC2yp/dYnUySeYSVcSPHz+efPny4ebmRtmyZdm0adMjj//+++/x9/fH3d2dIkWKMGvWrFjXL1q0iHLlyuHj40O6dOkoXbo0P/zwQ2KiiUgS+3nPBfaev006F0c+aFLM6jiSgnWp7Mc37Urj5GBj8f7L9P1hD2GRGtAngqMT1BoMPVZBhnwQfAFmNIa1oyAqwup0kkwluIifP38+AwcOZOjQoezbt4/q1avTsGFDzp8/H+/xEyZMYMiQIYwcOZJDhw4xatQoXn31VZYuXRpzTMaMGRk6dCjbtm3jr7/+onv37nTv3p1Vq1Yl/pGJyBO7eS+CsSuPAvBmvcJk83azOJGkdM1L52RKl3K4Ojnw+9FrdJm2k5CwSKtjiSQPuctDv01QuhNgwOYvzRlsbpywOpkkQzYjgSOMKlasSJkyZZgwYULMPn9/f1q0aMHYsWPjHF+lShWqVq3KuHHjYvYNHDiQ3bt3s3nz5oeep0yZMjRu3JgPP/zwsXKFhITg7e1NcHAwXl5aAl4kKQxe8Bfzd1+gaDZPlg2ohpOjeuBJ0th19iY9ZuziTlgUxbJ7MbNHBTJ7ulodSyT5OLQYlr4BYbfB2QMajIGy3cy+9CmY6rWkk6BP5IiICPbs2UP9+vVj7a9fvz5bt26N9zbh4eG4ucVuvXN3d2fnzp1ERsZtfTEMg3Xr1nHs2DFq1Kjx0Czh4eGEhITEuohI0tlz7ibzd18A4KMWJVTAS5Iq75eReX0q4ZvehcOBIbSZtI0LN0OtjiWSfBRvAf23Qr4aEBkKywbCvI5w74bVySSZSNCn8o0bN4iOjiZr1qyx9mfNmpUrV67Ee5sGDRowdepU9uzZg2EY7N69m4CAACIjI7lx4+8nYnBwMOnTp8fFxYXGjRvz7bffUq9evYdmGTt2LN7e3jGX3LlzJ+ShiMgjREXbGfrLQQDalMtFOb+MFieS1Kh4Dm9+7leFnD7unLlxj9YTt3Hi6h2rY4kkH945ofOvUP8jcHCGY8thQhU4udbqZJIMJKpp7d+rNBqG8dCVGz/44AMaNmxIpUqVcHZ2pnnz5nTr1g0AR8e/p6nz9PRk//797Nq1i48//phBgwaxYcOGh2YYMmQIwcHBMZcLFy4k5qGISDxmbD3L0St38PFw5r2G/lbHkVQsn286FvavQqEs6bkSEkbrSdvYf+G21bFEkg8HB6gyAHr/Dr5F4O5VmP0yrHwPIsOsTicWSlAR7+vri6OjY5xW92vXrsVpnX/A3d2dgIAAQkNDOXv2LOfPn8fPzw9PT098fX3/DuLgQMGCBSldujRvvfUWrVq1ireP/QOurq54eXnFuojIk7sSHMZXa44D8N6LRcmYzsXiRJLaZfN246e+lXkutw+3QyPpMGU7W06qy4BILNlLQZ8NUL63ub1jAkypDVcPWRpLrJOgIt7FxYWyZcuyZs2aWPvXrFlDlSpVHnlbZ2dncuXKhaOjI/PmzaNJkyY4ODz89IZhEB4enpB4IpIEPlx+mHsR0ZTJ40ObcuqmJs9GhnQuzOlVkaoFMxEaEU336bv47WD83TRF0iwXD2j8OXT4GdJlhmuHYXIt2DYe7Fp3Ia1JcHeaQYMGMXXqVAICAjhy5Ahvvvkm58+fp1+/foDZzaVLly4xxx8/fpzZs2dz4sQJdu7cSbt27Th48CBjxoyJOWbs2LGsWbOG06dPc/ToUb788ktmzZpFp06dkuAhisjj2nj8Osv/CsTBBh+1KImDQ8qeBUFSlvSuTgR0K8+LxbMREW3nlTl7+GmXukqKxFG4PvTfBoUaQHQErBoCs1tCSKDVyeQZckroDdq2bUtQUBCjR48mMDCQEiVKsGLFCvLmzQtAYGBgrDnjo6Oj+eKLLzh27BjOzs7Url2brVu34ufnF3PMvXv3eOWVV7h48SLu7u4ULVqU2bNn07Zt2yd/hCLyWMIioxn+qzmYtVuVfBTLoS5q8uy5OjnyXYfnGfrLQebvvsC7C/8i+H4kvWvktzqaSPKSPjN0mA+7p8GqoXB6vTnotdm34N/E6nTyDCR4nvjkSvOOijyZb9ae4Ku1x8ni6cq6t2ri6eZsdSRJwwzD4JOVR5m08TQAr9YuwNv1izx0EgWRNO36MVjYC678ZW6X6QINxoJremtzxUP1WtLRxM8iwrmge3y/4SQAHzQppgJeLGez2RjSyJ/BLxYF4Pv1pxi6+CDR9lTR7iSStDIXgV7roOobgA32zoJJNeDSHquTyVOkIl4kjTMMg+G/HiIiyk71Qr40KZXd6kgiMfrXKsCYl0pis8GPO87zxrx9RERpAJ9IHE4uUG80dF0Cnjng5imYVh82fg72aKvTyVOgIl4kjfvt4BX+OH4dF0cHRjUrru4Kkux0qJiHb9s/j7OjjWV/BdJr1m5CI6KsjiWSPOWrAf23QLEWYI+C3z+EGU3g9vn/vKmkLCriRdKwu+FRjFp6GIB+NfOTP3Py6z8pAtCkVA6mdi2Pu7MjG49fp9PUHQSHRlodSyR58sgIrWdAiwngkh7Ob4UJVeGvn61OJklIRbxIGvbN2uNcCQkjT0YPXqld0Oo4Io9Us3BmZveqiJebE3vP36bt5G1cC9GKlSLxstmgdAfotwlylYfwEFjUyxwAGxZsdTpJAiriRdKoo1dCCNhyFoBRzYvj5uxobSCRx1A2bwZ+6leZzJ6uHL1yh1YTt3E+KNTqWCLJV8b80P03qDUEbA5w4GeYUA3ObbU6mTwhFfEiaZDdbjDsF3OmjxeLZ6N2kSxWRxJ5bEWzebGwXxXyZPTg/M1QWk3cytErIVbHEkm+HJ2g1nvQYxVk8IPg8zCjMawbDdHqlpZSqYgXSYMW7r3I7nO38HBxZHjTYlbHEUmwPJk8WNCvMkWzeXLtTjhtJm5jz7lbVscSSd5yV4B+m6F0RzDssOkLcwaboFNWJ5NEUBEvksbcuhfB2JVHARhYtxA5fNwtTiSSOFm83JjfpzJl8vgQEhZFp6k72Hj8utWxRJI3V09oMd4c+OrmDZf3wpLXrU4liaAiXiSN+WzVMW7ei6Bw1vR0r5rP6jgiT8Tbw5nZvSpSo3Bm7kdG03PmLpb/FWh1LJHkr/hL0H8rFGoATb+2Oo0kgop4kTRk7/lbzNtlzhX8UYuSODvqLUBSPg8XJ6Z2KUeTUtmJjDZ4be5eftyhObFF/pN3Luj4E/gWsjqJJII+wUXSiKhoO8N+OYhhQKuyuaiQL6PVkUSSjIuTA9+0e54OFfNgGPD+LwcYv+EkhmFYHU1E5KlQES+SRvyw/RyHA0PwdndmSMOiVscRSXKODjY+blGCV2sXAOCz344xduVRFfIikiqpiBdJA66FhPHF6uMAvPtiETKld7U4kcjTYbPZeKdBUYY28gdg8sbTDF74F1HRdouTiYgkLRXxImnAR8uPcDc8iudy+9C+fB6r44g8db1r5Oezl0vhYIOfdl/ktR/3ER4VbXUsEZEkoyJeJJXbfOIGS/68jIMNPm5RAgcHm9WRRJ6JNuVzM75jGVwcHfjt0BV6zNjF3fAoq2OJiCQJFfEiqVh4VDTDfz0IQJfKfpTI6W1xIpFn68US2ZnevTzpXBzZcjKIjlN3cOtehNWxRESemIp4kVRsysbTnL5xj8yergyqX9jqOCKWqFrQlx97VyKDhzN/XrhNm0nbuBIcZnUsEZEnoiJeJJU6HxTKt7+fBGBYY3+83JwtTiRinedy+/BT38pk83LjxLW7vDxhK2du3LM6lohIoqmIF0mFDMNgxJKDhEfZqVIgE82ey2F1JBHLFcrqyYL+lcnnm45Lt+/TeuJWDl0OtjqWiEiiqIgXSYVWH77K+mPXcXa0Mbp5CWw2DWYVAciVwYOf+lamWHYvbtyNoN2k7ew8c9PqWCIiCaYiXiSVCY2IYtSSQwD0rVGAglnSW5xIJHnJ7OnKvL6VqOCXkTvhUXSetoPfj161OpaISIKoiBdJZb5Zd4LLwWHkyuDOq7ULWh1HJFnycnNmZo8KvFA0C+FRdvrM2sOv+y9ZHUtE5LGpiBdJRY5fvcO0TWcAGNWsOO4ujhYnEkm+3F0cmdS5LM1L5yDKbjBw/n5mbTtrdSwRkceiIl4klTAMg2GLDxJlN6hXLCt1/LNaHUkk2XN2dOCrNqXpWjkvhgHDfz3E/9adwDAMq6OJiDySiniRVGLR3kvsPHMTd2dHRjQtZnUckRTDwcHGyGbFeaNOIQC+XHOc0csOY7erkBeR5EtFvEgqEBwayZgVRwB4vU4hcmXwsDiRSMpis9l4s17hmC/A07ec5e0FfxIVbbc4mYhI/FTEi6QC41YfJeheBIWypKdntXxWxxFJsbpXzceXbZ7D0cHGor2X6Dd7L2GR0VbHEhGJQ0W8SAq3/8Jt5uw4D8CHLUrg4qSXtciTaFkmF5M6lcXFyYG1R67SbfpO7oRFWh1LRCQWfdqLpGDRdoNhiw9gGNDy+ZxUyp/J6kgiqULdYlmZ1aMC6V2d2H76Jh2m7CDobrjVsUREYqiIF0nB5uw4x8FLIXi6OTGkkb/VcURSlUr5MzGvTyUypXPhwKVgWk/axqXb962OJSICqIgXSbGu3Qlj3KpjALzboAiZPV0tTiSS+pTI6c1P/SqTw9uN09fv0WrCVk5eu2t1LBERFfEiKdWY5Ue4ExZFqVzedKiY1+o4IqlWgczpWdC/CgUypyMwOIw2k7Zx4GKw1bFEJI1TES+SAm09dYPF+y9js8FHLUrg6GCzOpJIqpbDx52f+lamVC5vbt6LoP2U7Ww7FWR1LBFJw1TEi6QwEVF2Plh8EIBOFfNSKpePtYFE0ohM6V35sXclKufPxN3wKLpO38maw1etjiUiaZSKeJEUZsqm05y6fg/f9C683aCI1XFE0pT0rk5M716eesWyEhFlp9/sPSzcc9HqWCKSBqmIF0lBLtwM5dvfTwAwtLE/3u7OFicSSXvcnB2Z0LEMrcrmItpu8NbPfxKw+YzVsUQkjVERL5KCjFp6mLBIOxXzZaRF6ZxWxxFJs5wcHfjs5VIxKySPXnaYL1cfwzAMi5OJSFqhIl4khVhz+Cprj1zFycHGRy1KYLNpMKuIlRwcbAxr7M/b9QsD8L/fTzJiySHsdhXyIvL0qYgXSQFCI6IYueQQAL1r5KdQVk+LE4kIgM1m47UXCvFhixLYbDBr2zkGzt9PZLTd6mgiksqpiBdJAb77/SSXbt8np487A14oaHUcEfmXzpXy8nXb0jg52Fjy52X6zNrN/Yhoq2OJSCqmIl4kmTt57Q5TNp0GYETTYni4OFmcSETi07x0TqZ0KYebswPrj12nS8AOgu9HWh1LRFIpFfEiyZhhGAxbfJDIaIO6/lmoXzyb1ZFE5BFqF83CDz0r4unmxK6zt2g3eTvX74RbHUtEUiEV8SLJ2K/7L7P99E3cnB0Y0bS41XFE5DGU98vI/D6V8U3vypHAEFpP3MqFm6FWxxKRVEZFvEgyFXw/ko+WHwFgwAuFyJ3Rw+JEIvK4iuXwYkG/yuTK4M7ZoFBaTdzKiat3rI4lIqmIiniRZOrL1ce4cTec/JnT0at6PqvjiEgC+fmmY0G/KhTOmp6rIeG0nrSN/RduWx1LRFIJFfEiydCBi8H8sP0cAB81L4Grk6PFiUQkMbJ5u/FT38qUzu3D7dBIOkzZzuYTN6yOJSKpgIp4kWQm2m4wbPEB7AY0L52DKgV9rY4kIk/Ax8OFOb0qUq2gL6ER0fSYsYvfDgZaHUtEUjgV8SLJzNyd5/nzYjCerk4MbexvdRwRSQLpXJ2Y1q0cDUtkIyLazitz9jJ/13mrY4lICqYiXiQZuX4nnM9+OwrA2w2KkMXTzeJEIpJUXJ0c+a5DGdqWy43dgMELDzDpj1NWxxKRFEpFvEgyMnblEULCoiiew4tOlfJaHUdEkpijg41PXi5J35r5ARi78iif/nYUwzAsTiYiKY2KeJFkYvvpIBbtvYTNBh+/VBJHB5vVkUTkKbDZbAxp6M97DYsCMGHDKd7/5SDRdhXyIvL4VMSLJAOR0XY+WHwQgA4V8lA6t4+1gUTkqetXswCftCyJg80cC/P63H1ERNmtjiUiKYSKeJFkYNrmM5y4dpdM6Vx4t0FRq+OIyDPSrkIevutQBmdHG8sPBNJz5i5CI6KsjiUiKYCKeBGLXbp9n2/WngBgSCN/vD2cLU4kIs9So5LZCehWHg8XRzaduEGnqTu4HRphdSwRSeZUxItYbPTSQ9yPjKaCX0ZeLpPT6jgiYoHqhTIzu1dFvN2d2Xv+Nm0nbedaSJjVsUQkGVMRL2Kh349eZdWhqzg52PiwRQlsNg1mFUmryuTJwE99K5PF05VjV+/w8sStnAu6Z3UsEUmmVMSLWOR+RDTDfz0EQM9q+SiSzdPiRCJitSLZPFnYvwp5M3lw4eZ9Wk3cxpHAEKtjiUgylKgifvz48eTLlw83NzfKli3Lpk2bHnn8999/j7+/P+7u7hQpUoRZs2bFun7KlClUr16dDBkykCFDBurWrcvOnTsTE00kxRi/4SQXb90nh7cbr9cpZHUcEUkmcmf04Od+lSmazZPrd8JpO2kbe87dtDqWiCQzCS7i58+fz8CBAxk6dCj79u2jevXqNGzYkPPn418+esKECQwZMoSRI0dy6NAhRo0axauvvsrSpUtjjtmwYQPt27dn/fr1bNu2jTx58lC/fn0uXbqU+Ecmkoydun6Xif+/UuPwpsVJ5+pkcSIRSU6yeLoxv09lyubNQEhYFB2n7mDDsWtWxxKRZMRmJHCZuIoVK1KmTBkmTJgQs8/f358WLVowduzYOMdXqVKFqlWrMm7cuJh9AwcOZPfu3WzevDnec0RHR5MhQwa+++47unTp8li5QkJC8Pb2Jjg4GC8vr4Q8JJFnyjAMOk3bwZaTQdQukpmAbuXVF15E4hUaEUX/2Xv54/h1nB1tfNmmNE2fy2F1LJFEU72WdBLUEh8REcGePXuoX79+rP3169dn69at8d4mPDwcNze3WPvc3d3ZuXMnkZGR8d4mNDSUyMhIMmbM+NAs4eHhhISExLqIpARL/wpky8kgXJ0cGNVMg1lF5OE8XJyY0qUcTZ/LQWS0wevz9jFnxzmrY4lIMpCgIv7GjRtER0eTNWvWWPuzZs3KlStX4r1NgwYNmDp1Knv27MEwDHbv3k1AQACRkZHcuHEj3tu899575MyZk7p16z40y9ixY/H29o655M6dOyEPRcQSIWGRfLjsMACv1i5InkweFicSkeTOxcmBr9uWplOlPBgGDP3lIN+vP0kCf0gXkVQmUQNb/91yaBjGQ1sTP/jgAxo2bEilSpVwdnamefPmdOvWDQBHR8c4x3/22WfMnTuXRYsWxWnB/6chQ4YQHBwcc7lw4UJiHorIM/Xl6uNcvxNOPt909K2Z3+o4IpJCODrY+LB5CQa8UBCAcauOMWbFERXyImlYgop4X19fHB0d47S6X7t2LU7r/APu7u4EBAQQGhrK2bNnOX/+PH5+fnh6euLr6xvr2M8//5wxY8awevVqSpUq9cgsrq6ueHl5xbqIJGcHLwUza9tZAEY3L46rU9wvsSIiD2Oz2XirfhGGNfYHYMqmMwxe+BdR0XaLk4mIFRJUxLu4uFC2bFnWrFkTa///tXfn4VHV9x7H3zOTyb5AEkgCCUnYA2GPhAR3LyheFCoKSkVQsWC1ldLb3qIiSLli3Wp7VRARBTdwqSJXXLAqLgmEsBP2sCQsISSQnWwz5/4Bpo2gmJDkzGQ+r+eZ5zHnnBk/4/H38MnhzHdWr15NamrqTz7XbrcTHR2NzWZj2bJljBw5Eqv1X//6J598kj//+c988sknJCUlNSSWiMtzOg0e/mA7TgNG9o3ism7tzI4kIm5q8mWdefLmvlgt8HbmYe57cyOVNQ6zY4lIC2vwXLvp06czYcIEkpKSSElJYeHCheTk5DB16lTgzG0uR44cqZsFv2fPHjIyMkhOTubUqVM888wzbN++nSVLltS95hNPPMHMmTN58803iYuLq7vSHxgYSGBgYFO8TxFTLVufy+bcIgJ9vJg5spfZcUTEzd2SFEOwn53fvLmJT7OOc9er61l4RxKBGlcr4jEafE/8uHHjePbZZ5kzZw79+/fn66+/ZtWqVcTGxgJw7NixejPjHQ4HTz/9NP369WPYsGFUVlaSlpZGXFxc3TEvvPAC1dXV3HzzzURFRdU9nnrqqYt/hyImKyyr4i+f7AJg+rDuRAT/+Gc9RER+rmt7R/LqnZcQ4G0jLbuQX760lpPl1WbHEpEW0uA58a5Kc0fFVf3hnS28s+EwCVHBrLx/KF62Rn2eXETkvLbkFjHplQxOVdTQtX0gr909mKgQP7NjiZyX+lrTUZsQaUbrD57knQ2HAZg7OlEFXkSaXL+YNrwzNYWoEF/25Zdx8/x09p8oMzuWiDQzNQqRZlLjcPLw+9sBuG1wDINi25qcSERaq67tg3j33lQ6hwdwpOg0tyxIZ/uRYrNjiUgzUokXaSavfneQ3cdLaetv54/X9jQ7joi0ch3b+PH21BR6dwimsLya2xauJePASbNjiUgzUYkXaQbHik/z18/3ADBjRAJtA7xNTiQiniA80Ie3fjWEwfGhlFbVMuHldXyx67jZsUSkGajEizSDOSt3UFHtICm2LTcPijY7joh4kGBfO0vvGsw1PdtTVevkV0s38MGmI2bHEpEmphIv0sS+3J3Px9vzznxN+uhErFaL2ZFExMP42m0smDCIXwzoSK3TYNryzSxJO2h2LBFpQirxIk2ossbBrBVZANyZGkdClMZniYg57DYrT9/Sj0mpcQDM+jCLv32+l1YyWVrE46nEizSh+V9lk3OygshgX6YN6252HBHxcFarhVk39GLaf3QD4K+f7+HRlTtwOlXkRdydSrxIEzlQUM78r7IBeOSGXvr6cxFxCRaLhWn/0Z3ZN/QC4NW0g/z+nS3UOJwmJxORi6ESL9IEDMPgkRXbqXY4ubx7O0YkRpodSUSknklD43l2XH9sVgvvbzrCva9voLLGYXYsEWkklXiRJrBqWx7f7C3A28vKnBt7Y7How6wi4npGD+jIwgmD8PGy8vnOfO5YnEFJZY3ZsUSkEVTiRS5SaWUNc/7vzIdZf31lF+LCA0xOJCLy465JiGDpXYMJ8vEi48BJblu4loKyKrNjiUgDqcSLXKRnP9/L8ZIqYsP8mXpFF7PjiIhcUHLnMN761RDCArzJOlrC2AXpHCk6bXYsEWkAlXiRi7DjaAmvnp29/OiNvfG128wNJCLyMyV2DOGdqSl0bOPH/oJybp6fxr78MrNjicjPpBIv0khOp8HDH2zD4TS4vk8kV/Zob3YkEZEG6dwukHfvTaFLuwCOFVcy9sV0th4uMjuWiPwMKvEijfTOhlw25hQR4G3jkZG9zY4jItIoUSF+vDM1lb7RIZwsr+a2hWtJyy4wO5aIXIBKvEgjnCyvZt7HuwD43bDuRIb4mpxIRKTxQgO8efOeIaR2CaO82sGkV9bzaVae2bFE5CeoxIs0whOf7KKoooaekUFMPPuV5iIi7izQx4vFky5heK8Iqmud3Pv6Bt7JzDU7loj8CJV4kQbacOgky9af+YNt7uhE7DYtIxFpHXztNl745UBuGRSN04A/vLuVRd/sNzuWiJyH2odIA9Q6nDz0/nYAxiZFkxQXanIiEZGm5WWz8sTNfbnnsngA5n60k6c+3Y1hGCYnE5F/pxIv0gBL0g+xK6+UNv52/jQiwew4IiLNwmKx8OD1Cfzh2h4APPflPmau2I7TqSIv4ipU4kV+prziSp75bDcAf7quJ6EB3iYnEhFpPhaLhfuu6src0YlYLPD62hweWL6Z6lqn2dFEBJV4kZ/tzx/toLzawYBObRibFGN2HBGRFnH7kFj+fusAvKwWVm45yq9ey+R0tcPsWCIeTyVe5Gf4es8JPtp6DKvlzIdZrVaL2ZFERFrMDf06sGhiEr52K1/tPsGEl9dRfLrG7FgiHk0lXuQCKmscPLLizIdZJ6bG0btDiMmJRERa3pU92vP63ckE+3qReegU415MJ7+00uxYIh5LJV7kAl5cs5+DhRW0D/Jh+rDuZscRETFNUlwoy6ekEB7ow668Um5ZkE7uyQqzY4l4JJV4kZ9wqLCc57/aB8DMkb0I8rWbnEhExFwJUcG8d28KMaF+HCqsYMz8NHbnlZodS8TjqMSL/AjDMHhkRRbVtU4u7RrOyL5RZkcSEXEJsWEBvDs1lR4RQeSXVjH2xXQ25pwyO5aIR1GJF/kRn2zPY82eE3jbrMwZ1RuLRR9mFRH5XkSwL8unDGFApzYUn67h9kXr+GbvCbNjiXgMlXiR8yirquXRlTsAmHpFZzq3CzQ5kYiI62nj780bk5O5rFs4FdUO7np1Pau2HTM7lohHUIkXOY+//3MveSWVxIT68eurupodR0TEZfl7e7FoYhL/2SeKGofB/W9uZFlGjtmxRFo9lXiRH9idV8rL3x4AYM6NifjabSYnEhFxbT5eNv5+2wBuGxyD04A//WMbC9Zkmx1LpFVTiRf5N06nwcMfbMPhNLiudyRX9WxvdiQREbdgs1p47Bd9uPfKLgA8/vEuHv94F4ZhmJxMpHVSiRf5N+9tPMz6g6fw97bxyA29zI4jIuJWLBYL/31dT2aM6AnAgjXZzPjHmQsjItK0VOJFziqqqGbex7sAeOCabnRo42dyIhER9zTlii48flMfrBZYtj6X37y1kapah9mxRFoVlXiRs/7yyW5OllfTPSKQuy6NNzuOiIhbu3VwJ54fPxBvm5VV2/KYvCST8qpas2OJtBoq8SLAxpxTLFt/ZprC3NF9sNu0NERELtaIPlEsnnQJ/t42vtlbwC8XraOootrsWCKtgpqKeLxah5OZH2zHMGDMwGgGx4eaHUlEpNW4tFs4b0xOpo2/nc25RYx9MZ3jJZVmxxJxeyrx4vFeW3uIrKMlhPjZmXF9T7PjiIi0OgM6teXtKSlEBPuw53gZY+ancbCg3OxYIm5NJV48Wn5JJU9/tgeAP17Xg/BAH5MTiYi0Tt0jgnh3aipxYf4cPnWamxeks/NYidmxRNyWSrx4tLkf7aSsqpZ+MW249ZJOZscREWnVYkL9eWdqKglRwRSUVTH2xXQyD540O5aIW1KJF4/13b4CPtxyFKsF/md0IjarxexIIiKtXrsgH5b9aghJsW0prazl9pfX8dXufLNjibgdlXjxSFW1DmZ+sB2AO1LiSOwYYnIiERHPEeJn57W7k7myRzsqa5xMXpLJh1uOmh1LxK2oxItHeunr/ewvKCc80Ifpw7ubHUdExOP4edtYOCGJG/t1oNZp8MCyTby29pDZsUTchkq8eJzckxX87xf7AJg5MoFgX7vJiUREPJO3l5Vnx/VnwpBYDANmfrCd577Yi2EYZkcTcXkq8eJRDMNg1odZVNU6Se0Sxo39OpgdSUTEo1mtFuaM6s1vr+4KwFOf7WHuRztxOlXkRX6KSrx4lM92HOeLXfnYbRbmjErEYtGHWUVEzGaxWJg+vAczR/YC4OVvD/DH97ZS63CanEzEdanEi8eoqK7l0Q+zAPjV5Z3p2j7Q5EQiIvLv7r40nqdv6YfNauHdDYf59RsbqaxxmB1LxCWpxIvH+Ps/93G0uJLotn7cf1U3s+OIiMh5jBkUzYLbB+HtZeWzHce585X1lFXVmh1LxOWoxItH2HO8lEXf7Afg0Rt74+dtMzmRiIj8mGG9Ilhy52ACfbxI31/I+JfWcrK82uxYIi5FJV5aPcMwePiD7dQ6DYb1iuCahAizI4mIyAWkdAnjrXuGEBrgzdbDxdyyII2jRafNjiXiMlTipdV7f9MRMg6cxM9uY9YNvcyOIyIiP1Of6BDenpJCVIgv2SfKuWVBOvtPlJkdS8QlqMRLq1ZcUcP/fLQTgN9e043otv4mJxIRkYbo2j6Qd+9NpXN4AEeKTnPLgnS2Hyk2O5aI6VTipVV78rNdFJZX07V9IHdfGm92HBERaYSObfx4Z2oKiR2DKSyv5taFa1m7v9DsWCKmUomXVmtLbhFvrMsB4M+jEvH20v/uIiLuKizQh7fuGUJyfChlVbVMXJzB5zuOmx1LxDRqNdIqOZxnPsxqGPCLAR1J6RJmdiQREblIQb52ltw1mP9IiKCq1smU1zfw/qbDZscSMYVKvLRKb6w7xLYjxQT5evHg9QlmxxERkSbia7ex4PaB3DSwIw6nwe+Wb+GV7w6YHUukxanES6uTX1rJk5/uBuCP1/agXZCPyYlERKQpedmsPHVzP+4cGgfAoyt38NfVezAMw9xgIi2oUSX+hRdeID4+Hl9fXwYNGsQ333zzk8c///zzJCQk4OfnR48ePVi6dGm9/VlZWYwZM4a4uDgsFgvPPvtsY2KJADBv1S5KK2vpGx3C+ORYs+OIiEgzsFotPDKyF9OHdQfgb//cy6Mrd+B0qsiLZ2hwiV++fDnTpk3joYceYtOmTVx22WWMGDGCnJyc8x4/f/58ZsyYwezZs8nKyuLRRx/lvvvuY+XKlXXHVFRU0LlzZx5//HEiIyMb/27E46VlF/D+piNYLDB3dCI2q8XsSCIi0kwsFgu/vaYbc0b1BuDVtIP8/p0t1DicJicTaX4Wo4F/95ScnMzAgQOZP39+3baEhARGjx7NvHnzzjk+NTWVoUOH8uSTT9ZtmzZtGpmZmXz77bfnHB8XF8e0adOYNm3aT+aoqqqiqqqq7ueSkhJiYmIoLi4mODi4IW9JWonqWicj/vY12SfKmTAklj+PTjQ7koiItJAVm4/w+7e3UOs0uKZne57/5UB87TazY8kPlJSUEBISor7WBBp0Jb66upoNGzYwfPjwetuHDx9OWlraeZ9TVVWFr69vvW1+fn5kZGRQU1PTwLj/Mm/ePEJCQuoeMTExjX4taR0Wfbuf7BPlhAd681/De5gdR0REWtCo/h1ZeMcgfLys/HNXPne8nEFJZeN7hoira1CJLygowOFwEBERUW97REQEeXl5533Otddey6JFi9iwYQOGYZCZmcnixYupqamhoKCg0cFnzJhBcXFx3SM3N7fRryXuL/dkBX//514AHrw+gRB/u8mJRESkpV3dM4LX7k4myMeLjIMnufXFtZworbrwE0XcUKM+2Gqx1L/P2DCMc7Z9b+bMmYwYMYIhQ4Zgt9sZNWoUkyZNAsBma/xfc/n4+BAcHFzvIZ7r0ZU7qKxxkhwfyi8GdDQ7joiImGRwfCjLpgwhPNCbHcdKGPtiOodPVZgdS6TJNajEh4eHY7PZzrnqnp+ff87V+e/5+fmxePFiKioqOHjwIDk5OcTFxREUFER4eHjjk4uctXrHcT7feRwvq4W5oxN/9BdKERHxDL07hPDO1FQ6tvHjQEE5N89PZ+/xUrNjiTSpBpV4b29vBg0axOrVq+ttX716NampqT/5XLvdTnR0NDabjWXLljFy5EisVo2pl4tzutrB7A+zAJh8WWe6RQSZnEhERFxBfHgA792bSrf2geSVVDL2xXS25BaZHUukyTS4RU+fPp1FixaxePFidu7cye9+9ztycnKYOnUqcOZe9TvuuKPu+D179vD666+zd+9eMjIyuPXWW9m+fTuPPfZY3THV1dVs3ryZzZs3U11dzZEjR9i8eTP79u1rgrcordn/frGXI0Wn6djGj99e09XsOCIi4kIiQ3x5e0oK/WLacKqihvEvrSVtX+M/jyfiShpc4seNG8ezzz7LnDlz6N+/P19//TWrVq0iNvbMl+ocO3as3sx4h8PB008/Tb9+/Rg2bBiVlZWkpaURFxdXd8zRo0cZMGAAAwYM4NixYzz11FMMGDCAyZMnX/w7lFZrX34pL32zH4BZN/TC39vL5EQiIuJq2gZ488bkZIZ2DaO82sGkV9bzadb5h3GIuJMGz4l3VZo76lkMw2D8S+tI31/INT3bs2hiku6FFxGRH1VV6+CBtzbzSVYeVgs8PqYvY5M0nrqlqa81Hd2ULm5pxeajpO8vxNduZfaNvVXgRUTkJ/l42Xhu/ADGJkXjNOCP725l0dm/zRVxRyrx4naKT9cw96OdAPzm6m7EhPqbnEhERNyBl83KX8b05VeXdwZg7kc7efLTXbSSmxLEw6jEi9t55rPdFJRV0bldAJMvizc7joiIuBGLxcKD1yfw39f1BOD5L7N56IPtOJwq8uJeVOLFrWw7XMxraw8BMHdUIj5ejf/CMBER8Vz3XtmFx37RB4sF3lyXwwPLNlFd6zQ7lsjPphIvbsPhNHj4g204DRjVvwOpXfVlYSIi0njjkzvxv7cNwG6z8H9bj3HP0kwqqmvNjiXys6jEi9t4KyOHLYeLCfLx4qHrE8yOIyIircDIvh1YNPES/Ow21uw5wYSXMyiuqDE7lsgFqcSLWygoq+KJT3YB8Pvh3Wkf7GtyIhERaS2u6N6O1ycnE+zrxYZDpxi3MJ38kkqzY4n8JJV4cQuPrdpJSWUtvTsEc/uQWLPjiIhIKzMoti1vT02hXZAPu/JKuXlBOjmFFWbHEvlRKvHi8tbuL+QfG49gscDc0Yl42fS/rYiINL2ekcG8NzWVTqH+5Jys4OYFaezOKzU7lsh5qQ2JS6txOJn5wXYAbhvciQGd2pqcSEREWrNOYf68OzWFHhFB5JdWMfbFdDbmnDI7lsg5VOLFpS3+9gB788sIC/Dmj9f2MDuOiIh4gPbBvrw9JYWBndpQfLqGX760jm/2njA7lkg9KvHiso4UnebZz/cCMOP6BNr4e5ucSEREPEWIv53XJydzefd2nK5xcNer6/lo6zGzY4nUUYkXlzVnZRanaxwMjgtlzMCOZscREREP4+/txaI7khjZN4oah8H9b23krYwcs2OJACrx4qK+2HWcT7OOY7Na+PPoRCwWi9mRRETEA3l7WfnbrQMYn9wJw4AZ/9jG/K+yzY4lohIvrud0tYNZH2YBcPel8fSIDDI5kYiIeDKb1cL/jE7kvqu6APCXT3Yxb9VODMMwOZl4MpV4cTkvfLWP3JOniQrx5YFrupkdR0REBIvFwh+u7Vn3jeEvfr2fP723DYdTRV7MoRIvLiX7RBkvrtkPwKwbehHg42VyIhERkX+55/LOPDGmL1YLLM/M5f43N1JV6zA7lngglXhxGYZh8MiK7VQ7nFzZox3X9o40O5KIiMg5xl4Swwu/HIi3zcrH2/O4+9VMyqtqzY4lHkYlXlzGyq3H+G5fIT5eVubcqA+zioiI67ouMYpX7rwEf28b3+4r4JeL1nGqvNrsWOJBVOLFJZRU1vDn/9sBwH1XdaVTmL/JiURERH7a0K7hvHnPENr629mcW8TYF9PJK640O5Z4CJV4cQl/Xb2HE6VVxIcHMOWKzmbHERER+Vn6x7Th7SkpRAb7sje/jJsXpHGgoNzsWOIBVOLFdNuPFLMk7SAAc0b1xsfLZm4gERGRBugWEcS796YQHx7A4VOnuWVBGllHi82OJa2cSryYyuk0ePiD7TgNGNk3isu6tTM7koiISINFt/Xn7Skp9IoKpqCsmlsXrmX9wZNmx5JWTCVeTLU8M5fNuUUE+ngxc2Qvs+OIiIg0WrsgH5ZNGcLguFBKK2uZ8PI6vtyVb3YsaaVU4sU0hWVVPP7xLgB+N6w7EcG+JicSERG5OMG+dpbcNZire7anssbJPUszWbH5iNmxpBVSiRfTPP7xLopP15AQFczElFiz44iIiDQJP28bL04YxKj+Hah1GkxbvpnX0g+aHUtaGX0dprQowzA4WFjBP3ce550NhwGYOzoRL5t+nxQRkdbDbrPy17H9aeNnZ0n6IWauyKKooob7r+6q70GRJqESL83uWPFp0vYVkpZdSHp2AUf/bYburZfEMCi2rYnpREREmofVamH2jb0J8ffm7//cy9Or93CqooaH/zMBq1VFXi6OSrw0uZPl1aRnF5KWXUB6diH7fzAv19tmZUCnNlzRox13DY03KaWIiEjzs1gsTB/Wnbb+dh5duYPF3x2g+HQNfxnTR38LLRdFJV4uWmllDesPnuS7s1fbdx4rqbffaoE+0W1I7RJGapcwkmJD8fPWLHgREfEcdw6NJ8TPzh/e3cp7Gw9TUlnD/942AF+7/jyUxlGJlwarrHGw8dApvssuIC27kK2Hi3E4jXrH9IwMIqVLGKldwhkcH0qIn92ktCIiIq7hpoHRBPva+fWbG1m94ziTXsngpTuSCPLVn5HScBbDMIwLH+b6SkpKCAkJobi4mODgYLPjtCo1DidbDxeTnl3Ad/sK2ZBziupaZ71jYsP8Se0STmqXMIZ0DqNdkI9JaUVERFzb2v2FTF6SSVlVLX06hvDqnZcQFugZf26qrzUdlXg5h9NpsDOv5Ox97YWs219IebWj3jERwT4M7RJOSpcwUrqEEd3W36S0IiIi7mf7kWImLs6gsLyazu0CeP3uZDq08TM7VrNTX2s6KvGCYRjsLyivmx6Tnl3IqYqaese08beT0jmM1K5nrrZ3Dg/QiCwREZGLkH2ijAmL1nG0uJIOIb68NjmZLu0CzY7VrNTXmo5KvIc6WnSa7/YV1F1tzyuprLc/wNvG4PjQM7fIdA0jITJY47BERESa2NGi00x4eR3ZJ8oJDfBm6V2DSewYYnasZqO+1nRU4j1EQVlVXWFPzy7gYGFFvf3eXlYGdWp7ZoJM13D6Rodg1+grERGRZldYVsWkV9az7UgxgT5eLJqYxJDOYWbHahbqa01HJb6VKqmsIWP/Sb47e3vMrrzSevttVgt9o0POjn0MZ1BsW425EhERMUlpZQ33LM1k7f6TeHtZeX78QIb1ijA7VpNTX2s6KvGtxOlqBxsOnSItu4DvsgvZdriIH0x9JCEquG5W++D4UI20EhERcSGVNQ5+89YmVu84js1q4cmb+3LTwGizYzUpT+9rTUlz4t1UjcPJltwi0rIL+W5fAZtyiqh21B/7GB8eUHelfUjnUI8ZXyUiIuKOfO025v9yIH/6xzbe3XCY6W9voaiihrsu1beby7lU4t2Ew2mw81gJaWe/YCnjwEkqfjD2MSrEt25We0qXMI8YVSUiItKaeNmsPDGmLyF+dl7+9gBz/m8HRRXV/G5Yd02Fk3pU4l2UYRhknygjLbuQtH2FpO8vpPh0/bGPoQHeZ8c+nrnaHhfmrwUuIiLi5qxWCw//ZwJt/e089dke/v7FPopO1zD7ht6aFCd1VOJdyOFTFaTtK6y72p5fWlVvf6CPF8nxoXWz2ntEBGkxi4iItEIWi4X7r+5GiL83j6zYztL0QxSfruGpW/ppepwAKvGmOlFaRVr2v2a155ysP/bRx8tKUlzbultk+nQMwUsLV0RExGNMGBJLsK8Xv397Cys2H6W0spbnxw/Ez1sT5TydSnwLKj5dw7r9Zwp7WnYBe46X1dtvs1roH9Om7p72gZ009lFERMTTjerfkWBfO/e+sYEvduVzx+J1LJp4CSF+mjLnyTRishlVVNeSefBU3az27UeK6419tFigV93Yx3AuiQ8l0Ee/V4mIiMi51h88yV2vrqe0spZeUcEsuWsw7YLca/KcK/Y1d6US34Sqa51szi06c0/7vkI25Z6ixlH/P2/ndgEMPXt7zJDOYbQN8DYlq4iIiLifHUdLuGNxBgVlVcSF+fPa3cnEhPqbHetnc4W+1lqoxF8Eh9Mg62jx2dtjCll/4CSna+qPfewQ4ktq13CGdg0jpXM4kSG+LZJNREREWqeDBeXc/vI6Dp86TWSwL6/dPZhuEUFmx/pZVOKbjkp8I+zLL+WJT3azdn8hJZW19faFBXiT0iWMoWcnyHQK1dhHERERaVp5xZXcsXgde46X0cbfzqt3DqZ/TBuzY12QSnzT0Q3YjeBrt/HZjuMABPl4kdw5jNSzxb17RKBKu4iIiDSryBBf3p6SwqRX1rM5t4jxL61l4YQkLu0WbnY0aSG6Et9IS9IO0i+mDYkdgjX2UURERExRXlXLlNc28O2+ArxtVv5+W3+uS4wyO9aP0pX4pqP22UgTU+PoH9NGBV5ERERME+DjxcuTkhiRGEm1w8mv39jI2+tzzY4lLUANVERERMSN+XjZeG78QMYlxeA04I/vbWXh19lmx5JmphIvIiIi4uZsVguPj+nDlCs6A/DYql088ckuWsld03IeKvEiIiIirYDFYmHGiAT++7qeALzwVTYPfbAdh1NFvjVSiRcRERFpRe69sgvzbuqD1QJvrsvht8s2UV3rNDuWNDGVeBEREZFW5rbBnXhu/EDsNgsfbT3G5KWZVFTXXviJ4jZU4kVERERaoev7RLF40iX4e9v4es8Jbl+0juKKGrNjSRNpVIl/4YUXiI+Px9fXl0GDBvHNN9/85PHPP/88CQkJ+Pn50aNHD5YuXXrOMe+99x69evXCx8eHXr168f777zcmmoiIiIicdVm3drw+OZkQPzsbc4oY+2I6+SWVZseSJtDgEr98+XKmTZvGQw89xKZNm7jssssYMWIEOTk55z1+/vz5zJgxg9mzZ5OVlcWjjz7Kfffdx8qVK+uOSU9PZ9y4cUyYMIEtW7YwYcIExo4dy7p16xr/zkRERESEgZ3a8vaUFNoH+bD7eCljFqRxqLDc7FhykRr8ja3JyckMHDiQ+fPn121LSEhg9OjRzJs375zjU1NTGTp0KE8++WTdtmnTppGZmcm3334LwLhx4ygpKeHjjz+uO+a6666jbdu2vPXWWz8rl74BTEREROTH5Z6s4PaX13GosIJ2QT68dvdgeka2bGdSX2s6DboSX11dzYYNGxg+fHi97cOHDyctLe28z6mqqsLX17feNj8/PzIyMqipOXNfVnp6+jmvee211/7oa37/uiUlJfUeIiIiInJ+MaH+vDM1hZ6RQZworWLsgnQ2HDppdixppAaV+IKCAhwOBxEREfW2R0REkJeXd97nXHvttSxatIgNGzZgGAaZmZksXryYmpoaCgoKAMjLy2vQawLMmzePkJCQukdMTExD3oqIiIiIx2kf5MvyX6UwKLYtJZW13L4ogzV7TpgdSxqhUR9stVgs9X42DOOcbd+bOXMmI0aMYMiQIdjtdkaNGsWkSZMAsNlsjXpNgBkzZlBcXFz3yM3NbcxbEREREfEoIf52Xrt7MFd0b8fpGgevfndA3+zqhhpU4sPDw7HZbOdcIc/Pzz/nSvr3/Pz8WLx4MRUVFRw8eJCcnBzi4uIICgoiPDwcgMjIyAa9JoCPjw/BwcH1HiIiIiJyYf7eXrx0RxIPXNON58YP/MkLp+KaGlTivb29GTRoEKtXr663ffXq1aSmpv7kc+12O9HR0dhsNpYtW8bIkSOxWs/861NSUs55zc8+++yCrykiIiIijePtZeV3w7oT4ONldhRphAaftenTpzNhwgSSkpJISUlh4cKF5OTkMHXqVODMbS5HjhypmwW/Z88eMjIySE5O5tSpUzzzzDNs376dJUuW1L3mAw88wOWXX85f/vIXRo0axYoVK/j888/rpteIiIiIiMi/NLjEjxs3jsLCQubMmcOxY8dITExk1apVxMbGAnDs2LF6M+MdDgdPP/00u3fvxm63c9VVV5GWlkZcXFzdMampqSxbtoyHH36YmTNn0qVLF5YvX05ycvLFv0MRERERkVamwXPiXZXmjoqIiIi4NvW1ptOo6TQiIiIiImIelXgRERERETejEi8iIiIi4mZU4kVERERE3IxKvIiIiIiIm1GJFxERERFxMyrxIiIiIiJuRiVeRERERMTNqMSLiIiIiLgZlXgRERERETejEi8iIiIi4mZU4kVERERE3IxKvIiIiIiIm1GJFxERERFxM15mB2gqhmEAUFJSYnISERERETmf73va971NGq/VlPjS0lIAYmJiTE4iIiIiIj+ltLSUkJAQs2O4NYvRSn4VcjqdHD16lKCgICwWi9lxGqykpISYmBhyc3MJDg42O46gc+JqdD5cj86J69E5cS06H+cyDIPS0lI6dOiA1aq7ui9Gq7kSb7VaiY6ONjvGRQsODtZCdzE6J65F58P16Jy4Hp0T16LzUZ+uwDcN/QokIiIiIuJmVOJFRERERNyMSryL8PHxYdasWfj4+JgdRc7SOXEtOh+uR+fE9eicuBadD2lOreaDrSIiIiIinkJX4kVERERE3IxKvIiIiIiIm1GJFxERERFxMyrxIiIiIiJuRiVeRERERMTNqMSbbPbs2VgslnqPyMhIs2N5jK+//pobbriBDh06YLFY+OCDD+rtNwyD2bNn06FDB/z8/LjyyivJysoyJ6yHuNA5mTRp0jlrZsiQIeaE9QDz5s3jkksuISgoiPbt2zN69Gh2795d7xitk5b1c86J1knLmT9/Pn379q37VtaUlBQ+/vjjuv1aH9JcVOJdQO/evTl27FjdY9u2bWZH8hjl5eX069eP55577rz7n3jiCZ555hmee+451q9fT2RkJMOGDaO0tLSFk3qOC50TgOuuu67emlm1alULJvQsa9as4b777mPt2rWsXr2a2tpahg8fTnl5ed0xWict6+ecE9A6aSnR0dE8/vjjZGZmkpmZydVXX82oUaPqirrWhzQbQ0w1a9Yso1+/fmbHEMMwAOP999+v+9npdBqRkZHG448/XretsrLSCAkJMRYsWGBCQs/zw3NiGIYxceJEY9SoUabkEcPIz883AGPNmjWGYWiduIIfnhPD0DoxW9u2bY1FixZpfUiz0pV4F7B37146dOhAfHw8t956K/v37zc7kgAHDhwgLy+P4cOH123z8fHhiiuuIC0tzcRk8tVXX9G+fXu6d+/OPffcQ35+vtmRPEZxcTEAoaGhgNaJK/jhOfme1knLczgcLFu2jPLyclJSUrQ+pFmpxJssOTmZpUuX8umnn/LSSy+Rl5dHamoqhYWFZkfzeHl5eQBERETU2x4REVG3T1reiBEjeOONN/jiiy94+umnWb9+PVdffTVVVVVmR2v1DMNg+vTpXHrppSQmJgJaJ2Y73zkBrZOWtm3bNgIDA/Hx8WHq1Km8//779OrVS+tDmpWX2QE83YgRI+r+uU+fPqSkpNClSxeWLFnC9OnTTUwm37NYLPV+NgzjnG3ScsaNG1f3z4mJiSQlJREbG8tHH33ETTfdZGKy1u/+++9n69atfPvtt+fs0zoxx4+dE62TltWjRw82b95MUVER7733HhMnTmTNmjV1+7U+pDnoSryLCQgIoE+fPuzdu9fsKB7v+ylBP7xakp+ff85VFTFPVFQUsbGxWjPN7De/+Q0ffvghX375JdHR0XXbtU7M82Pn5Hy0TpqXt7c3Xbt2JSkpiXnz5tGvXz/+9re/aX1Is1KJdzFVVVXs3LmTqKgos6N4vPj4eCIjI1m9enXdturqatasWUNqaqqJyeTfFRYWkpubqzXTTAzD4P777+cf//gHX3zxBfHx8fX2a520vAudk/PROmlZhmFQVVWl9SHNSrfTmOy//uu/uOGGG+jUqRP5+fnMnTuXkpISJk6caHY0j1BWVsa+ffvqfj5w4ACbN28mNDSUTp06MW3aNB577DG6detGt27deOyxx/D392f8+PEmpm7dfuqchIaGMnv2bMaMGUNUVBQHDx7kwQcfJDw8nF/84hcmpm697rvvPt58801WrFhBUFBQ3RXFkJAQ/Pz8sFgsWict7ELnpKysTOukBT344IOMGDGCmJgYSktLWbZsGV999RWffPKJ1oc0L/MG44hhGMa4ceOMqKgow263Gx06dDBuuukmIysry+xYHuPLL780gHMeEydONAzjzPi8WbNmGZGRkYaPj49x+eWXG9u2bTM3dCv3U+ekoqLCGD58uNGuXTvDbrcbnTp1MiZOnGjk5OSYHbvVOt+5AIxXXnml7hitk5Z1oXOiddKy7rrrLiM2Ntbw9vY22rVrZ1xzzTXGZ599Vrdf60Oai8UwDKMlf2kQEREREZGLo3viRURERETcjEq8iIiIiIibUYkXEREREXEzKvEiIiIiIm5GJV5ERERExM2oxIuIiIiIuBmVeBERERERN6MSLyIiIiLiZlTiRURERETcjEq8iIiIiIibUYkXEREREXEz/w+5a9Zug6hcSgAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 800x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "mlp_dim, mlp_model, mlp_dim_history, mlp_dim_curves = tune_dims('mlp')\n",
    "title = f\"Train and Test RMSE vs. Embedding Dim for MLP on {dataset}\"\n",
    "plot_learning_curve(mlp_dim_history, mlp_dim_curves, title, True)\n",
    "print(f\"Best Embedding Dimension for MLP on {dataset}\", mlp_dim)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "8f40a878",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1\n",
      "------------------------\n",
      "loss: 0.125934  [ 1024/15945812]\n",
      "loss: 0.034734  [1025024/15945812]\n",
      "loss: 0.034944  [2049024/15945812]\n",
      "loss: 0.031469  [3073024/15945812]\n",
      "loss: 0.028040  [4097024/15945812]\n",
      "loss: 0.029949  [5121024/15945812]\n",
      "loss: 0.029912  [6145024/15945812]\n",
      "loss: 0.029631  [7169024/15945812]\n",
      "loss: 0.027694  [8193024/15945812]\n",
      "loss: 0.028323  [9217024/15945812]\n",
      "loss: 0.030648  [10241024/15945812]\n",
      "loss: 0.029718  [11265024/15945812]\n",
      "loss: 0.028390  [12289024/15945812]\n",
      "loss: 0.029991  [13313024/15945812]\n",
      "loss: 0.029014  [14337024/15945812]\n",
      "loss: 0.029624  [15361024/15945812]\n",
      "Test MSE 0.7700475713923641\n",
      "Test RMSE 0.8775235446370451\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.024776  [ 1024/15945812]\n",
      "loss: 0.026746  [1025024/15945812]\n",
      "loss: 0.028918  [2049024/15945812]\n",
      "loss: 0.027846  [3073024/15945812]\n",
      "loss: 0.026914  [4097024/15945812]\n",
      "loss: 0.027386  [5121024/15945812]\n",
      "loss: 0.030408  [6145024/15945812]\n",
      "loss: 0.027526  [7169024/15945812]\n",
      "loss: 0.028831  [8193024/15945812]\n",
      "loss: 0.028979  [9217024/15945812]\n",
      "loss: 0.028382  [10241024/15945812]\n",
      "loss: 0.028117  [11265024/15945812]\n",
      "loss: 0.024525  [12289024/15945812]\n",
      "loss: 0.027395  [13313024/15945812]\n",
      "loss: 0.028905  [14337024/15945812]\n",
      "loss: 0.028658  [15361024/15945812]\n",
      "Test MSE 0.7529413479708044\n",
      "Test RMSE 0.8677219300967357\n",
      "Test MSE 0.6761272288612026\n",
      "Test RMSE 0.822269559731602\n",
      "Test MSE 0.7529413479708041\n",
      "Test RMSE 0.8677219300967356\n",
      "Epoch 1\n",
      "------------------------\n",
      "loss: 0.225583  [ 1024/15945812]\n",
      "loss: 0.031466  [1025024/15945812]\n",
      "loss: 0.029348  [2049024/15945812]\n",
      "loss: 0.027385  [3073024/15945812]\n",
      "loss: 0.028571  [4097024/15945812]\n",
      "loss: 0.028150  [5121024/15945812]\n",
      "loss: 0.029806  [6145024/15945812]\n",
      "loss: 0.029089  [7169024/15945812]\n",
      "loss: 0.026780  [8193024/15945812]\n",
      "loss: 0.028512  [9217024/15945812]\n",
      "loss: 0.024440  [10241024/15945812]\n",
      "loss: 0.028209  [11265024/15945812]\n",
      "loss: 0.030888  [12289024/15945812]\n",
      "loss: 0.030143  [13313024/15945812]\n",
      "loss: 0.032009  [14337024/15945812]\n",
      "loss: 0.027438  [15361024/15945812]\n",
      "Test MSE 0.7769832185501827\n",
      "Test RMSE 0.8814665158417435\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.028922  [ 1024/15945812]\n",
      "loss: 0.029826  [1025024/15945812]\n",
      "loss: 0.027720  [2049024/15945812]\n",
      "loss: 0.028975  [3073024/15945812]\n",
      "loss: 0.028224  [4097024/15945812]\n",
      "loss: 0.029604  [5121024/15945812]\n",
      "loss: 0.027985  [6145024/15945812]\n",
      "loss: 0.028563  [7169024/15945812]\n",
      "loss: 0.026653  [8193024/15945812]\n",
      "loss: 0.029466  [9217024/15945812]\n",
      "loss: 0.029061  [10241024/15945812]\n",
      "loss: 0.027989  [11265024/15945812]\n",
      "loss: 0.029885  [12289024/15945812]\n",
      "loss: 0.030247  [13313024/15945812]\n",
      "loss: 0.024476  [14337024/15945812]\n",
      "loss: 0.033896  [15361024/15945812]\n",
      "Test MSE 0.7772361510008718\n",
      "Test RMSE 0.8816099766908675\n",
      "Test MSE 0.700845643652505\n",
      "Test RMSE 0.8371652427403475\n",
      "Test MSE 0.7772361510008711\n",
      "Test RMSE 0.881609976690867\n",
      "Epoch 1\n",
      "------------------------\n",
      "loss: 0.114147  [ 1024/15945812]\n",
      "loss: 0.030007  [1025024/15945812]\n",
      "loss: 0.031534  [2049024/15945812]\n",
      "loss: 0.036140  [3073024/15945812]\n",
      "loss: 0.031649  [4097024/15945812]\n",
      "loss: 0.030067  [5121024/15945812]\n",
      "loss: 0.034920  [6145024/15945812]\n",
      "loss: 0.032212  [7169024/15945812]\n",
      "loss: 0.030351  [8193024/15945812]\n",
      "loss: 0.032645  [9217024/15945812]\n",
      "loss: 0.030513  [10241024/15945812]\n",
      "loss: 0.030986  [11265024/15945812]\n",
      "loss: 0.030582  [12289024/15945812]\n",
      "loss: 0.033307  [13313024/15945812]\n",
      "loss: 0.030839  [14337024/15945812]\n",
      "loss: 0.032352  [15361024/15945812]\n",
      "Test MSE 0.8623468134204565\n",
      "Test RMSE 0.9286263045059926\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.030818  [ 1024/15945812]\n",
      "loss: 0.032293  [1025024/15945812]\n",
      "loss: 0.032817  [2049024/15945812]\n",
      "loss: 0.031430  [3073024/15945812]\n",
      "loss: 0.035122  [4097024/15945812]\n",
      "loss: 0.032091  [5121024/15945812]\n",
      "loss: 0.031407  [6145024/15945812]\n",
      "loss: 0.033703  [7169024/15945812]\n",
      "loss: 0.033919  [8193024/15945812]\n",
      "loss: 0.028994  [9217024/15945812]\n",
      "loss: 0.033706  [10241024/15945812]\n",
      "loss: 0.035668  [11265024/15945812]\n",
      "loss: 0.036300  [12289024/15945812]\n",
      "loss: 0.031762  [13313024/15945812]\n",
      "loss: 0.032786  [14337024/15945812]\n",
      "loss: 0.034740  [15361024/15945812]\n",
      "Test MSE 0.8629428612254457\n",
      "Test RMSE 0.9289471789210868\n",
      "Test MSE 0.8050269129685381\n",
      "Test RMSE 0.8972329201319678\n",
      "Test MSE 0.8629428612254458\n",
      "Test RMSE 0.9289471789210869\n",
      "Epoch 1\n",
      "------------------------\n",
      "loss: 0.213245  [ 1024/15945812]\n",
      "loss: 0.033215  [1025024/15945812]\n",
      "loss: 0.030653  [2049024/15945812]\n",
      "loss: 0.031353  [3073024/15945812]\n",
      "loss: 0.037701  [4097024/15945812]\n",
      "loss: 0.035806  [5121024/15945812]\n",
      "loss: 0.032804  [6145024/15945812]\n",
      "loss: 0.032732  [7169024/15945812]\n",
      "loss: 0.035839  [8193024/15945812]\n",
      "loss: 0.037102  [9217024/15945812]\n",
      "loss: 0.033037  [10241024/15945812]\n",
      "loss: 0.034696  [11265024/15945812]\n",
      "loss: 0.033134  [12289024/15945812]\n",
      "loss: 0.031772  [13313024/15945812]\n",
      "loss: 0.034750  [14337024/15945812]\n",
      "loss: 0.032233  [15361024/15945812]\n",
      "Test MSE 0.898124532109108\n",
      "Test RMSE 0.9476943241937814\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.034446  [ 1024/15945812]\n",
      "loss: 0.035117  [1025024/15945812]\n",
      "loss: 0.032528  [2049024/15945812]\n",
      "loss: 0.034724  [3073024/15945812]\n",
      "loss: 0.033601  [4097024/15945812]\n",
      "loss: 0.033610  [5121024/15945812]\n",
      "loss: 0.031947  [6145024/15945812]\n",
      "loss: 0.033297  [7169024/15945812]\n",
      "loss: 0.029171  [8193024/15945812]\n",
      "loss: 0.033768  [9217024/15945812]\n",
      "loss: 0.035485  [10241024/15945812]\n",
      "loss: 0.032811  [11265024/15945812]\n",
      "loss: 0.032212  [12289024/15945812]\n",
      "loss: 0.032249  [13313024/15945812]\n",
      "loss: 0.032904  [14337024/15945812]\n",
      "loss: 0.034200  [15361024/15945812]\n",
      "Test MSE 0.8805934715968883\n",
      "Test RMSE 0.9383994200748892\n",
      "Test MSE 0.8252700904895027\n",
      "Test RMSE 0.9084437739835651\n",
      "Test MSE 0.8805934715968875\n",
      "Test RMSE 0.9383994200748887\n",
      "Epoch 1\n",
      "------------------------\n",
      "loss: 0.196707  [ 1024/15945812]\n",
      "loss: 0.035806  [1025024/15945812]\n",
      "loss: 0.033536  [2049024/15945812]\n",
      "loss: 0.034522  [3073024/15945812]\n",
      "loss: 0.036021  [4097024/15945812]\n",
      "loss: 0.034417  [5121024/15945812]\n",
      "loss: 0.032501  [6145024/15945812]\n",
      "loss: 0.036193  [7169024/15945812]\n",
      "loss: 0.035029  [8193024/15945812]\n",
      "loss: 0.034938  [9217024/15945812]\n",
      "loss: 0.034154  [10241024/15945812]\n",
      "loss: 0.034933  [11265024/15945812]\n",
      "loss: 0.035548  [12289024/15945812]\n",
      "loss: 0.033469  [13313024/15945812]\n",
      "loss: 0.035485  [14337024/15945812]\n",
      "loss: 0.036225  [15361024/15945812]\n",
      "Test MSE 0.9285154410701175\n",
      "Test RMSE 0.9635950607335623\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.032657  [ 1024/15945812]\n",
      "loss: 0.043019  [1025024/15945812]\n",
      "loss: 0.044740  [2049024/15945812]\n",
      "loss: 0.039184  [3073024/15945812]\n",
      "loss: 0.041695  [4097024/15945812]\n",
      "loss: 0.044668  [5121024/15945812]\n",
      "loss: 0.043303  [6145024/15945812]\n",
      "loss: 0.042430  [7169024/15945812]\n",
      "loss: 0.045760  [8193024/15945812]\n",
      "loss: 0.044782  [9217024/15945812]\n",
      "loss: 0.046938  [10241024/15945812]\n",
      "loss: 0.042240  [11265024/15945812]\n",
      "loss: 0.044830  [12289024/15945812]\n",
      "loss: 0.040848  [13313024/15945812]\n",
      "loss: 0.044033  [14337024/15945812]\n",
      "loss: 0.044281  [15361024/15945812]\n",
      "Test MSE 1.0734947318060344\n",
      "Test RMSE 1.03609590859439\n",
      "Test MSE 1.1015001895777092\n",
      "Test RMSE 1.0495237918111762\n",
      "Test MSE 1.0734947318060324\n",
      "Test RMSE 1.036095908594389\n",
      "Epoch 1\n",
      "------------------------\n",
      "loss: 0.133346  [ 1024/15945812]\n",
      "loss: 0.049927  [1025024/15945812]\n",
      "loss: 0.045288  [2049024/15945812]\n",
      "loss: 0.048058  [3073024/15945812]\n",
      "loss: 0.044093  [4097024/15945812]\n",
      "loss: 0.045168  [5121024/15945812]\n",
      "loss: 0.041363  [6145024/15945812]\n",
      "loss: 0.044063  [7169024/15945812]\n",
      "loss: 0.042873  [8193024/15945812]\n",
      "loss: 0.047939  [9217024/15945812]\n",
      "loss: 0.043210  [10241024/15945812]\n",
      "loss: 0.043189  [11265024/15945812]\n",
      "loss: 0.046448  [12289024/15945812]\n",
      "loss: 0.042111  [13313024/15945812]\n",
      "loss: 0.045124  [14337024/15945812]\n",
      "loss: 0.044373  [15361024/15945812]\n",
      "Test MSE 1.0813065054231807\n",
      "Test RMSE 1.0398588872646042\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.047415  [ 1024/15945812]\n",
      "loss: 0.043120  [1025024/15945812]\n",
      "loss: 0.044834  [2049024/15945812]\n",
      "loss: 0.045904  [3073024/15945812]\n",
      "loss: 0.043379  [4097024/15945812]\n",
      "loss: 0.046050  [5121024/15945812]\n",
      "loss: 0.046792  [6145024/15945812]\n",
      "loss: 0.044002  [7169024/15945812]\n",
      "loss: 0.047265  [8193024/15945812]\n",
      "loss: 0.044836  [9217024/15945812]\n",
      "loss: 0.048117  [10241024/15945812]\n",
      "loss: 0.042228  [11265024/15945812]\n",
      "loss: 0.048340  [12289024/15945812]\n",
      "loss: 0.047364  [13313024/15945812]\n",
      "loss: 0.047419  [14337024/15945812]\n",
      "loss: 0.046352  [15361024/15945812]\n",
      "Test MSE 1.0981672889621645\n",
      "Test RMSE 1.047934773238375\n",
      "Test MSE 1.1278382650898198\n",
      "Test RMSE 1.0619972999446938\n",
      "Test MSE 1.0981672889621645\n",
      "Test RMSE 1.047934773238375\n",
      "Test MSE 3.1613927039582688\n",
      "Test RMSE 1.7780305689043339\n",
      "Best Learning Rate for NCF on 20m 0.001\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ncf_lr, ncf_model, ncf_lr_history, ncf_lr_curves = tune_lr('ncf', gmf_model=gmf_model, mlp_model=mlp_model)\n",
    "title = f\"Train and Test RMSE vs. Learning Rate for NCF on {dataset}\"\n",
    "plot_learning_curve(ncf_lr_history, ncf_lr_curves, title, True)\n",
    "print(f\"Best Learning Rate for NCF on {dataset}\", ncf_lr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "1d699807",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1\n",
      "------------------------\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.097825  [ 1024/15945812]\n",
      "loss: 0.031000  [1025024/15945812]\n",
      "loss: 0.032747  [2049024/15945812]\n",
      "loss: 0.030965  [3073024/15945812]\n",
      "loss: 0.028762  [4097024/15945812]\n",
      "loss: 0.034955  [5121024/15945812]\n",
      "loss: 0.032868  [6145024/15945812]\n",
      "loss: 0.030627  [7169024/15945812]\n",
      "loss: 0.033926  [8193024/15945812]\n",
      "loss: 0.030145  [9217024/15945812]\n",
      "loss: 0.034415  [10241024/15945812]\n",
      "loss: 0.033199  [11265024/15945812]\n",
      "loss: 0.031839  [12289024/15945812]\n",
      "loss: 0.033863  [13313024/15945812]\n",
      "loss: 0.033663  [14337024/15945812]\n",
      "loss: 0.032440  [15361024/15945812]\n",
      "Test MSE 0.8647577468867794\n",
      "Test RMSE 0.9299235166866033\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.033203  [ 1024/15945812]\n",
      "loss: 0.033691  [1025024/15945812]\n",
      "loss: 0.030753  [2049024/15945812]\n",
      "loss: 0.033331  [3073024/15945812]\n",
      "loss: 0.032072  [4097024/15945812]\n",
      "loss: 0.033230  [5121024/15945812]\n",
      "loss: 0.032400  [6145024/15945812]\n",
      "loss: 0.028904  [7169024/15945812]\n",
      "loss: 0.033661  [8193024/15945812]\n",
      "loss: 0.033345  [9217024/15945812]\n",
      "loss: 0.031738  [10241024/15945812]\n",
      "loss: 0.033528  [11265024/15945812]\n",
      "loss: 0.033680  [12289024/15945812]\n",
      "loss: 0.031803  [13313024/15945812]\n",
      "loss: 0.032007  [14337024/15945812]\n",
      "loss: 0.030548  [15361024/15945812]\n",
      "Test MSE 0.8773502125164228\n",
      "Test RMSE 0.9366697457035872\n",
      "Test MSE 0.8146768551989464\n",
      "Test RMSE 0.9025945131668741\n",
      "Test MSE 0.8773502125164231\n",
      "Test RMSE 0.9366697457035874\n",
      "Epoch 1\n",
      "------------------------\n",
      "loss: 0.054853  [ 1024/15945812]\n",
      "loss: 0.035066  [1025024/15945812]\n",
      "loss: 0.031106  [2049024/15945812]\n",
      "loss: 0.031638  [3073024/15945812]\n",
      "loss: 0.032045  [4097024/15945812]\n",
      "loss: 0.034326  [5121024/15945812]\n",
      "loss: 0.032355  [6145024/15945812]\n",
      "loss: 0.032551  [7169024/15945812]\n",
      "loss: 0.031155  [8193024/15945812]\n",
      "loss: 0.030682  [9217024/15945812]\n",
      "loss: 0.032104  [10241024/15945812]\n",
      "loss: 0.031724  [11265024/15945812]\n",
      "loss: 0.032597  [12289024/15945812]\n",
      "loss: 0.034489  [13313024/15945812]\n",
      "loss: 0.031432  [14337024/15945812]\n",
      "loss: 0.029518  [15361024/15945812]\n",
      "Test MSE 0.8782503517335785\n",
      "Test RMSE 0.9371501223035605\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.034149  [ 1024/15945812]\n",
      "loss: 0.031137  [1025024/15945812]\n",
      "loss: 0.032139  [2049024/15945812]\n",
      "loss: 0.032771  [3073024/15945812]\n",
      "loss: 0.032661  [4097024/15945812]\n",
      "loss: 0.032155  [5121024/15945812]\n",
      "loss: 0.033051  [6145024/15945812]\n",
      "loss: 0.032791  [7169024/15945812]\n",
      "loss: 0.034593  [8193024/15945812]\n",
      "loss: 0.029771  [9217024/15945812]\n",
      "loss: 0.031074  [10241024/15945812]\n",
      "loss: 0.032775  [11265024/15945812]\n",
      "loss: 0.034291  [12289024/15945812]\n",
      "loss: 0.033398  [13313024/15945812]\n",
      "loss: 0.033987  [14337024/15945812]\n",
      "loss: 0.031663  [15361024/15945812]\n",
      "Test MSE 0.8662313799637821\n",
      "Test RMSE 0.9307155204270433\n",
      "Test MSE 0.8166579462227358\n",
      "Test RMSE 0.9036912892258815\n",
      "Test MSE 0.8662313799637814\n",
      "Test RMSE 0.930715520427043\n",
      "Epoch 1\n",
      "------------------------\n",
      "loss: 0.061798  [ 1024/15945812]\n",
      "loss: 0.034385  [1025024/15945812]\n",
      "loss: 0.033566  [2049024/15945812]\n",
      "loss: 0.033982  [3073024/15945812]\n",
      "loss: 0.033958  [4097024/15945812]\n",
      "loss: 0.032296  [5121024/15945812]\n",
      "loss: 0.031649  [6145024/15945812]\n",
      "loss: 0.032817  [7169024/15945812]\n",
      "loss: 0.030662  [8193024/15945812]\n",
      "loss: 0.032247  [9217024/15945812]\n",
      "loss: 0.032138  [10241024/15945812]\n",
      "loss: 0.032935  [11265024/15945812]\n",
      "loss: 0.031341  [12289024/15945812]\n",
      "loss: 0.034592  [13313024/15945812]\n",
      "loss: 0.031995  [14337024/15945812]\n",
      "loss: 0.031555  [15361024/15945812]\n",
      "Test MSE 0.881238965996361\n",
      "Test RMSE 0.9387432907863369\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.031426  [ 1024/15945812]\n",
      "loss: 0.032027  [1025024/15945812]\n",
      "loss: 0.032061  [2049024/15945812]\n",
      "loss: 0.031782  [3073024/15945812]\n",
      "loss: 0.033633  [4097024/15945812]\n",
      "loss: 0.035739  [5121024/15945812]\n",
      "loss: 0.033492  [6145024/15945812]\n",
      "loss: 0.032456  [7169024/15945812]\n",
      "loss: 0.029911  [8193024/15945812]\n",
      "loss: 0.028938  [9217024/15945812]\n",
      "loss: 0.031202  [10241024/15945812]\n",
      "loss: 0.031500  [11265024/15945812]\n",
      "loss: 0.033096  [12289024/15945812]\n",
      "loss: 0.031287  [13313024/15945812]\n",
      "loss: 0.028514  [14337024/15945812]\n",
      "loss: 0.030496  [15361024/15945812]\n",
      "Test MSE 0.8708295866834227\n",
      "Test RMSE 0.9331825044884965\n",
      "Test MSE 0.8193766748045644\n",
      "Test RMSE 0.9051942746198545\n",
      "Test MSE 0.870829586683423\n",
      "Test RMSE 0.9331825044884966\n",
      "Epoch 1\n",
      "------------------------\n",
      "loss: 0.170125  [ 1024/15945812]\n",
      "loss: 0.033915  [1025024/15945812]\n",
      "loss: 0.035039  [2049024/15945812]\n",
      "loss: 0.033671  [3073024/15945812]\n",
      "loss: 0.031855  [4097024/15945812]\n",
      "loss: 0.033759  [5121024/15945812]\n",
      "loss: 0.034821  [6145024/15945812]\n",
      "loss: 0.033641  [7169024/15945812]\n",
      "loss: 0.034402  [8193024/15945812]\n",
      "loss: 0.033612  [9217024/15945812]\n",
      "loss: 0.031465  [10241024/15945812]\n",
      "loss: 0.031507  [11265024/15945812]\n",
      "loss: 0.033738  [12289024/15945812]\n",
      "loss: 0.030941  [13313024/15945812]\n",
      "loss: 0.033162  [14337024/15945812]\n",
      "loss: 0.035167  [15361024/15945812]\n",
      "Test MSE 0.8617144562842479\n",
      "Test RMSE 0.9282857621897731\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.031211  [ 1024/15945812]\n",
      "loss: 0.033396  [1025024/15945812]\n",
      "loss: 0.031518  [2049024/15945812]\n",
      "loss: 0.033193  [3073024/15945812]\n",
      "loss: 0.033879  [4097024/15945812]\n",
      "loss: 0.029397  [5121024/15945812]\n",
      "loss: 0.031703  [6145024/15945812]\n",
      "loss: 0.035991  [7169024/15945812]\n",
      "loss: 0.033317  [8193024/15945812]\n",
      "loss: 0.032106  [9217024/15945812]\n",
      "loss: 0.028594  [10241024/15945812]\n",
      "loss: 0.032193  [11265024/15945812]\n",
      "loss: 0.032658  [12289024/15945812]\n",
      "loss: 0.032194  [13313024/15945812]\n",
      "loss: 0.032393  [14337024/15945812]\n",
      "loss: 0.032467  [15361024/15945812]\n",
      "Test MSE 0.8770688404264344\n",
      "Test RMSE 0.9365195355284557\n",
      "Test MSE 0.8133183778213858\n",
      "Test RMSE 0.90184165895205\n",
      "Test MSE 0.8770688404264344\n",
      "Test RMSE 0.9365195355284557\n",
      "Test MSE 0.8770688404264342\n",
      "Test RMSE 0.9365195355284556\n",
      "Best Learning Rate for NCF on 20m 0.001\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAvoAAAI6CAYAAABW9+xhAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8fJSN1AAAACXBIWXMAAA9hAAAPYQGoP6dpAABvgUlEQVR4nO3deVxU5eLH8e8AAww7iLK44ZK7VmqhmKllmqmpWWmLqe2b95p5K7NyySQ1vXYrvaWR2qaVLTe1Uksrf2qmV2+pZbYYLiCBAio7nN8fIyPjDAqIgIfP+/WaF8yZ55zzDDNn+D7PPOc5FsMwDAEAAAAwFY/qrgAAAACAykfQBwAAAEyIoA8AAACYEEEfAAAAMCGv6q4AAAAAKkdhYaHy8/OruxqoIlarVZ6enqU+TtAHAAC4wBmGoeTkZKWnp1d3VVDFQkJCFBkZKYvF4vIYQR8AAOACVxzy69WrJz8/P7ehD+ZiGIaysrKUkpIiSYqKinIpQ9AHAAC4gBUWFjpCfp06daq7OqhCNptNkpSSkqJ69eq5DOPhZFwAAIALWPGYfD8/v2quCapD8evu7twMgj4AAIAJMFyndjrT607QBwAAAEyIoA8AAACYEEEfAAAAptGzZ0+NHTu2uqtRIxD0AQAAUOUsFssZb6NGjarQdj/88EM9++yz51S3UaNGOerh5eWlRo0a6YEHHtDRo0edysXExMhisWjp0qUu22jbtq0sFosWLVrkWLZ9+3YNGDBA9erVk6+vr2JiYjRs2DClpqZKkvbt21fq32Pz5s3lfh4E/VrmbAdV8W39+vXntJ/JkydfsCcFLVq0SBaLRfv27XP7+JkOwtNvpW2jPA4dOqTJkydrx44dZSq/fv16pzp4enqqbt26GjhwoLZu3epSvvjDLDAwUMePH3d5/M8//5SHh4csFosmT57s9NhPP/2kESNGqGnTpvL19VV4eLg6duyohx9+WJmZmS77KO1WE/Xs2bPU+sbExFTaforfTy+88EKlbbM0xe+NshzfPXv2VM+ePR33i+tZ8h9WVTn9mLNarapTp44uu+wyPfLII9q1a5fLOuV5rpVRv/79+yssLEwWi+W89yQWB4v777/f5bHi5/3BBx+4PPbDDz9o9OjRatKkiXx9fRUQEKCOHTtq5syZOnLkiKPcmd77O3fuPK/PrSx++eUXjR8/Xp06dVJISIjCwsLUrVs3t89Zsk87OGrUKIWHh8vPz09du3bVl19+WcW1hjtJSUmO29y5cxUUFOS07MUXX3QqX9Yr/oaFhSkwMPCc63fttdcqKSlJ+/bt08KFC/Xpp5/qwQcfdCnXsGFDvfHGG07LNm/erOTkZPn7+zuWpaSkqHfv3goPD9cXX3yhn376SQkJCYqKilJWVpbT+mvXrnX6WyQlJalTp07lfg7Mo1/LbNq0yen+s88+q3Xr1umrr75yWt6mTZtz2s/dd9+ta6+99py2UVNFRUW5/B0ffPBBZWRk6O2333Ype64OHTqkKVOmKCYmRpdcckmZ15s+fbp69eql/Px8bd++XVOmTFGPHj20Y8cOXXTRRU5lrVarCgoKtGzZMt11111Oj73xxhsKDAx0Cu6SvVeiW7duat26tZ555hnFxMQoNTVV//vf/7R06VKNHz9eQUFBjvI2m83lfVbTNW3a1OU1lSQfH59qqE31Kn7fN2vWrNrqMGbMGN16660qKipSenq6tm/froSEBL300kuKj4/XP/7xD0fZjh07atOmTef8WVYWjzzyiL777jslJCQoMjKyUo77snj99df1yCOPqGXLlmctu2DBAj344INq2bKl/vGPf6hNmzbKz8/X1q1b9e9//1ubNm3SRx995Chf2nu/Ol//YqtXr9bKlSs1YsQIXXbZZY7PrptuuklTpkzRM8884yibm5urq6++Wunp6XrxxRdVr149vfLKK7r22mu1du1a9ejRoxqfyfllGIay8wurZd82q2eZOnEiIyMdvwcHB8tisTiW7du3T1FRUVq2bJnmzZunzZs3a/78+br++uv18MMP69tvv9WRI0fUrFkzPfnkk7rlllsc2+rZs6cuueQSzZ07V5K9cXzvvffq119/1fvvv6/Q0FA99dRTuvfee89YPx8fH0d9GjRooGHDhrnt7Ljtttv0z3/+U/v371fDhg0lSQkJCbrtttu0ZMkSR7mNGzcqMzNTCxculJeXPYI3adJEV111lcs269Sp4/T3qSiCfi3TpUsXp/t169aVh4eHy/LTZWVllWt+3gYNGqhBgwYVqmNN5+Pj4/L3CgoKUl5e3ln/jlXpoosuctSne/fuCgkJ0ciRI/XWW29pypQpTmW9vb01cOBAJSQkOAV9wzC0aNEiDRs2TAsWLHBaZ+7cufLw8ND69eudek5uvPFGPfvsszIMw6l8Wd5nNY3NZrvg6ny+uHvfV7VGjRo51eG6667TuHHjdMMNN+ixxx5Tu3bt1K9fP0n2Y7Kq6rtz505dfvnlGjx4cKVsr7CwUAUFBWdsUHbt2lW7d+/Wk08+qeXLl59xe5s2bdIDDzyga665Rh9//LHTdq+55ho9+uij+vzzz53Wqcnv/eHDh+uhhx5yCpL9+vVTamqqZsyYoccff9zxHF9//XXt3LlTGzduVNeuXSVJvXr10sUXX6zHHntM3333XbU8h6qQnV+oNs98US373j21r/y8KydiPv7445o9e7beeOMN+fj4KCcnR506ddLjjz+uoKAgR6OvadOmio2NLXU7s2fP1rPPPqsnn3xSH3zwgR544AFdeeWVatWqVZnq8fvvv+vzzz+X1Wp1eSwiIkJ9+/bV4sWL9dRTTykrK0vLli3T119/7RT0IyMjVVBQoI8++kg33nhjlXyjzdAduOjZs6fatWunb775RnFxcfLz89Odd94pSVq2bJn69OmjqKgo2Ww2tW7dWk888YROnDjhtA13Q3diYmI0YMAAff755+rYsaNsNptatWqlhISEMtVrypQpio2NVVhYmIKCgtSxY0e9/vrrLoGyPPvZvHmzunXrJl9fX0VHR2vChAll/mrwbDIzMzV+/Hg1adJE3t7eql+/vsaOHevyt3r//fcVGxur4OBg+fn5qWnTpo6/9/r163XZZZdJkkaPHu34+vz0ITRl0blzZ0nS4cOH3T5+5513auPGjdqzZ49j2dq1a/Xnn39q9OjRLuXT0tIUFBSkgIAAt9urrA+wSy+9VN27d3dZXlhYqPr16+uGG25wLJs/f74uvvhiBQQEKDAwUK1atdKTTz5ZKfUoTfFQr6+++kr33HOP6tSpo6CgIN1xxx06ceKEkpOTdfPNNyskJERRUVEaP3682/dYUVGRnnvuOTVq1Ei+vr7q3Lmz2+EFe/fu1a233qp69erJx8dHrVu31iuvvOJS7ueff9a1114rPz8/hYeH6/7779exY8dcyhmGoZkzZ6px48by9fVVx44d9dlnn7mUczd0p/g437Vrl2655RYFBwcrIiJCd955pzIyMpzWT09P11133aWwsDAFBASof//++v333yv8fi5ms9n0+uuvy2q1atasWY7l7obujBo1SgEBAfr555/Vt29f+fv7KyoqSs8//7wk++fBFVdcIX9/f7Vo0UKLFy8+476L9/Hrr7/qs88+cxmyl5iYqNtvv93ptZo9e7aKiooc2yj+u86cOVPTpk1TkyZN5OPjo3Xr1p1x32FhYXriiSf04YcfnnXc7vTp02WxWPTaa6+5bTx4e3vr+uuvP+M2yurIkSN68MEHVb9+fXl7e6tp06aaOHGicnNzncpZLBY9/PDDevPNN9W6dWv5+fnp4osv1ooVK866j/DwcLefL5dffrmysrKchiF99NFHatmypSPkS5KXl5duv/12bdmyRQcPHnSp0xtvvKGWLVvKZrOpc+fO2rx5swzD0KxZs9SkSRMFBAToqquu0q+//lqRPxHKaezYsbrhhhvUpEkTRUdHq379+ho/frwuueQSNW3aVGPGjFHfvn31/vvvn3E71113nR588EE1b95cjz/+uMLDw886tG/FihUKCAiQzWZTs2bNtHv3bj3++ONuy955551atGiRDMPQBx98oGbNmrl8C9+lSxc9+eSTuvXWWxUeHq5+/fpp1qxZbv8vx8XFKSAgwOlWWFj+b2jo0YdbSUlJuv322/XYY49p+vTp8vCwtwn37t2r6667TmPHjpW/v79+/vlnzZgxQ1u2bCnTsIz//e9/evTRR/XEE08oIiJCCxcu1F133aXmzZvryiuvPOO6+/bt03333adGjRpJsv9THjNmjA4ePOj0VW1Z97N7925dffXViomJ0aJFi+Tn56d58+bpnXfeqcifzElWVpZ69OihAwcO6Mknn1SHDh20a9cuPfPMM/rxxx+1du1aWSwWbdq0ScOGDdOwYcM0efJk+fr66s8//3T8LTt27Kg33nhDo0eP1lNPPaX+/ftLUoW+Lfnjjz8kSS1atHD7eO/evdW4cWMlJCRoxowZkuy9YVdeeaXLUB/J3qO4cuVK3Xbbbbrvvvt0+eWXOy7FXZqCggKXZR4eHo73lzujR4/W3//+d+3du9epHqtXr9ahQ4ccjZClS5fqwQcf1JgxY/TCCy/Iw8NDv/76q3bv3n3GOp1NWet8991364YbbtDSpUu1fft2PfnkkyooKNCePXt0ww036N5779XatWs1Y8YMRUdHa9y4cU7rv/zyy2rcuLHmzp2roqIizZw5U/369dPXX3/tCCm7d+9WXFycGjVqpNmzZysyMlJffPGF/va3vyk1NVWTJk2SZG/M9ejRQ1arVfPmzVNERITefvttPfzwwy7PZcqUKZoyZYruuusu3Xjjjdq/f7/uueceFRYWlmlIiCQNHTpUw4YN01133aUff/xREyZMkCRH47qoqMhxjsjkyZMdw2oqa3hfdHS0OnXqpI0bN6qgoMDxlbg7+fn5uuGGG3T//ffrH//4h9555x1NmDBBmZmZWr58uR5//HE1aNBAL730kkaNGqV27dqVOi62+HkMGTJEzZo1c5xnERUVpb/++ktxcXHKy8vTs88+q5iYGK1YsULjx4/Xb7/9pnnz5jlt61//+pdatGihF154QUFBQW6PudP9/e9/18svv6zHHntM33zzjdsyhYWF+uqrr9SpUyfHkIKyOv29f7ZjNScnR7169dJvv/2mKVOmqEOHDvr2228VHx+vHTt2aOXKlU7lV65cqe+//15Tp05VQECAZs6cqSFDhmjPnj1q2rRpueoqSevWrVPdunVVr149x7KdO3e67Sjo0KGDJGnXrl2qX7++Y/mKFSu0fft2Pf/887JYLHr88cfVv39/jRw5Ur///rtefvllZWRkaNy4cRo6dKh27NhRY88zslk9tXtq32rbd2Up7qQqVlhYqOeff17Lli3TwYMHlZubq9zcXKex8O4Uv+aSHEOEUlJSzrhOr169NH/+fGVlZWnhwoX65ZdfNGbMGLdl+/fvr/vuu0/ffPONEhISHB12p3vuuec0btw4ffXVV9q8ebP+/e9/a/r06frmm2/Uvn17R7lly5apdevWTut6elbg72qgVhs5cqTh7+/vtKxHjx6GJOPLL78847pFRUVGfn6+8fXXXxuSjP/973+OxyZNmmSc/vZq3Lix4evra/z555+OZdnZ2UZYWJhx3333lavehYWFRn5+vjF16lSjTp06RlFRUbn3M2zYMMNmsxnJycmOZQUFBUarVq0MScYff/xR5vr06NHDaNu2reN+fHy84eHhYXz//fdO5T744ANDkrFq1SrDMAzjhRdeMCQZ6enppW77+++/NyQZb7zxRpnqsm7dOkOSsWzZMiM/P9/Iysoy/u///s9o2bKl0aZNG+Po0aNO5Uu+ByZNmmRERkYa+fn5RlpamuHj42MsWrTI+OuvvwxJxqRJkxzr5eTkGIMHDzYkGZIMT09P49JLLzUmTpxopKSkuOyjuNzpt6uvvvqMzyc1NdXw9vY2nnzySaflN998sxEREWHk5+cbhmEYDz/8sBESElKmv1FZFB8H7m533XWXo9wbb7xhSDLGjBnjtH7x32bOnDlOyy+55BKjY8eOjvt//PGHIcmIjo42srOzHcszMzONsLAwo3fv3o5lffv2NRo0aGBkZGQ4bfPhhx82fH19jSNHjhiGYRiPP/64YbFYjB07djiVu+aaawxJxrp16wzDMIyjR48avr6+xpAhQ5zK/d///Z8hyejRo4dLPUu+D4uP85kzZzqt/+CDDxq+vr6O43LlypWGJGP+/PlO5eLj413eV+4U73vWrFmllhk2bJghyTh8+LBhGKeOg+Lnahin3ofLly93LMvPzzfq1q1rSDL++9//OpanpaUZnp6exrhx485YN8Owf+b079/fadkTTzxhSDK+++47p+UPPPCAYbFYjD179jg9t2bNmhl5eXln3dfp+1uwYIEhyfj000+dnvf7779vGIZhJCcnG5KM4cOHl2nbhlH6e/+2224743r//ve/DUnGe++957R8xowZhiRj9erVjmWSjIiICCMzM9OxLDk52fDw8DDi4+PLXNdixX+HF1980Wm51Wp1+/9l48aNhiTjnXfecapTZGSkcfz4cceyjz/+2JBkXHLJJU7/Z+bOnWtIMn744Ydy1/V8yM7ONnbv3u30GXKheeONN4zg4GDH/eJjY/v27U7lZsyYYdSpU8d48803jR07dhh79+41+vfvbwwaNMhRpkePHsbf//53x/3GjRsb//znP522c/HFF5/xs2fkyJFO2zQMw+jZs6fx1FNPOS0rue3x48cbPXr0cPo8Dg4OPuP/79zcXKNNmzbGHXfcccbnfSZnev0ZugO3QkND3Z4c8vvvv+vWW29VZGSkPD09ZbVaHScz/fTTT2fd7iWXXOLokZckX19ftWjRQn/++edZ1/3qq6/Uu3dvBQcHO/b9zDPPKC0tzaVVXpb9rFu3TldffbUiIiIcyzw9PTVs2LCz1uVsVqxYoXbt2umSSy5RQUGB49a3b1+n4QTFw3Juvvlmvffee05fI5+rYcOGyWq1ys/PT926dVNmZqZWrlypkJCQUtcZPXq0Dh8+rM8++0xvv/22vL29ddNNN7kt6+Pjo48++ki7d+/WP//5Tw0fPlx//fWXnnvuObVu3dppCJBkH2Lx/fffu9xO79k8XZ06dTRw4EAtXrzYMeTh6NGj+uSTT3THHXc4em8vv/xypaen65ZbbtEnn3zimKrsXDRr1sxtnZ9++mmXsgMGDHC6X9wTU/wtTMnl7t7vN9xwg3x9fR33AwMDNXDgQH3zzTcqLCxUTk6OvvzySw0ZMkR+fn5O76vrrrtOOTk5jiEc69atU9u2bXXxxRc77ePWW291ur9p0ybl5OTotttuc1oeFxenxo0bn+3P43D6sI8OHTooJyfHcVx+/fXXkuzv85JKnjx3rozThvCVxmKx6LrrrnPc9/LyUvPmzRUVFaVLL73UsTwsLEz16tUr02eTO1999ZXatGmjyy+/3Gn5qFGjZBiGyzeg119/vduxv2czevRotWnTRk888YTTkKBz5e69f7bpCr/66iv5+/vrxhtvdFpePEXi6UPRevXq5XR+T0RERIX+5p999pkeeugh3XjjjW57W8/U4376Y7169XLqGS4+jvv16+dUtnh5Rd8fqLhvv/1WgwYN0u23366LL75YTZs21d69e6tk35MmTdILL7ygQ4cOuX38zjvv1Ndff61BgwYpNDS0TNv09vZWs2bNXIb1VhaG7sAtd7NGHD9+XN27d5evr6+mTZumFi1ayM/PT/v379cNN9yg7Ozss263Tp06Lst8fHzOuu6WLVvUp08f9ezZUwsWLFCDBg3k7e2tjz/+WM8995zL+mXZT1pamtsz2ivjLPfDhw/r119/LfUfd3EIvfLKK/Xxxx/rX//6l+644w7l5uaqbdu2mjhx4jmHoBkzZuiqq65SVlaWVq9erfj4eA0ePFjfffddqSf5NW7cWFdffbUSEhK0b98+DR8+XH5+fi7TfpXUunVrxz89wzA0d+5cjRs3Tk8//bTee+89RzkPDw+Xr2DL6s4779Ty5cu1Zs0a9e3bV++++65yc3Od5lgeMWKECgoKtGDBAg0dOlRFRUW67LLLNG3aNF1zzTUV2m/xWPmyCAsLc7rv7e1d6vKcnByX9Ut7L+bl5en48eM6fvy4CgoK9NJLL+mll15yW4fi91VaWpqaNGly1n2kpaWdcd9ldfrxVvz+Kj7e0tLS5OXl5fK3KNnIPld//vmnfHx8XPZxOj8/P6cGlWR/TdytV9prVRZpaWlup2GNjo52PF5SRWfq8fT01PTp0zV48GAtXrzY5XUvnlKyeOheWZXnvV+s+DP19PBcr149eXl5uTzniv4/KOmLL77QDTfcoGuuuUZvv/22y77r1Knjsl9JjnH8p7/u5TmOJVX4/YGKa968uZYvX66NGzcqNDRUc+bMUXJyssswl/OhZ8+eatu2raZPn66XX37Z5fHWrVsrNTW11MlLVqxYoaVLl2r48OFq0aKFDMPQp59+qlWrVrlMz5mWlqbk5GSnZSEhIS6fX2dD0Idb7npAvvrqKx06dEjr1693mpIsPT39vNdn6dKlslqtWrFihdOb/OOPP67wNuvUqeNyEElyu6y8wsPDZbPZSj3RODw83PH7oEGDNGjQIOXm5mrz5s2Kj4/XrbfeqpiYGKcTyMqradOmjn/UV155pWw2m5566im99NJLGj9+fKnr3Xnnnbr99ttVVFSk+fPnl2ufFotFjzzyiKZOnVqp82337dtX0dHReuONN9S3b1+98cYbio2NdZk6cfTo0Ro9erROnDihb775RpMmTdKAAQP0yy+/lKuHujqU9l709vZWQECArFarPD09NWLECD300ENut1Ec8sr63i4OWqWVrazrBdSpU0cFBQU6cuSIU2CqjGNNkg4ePKht27apR48eZxyfX5Xq1KmjpKQkl+XFPYElPwOkczt5fdCgQerWrZsmTZqk1157zekxT09PXX311frss8904MCB8zobWp06dfTdd9/JMAyn55OSkqKCggKX53yuvvjiCw0ePFg9evTQ8uXLHeG7pPbt2+vHH390WV68rF27dpVaJ5x/Tz/9tP744w/17dtXfn5+uvfeezV48GCXCQDOl3Hjxmn06NF6/PHH3Z734q4BW6xNmzby8/PTo48+qv3798vHx0cXXXSRFi5cqBEjRjiV7d27t8v67777roYPH16u+taMT0RcEIo/uE/vDX711VerZN9eXl5OJ6JkZ2frzTffrPA2e/Xqpf/85z86fPiwo2exsLBQy5YtO+f6DhgwQNOnT1edOnXc9qy64+Pjox49eigkJERffPGFtm/frq5du7r0jlbUY489pkWLFun555/XfffdV+rFRIYMGaIhQ4YoODj4jNPrJSUlue2FPHTokDIzMyt0YY/SFAfcuXPn6ttvv9XWrVvP+L7z9/dXv379lJeXp8GDB2vXrl01Puh/+OGHmjVrlqMhe+zYMX366afq3r27PD095efnp169emn79u3q0KGD21BTrFevXpo5c6b+97//OQ3fOf1E8y5dusjX11dvv/22hg4d6li+ceNG/fnnn5UW9Hv06KGZM2dq2bJleuCBBxzL3V1Jsryys7N19913q6CgQI899tg5b6+yXH311YqPj9d///tfdezY0bF8yZIlslgs6tWrV6Xub8aMGbriiiv0r3/9y+WxCRMmaNWqVbrnnnv0ySefuLx38vPz9fnnn2vgwIHnVIerr75a7733nj7++GMNGTLEsbx4esGrr776nLZf0urVqzV48GBdccUVLlOGljRkyBA9+OCD+u677xxTLxYUFOitt95SbGys4xsWVL9Ro0Y5fUsbExPjdkheWFjYWTv5Tp9Nx93FK892EcrSLg546623Og2DPNuFMUt2hjZt2tSlMX660p53RRH0UWZxcXEKDQ3V/fffr0mTJslqtertt9/W//73v/O+7/79+2vOnDm69dZbde+99yotLU0vvPDCOV246KmnntJ//vMfXXXVVXrmmWfk5+enV155pVLGyY0dO1bLly/XlVdeqUceeUQdOnRQUVGREhMTtXr1aj366KOKjY3VM888owMHDujqq69WgwYNHBd1KXnuQ7NmzWSz2fT222+rdevWCggIUHR0dLn/QVmtVk2fPl0333yzXnzxRT311FNuy/n6+pZ6hcmS7r33XqWnp2vo0KFq166dPD099fPPP+uf//ynPDw8XKYgKyoqKnUawEsvvfSsr+Wdd96pGTNm6NZbb5XNZnM5l+Kee+6RzWZTt27dFBUVpeTkZMXHxys4ONhxLsSff/6pZs2aaeTIkXr99dfP+hyzs7NLrXNlzzHu6empa665RuPGjVNRUZFmzJihzMxMp2sevPjii7riiivUvXt3PfDAA4qJidGxY8f066+/6tNPP3WM+x47dqwSEhLUv39/TZs2zTHrzs8//+y0z9DQUI0fP17Tpk3T3XffrZtuukn79+/X5MmTK2UIW7Frr71W3bp106OPPupoBG7atMkRAM80k0tJiYmJ2rx5s4qKipSRkeG4YNaff/6p2bNnq0+fPpVW53P1yCOPaMmSJerfv7+mTp2qxo0ba+XKlZo3b54eeOCBUme/qqhu3bpp0KBB+uSTT1we69q1q+bPn68HH3xQnTp10gMPPKC2bds6Lqb32muvqV27ducc9O+44w698sorGjlypPbt26f27dtrw4YNmj59uq677jq3PZQVsWHDBg0ePFiRkZF68sknXQJbmzZtHBfru/POO/XKK6/opptu0vPPP6969epp3rx52rNnj9auXVsp9QFqMoI+yqxOnTpauXKlHn30Ud1+++3y9/fXoEGDtGzZMqceq/Phqquuckz7OHDgQNWvX1/33HOP6tWr53Il17Jq166d1q5dq0cffVQjR45UaGioRowYoaFDh571anln4+/vr2+//VbPP/+8XnvtNf3xxx+y2Wxq1KiRevfu7egpjY2N1datW/X444/rr7/+UkhIiDp37qyvvvpKbdu2lWQfU5yQkKApU6aoT58+ys/P16RJkyo09/hNN92k2NhYzZkzR2PGjFFwcHCFn+OYMWO0bNkyLViwQAcPHtSJEydUt25dde3aVUuWLHEJwtnZ2aUORdq7d6+aN29+xv21aNFCcXFx2rhxo2677TaXunfv3l2LFi3Se++9p6NHjyo8PFxXXHGFlixZorp160qyn0NQWFhY5rmIf//991LrnJ+fX6nDRB5++GHl5OTob3/7m1JSUtS2bVutXLlS3bp1c5Rp06aN/vvf/+rZZ5/VU089pZSUFIWEhOiiiy5yOsE0MjJSX3/9tf7+97/rgQcekJ+fn4YMGaKXX35ZgwYNctrv1KlT5e/vr3nz5unNN99Uq1at9O9//9sxVWRl8PDw0KeffqpHH31Uzz//vPLy8tStWze99dZb6tKlyxlPEC+p+PwET09PBQUFqWnTpho4cKDuueeeKrkCbnnUrVtXGzdu1IQJExzTdzZt2lQzZ850mVq1ssTHx2vFihVu39/33HOPLr/8cv3zn//UjBkzlJycLKvVqhYtWujWW291O/Vqefn6+mrdunWaOHGiZs2apb/++ssx53nx1K+VYe3atcrOzta+ffvcThqxbt069ezZU5L9m9Ivv/xSjz32mMaMGaOsrCxdcskl+uyzz0x9VVygmMWozO8HAAAoo3feeUe33Xab/u///k9xcXHVXR3ggpWTk6M//vhDTZo0KffJmrjwnen1p0cfAHDevfvuuzp48KDat28vDw8Pbd68WbNmzdKVV15JyAeA84SgDwA47wIDA7V06VJNmzZNJ06cUFRUlEaNGqVp06ZVd9UAwLQI+gCA827AgAEuFxUDAJxfXBkXAAAAMCGCPgAAAGBCBH0AAADAhAj6AAAAqHIWi+WMt5JXyi2vmJgYzZ07t0zlivdns9nUqlUrzZo1y+nqtPv27ZPFYpGXl5cOHjzotH5SUpK8vLxksVicrpK7fPlyxcbGKjg4WIGBgWrbtq0effRRx+OLFi1y+5wre3pUTsYFAABAlUtKSnL8vmzZMj3zzDPas2ePY5nNZquSekydOlX33HOPcnJytHbtWj3wwAMKCgrSfffd51QuOjpaS5Ys0YQJExzLFi9erPr16ysxMdGxbO3atRo+fLimT5+u66+/XhaLRbt379aXX37ptL2goCCn5yvZGz+ViR59AAAAVLnIyEjHLTg4WBaLxWnZN998o06dOsnX11dNmzbVlClTVFBQ4Fh/8uTJatSokXx8fBQdHa2//e1vkqSePXvqzz//1COPPOLoKT+TwMBARUZGKiYmRnfffbc6dOig1atXu5QbOXKk3njjDadlixYt0siRI52WrVixQldccYX+8Y9/qGXLlmrRooUGDx6sl156yanc6c83MjJSERER5fobng1BHwAAwGwMQ8o7UT23EsNeKuqLL77Q7bffrr/97W/avXu3Xn31VS1atEjPPfecJOmDDz7QP//5T7366qvau3evPv74Y7Vv316S9OGHH6pBgwaaOnWqkpKSnL45OPOfzND69ev1008/yWq1ujx+/fXX6+jRo9qwYYMkacOGDTpy5IgGDhzoVC4yMlK7du3Szp07z+VPUCkYugMAAGA2+VnS9Ojq2feThyRv/3PaxHPPPacnnnjC0VvetGlTPfvss3rsscc0adIkJSYmKjIyUr1795bValWjRo10+eWXS5LCwsLk6enp6Kk/m8cff1xPPfWU8vLylJ+fL19fX8e3AyVZrVbdfvvtSkhI0BVXXKGEhATdfvvtLo2CMWPG6Ntvv1X79u3VuHFjdenSRX369NFtt90mHx8fR7mMjAwFBAQ4rRsXF+f224SKIugDAACgRtm2bZu+//57Rw++JBUWFionJ0dZWVm66aabNHfuXDVt2lTXXnutrrvuOg0cOFBeXuWPtv/4xz80atQo/fXXX5o4caKuuuoqxcXFuS171113qWvXrpo+fbref/99bdq0yWk4kST5+/tr5cqV+u2337Ru3Tpt3rxZjz76qF588UVt2rRJfn5+kuxDhv773/86rVvZ5yUQ9AEAAMzG6mfvWa+ufZ+joqIiTZkyRTfccIPLY76+vmrYsKH27NmjNWvWaO3atXrwwQc1a9Ysff31126H3ZxJeHi4mjdvrubNm2v58uVq3ry5unTpot69e7uUbdeunVq1aqVbbrlFrVu3Vrt27bRjxw63223WrJmaNWumu+++WxMnTlSLFi20bNkyjR49WpLk4eGh5s2bl6uu5UXQBwAAMBuL5ZyHz1Snjh07as+ePWcMwjabTddff72uv/56PfTQQ2rVqpV+/PFHdezYUd7e3iosLCz3fkNDQzVmzBiNHz9e27dvd3si75133qkHH3xQ8+fPL/N2Y2Ji5OfnpxMnTpS7TueCoA8AAIAa5ZlnntGAAQPUsGFD3XTTTfLw8NAPP/ygH3/8UdOmTdOiRYtUWFio2NhY+fn56c0335TNZlPjxo0l2YP1N998o+HDh8vHx0fh4eFl3vdDDz2kGTNmaPny5brxxhtdHr/nnnt00003KSQkxO36kydPVlZWlq677jo1btxY6enp+te//qX8/Hxdc801jnKGYSg5Odll/Xr16snDo3Lmy2HWHQAAANQoffv21YoVK7RmzRpddtll6tKli+bMmeMI8iEhIVqwYIG6deumDh066Msvv9Snn36qOnXqSLLPjb9v3z41a9ZMdevWLde+69atqxEjRmjy5MkqKipyedzLy0vh4eGlng/Qo0cP/f7777rjjjvUqlUr9evXT8nJyVq9erVatmzpKJeZmamoqCiXW0pKSrnqeyYWw6iEOZAAAABQLXJycvTHH3+oSZMmlX5lVdR8Z3r96dEHAAAATIigDwAAAJgQQR8AAAAwIYI+AAAAYEIEfQAAABNgfpXa6UyvO0EfAADgAlZ8JdisrKxqrgmqQ/Hr7u6KwFwwCwAA4ALm6empkJAQx/zrfn5+bq/oCnMxDENZWVlKSUlRSEiIPD09Xcowjz4AAMAFrvgqq+np6dVdFVSxkJAQRUZGum3cEfQBAABMorCwUPn5+dVdDVQRq9Xqtie/GEEfAAAAMCFOxgUAAABMiKAPAAAAmBBBHwAAADAhgj4AAABgQgR9AAAAwIQI+gAAAIAJEfQBAAAAEyLoAwAAACZE0AcAAABMiKAPAAAAmBBBHwAAADAhgj4AAABgQgR9AAAAwIQI+gAAAIAJEfQBAAAAEyLoAwAAACZE0AcAAABMiKAPAAAAmJBXdVegKhUVFenQoUMKDAyUxWKp7uoAAADgNIZh6NixY4qOjpaHB33S56JWBf1Dhw6pYcOG1V0NAAAAnMX+/fvVoEGD6q7GBa1WBf3AwEBJ9jdOUFBQNdcGAAAAp8vMzFTDhg0duQ0VV6uCfvFwnaCgIII+AABADcYw63PHwCcAAADAhAj6AAAAgAkR9AEAAAATIugDAAAAJkTQBwAAAEyIoA8AAACYEEEfAAAAMCGCPgAAAGBCBH0AAADAhAj6AAAAgAkR9AEAAAATIugDAAAAJkTQBwAAAEyIoA8AAACYEEEfAAAAMCGCPgAAAGBCBP3zKS+rumsAAACAWoqgf76cSJNe7S5984JkGNVdGwAAANQyBP3zZfdHUtqv0lfPSqvGS0WF1V0jAACA8isqkgrzq7sWqACv6q6AaV12t1RYIH3+hPT9QulYsjR0oWS1VXfNAAAATikqkk6kSOmJJ29/lvg9UUrfL103S+o0srprinIi6J9PXe6XAiOkD++Vfl4hLRkk3bJU8gur7poBAIDaoixBvjD3zNtI/7Nq6opKRdA/39oOkfzrSktvlfZ/JyX0lW5fLoU0qu6aAQAAMygqko4fLj3IZ+yXCvPOvA2LhxTUwJ5P3N2CoqvmuaBSEfSrQswV0p1fSG8NlVJ/kRZeI93+gRTZvrprBgAAarqqCvKe1qp5PqgyBP2qUq+1dNca6e0bpZTdUkI/afjbUtMe1V0zAABQnQjyOE8shlF75n7MzMxUcHCwMjIyFBQUVD2VyE6Xlt4m/blB8rBKQ/4ttb+xeuoCAADOv0oJ8p5SUP1aEeRrRF4zCXr0q5otxD5G/6P7pN0fS8vvko4lSXFjqrtmAACgIioryAfXl0Iauw/ygdGSJ7EN5cM7pjpYfaUb35C+iJK+my+tfkrKTJL6TJM8uLQBAAA1CkEeFyjeUdXFw0O6Nt7+Vduap6XNr0jHDklDXpW8fKq7dgAA1B4EeZgU77jqZLFI3f4mBUZJHz8g7fpIOpEqDXvLPsQHAACcO4I8ainekTVBh5ukgLrS0tulfd9Kb1xnn36TOWsBADg7gjzgFrPu1CRJP9in3zx+2D5F1u0f2KflBACgNisqko4nnyHIHyDIm0iNz2sXEN7RNUlUB/tc+28NldL22q+ie8tSqXFcddcMAIDzhyAPnBf06NdEWUekd4dL+7+TPH2kG16T2g6u7loBAFAxBHmUwwWT1y4AHBE1kV+YdMcn0gd3SXtWSu+Pko7PkGLvq+6aAQDgiiAP1EgcMTWV1SYNe1Na9Q9p6+vSZ49JmYekqycx1z4AoGpVWpBvcDK4uwnzgVEEeaCSVeiImjdvnmbNmqWkpCS1bdtWc+fOVffu3Ust/8orr+jll1/Wvn371KhRI02cOFF33HGH4/EPP/xQ06dP16+//qr8/HxddNFFevTRRzVixAhHmcmTJ2vKlClO242IiFBycnJFnsKFwcNT6j/bPvvOV89K/zfXfhXd61+WvLyru3YAALMgyAOmVO4jbtmyZRo7dqzmzZunbt266dVXX1W/fv20e/duNWrUyKX8/PnzNWHCBC1YsECXXXaZtmzZonvuuUehoaEaOHCgJCksLEwTJ05Uq1at5O3trRUrVmj06NGqV6+e+vbt69hW27ZttXbtWsd9T0/PijznC4vFIl053v4B+Z8x0g/LpOMp9t5+n8Dqrh0A4EJwtiCfvl8qyj/zNgjywAWn3CfjxsbGqmPHjpo/f75jWevWrTV48GDFx8e7lI+Li1O3bt00a9Ysx7KxY8dq69at2rBhQ6n76dixo/r3769nn31Wkr1H/+OPP9aOHTvKXNfc3Fzl5uY67mdmZqphw4YX7skde9dK790h5Z+QIjtIt30gBUZUd60AANWNIA8T4WTcylOuIzYvL0/btm3TE0884bS8T58+2rhxo9t1cnNz5evr67TMZrNpy5Ytys/Pl9VqdXrMMAx99dVX2rNnj2bMmOH02N69exUdHS0fHx/FxsZq+vTpatq0aan1jY+Pdxnuc0G7qLc0aoX0zs1S8g/S672l2z+Uwi+q7poBAM4ngjyACijXEZ2amqrCwkJFRDj3Ip9prHzfvn21cOFCDR48WB07dtS2bduUkJCg/Px8paamKioqSpKUkZGh+vXrKzc3V56enpo3b56uueYax3ZiY2O1ZMkStWjRQocPH9a0adMUFxenXbt2qU6dOm73PWHCBI0bN85xv7hH/4JWv6N012r7XPtHfpde7yPd+p7U8LLqrhkAoKKKCqVjyaeF9z+dx8gT5AGUU4WOeIvF4nTfMAyXZcWefvppJScnq0uXLjIMQxERERo1apRmzpzpNMY+MDBQO3bs0PHjx/Xll19q3Lhxatq0qXr27ClJ6tevn6Ns+/bt1bVrVzVr1kyLFy92CvMl+fj4yMfHpyJPsWYLayrdudres3/ov9LigdKNCVKr66q7ZgAAdyojyHt4SUH1CfIAyqxcnwjh4eHy9PR06b1PSUlx6eUvZrPZlJCQoFdffVWHDx9WVFSUXnvtNQUGBio8PNxRzsPDQ82bN5ckXXLJJfrpp58UHx/vCPqn8/f3V/v27bV3797yPAXzCKhrH8bz/ihp72pp2W1S/zlS59HVXTMAqH0I8gBqoHJ9Ynh7e6tTp05as2aNhgwZ4li+Zs0aDRo06IzrWq1WNWjQQJK0dOlSDRgwQB5nmA/eMAynE2lPl5ubq59++umM03qanre/NPxdacXfpe1vSSvG2qff7DnBPlsPAKByVFaQdwytcRPmA6Ps0yoDQCUpd9fAuHHjNGLECHXu3Fldu3bVa6+9psTERN1///2S7OPiDx48qCVLlkiSfvnlF23ZskWxsbE6evSo5syZo507d2rx4sWObcbHx6tz585q1qyZ8vLytGrVKi1ZssRpZp/x48dr4MCBatSokVJSUjRt2jRlZmZq5MiR5/o3uLB5etnn1Q+qL309w37LPCQNmEvPDwCUFUEegAmVOwkOGzZMaWlpmjp1qpKSktSuXTutWrVKjRs3liQlJSUpMTHRUb6wsFCzZ8/Wnj17ZLVa1atXL23cuFExMTGOMidOnNCDDz6oAwcOyGazqVWrVnrrrbc0bNgwR5kDBw7olltuUWpqqurWrasuXbpo8+bNjv3WahaL1OtJ+z+RleOk7W9Kxw9LNy2y9/oDQG1HkAdQC5V7Hv0LWa2Yl/XnVdIHd0oF2VJ0R/uMPAF1q7tWAHB+EeQB06gVea2KEPTNaP8W6Z1hUvYR+ww9ty+3/wSACxVBHqg1ak1eqwIM4jajhpefnGv/Bvtc+wuvkW573z4HPwDURAR5AKh0BH2zCr9Iumut9PaN9qvoLhog3bxYuuias68LAJWtqNA+K9gZg3zBmbdBkAeAcmHojtnlHpOWjZB+X2e/auL1L0mX3lbdtQJgNgR5AJWkVua184QefbPzCbSfkPufh6UflkmfPCgdOyR1H89c+wDKjiAPABccgn5t4OUtDXlVCoqWNvxT+mqalJkkXTeLf6oA7CotyDc8Q5CP5DMHAKoQQb+2sFik3pOlwGjps8ekra/b59ofulCy2qq7dgDOt0oJ8taz9MgT5AGgJiHo1zax90qBEdLye6SfV0hLBkm3LJX8wqq7ZgDOBUEeAHAagn5t1GaQ5F9Xene4tP87KaGvfa79kEbVXTMApSHIAwDKiaBfWzWOk+78QnprqJT6y6m59qM6VHfNgNqJIA8AqGQE/dqsXmvprjX2ufZTdktvXCcNf0tq2rO6awaYD0EeAFDFmEcfUna6tOx2ad+39iAxeL7U4abqrhVwYSkqlDIPnRbkS4T5zIMEeQAoA/Ja5aFHH5ItxD5G/6P7pF0fSR/ebe95jBvDXPtAMYI8AOACQ9CHnZePNDTBfsGazfOkNU/bQ03f6ZKHR3XXDqg6+dn2oWzJO6XkH6W/fibIAwAuSAR9nOLhIV0bLwXVl1ZPlL6bb+/ZH/KqZPWt7toBlcsw7NeSSP7Rfjt8Mtin/SoZRe7X8bBKIQ2dw3vJMB8QQZAHANQYBH24invY3vP40f3S7o+lE6nS8LftQ3yAC1Fhvn12qeSdUvIPJ0P9Tikr1X15v3Apsr0U2U6q11YKa0KQBwBccAj6cK/9jfa59pfdLv25QUq41j6OP7h+ddcMOLOsI6eC/OESw28K81zLWjykOhfZA31keyniZLgPiOD8FADABY9Zd3BmyT9Kb90oHU+2D+m57QMpok111wqQioqko384D7tJ3illHnBf3idIimhnD/IRJ4N9vdaS1Va19QYAnBF5rfLQo48zi2wv3b3GHvZT90hvXCsNf1eK6VbdNUNtkndCOrzbedjN4V1S/gn35UMan+yhb3dqCE5IY3rpAQC1Cj36KJusI9K7t0j7N0ue3tINC6S2g6u7VjAbw7DPbpO8Uzr846mZb478LsnNR5WXr71XvuSwm4i2km9wlVcdAFA5yGuVhx59lI1fmHTHx9Lyu6WfV0jvj5KOz5Bi76vumuFCVZBnHzvvGHZzcghO9lH35QMiTuulby+FNZM8+RgDAMAd/kOi7Kw26eYl0mePSd8vtP/MPChdPZm59nFmJ1JLjKU/eZLsXz+7n5fe4inVbek87CaivRRQt+rrDQDABYygj/Lx8JSue0EKipa+nCr934tSZpI06BXJy7u6a4fqVlQopf3mPOzm8E779Rjc8Q2WIjs4nyRbtxXXbQAAoBIQ9FF+FovU/VH7VXT/M0b68T3pRIp085uSL2Ppao3cY/YTYksOuzm8WyrIdl8+rKnzsJuIdvaryHKCLAAA5wVBHxV3ya2Sfz3pvTuk39dLi66zT78ZGFndNUNlMgwpPbHEsJuTwf7oPvflrX5SvTbOw24i2kg+gVVabQAAajtm3cG5O7Rdevsm6cRf9quH3v6hFH5RddcKFZGfI/31k/Owm8M7pZwM9+WD6p82N30H+1VkuXosAKCCyGuVhx59nLvoS6W71khv3WCfBvH1a6Rb35MaXl7dNcOZHE+xz0tf8gqyqXslo9C1rIfVPnbecQXZkz/9wqq+3gAAoEwI+qgcYU3sYf+dm6WD26TFA6UbE6RW/au7ZigskNL2Og+7Sd5pP6/CHVuY8zj6yHZSeEtOtgYA4ALD0B1UrrwT0gd3Sr98Llk8pP6zpc53Vnetao/s9FMnyBbPfJPyk1SY66awRarT3HnYTWQ7+0nWnCALAKgm5LXKQ48+Kpe3vzTsbWnlI9J/l0grHpEyD0m9JhIeK1NRkZS+z3nYTfJOKSPRfXnvAPsVY0sOu6nX2v56AQAAUyLoo/J5ekkD/2U/UXN9vPTNLPtc+wPnSp7W6q7dhScvy94rn/xDiZlvdkl5x9yXD25Uopf+ZKgPieGiZgAA1DIEfZwfFovU8wn7VJsrHpF2vCUdPyzdtEjyCaju2tVMhiEdS3YedpP8o3TkN8koci3v6SPVa2WfvtIxlWVbyRZa9XUHAAA1DkEf51enUVJApPT+KOnXNdLiAdKt70sBdau7ZtWrMF/6a0+JYTcnp7LMSnNf3r+u87CbyPb28fV8QwIAAErBybioGge22ufazz4ihTaRbl8u1WlW3bWqGllHTgX54plvUn6WivJdy1o87dcgKDnsJqK9FBhR9fUGAKAakNcqDz36qBoNOp+aa//oH9LrfaTb3pPqd6rumlWeoiL7dQRKDrs5vFPKPOi+vE9QiR76k2Pq67WWrLaqrTcAADAlgj6qTnjzk3Pt3yQl/U9aNEC6eYl00TXVXbPyyz0upex2HnZzeLeUf8J9+dCYU6G++GdII2YiAgAA5w1Dd1D1co9J790h/faVfajK9f+SLr29umvlnmHYe+SLp68svuDUkT8kuTl0vHylem1ODrvpYA/1EW0lX95vAACUBXmt8tCjj6rnEyjdskz6zxjph6XSJw/Zp9+8cnz19nAX5Ep//ew87Cb5Rykn3X35gEjnYTeRHeznHXh4Vmm1AQAA3CHoo3p4eUtD/i0FRUsb5kjrpknHDknXvVA1QflEqvOwm+SdUuoeqajAtayHlxTeskSgPznrjX/4+a8nAABABRH0UX0sFqn3JCkwSvrsMWlrgnTssDR0oeTtVzn7KCqU0n517qFP3ikdT3Zf3jfkVJAvnvmmbivJy6dy6gMAAFBFCPqofrH32i+stfxuac9Kackg6dZlkl9Y+baTk2m/YmzJC06l/CQVZLspbJHCmp7spW9/airLoPqcIAsAAEyBoI+aoc31kv8n0rvDpQNb7NNv3r5cCm3sWtYwpPTE03rpf5TS/3S/bauf/YTYksNu6rXhCr0AAMDUCPqoORp3le78QnprqJS2V3r9GmnY2/Yx+yWH3RzeJeVmuN9GUH3nYTcR7aWwJpwgCwAAah2m10TNk3nIfhXdwztLL+Nhleq1ch52E9Gu/MN9AABAjUJeqzz06KPmCYqWRq+S3h8t/fal5FfHedhNRDspvIV95h4AAAC4RdBHzeQbLI340H5xLe8ATpAFAAAoJ4I+ajafwOquAQAAwAXJo7orAAAAAKDyEfQBAAAAE6pQ0J83b56aNGkiX19fderUSd9+++0Zy7/yyitq3bq1bDabWrZsqSVLljg9/uGHH6pz584KCQmRv7+/LrnkEr355pvnvF8AAACgtip30F+2bJnGjh2riRMnavv27erevbv69eunxMREt+Xnz5+vCRMmaPLkydq1a5emTJmihx56SJ9++qmjTFhYmCZOnKhNmzbphx9+0OjRozV69Gh98cUXFd4vAAAAUJuVex792NhYdezYUfPnz3csa926tQYPHqz4+HiX8nFxcerWrZtmzZrlWDZ27Fht3bpVGzZsKHU/HTt2VP/+/fXss89WaL/uMC8rAABAzUZeqzzl6tHPy8vTtm3b1KdPH6flffr00caNG92uk5ubK19fX6dlNptNW7ZsUX5+vkt5wzD05Zdfas+ePbryyisrvN/ifWdmZjrdAAAAgNqgXEE/NTVVhYWFioiIcFoeERGh5ORkt+v07dtXCxcu1LZt22QYhrZu3aqEhATl5+crNTXVUS4jI0MBAQHy9vZW//799dJLL+maa66p8H4lKT4+XsHBwY5bw4YNy/N0AQAAgAtWhU7GtZx28SLDMFyWFXv66afVr18/denSRVarVYMGDdKoUaMkSZ6eno5ygYGB2rFjh77//ns999xzGjdunNavX1/h/UrShAkTlJGR4bjt37+/HM8SAAAAuHCVK+iHh4fL09PTpRc9JSXFpbe9mM1mU0JCgrKysrRv3z4lJiYqJiZGgYGBCg8PP1URDw81b95cl1xyiR599FHdeOONjrH3FdmvJPn4+CgoKMjpBgAAANQG5Qr63t7e6tSpk9asWeO0fM2aNYqLizvjularVQ0aNJCnp6eWLl2qAQMGyMOj9N0bhqHc3Nxz3i8AAABQG3mVd4Vx48ZpxIgR6ty5s7p27arXXntNiYmJuv/++yXZh8scPHjQMVf+L7/8oi1btig2NlZHjx7VnDlztHPnTi1evNixzfj4eHXu3FnNmjVTXl6eVq1apSVLljjNsHO2/QIAAAA4pdxBf9iwYUpLS9PUqVOVlJSkdu3aadWqVWrcuLEkKSkpyWlu+8LCQs2ePVt79uyR1WpVr169tHHjRsXExDjKnDhxQg8++KAOHDggm82mVq1a6a233tKwYcPKvF8AAAAAp5R7Hv0LGfOyAgAA1GzktcpToVl3AAAAANRsBH0AAADAhAj6AAAAgAkR9AEAAAATIugDAAAAJkTQBwAAAEyIoA8AAACYEEEfAAAAMCGCPgAAAGBCBH0AAADAhAj6AAAAgAkR9AEAAAATIugDAAAAJkTQBwAAAEyIoA8AAACYEEEfAAAAMCGCPgAAAGBCBH0AAADAhAj6AAAAgAkR9AEAAAATIugDAAAAJkTQBwAAAEyIoA8AAACYEEEfAAAAMCGCPgAAAGBCBH0AAADAhAj6AAAAgAkR9AEAAAATIugDAAAAJkTQBwAAAEyIoA8AAACYEEEfAAAAMCGCPgAAAGBCBH0AAADAhAj6AAAAgAkR9AEAAAATIugDAAAAJkTQBwAAAEyIoA8AAACYEEEfAAAAMCGCPgAAAGBCBH0AAADAhAj6AAAAgAkR9AEAAAATIugDAAAAJkTQBwAAAEyIoA8AAACYEEEfAAAAMCGCPgAAAGBCFQr68+bNU5MmTeTr66tOnTrp22+/PWP5V155Ra1bt5bNZlPLli21ZMkSp8cXLFig7t27KzQ0VKGhoerdu7e2bNniVGby5MmyWCxOt8jIyIpUHwAAADC9cgf9ZcuWaezYsZo4caK2b9+u7t27q1+/fkpMTHRbfv78+ZowYYImT56sXbt2acqUKXrooYf06aefOsqsX79et9xyi9atW6dNmzapUaNG6tOnjw4ePOi0rbZt2yopKclx+/HHH8tbfQAAAKBWsBiGYZRnhdjYWHXs2FHz5893LGvdurUGDx6s+Ph4l/JxcXHq1q2bZs2a5Vg2duxYbd26VRs2bHC7j8LCQoWGhurll1/WHXfcIcneo//xxx9rx44d5amuk8zMTAUHBysjI0NBQUEV3g4AAADOD/Ja5SlXj35eXp62bdumPn36OC3v06ePNm7c6Had3Nxc+fr6Oi2z2WzasmWL8vPz3a6TlZWl/Px8hYWFOS3fu3evoqOj1aRJEw0fPly///77Geubm5urzMxMpxsAAABQG5Qr6KempqqwsFARERFOyyMiIpScnOx2nb59+2rhwoXatm2bDMPQ1q1blZCQoPz8fKWmprpd54knnlD9+vXVu3dvx7LY2FgtWbJEX3zxhRYsWKDk5GTFxcUpLS2t1PrGx8crODjYcWvYsGF5ni4AAABwwarQybgWi8XpvmEYLsuKPf300+rXr5+6dOkiq9WqQYMGadSoUZIkT09Pl/IzZ87Uu+++qw8//NDpm4B+/fpp6NChat++vXr37q2VK1dKkhYvXlxqPSdMmKCMjAzHbf/+/eV9qgAAAMAFqVxBPzw8XJ6eni699ykpKS69/MVsNpsSEhKUlZWlffv2KTExUTExMQoMDFR4eLhT2RdeeEHTp0/X6tWr1aFDhzPWxd/fX+3bt9fevXtLLePj46OgoCCnGwAAAFAblCvoe3t7q1OnTlqzZo3T8jVr1iguLu6M61qtVjVo0ECenp5aunSpBgwYIA+PU7ufNWuWnn32WX3++efq3LnzWeuSm5urn376SVFRUeV5CgAAAECt4FXeFcaNG6cRI0aoc+fO6tq1q1577TUlJibq/vvvl2QfLnPw4EHHXPm//PKLtmzZotjYWB09elRz5szRzp07nYbczJw5U08//bTeeecdxcTEOL4xCAgIUEBAgCRp/PjxGjhwoBo1aqSUlBRNmzZNmZmZGjly5Dn/EQAAAACzKXfQHzZsmNLS0jR16lQlJSWpXbt2WrVqlRo3bixJSkpKcppTv7CwULNnz9aePXtktVrVq1cvbdy4UTExMY4y8+bNU15enm688UanfU2aNEmTJ0+WJB04cEC33HKLUlNTVbduXXXp0kWbN2927BcAAADAKeWeR/9CxrysAAAANRt5rfJUaNYdAAAAADUbQR8AAAAwIYI+AAAAYEIEfQAAAMCECPoAAACACRH0AQAAABMi6AMAAAAmRNAHAAAATIigDwAAAJgQQR8AAAAwIYI+AAAAYEIEfQAAAMCECPoAAACACRH0AQAAABMi6AMAAAAmRNAHAAAATIigDwAAAJgQQR8AAAAwIYI+AAAAYEIEfQAAAMCECPoAAACACRH0AQAAABMi6AMAAAAmRNAHAAAATIigDwAAAJgQQR8AAAAwIYI+AAAAYEIEfQAAAMCECPoAAACACRH0AQAAABMi6AMAAAAmRNAHAAAATIigDwAAAJgQQR8AAAAwIYI+AAAAYEIEfQAAAMCECPoAAACACRH0AQAAABMi6AMAAAAmRNAHAAAATIigDwAAAJgQQR8AAAAwIYI+AAAAYEIEfQAAAMCECPoAAACACRH0AQAAABMi6AMAAAAmRNAHAAAATIigDwAAAJgQQR8AAAAwoQoF/Xnz5qlJkyby9fVVp06d9O23356x/CuvvKLWrVvLZrOpZcuWWrJkidPjCxYsUPfu3RUaGqrQ0FD17t1bW7ZsOef9AgAAALVVuYP+smXLNHbsWE2cOFHbt29X9+7d1a9fPyUmJrotP3/+fE2YMEGTJ0/Wrl27NGXKFD300EP69NNPHWXWr1+vW265RevWrdOmTZvUqFEj9enTRwcPHqzwfgEAAIDazGIYhlGeFWJjY9WxY0fNnz/fsax169YaPHiw4uPjXcrHxcWpW7dumjVrlmPZ2LFjtXXrVm3YsMHtPgoLCxUaGqqXX35Zd9xxR4X2K0m5ubnKzc113M/MzFTDhg2VkZGhoKCg8jxtAAAAVIHMzEwFBweT1ypBuXr08/LytG3bNvXp08dpeZ8+fbRx40a36+Tm5srX19dpmc1m05YtW5Sfn+92naysLOXn5yssLKzC+5Wk+Ph4BQcHO24NGzY863MEAAAAzKBcQT81NVWFhYWKiIhwWh4REaHk5GS36/Tt21cLFy7Utm3bZBiGtm7dqoSEBOXn5ys1NdXtOk888YTq16+v3r17V3i/kjRhwgRlZGQ4bvv37y/P0wUAAAAuWF4VWclisTjdNwzDZVmxp59+WsnJyerSpYsMw1BERIRGjRqlmTNnytPT06X8zJkz9e6772r9+vUu3wSUZ7+S5OPjIx8fn7I+LQAAAMA0ytWjHx4eLk9PT5de9JSUFJfe9mI2m00JCQnKysrSvn37lJiYqJiYGAUGBio8PNyp7AsvvKDp06dr9erV6tChwzntFwAAAKjNyhX0vb291alTJ61Zs8Zp+Zo1axQXF3fGda1Wqxo0aCBPT08tXbpUAwYMkIfHqd3PmjVLzz77rD7//HN17ty50vYLAAAA1EblHrozbtw4jRgxQp07d1bXrl312muvKTExUffff78k+7j4gwcPOubK/+WXX7RlyxbFxsbq6NGjmjNnjnbu3KnFixc7tjlz5kw9/fTTeueddxQTE+PouQ8ICFBAQECZ9gsAAADglHIH/WHDhiktLU1Tp05VUlKS2rVrp1WrVqlx48aSpKSkJKe57QsLCzV79mzt2bNHVqtVvXr10saNGxUTE+MoM2/ePOXl5enGG2902tekSZM0efLkMu0XAAAAwCnlnkf/Qsa8rAAAADUbea3ylPvKuAAAAABqPoI+AAAAYEIEfQAAAMCECPoAAACACRH0AQAAABMi6AMAAAAmRNAHAAAATIigDwAAAJgQQR8AAAAwIYI+AAAAYEIEfQAAAMCECPoAAACACRH0AQAAABMi6AMAAAAmRNAHAAAATIigDwAAAJgQQR8AAAAwIYI+AAAAYEIEfQAAAMCECPoAAACACRH0AQAAABMi6AMAAAAmRNAHAAAATIigDwAAAJgQQR8AAAAwIYI+AAAAYEIEfQAAAMCECPoAAACACRH0AQAAABMi6AMAAAAmRNAHAAAATIigDwAAAJgQQR8AAAAwIYI+AAAAYEIEfQAAAMCECPoAAACACRH0AQAAABMi6AMAAAAmRNAHAAAATIigDwAAAJgQQR8AAAAwIYI+AAAAYEIEfQAAAMCECPoAAACACRH0AQAAABMi6AMAAAAmRNAHAAAATIigDwAAAJgQQR8AAAAwoQoF/Xnz5qlJkyby9fVVp06d9O23356x/CuvvKLWrVvLZrOpZcuWWrJkidPju3bt0tChQxUTEyOLxaK5c+e6bGPy5MmyWCxOt8jIyIpUHwAAADC9cgf9ZcuWaezYsZo4caK2b9+u7t27q1+/fkpMTHRbfv78+ZowYYImT56sXbt2acqUKXrooYf06aefOspkZWWpadOmev75588Y3tu2baukpCTH7ccffyxv9QEAAIBawWIYhlGeFWJjY9WxY0fNnz/fsax169YaPHiw4uPjXcrHxcWpW7dumjVrlmPZ2LFjtXXrVm3YsMGlfExMjMaOHauxY8c6LZ88ebI+/vhj7dixozzVdZKZmang4GBlZGQoKCiowtsBAADA+UFeqzzl6tHPy8vTtm3b1KdPH6flffr00caNG92uk5ubK19fX6dlNptNW7ZsUX5+frkqu3fvXkVHR6tJkyYaPny4fv/99zOWz83NVWZmptMNAAAAqA3KFfRTU1NVWFioiIgIp+URERFKTk52u07fvn21cOFCbdu2TYZhaOvWrUpISFB+fr5SU1PLvO/Y2FgtWbJEX3zxhRYsWKDk5GTFxcUpLS2t1HXi4+MVHBzsuDVs2LDM+wMAAAAuZBU6GddisTjdNwzDZVmxp59+Wv369VOXLl1ktVo1aNAgjRo1SpLk6elZ5n3269dPQ4cOVfv27dW7d2+tXLlSkrR48eJS15kwYYIyMjIct/3795d5fwAAAMCFrFxBPzw8XJ6eni699ykpKS69/MVsNpsSEhKUlZWlffv2KTExUTExMQoMDFR4eHiFK+7v76/27dtr7969pZbx8fFRUFCQ0w0AAACoDcoV9L29vdWpUyetWbPGafmaNWsUFxd3xnWtVqsaNGggT09PLV26VAMGDJCHR8Wn8c/NzdVPP/2kqKioCm8DAAAAMCuv8q4wbtw4jRgxQp07d1bXrl312muvKTExUffff78k+3CZgwcPOubK/+WXX7RlyxbFxsbq6NGjmjNnjnbu3Ok05CYvL0+7d+92/H7w4EHt2LFDAQEBat68uSRp/PjxGjhwoBo1aqSUlBRNmzZNmZmZGjly5Dn/EQAAAACzKXfQHzZsmNLS0jR16lQlJSWpXbt2WrVqlRo3bixJSkpKcppTv7CwULNnz9aePXtktVrVq1cvbdy4UTExMY4yhw4d0qWXXuq4/8ILL+iFF15Qjx49tH79eknSgQMHdMsttyg1NVV169ZVly5dtHnzZsd+AQAAAJxS7nn0L2TMywoAAFCzkdcqT8UHyQMAAACosQj6AAAAgAkR9AEAAAATIugDAAAAJkTQBwAAAEyIoA8AAACYEEEfAAAAMCGCPgAAAGBCBH0AAADAhAj6AAAAgAkR9AEAAAATIugDAAAAJkTQBwAAAEyIoA8AAACYEEEfAAAAMCGCPgAAAGBCBH0AAADAhAj6AAAAgAkR9AEAAAATIugDAAAAJkTQBwAAAEyIoA8AAACYEEEfAAAAMCGCPgAAAGBCBH0AAADAhAj6AAAAgAkR9AEAAAATIugDAAAAJkTQBwAAAEyIoA8AAACYEEEfAAAAMCGCPgAAAGBCBH0AAADAhAj6AAAAgAkR9AEAAAATIugDAAAAJkTQBwAAAEyIoA8AAACYEEEfAAAAMCGCPgAAAGBCBH0AAADAhAj6AAAAgAkR9AEAAAATIugDAAAAJkTQBwAAAEyIoA8AAACYEEEfAAAAMCGCPgAAAGBCBH0AAADAhCoU9OfNm6cmTZrI19dXnTp10rfffnvG8q+88opat24tm82mli1basmSJU6P79q1S0OHDlVMTIwsFovmzp1bKfsFAAAAaqtyB/1ly5Zp7NixmjhxorZv367u3burX79+SkxMdFt+/vz5mjBhgiZPnqxdu3ZpypQpeuihh/Tpp586ymRlZalp06Z6/vnnFRkZWSn7BQAAAGozi2EYRnlWiI2NVceOHTV//nzHstatW2vw4MGKj493KR8XF6du3bpp1qxZjmVjx47V1q1btWHDBpfyMTExGjt2rMaOHXtO+3UnMzNTwcHBysjIUFBQUJnWAQAAQNUhr1WecvXo5+Xladu2berTp4/T8j59+mjjxo1u18nNzZWvr6/TMpvNpi1btig/P/+87bd435mZmU43AAAAoDYoV9BPTU1VYWGhIiIinJZHREQoOTnZ7Tp9+/bVwoULtW3bNhmGoa1btyohIUH5+flKTU09b/uVpPj4eAUHBztuDRs2LNP+AAAAgAtdhU7GtVgsTvcNw3BZVuzpp59Wv3791KVLF1mtVg0aNEijRo2SJHl6ep63/UrShAkTlJGR4bjt37+/XPsDAAAALlTlCvrh4eHy9PR06UVPSUlx6W0vZrPZlJCQoKysLO3bt0+JiYmKiYlRYGCgwsPDz9t+JcnHx0dBQUFONwAAAKA2KFfQ9/b2VqdOnbRmzRqn5WvWrFFcXNwZ17VarWrQoIE8PT21dOlSDRgwQB4eZdv9uewXAAAAqI28yrvCuHHjNGLECHXu3Fldu3bVa6+9psTERN1///2S7MNlDh486Jgr/5dfftGWLVsUGxuro0ePas6cOdq5c6cWL17s2GZeXp52797t+P3gwYPasWOHAgIC1Lx58zLtFwAAAMAp5Q76w4YNU1pamqZOnaqkpCS1a9dOq1atUuPGjSVJSUlJTnPbFxYWavbs2dqzZ4+sVqt69eqljRs3KiYmxlHm0KFDuvTSSx33X3jhBb3wwgvq0aOH1q9fX6b9AgAAADil3PPoX8iYlxUAAKBmI69VngrNugMAAACgZiPoAwAAACZE0AcAAABMiKAPAAAAmBBBHwAAADAhgj4AAABgQgR9AAAAwIQI+gAAAIAJEfQBAAAAEyLoAwAAACZE0AcAAABMiKAPAAAAmBBBHwAAADAhgj4AAABgQgR9AAAAwIQI+gAAAIAJEfQBAAAAEyLoAwAAACZE0AcAAABMiKAPAAAAmBBBHwAAADAhgj4AAABgQgR9AAAAwIQI+gAAAIAJEfQBAAAAEyLoAwAAACZE0AcAAABMiKAPAAAAmBBBHwAAADAhgj4AAABgQgR9AAAAwIQI+gAAAIAJEfQBAAAAEyLoAwAAACZE0AcAAABMiKAPAAAAmBBBHwAAADAhgj4AAABgQgR9AAAAwIQI+gAAAIAJEfQBAAAAEyLoAwAAACZE0AcAAABMiKAPAAAAmBBBHwAAADAhgj4AAABgQgR9AAAAwIQI+gAAAIAJEfQBAAAAE6pQ0J83b56aNGkiX19fderUSd9+++0Zy7/yyitq3bq1bDabWrZsqSVLlriUWb58udq0aSMfHx+1adNGH330kdPjkydPlsVicbpFRkZWpPoAcFaFRYaSMrL1y+FjyszJr+7qAABQbl7lXWHZsmUaO3as5s2bp27duunVV19Vv379tHv3bjVq1Mil/Pz58zVhwgQtWLBAl112mbZs2aJ77rlHoaGhGjhwoCRp06ZNGjZsmJ599lkNGTJEH330kW6++WZt2LBBsbGxjm21bdtWa9euddz39PSsyHMGUMsVFRlKPZ6rQxk5Ss7I1qH0HCVlZOtQRo6S0rOVlJGjlGO5KiwyHOsE+XqpQaifGoTaSvy0qWGY/fdAX2s1PiMAAFxZDMMwzl7slNjYWHXs2FHz5893LGvdurUGDx6s+Ph4l/JxcXHq1q2bZs2a5Vg2duxYbd26VRs2bJAkDRs2TJmZmfrss88cZa699lqFhobq3XfflWTv0f/444+1Y8eOcj3BkjIzMxUcHKyMjAwFBQVVeDsAai7DMHTkRJ6SMnJ06GRoP5SRraT0HCWf/P1wZo7yC8/+0efpYZG/t6cycwrOWjbYZnWE/+KGQMNQPzUIs98P8Cl3vwoA1ErktcpTrv88eXl52rZtm5544gmn5X369NHGjRvdrpObmytfX1+nZTabTVu2bFF+fr6sVqs2bdqkRx55xKlM3759NXfuXKdle/fuVXR0tHx8fBQbG6vp06eradOmpdY3NzdXubm5jvuZmZlleZoAaijDMJSRne/UA598MsQfyrCH+qSMHOUVFJ11Wx4WqV6gryKDfRUd4quoYJuign0VHWL/GRVsU91AH3l6WHQit0AH07N14GiW9h+x/zxwNPvkLUtHs/KVkW2/7Trk/nMmxO9kQyDEz+mbgAahfqofaqMhAACodOX6z5KamqrCwkJFREQ4LY+IiFBycrLbdfr27auFCxdq8ODB6tixo7Zt26aEhATl5+crNTVVUVFRSk5OPus2Y2NjtWTJErVo0UKHDx/WtGnTFBcXp127dqlOnTpu9x0fH68pU6aU5ykCqEaZOflKOhnik04OozmUcfJ+uj3EZ+cXlmlb4QE+JwO8PbRHh/gqMtim6GBfRYXYVC/QR1bPsp2m5O/jpRYRgWoREej28eO5BTp4tLghUKIRkG7/PT0r33HbedB9QyDUz+p2SFCDUD/VD7HJn4YAAKCcKvSfw2KxON03DMNlWbGnn35aycnJ6tKliwzDUEREhEaNGqWZM2c6jbE/2zb79evn+L19+/bq2rWrmjVrpsWLF2vcuHFu9z1hwgSnxzIzM9WwYcOyP1EAlSYrr8DRE+/ogU/PUVLmqXHxx3PPPkRGksL8vU8L8L6KLtEjHxHkK2+vqptULMDHSy0jA9Uy0n1D4FhOvg6mZ5/2bcCpBkFGdr6OZuXraFaGfjyY4XYbYf7epxoBp50rUD/UJj9vGgIAAGfl+s8QHh4uT09Pl977lJQUlx75YjabTQkJCXr11Vd1+PBhRUVF6bXXXlNgYKDCw8MlSZGRkeXapiT5+/urffv22rt3b6llfHx85OPjU9anB6CCcvILnXrgk087sfVQenaZxrlL9rHu9hBv73mPPhnoo0LsYT4y2Fe+1gvrRPxAX6taRVrVKtL9WNPMnHwdcDMkqPhnZk6BjpzI05ETefrhgPuGQJ3ihkCY32nnCdhUP8RPNu8L628GADh35Qr63t7e6tSpk9asWaMhQ4Y4lq9Zs0aDBg0647pWq1UNGjSQJC1dulQDBgyQh4e9x61r165as2aN0zj91atXKy4urtTt5ebm6qefflL37t3L8xQAlFNuQaEOZ+Q6htM4euJPzlaTnJmjIyfyyrStQB8vRboJ8MW981HBvrVyiEqQr1Vtoq1qE+2+IZCRnV9KIyBbB45k6VhugdJO5CntRJ7+V0pDIDzA2+2sQcW/X2iNJwDA2ZX7P+q4ceM0YsQIde7cWV27dtVrr72mxMRE3X///ZLsw2UOHjzomCv/l19+0ZYtWxQbG6ujR49qzpw52rlzpxYvXuzY5t///nddeeWVmjFjhgYNGqRPPvlEa9eudczKI0njx4/XwIED1ahRI6WkpGjatGnKzMzUyJEjz/VvANRaBYVFOnws99RY+PTiE1qLe+JzlHo89+wbkmSzejp63Yt744t75otPcGUKyooJtlkVbAtW2+hgt49nZOVrv5shQcW/H88tUOrxPKUez9OO/elut1E30KfURkD9EBoCAHAhKnfQHzZsmNLS0jR16lQlJSWpXbt2WrVqlRo3bixJSkpKUmJioqN8YWGhZs+erT179shqtapXr17auHGjYmJiHGXi4uK0dOlSPfXUU3r66afVrFkzLVu2zGkO/QMHDuiWW25Ramqq6tatqy5dumjz5s2O/QJwVlhk6K9juU498MUhvnis/F/HclVUhgl2vb083PbAF89WEx1sU5DNq9RzdXB+BftZFewXrHb1XRsCxTMVldYI2H8kSyfyCvXXsVz9dSxX2xPT3e6jnktD4FSDoH6oTT5eNAQAoKYp9zz6FzLmZYVZFBUZSj2Ra58XPr1kD7z9Z3JGjg5n5qigDCne6mmxD6cJKg7xNqfpJqOCfRXm702INynDMJSedXpD4GQj4OTPrLyzz3QUEeTj9tuAhqF+igrxpSEAoMzIa5Wn9g2GBWo4wzB0NCvfEdodY+EzTk01eTgjV3mFZ58r3tPDoohAH8cwmugSw2mKe+fD/X3k4UGIr60sFotC/b0V6u+t9g3cfyNwNCvfpRFQ/G3AgaPZys4v1OHMXB3OzNW2P4+62YcUEejr2gg4eeJwVLCtSmdJAoDagh59oAoZhqHM7IKTF3cqnis+x2V4TW4ZLvhksdiHUzimmAwq0RN/cqx88QWfgPOl+ErE7mYL2n/yZ07+md/PFosUGeTrMiTIPo2o/RuBsl7zAMCFj7xWeQj6QCU6lpPvuDpraSe4lmUYhHTqgk+RQSV64k/OVhMZ7KuIIMIPaj7DMJTmaAiUaASUmE70bA1bD0dDoMTQoLBTQ4MigzkWADMhr1Uegj5QRtl5hY6e9+KfyZnZTheBOlbGCz6F+lmdx8KXPME12KaIYB/GNKNWMAxDqcfznIcEnTZMKK8MDYGoYPtJwcXfCjQs8e1AVLCvvGgIABcM8lrlIegDsl/wKfnkHPHJJ3vkD6U7n+CakZ1fpm0F+Xo59cBHBZWYN/7kcqYqBMqm+MTzkkODir8NOHg0WwfSz94Q8PSwOA0NahjmPEQoMoiGAFCTkNcqD0EfppdXUKTDmTkuU0seOtkjn5Seo7QyXvApwMfrtAB/ct74kNp9wSeguhQVGUo9nus4H+D0E4YPHs0+64nrnh4WRQX7Op0XUHKIUGSQL+e6AFWIvFZ5CPq4oBUUFinlWK5TgC8+wTXp5Cw1qcdzVZZ3ua/Vwym0Rztd9Mm+PIgLPgEXlKIiQ38dz3W6bkDJbwcOpmcrv/DMHxBeHhZFhfiebAS4XksggoYAUKnIa5WHrkfUWIZhKOVYrg6lZ58cVnPqxNbiMfIpx3LKfMEnx1VaT4b2yOIwf3KsfLDNylzxgMl4eFgUEWQ/eb2Tm+srFhXZP2fsswRl6cCRk42AdHuD4NDJhsD+I9nafyTb7T68PCyKDrHZhwSFnGwElBgeVC+QhgCA6kHQR7UrKjJ04Gi29qYc096U49p7+Lh+TTmmX1OO60QZZqixetr/kZ8K8Cd/LzFvPBd8AuCOh4f9gnGRwb7qHBPm8nhhkaGUYzmnfRtw6luBQ+nZKigylHgkS4lHsiSluWzD6mlR/RDn8wJKfitQL5BrWQA4Pwj6qDIFhUXafzRbvxy2h/i9h+3B/re/jpc6z3bxBZ8iS0wtefqVW8MD+CcJ4Pywj9+3X9TrslIaAoczc1yGBBV/K3AoPUf5hYb2pWVpX1qW2314e3qUmDHI9VoCfMYBqCiCPipdfmGR/kw7ob2Hj9t76E+G+t9TT5Q6O4a3l4ea1Q3QRfVO3iIC1LxeoBrX8WN+bAA1lufJYTvRITbFunm8oLBIh4/luvk2wP4zKSNHeYVF+iP1hP5IPeF2H95eHmoQUjx9qPO3Ag3DbKob4MM3lgDcIuijwnILCvVH6qlA/2vKMe09fFx/pJ5QQSkD532tHmpeL0AX1QvURREnf9YLUMMwP8awAjAdL08P1Q+xqX6Ize3jBYVFSsrIcRkSdKohYJ8+9PfUE/q9lIaAj5eH+0bAyZ/hAQxdBGorgj7OKjuvUL/9ddw+3OZkmP815bj+PJKlwlICvb+3p5pHBDr10F9UL1D1Q2x8BQ0AJ3l5eqhhmJ8ahvlJquPyeH5hkZIzckpcROxUI+DgyYZAbkGRfv/rhH7/q/SGwOnnBZS8lkAdzmECTIugD4cTuQX67a/jp4bcnBxDv/9oVqnTUwb6eqnFyUDfvF6ALjr5e1SwL/84AOAcWZ0aAq7yCuwNAXfXENh/NEvJmTnKLSjSb3+d0G+lNAR8rR4u5wWUvM9kBsCFi6BfC2Xm5OvXlOP69fBxp5luDqa7nzpOkkL9rI4Qf1GJQF83kLGhAFBdvL081KiOnxrVKb0hkJSR7XZo0P4j2Tp8LEc5+UX2/wkpx91uw2b1PNUICHO9lkCoH1MTAzUVQd/E0rPyHCF+b0rxTDfHlZyZU+o64QE+JYba2E+IvSgiQOEBPlVYcwBAZfD28lDjOv5qXMff7eO5BYVKSj/9HIEsx5WGD2fmKju/0DGxgjt+3p4u5wWUbAyE0BAAqg1B3wTSjuc6DbUpHnqTejy31HUig3xPzmxz6sTY5nUDFOrvXYU1BwBUJx8vT8WE+ysmvPSGwKF016FBxbMIpRzLVVZeoX45fFy/HHbfEPD39nQ5L6BkQ4CLFQLnD0H/AmEYhv46dlqgP/lV65ETeaWuVz/E5uidv6heoJqfDPdBvtYqrD0A4ELk4+WpJuH+alJKQyAnv1CH0rOdhwSVaBD8dSxXJ/IKtefwMe05fMztNgJ8vFyuH1D8e8NQPwX78f8KqCiCfg1jGIaSMnIcgf7XEvPQZ+YUuF3HYpEahvrZh9qUmLKyWb0ABfjwEgMAzg9fq6ea1g1Q07oBbh/PyS/UwfRst98GHDiardTjuTqeW6Cfk4/p52T3DYFAX69Srips/z3YRkMAKA0psJoUFRk6mJ7tmK6yuIf+t5TjOp7rPtB7WKSYOv4nZ7c52UNfL0DN6gbI5u1Zxc8AAIAz87V6qlld+/8pd7LzihsCWY6ZgoobAQePZin1eJ6O5RTop6RM/ZSU6XYbQU4NgdMaBGE2vsFGrUbQP88KiwztP5J1MsgfOznTjX3ITXZ+odt1vDwsahLu77g6bPHJsTF1/OVrJdADAMzB5u2p5ienZ3bH3hA4ORzoiOu1BNJO5Ckzp0C7kzK1u5SGQLDN6ubbgFMNgkAaAjAxgv55svHXVD278if99tdx5RUUuS3j7emhpnX9HSfEtoiwB/rGdfxl9fSo4hoDAFCz2BsCgWpeL9Dt41l5BTpY4roBp08jeuREnjKy85WRna9dh9w3BEL8TjYEQk67snCYn+qH2hgCiwsa797zxNvLw/E1o4+Xx8kwb59/vvj3RmF+8iLQAwBQIX7eXvbrukS4bwicyC3QwfSS5wU4fytwNCtf6SdvOw+6bwiE+lldhgQVzyBUP8QmfxoCqMEshlHaNU/NJzMzU8HBwcrIyFBQUNB53deJ3AJt/j1NF9ULVP1Qmzw9mDoMAICa5Hiu/RsBe0OgRCMg3f57elb+WbcR5u/t3AgoMTSofqhNft40BMqrKvOa2RH0AQAA3DiWk+9yXkDJBkFG9tkbAnUcDQH7ycGnpg61qX6IH5NpuEFeqzw0MwEAANwI9LWqdZRVraPch83MnHwdOOI6JKj4Z2ZOgdJO5CntRJ7+dyDD7TbCA7xVv8R1A04/cZhJOHAuCPoAAAAVEORrVZtoq9pEu28IZGTnl9IIsM8idCy3QKnH85R6PE//25/udhvhAT72RkCYayOgfggNAZwZQR8AAOA8CLZZFWwLVtvoYJfHDMNQZnaB29mCin8/nlug1OO5Sj2eqx2lNATqBvo4nRdQ8sThaBoCtR5BHwAAoIpZLBYF+1kV7BesdvXdNwTs3wi4bwTsP5KlE3mF+utYrv46lqv/Jqa73U9EkE8pVxb2U3SIr3y8aAiYGUEfAACghrFYLArx81aIn3epDYH0rNMbAs5XGM7KK9ThzFwdzszVtj+PutmHFBHoW+oFxaJDbPL2YhrwCxlBHwAA4AJjsVgU6u+tUH9vtW/gviFwNCvftRFQ4grD2fmFSs7MUXJmjraW0hCIDLI3BG6NbaQhlzaoiqeGSkTQBwAAMBmLxaIwf2+F+XurQ4MQl8cNw9CRE3muQ4JKNAxy8ouUlJGjpIwcXdsuquqfBM4ZQR8AAKCWsVgsqhPgozoBPrq4YYjL44ZhKM3REMhSm1KmGEXNRtAHAACAE4vFovAAH4UH+OgSNw0BXBg4wwIAAAAwIYI+AAAAYEIEfQAAAMCECPoAAACACRH0AQAAABMi6AMAAAAmRNAHAAAATIigDwAAAJgQQR8AAAAwIYI+AAAAYEIEfQAAAMCECPoAAACACRH0AQAAABMi6AMAAAAmRNAHAAAATIigDwAAAJgQQR8AAAAwIa/qrkBVMgxDkpSZmVnNNQEAAIA7xTmtOLeh4mpV0D927JgkqWHDhtVcEwAAAJzJsWPHFBwcXN3VuKBZjFrUXCoqKtKhQ4cUGBgoi8VS3dUpt8zMTDVs2FD79+9XUFBQdVcH4jWpaXg9ah5ek5qH16Rm4fVwZRiGjh07pujoaHl4MMr8XNSqHn0PDw81aNCguqtxzoKCgvgwqGF4TWoWXo+ah9ek5uE1qVl4PZzRk185aCYBAAAAJkTQBwAAAEyIoH8B8fHx0aRJk+Tj41PdVcFJvCY1C69HzcNrUvPwmtQsvB44n2rVybgAAABAbUGPPgAAAGBCBH0AAADAhAj6AAAAgAkR9AEAAAATIugDAAAAJkTQvwBMnjxZFovF6RYZGVnd1ao1vvnmGw0cOFDR0dGyWCz6+OOPnR43DEOTJ09WdHS0bDabevbsqV27dlVPZWuJs70mo0aNcjlmunTpUj2VrQXi4+N12WWXKTAwUPXq1dPgwYO1Z88epzIcJ1WrLK8Jx0nVmT9/vjp06OC4+m3Xrl312WefOR7n+MD5QtC/QLRt21ZJSUmO248//ljdVao1Tpw4oYsvvlgvv/yy28dnzpypOXPm6OWXX9b333+vyMhIXXPNNTp27FgV17T2ONtrIknXXnut0zGzatWqKqxh7fL111/roYce0ubNm7VmzRoVFBSoT58+OnHihKMMx0nVKstrInGcVJUGDRro+eef19atW7V161ZdddVVGjRokCPMc3zgvDFQ402aNMm4+OKLq7saMAxDkvHRRx857hcVFRmRkZHG888/71iWk5NjBAcHG//+97+roYa1z+mviWEYxsiRI41BgwZVS31gGCkpKYYk4+uvvzYMg+OkJjj9NTEMjpPqFhoaaixcuJDjA+cVPfoXiL179yo6OlpNmjTR8OHD9fvvv1d3lSDpjz/+UHJysvr06eNY5uPjox49emjjxo3VWDOsX79e9erVU4sWLXTPPfcoJSWluqtUa2RkZEiSwsLCJHGc1ASnvybFOE6qXmFhoZYuXaoTJ06oa9euHB84rwj6F4DY2FgtWbJEX3zxhRYsWKDk5GTFxcUpLS2tuqtW6yUnJ0uSIiIinJZHREQ4HkPV69evn95++2199dVXmj17tr7//ntdddVVys3Nre6qmZ5hGBo3bpyuuOIKtWvXThLHSXVz95pIHCdV7ccff1RAQIB8fHx0//3366OPPlKbNm04PnBeeVV3BXB2/fr1c/zevn17de3aVc2aNdPixYs1bty4aqwZilksFqf7hmG4LEPVGTZsmOP3du3aqXPnzmrcuLFWrlypG264oRprZn4PP/ywfvjhB23YsMHlMY6T6lHaa8JxUrVatmypHTt2KD09XcuXL9fIkSP19ddfOx7n+MD5QI/+Bcjf31/t27fX3r17q7sqtV7x7Een97qkpKS49M6g+kRFRalx48YcM+fZmDFj9J///Efr1q1TgwYNHMs5TqpPaa+JOxwn55e3t7eaN2+uzp07Kz4+XhdffLFefPFFjg+cVwT9C1Bubq5++uknRUVFVXdVar0mTZooMjJSa9ascSzLy8vT119/rbi4uGqsGUpKS0vT/v37OWbOE8Mw9PDDD+vDDz/UV199pSZNmjg9znFS9c72mrjDcVK1DMNQbm4uxwfOK4buXADGjx+vgQMHqlGjRkpJSdG0adOUmZmpkSNHVnfVaoXjx4/r119/ddz/448/tGPHDoWFhalRo0YaO3aspk+frosuukgXXXSRpk+fLj8/P916663VWGtzO9NrEhYWpsmTJ2vo0KGKiorSvn379OSTTyo8PFxDhgypxlqb10MPPaR33nlHn3zyiQIDAx09k8HBwbLZbLJYLBwnVexsr8nx48c5TqrQk08+qX79+qlhw4Y6duyYli5dqvXr1+vzzz/n+MD5VX0T/qCshg0bZkRFRRlWq9WIjo42brjhBmPXrl3VXa1aY926dYYkl9vIkSMNw7BPHThp0iQjMjLS8PHxMa688krjxx9/rN5Km9yZXpOsrCyjT58+Rt26dQ2r1Wo0atTIGDlypJGYmFjd1TYtd6+FJOONN95wlOE4qVpne004TqrWnXfeaTRu3Njw9vY26tata1x99dXG6tWrHY9zfOB8sRiGYVRlwwIAAADA+ccYfQAAAMCECPoAAACACRH0AQAAABMi6AMAAAAmRNAHAAAATIigDwAAAJgQQR8AAAAwIYI+AAAAYEIEfQAAAMCECPoAAACACRH0AQAAABP6f4gP1TYYoYFuAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 800x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Note: by specifying None for the pretrained models, they will be initialized from scratch, using\n",
    "# the specified embedding dimensionality.\n",
    "ncf_dim, ncf_model, ncf_dim_history, ncf_dim_curves = tune_dims('ncf', gmf_model=None, mlp_model=None)\n",
    "title = f\"Train and Test RMSE vs. Embedding Dim for NCF on {dataset}\"\n",
    "plot_learning_curve(ncf_dim_history, ncf_dim_curves, title, True)\n",
    "print(f\"Best Embedding Dimensionality for NCF on {dataset}\", ncf_dim)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "7f1e5deb",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best Embedding Dimensionality for NCF on 20m 32\n"
     ]
    }
   ],
   "source": [
    "print(f\"Best Embedding Dimensionality for NCF on {dataset}\", ncf_dim)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f1fe3b5c",
   "metadata": {},
   "source": [
    "# Final Training of Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "a0b80929",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TRAINING GMF\n",
      "Epoch 1\n",
      "------------------------\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.917677  [ 1024/15945812]\n",
      "loss: 0.046885  [1025024/15945812]\n",
      "loss: 0.046660  [2049024/15945812]\n",
      "loss: 0.044594  [3073024/15945812]\n",
      "loss: 0.044760  [4097024/15945812]\n",
      "loss: 0.044113  [5121024/15945812]\n",
      "loss: 0.047092  [6145024/15945812]\n",
      "loss: 0.045372  [7169024/15945812]\n",
      "loss: 0.044252  [8193024/15945812]\n",
      "loss: 0.043717  [9217024/15945812]\n",
      "loss: 0.043179  [10241024/15945812]\n",
      "loss: 0.044244  [11265024/15945812]\n",
      "loss: 0.045206  [12289024/15945812]\n",
      "loss: 0.045637  [13313024/15945812]\n",
      "loss: 0.043815  [14337024/15945812]\n",
      "loss: 0.038122  [15361024/15945812]\n",
      "Avg loss: 0.039577 \n",
      "\n",
      "Test MSE 0.9894309105309593\n",
      "Test RMSE 0.9947014177787017\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.038665  [ 1024/15945812]\n",
      "loss: 0.037367  [1025024/15945812]\n",
      "loss: 0.033921  [2049024/15945812]\n",
      "loss: 0.032917  [3073024/15945812]\n",
      "loss: 0.028497  [4097024/15945812]\n",
      "loss: 0.030026  [5121024/15945812]\n",
      "loss: 0.027015  [6145024/15945812]\n",
      "loss: 0.030507  [7169024/15945812]\n",
      "loss: 0.028102  [8193024/15945812]\n",
      "loss: 0.028857  [9217024/15945812]\n",
      "loss: 0.025413  [10241024/15945812]\n",
      "loss: 0.025373  [11265024/15945812]\n",
      "loss: 0.026715  [12289024/15945812]\n",
      "loss: 0.031415  [13313024/15945812]\n",
      "loss: 0.025472  [14337024/15945812]\n",
      "loss: 0.025482  [15361024/15945812]\n",
      "Avg loss: 0.029965 \n",
      "\n",
      "Test MSE 0.7491185259924344\n",
      "Test RMSE 0.8655163349079176\n",
      "Epoch 3\n",
      "------------------------\n",
      "loss: 0.028057  [ 1024/15945812]\n",
      "loss: 0.027066  [1025024/15945812]\n",
      "loss: 0.027635  [2049024/15945812]\n",
      "loss: 0.027477  [3073024/15945812]\n",
      "loss: 0.026655  [4097024/15945812]\n",
      "loss: 0.027836  [5121024/15945812]\n",
      "loss: 0.027406  [6145024/15945812]\n",
      "loss: 0.027523  [7169024/15945812]\n",
      "loss: 0.025616  [8193024/15945812]\n",
      "loss: 0.025474  [9217024/15945812]\n",
      "loss: 0.025372  [10241024/15945812]\n",
      "loss: 0.025492  [11265024/15945812]\n",
      "loss: 0.026921  [12289024/15945812]\n",
      "loss: 0.028209  [13313024/15945812]\n",
      "loss: 0.025116  [14337024/15945812]\n",
      "loss: 0.028277  [15361024/15945812]\n",
      "Avg loss: 0.029288 \n",
      "\n",
      "Test MSE 0.7322060790968331\n",
      "Test RMSE 0.8556904107776557\n",
      "Epoch 4\n",
      "------------------------\n",
      "loss: 0.024379  [ 1024/15945812]\n",
      "loss: 0.025760  [1025024/15945812]\n",
      "loss: 0.027077  [2049024/15945812]\n",
      "loss: 0.027008  [3073024/15945812]\n",
      "loss: 0.028560  [4097024/15945812]\n",
      "loss: 0.024901  [5121024/15945812]\n",
      "loss: 0.026333  [6145024/15945812]\n",
      "loss: 0.026529  [7169024/15945812]\n",
      "loss: 0.025784  [8193024/15945812]\n",
      "loss: 0.029564  [9217024/15945812]\n",
      "loss: 0.027054  [10241024/15945812]\n",
      "loss: 0.026853  [11265024/15945812]\n",
      "loss: 0.028429  [12289024/15945812]\n",
      "loss: 0.025345  [13313024/15945812]\n",
      "loss: 0.026574  [14337024/15945812]\n",
      "loss: 0.026783  [15361024/15945812]\n",
      "Avg loss: 0.029127 \n",
      "\n",
      "Test MSE 0.7281695633639222\n",
      "Test RMSE 0.8533285201866408\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# GMF Hyperparameters\n",
    "learning_rate = 0.001\n",
    "weight_decay = 0.0000001\n",
    "epochs = 4\n",
    "batch_size = 1024\n",
    "latent_dims = 4\n",
    "alpha = 0.5\n",
    "loss = nn.MSELoss()\n",
    "\n",
    "print('TRAINING GMF')\n",
    "train_model = train_gmf\n",
    "final_gmf_model, train_loss, test_loss, test_rmse = train_model(train_dataloader,\n",
    "                                                      test_dataloader,\n",
    "                                                      num_users,\n",
    "                                                      num_items,\n",
    "                                                      epochs,\n",
    "                                                      latent_dims,\n",
    "                                                      learning_rate,\n",
    "                                                      'adam',\n",
    "                                                      criterion=loss,\n",
    "                                                      device=device,\n",
    "                                                      weight_decay=weight_decay,\n",
    "                                                      rescale_data=True)\n",
    "\n",
    "\n",
    "model_name = f\"models/final_gmf_model_{dataset}\"\n",
    "torch.save(final_gmf_model.state_dict(), model_name)\n",
    "\n",
    "curve_data = {\n",
    "    'Train Loss': train_loss,\n",
    "    'Test Loss': test_loss,\n",
    "    'Test RMSE': test_rmse,\n",
    "}\n",
    "title = f\"Training Curves for tuned GMF on {dataset}\"\n",
    "plot_learning_curve(np.arange(1, epochs+1), curve_data, title, True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "10a2d4d0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TRAINING MLP\n",
      "Epoch 1\n",
      "-------------------------------\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.568002  [ 1024/15945812]\n",
      "loss: 0.039790  [1025024/15945812]\n",
      "loss: 0.035519  [2049024/15945812]\n",
      "loss: 0.034196  [3073024/15945812]\n",
      "loss: 0.035445  [4097024/15945812]\n",
      "loss: 0.031850  [5121024/15945812]\n",
      "loss: 0.029452  [6145024/15945812]\n",
      "loss: 0.030576  [7169024/15945812]\n",
      "loss: 0.030913  [8193024/15945812]\n",
      "loss: 0.031117  [9217024/15945812]\n",
      "loss: 0.028190  [10241024/15945812]\n",
      "loss: 0.029128  [11265024/15945812]\n",
      "loss: 0.029806  [12289024/15945812]\n",
      "loss: 0.030743  [13313024/15945812]\n",
      "loss: 0.025083  [14337024/15945812]\n",
      "loss: 0.029603  [15361024/15945812]\n",
      "Avg loss: 0.032123 \n",
      "\n",
      "Test MSE 0.8030537716800017\n",
      "Test RMSE 0.8961326752663367\n",
      "Epoch 2\n",
      "-------------------------------\n",
      "loss: 0.030763  [ 1024/15945812]\n",
      "loss: 0.027192  [1025024/15945812]\n",
      "loss: 0.027239  [2049024/15945812]\n",
      "loss: 0.030660  [3073024/15945812]\n",
      "loss: 0.030782  [4097024/15945812]\n",
      "loss: 0.026741  [5121024/15945812]\n",
      "loss: 0.026860  [6145024/15945812]\n",
      "loss: 0.029238  [7169024/15945812]\n",
      "loss: 0.027288  [8193024/15945812]\n",
      "loss: 0.029520  [9217024/15945812]\n",
      "loss: 0.030394  [10241024/15945812]\n",
      "loss: 0.027987  [11265024/15945812]\n",
      "loss: 0.026162  [12289024/15945812]\n",
      "loss: 0.027765  [13313024/15945812]\n",
      "loss: 0.025944  [14337024/15945812]\n",
      "loss: 0.027897  [15361024/15945812]\n",
      "Avg loss: 0.030409 \n",
      "\n",
      "Test MSE 0.76023144620032\n",
      "Test RMSE 0.8719125221031753\n",
      "Epoch 3\n",
      "-------------------------------\n",
      "loss: 0.027761  [ 1024/15945812]\n",
      "loss: 0.028046  [1025024/15945812]\n",
      "loss: 0.029597  [2049024/15945812]\n",
      "loss: 0.027257  [3073024/15945812]\n",
      "loss: 0.026840  [4097024/15945812]\n",
      "loss: 0.027018  [5121024/15945812]\n",
      "loss: 0.029264  [6145024/15945812]\n",
      "loss: 0.027655  [7169024/15945812]\n",
      "loss: 0.027880  [8193024/15945812]\n",
      "loss: 0.026856  [9217024/15945812]\n",
      "loss: 0.029783  [10241024/15945812]\n",
      "loss: 0.027990  [11265024/15945812]\n",
      "loss: 0.029151  [12289024/15945812]\n",
      "loss: 0.024449  [13313024/15945812]\n",
      "loss: 0.029092  [14337024/15945812]\n",
      "loss: 0.028161  [15361024/15945812]\n",
      "Avg loss: 0.029665 \n",
      "\n",
      "Test MSE 0.7416087680019071\n",
      "Test RMSE 0.8611670964463907\n",
      "Epoch 4\n",
      "-------------------------------\n",
      "loss: 0.025092  [ 1024/15945812]\n",
      "loss: 0.024701  [1025024/15945812]\n",
      "loss: 0.028636  [2049024/15945812]\n",
      "loss: 0.026855  [3073024/15945812]\n",
      "loss: 0.026648  [4097024/15945812]\n",
      "loss: 0.026521  [5121024/15945812]\n",
      "loss: 0.025873  [6145024/15945812]\n",
      "loss: 0.028524  [7169024/15945812]\n",
      "loss: 0.027724  [8193024/15945812]\n",
      "loss: 0.026044  [9217024/15945812]\n",
      "loss: 0.025766  [10241024/15945812]\n",
      "loss: 0.026679  [11265024/15945812]\n",
      "loss: 0.024328  [12289024/15945812]\n",
      "loss: 0.024551  [13313024/15945812]\n",
      "loss: 0.023638  [14337024/15945812]\n",
      "loss: 0.026061  [15361024/15945812]\n",
      "Avg loss: 0.029620 \n",
      "\n",
      "Test MSE 0.7405072479921461\n",
      "Test RMSE 0.8605273081036685\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAugAAAI6CAYAAAB1rL20AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8fJSN1AAAACXBIWXMAAA9hAAAPYQGoP6dpAABJe0lEQVR4nO3deXhU1eH/8c9knSRkYU1YQthkURYlFEgQAYEgCEgpCCpLEFAEoSyioP7YRCm2KqKC+pWlKlUqiFVEJCibBS1iUBRMrSBBCCIoCRISstzfH2mmTGYSMpNATuD9ep55zJw5594zJzf4uWfOvWOzLMsSAAAAACP4VHQHAAAAAPwPAR0AAAAwCAEdAAAAMAgBHQAAADCIX0V3AAAAAMXLy8tTTk5ORXcD5cjf31++vr7Fvk5ABwAAMJBlWTp+/LhOnz5d0V3BJRAREaGoqCjZbDaX1wjoAAAABioM57Vq1VJwcLDbIIfKx7IsZWZm6sSJE5Kk2rVru9QhoAMAABgmLy/PEc6rV69e0d1BOQsKCpIknThxQrVq1XJZ7sJFogAAAIYpXHMeHBxcwT3BpVL4u3V3fQEBHQAAwFAsa7lylfS7JaADAAAABiGgAwAAAAYhoAMAAMBYXbt21eTJkyu6G5cVd3EBAABAmV1svfzIkSO1cuVKj7f79ttvy9/f38teFUhMTNTp06f1zjvvlGk7lwsz6Lhi2Wy2Uj22bt1apv3MmTPH64t4tm7dWi59KIuDBw/q/vvvV9OmTRUUFKTg4GBdd911evTRR3X06NEK69ellpycrC5duig8PFw2m02LFi26ZPvKzMzUnDlzKvT37InSHpcrV64s8e/Isiw1adJENptNXbt2dXrNZrPp/vvvL3H7Xbt2dfpbDQoKUps2bbRo0SLl5+d7+K4ujbffflt33HGHmjRpoqCgIDVo0EB33XWXvvvuO7f1N2/erLi4OAUHB6tGjRpKTEx03AsZqOzS0tIcj0WLFiksLMyp7Nlnn3WqX9pvR61WrZpCQ0MvRZeNRUDHFWvXrl1Ojz59+igoKMilvG3btmXaz5gxY7Rr1y6v2rZt27Zc+uCt9evXq3Xr1lq/fr3uuecerV+/3vHze++9p759+1ZIvy6Hu+++W2lpaXrzzTe1a9cuDR069JLtKzMzU3Pnzq00Ad1ToaGhWrZsmUv5tm3b9P3335fpf6yNGjVy/K2uXr1adevW1ZQpUzRz5syydLncLFy4UJmZmXrkkUe0ceNGzZ8/X8nJyWrbtq2++eYbp7rbtm1T7969FRkZqX/84x969tlntXnzZnXv3l3Z2dkV9A5QmViWpczzuZf9YVlWqfoXFRXleBROfhQ+z8rKUkREhP7+97+ra9eustvtev3113Xq1CndcccdqlevnoKDg9WqVSu98cYbTtstusSlQYMGeuKJJ3T33XcrNDRU9evX18svv1ymsd22bZvat2+vwMBA1a5dWzNmzFBubq7j9TVr1qhVq1YKCgpS9erV1aNHD509e1ZSwaRG+/btFRISooiICHXq1EmHDx8uU39Y4oIrVseOHZ2e16xZUz4+Pi7lRWVmZnp039l69eqpXr16XvUxLCzsov25VA4dOqShQ4eqadOm2rJli8LDwx2v3XzzzZo0aZLWrVtXLvvKycmRzWaTn585/+R8/fXXGjt2rHr37l0u2zPxPV4uQ4YM0apVq/TCCy8oLCzMUb5s2TLFxcUpIyPD620HBQU5/Y307t1bzZs31/PPP6/58+eX+WPvsnrvvfdUq1Ytp7Kbb75ZDRo00DPPPKNXXnnFUT59+nQ1bdpUa9ascRwnDRs2VKdOnbR8+XLdd999l7XvqHzO5eTp2lkfXvb97p/XS8EB5fNv20MPPaSnnnpKK1asUGBgoLKyshQbG6uHHnpIYWFhev/99zV8+HA1atRIHTp0KHY7Tz31lB577DE9/PDDWrNmje677z7ddNNNat68ucd9Onr0qPr06aPExES9+uqr+vbbbzV27FjZ7XbNmTNHaWlpuuOOO/Tkk0/q97//vc6cOaMdO3bIsizl5uZqwIABGjt2rN544w2dP39e//rXv8p8e0xm0HFV69q1q1q2bKnt27crPj5ewcHBuvvuuyVJq1evVkJCgmrXrq2goCC1aNFCM2bMcJwxF3K3xKVBgwbq27evNm7cqLZt2yooKEjNmzfX8uXLneq5W0qQmJioKlWq6D//+Y/69OmjKlWqKDo6WtOmTXOZZfvxxx81aNAghYaGKiIiQnfddZd2794tm8120XV+Tz/9tM6ePaslS5Y4hfNCNptNAwcOdHpPiYmJbsfwwuULhe/ptdde07Rp01S3bl0FBgbqm2++kc1mczvT+sEHH8hms+ndd991lH333Xe68847VatWLQUGBqpFixZ64YUXnNrl5+dr/vz5atasmYKCghQREaHWrVu7fIx6ocJlGbm5uVq6dKlj+UShr7/+WrfddpuqVq0qu92u66+/Xn/961+dtlHce/zPf/7jsr8ffvhBNWvWlCTNnTvXsb/CsUxMTFSDBg1c2rk7rgqXhbz22mtq0aKFgoOD1aZNG61fv96lfWnGT5K+/fZb3XLLLY4lF+PGjdOZM2eKHT937rjjDklymvVKT0/X2rVrHX9P5cXf31+xsbHKzMzUzz//XGLdTz75RN27d1doaKiCg4MVHx+v999/36lO4fGwZcsW3XfffapRo4aqV6+ugQMH6tixYxftT9FwLkl16tRRvXr1dOTIEUfZ0aNHtXv3bg0fPtzpJC4+Pl5NmzZ1Ohku7NPHH3+ssWPHqnr16goLC9OIESN09uxZHT9+XLfffrsiIiJUu3ZtPfDAA6VeKgBUtMmTJ2vgwIFq2LCh6tSpo7p16+qBBx7Q9ddfr0aNGmnixInq1auX3nrrrRK306dPH40fP15NmjTRQw89pBo1anj9KeWSJUsUHR2t559/Xs2bN9eAAQM0d+5cPfXUU8rPz1daWppyc3M1cOBANWjQQK1atdL48eNVpUoVZWRkKD09XX379lXjxo3VokULjRw5UvXr1/eqL4WuvqkeoIi0tDQNGzZMDz74oJ544gn5+BSct3733Xfq06ePJk+erJCQEH377bdauHCh/vWvf+njjz++6Ha//PJLTZs2TTNmzFBkZKReeeUVjR49Wk2aNNFNN91UYtucnBz1799fo0eP1rRp07R9+3Y99thjCg8P16xZsyRJZ8+eVbdu3fTLL79o4cKFatKkiTZu3KghQ4aU6n1v2rRJkZGRl2wGf+bMmYqLi9OLL74oHx8fRUdH64YbbtCKFSs0evRop7orV65UrVq11KdPH0nS/v37FR8fr/r16+upp55SVFSUPvzwQ02aNEknT57U7NmzJUlPPvmk5syZo0cffVQ33XSTcnJy9O233+r06dPF9uvWW2/Vrl27FBcXp0GDBmnatGmO11JSUhQfH69atWpp8eLFql69ul5//XUlJibqp59+0oMPPljie3QX1mrXrq2NGzfqlltu0ejRozVmzBhJcoR2T73//vvavXu35s2bpypVqjhmdFJSUtSoUSOPxu+nn35Sly5d5O/vryVLligyMlKrVq266NrwosLCwjRo0CAtX75c9957r6SCsO7j46MhQ4aU+/r+77//Xn5+fqpatWqxdbZt26aePXuqdevWWrZsmQIDA7VkyRL169dPb7zxhsvfyZgxY3Trrbfqb3/7m44cOaLp06dr2LBhpfpbL+rgwYM6fPiwBgwY4Cj7+uuvJUmtW7d2qd+6dWv985//dCkfM2aMBg4cqDfffFPJycl6+OGHlZubq5SUFA0cOFD33HOPNm/erIULF6pOnTqaOnWqx31F5RLk76v983pVyH7LS7t27Zye5+Xl6U9/+pNWr16to0ePKjs7W9nZ2QoJCSlxOxf+LRUupfH2eo4DBw4oLi7OaVKkU6dO+u233/Tjjz+qTZs26t69u1q1aqVevXopISFBgwYNUtWqVVWtWjUlJiaqV69e6tmzp3r06KHbb79dtWvX9qovDhZwlRg5cqQVEhLiVNalSxdLkvXRRx+V2DY/P9/Kycmxtm3bZkmyvvzyS8drs2fPtor+KcXExFh2u906fPiwo+zcuXNWtWrVrHvvvddRtmXLFkuStWXLFqd+SrL+/ve/O22zT58+VrNmzRzPX3jhBUuS9cEHHzjVu/feey1J1ooVK0p8T3a73erYsWOJdYq+p5EjR7qUd+nSxerSpYvLe7rppptc6i5evNiSZKWkpDjKfvnlFyswMNCaNm2ao6xXr15WvXr1rPT0dKf2999/v2W3261ffvnFsizL6tu3r3X99deX+j1cSJI1YcIEp7KhQ4dagYGBVmpqqlN57969reDgYOv06dMXfY/u/Pzzz5Yka/bs2S6vjRw50oqJiXEpd3dcSbIiIyOtjIwMR9nx48ctHx8fa8GCBY6y0o7fQw89ZNlsNmvv3r1O9Xr27OlyXLqzYsUKS5K1e/dux5h8/fXXlmVZ1u9+9zsrMTHRsizLuu6665yOkcL3UnT8i+rSpYt13XXXWTk5OVZOTo517Ngxa8aMGZYka/DgwSW27dixo1WrVi3rzJkzjrLc3FyrZcuWVr169az8/Hyn9zB+/Hin9k8++aQlyUpLSytxP0Xl5ORYXbt2tcLCwpyOo1WrVlmSrF27drm0ueeee6yAgADH88I+TZw40anegAEDLEnW008/7VR+/fXXW23btvWonzDfuXPnrP3791vnzp2r6K54ZcWKFVZ4eLjj+aFDhyxJVnJyslO9hQsXWtWrV7dee+01a+/evdZ3331n3XrrrdZtt93mqNOlSxfrj3/8o+N5TEyM9cwzzzhtp02bNm7/jS00cuRIp21eaMCAAdaoUaOcypKTky1Jjr/j/Px865NPPrFmzZpltWrVyqpZs6Z18OBBR/0vvvjCeuKJJ6y4uDirSpUqbv/Wiyrpd8wSF1z1qlatqptvvtml/ODBg7rzzjsVFRUlX19f+fv7q0uXLpIKzrYv5vrrr3f6iMtut6tp06alunDEZrOpX79+TmWtW7d2artt2zaFhobqlltucapXuNygov3hD39wKbvrrrsUGBjotPzmjTfeUHZ2tkaNGiVJysrK0kcffaTf//73Cg4OVm5uruPRp08fZWVl6dNPP5UktW/fXl9++aXGjx+vDz/8sExrnSXp448/Vvfu3RUdHe1UnpiYqMzMTJeLgd29x0utW7duThddRkZGqlatWo5jw5Px27Jli6677jq1adPGaR933nmnx/3q0qWLGjdurOXLl2vfvn3avXt3uSxv+eabb+Tv7y9/f3/VqVNHTz31lO666y793//9X7Ftzp49q88++0yDBg1SlSpVHOW+vr4aPny4fvzxR6WkpDi16d+/v9Pzwtk5Ty70sixLo0eP1o4dO/Tqq6+6HEdS8behc1de9CLtFi1aSCr4FKhoeVkvSAMqyo4dO3Tbbbdp2LBhatOmjRo1alTsXZAulWuvvVY7d+50uhh2586dCg0NVd26dSUV/I126tRJc+fOVXJysgICApyWpt1www2aOXOmdu7cqZYtW+pvf/tbmfrEEhdc9dx9DPXbb7+pc+fOstvtmj9/vpo2barg4GAdOXJEAwcO1Llz5y663erVq7uUBQYGlqptcHCw7Ha7S9usrCzH81OnTikyMtKlrbsyd+rXr69Dhw6Vqq433I1rtWrV1L9/f7366qt67LHH5Ovrq5UrV6p9+/a67rrrJBW8r9zcXD333HN67rnn3G775MmTkgqWmISEhOj111/Xiy++KF9fX910001auHChy8eopXHq1Cm3/a5Tp47j9Yu9x0vtYseVJ+N36tQpNWzY0OX1qKgoj/tls9k0atQoLV68WFlZWWratKk6d+7s8XaKaty4sd58803ZbDbZ7XY1bNjwohdx//rrr7Isy6PfZdFxDQwMlKRS/b1KBeF8zJgxev311/XXv/5Vt912m9vtF92vJP3yyy+qVq2aS3nRsoCAgGLLL/y3AahMmjRporVr12rnzp2qWrWqnn76aR0/ftxxQlqe0tPTtXfvXqeyatWqafz48Vq0aJEmTpyo+++/XykpKZo9e7amTp0qHx8fffbZZ/roo4+UkJCgWrVq6bPPPtPPP/+sFi1a6NChQ3r55ZfVv39/1alTRykpKfr3v/+tESNGlKmvBHRc9dzNXH388cc6duyYtm7d6pg1l1Ti2ubLrXr16vrXv/7lUn78+PFSte/Vq5eee+45ffrpp6Vah263293eCu7kyZOqUaOGS3lxM4WjRo3SW2+9paSkJNWvX1+7d+/W0qVLHa9XrVrVMdM5YcIEt9soDJV+fn6aOnWqpk6dqtOnT2vz5s16+OGH1atXLx05csSju/FIBWOalpbmUl54sWDR91nWq/SlksfVG56MX/Xq1d0eL6U9hopKTEzUrFmz9OKLL+rxxx/3ahtF2e12j0+2qlatKh8fH49+l2VRGM5XrFihZcuWadiwYS51WrZsKUnat2+f41qLQvv27XO8Dlxt/t//+386dOiQevXqpeDgYN1zzz0aMGCA0tPTy31fW7du1Q033OBUVvjlSRs2bND06dPVpk0bVatWTaNHj9ajjz4qqeA6m+3bt2vRokXKyMhQTEyMnnrqKfXu3Vs//fSTvv32W/31r391TPLcf//9jutxvEVAB9woDF6Fs2iFXnrppYrojltdunTR3//+d33wwQdOtwp88803S9V+ypQpWr58ucaPH+9ym0WpIHS88847+v3vfy+p4C4uX331lVOdf//730pJSfEo7CQkJKhu3bpasWKF6tevL7vd7rQsJzg4WN26dVNycrJat27tmDW8mIiICA0aNEhHjx7V5MmT9cMPP+jaa68tdb8kqXv37lq3bp2OHTvmmGmVpFdffVXBwcFeX1Bb0mxsgwYNdOLECf3000+OTz/Onz+vDz/07lZqnoxft27d9OSTT+rLL790Wubi7UezdevW1fTp0/Xtt99q5MiRXm2jPISEhKhDhw56++239Ze//EVBQUGSCu768/rrr6tevXpq2rRpuezLsiyNHTtWK1as0EsvveRYqlVU3bp11b59e73++ut64IEH5OtbcNHdp59+qpSUlKvua8xx5UtMTHS681eDBg3c3k+9WrVqF/12z6J3Z/nhhx9c6hSdGS9q5cqVJd7drEuXLm4nvaSCZWQbN250+1pkZGS53ZL4QgR0wI34+HhVrVpV48aN0+zZs+Xv769Vq1bpyy+/rOiuOYwcOVLPPPOMhg0bpvnz56tJkyb64IMPHMGu8G40xWnYsKHefPNNDRkyRNdff73uv/9+x8zC/v37tXz5clmW5Qjow4cP17BhwzR+/Hj94Q9/0OHDh/Xkk096fDcSX19fjRgxQk8//bTCwsI0cOBAl5ODZ599VjfeeKM6d+6s++67Tw0aNNCZM2f0n//8R++9957jzhr9+vVTy5Yt1a5dO9WsWVOHDx/WokWLFBMTo2uuucajfknS7NmztX79enXr1k2zZs1StWrVtGrVKr3//vt68skn3d6OsjRCQ0MVExOjf/zjH+revbuqVaumGjVqqEGDBhoyZIhmzZqloUOHavr06crKytLixYuVl5fn1b6k0o/f5MmTtXz5ct16662aP3++4y4u3377rdf7/tOf/lTqut9//73WrFnjUn7ttdd6fHJV1IIFC9SzZ09169ZNDzzwgAICArRkyRJ9/fXXeuONN8rl0w9JmjRpkpYtW6a7775brVq1cqzvlwpOzC6crVu4cKF69uypwYMHa/z48Tpx4oRmzJihli1bFhvsAVyduEgUcKN69ep6//33FRwcrGHDhunuu+9WlSpVtHr16orumkNISIg+/vhjde3aVQ8++KD+8Ic/KDU1VUuWLJFUMKN8MX379nV85P7iiy+qT58+6tu3r5YuXapu3bo5zQrceeedevLJJ/Xhhx866ixdutSrmchRo0YpOztbP//8s9tgcu211+qLL75Qy5Yt9eijjyohIUGjR4/WmjVr1L17d0e9bt26afv27Ro3bpx69uypRx99VN27d9e2bdu8+gKbZs2aaefOnWrWrJkmTJigAQMG6Ouvv9aKFSs0ffp0j7d3oWXLlik4OFj9+/fX7373O82ZM0dSwYnSP/7xD50+fVqDBg3S9OnTNXjw4DKtXyzt+EVFRWnbtm269tprdd9992nYsGGy2+16/vnny/ReS2vjxo0aPHiwy+Pvf/97mbfdpUsXffzxxwoJCVFiYqKGDh2q9PR0vfvuu6W+FWlpvPfee5Kk5cuXKy4uzulReHJbqGvXrtqwYYPS0tLUr18/TZw4Ud26ddNHH33k8mkdgKubzXL3eQOASuuJJ57Qo48+qtTUVK+/4RQAULGysrJ06NAhNWzY0OWmAbgylPQ7ZokLUIkVznQ2b95cOTk5+vjjj7V48WINGzaMcA4AQCVFQAcqseDgYD3zzDP64YcflJ2drfr16+uhhx5yXHkOAAAqH5a4AAAAGIYlLle+kn7HXCQKAAAAGISADgAAABiEgA4AAAAYhIAOAACAMrPZbCU+LvxmUU81aNBAixYtKrd6puMuLgAAACiztLQ0x8+rV6/WrFmzlJKS4igLCgqqiG5VSsygAwAAoMyioqIcj/DwcNlsNqey7du3KzY2Vna7XY0aNdLcuXOVm5vraD9nzhzVr19fgYGBqlOnjiZNmiSp4Ft4Dx8+rClTpjhm4721dOlSNW7cWAEBAWrWrJlee+01p9eL64MkLVmyRNdcc43sdrsiIyM1aNAgr/txMcygAwAAVAaWJeVkXv79+gdLZQjFkvThhx9q2LBhWrx4sTp37qzvv/9e99xzjyRp9uzZWrNmjZ555hm9+eabuu6663T8+HF9+eWXkqS3335bbdq00T333KOxY8d63Yd169bpj3/8oxYtWqQePXpo/fr1GjVqlOrVq6du3bqV2IfPP/9ckyZN0muvvab4+Hj98ssv2rFjR5nGpCQEdAAAgMogJ1N6os7l3+/Dx6SAkDJt4vHHH9eMGTM0cuRISVKjRo302GOP6cEHH9Ts2bOVmpqqqKgo9ejRQ/7+/qpfv77at28vSapWrZp8fX0VGhqqqKgor/vwl7/8RYmJiRo/frwkaerUqfr000/1l7/8Rd26dSuxD6mpqQoJCVHfvn0VGhqqmJgY3XDDDWUak5KwxAUAAACX1J49ezRv3jxVqVLF8Rg7dqzS0tKUmZmpwYMH69y5c2rUqJHGjh2rdevWOS1/KQ8HDhxQp06dnMo6deqkAwcOSFKJfejZs6diYmLUqFEjDR8+XKtWrVJm5qX7NIMZdAAAgMrAP7hgNrsi9ltG+fn5mjt3rgYOHOjymt1uV3R0tFJSUpSUlKTNmzdr/Pjx+vOf/6xt27bJ39+/zPsvVHT9umVZjrKS+hAaGqovvvhCW7du1aZNmzRr1izNmTNHu3fvVkRERLn1rxABHQAAoDKw2cq81KSitG3bVikpKWrSpEmxdYKCgtS/f3/1799fEyZMUPPmzbVv3z61bdtWAQEBysvLK1MfWrRooU8++UQjRoxwlO3cuVMtWrQoVR/8/PzUo0cP9ejRQ7Nnz1ZERIQ+/vhjtycdZUVABwAAwCU1a9Ys9e3bV9HR0Ro8eLB8fHz01Vdfad++fZo/f75WrlypvLw8dejQQcHBwXrttdcUFBSkmJgYSQX3N9++fbuGDh2qwMBA1ahRo9h9HT16VHv37nUqq1+/vqZPn67bb79dbdu2Vffu3fXee+/p7bff1ubNmyWpxD6sX79eBw8e1E033aSqVatqw4YNys/PV7NmzS7NgFkAAAAwyrlz56z9+/db586dq+iueGXFihVWeHi4U9nGjRut+Ph4KygoyAoLC7Pat29vvfzyy5ZlWda6deusDh06WGFhYVZISIjVsWNHa/PmzY62u3btslq3bm0FBgZaJcXXmJgYS5LLY8WKFZZlWdaSJUusRo0aWf7+/lbTpk2tV1991dG2pD7s2LHD6tKli1W1alUrKCjIat26tbV69eoyjVFJv2ObZVnWpYn+AAAA8EZWVpYOHTqkhg0bym63V3R3cAmU9DvmLi4AAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAKDObzVbiIzEx0ettN2jQQIsWLSpVvcL9BQUFqXnz5vrzn/8sy7IcdX744QfZbDb5+fnp6NGjTu3T0tLk5+cnm82mH374wVG+du1adejQQeHh4QoNDdV1112nadOmOV5fuXKl2/dst9u9er9+XrUCAAAALpCWlub4efXq1Zo1a5ZSUlIcZUFBQZelH/PmzdPYsWOVlZWlzZs367777lNYWJjuvfdep3p16tTRq6++qpkzZzrK/vrXv6pu3bpKTU11lG3evFlDhw7VE088of79+8tms2n//v366KOPnLYXFhbm9H6lgpMWbzCDDgAAgDKLiopyPMLDw2Wz2ZzKtm/frtjYWNntdjVq1Ehz585Vbm6uo/2cOXNUv359BQYGqk6dOpo0aZIkqWvXrjp8+LCmTJnimJkuSWhoqKKiotSgQQONGTNGrVu31qZNm1zqjRw5UitWrHAqW7lypUaOHOlUtn79et14442aPn26mjVrpqZNm2rAgAF67rnnnOoVfb9RUVGKjIz0aAwLEdABAAAqAcuylJmTedkfFy4P8daHH36oYcOGadKkSdq/f79eeuklrVy5Uo8//rgkac2aNXrmmWf00ksv6bvvvtM777yjVq1aSZLefvtt1atXT/PmzVNaWprTTP3Fxmvr1q06cOCA/P39XV7v37+/fv31V33yySeSpE8++US//PKL+vXr51QvKipK33zzjb7++uuyDIFHWOICAABQCZzLPacOf+tw2ff72Z2fKdg/uEzbePzxxzVjxgzH7HSjRo302GOP6cEHH9Ts2bOVmpqqqKgo9ejRQ/7+/qpfv77at28vSapWrZp8fX0dM+MX89BDD+nRRx/V+fPnlZOTI7vd7piNv5C/v7+GDRum5cuX68Ybb9Ty5cs1bNgwlzA/ceJE7dixQ61atVJMTIw6duyohIQE3XXXXQoMDHTUS09PV5UqVZzaxsfHu529vxgCOgAAAC6pPXv2aPfu3Y4Zc0nKy8tTVlaWMjMzNXjwYC1atEiNGjXSLbfcoj59+qhfv37y8/M8qk6fPl2JiYn6+eef9cgjj+jmm29WfHy827qjR49WXFycnnjiCb311lvatWuX07IbSQoJCdH777+v77//Xlu2bNGnn36qadOm6dlnn9WuXbsUHFxw8hIaGqovvvjCqa236+4J6AAAAJVAkF+QPrvzswrZb1nl5+dr7ty5GjhwoMtrdrtd0dHRSklJUVJSkjZv3qzx48frz3/+s7Zt2+Z2eUpJatSooSZNmqhJkyZau3atmjRpoo4dO6pHjx4udVu2bKnmzZvrjjvuUIsWLdSyZUvt3bvX7XYbN26sxo0ba8yYMXrkkUfUtGlTrV69WqNGjZIk+fj4qEmTJh71tTgEdAAAgErAZrOVealJRWnbtq1SUlJKDLBBQUHq37+/+vfvrwkTJqh58+bat2+f2rZtq4CAAOXl5Xm836pVq2rixIl64IEHlJyc7PYC07vvvlvjx4/X0qVLS73dBg0aKDg4WGfPnvW4T6VBQAcAAMAlNWvWLPXt21fR0dEaPHiwfHx89NVXX2nfvn2aP3++Vq5cqby8PHXo0EHBwcF67bXXFBQUpJiYGEkFgXj79u0aOnSoAgMDVaNGjVLve8KECVq4cKHWrl2rQYMGubw+duxYDR48WBEREW7bz5kzR5mZmerTp49iYmJ0+vRpLV68WDk5OerZs6ejnmVZOn78uEv7WrVqycfHs/uycBcXAAAAXFK9evXS+vXrlZSUpN/97nfq2LGjnn76aUcAj4iI0P/93/+pU6dOat26tT766CO99957ql69uqSCe5v/8MMPaty4sWrWrOnRvmvWrKnhw4drzpw5ys/Pd3ndz89PNWrUKHa9e5cuXXTw4EGNGDFCzZs3V+/evXX8+HFt2rRJzZo1c9TLyMhQ7dq1XR4nTpzwqL+SZLPK4945AAAAKDdZWVk6dOiQGjZs6PW3UcJsJf2OmUEHAAAADEJABwAAAAxCQAcAAAAMQkAHAAAADEJABwAAMBT38rhylfS7JaADAAAYpvDbMzMzMyu4J7hUCn+37r4plS8qAgAAMIyvr68iIiIc99AODg52+y2YqHwsy1JmZqZOnDihiIgI+fr6utSpFPdBz8/P17FjxxQaGsrBCQAArgqWZenXX3+9ZF8nfynk5+fL19eXvFYKERERioqKcjtWlWIG/dixY4qOjq7obgAAAFx2wcHBqlGjRqUJvVu3blVUVFRFd8No/v7+bmfOC1WKgB4aGipJOnLkiMLCwiq4NwAAACgqIyND0dHRqlq1Kt9+WkaVIqAXnjGGhYUR0AEAAAxWWWb6TcZdXAAAAACDENABAAAAgxDQAQAAAIMQ0AEAAACDENABAAAAgxDQAQAAAIMQ0AEAAACDENABAAAAgxDQAQAAAIMQ0AEAAACDENABAAAAgxDQAQAAAIMQ0AEAAACDENABAAAAgxDQAQAAAIMQ0AEAAACDENABAAAAg/hVdAdMdNeGuxTkG6QqAVUUGhCqKv5VFBYQpioBVZx+Dg0IVah/aEGdgCry82E4AQAAUDYkyiKy87L11c9fedU2yC/IKbAXDfCFzx0/X/A8LCBMQX5Bstls5fyOAAAAUJkQ0Ivwsfnohe4v6Lfzv+nM+TM6k3NGZ86fKXh+4c8XvHYu95wk6VzuOZ3LPacT5054tW9fm6/rLP0FAf5iM/ih/qHy9/Uvz+EAAADAZUZAL8Lfx1831bvJozY5+Tk6e/5ssQH+t/O/KeN8hn7L+a3YsJ9r5SrPylN6drrSs9N1VEe96r/d1176GfwLXgsLCFMV/yoK9g+Wj41LEwAAACoKAb0c+Pv4K8IeoQh7hFftLctSVl6W2zBf+Cjp+W85v+lszllJUlZelrLOZenncz971Rcfm49C/EMcgb1ogL/w56KvFS7bCfAN8GrfAAAAIKAbwWazKcgvSEF+QaoVXMurbeTl5/1vht5NmM84n+GYrb/weWHdjPMZys3PVb6V72jnrUDfQKfAXnihrbuf3T0P8Q9hFh8AAFy1COhXCF8fX4UHhis8MNyr9pZlKTsv2yXMuyzb+W/Zha9fGPylggtts/OydSrrlFd9scnmmK13dxcdp3AfUEVh/mHOF94GhCrQN9CrfQMAAFQ0AjokFczi2/3ssvvZVSOohlfbyMvP09ncs85hvpQz+IWP8/nnZckqODHIOaO0s2le9cXfx9/lTjmlncGv4l9wIuDr4+vVvgEAAMqCgI5y4+vjq7CAMIUFhHm9jey87GIvtL3o8//O5luylJOfo1+yftEvWb943ZcQ/5BSLc8p7q46dl87t80EAAAeI6DDKIG+gQoMCvR6Fj/fyldmTuZF76JT3PPfzv+mrLwsSdLZnLOOi2+94efj53IbzIvdYefCC29D/EP48isAAK5C/N8fVxQfm0/BWvWAKqqt2l5tIycvp9hwX9q76uRb+crNz9Wv2b/q1+xfvX4/wX7BFw33Re+ic+FMP19+BQBA5UNAB4rw9/VXNd9qqmav5lV7y7KUmZtZYpgv6ULc33J+c3z5VWZupjJzM3VC3n/5VWmW6QT5BynAJ0ABvgHy9/F3+q+j3Ne/oOy/zwtf8/Px4yQAAIByREAHypnNZlOIf4hC/EMUFRLl1TZy8nMKLqI9/5sycjLcXnhb7PP/rsXPs/KUZ+XpdPZpnc4+Xb5vsogAn4IAf+F/LwzxjnB/wfML6/j7XPD6hScFPv7OdQvbFv58QZuiP3OrTgBAZUVABwzk7+Ovqvaqqmqv6lV7y7J0LvdcqWfws3KzdD7/vHLycnQ+77zO55/X+bzzys3PdXqek1/wep6V57S/8/kFdc7K+zX75c3Px8/1UwA3JwmlOVm4sI7jE4YiJw7Ffcpw4T65MxAAoDQI6MAVyGazKdg/WMH+wYpUZLlvPy8/ryCsFwb3vP/9XBj0C8O8u4DveM3NSUFOfo7b7TnaFbPP3Pxcpz7m5ucqNz/XsVzIBD42n7KdFJThUwa3JxQ+/ixRAgADEdABeMzXx1e+Pr6yy17RXXGwLMsp+F8Y4oueLFwY7l1OGtydZBSp66hTeDJRwkmIJcvRx3wrX1l5WY47BZnAJpvbpUKeLkkqzQlF0aVORU8oLnyNkwYAVzMCOoArgs1mc4RBU1iWpVwr96Ih/sITigs/JXD7aUMZThbcLVGyZDmWKCmnAgerCD8fP5cQ7+7CZW8ubr7w+gSbCk4ELjwhKCy7UOHrjvoX1Cn62n+fuNYrYV8ltb3YvtydzHi9r4vU93hfJYxrafdVlnG9nPuy/a+BZ/VVunH19vfg7+OvKgFVXOrDbAR0ALhEbDab/G0FATHYP7iiu+OQl59X+iVFZfwEobTLmIpbopSZm1lBowRcGTrX7awlPZZUdDfgIQI6AFxlfH18FeQTpCC/oIruikO+le8S3N2eOBT5JMER+D1YxpRv5UuSY/nRhcuQCn+8sMyyLM/qF9azSi4r+ppX+7Jc65XHvty2Lc99uRkbb/Zl/W9QXOtd4n2Vuv7F9lXC/suyL1RuBHQAQIXzsfkUfJOwb2BFdwW4IrgL/qg8COgAAABXGLfr6lFp8E0eAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEG8CuhLlixRw4YNZbfbFRsbqx07dpRYf9WqVWrTpo2Cg4NVu3ZtjRo1SqdOnfKqwwAAAMCVzOOAvnr1ak2ePFmPPPKIkpOT1blzZ/Xu3Vupqalu63/yyScaMWKERo8erW+++UZvvfWWdu/erTFjxpS58wAAAMCVxuOA/vTTT2v06NEaM2aMWrRooUWLFik6OlpLly51W//TTz9VgwYNNGnSJDVs2FA33nij7r33Xn3++efF7iM7O1sZGRlODwAAAOBq4FFAP3/+vPbs2aOEhASn8oSEBO3cudNtm/j4eP3444/asGGDLMvSTz/9pDVr1ujWW28tdj8LFixQeHi44xEdHe1JNwEAAIBKy6OAfvLkSeXl5SkyMtKpPDIyUsePH3fbJj4+XqtWrdKQIUMUEBCgqKgoRURE6Lnnnit2PzNnzlR6errjceTIEU+6CQAAAFRaXl0karPZnJ5bluVSVmj//v2aNGmSZs2apT179mjjxo06dOiQxo0bV+z2AwMDFRYW5vQAAAAArgZ+nlSuUaOGfH19XWbLT5w44TKrXmjBggXq1KmTpk+fLklq3bq1QkJC1LlzZ82fP1+1a9f2susAAADAlcejGfSAgADFxsYqKSnJqTwpKUnx8fFu22RmZsrHx3k3vr6+kgpm3gEAAAD8j8dLXKZOnapXXnlFy5cv14EDBzRlyhSlpqY6lqzMnDlTI0aMcNTv16+f3n77bS1dulQHDx7UP//5T02aNEnt27dXnTp1yu+dAAAAAFcAj5a4SNKQIUN06tQpzZs3T2lpaWrZsqU2bNigmJgYSVJaWprTPdETExN15swZPf/885o2bZoiIiJ08803a+HCheX3LgAAAIArhM2qBOtMMjIyFB4ervT0dC4YBQAAMBB5rfx4dRcXAAAAAJcGAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADCIVwF9yZIlatiwoex2u2JjY7Vjx44S62dnZ+uRRx5RTEyMAgMD1bhxYy1fvtyrDgMAAABXMj9PG6xevVqTJ0/WkiVL1KlTJ7300kvq3bu39u/fr/r167ttc/vtt+unn37SsmXL1KRJE504cUK5ubll7jwAAABwpbFZlmV50qBDhw5q27atli5d6ihr0aKFBgwYoAULFrjU37hxo4YOHaqDBw+qWrVqXnUyIyND4eHhSk9PV1hYmFfbAAAAwKVDXis/Hi1xOX/+vPbs2aOEhASn8oSEBO3cudNtm3fffVft2rXTk08+qbp166pp06Z64IEHdO7cuWL3k52drYyMDKcHAAAAcDXwaInLyZMnlZeXp8jISKfyyMhIHT9+3G2bgwcP6pNPPpHdbte6det08uRJjR8/Xr/88kux69AXLFiguXPnetI1AAAA4Irg1UWiNpvN6bllWS5lhfLz82Wz2bRq1Sq1b99effr00dNPP62VK1cWO4s+c+ZMpaenOx5HjhzxppsAAABApePRDHqNGjXk6+vrMlt+4sQJl1n1QrVr11bdunUVHh7uKGvRooUsy9KPP/6oa665xqVNYGCgAgMDPekaAAAAcEXwaAY9ICBAsbGxSkpKcipPSkpSfHy82zadOnXSsWPH9NtvvznK/v3vf8vHx0f16tXzossAAADAlcvjJS5Tp07VK6+8ouXLl+vAgQOaMmWKUlNTNW7cOEkFy1NGjBjhqH/nnXeqevXqGjVqlPbv36/t27dr+vTpuvvuuxUUFFR+7wQAAAC4Anh8H/QhQ4bo1KlTmjdvntLS0tSyZUtt2LBBMTExkqS0tDSlpqY66lepUkVJSUmaOHGi2rVrp+rVq+v222/X/Pnzy+9dAAAAAFcIj++DXhG4ryYAAIDZyGvlx6u7uAAAAAC4NAjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEG8CuhLlixRw4YNZbfbFRsbqx07dpSq3T//+U/5+fnp+uuv92a3AAAAwBXP44C+evVqTZ48WY888oiSk5PVuXNn9e7dW6mpqSW2S09P14gRI9S9e3evOwsAAABc6WyWZVmeNOjQoYPatm2rpUuXOspatGihAQMGaMGCBcW2Gzp0qK655hr5+vrqnXfe0d69e0u9z4yMDIWHhys9PV1hYWGedBcAAACXAXmt/Hg0g37+/Hnt2bNHCQkJTuUJCQnauXNnse1WrFih77//XrNnzy7VfrKzs5WRkeH0AAAAAK4GHgX0kydPKi8vT5GRkU7lkZGROn78uNs23333nWbMmKFVq1bJz8+vVPtZsGCBwsPDHY/o6GhPugkAAABUWl5dJGqz2ZyeW5blUiZJeXl5uvPOOzV37lw1bdq01NufOXOm0tPTHY8jR454000AAACg0indlPZ/1ahRQ76+vi6z5SdOnHCZVZekM2fO6PPPP1dycrLuv/9+SVJ+fr4sy5Kfn582bdqkm2++2aVdYGCgAgMDPekaAAAAcEXwaAY9ICBAsbGxSkpKcipPSkpSfHy8S/2wsDDt27dPe/fudTzGjRunZs2aae/everQoUPZeg8AAABcYTyaQZekqVOnavjw4WrXrp3i4uL08ssvKzU1VePGjZNUsDzl6NGjevXVV+Xj46OWLVs6ta9Vq5bsdrtLOQAAAAAvAvqQIUN06tQpzZs3T2lpaWrZsqU2bNigmJgYSVJaWtpF74kOAAAAwD2P74NeEbivJgAAgNnIa+XHq7u4AAAAALg0COgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQbwK6EuWLFHDhg1lt9sVGxurHTt2FFv37bffVs+ePVWzZk2FhYUpLi5OH374odcdBgAAAK5kHgf01atXa/LkyXrkkUeUnJyszp07q3fv3kpNTXVbf/v27erZs6c2bNigPXv2qFu3burXr5+Sk5PL3HkAAADgSmOzLMvypEGHDh3Utm1bLV261FHWokULDRgwQAsWLCjVNq677joNGTJEs2bNKlX9jIwMhYeHKz09XWFhYZ50FwAAAJcBea38eDSDfv78ee3Zs0cJCQlO5QkJCdq5c2eptpGfn68zZ86oWrVqxdbJzs5WRkaG0wMAAAC4GngU0E+ePKm8vDxFRkY6lUdGRur48eOl2sZTTz2ls2fP6vbbby+2zoIFCxQeHu54REdHe9JNAAAAoNLy6iJRm83m9NyyLJcyd9544w3NmTNHq1evVq1atYqtN3PmTKWnpzseR44c8aabAAAAQKXj50nlGjVqyNfX12W2/MSJEy6z6kWtXr1ao0eP1ltvvaUePXqUWDcwMFCBgYGedA0AAAC4Ing0gx4QEKDY2FglJSU5lSclJSk+Pr7Ydm+88YYSExP1t7/9Tbfeeqt3PQUAAACuAh7NoEvS1KlTNXz4cLVr105xcXF6+eWXlZqaqnHjxkkqWJ5y9OhRvfrqq5IKwvmIESP07LPPqmPHjo7Z96CgIIWHh5fjWwEAAAAqP48D+pAhQ3Tq1CnNmzdPaWlpatmypTZs2KCYmBhJUlpamtM90V966SXl5uZqwoQJmjBhgqN85MiRWrlyZdnfAQAAAHAF8fg+6BWB+2oCAACYjbxWfry6iwsAAACAS4OADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGMSvojtgnPw86aUukk2SbJLNVvx/bT4l1NEFz31K3o7LdktTp7h9u+l3if30Yt/FvqeK3HfRMb/c+3b3/kv5e7f5uPb9ch1zTtvTBdsAAAAVhYBelGVJP+2r6F4ABrkw/Hvw3Js2Ls/LoR/l0q/i+nC5+1XafV6ufpW235e6X572o4JPQiv0JJj3XnG7r6D9R7WS4idWzL7hNQJ6UTYfadhayZIkqyCwl/jf/GJekwd1i/mvR3Xzi5SV0H+XuiX0v1R1i263FGNX4nblQV1TxkyXv7+XzYW/E9diAIDBzv1KQK+ECOhF+fhITXpUdC+A0iku5Ht8kiA5Je6iZeX2vLR1vNjHJel3kT5fkn14MjbF9OuS7MOTbeoidcrpuTf9Kk2/jWBovxiv0jN1rKrGVHQP4AUCOlCZOdbPAwCAKwV3cQEAAAAMQkAHAAAADEJABwAAAAxCQAcAAAAMQkAHAAAADEJABwAAAAxCQAcAAAAMQkAHAAAADOJVQF+yZIkaNmwou92u2NhY7dixo8T627ZtU2xsrOx2uxo1aqQXX3zRq84CAAAAVzqPA/rq1as1efJkPfLII0pOTlbnzp3Vu3dvpaamuq1/6NAh9enTR507d1ZycrIefvhhTZo0SWvXri1z5wEAAIArjc2yLMuTBh06dFDbtm21dOlSR1mLFi00YMAALViwwKX+Qw89pHfffVcHDhxwlI0bN05ffvmldu3a5XYf2dnZys7OdjzPyMhQdHS00tPTFRYW5kl3PZaXb2nIS+77BQAAUJm0jamqh/u0uCz7ysjIUHh4+GXJa1c6P08qnz9/Xnv27NGMGTOcyhMSErRz5063bXbt2qWEhASnsl69emnZsmXKycmRv7+/S5sFCxZo7ty5nnStXH1++NcK2zcAAEB5CbV7FPVgCI9+aydPnlReXp4iIyOdyiMjI3X8+HG3bY4fP+62fm5urk6ePKnatWu7tJk5c6amTp3qeF44g345+NikF4fFXpZ9AWbw6EM0AEAlUjM0sKK7AC94dVpls9mcnluW5VJ2sfruygsFBgYqMLBiDiibzaZbWkZVyL4BAAAAjy4SrVGjhnx9fV1my0+cOOEyS14oKirKbX0/Pz9Vr17dw+4CAAAAVzaPAnpAQIBiY2OVlJTkVJ6UlKT4+Hi3beLi4lzqb9q0Se3atXO7/hwAAAC4mnl8m8WpU6fqlVde0fLly3XgwAFNmTJFqampGjdunKSC9eMjRoxw1B83bpwOHz6sqVOn6sCBA1q+fLmWLVumBx54oPzeBQAAAHCF8HgN+pAhQ3Tq1CnNmzdPaWlpatmypTZs2KCYmBhJUlpamtM90Rs2bKgNGzZoypQpeuGFF1SnTh0tXrxYf/jDH8rvXQAAAABXCI/vg14RuK8mAACA2chr5cfjJS4AAAAALh0COgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEL+K7kBpWJYlScrIyKjgngAAAMCdwpxWmNvgvUoR0M+cOSNJio6OruCeAAAAoCRnzpxReHh4RXejUrNZleA0Jz8/X8eOHVNoaKhsNtsl319GRoaio6N15MgRhYWFXfL9oQDjXjEY94rBuFcMxr1iMO4V43KPu2VZOnPmjOrUqSMfH1ZRl0WlmEH38fFRvXr1Lvt+w8LC+IekAjDuFYNxrxiMe8Vg3CsG414xLue4M3NePji9AQAAAAxCQAcAAAAMQkB3IzAwULNnz1ZgYGBFd+WqwrhXDMa9YjDuFYNxrxiMe8Vg3CuvSnGRKAAAAHC1YAYdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADDIVRnQt2/frn79+qlOnTqy2Wx65513Ltpm27Ztio2Nld1uV6NGjfTiiy9e+o5eQTwd861bt8pms7k8vv3228vT4SvEggUL9Lvf/U6hoaGqVauWBgwYoJSUlIu243gvG2/GnWO+7JYuXarWrVs7vjUxLi5OH3zwQYltONbLztNx51gvfwsWLJDNZtPkyZNLrMfxXnlclQH97NmzatOmjZ5//vlS1T906JD69Omjzp07Kzk5WQ8//LAmTZqktWvXXuKeXjk8HfNCKSkpSktLczyuueaaS9TDK9O2bds0YcIEffrpp0pKSlJubq4SEhJ09uzZYttwvJedN+NeiGPee/Xq1dOf/vQnff755/r88891880367bbbtM333zjtj7HevnwdNwLcayXj927d+vll19W69atS6zH8V7JWFc5Sda6detKrPPggw9azZs3dyq79957rY4dO17Cnl25SjPmW7ZssSRZv/7662Xp09XixIkTliRr27ZtxdbheC9/pRl3jvlLo2rVqtYrr7zi9jWO9UunpHHnWC8/Z86csa655horKSnJ6tKli/XHP/6x2Loc75XLVTmD7qldu3YpISHBqaxXr176/PPPlZOTU0G9ujrccMMNql27trp3764tW7ZUdHcqvfT0dElStWrViq3D8V7+SjPuhTjmy0deXp7efPNNnT17VnFxcW7rcKyXv9KMeyGO9bKbMGGCbr31VvXo0eOidTneKxe/iu5AZXD8+HFFRkY6lUVGRio3N1cnT55U7dq1K6hnV67atWvr5ZdfVmxsrLKzs/Xaa6+pe/fu2rp1q2666aaK7l6lZFmWpk6dqhtvvFEtW7Ysth7He/kq7bhzzJePffv2KS4uTllZWapSpYrWrVuna6+91m1djvXy48m4c6yXjzfffFNffPGFdu/eXar6HO+VCwG9lGw2m9Nzy7LclqN8NGvWTM2aNXM8j4uL05EjR/SXv/yFf8C9dP/99+urr77SJ598ctG6HO/lp7TjzjFfPpo1a6a9e/fq9OnTWrt2rUaOHKlt27YVGxY51suHJ+POsV52R44c0R//+Edt2rRJdru91O043isPlriUQlRUlI4fP+5UduLECfn5+al69eoV1KurT8eOHfXdd99VdDcqpYkTJ+rdd9/Vli1bVK9evRLrcryXH0/G3R2Oec8FBASoSZMmateunRYsWKA2bdro2WefdVuXY738eDLu7nCse2bPnj06ceKEYmNj5efnJz8/P23btk2LFy+Wn5+f8vLyXNpwvFcuzKCXQlxcnN577z2nsk2bNqldu3by9/evoF5dfZKTk/kIzkOWZWnixIlat26dtm7dqoYNG160Dcd72Xkz7u5wzJedZVnKzs52+xrH+qVT0ri7w7Hume7du2vfvn1OZaNGjVLz5s310EMPydfX16UNx3slU1FXp1akM2fOWMnJyVZycrIlyXr66aet5ORk6/Dhw5ZlWdaMGTOs4cOHO+ofPHjQCg4OtqZMmWLt37/fWrZsmeXv72+tWbOmot5CpePpmD/zzDPWunXrrH//+9/W119/bc2YMcOSZK1du7ai3kKldN9991nh4eHW1q1brbS0NMcjMzPTUYfjvfx5M+4c82U3c+ZMa/v27dahQ4esr776ynr44YctHx8fa9OmTZZlcaxfKp6OO8f6pVH0Li4c75XbVRnQC2/xVPQxcuRIy7Isa+TIkVaXLl2c2mzdutW64YYbrICAAKtBgwbW0qVLL3/HKzFPx3zhwoVW48aNLbvdblWtWtW68cYbrffff79iOl+JuRtzSdaKFSscdTjey583484xX3Z33323FRMTYwUEBFg1a9a0unfv7giJlsWxfql4Ou4c65dG0YDO8V652Szrv1cIAAAAAKhwXCQKAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABjk/wM0tgQTHC6PcAAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 800x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# MLP Hyperparameters\n",
    "learning_rate = 0.001\n",
    "weight_decay = 0.0000001\n",
    "epochs = 4\n",
    "batch_size = 1024\n",
    "latent_dims = 32\n",
    "alpha = 0.5\n",
    "loss = nn.MSELoss()\n",
    "\n",
    "print('TRAINING MLP')\n",
    "train_model = train_mlp\n",
    "final_mlp_model, train_loss, test_loss, test_rmse = train_model(train_dataloader,\n",
    "                                                      test_dataloader,\n",
    "                                                      num_users,\n",
    "                                                      num_items,\n",
    "                                                      epochs,\n",
    "                                                      latent_dims,\n",
    "                                                      learning_rate,\n",
    "                                                      'adam',\n",
    "                                                      criterion=loss,\n",
    "                                                      device=device,\n",
    "                                                      weight_decay=weight_decay,\n",
    "                                                      rescale_data=True)\n",
    "\n",
    "\n",
    "model_name = f\"models/final_mlp_model_{dataset}\"\n",
    "torch.save(final_mlp_model.state_dict(), model_name)\n",
    "\n",
    "curve_data = {\n",
    "    'Train Loss': train_loss,\n",
    "    'Test Loss': test_loss,\n",
    "    'Test RMSE': test_rmse,\n",
    "}\n",
    "title = f\"Training Curves for tuned MLP on {dataset}\"\n",
    "plot_learning_curve(np.arange(1, epochs+1), curve_data, title, True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "86aa44a0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "PRETRAINING GMF\n",
      "Epoch 1\n",
      "------------------------\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.971519  [ 1024/15945812]\n",
      "loss: 0.055000  [1025024/15945812]\n",
      "loss: 0.048120  [2049024/15945812]\n",
      "loss: 0.042417  [3073024/15945812]\n",
      "loss: 0.044311  [4097024/15945812]\n",
      "loss: 0.043971  [5121024/15945812]\n",
      "loss: 0.048192  [6145024/15945812]\n",
      "loss: 0.043231  [7169024/15945812]\n",
      "loss: 0.039006  [8193024/15945812]\n",
      "loss: 0.033176  [9217024/15945812]\n",
      "loss: 0.029561  [10241024/15945812]\n",
      "loss: 0.031623  [11265024/15945812]\n",
      "loss: 0.027647  [12289024/15945812]\n",
      "loss: 0.031495  [13313024/15945812]\n",
      "loss: 0.027189  [14337024/15945812]\n",
      "loss: 0.027853  [15361024/15945812]\n",
      "Avg loss: 0.031862 \n",
      "\n",
      "Test MSE 0.7965283445540001\n",
      "Test RMSE 0.8924843665599976\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.028991  [ 1024/15945812]\n",
      "loss: 0.026254  [1025024/15945812]\n",
      "loss: 0.028060  [2049024/15945812]\n",
      "loss: 0.025611  [3073024/15945812]\n",
      "loss: 0.030032  [4097024/15945812]\n",
      "loss: 0.028550  [5121024/15945812]\n",
      "loss: 0.027172  [6145024/15945812]\n",
      "loss: 0.028619  [7169024/15945812]\n",
      "loss: 0.027485  [8193024/15945812]\n",
      "loss: 0.025876  [9217024/15945812]\n",
      "loss: 0.030403  [10241024/15945812]\n",
      "loss: 0.026613  [11265024/15945812]\n",
      "loss: 0.026908  [12289024/15945812]\n",
      "loss: 0.024620  [13313024/15945812]\n",
      "loss: 0.027182  [14337024/15945812]\n",
      "loss: 0.026805  [15361024/15945812]\n",
      "Avg loss: 0.029142 \n",
      "\n",
      "Test MSE 0.7285624713173602\n",
      "Test RMSE 0.8535587099417123\n",
      "TRAINING NMF\n",
      "Epoch 1\n",
      "------------------------\n",
      "loss: 0.074664  [ 1024/15945812]\n",
      "loss: 0.026165  [1025024/15945812]\n",
      "loss: 0.026977  [2049024/15945812]\n",
      "loss: 0.026655  [3073024/15945812]\n",
      "loss: 0.026677  [4097024/15945812]\n",
      "loss: 0.025410  [5121024/15945812]\n",
      "loss: 0.026343  [6145024/15945812]\n",
      "loss: 0.027717  [7169024/15945812]\n",
      "loss: 0.023873  [8193024/15945812]\n",
      "loss: 0.025444  [9217024/15945812]\n",
      "loss: 0.025024  [10241024/15945812]\n",
      "loss: 0.027767  [11265024/15945812]\n",
      "loss: 0.026698  [12289024/15945812]\n",
      "loss: 0.027317  [13313024/15945812]\n",
      "loss: 0.025064  [14337024/15945812]\n",
      "loss: 0.025945  [15361024/15945812]\n",
      "Avg loss: 0.028422 \n",
      "\n",
      "Test MSE 0.710546368369142\n",
      "Test RMSE 0.8429391249486181\n",
      "Epoch 2\n",
      "------------------------\n",
      "loss: 0.025715  [ 1024/15945812]\n",
      "loss: 0.023941  [1025024/15945812]\n",
      "loss: 0.026948  [2049024/15945812]\n",
      "loss: 0.024835  [3073024/15945812]\n",
      "loss: 0.023334  [4097024/15945812]\n",
      "loss: 0.024045  [5121024/15945812]\n",
      "loss: 0.026019  [6145024/15945812]\n",
      "loss: 0.023346  [7169024/15945812]\n",
      "loss: 0.023581  [8193024/15945812]\n",
      "loss: 0.023836  [9217024/15945812]\n",
      "loss: 0.023438  [10241024/15945812]\n",
      "loss: 0.026391  [11265024/15945812]\n",
      "loss: 0.025728  [12289024/15945812]\n",
      "loss: 0.026920  [13313024/15945812]\n",
      "loss: 0.024921  [14337024/15945812]\n",
      "loss: 0.025062  [15361024/15945812]\n",
      "Avg loss: 0.028325 \n",
      "\n",
      "Test MSE 0.7081430503220998\n",
      "Test RMSE 0.8415123589835741\n",
      "Epoch 3\n",
      "------------------------\n",
      "loss: 0.024730  [ 1024/15945812]\n",
      "loss: 0.021703  [1025024/15945812]\n",
      "loss: 0.025325  [2049024/15945812]\n",
      "loss: 0.026393  [3073024/15945812]\n",
      "loss: 0.021792  [4097024/15945812]\n",
      "loss: 0.024877  [5121024/15945812]\n",
      "loss: 0.026041  [6145024/15945812]\n",
      "loss: 0.024317  [7169024/15945812]\n",
      "loss: 0.024401  [8193024/15945812]\n",
      "loss: 0.024903  [9217024/15945812]\n",
      "loss: 0.022407  [10241024/15945812]\n",
      "loss: 0.024659  [11265024/15945812]\n",
      "loss: 0.022679  [12289024/15945812]\n",
      "loss: 0.021218  [13313024/15945812]\n",
      "loss: 0.023889  [14337024/15945812]\n",
      "loss: 0.024979  [15361024/15945812]\n",
      "Avg loss: 0.027831 \n",
      "\n",
      "Test MSE 0.6957575184532006\n",
      "Test RMSE 0.8341208056709775\n",
      "Epoch 4\n",
      "------------------------\n",
      "loss: 0.022267  [ 1024/15945812]\n",
      "loss: 0.022510  [1025024/15945812]\n",
      "loss: 0.021708  [2049024/15945812]\n",
      "loss: 0.022731  [3073024/15945812]\n",
      "loss: 0.025120  [4097024/15945812]\n",
      "loss: 0.025251  [5121024/15945812]\n",
      "loss: 0.026490  [6145024/15945812]\n",
      "loss: 0.023983  [7169024/15945812]\n",
      "loss: 0.026135  [8193024/15945812]\n",
      "loss: 0.023734  [9217024/15945812]\n",
      "loss: 0.025484  [10241024/15945812]\n",
      "loss: 0.026196  [11265024/15945812]\n",
      "loss: 0.024373  [12289024/15945812]\n",
      "loss: 0.023167  [13313024/15945812]\n",
      "loss: 0.023431  [14337024/15945812]\n",
      "loss: 0.021904  [15361024/15945812]\n",
      "Avg loss: 0.027904 \n",
      "\n",
      "Test MSE 0.6975809341486864\n",
      "Test RMSE 0.8352131070263962\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAugAAAI6CAYAAAB1rL20AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8fJSN1AAAACXBIWXMAAA9hAAAPYQGoP6dpAABGK0lEQVR4nO3deVyVZf7/8fdhPaCAO7ggSOaSa2IuqKmpmGZmZWnlbpapmUvuft0ySafFtLScXNJxyjJtSk3Fcmu0Rg3NcqnJBVOItFGcVBS4f384nJ9HDshhkQt9PWfux3iuc133/bnvc8O8z32uc2OzLMsSAAAAACN4FHYBAAAAAP4/AjoAAABgEAI6AAAAYBACOgAAAGAQr8IuAAAAAFlLS0vTlStXCrsM5CNvb295enpm+TwBHQAAwECWZSkxMVFnz54t7FJQAEqUKKGQkBDZbLZMzxHQAQAADJQRzsuVKyd/f3+XQQ5Fj2VZunDhgpKSkiRJ5cuXz9SHgA4AAGCYtLQ0RzgvXbp0YZeDfObn5ydJSkpKUrly5TJNd+FLogAAAIbJmHPu7+9fyJWgoGS8tq6+X0BABwAAMBTTWm5d2b22BHQAAADAIAR0AAAAwCAEdAAAABirVatWGjZsWGGXcVNxFxcAAADk2Y3my/fu3VtLlixxe72rVq2St7d3Lqu6qk+fPjp79qw+/fTTPK3nZuEKOm45NpstR8uWLVvytJ0pU6bk+ss7W7ZsyZca8uLIkSMaMmSIqlWrJj8/P/n7+6tWrVqaOHGiTp48WWh1FbS4uDi1bNlSQUFBstlsmj17doFt68KFC5oyZUqhvs7uyOl5uWTJEtlsNtntdh0/fjzT861atVLt2rWd2sLDw2Wz2dSqVSuX61y6dKnLn82MnzNXy1tvveXuLua75ORkvfzyy2rVqpVCQkJUvHhx1alTRzNnztSlS5cy9b9y5YqmTp2q8PBw+fr6qkaNGpo7d24hVA7kv4SEBMcye/ZsBQYGOrW9+eabTv1z+tdRS5UqpYCAgIIo2VgEdNxydu7c6bR07NhRfn5+mdobNGiQp+08/fTT2rlzZ67GNmjQIF9qyK01a9aobt26WrNmjZ555hmtWbPG8e/PP/9cnTp1KpS6boZ+/fopISFBH374oXbu3Knu3bsX2LYuXLigqVOnFpmA7q6UlBRNnDgxx/0DAgK0bds2/fLLL5meW7RokQIDA7Mcu379+kw/w4899liu6s5P8fHxmj17tho0aKAFCxbos88+U9euXTVlyhR16tRJlmU59R80aJBiYmI0ePBgbdiwQQ8//LBeeOEFzZgxo5D2AEWJZVm6cDn1pi/Xn8dZCQkJcSwZF0EyHl+6dEklSpTQRx99pFatWslut+tvf/ubzpw5oyeeeEKVKlWSv7+/6tSpow8++MBpvddPcQkPD9eMGTPUr18/BQQEqHLlylqwYEGeju3WrVvVqFEj+fr6qnz58ho7dqxSU1Mdz69cuVJ16tSRn5+fSpcurbZt2+rPP/+UdPXiRqNGjVSsWDGVKFFCzZo1c3nxwh1MccEtp0mTJk6Py5YtKw8Pj0zt17tw4YJb95utVKmSKlWqlKsaAwMDb1hPQTl69Ki6d++uatWqafPmzQoKCnI8d99992no0KFavXp1vmzrypUrstls8vIy51fNDz/8oAEDBqhDhw75sj4T9/Fmuf/++/X3v/9dL774ourVq3fD/s2bN9f+/fu1aNEivfzyy472X375Rdu2bdPTTz+tv/71ry7HRkZGqkyZMvlWe36pUqWKjh07pmLFijna7rvvPhUrVkyjRo3SP//5TzVv3lyS9OOPP2rhwoV6+eWXNWrUKElXg8eZM2c0ffp0DRw4UKVKlSqU/UDRcPFKmu6atOGmb/fAtPby98mf33FjxozRa6+9psWLF8vX11eXLl1SZGSkxowZo8DAQK1du1Y9e/ZURESEGjdunOV6XnvtNb300ksaP368Vq5cqeeee0733nuvatSo4XZNJ0+eVMeOHdWnTx8tXbpUhw4d0oABA2S32zVlyhQlJCToiSee0KxZs/Twww/r/Pnz2r59uyzLUmpqqrp06aIBAwbogw8+0OXLl/Wvf/0rz7fH5Ao6bksZH8Fv27ZNUVFR8vf3V79+/SRJK1asUHR0tMqXLy8/Pz/VrFlTY8eOdbxTzuBqikt4eLg6deqk9evXq0GDBvLz81ONGjW0aNEip36uphL06dNHxYsX17///W917NhRxYsXV2hoqEaOHKmUlBSn8b/++qu6du2qgIAAlShRQk899ZR27dolm812w/l9r7/+uv7880/NmzfPKZxnsNlseuSRR5z2qU+fPi6P4bXTFTL2admyZRo5cqQqVqwoX19f/fjjj7LZbFq4cGGmdXzxxRey2Wz67LPPHG0///yznnzySZUrV06+vr6qWbOm3n77badx6enpmj59uqpXry4/Pz+VKFFCdevWzfTx6bUypmWkpqZq/vz5jmkSGX744Qc99NBDKlmypOx2u+rXr6/333/faR1Z7eO///3vTNs7duyYypYtK0maOnWqY3sZx7JPnz4KDw/PNM7VeWWz2TRkyBAtW7ZMNWvWlL+/v+rVq6c1a9ZkGp+T4ydJhw4d0v333y9/f3+VKVNGAwcO1Pnz57M8fq6MHj1apUuX1pgxY3LU38PDQ7169dL777+v9PR0R/uiRYsUGhqqtm3burX9G0lPT9esWbNUo0YN+fr6qly5curVq5d+/fVXp34Zvw927dqlFi1ayN/fXxEREXrllVec6nSlWLFiTuE8Q6NGjSRJJ06ccLR9+umnsixLffv2derbt29fXbx4UevXr89U086dOxUVFSU/Pz+Fh4dr8eLFkqS1a9eqQYMGjiuO144FTDZs2DA98sgjqlKliipUqKCKFSvqxRdfVP369RUREaHnn39e7du318cff5ztejp27KhBgwapatWqGjNmjMqUKZPrTyvnzZun0NBQvfXWW6pRo4a6dOmiqVOn6rXXXlN6eroSEhKUmpqqRx55ROHh4apTp44GDRqk4sWLKzk5WefOnVOnTp10xx13qGbNmurdu7cqV66cq1oy3H6XfID/SUhIUI8ePTR69GjNmDFDHh5X36/+/PPP6tixo4YNG6ZixYrp0KFDmjlzpv71r3/pq6++uuF69+3bp5EjR2rs2LEKDg7We++9p/79+6tq1aq69957sx175coVde7cWf3799fIkSO1bds2vfTSSwoKCtKkSZMkSX/++adat26tP/74QzNnzlTVqlW1fv16devWLUf7vXHjRgUHBxfYFfxx48apadOmeuedd+Th4aHQ0FDdfffdWrx4sfr37+/Ud8mSJSpXrpw6duwoSTpw4ICioqJUuXJlvfbaawoJCdGGDRs0dOhQnT59WpMnT5YkzZo1S1OmTNHEiRN177336sqVKzp06JDOnj2bZV0PPPCAdu7cqaZNm6pr164aOXKk47nDhw8rKipK5cqV05w5c1S6dGn97W9/U58+ffTbb79p9OjR2e5juXLlMm2vfPnyWr9+ve6//371799fTz/9tCQ5Qru71q5dq127dmnatGkqXry440rO4cOHFRER4dbx++2339SyZUt5e3tr3rx5Cg4O1vLlyzVkyBC3agoICNDEiRP1wgsv6KuvvtJ99913wzH9+vVTTEyMNmzYoA4dOigtLU3vv/+++vfv7/gZdCUtLc3p42abzZbpT2Nf77nnntOCBQs0ZMgQderUSceOHdP//d//acuWLfruu++crsgnJibqqaee0siRIzV58mStXr1a48aNU4UKFdSrV68cHA1nGb8ratWq5Wj74YcfVLZsWYWEhDj1rVu3ruP5ayUmJqpv374aPXq0KlWqpLlz56pfv346ceKEVq5cqfHjxysoKEjTpk1Tly5ddOTIEVWoUMHtWlF0+Hl76sC09oWy3fzSsGFDp8dpaWl65ZVXtGLFCp08eVIpKSlKSUlx+cb3Whk/N5IcU2mSkpJyVdPBgwfVtGlTp4sjzZo103//+1/9+uuvqlevntq0aaM6deqoffv2io6OVteuXVWyZEmVKlVKffr0Ufv27dWuXTu1bdtWjz/+uMqXL5+rWhws4BbXu3dvq1ixYk5tLVu2tCRZX375ZbZj09PTrStXrlhbt261JFn79u1zPDd58mTr+h+hsLAwy263W8ePH3e0Xbx40SpVqpT17LPPOto2b95sSbI2b97sVKck66OPPnJaZ8eOHa3q1as7Hr/99tuWJOuLL75w6vfss89akqzFixdnu092u91q0qRJtn2u36fevXtnam/ZsqXVsmXLTPt07733Zuo7Z84cS5J1+PBhR9sff/xh+fr6WiNHjnS0tW/f3qpUqZJ17tw5p/FDhgyx7Ha79ccff1iWZVmdOnWy6tevn+N9uJYka/DgwU5t3bt3t3x9fa34+Hin9g4dOlj+/v7W2bNnb7iPrvz++++WJGvy5MmZnuvdu7cVFhaWqd3VeSXJCg4OtpKTkx1tiYmJloeHhxUTE+Noy+nxGzNmjGWz2ay9e/c69WvXrl2m89KVxYsXW5KsXbt2WSkpKVZERITVsGFDKz093bKsq+dGrVq1nMaEhYVZDzzwgOP5rl27WpZlWWvXrrVsNpt19OhR6+OPP860/Yzjcf1SsWLFbGs8ePCgJckaNGiQU/u3335rSbLGjx/vaMv4ffDtt9869b3rrrus9u3bZ7sdV/bt22f5+flZDz/8sFN7u3btnH6Wr+Xj42M988wzmWravXu3o+3MmTOWp6en5efnZ508edLRvnfvXkuSNWfOHLdrhbkuXrxoHThwwLp48WJhl5IrixcvtoKCghyPjx49akmy4uLinPrNnDnTKl26tLVs2TJr79691s8//2w98MAD1kMPPeTo07JlS+uFF15wPA4LC7PeeOMNp/XUq1fP5e/aDL1793Za57W6dOli9e3b16ktLi7OkuT4/4X09HTr66+/tiZNmmTVqVPHKlu2rHXkyBFH/++++86aMWOG1bRpU6t48eLWzp07s6wlQ3avMVNccNsqWbKkyyt+R44c0ZNPPqmQkBB5enrK29tbLVu2lHT1XfaN1K9f3+mjLbvdrmrVquXoCyM2m00PPvigU1vdunWdxm7dulUBAQG6//77nfo98cQTN1z/zfDoo49manvqqafk6+vrNP3mgw8+UEpKiuPj/kuXLunLL7/Uww8/LH9/f6WmpjqWjh076tKlS/rmm28kXZ0+sG/fPg0aNEgbNmxQcnJynmr+6quv1KZNG4WGhjq19+nTRxcuXMj0ZWBX+1jQWrdu7XQXg+DgYJUrV85xbrhz/DZv3qxatWplmjf+5JNPul2Xj4+Ppk+frt27d+ujjz7K0Zh+/frps88+05kzZ7Rw4UK1bt3a5XSfa23atEm7du1yLOvWrcu2/+bNmyUp0/SsRo0aqWbNmvryyy+d2kNCQhzTUjJc/7OXE8eOHVOnTp0UGhqq9957L9Pz2c1Lvf658uXLKzIy0vG4VKlSKleunOrXr+90pbxmzZqSlOcvpQGFYfv27XrooYfUo0cP1atXTxEREfr5559vag133XWXduzY4fRl2B07diggIEAVK1aUdPXns1mzZpo6dari4uLk4+Pj9H2tu+++W+PGjdOOHTtUu3Zt/f3vf89TTQR03LZcffz03//+Vy1atNC3336r6dOna8uWLdq1a5dWrVolSbp48eIN11u6dOlMbb6+vjka6+/vL7vdnmnstbdrO3PmjIKDgzONddXmSuXKlXX06NEc9c0NV8e1VKlS6ty5s5YuXaq0tDRJV6e3NGrUyDEF4MyZM0pNTdXcuXPl7e3ttGRMgTl9+rSkq1NMXn31VX3zzTfq0KGDSpcurTZt2mj37t25qvnMmTMu684IQWfOnLnhPha0G51X7hy/M2fOZJpmIcllW050795dDRo00IQJE3J027SuXbvKbrfrjTfe0Oeff55p6pMr9erVU8OGDR3LtR9vu5LxmmX1ul7/mubl5zbD8ePH1bp1a3l5eenLL7/M9IXP0qVLZ9qudHXa2uXLlzP1d/WFUR8fn0ztPj4+kuTyto6A6apWrarY2Fjt2LFDBw8e1LPPPqvExMQC2da5c+e0d+9epyU+Pl6DBg3SiRMn9Pzzz+vQoUP6xz/+ocmTJ2vEiBHy8PDQt99+qxkzZmj37t2Kj4/XqlWr9Pvvv6tmzZo6evSoxo0bp507d+r48ePauHGjfvrpJ8cb59xiDjpuW66uZH311Vc6deqUtmzZ4rhqLinbuc03W+nSpfWvf/0rU3tOf6G1b99ec+fO1TfffJOjeeh2uz3Tl1Slq2HP1V01srpC2LdvX3388ceKjY1V5cqVtWvXLs2fP9/xfMmSJeXp6amePXtq8ODBLtdRpUoVSZKXl5dGjBihESNG6OzZs9q0aZPGjx+v9u3b68SJE27djUe6ekwTEhIytZ86dUqSMu1nXr+dL2V/XHPDneNXunRpl+dLbv9P0WazaebMmWrXrl2ObnXm7++v7t27KyYmRoGBgU5fSs4vGYE7ISEh092WTp06le93hDl+/LhatWoly7K0ZcsWl3d4qlOnjj788EMlJiY6vRnav3+/JGW6dzxwO/i///s/HT16VO3bt5e/v7+eeeYZdenSRefOncv3bW3ZskV33323U1vGH09at26dRo0apXr16qlUqVLq37+/4zaygYGB2rZtm2bPnq3k5GSFhYXptddeU4cOHfTbb7/p0KFDev/99x0Xe4YMGaJnn302T7US0IFrZAQvX19fp/Z33323MMpxqWXLlvroo4/0xRdfON0q8MMPP8zR+OHDh2vRokUaNGhQptssSlfvs/vpp5/q4YcflnT1Li7ff/+9U5+ffvpJhw8fdivkREdHq2LFilq8eLEqV64su93uNC3H399frVu3VlxcnOrWreu4KngjJUqUUNeuXXXy5EkNGzZMx44d01133ZXjuiSpTZs2Wr16tU6dOuU0dWDp0qXy9/fP9RdqM84jV1dhw8PDlZSUpN9++83x6cfly5e1YUPubqHmzvFr3bq1Zs2apX379jlNc8nLR7Jt27ZVu3btNG3atExThVx57rnnHF9Wvf5To/yQMX3tb3/7m+655x5H+65du3Tw4EFNmDAh37YVHx+vVq1aKS0tTVu2bFFYWJjLfg899JAmTpyo999/3+nON0uWLJGfn1+maWtAUdanTx+nKWbh4eEu76deqlSpG/51z+vvznLs2LFMffbu3ZvtOpYsWZLtXc5atmzp8uKXdHUaWVZ3SgoODs63WxNfi4AOXCMqKkolS5bUwIEDNXnyZHl7e2v58uXat29fYZfm0Lt3b73xxhvq0aOHpk+frqpVq+qLL75wBLvs7oQhXb2K+uGHH6pbt26qX7++hgwZ4riicODAAS1atEiWZTkCes+ePdWjRw8NGjRIjz76qI4fP65Zs2a5fTcST09P9erVS6+//rrjqun1bw7efPNNNW/eXC1atNBzzz2n8PBwnT9/Xv/+97/1+eefO+6M8eCDD6p27dpq2LChypYtq+PHj2v27NkKCwvTnXfe6VZdkjR58mStWbNGrVu31qRJk1SqVCktX75ca9eu1axZs1zejjInAgICFBYWpn/84x9q06aNSpUqpTJlyig8PFzdunXTpEmT1L17d40aNUqXLl3SnDlzHFOAciOnx2/YsGFatGiRHnjgAU2fPt1xF5dDhw7letuSNHPmTEVGRiopKcnp7iWu1K9fv0D/5Hb16tX1zDPPaO7cufLw8FCHDh0cd3EJDQ3V8OHD82U7SUlJat26tRISErRw4UIlJSU53Uni2r+XUKtWLfXv31+TJ0+Wp6en7rnnHm3cuFELFizQ9OnTuQc6AAfmoAPXKF26tNauXSt/f3/16NFD/fr1U/HixbVixYrCLs2hWLFi+uqrr9SqVSuNHj1ajz76qOLj4zVv3jxJV68o30inTp20f/9+dezYUe+88446duyoTp06af78+WrdurXT1YAnn3xSs2bN0oYNGxx95s+fr2rVqrlde9++fZWSkqLff/89072gpatf1Pnuu+9Uu3ZtTZw4UdHR0erfv79WrlypNm3aOPq1bt1a27Zt08CBA9WuXTtNnDhRbdq00datW+Xt7e12XdWrV9eOHTtUvXp1DR48WF26dNEPP/ygxYsXO/6gTG4tXLhQ/v7+6ty5s+655x5NmTJF0tU3Sv/4xz909uxZde3aVaNGjdJjjz2Wq1v6Zcjp8QsJCdHWrVt111136bnnnlOPHj1kt9v11ltv5Wlf7777bmO+rCxJ8+fP1yuvvKJ169apU6dOmjBhgqKjo7Vjxw6Xc85z48CBAzpy5IhSUlLUo0cPNW3a1Gm5/oui8+bN09ixYzV37lxFR0dr5cqVevPNNzV+/Ph8qQfArcFmufq8AUCRM2PGDE2cOFHx8fG5/gunAAAzXLp0SUePHlWVKlUKZBoYCl92rzFTXIAiKONKZ40aNXTlyhV99dVXmjNnjnr06EE4BwCgiCOgA0WQv7+/3njjDR07dkwpKSmqXLmyxowZ4/jGOQAAKLqY4gIAAGAYprjc+rJ7jfmSKAAAAGAQAjoAAABgEAI6AAAAYBACOgAAAPLMZrNlu1z7l0XdFR4ertmzZ+dbP9NxFxcAAADkWUJCguPfK1as0KRJk3T48GFHm5+fX2GUVSRxBR0AAAB5FhIS4liCgoJks9mc2rZt26bIyEjZ7XZFRERo6tSpSk1NdYyfMmWKKleuLF9fX1WoUEFDhw6VJLVq1UrHjx/X8OHDHVfjc2v+/Pm644475OPjo+rVq2vZsmVOz2dVg3T1LwHfeeedstvtCg4OVteuXXNdx41wBR0AAKAosCzpyoWbv11vfykPoViSNmzYoB49emjOnDlq0aKFfvnlFz3zzDOSpMmTJ2vlypV644039OGHH6pWrVpKTEzUvn37JEmrVq1SvXr19Mwzz2jAgAG5rmH16tV64YUXNHv2bLVt21Zr1qxR3759ValSJbVu3TrbGnbv3q2hQ4dq2bJlioqK0h9//KHt27fn6Zhkh4AOAABQFFy5IM2ocPO3O/6U5FMsT6t4+eWXNXbsWPXu3VuSFBERoZdeekmjR4/W5MmTFR8fr5CQELVt21be3t6qXLmyGjVqJEkqVaqUPD09FRAQoJCQkFzX8Oqrr6pPnz4aNGiQJGnEiBH65ptv9Oqrr6p169bZ1hAfH69ixYqpU6dOCggIUFhYmO6+++48HZPsMMUFAAAABWrPnj2aNm2aihcv7lgGDBighIQEXbhwQY899pguXryoiIgIDRgwQKtXr3aa/pIfDh48qGbNmjm1NWvWTAcPHpSkbGto166dwsLCFBERoZ49e2r58uW6cKHgPs3gCjoAAEBR4O1/9Wp2YWw3j9LT0zV16lQ98sgjmZ6z2+0KDQ3V4cOHFRsbq02bNmnQoEH6y1/+oq1bt8rb2zvP289w/fx1y7IcbdnVEBAQoO+++05btmzRxo0bNWnSJE2ZMkW7du1SiRIl8q2+DAR0AACAosBmy/NUk8LSoEEDHT58WFWrVs2yj5+fnzp37qzOnTtr8ODBqlGjhvbv368GDRrIx8dHaWlpeaqhZs2a+vrrr9WrVy9H244dO1SzZs0c1eDl5aW2bduqbdu2mjx5skqUKKGvvvrK5ZuOvCKgAwAAoEBNmjRJnTp1UmhoqB577DF5eHjo+++/1/79+zV9+nQtWbJEaWlpaty4sfz9/bVs2TL5+fkpLCxM0tX7m2/btk3du3eXr6+vypQpk+W2Tp48qb179zq1Va5cWaNGjdLjjz+uBg0aqE2bNvr888+1atUqbdq0SZKyrWHNmjU6cuSI7r33XpUsWVLr1q1Tenq6qlevXjAHzAIAAIBRLl68aB04cMC6ePFiYZeSK4sXL7aCgoKc2tavX29FRUVZfn5+VmBgoNWoUSNrwYIFlmVZ1urVq63GjRtbgYGBVrFixawmTZpYmzZtcozduXOnVbduXcvX19fKLr6GhYVZkjItixcvtizLsubNm2dFRERY3t7eVrVq1aylS5c6xmZXw/bt262WLVtaJUuWtPz8/Ky6detaK1asyNMxyu41tlmWZRVM9AcAAEBuXLp0SUePHlWVKlVkt9sLuxwUgOxeY+7iAgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAAAYhIAOAAAAGISADgAAABiEgA4AAIA8s9ls2S59+vTJ9brDw8M1e/bsHPXL2J6fn59q1Kihv/zlL7Isy9Hn2LFjstls8vLy0smTJ53GJyQkyMvLSzabTceOHXO0f/LJJ2rcuLGCgoIUEBCgWrVqaeTIkY7nlyxZ4nKf7XZ7rvbXK1ejAAAAgGskJCQ4/r1ixQpNmjRJhw8fdrT5+fndlDqmTZumAQMG6NKlS9q0aZOee+45BQYG6tlnn3XqV6FCBS1dulTjxo1ztL3//vuqWLGi4uPjHW2bNm1S9+7dNWPGDHXu3Fk2m00HDhzQl19+6bS+wMBAp/2Vrr5pyQ2uoAMAACDPQkJCHEtQUJBsNptT27Zt2xQZGSm73a6IiAhNnTpVqampjvFTpkxR5cqV5evrqwoVKmjo0KGSpFatWun48eMaPny448p0dgICAhQSEqLw8HA9/fTTqlu3rjZu3JipX+/evbV48WKntiVLlqh3795ObWvWrFHz5s01atQoVa9eXdWqVVOXLl00d+5cp37X729ISIiCg4PdOoYZCOgAAABFgGVZunDlwk1frp0eklsbNmxQjx49NHToUB04cEDvvvuulixZopdfflmStHLlSr3xxht699139fPPP+vTTz9VnTp1JEmrVq1SpUqVNG3aNCUkJDhdqb/R8dqyZYsOHjwob2/vTM937txZ//nPf/T1119Lkr7++mv98ccfevDBB536hYSE6Mcff9QPP/yQl0PgFqa4AAAAFAEXUy+q8d8b3/Ttfvvkt/L39s/TOl5++WWNHTvWcXU6IiJCL730kkaPHq3JkycrPj5eISEhatu2rby9vVW5cmU1atRIklSqVCl5eno6rozfyJgxYzRx4kRdvnxZV65ckd1ud1yNv5a3t7d69OihRYsWqXnz5lq0aJF69OiRKcw///zz2r59u+rUqaOwsDA1adJE0dHReuqpp+Tr6+vod+7cORUvXtxpbFRUlMur9zdCQAcAAECB2rNnj3bt2uW4Yi5JaWlpunTpki5cuKDHHntMs2fPVkREhO6//3517NhRDz74oLy83I+qo0aNUp8+ffT7779rwoQJuu+++xQVFeWyb//+/dW0aVPNmDFDH3/8sXbu3Ok07UaSihUrprVr1+qXX37R5s2b9c0332jkyJF68803tXPnTvn7X33zEhAQoO+++85pbG7n3RPQAQAAigA/Lz99++S3hbLdvEpPT9fUqVP1yCOPZHrObrcrNDRUhw8fVmxsrDZt2qRBgwbpL3/5i7Zu3epyekp2ypQpo6pVq6pq1ar65JNPVLVqVTVp0kRt27bN1Ld27dqqUaOGnnjiCdWsWVO1a9fW3r17Xa73jjvu0B133KGnn35aEyZMULVq1bRixQr17dtXkuTh4aGqVau6VWtWCOgAAABFgM1my/NUk8LSoEEDHT58ONsA6+fnp86dO6tz584aPHiwatSoof3796tBgwby8fFRWlqa29stWbKknn/+eb344ouKi4tz+QXTfv36adCgQZo/f36O1xseHi5/f3/9+eefbteUEwR0AAAAFKhJkyapU6dOCg0N1WOPPSYPDw99//332r9/v6ZPn64lS5YoLS1NjRs3lr+/v5YtWyY/Pz+FhYVJuhqIt23bpu7du8vX11dlypTJ8bYHDx6smTNn6pNPPlHXrl0zPT9gwAA99thjKlGihMvxU6ZM0YULF9SxY0eFhYXp7NmzmjNnjq5cuaJ27do5+lmWpcTExEzjy5UrJw8P9+7Lwl1cAAAAUKDat2+vNWvWKDY2Vvfcc4+aNGmi119/3RHAS5Qoob/+9a9q1qyZ6tatqy+//FKff/65SpcuLenqvc2PHTumO+64Q2XLlnVr22XLllXPnj01ZcoUpaenZ3rey8tLZcqUyXK+e8uWLXXkyBH16tVLNWrUUIcOHZSYmKiNGzeqevXqjn7JyckqX758piUpKcmteiXJZuXHvXMAAACQby5duqSjR4+qSpUquf5rlDBbdq8xV9ABAAAAgxDQAQAAAIMQ0AEAAACDENABAAAAgxDQAQAADMW9PG5d2b22BHQAAADDZPz1zAsXLhRyJSgoGa+tq7+Uyh8qAgAAMIynp6dKlCjhuIe2v7+/y7+CiaLHsixduHBBSUlJKlGihDw9PTP1KRL3QU9PT9epU6cUEBDAyQkAAG4LlmXpP//5T4H9OfmCkJ6eLk9PT/JaDpQoUUIhISEuj1WRuIJ+6tQphYaGFnYZAAAAN52/v7/KlClTZELvli1bFBISUthlGM3b29vllfMMRSKgBwQESJJOnDihwMDAQq4GAAAA10tOTlZoaKhKlizJXz/NoyIR0DPeMQYGBhLQAQAADFZUrvSbjLu4AAAAAAYhoAMAAAAGIaADAAAABiGgAwAAAAYhoAMAAAAGIaADAAAABiGgAwAAAAYhoAMAAAAGIaADAAAABiGgAwAAAAYhoAMAAAAGIaADAAAABiGgAwAAAAYhoAMAAAAGIaADAAAABiGgAwAAAAYhoAMAAAAG8SrsAkyTbqWr9xe9JUk2m0022Zz+7fhf2XT1v87trtqybLf9b90u+rpqd9nfVY256HttjdePc2f/s9qnrOpxuf//216Wdbqo50b75E7tWb0meX09stru9X2dzr0b1Z/LY3z98fW0ecrLw0teHl7ysPG+HQCAwkRAv066la69v+8t7DKAQuNh85CXzcsR2DMWbw/vq/928Zyjj83b9XO269Zx/Tpv0C+nz13/PG82AABFEQH9Oh42D81uNVtWxn8sx7909b/ObZZlSbqu3VXbte3XtVlybs/q31n1d1ljDurJbn9c7VNW9bjTN6sas9tnl/uURf0Ftv/ZvH4Fsv9u9L/R/rsr3UrXZeuyLqdfdnusaW7mm41MbbzZAADkEgH9Oh42D7UJa1PYZQD56kZhPi09TalWqlLTMy9X0q9k+VxqeqquWFeyfC7V+t/49P//vzfq5+o5p1pctKempyrNSsu037zZyP2bDS+bl7w9vXP8hoI3GwCQfwjowG3g2jntLnne1HIKRLqVrrT0tGzfUOTXm41s33DwZuOGCuPNhpfH/95w2Lxks9mcPv2SrvvU7n+Pnf7XxadmGVx9WpeTsa4+gXM8b8np8fV9Hdu+0dis9svFJ3FZ7pebY7Pdbxf7df023TkmOd3vay9IXPt8tvt1/dgs9svla3T98+6MdXFMsjxm2exXxuMmFZpo1r2zhKKFgA7gluBh85CHp4e8Pb0Lu5Q8u9lvNrJ808GbDaDI++/l/xZ2CcgFAjoAGIY3G3l/s+HqjUPG4wyZ7qZ0zR2SXD5/zV2RMh5nOTarO1O5uBPV9e2Obbu4u1Ruxt5ovzLtSz4ck5zu97XrvLZ2p7HZ7df1z2e3X9c/n91+5XG/M8Zmu18F/HpmPFfMu5hQ9BDQAQAF5lZ6swEANwvf1AEAAAAMQkAHAAAADEJABwAAAAxCQAcAAAAMQkAHAAAADEJABwAAAAxCQAcAAAAMQkAHAAAADEJABwAAAAxCQAcAAAAMQkAHAAAADJKrgD5v3jxVqVJFdrtdkZGR2r59e7b9ly9frnr16snf31/ly5dX3759debMmVwVDAAAANzK3A7oK1as0LBhwzRhwgTFxcWpRYsW6tChg+Lj4132//rrr9WrVy/1799fP/74oz7++GPt2rVLTz/9dJ6LBwAAAG41bgf0119/Xf3799fTTz+tmjVravbs2QoNDdX8+fNd9v/mm28UHh6uoUOHqkqVKmrevLmeffZZ7d69O8/FAwAAALcatwL65cuXtWfPHkVHRzu1R0dHa8eOHS7HREVF6ddff9W6detkWZZ+++03rVy5Ug888ECW20lJSVFycrLTAgAAANwO3Arop0+fVlpamoKDg53ag4ODlZiY6HJMVFSUli9frm7dusnHx0chISEqUaKE5s6dm+V2YmJiFBQU5FhCQ0PdKRMAAAAosnL1JVGbzeb02LKsTG0ZDhw4oKFDh2rSpEnas2eP1q9fr6NHj2rgwIFZrn/cuHE6d+6cYzlx4kRuygQAAACKHC93OpcpU0aenp6ZrpYnJSVluqqeISYmRs2aNdOoUaMkSXXr1lWxYsXUokULTZ8+XeXLl880xtfXV76+vu6UBgAAANwS3LqC7uPjo8jISMXGxjq1x8bGKioqyuWYCxcuyMPDeTOenp6Srl55BwAAAPD/uT3FZcSIEXrvvfe0aNEiHTx4UMOHD1d8fLxjysq4cePUq1cvR/8HH3xQq1at0vz583XkyBH985//1NChQ9WoUSNVqFAh//YEAAAAuAW4NcVFkrp166YzZ85o2rRpSkhIUO3atbVu3TqFhYVJkhISEpzuid6nTx+dP39eb731lkaOHKkSJUrovvvu08yZM/NvLwAAAIBbhM0qAvNMkpOTFRQUpHPnzikwMLCwywEAAMB1yGv5J1d3cQEAAABQMAjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEEI6AAAAIBBCOgAAACAQQjoAAAAgEFyFdDnzZunKlWqyG63KzIyUtu3b8+2f0pKiiZMmKCwsDD5+vrqjjvu0KJFi3JVMAAAAHAr83J3wIoVKzRs2DDNmzdPzZo107vvvqsOHTrowIEDqly5sssxjz/+uH777TctXLhQVatWVVJSklJTU/NcPAAAAHCrsVmWZbkzoHHjxmrQoIHmz5/vaKtZs6a6dOmimJiYTP3Xr1+v7t2768iRIypVqlSOtpGSkqKUlBTH4+TkZIWGhurcuXMKDAx0p1wAAADcBMnJyQoKCiKv5QO3prhcvnxZe/bsUXR0tFN7dHS0duzY4XLMZ599poYNG2rWrFmqWLGiqlWrphdffFEXL17McjsxMTEKCgpyLKGhoe6UCQAAABRZbk1xOX36tNLS0hQcHOzUHhwcrMTERJdjjhw5oq+//lp2u12rV6/W6dOnNWjQIP3xxx9ZzkMfN26cRowY4XiccQUdAAAAuNW5PQddkmw2m9Njy7IytWVIT0+XzWbT8uXLFRQUJEl6/fXX1bVrV7399tvy8/PLNMbX11e+vr65KQ0AAAAo0tya4lKmTBl5enpmulqelJSU6ap6hvLly6tixYqOcC5dnbNuWZZ+/fXXXJQMAAAA3LrcCug+Pj6KjIxUbGysU3tsbKyioqJcjmnWrJlOnTql//73v462n376SR4eHqpUqVIuSgYAAABuXW7fB33EiBF67733tGjRIh08eFDDhw9XfHy8Bg4cKOnq/PFevXo5+j/55JMqXbq0+vbtqwMHDmjbtm0aNWqU+vXr53J6CwAAAHA7c3sOerdu3XTmzBlNmzZNCQkJql27ttatW6ewsDBJUkJCguLj4x39ixcvrtjYWD3//PNq2LChSpcurccff1zTp0/Pv70AAAAAbhFu3we9MHBfTQAAALOR1/KP21NcAAAAABQcAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGCQXAX0efPmqUqVKrLb7YqMjNT27dtzNO6f//ynvLy8VL9+/dxsFgAAALjluR3QV6xYoWHDhmnChAmKi4tTixYt1KFDB8XHx2c77ty5c+rVq5fatGmT62IBAACAW53NsizLnQGNGzdWgwYNNH/+fEdbzZo11aVLF8XExGQ5rnv37rrzzjvl6empTz/9VHv37s3xNpOTkxUUFKRz584pMDDQnXIBAABwE5DX8o9bV9AvX76sPXv2KDo62qk9OjpaO3bsyHLc4sWL9csvv2jy5Mk52k5KSoqSk5OdFgAAAOB24FZAP336tNLS0hQcHOzUHhwcrMTERJdjfv75Z40dO1bLly+Xl5dXjrYTExOjoKAgxxIaGupOmQAAAECRlasvidpsNqfHlmVlapOktLQ0Pfnkk5o6daqqVauW4/WPGzdO586dcywnTpzITZkAAABAkZOzS9r/U6ZMGXl6ema6Wp6UlJTpqroknT9/Xrt371ZcXJyGDBkiSUpPT5dlWfLy8tLGjRt13333ZRrn6+srX19fd0oDAAAAbgluXUH38fFRZGSkYmNjndpjY2MVFRWVqX9gYKD279+vvXv3OpaBAweqevXq2rt3rxo3bpy36gEAAIBbjFtX0CVpxIgR6tmzpxo2bKimTZtqwYIFio+P18CBAyVdnZ5y8uRJLV26VB4eHqpdu7bT+HLlyslut2dqBwAAAJCLgN6tWzedOXNG06ZNU0JCgmrXrq1169YpLCxMkpSQkHDDe6IDAAAAcM3t+6AXBu6rCQAAYDbyWv7J1V1cAAAAABQMAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgkFwF9Hnz5qlKlSqy2+2KjIzU9u3bs+y7atUqtWvXTmXLllVgYKCaNm2qDRs25LpgAAAA4FbmdkBfsWKFhg0bpgkTJiguLk4tWrRQhw4dFB8f77L/tm3b1K5dO61bt0579uxR69at9eCDDyouLi7PxQMAAAC3GptlWZY7Axo3bqwGDRpo/vz5jraaNWuqS5cuiomJydE6atWqpW7dumnSpEk56p+cnKygoCCdO3dOgYGB7pQLAACAm4C8ln/cuoJ++fJl7dmzR9HR0U7t0dHR2rFjR47WkZ6ervPnz6tUqVJZ9klJSVFycrLTAgAAANwO3Arop0+fVlpamoKDg53ag4ODlZiYmKN1vPbaa/rzzz/1+OOPZ9knJiZGQUFBjiU0NNSdMgEAAIAiK1dfErXZbE6PLcvK1ObKBx98oClTpmjFihUqV65clv3GjRunc+fOOZYTJ07kpkwAAACgyPFyp3OZMmXk6emZ6Wp5UlJSpqvq11uxYoX69++vjz/+WG3bts22r6+vr3x9fd0pDQAAALgluHUF3cfHR5GRkYqNjXVqj42NVVRUVJbjPvjgA/Xp00d///vf9cADD+SuUgAAAOA24NYVdEkaMWKEevbsqYYNG6pp06ZasGCB4uPjNXDgQElXp6ecPHlSS5culXQ1nPfq1UtvvvmmmjRp4rj67ufnp6CgoHzcFQAAAKDoczugd+vWTWfOnNG0adOUkJCg2rVra926dQoLC5MkJSQkON0T/d1331VqaqoGDx6swYMHO9p79+6tJUuW5H0PAAAAgFuI2/dBLwzcVxMAAMBs5LX8k6u7uAAAAAAoGAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCBehV2AcdLTpFer5W0dNlsei8jj+DxtvzC3XdjbL+r7nk+DnfbD5HZl0W5anTlpVxbtptWZH/UbUHu+swpglQWwTurM75UWwCoLYJ3l60rNh+f/elGgCOiuXDhd2BUAAADk3eX/EtCLIAL69Wwe0qBvcjc2T+98czmWbd5i28zDJvNlP60i2K4s2k2rMyftyqLdtDpzuV9G1GIpX6+m5/mTM6eV5eOqqMu9VeXXugysKSg0f9aDm4qAfj2bTSpXs7CrAAAAwG2KL4kCAAAABiGgAwAAAAYhoAMAAAAGIaADAAAABiGgAwAAAAYhoAMAAAAGIaADAAAABiGgAwAAAAbJVUCfN2+eqlSpIrvdrsjISG3fvj3b/lu3blVkZKTsdrsiIiL0zjvv5KpYAAAA4FbndkBfsWKFhg0bpgkTJiguLk4tWrRQhw4dFB8f77L/0aNH1bFjR7Vo0UJxcXEaP368hg4dqk8++STPxQMAAAC3GptlWZY7Axo3bqwGDRpo/vz5jraaNWuqS5cuiomJydR/zJgx+uyzz3Tw4EFH28CBA7Vv3z7t3LnT5TZSUlKUkpLieJycnKzQ0FCdO3dOgYGB7pTrtrR0S93edV0XAABAUdIgrKTGd6x5U7aVnJysoKCgm5LXbnVe7nS+fPmy9uzZo7Fjxzq1R0dHa8eOHS7H7Ny5U9HR0U5t7du318KFC3XlyhV5e3tnGhMTE6OpU6e6U1q+2n38P4W2bQAAgPwSYHcr6sEQbr1qp0+fVlpamoKDg53ag4ODlZiY6HJMYmKiy/6pqak6ffq0ypcvn2nMuHHjNGLECMfjjCvoN4OHTXqnR+RN2RZgBrc+RAMAFCFlA3wLuwTkQq7eVtlsNqfHlmVlartRf1ftGXx9feXrWzgnlM1m0/21Qwpl2wAAAIBbXxItU6aMPD09M10tT0pKynSVPENISIjL/l5eXipdurSb5QIAAAC3NrcCuo+PjyIjIxUbG+vUHhsbq6ioKJdjmjZtmqn/xo0b1bBhQ5fzzwEAAIDbmdu3WRwxYoTee+89LVq0SAcPHtTw4cMVHx+vgQMHSro6f7xXr16O/gMHDtTx48c1YsQIHTx4UIsWLdLChQv14osv5t9eAAAAALcIt+egd+vWTWfOnNG0adOUkJCg2rVra926dQoLC5MkJSQkON0TvUqVKlq3bp2GDx+ut99+WxUqVNCcOXP06KOP5t9eAAAAALcIt++DXhi4ryYAAIDZyGv5x+0pLgAAAAAKDgEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADAIAR0AAAAwCAEdAAAAMAgBHQAAADCIV2EXkBOWZUmSkpOTC7kSAAAAuJKR0zJyG3KvSAT08+fPS5JCQ0MLuRIAAABk5/z58woKCirsMoo0m1UE3uakp6fr1KlTCggIkM1mK/DtJScnKzQ0VCdOnFBgYGCBbw9XcdwLB8e9cHDcCwfHvXBw3AvHzT7ulmXp/PnzqlChgjw8mEWdF0XiCrqHh4cqVap007cbGBjIL5JCwHEvHBz3wsFxLxwc98LBcS8cN/O4c+U8f/D2BgAAADAIAR0AAAAwCAHdBV9fX02ePFm+vr6FXcptheNeODjuhYPjXjg47oWD4144OO5FV5H4kigAAABwu+AKOgAAAGAQAjoAAABgEAI6AAAAYBACOgAAAGAQAjoAAABgkNsyoG/btk0PPvigKlSoIJvNpk8//fSGY7Zu3arIyEjZ7XZFRETonXfeKfhCbyHuHvMtW7bIZrNlWg4dOnRzCr5FxMTE6J577lFAQIDKlSunLl266PDhwzccx/meN7k57pzzeTd//nzVrVvX8VcTmzZtqi+++CLbMZzreefucedcz38xMTGy2WwaNmxYtv0434uO2zKg//nnn6pXr57eeuutHPU/evSoOnbsqBYtWiguLk7jx4/X0KFD9cknnxRwpbcOd495hsOHDyshIcGx3HnnnQVU4a1p69atGjx4sL755hvFxsYqNTVV0dHR+vPPP7Mcw/med7k57hk453OvUqVKeuWVV7R7927t3r1b9913nx566CH9+OOPLvtzrucPd497Bs71/LFr1y4tWLBAdevWzbYf53sRY93mJFmrV6/Ots/o0aOtGjVqOLU9++yzVpMmTQqwsltXTo755s2bLUnWf/7zn5tS0+0iKSnJkmRt3bo1yz6c7/kvJ8edc75glCxZ0nrvvfdcPse5XnCyO+6c6/nn/Pnz1p133mnFxsZaLVu2tF544YUs+3K+Fy235RV0d+3cuVPR0dFObe3bt9fu3bt15cqVQqrq9nD33XerfPnyatOmjTZv3lzY5RR5586dkySVKlUqyz6c7/kvJ8c9A+d8/khLS9OHH36oP//8U02bNnXZh3M9/+XkuGfgXM+7wYMH64EHHlDbtm1v2JfzvWjxKuwCioLExEQFBwc7tQUHBys1NVWnT59W+fLlC6myW1f58uW1YMECRUZGKiUlRcuWLVObNm20ZcsW3XvvvYVdXpFkWZZGjBih5s2bq3bt2ln243zPXzk97pzz+WP//v1q2rSpLl26pOLFi2v16tW66667XPblXM8/7hx3zvX88eGHH+q7777Trl27ctSf871oIaDnkM1mc3psWZbLduSP6tWrq3r16o7HTZs21YkTJ/Tqq6/yCzyXhgwZou+//15ff/31DftyvuefnB53zvn8Ub16de3du1dnz57VJ598ot69e2vr1q1ZhkXO9fzhznHnXM+7EydO6IUXXtDGjRtlt9tzPI7zvehgiksOhISEKDEx0aktKSlJXl5eKl26dCFVdftp0qSJfv7558Iuo0h6/vnn9dlnn2nz5s2qVKlStn053/OPO8fdFc559/n4+Khq1apq2LChYmJiVK9ePb355psu+3Ku5x93jrsrnOvu2bNnj5KSkhQZGSkvLy95eXlp69atmjNnjry8vJSWlpZpDOd70cIV9Bxo2rSpPv/8c6e2jRs3qmHDhvL29i6kqm4/cXFxfATnJsuy9Pzzz2v16tXasmWLqlSpcsMxnO95l5vj7grnfN5ZlqWUlBSXz3GuF5zsjrsrnOvuadOmjfbv3+/U1rdvX9WoUUNjxoyRp6dnpjGc70VMYX07tTCdP3/eiouLs+Li4ixJ1uuvv27FxcVZx48ftyzLssaOHWv17NnT0f/IkSOWv7+/NXz4cOvAgQPWwoULLW9vb2vlypWFtQtFjrvH/I033rBWr15t/fTTT9YPP/xgjR071pJkffLJJ4W1C0XSc889ZwUFBVlbtmyxEhISHMuFCxccfTjf819ujjvnfN6NGzfO2rZtm3X06FHr+++/t8aPH295eHhYGzdutCyLc72guHvcOdcLxvV3ceF8L9puy4CecYun65fevXtblmVZvXv3tlq2bOk0ZsuWLdbdd99t+fj4WOHh4db8+fNvfuFFmLvHfObMmdYdd9xh2e12q2TJklbz5s2ttWvXFk7xRZirYy7JWrx4saMP53v+y81x55zPu379+llhYWGWj4+PVbZsWatNmzaOkGhZnOsFxd3jzrleMK4P6JzvRZvNsv73DQEAAAAAhY4viQIAAAAGIaADAAAABiGgAwAAAAYhoAMAAAAGIaADAAAABiGgAwAAAAYhoAMAAAAGIaADAAAABiGgAwAAAAYhoAMAAAAGIaADAAAABvl/h3MIgK+T3sMAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 800x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# NMF Hyperparameters\n",
    "learning_rate = 0.001\n",
    "weight_decay = 0.0000001\n",
    "epochs = 4\n",
    "batch_size = 1024\n",
    "latent_dims = 32\n",
    "alpha = 0.5\n",
    "loss = nn.MSELoss()\n",
    "\n",
    "print('PRETRAINING GMF')\n",
    "pretrain_gmf, _, __, ___ = train_gmf(train_dataloader,\n",
    "                   test_dataloader,\n",
    "                   num_users,\n",
    "                   num_items,\n",
    "                   2, # Just pretrain a few epochs instead of the full amount\n",
    "                   latent_dims,\n",
    "                   learning_rate,\n",
    "                   'adam',\n",
    "                   criterion=loss,\n",
    "                   device=device,\n",
    "                   weight_decay=weight_decay,\n",
    "                   rescale_data=True)\n",
    "# Can reuse final MLP since the expected latent dims happen to match\n",
    "\n",
    "print('TRAINING NMF')\n",
    "train_model = train_neural_mf\n",
    "final_nmf_model, train_loss, test_loss, test_rmse = train_model(train_dataloader,\n",
    "                                                      test_dataloader,\n",
    "                                                      pretrain_gmf,\n",
    "                                                      final_mlp_model,\n",
    "                                                      num_users,\n",
    "                                                      num_items,\n",
    "                                                      epochs,\n",
    "                                                      latent_dims,\n",
    "                                                      learning_rate,\n",
    "                                                      'adam',\n",
    "                                                      criterion=loss,\n",
    "                                                      device=device,\n",
    "                                                      alpha=alpha,\n",
    "                                                      weight_decay=weight_decay,\n",
    "                                                      rescale_data=True)\n",
    "\n",
    "\n",
    "model_name = f\"models/final_nmf_model_{dataset}\"\n",
    "torch.save(final_nmf_model.state_dict(), model_name)\n",
    "\n",
    "curve_data = {\n",
    "    'Train Loss': train_loss,\n",
    "    'Test Loss': test_loss,\n",
    "    'Test RMSE': test_rmse,\n",
    "}\n",
    "title = f\"Training Curves for tuned NMF on {dataset}\"\n",
    "plot_learning_curve(np.arange(1, epochs+1), curve_data, title, True)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ecc4e557-d35b-45d0-9999-30cd6278fd89",
   "metadata": {},
   "source": [
    "# Evaluation of Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "eb2a9101-60e5-4417-bd42-0b00e6d67a91",
   "metadata": {},
   "outputs": [],
   "source": [
    "from eval import hit_rate, precision_at_k\n",
    "from train import train_gmf, train_mlp, train_neural_mf\n",
    "from ipywidgets import FloatProgress"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "d9691555-cb31-46c1-8cb7-fe825100d935",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "dcdaae53e27c44e48890c36ce3611421",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/943 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "099ae51c56234f58990a0cf125d41416",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/943 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tuned GMF Hit Rate (k=20)\t 0.7285259809119831\n",
      "Tuned GMF Precision at k=20\t 0.05503711558854719\n"
     ]
    }
   ],
   "source": [
    "final_gmf_model = GMF(num_users, num_items, 4)\n",
    "final_gmf_model.load_state_dict(torch.load(f\"models/final_gmf_model_{dataset}\"), strict=False)\n",
    "final_gmf_model = final_gmf_model.to(device)\n",
    "\n",
    "# user_ratio = 0.001 # set to 1 to compute this exactly\n",
    "user_ratio = 1 # set to 1 to compute this exactly\n",
    "eval_users = int(user_ratio * num_users)\n",
    "\n",
    "hr, users_to_ratings = hit_rate(final_gmf_model, 40, eval_users, num_items, data, device, 10_000)\n",
    "pk, users_to_ratings = precision_at_k(final_gmf_model, 40, eval_users, num_items, data, device, 10_000)\n",
    "\n",
    "print(\"Tuned GMF Hit Rate (k=20)\\t\", hr)\n",
    "print(\"Tuned GMF Precision at k=20\\t\", pk)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "4c629f43",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "bfe21c622139462d9ddf14ec2e584988",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/943 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4faa7f94660c40298030e7d2bbc6eb28",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/943 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tuned MLP Hit Rate (k=20)\t 0.5259809119830329\n",
      "Tuned MLP Precision at k=20\t 0.034915164369034996\n"
     ]
    }
   ],
   "source": [
    "final_mlp_model = NCF_MLP(num_users, num_items, 16)\n",
    "final_mlp_model.load_state_dict(torch.load(f\"models/final_mlp_model_{dataset}\"), strict=False)\n",
    "final_mlp_model = final_mlp_model.to(device)\n",
    "\n",
    "# user_ratio = 0.001 # set to 1 to compute this exactly\n",
    "user_ratio = 1 # set to 1 to compute this exactly\n",
    "eval_users = int(user_ratio * num_users)\n",
    "\n",
    "hr, users_to_ratings = hit_rate(final_mlp_model, 40, eval_users, num_items, data, device, 10_000)\n",
    "pk, users_to_ratings = precision_at_k(final_mlp_model, 40, eval_users, num_items, data, device, 10_000)\n",
    "\n",
    "print(\"Tuned MLP Hit Rate (k=20)\\t\", hr)\n",
    "print(\"Tuned MLP Precision at k=20\\t\", pk)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "1c2e2ac8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b7812715b06b428298159f7892667adf",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/943 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a40d9838291d4df28256c0ac97ae3cd3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/943 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tuned GMF Hit Rate (k=20)\t\t 0.6585365853658537\n",
      "Tuned GMF Precision at k=20\t 0.05312831389183457\n"
     ]
    }
   ],
   "source": [
    "final_nmf_model = NEURAL_MF(num_users, num_items, 16)\n",
    "final_nmf_model.load_state_dict(torch.load(f\"models/final_nmf_model_{dataset}\"), strict=False)\n",
    "final_nmf_model = final_nmf_model.to(device)\n",
    "\n",
    "# user_ratio = 0.001 # set to 1 to compute this exactly\n",
    "user_ratio = 1 # set to 1 to compute this exactly\n",
    "eval_users = int(user_ratio * num_users)\n",
    "\n",
    "hr, users_to_ratings = hit_rate(final_nmf_model, 40, eval_users, num_items, data, device, 10_000)\n",
    "pk, users_to_ratings = precision_at_k(final_nmf_model, 40, eval_users, num_items, data, device, 10_000)\n",
    "\n",
    "print(\"Tuned GMF Hit Rate (k=20)\\t\\t\", hr)\n",
    "print(\"Tuned GMF Precision at k=20\\t\", pk)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3e2a782e",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.19"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}