Patient_Survival_Predictor_(using_Pytorch).ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/DataMinati/Super-Notebooks-100/blob/main/Patient_Survival_Predictor_(using_Pytorch).ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "ee7eb627",
      "metadata": {
        "papermill": {
          "duration": 0.021409,
          "end_time": "2022-01-27T11:15:45.960275",
          "exception": false,
          "start_time": "2022-01-27T11:15:45.938866",
          "status": "completed"
        },
        "tags": [],
        "id": "ee7eb627"
      },
      "source": [
        "# **Dependency**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "id": "67342560",
      "metadata": {
        "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
        "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5",
        "execution": {
          "iopub.execute_input": "2022-01-27T11:15:46.010000Z",
          "iopub.status.busy": "2022-01-27T11:15:46.008679Z",
          "iopub.status.idle": "2022-01-27T11:15:47.043294Z",
          "shell.execute_reply": "2022-01-27T11:15:47.042700Z",
          "shell.execute_reply.started": "2022-01-27T11:00:47.541216Z"
        },
        "papermill": {
          "duration": 1.060057,
          "end_time": "2022-01-27T11:15:47.043488",
          "exception": false,
          "start_time": "2022-01-27T11:15:45.983431",
          "status": "completed"
        },
        "tags": [],
        "id": "67342560"
      },
      "outputs": [],
      "source": [
        "import os\n",
        "import numpy as np # linear algebra\n",
        "import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "from sklearn.model_selection import train_test_split\n",
        "import matplotlib.pyplot as plt\n"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "df2ae2cf",
      "metadata": {
        "papermill": {
          "duration": 0.019355,
          "end_time": "2022-01-27T11:15:47.083235",
          "exception": false,
          "start_time": "2022-01-27T11:15:47.063880",
          "status": "completed"
        },
        "tags": [],
        "id": "df2ae2cf"
      },
      "source": [
        "# **Code Block - Preprocessing**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "id": "924a2eb8",
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-01-27T11:15:47.142231Z",
          "iopub.status.busy": "2022-01-27T11:15:47.141245Z",
          "iopub.status.idle": "2022-01-27T11:15:47.144033Z",
          "shell.execute_reply": "2022-01-27T11:15:47.143595Z",
          "shell.execute_reply.started": "2022-01-27T11:00:48.559141Z"
        },
        "papermill": {
          "duration": 0.041598,
          "end_time": "2022-01-27T11:15:47.144173",
          "exception": false,
          "start_time": "2022-01-27T11:15:47.102575",
          "status": "completed"
        },
        "tags": [],
        "id": "924a2eb8"
      },
      "outputs": [],
      "source": [
        "# solve_missing {'drop' , 'mean', 'median'}\n",
        "def clean_data(dataframe, \n",
        "               solve_missing : str ='drop', \n",
        "               solve_missing_cat : str = 'drop',\n",
        "               drop_list : list = [], \n",
        "               category_list : list = []): \n",
        "    df = dataframe.drop_duplicates()\n",
        "    df = df.drop(drop_list, axis=1)\n",
        "    if solve_missing == 'drop' or solve_missing_cat == 'drop':\n",
        "        df = df.dropna()\n",
        "        object_dtype = list(df.select_dtypes(include='object').columns)\n",
        "        for col in df.columns:\n",
        "            if col in category_list or col in object_dtype:\n",
        "                df[col] = df[col].astype('category')\n",
        "    else:\n",
        "        object_dtype = list(df.select_dtypes(include='object').columns)\n",
        "        for col in df.columns:\n",
        "            if col in category_list or col in object_dtype:\n",
        "                df[col] = df[col].astype('category')\n",
        "                if solve_missing_cat == 'mode':\n",
        "                    df[col] = df[col].fillna(df[col].mode()[0], inplace=False)\n",
        "            else:\n",
        "                if solve_missing == 'mean':\n",
        "                    df[col] = df[col].fillna(df[col].mean(), inplace=False)\n",
        "                else:\n",
        "                    df[col] = df[col].fillna(df[col].median(), inplace=False)\n",
        "    return df\n",
        "\n",
        "def split_xy(dataframe, label : str):\n",
        "    return dataframe.drop(labels=label, axis=1), dataframe[label]\n",
        "\n",
        "def encode(df):\n",
        "    category_dtype = list(df.select_dtypes(include='category').columns)\n",
        "    cat = pd.get_dummies(df, columns = category_dtype, drop_first = True)\n",
        "    for col in cat.columns:\n",
        "        if cat[col].dtype == np.uint8:\n",
        "            cat[col] = cat[col].astype('category')\n",
        "    return cat\n",
        "\n",
        "def scale(features : tuple):\n",
        "    trainFeatures, valFeatures, testFeatures = features # type pandas DataFrame\n",
        "    scaler = StandardScaler()\n",
        "    category_dtype = list(trainFeatures.select_dtypes(include='category').columns)\n",
        "    continuous_dtype = list(filter(lambda c: c not in category_dtype, trainFeatures.columns))\n",
        "    # SCALING\n",
        "    scaler.fit(trainFeatures[continuous_dtype])\n",
        "    cont_xtrain = scaler.transform(trainFeatures[continuous_dtype])\n",
        "    cont_xval = scaler.transform(valFeatures[continuous_dtype])\n",
        "    cont_xtest = scaler.transform(testFeatures[continuous_dtype])\n",
        "    # ENCODING\n",
        "    cat_xtrain = trainFeatures[category_dtype]\n",
        "    cat_xval = valFeatures[category_dtype]\n",
        "    cat_xtest = testFeatures[category_dtype]\n",
        "    print(cat_xtrain.shape, cat_xval.shape, cat_xtrain.shape)\n",
        "    xtrain = np.concatenate((cont_xtrain, cat_xtrain), axis=1)\n",
        "    xval = np.concatenate((cont_xval, cat_xval), axis=1)\n",
        "    xtest = np.concatenate((cont_xtest, cat_xtest), axis=1)\n",
        "    return scaler, xtrain, xval, xtest"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "04357cc7",
      "metadata": {
        "papermill": {
          "duration": 0.021563,
          "end_time": "2022-01-27T11:15:47.188202",
          "exception": false,
          "start_time": "2022-01-27T11:15:47.166639",
          "status": "completed"
        },
        "tags": [],
        "id": "04357cc7"
      },
      "source": [
        "# **Data Loading**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "id": "62d15f10",
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-01-27T11:15:47.235992Z",
          "iopub.status.busy": "2022-01-27T11:15:47.235244Z",
          "iopub.status.idle": "2022-01-27T11:15:48.611868Z",
          "shell.execute_reply": "2022-01-27T11:15:48.610814Z",
          "shell.execute_reply.started": "2022-01-27T11:00:48.581083Z"
        },
        "papermill": {
          "duration": 1.403214,
          "end_time": "2022-01-27T11:15:48.612018",
          "exception": false,
          "start_time": "2022-01-27T11:15:47.208804",
          "status": "completed"
        },
        "tags": [],
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 466
        },
        "id": "62d15f10",
        "outputId": "a8b6d8be-cd0f-412c-ab67-760f339682fa"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x576 with 1 Axes>"
            ]
          },
          "metadata": {}
        }
      ],
      "source": [
        "df = pd.read_csv('https://raw.githubusercontent.com/MainakRepositor/Datasets-/master/patient_survival.csv')\n",
        "TARGET = 'hospital_death'\n",
        "\n",
        "#df.info()\n",
        "plt.figure(figsize=(10,8))\n",
        "unique_labels = np.unique(df[TARGET])\n",
        "plt.pie(np.array([len(df[df[TARGET]==label]) for label in unique_labels ]), labels = list(unique_labels), autopct='%1.1f%%')\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "id": "32607f1c",
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-01-27T11:15:48.658311Z",
          "iopub.status.busy": "2022-01-27T11:15:48.657397Z",
          "iopub.status.idle": "2022-01-27T11:15:49.520933Z",
          "shell.execute_reply": "2022-01-27T11:15:49.520376Z",
          "shell.execute_reply.started": "2022-01-27T11:00:49.997285Z"
        },
        "papermill": {
          "duration": 0.887737,
          "end_time": "2022-01-27T11:15:49.521077",
          "exception": false,
          "start_time": "2022-01-27T11:15:48.633340",
          "status": "completed"
        },
        "tags": [],
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 496
        },
        "id": "32607f1c",
        "outputId": "a58f8884-72ec-4d12-ffec-671f2b74bd7d"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "category_list = ['elective_surgery', 'ethnicity', 'icu_admit_source', 'icu_stay_type', 'icu_type', 'apache_3j_bodysystem', 'apache_2_bodysystem', \n",
        "                 'aids', 'cirrhosis', 'diabetes_mellitus', 'hepatic_failure', 'immunosuppression', 'leukemia', 'lymphoma', 'solid_tumor_with_metastasis', 'hospital_death']\n",
        "drop_list = ['encounter_id', 'patient_id', 'hospital_id', 'Unnamed: 83']\n",
        "\n",
        "clean_df = clean_data(df, 'mean', 'mode', drop_list, category_list)\n",
        "plt.pie(np.array([len(clean_df[clean_df[TARGET]==label]) for label in unique_labels ]), labels = list(unique_labels), autopct='%1.1f%%')\n",
        "plt.show()\n",
        "plt.bar(list(unique_labels), [len(clean_df[clean_df[TARGET]==label]) for label in unique_labels ], color = ['cyan', 'green'])\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "1f538980",
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-01-27T11:15:49.606868Z",
          "iopub.status.busy": "2022-01-27T11:15:49.605046Z",
          "iopub.status.idle": "2022-01-27T11:15:51.385334Z",
          "shell.execute_reply": "2022-01-27T11:15:51.384797Z",
          "shell.execute_reply.started": "2022-01-27T11:00:51.054375Z"
        },
        "papermill": {
          "duration": 1.84308,
          "end_time": "2022-01-27T11:15:51.385510",
          "exception": false,
          "start_time": "2022-01-27T11:15:49.542430",
          "status": "completed"
        },
        "tags": [],
        "id": "1f538980",
        "outputId": "3250a225-750a-4d63-ba93-8aa52cbd6239"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Encoded Features Shape : (91713, 111)\n",
            "(45856, 47) (18343, 47) (45856, 47)\n"
          ]
        }
      ],
      "source": [
        "features, labels = split_xy(clean_df, TARGET)\n",
        "VAL_SIZE = 0.2\n",
        "TEST_SIZE = 0.3\n",
        "encoded_features = encode(features)\n",
        "print('Encoded Features Shape : {}'.format(encoded_features.shape))\n",
        "# split all samples to train and test with respect to the ratio of each class\n",
        "X_train, X_test, ytrain, ytest = train_test_split(encoded_features, labels, test_size = (VAL_SIZE+TEST_SIZE), stratify = labels)\n",
        "# split test sample to test and validation with respect to the ratio of each class\n",
        "X_test, X_val, ytest, yval = train_test_split(X_test, ytest, test_size = VAL_SIZE/(VAL_SIZE+TEST_SIZE), stratify = ytest)\n",
        "# scale continous variables with standard scaler\n",
        "scaler, xtrain, xval, xtest = scale((X_train, X_val, X_test))\n",
        "#print(scaler.mean_)\n",
        "#print(scaler.var_)\n",
        "#print(xtrain.shape, ytrain.shape)\n",
        "#print(xval.shape, yval.shape)\n",
        "#print(xtest.shape, ytest.shape)\n",
        "# convert features to numpy array format\n",
        "xtrain = np.array(xtrain, dtype='float32')\n",
        "xval = np.array(xval, dtype='float32')\n",
        "xtest = np.array(xtest, dtype='float32')\n",
        "ytrain = np.array(ytrain, dtype='float32')\n",
        "yval = np.array(yval, dtype='float32')\n",
        "ytest = np.array(ytest, dtype='float32')"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "20caefd8",
      "metadata": {
        "papermill": {
          "duration": 0.023244,
          "end_time": "2022-01-27T11:15:51.431346",
          "exception": false,
          "start_time": "2022-01-27T11:15:51.408102",
          "status": "completed"
        },
        "tags": [],
        "id": "20caefd8"
      },
      "source": [
        "# **Model**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "2ebed76d",
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-01-27T11:15:51.482898Z",
          "iopub.status.busy": "2022-01-27T11:15:51.482195Z",
          "iopub.status.idle": "2022-01-27T11:15:52.572118Z",
          "shell.execute_reply": "2022-01-27T11:15:52.571561Z",
          "shell.execute_reply.started": "2022-01-27T11:00:52.946783Z"
        },
        "papermill": {
          "duration": 1.116493,
          "end_time": "2022-01-27T11:15:52.572274",
          "exception": false,
          "start_time": "2022-01-27T11:15:51.455781",
          "status": "completed"
        },
        "tags": [],
        "id": "2ebed76d"
      },
      "outputs": [],
      "source": [
        "import torch\n",
        "# locate device\n",
        "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "428aa385",
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-01-27T11:15:52.627207Z",
          "iopub.status.busy": "2022-01-27T11:15:52.626610Z",
          "iopub.status.idle": "2022-01-27T11:15:52.628833Z",
          "shell.execute_reply": "2022-01-27T11:15:52.629363Z",
          "shell.execute_reply.started": "2022-01-27T11:00:54.106539Z"
        },
        "papermill": {
          "duration": 0.034467,
          "end_time": "2022-01-27T11:15:52.629529",
          "exception": false,
          "start_time": "2022-01-27T11:15:52.595062",
          "status": "completed"
        },
        "tags": [],
        "id": "428aa385"
      },
      "outputs": [],
      "source": [
        "import torch.nn.functional as F\n",
        "import torch.nn as nn\n",
        "#from torch.autograd import Variable\n",
        "\n",
        "class Model(nn.Module):\n",
        "    def __init__(self, input_dim, output_dim):\n",
        "        super(Model, self).__init__()\n",
        "        self.layer1 = nn.Linear(input_dim, 1024)\n",
        "        self.gnorm1 = nn.GroupNorm(32, 1024)\n",
        "        \n",
        "        self.layer2 = nn.Linear(1024, 512)\n",
        "        self.gnorm2 = nn.GroupNorm(16, 512)\n",
        "        \n",
        "        self.layer3 = nn.Linear(512, 128)\n",
        "        self.gnorm3 = nn.GroupNorm(4, 128)\n",
        "        self.layer4 = nn.Linear(128, output_dim)\n",
        "        \n",
        "    def forward(self, x):\n",
        "        x = F.relu(self.gnorm1(self.layer1(x)))\n",
        "        x = F.relu(self.gnorm2(self.layer2(x)))\n",
        "        x = F.relu(self.gnorm3(self.layer3(x)))\n",
        "        x = torch.sigmoid(self.layer4(x))\n",
        "        return x"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "4b2e386e",
      "metadata": {
        "papermill": {
          "duration": 0.022626,
          "end_time": "2022-01-27T11:15:52.674580",
          "exception": false,
          "start_time": "2022-01-27T11:15:52.651954",
          "status": "completed"
        },
        "tags": [],
        "id": "4b2e386e"
      },
      "source": [
        "# **Code Block - Convert Dataset to PyTorch Compatible Dataset**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "8d9d0096",
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-01-27T11:15:52.726354Z",
          "iopub.status.busy": "2022-01-27T11:15:52.725690Z",
          "iopub.status.idle": "2022-01-27T11:15:52.727500Z",
          "shell.execute_reply": "2022-01-27T11:15:52.727971Z",
          "shell.execute_reply.started": "2022-01-27T11:00:54.119808Z"
        },
        "papermill": {
          "duration": 0.031282,
          "end_time": "2022-01-27T11:15:52.728126",
          "exception": false,
          "start_time": "2022-01-27T11:15:52.696844",
          "status": "completed"
        },
        "tags": [],
        "id": "8d9d0096"
      },
      "outputs": [],
      "source": [
        "from torch.utils.data import Dataset, DataLoader\n",
        "\n",
        "class CustomDataset(Dataset):\n",
        "    def __init__(self, x, y, device):\n",
        "        self._x = x\n",
        "        self._y = y\n",
        "        self._device = device\n",
        "\n",
        "    def __len__(self):\n",
        "        return len(self._x)\n",
        "\n",
        "    def __getitem__(self, idx):\n",
        "        X, Y = self._x[idx], self._y[idx].ravel()\n",
        "        return torch.as_tensor(X, dtype=torch.float32, device=self._device), torch.as_tensor(Y, dtype=torch.float32, device=self._device)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "81e77ab3",
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-01-27T11:15:52.780947Z",
          "iopub.status.busy": "2022-01-27T11:15:52.779981Z",
          "iopub.status.idle": "2022-01-27T11:15:52.782099Z",
          "shell.execute_reply": "2022-01-27T11:15:52.782688Z",
          "shell.execute_reply.started": "2022-01-27T11:00:54.128147Z"
        },
        "papermill": {
          "duration": 0.031891,
          "end_time": "2022-01-27T11:15:52.782850",
          "exception": false,
          "start_time": "2022-01-27T11:15:52.750959",
          "status": "completed"
        },
        "tags": [],
        "id": "81e77ab3"
      },
      "outputs": [],
      "source": [
        "# batch size of train, validation, test don't have to be the same\n",
        "TRAIN_BATCH_SIZE, VAL_BATCH_SIZE, TEST_BATCH_SIZE = 256, 256, 1\n",
        "\n",
        "train_dataset = CustomDataset(xtrain, ytrain, device)\n",
        "val_dataset = CustomDataset(xval, yval, device)\n",
        "test_dataset = CustomDataset(xtest, ytest, device)\n",
        "\n",
        "train_loader = DataLoader(train_dataset, batch_size=TRAIN_BATCH_SIZE, shuffle=True)\n",
        "val_loader = DataLoader(val_dataset, batch_size=VAL_BATCH_SIZE, shuffle=False)\n",
        "test_loader = DataLoader(test_dataset, batch_size=TEST_BATCH_SIZE, shuffle=False)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "2a8f2b4d",
      "metadata": {
        "papermill": {
          "duration": 0.022198,
          "end_time": "2022-01-27T11:15:52.827302",
          "exception": false,
          "start_time": "2022-01-27T11:15:52.805104",
          "status": "completed"
        },
        "tags": [],
        "id": "2a8f2b4d"
      },
      "source": [
        "# **Declare Metrics to Monitor and Load to Device**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "f8619013",
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-01-27T11:15:52.875929Z",
          "iopub.status.busy": "2022-01-27T11:15:52.874991Z",
          "iopub.status.idle": "2022-01-27T11:16:00.012114Z",
          "shell.execute_reply": "2022-01-27T11:16:00.011555Z",
          "shell.execute_reply.started": "2022-01-27T11:00:54.144941Z"
        },
        "papermill": {
          "duration": 7.162482,
          "end_time": "2022-01-27T11:16:00.012276",
          "exception": false,
          "start_time": "2022-01-27T11:15:52.849794",
          "status": "completed"
        },
        "tags": [],
        "id": "f8619013"
      },
      "outputs": [],
      "source": [
        "# unable to import F1Score\n",
        "from torchmetrics import Accuracy, MeanSquaredError, Precision, Recall\n",
        "if torch.cuda.is_available():\n",
        "    loss_fn   = nn.BCELoss().cuda()\n",
        "    metrics = [\n",
        "        {'name': 'Accuracy', 'fn' : Accuracy().cuda(), 'type' : 'int'},\n",
        "        {'name': 'MeanSquaredError', 'fn': MeanSquaredError().cuda(), 'type' : 'float'},\n",
        "        {'name': 'Precision', 'fn': Precision().cuda(), 'type': 'int'},\n",
        "        {'name': 'Recall', 'fn': Recall().cuda(), 'type':'int'}\n",
        "    ]\n",
        "else:\n",
        "    loss_fn = nn.BCELoss().cpu()\n",
        "    metrics = [\n",
        "        {'name': 'Accuracy', 'fn' : Accuracy().cpu(), 'type' : 'int'},\n",
        "        {'name': 'MeanSquaredError', 'fn': MeanSquaredError().cpu(), 'type' : 'float'},\n",
        "        {'name': 'Precision', 'fn': Precision().cpu(), 'type' : 'int'},\n",
        "        {'name': 'Recall', 'fn': Recall().cpu(), 'type' : 'int'}\n",
        "    ]"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "391dc047",
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-01-27T11:16:00.067387Z",
          "iopub.status.busy": "2022-01-27T11:16:00.066762Z",
          "iopub.status.idle": "2022-01-27T11:16:00.102463Z",
          "shell.execute_reply": "2022-01-27T11:16:00.101898Z",
          "shell.execute_reply.started": "2022-01-27T11:01:01.849620Z"
        },
        "papermill": {
          "duration": 0.067391,
          "end_time": "2022-01-27T11:16:00.102612",
          "exception": false,
          "start_time": "2022-01-27T11:16:00.035221",
          "status": "completed"
        },
        "tags": [],
        "id": "391dc047"
      },
      "outputs": [],
      "source": [
        "model     = Model(xtrain.shape[-1], 1)\n",
        "model.to(device)\n",
        "optimizer = torch.optim.Adam(model.parameters(), lr=0.000001)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "d3cab7ae",
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-01-27T11:16:00.195589Z",
          "iopub.status.busy": "2022-01-27T11:16:00.194866Z",
          "iopub.status.idle": "2022-01-27T11:16:00.225425Z",
          "shell.execute_reply": "2022-01-27T11:16:00.224841Z",
          "shell.execute_reply.started": "2022-01-27T11:01:01.896952Z"
        },
        "papermill": {
          "duration": 0.056023,
          "end_time": "2022-01-27T11:16:00.225598",
          "exception": false,
          "start_time": "2022-01-27T11:16:00.169575",
          "status": "completed"
        },
        "tags": [],
        "id": "d3cab7ae"
      },
      "outputs": [],
      "source": [
        "def L1(params, device, l1 = 0.0001):\n",
        "    reg = torch.tensor(0., device=device)\n",
        "    for param in params:\n",
        "        reg += torch.sum(torch.abs(param))\n",
        "    return l1 * reg\n",
        "\n",
        "def L2(params, device, l2 = 0.0001):\n",
        "    reg = torch.tensor(0., device=device)\n",
        "    for param in params:\n",
        "        reg += torch.sum(torch.pow(param, 2))\n",
        "    return l2 * reg\n",
        "\n",
        "def train_loop(dataloader, valloader, model, loss_fn, optimizer, max_iter, metrics : dict, device):\n",
        "    model.train()\n",
        "    history = dict()\n",
        "    history['Loss'] = []\n",
        "    history['val_Loss'] = []\n",
        "    for m in metrics:\n",
        "        history[m['name']] = []\n",
        "        history['val_{}'.format(m['name'])] = []\n",
        "    size = len(dataloader.dataset)\n",
        "    for itr in range(max_iter):\n",
        "        real_time = dict()\n",
        "        real_time['Loss'] = []\n",
        "        for m in metrics:\n",
        "            real_time[m['name']] = []\n",
        "        for batch, (X, y) in enumerate(dataloader):\n",
        "            # Compute prediction and loss\n",
        "            X.to(device)\n",
        "            y.to(device)\n",
        "            pred = model(X)\n",
        "            loss = loss_fn(pred, y)\n",
        "            loss += L2(model.parameters(), device) + L1(model.parameters(), device)\n",
        "            # Backpropagation\n",
        "            optimizer.zero_grad()\n",
        "            loss.backward()\n",
        "            optimizer.step()\n",
        "            # compute metrics\n",
        "            real_time['Loss'].append(loss.item())\n",
        "            for m in metrics:\n",
        "                if m['type'] == 'float':\n",
        "                    real_time[m['name']].append(m['fn'](pred, y).item())\n",
        "                else:\n",
        "                    real_time[m['name']].append(m['fn'](pred, y.type(torch.uint8)).item())\n",
        "        history['Loss'].append(np.mean(real_time['Loss']))\n",
        "        for m in metrics:\n",
        "            history[m['name']].append(np.mean(real_time[m['name']]))\n",
        "            print(f\"{m['name']}: {np.mean(real_time[m['name']]):>8f}\", end='\\t')\n",
        "        print('')\n",
        "        epoch_loss = np.mean(real_time['Loss'])\n",
        "        print(f\"Loss: {epoch_loss:>7f} [{itr:>5d}/{max_iter:>5d}]\")\n",
        "        print('Val', end = '\\t')\n",
        "        tmp = dict()\n",
        "        tmp['Loss'] = []\n",
        "        for m in metrics:\n",
        "            tmp[m['name']] = []\n",
        "        model.eval()\n",
        "        with torch.no_grad():\n",
        "            for X, y in valloader:\n",
        "                X.to(device)\n",
        "                y.to(device)\n",
        "                pred = model(X)\n",
        "                test_loss = loss_fn(pred, y)\n",
        "                tmp['Loss'].append(test_loss.item())\n",
        "                for m in metrics:\n",
        "                    if m['type'] == 'float':\n",
        "                        tmp[m['name']].append(m['fn'](pred, y).item())\n",
        "                    else:\n",
        "                        tmp[m['name']].append(m['fn'](pred, y.type(torch.uint8)).item())\n",
        "        for k, v in tmp.items():\n",
        "            print(f\"{k}: {np.mean(v):>8f}\", end='\\t')\n",
        "            history['val_{}'.format(k)].append(np.mean(v))\n",
        "        print('###END')\n",
        "        #torch.save(model, f'./model_E{itr:>5d}.pt')\n",
        "    return history\n",
        "\n",
        "def test_loop(dataloader, model, loss_fn, metrics : dict, device):\n",
        "    score = dict()\n",
        "    tmp = dict()\n",
        "    tmp['Loss'] = []\n",
        "    for m in metrics:\n",
        "        tmp[m['name']] = []\n",
        "    model.eval()\n",
        "    with torch.no_grad():\n",
        "        for X, y in dataloader:\n",
        "            X.to(device)\n",
        "            y.to(device)\n",
        "            pred = model(X)\n",
        "            test_loss = loss_fn(pred, y)\n",
        "            tmp['Loss'].append(test_loss.item())\n",
        "            for m in metrics:\n",
        "                if m['type'] == 'float':\n",
        "                    tmp[m['name']].append(m['fn'](pred, y).item())\n",
        "                else:\n",
        "                    tmp[m['name']].append(m['fn'](pred, y.type(torch.uint8)).item())\n",
        "    for k, v in tmp.items():\n",
        "        print(f\"{k}: {np.mean(v):>8f}\", end='\\t')\n",
        "        score[k] = np.mean(v)\n",
        "    return score\n",
        "\n",
        "def predict(dataloader, model, device):\n",
        "    ypred = []\n",
        "    model.eval()\n",
        "    with torch.no_grad():\n",
        "        for X, y in dataloader:\n",
        "            X.to(device)\n",
        "            pred = model(X)\n",
        "            ypred.append([torch.ravel(pred), torch.ravel(y)])\n",
        "    return ypred"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "76b53600",
      "metadata": {
        "papermill": {
          "duration": 0.021713,
          "end_time": "2022-01-27T11:16:00.270078",
          "exception": false,
          "start_time": "2022-01-27T11:16:00.248365",
          "status": "completed"
        },
        "tags": [],
        "id": "76b53600"
      },
      "source": [
        "# **Train**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "93b858d8",
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-01-27T11:16:00.320672Z",
          "iopub.status.busy": "2022-01-27T11:16:00.319943Z",
          "iopub.status.idle": "2022-01-27T11:20:59.920149Z",
          "shell.execute_reply": "2022-01-27T11:20:59.919543Z",
          "shell.execute_reply.started": "2022-01-27T11:01:01.931474Z"
        },
        "papermill": {
          "duration": 299.628171,
          "end_time": "2022-01-27T11:20:59.920307",
          "exception": false,
          "start_time": "2022-01-27T11:16:00.292136",
          "status": "completed"
        },
        "tags": [],
        "id": "93b858d8",
        "outputId": "d203a5c9-cfda-46f9-a25f-c39589a5cbe7"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Accuracy: 0.815408\tMeanSquaredError: 0.189324\tPrecision: 0.064894\tRecall: 0.112541\t\n",
            "Loss: 2.464778 [    0/   50]\n",
            "Val\tLoss: 0.454103\tAccuracy: 0.913712\tMeanSquaredError: 0.135534\tPrecision: 0.013889\tRecall: 0.000661\t###END\n",
            "Accuracy: 0.913520\tMeanSquaredError: 0.111507\tPrecision: 0.000000\tRecall: 0.000000\t\n",
            "Loss: 2.290220 [    1/   50]\n",
            "Val\tLoss: 0.350884\tAccuracy: 0.913712\tMeanSquaredError: 0.094809\tPrecision: 0.000000\tRecall: 0.000000\t###END\n",
            "Accuracy: 0.914128\tMeanSquaredError: 0.087392\tPrecision: 0.000000\tRecall: 0.000000\t\n",
            "Loss: 2.220771 [    2/   50]\n",
            "Val\tLoss: 0.309750\tAccuracy: 0.913712\tMeanSquaredError: 0.082276\tPrecision: 0.000000\tRecall: 0.000000\t###END\n",
            "Accuracy: 0.913824\tMeanSquaredError: 0.079618\tPrecision: 0.000000\tRecall: 0.000000\t\n",
            "Loss: 2.190453 [    3/   50]\n",
            "Val\tLoss: 0.289990\tAccuracy: 0.913712\tMeanSquaredError: 0.077393\tPrecision: 0.000000\tRecall: 0.000000\t###END\n",
            "Accuracy: 0.913368\tMeanSquaredError: 0.076355\tPrecision: 0.000000\tRecall: 0.000000\t\n",
            "Loss: 2.173694 [    4/   50]\n",
            "Val\tLoss: 0.277724\tAccuracy: 0.913712\tMeanSquaredError: 0.074631\tPrecision: 0.000000\tRecall: 0.000000\t###END\n",
            "Accuracy: 0.912912\tMeanSquaredError: 0.074377\tPrecision: 0.000000\tRecall: 0.000000\t\n",
            "Loss: 2.161742 [    5/   50]\n",
            "Val\tLoss: 0.268509\tAccuracy: 0.913712\tMeanSquaredError: 0.072539\tPrecision: 0.000000\tRecall: 0.000000\t###END\n",
            "Accuracy: 0.914128\tMeanSquaredError: 0.071596\tPrecision: 0.000000\tRecall: 0.000000\t\n",
            "Loss: 2.148708 [    6/   50]\n",
            "Val\tLoss: 0.260881\tAccuracy: 0.913712\tMeanSquaredError: 0.070702\tPrecision: 0.000000\tRecall: 0.000000\t###END\n",
            "Accuracy: 0.913976\tMeanSquaredError: 0.070052\tPrecision: 0.000000\tRecall: 0.000000\t\n",
            "Loss: 2.139280 [    7/   50]\n",
            "Val\tLoss: 0.254372\tAccuracy: 0.913712\tMeanSquaredError: 0.069113\tPrecision: 0.000000\tRecall: 0.000000\t###END\n",
            "Accuracy: 0.913672\tMeanSquaredError: 0.068786\tPrecision: 0.000000\tRecall: 0.000000\t\n",
            "Loss: 2.131003 [    8/   50]\n",
            "Val\tLoss: 0.248870\tAccuracy: 0.913983\tMeanSquaredError: 0.067722\tPrecision: 0.069444\tRecall: 0.003195\t###END\n",
            "Accuracy: 0.914214\tMeanSquaredError: 0.067423\tPrecision: 0.100000\tRecall: 0.004811\t\n",
            "Loss: 2.122707 [    9/   50]\n",
            "Val\tLoss: 0.244210\tAccuracy: 0.914689\tMeanSquaredError: 0.066471\tPrecision: 0.270833\tRecall: 0.012778\t###END\n",
            "Accuracy: 0.914562\tMeanSquaredError: 0.066610\tPrecision: 0.312963\tRecall: 0.018482\t\n",
            "Loss: 2.116061 [   10/   50]\n",
            "Val\tLoss: 0.240329\tAccuracy: 0.916074\tMeanSquaredError: 0.065487\tPrecision: 0.504630\tRecall: 0.033617\t###END\n",
            "Accuracy: 0.916450\tMeanSquaredError: 0.065501\tPrecision: 0.522407\tRecall: 0.044382\t\n",
            "Loss: 2.108705 [   11/   50]\n",
            "Val\tLoss: 0.237029\tAccuracy: 0.917539\tMeanSquaredError: 0.064616\tPrecision: 0.648843\tRecall: 0.061395\t###END\n",
            "Accuracy: 0.917556\tMeanSquaredError: 0.064902\tPrecision: 0.637526\tRecall: 0.070743\t\n",
            "Loss: 2.102764 [   12/   50]\n",
            "Val\tLoss: 0.234237\tAccuracy: 0.918544\tMeanSquaredError: 0.063939\tPrecision: 0.715278\tRecall: 0.079365\t###END\n",
            "Accuracy: 0.918446\tMeanSquaredError: 0.064117\tPrecision: 0.671396\tRecall: 0.097912\t\n",
            "Loss: 2.096311 [   13/   50]\n",
            "Val\tLoss: 0.231824\tAccuracy: 0.919058\tMeanSquaredError: 0.063325\tPrecision: 0.696627\tRecall: 0.100586\t###END\n",
            "Accuracy: 0.918533\tMeanSquaredError: 0.063907\tPrecision: 0.678380\tRecall: 0.109296\t\n",
            "Loss: 2.091338 [   14/   50]\n",
            "Val\tLoss: 0.229741\tAccuracy: 0.920172\tMeanSquaredError: 0.062818\tPrecision: 0.710483\tRecall: 0.121005\t###END\n",
            "Accuracy: 0.919184\tMeanSquaredError: 0.063396\tPrecision: 0.661077\tRecall: 0.124159\t\n",
            "Loss: 2.085841 [   15/   50]\n",
            "Val\tLoss: 0.227969\tAccuracy: 0.921148\tMeanSquaredError: 0.062390\tPrecision: 0.719252\tRecall: 0.141576\t###END\n",
            "Accuracy: 0.919748\tMeanSquaredError: 0.062816\tPrecision: 0.644370\tRecall: 0.142977\t\n",
            "Loss: 2.079830 [   16/   50]\n",
            "Val\tLoss: 0.226376\tAccuracy: 0.921582\tMeanSquaredError: 0.062017\tPrecision: 0.712296\tRecall: 0.154518\t###END\n",
            "Accuracy: 0.920573\tMeanSquaredError: 0.062400\tPrecision: 0.661244\tRecall: 0.155138\t\n",
            "Loss: 2.074300 [   17/   50]\n",
            "Val\tLoss: 0.224973\tAccuracy: 0.921825\tMeanSquaredError: 0.061695\tPrecision: 0.700959\tRecall: 0.166039\t###END\n",
            "Accuracy: 0.919596\tMeanSquaredError: 0.062765\tPrecision: 0.629719\tRecall: 0.162847\t\n",
            "Loss: 2.070838 [   18/   50]\n",
            "Val\tLoss: 0.223729\tAccuracy: 0.922396\tMeanSquaredError: 0.061406\tPrecision: 0.698692\tRecall: 0.178783\t###END\n",
            "Accuracy: 0.920595\tMeanSquaredError: 0.062093\tPrecision: 0.650674\tRecall: 0.181567\t\n",
            "Loss: 2.064971 [   19/   50]\n",
            "Val\tLoss: 0.222630\tAccuracy: 0.922559\tMeanSquaredError: 0.061154\tPrecision: 0.692560\tRecall: 0.185394\t###END\n",
            "Accuracy: 0.920595\tMeanSquaredError: 0.061860\tPrecision: 0.641919\tRecall: 0.185566\t\n",
            "Loss: 2.059848 [   20/   50]\n",
            "Val\tLoss: 0.221664\tAccuracy: 0.923210\tMeanSquaredError: 0.060939\tPrecision: 0.703557\tRecall: 0.198112\t###END\n",
            "Accuracy: 0.920312\tMeanSquaredError: 0.062049\tPrecision: 0.624500\tRecall: 0.189750\t\n",
            "Loss: 2.056056 [   21/   50]\n",
            "Val\tLoss: 0.220799\tAccuracy: 0.923318\tMeanSquaredError: 0.060742\tPrecision: 0.693390\tRecall: 0.205024\t###END\n",
            "Accuracy: 0.920898\tMeanSquaredError: 0.061455\tPrecision: 0.638154\tRecall: 0.202994\t\n",
            "Loss: 2.050066 [   22/   50]\n",
            "Val\tLoss: 0.220025\tAccuracy: 0.923210\tMeanSquaredError: 0.060574\tPrecision: 0.689762\tRecall: 0.207074\t###END\n",
            "Accuracy: 0.921159\tMeanSquaredError: 0.061393\tPrecision: 0.629296\tRecall: 0.209116\t\n",
            "Loss: 2.045572 [   23/   50]\n",
            "Val\tLoss: 0.219336\tAccuracy: 0.923210\tMeanSquaredError: 0.060439\tPrecision: 0.689290\tRecall: 0.209724\t###END\n",
            "Accuracy: 0.921332\tMeanSquaredError: 0.061113\tPrecision: 0.636479\tRecall: 0.205885\t\n",
            "Loss: 2.040567 [   24/   50]\n",
            "Val\tLoss: 0.218692\tAccuracy: 0.923427\tMeanSquaredError: 0.060284\tPrecision: 0.676406\tRecall: 0.218937\t###END\n",
            "Accuracy: 0.922092\tMeanSquaredError: 0.060895\tPrecision: 0.643258\tRecall: 0.219106\t\n",
            "Loss: 2.035524 [   25/   50]\n",
            "Val\tLoss: 0.218111\tAccuracy: 0.923644\tMeanSquaredError: 0.060174\tPrecision: 0.680915\tRecall: 0.220088\t###END\n",
            "Accuracy: 0.922526\tMeanSquaredError: 0.060694\tPrecision: 0.642023\tRecall: 0.227693\t\n",
            "Loss: 2.030823 [   26/   50]\n",
            "Val\tLoss: 0.217572\tAccuracy: 0.924024\tMeanSquaredError: 0.060059\tPrecision: 0.683274\tRecall: 0.224960\t###END\n",
            "Accuracy: 0.922743\tMeanSquaredError: 0.060581\tPrecision: 0.644632\tRecall: 0.227713\t\n",
            "Loss: 2.026135 [   27/   50]\n",
            "Val\tLoss: 0.217070\tAccuracy: 0.924132\tMeanSquaredError: 0.059938\tPrecision: 0.676924\tRecall: 0.232688\t###END\n",
            "Accuracy: 0.922114\tMeanSquaredError: 0.060744\tPrecision: 0.631345\tRecall: 0.223651\t\n",
            "Loss: 2.022205 [   28/   50]\n",
            "Val\tLoss: 0.216607\tAccuracy: 0.924053\tMeanSquaredError: 0.059822\tPrecision: 0.671178\tRecall: 0.239280\t###END\n",
            "Accuracy: 0.922700\tMeanSquaredError: 0.060362\tPrecision: 0.634311\tRecall: 0.237581\t\n",
            "Loss: 2.016844 [   29/   50]\n",
            "Val\tLoss: 0.216148\tAccuracy: 0.924053\tMeanSquaredError: 0.059731\tPrecision: 0.669929\tRecall: 0.239297\t###END\n",
            "Accuracy: 0.922461\tMeanSquaredError: 0.060358\tPrecision: 0.648318\tRecall: 0.238552\t\n",
            "Loss: 2.012468 [   30/   50]\n",
            "Val\tLoss: 0.215744\tAccuracy: 0.924161\tMeanSquaredError: 0.059654\tPrecision: 0.671591\tRecall: 0.240861\t###END\n",
            "Accuracy: 0.922852\tMeanSquaredError: 0.060140\tPrecision: 0.644004\tRecall: 0.236680\t\n",
            "Loss: 2.007574 [   31/   50]\n",
            "Val\tLoss: 0.215373\tAccuracy: 0.924324\tMeanSquaredError: 0.059572\tPrecision: 0.670153\tRecall: 0.246056\t###END\n",
            "Accuracy: 0.923286\tMeanSquaredError: 0.059858\tPrecision: 0.631782\tRecall: 0.243009\t\n",
            "Loss: 2.002512 [   32/   50]\n",
            "Val\tLoss: 0.215015\tAccuracy: 0.924161\tMeanSquaredError: 0.059511\tPrecision: 0.669676\tRecall: 0.243587\t###END\n",
            "Accuracy: 0.922635\tMeanSquaredError: 0.060006\tPrecision: 0.638244\tRecall: 0.244780\t\n",
            "Loss: 1.998660 [   33/   50]\n",
            "Val\tLoss: 0.214665\tAccuracy: 0.924378\tMeanSquaredError: 0.059437\tPrecision: 0.671896\tRecall: 0.245288\t###END\n",
            "Accuracy: 0.922938\tMeanSquaredError: 0.059833\tPrecision: 0.634943\tRecall: 0.242495\t\n",
            "Loss: 1.993989 [   34/   50]\n",
            "Val\tLoss: 0.214336\tAccuracy: 0.924487\tMeanSquaredError: 0.059365\tPrecision: 0.670616\tRecall: 0.250700\t###END\n",
            "Accuracy: 0.923459\tMeanSquaredError: 0.059448\tPrecision: 0.649895\tRecall: 0.248957\t\n",
            "Loss: 1.988564 [   35/   50]\n",
            "Val\tLoss: 0.214013\tAccuracy: 0.924649\tMeanSquaredError: 0.059296\tPrecision: 0.669991\tRecall: 0.254353\t###END\n",
            "Accuracy: 0.923090\tMeanSquaredError: 0.059816\tPrecision: 0.637427\tRecall: 0.253966\t\n",
            "Loss: 1.985493 [   36/   50]\n",
            "Val\tLoss: 0.213725\tAccuracy: 0.924812\tMeanSquaredError: 0.059253\tPrecision: 0.673591\tRecall: 0.253120\t###END\n",
            "Accuracy: 0.923503\tMeanSquaredError: 0.059407\tPrecision: 0.633291\tRecall: 0.248591\t\n",
            "Loss: 1.979971 [   37/   50]\n",
            "Val\tLoss: 0.213422\tAccuracy: 0.924921\tMeanSquaredError: 0.059170\tPrecision: 0.669547\tRecall: 0.261331\t###END\n",
            "Accuracy: 0.923329\tMeanSquaredError: 0.059442\tPrecision: 0.644132\tRecall: 0.250282\t\n",
            "Loss: 1.975938 [   38/   50]\n",
            "Val\tLoss: 0.213144\tAccuracy: 0.925029\tMeanSquaredError: 0.059118\tPrecision: 0.669963\tRecall: 0.262099\t###END\n",
            "Accuracy: 0.923177\tMeanSquaredError: 0.059597\tPrecision: 0.639379\tRecall: 0.254874\t\n",
            "Loss: 1.972192 [   39/   50]\n",
            "Val\tLoss: 0.212863\tAccuracy: 0.925517\tMeanSquaredError: 0.059053\tPrecision: 0.677094\tRecall: 0.265548\t###END\n",
            "Accuracy: 0.923394\tMeanSquaredError: 0.059443\tPrecision: 0.645817\tRecall: 0.256379\t\n",
            "Loss: 1.967679 [   40/   50]\n",
            "Val\tLoss: 0.212611\tAccuracy: 0.925680\tMeanSquaredError: 0.059000\tPrecision: 0.678319\tRecall: 0.269515\t###END\n",
            "Accuracy: 0.924262\tMeanSquaredError: 0.058810\tPrecision: 0.637209\tRecall: 0.256563\t\n",
            "Loss: 1.961726 [   41/   50]\n",
            "Val\tLoss: 0.212355\tAccuracy: 0.925734\tMeanSquaredError: 0.058953\tPrecision: 0.676948\tRecall: 0.271295\t###END\n",
            "Accuracy: 0.924023\tMeanSquaredError: 0.058882\tPrecision: 0.646745\tRecall: 0.257657\t\n",
            "Loss: 1.957852 [   42/   50]\n",
            "Val\tLoss: 0.212106\tAccuracy: 0.925680\tMeanSquaredError: 0.058895\tPrecision: 0.671929\tRecall: 0.274603\t###END\n",
            "Accuracy: 0.924392\tMeanSquaredError: 0.058764\tPrecision: 0.641833\tRecall: 0.261255\t\n",
            "Loss: 1.953484 [   43/   50]\n",
            "Val\tLoss: 0.211864\tAccuracy: 0.926006\tMeanSquaredError: 0.058861\tPrecision: 0.680067\tRecall: 0.271608\t###END\n",
            "Accuracy: 0.924197\tMeanSquaredError: 0.058940\tPrecision: 0.658513\tRecall: 0.262244\t\n",
            "Loss: 1.949910 [   44/   50]\n",
            "Val\tLoss: 0.211622\tAccuracy: 0.926223\tMeanSquaredError: 0.058806\tPrecision: 0.679257\tRecall: 0.276701\t###END\n",
            "Accuracy: 0.924219\tMeanSquaredError: 0.058750\tPrecision: 0.652384\tRecall: 0.264002\t\n",
            "Loss: 1.945312 [   45/   50]\n",
            "Val\tLoss: 0.211388\tAccuracy: 0.926223\tMeanSquaredError: 0.058748\tPrecision: 0.676050\tRecall: 0.279219\t###END\n",
            "Accuracy: 0.924284\tMeanSquaredError: 0.058627\tPrecision: 0.650540\tRecall: 0.264136\t\n",
            "Loss: 1.941009 [   46/   50]\n",
            "Val\tLoss: 0.211173\tAccuracy: 0.926223\tMeanSquaredError: 0.058708\tPrecision: 0.675987\tRecall: 0.279290\t###END\n",
            "Accuracy: 0.924870\tMeanSquaredError: 0.058341\tPrecision: 0.652057\tRecall: 0.267514\t\n",
            "Loss: 1.936180 [   47/   50]\n",
            "Val\tLoss: 0.210963\tAccuracy: 0.926168\tMeanSquaredError: 0.058667\tPrecision: 0.676044\tRecall: 0.278734\t###END\n",
            "Accuracy: 0.925022\tMeanSquaredError: 0.058246\tPrecision: 0.661546\tRecall: 0.270708\t\n",
            "Loss: 1.931963 [   48/   50]\n",
            "Val\tLoss: 0.210755\tAccuracy: 0.926223\tMeanSquaredError: 0.058633\tPrecision: 0.676886\tRecall: 0.278734\t###END\n",
            "Accuracy: 0.924436\tMeanSquaredError: 0.058491\tPrecision: 0.656856\tRecall: 0.266947\t\n",
            "Loss: 1.928626 [   49/   50]\n",
            "Val\tLoss: 0.210535\tAccuracy: 0.926223\tMeanSquaredError: 0.058581\tPrecision: 0.675772\tRecall: 0.280509\t###END\n"
          ]
        }
      ],
      "source": [
        "EPOCHS = 50\n",
        "train_history = train_loop(train_loader, val_loader, model, loss_fn, optimizer, EPOCHS, metrics, device)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "ee22952c",
      "metadata": {
        "papermill": {
          "duration": 0.052082,
          "end_time": "2022-01-27T11:21:00.027902",
          "exception": false,
          "start_time": "2022-01-27T11:20:59.975820",
          "status": "completed"
        },
        "tags": [],
        "id": "ee22952c"
      },
      "source": [
        "# **Evaluate**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "2ce1b7ed",
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-01-27T11:21:00.139761Z",
          "iopub.status.busy": "2022-01-27T11:21:00.139006Z",
          "iopub.status.idle": "2022-01-27T11:22:24.669551Z",
          "shell.execute_reply": "2022-01-27T11:22:24.670086Z",
          "shell.execute_reply.started": "2022-01-27T11:06:17.896834Z"
        },
        "papermill": {
          "duration": 84.588724,
          "end_time": "2022-01-27T11:22:24.670254",
          "exception": false,
          "start_time": "2022-01-27T11:21:00.081530",
          "status": "completed"
        },
        "tags": [],
        "id": "2ce1b7ed",
        "outputId": "cf65838f-6bec-41fd-b3bd-99196dab1fec"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Loss: 0.209250\tAccuracy: 0.925747\tMeanSquaredError: 0.058181\tPrecision: 0.024133\tRecall: 0.024133\t"
          ]
        }
      ],
      "source": [
        "score = test_loop(test_loader, model, loss_fn, metrics, device)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "43273426",
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-01-27T11:22:24.785676Z",
          "iopub.status.busy": "2022-01-27T11:22:24.784670Z",
          "iopub.status.idle": "2022-01-27T11:22:35.997694Z",
          "shell.execute_reply": "2022-01-27T11:22:35.998375Z",
          "shell.execute_reply.started": "2022-01-27T11:07:45.144669Z"
        },
        "papermill": {
          "duration": 11.273047,
          "end_time": "2022-01-27T11:22:35.998609",
          "exception": false,
          "start_time": "2022-01-27T11:22:24.725562",
          "status": "completed"
        },
        "tags": [],
        "id": "43273426",
        "outputId": "58a18a3f-15e8-43ab-de3c-92e8ae6d7e5b"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Confusion Matrix\n",
            "[[25139     0]\n",
            " [ 2375     0]]\n"
          ]
        }
      ],
      "source": [
        "from sklearn.metrics import confusion_matrix, plot_confusion_matrix\n",
        "\n",
        "ypred = predict(test_loader, model, device)\n",
        "ypred_np = np.array(ypred, dtype='int64')\n",
        "predictions, gt = ypred_np[:,0], ypred_np[:,1]\n",
        "\n",
        "cfm = confusion_matrix(gt, predictions)\n",
        "plt.imshow(cfm, cmap='cool', interpolation='nearest')\n",
        "plt.show()\n",
        "\n",
        "print('Confusion Matrix')\n",
        "print(cfm)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "c1623624",
      "metadata": {
        "papermill": {
          "duration": 0.052598,
          "end_time": "2022-01-27T11:22:36.105612",
          "exception": false,
          "start_time": "2022-01-27T11:22:36.053014",
          "status": "completed"
        },
        "tags": [],
        "id": "c1623624"
      },
      "source": [
        "# **Analysis**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "fb64fd1b",
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-01-27T11:22:36.214410Z",
          "iopub.status.busy": "2022-01-27T11:22:36.213749Z",
          "iopub.status.idle": "2022-01-27T11:22:36.216635Z",
          "shell.execute_reply": "2022-01-27T11:22:36.217093Z",
          "shell.execute_reply.started": "2022-01-27T11:07:56.331344Z"
        },
        "papermill": {
          "duration": 0.058882,
          "end_time": "2022-01-27T11:22:36.217235",
          "exception": false,
          "start_time": "2022-01-27T11:22:36.158353",
          "status": "completed"
        },
        "tags": [],
        "id": "fb64fd1b"
      },
      "outputs": [],
      "source": [
        "import matplotlib.pyplot as plt"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "df344151",
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-01-27T11:22:36.325856Z",
          "iopub.status.busy": "2022-01-27T11:22:36.325203Z",
          "iopub.status.idle": "2022-01-27T11:22:37.137751Z",
          "shell.execute_reply": "2022-01-27T11:22:37.137214Z",
          "shell.execute_reply.started": "2022-01-27T11:07:56.337495Z"
        },
        "papermill": {
          "duration": 0.867502,
          "end_time": "2022-01-27T11:22:37.137893",
          "exception": false,
          "start_time": "2022-01-27T11:22:36.270391",
          "status": "completed"
        },
        "tags": [],
        "id": "df344151",
        "outputId": "68578d3c-340e-440c-f559-6363e9468099"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "5\n"
          ]
        },
        {
          "data": {
            "text/plain": [
              "Text(0, 0.5, 'Loss')"
            ]
          },
          "execution_count": 17,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAA7YAAAZ+CAYAAABzebjmAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/YYfK9AAAACXBIWXMAAAsTAAALEwEAmpwYAAEAAElEQVR4nOz9eZzddX33/z9fc87sM5nJMklIJjuBJMimQwBFUVBEFFDBsmiB1kpblauLttW2X2vpzV972V61rXK1UmrFDaS4oWAjcCkWRc2EsBhCQhKSkCHrbMnsZ3n9/vicM3Nm5szMmeVsM4/77fa+fdZzzutMTjJ5nvf78/6YuwsAAAAAgGJVku8CAAAAAACYDoItAAAAAKCoEWwBAAAAAEWNYAsAAAAAKGoEWwAAAABAUSPYAgAAAACKWjjfBcyURYsW+erVq/NdBgAAAAAgC7Zt23bC3RvSHZs1wXb16tVqbm7OdxkAAAAAgCwwswNjHWMoMgAAAACgqBFsAQAAAABFjWALAAAAAChqBFsAAAAAQFEj2AIAAAAAihrBFgAAAABQ1Ai2AAAAAICiRrAFAAAAABQ1gi0AAAAAoKgRbAEAAAAARS2c7wIAAAAAzH7xuNTdLXV1DS17eqS+vqHW3z/29ljH4vF8v7NAScnUmlmwdA/ey2Rb8nHu038PX/yitGrV9J8nHwi2AAAAQI7E49LJk0GwGxgY3iKRye2bagBK3Z5M8Eq3v69vKKSmBtZ0+/r6pKoqqaZGqq4OllVVUmWlVF4uVVQMtdTt8nKprm7sY6FQvv9Ug5/lVINpsk0nFCfXp2vBguk/R74QbAEAAIBJcJd6e6W2Nqm9ffQy3b7ksrMzCHXV1UEoKyuTSkuDZWqbaF9paRDoUkNOODy5cJp8L5MNxdHo0HpFhbRs2VBQTQ2tI5eVlUOvC8w0gi0AAACywj0IQZPtiRxvXyQSPOfIFoul3z/Rsak8dmAgCJXz5wc9XOmWZ50VLEceq68PAijmnv7+fh0/fjxta21tVX9/vwYGBhSJRBSJRAbXRy7HOlZaWqpFixapoaFhwrZo0SKVl5fn+0cyo/hrBQAAkAH3qYexaDQIM5n2xqXuKy0d6uWKxaZ2DWKyTTfQRSLpg+Z47z8UGv5+ysun1kOZbOFwsD8cDnoLQ6Fgfaw28ngoNPT48R473rHk+yhWkUhELS0t6uzsVEVFxWCrrKxURUWFSktLZdMc1xqPx9Xd3a1Tp06NaidPnhxc7+/vV1lZmcrKylReXj6lVlJSIncf1iSNuz1yXzweVzweVywWm3A93bGurq4xQ2uy9ff3jxk8zznnHFVUVKisrEylpaWjlun2jVwODAzoxIkTo153z549euqpp4btO3HihKqqqkbV8zd/8zdavnz59D5geUKwBQBgjojFguvcTp6UTp1Kvxy5LxnKptobFo8PDZcMhUa3TPZPd+histdwOr13yV7CkUFrMmEs2dM3lR7LZA9fLJb+OsOJtsvLh4a9JoNZZeXkA10oNLnhs6mhHLnT09OjAwcOjNmOHj2qJUuWaMGCBerr61Nvb++wZTweHxZ0x1qWl5erv78/bWDt7u5WZWWl5s2bp9ra2lEtub+iokJdXV3q7++fckuGVDMb1kbuG++cUCikkpISlZSUjLk+3rHq6uphAfHMM88cFV7nzZs37S8MJrJ48eKMznN3dXZ2jgrB1dXVWa0vmwi2AIBZwz34j3/yWrIs//9hytyDwDLVXrdMe+x6eoYH1d7e4Fq3efOk2tr0y3nzgiGTq1YF+5JBaCq9WslQGo8Hfy4jW6b7Z2Kmz+n00CV7HMPh/HymksHcPXgfhfq5Hsnd1dvbq9bWVrW1tQ0uw+HwsGCTbDU1NSqZoRQ8MDCgzs5OdXZ26uTJk6PW4/H4YA9hsrcwtdcwk33JXrJsB5VoNKr+/n719fUNtv7+fnV3d6ulpSVtcD116pRWrlypVatWadWqVVq5cqWuuOKKwe3GxkaVlpaO+5rpAu/IZV9fn8rLy8f88wwVwqxOSMvMVF9fr/r6eq1fvz7f5cwIgi0AIGfcpY4O6cgR6ejR4S25r61t6tfBxWJDvUPx+PSCzHT+r5oMImMFzv7+ICiN18s2cn+6fbW14/fWVVUND63V1fSe5VMsFhsMVh0dHero6Bhcz2RpZlqyZMmwtnTp0lH7GhoaFJ6hizjdXX19feru7h4cVpoaUlOX6fZJ0sKFCwfb/PnzB38OI4en9vT0qLKyMm2vXmorLy/XqVOnxgytnZ2disViqqurU11dnebNmzdqPRQKDV7PmNpG7hvvnOS1jqFQaNxho2MNK5U0Kqym23b3waHC5eXlw4YNL1u2bDCsXnTRRYPrixcvntaXBOFwWDU1NaqpqZmRzxGQCwRbAMCYkj2g4w2TTD3W3y+1to4Oq8l27FgQupYulZYsGd4uvDDYv2DB6F7CTK+hSw2kyV6/qQ6jna5QKBjqOdaQ0EINmLFYTKdOnRoMU8membH+051uO3VdkhoaGrRkyRItXrx4cJlcr6urm5Eer97e3lHXjyXXOzs7VVZWltHQyrGWyTA2VksOwxyv9fT0qKamRvX19aqrqxu1rKur02mnnaYNGzakPebuOnr06GA7cuSIjhw5omeffXbY/ra2Ns2fP39U4F28eLGi0ehgSO3u7lZPT8+w7ZH7e3p6FA6HVV1drerqatXW1mrhwoVasGDBsOWaNWsGw2vqsaqqqoz/DFOvyUwXfJOtr69Pa9asGTO01tXVqaKiIus9qVIQ/KPRaNpJfSaa7EeSKisrh4XVkeG1oqJixr6kAGY785kY31MAmpqavLm5Od9lAEBBcQ/uH3jihHT8eLAcb72zc3RgLSmZ3PWECxYMD6wjQ2xFRb5/KrNfX1+fXn31VR07dmwwoGbSI9jZ2amuri7V1NQMBoSqqqpx/9M90XY8Htfx48d17NgxHT16dNSyv79/WNAduWxoaNDAwMCo68BGTpASjUaHzfaZel1bXV2dIpHIhMMqx1uGw2HNmzdvzJbsXRzveG1t7YwNtR1PNBrViRMnhoXd5M88NaQmW1VV1ah9qccIVgAKhZltc/emdMf4lwoACoB7MES1t3f0cqz1kft6eoJhvKlB9cSJIJguWiQ1NATL1PXVq4fvq6sbPjFM8j6JKAzurra2NrW0tKRthw4dUktLi06ePKnTTjtNS5YsSds7uHTp0jF7DWtra3N6XVyyp3Vk6D106JCefvppHTt2TGVlZYMhdfHixTrrrLNGTcpSU1OTkx66YhAOh7V06VItXbo036UAQM5ktcfWzK6U9M+SQpLucfe/G3F8laQvSWqQ1CbpA+5+yMzOk/SvkuZJikn6jLt/c7zXoscWQD5FIkFvZ0fH1JadnUNDV5OtomL0erp9qesLF44OsZMYCYgZ4O6KRCLq6+ub8Pq98fb39vbqyJEjo8JreXm5GhsbtXz58jFbQ0NDTnoGAQDIpbz02JpZSNJdkt4m6ZCkrWb2kLu/kHLaP0j6irvfa2aXSfpbSb8pqUfSLe7+kpktk7TNzLa4e0e26gWA8bhL7e3S3r3D2759wfLo0WCCnvr6oNdz5LKuLphl9pxzxj5nnAkqkUU9PT2jhrm2traqp6dHvb29g8tkS90e61hJScngsNypzLiavDfjkiVLdO655w4LrcV8KwYAALIlm0ORN0va4+77JMnM7pd0raTUYLtJ0h8n1n8s6buS5O67kye4+6tmdkxBr25HFusFMMfFYtKhQ8MDa2pzl9atG2oXXSS9//3B+ooVuRuyG4vFdOTIEb3yyis6ePDg4PLIkSNqaGjQunXrBtvatWtVWVmZlTr6+/v18ssva+/evdqzZ4/27Nmjffv2DU4YNFUlJSWqrKwcvO6vqqoqo5Y8t7KyctREQula8hrNWCw2aljrwoULVV1drcrKStXX1w8+b7KudOup21yTCABAbmXzN+9ySa+kbB+SdOGIc56V9F4Fw5XfI6nWzBa6e2vyBDPbLKlM0t4s1gpgjnAPrjt98cWhtmuX9NJL0oEDwVDe1PD6nvdIa9cG6wsXZv/+ke6u9vb2wcA6Mry+8sorOnz4sBYsWKAVK1Zo5cqVWrFihVatWqULL7xQx44d0969e/Xoo49q37592r9/vxYuXDgYclND77p167Rw4cJxr0vs7u4eFlyT63v37tXhw4e1cuVKrVu3TqeffrpOP/10ve1tb5t2j2IsFhvs/RzZjh07Nmw7OWvryFZZWTkqrC5dulRnn332qP3V1dVcmwkAQJHL91fKH5f0BTO7TdJPJbUouKZWkmRmp0n6qqRb3T0+8sFmdruk2yVp5cqVuagXQJGIRIJe12RwTQ2y7tLGjdKZZ0obNkhvfKO0fr20Zk12rkft7e0d7B08ceLEmO3w4cM6ePCgysrKtGLFimHB9corrxxcX758ucrLyzN67VgsppaWFu3du3ewfe973xtcj8fjw0JvTU3N4LE9e/aoo6NDa9euHQyu5557rq677jqdfvrpWrlyJT2TAACgIGRt8igzu1jSp9397YntT0qSu//tGOfXSHrR3RsT2/Mk/UTS/8/dH5zo9Zg8Cphb3IMJl9ragutbU8Prrl3Syy9LjY1D4XXDhqH1hobp9bz29vYOu49kcn2swJoc6rpo0aLBNnJ70aJFWrJkiVauXKna2tqZ+0FNoK2tbTDI7tu3T11dXYO9uaeffrqWLVvGJEQAAKAgjDd5VDaDbVjSbkmXK+iJ3SrpZnffkXLOIklt7h43s89Iirn7p8ysTNIPJX3f3f8pk9cj2ALFyV06eVJqbQ1Calvb2Oup2x0dQe/qggXS4sXSGWcMD7Cnnz65+6X29PQMu99jamAdua+/v19LlizR0qVLtWTJksE2VnitqqpiqCsAAMA05WVWZHePmtlHJW1RcLufL7n7DjO7U1Kzuz8k6c2S/tbMXMFQ5I8kHv4bkt4kaWFimLIk3ebuz2SrXgDZE41KBw9Ke/YEkzDt2TPU9u0L7pm6YEFwDeuCBcPXV62Szj8/2C4tPaXe3hadPPmyTpzYo0OHXtb+/ft1+PBh7dwZ0fPPRxWLxTJu0ejQ+aFQaFhITYbWDRs26NJLLx0WYuvq6giqAAAABSSr97HNJXpsgfwaGAiG/6YLrwcPSkuXBr2o69YFy2Rbu1aqrh6aNOnAgQPav3//YEvdjsViWr169WBbtWqVVq9erWXLlqm0tFThcFihUGhSLfmY0tJSwioAAEABy0uPLYDZyT0Irs3N0tat0nPPBeH18OHgljepwfWKK4LlmjVScq6j1tZWvfjii9q1a5e+9rVd2rVr1+DsvWamNWvWDAbW1atX69JLLx1cnz9/PuETAAAAoxBsAYzJPeht3bo1CLLNzdK2bdK8eVJTk3TBBdIf/3FwfevKlVJpafC4SCSiffv2adeuXfr+93cNBtkXX3xR0WhUZ555pjZs2KAzzzxTv/mbv6l169Zp9erVqq+vz+v7BQAAQHEi2AIY9Oqrw0Nsc7MUDgcB9oILpI99THrd64LJmqRgduDt27friSeC4JoMr/v379fy5csHw+uFF16oW265RRs2bNDixYvpdQUAAMCMItgCc9Thw9L27cNDbCQy1BP7+78frC9bNvSYaDSqbdu26Z57Htdjjz2mrVu3asOGDdq4caM2bNig3/zN39SZZ56p008/XRWTmZIYAAAAmAaCLTDLuQeTOm3fLj39dLDcvj2Y7On884MQe+ut0uc/HwwnTu1MdXe98MJOPf7443r88cf1xBNPaMWKFbr88sv1sY99TG9605tyes9VAAAAIB2CLTCLRKPSrl3DQ+wzzwSzDr/2tUGQ/d3fDZYrVgwPsUmvvPLKYJB9/PHHVVZWpre+9a264YYb9MUvflFLlizJ+fsCAAAAxkOwBYqQu3TypPTSS8ND7PPPB0OHkyH2k58Mlg0NYz9XW1ubfvKTn+ixxx7T448/rtbWVl122WV661vfqk9/+tNau3Yt18QCAACgoBFsgQLiLrW2Bte/prZXXx29Lxx2rVzp2rSpX2ec0aUPfahDjY0nFI93qKurS11dXXrxxVNqbg7WT506Nbg/ud7e3q5XXnlFb3jDG3T55Zfr9ttv1znnnKOSkpJ8/ygAAACAjBFsZ4loNKrW1lYdO3ZMx44d0/Hjx0etHz9+XN3d3aqqqlJ1dXXaNtGxmpoa1dXVqa6uTqFQKN9vOyORSES7d+/Wjh079Otf/1o7duzQvn37VFJSotLSUpWVlY27HOtYRUXF4M+rqqpapaW1CoVqVVJSLbNkq1I8XqFIpFR9faaeHqm3V+rpkbq6pKNHk0HV1dIS19GjpoqKuOrre1Vbe0qVle0Kh0/I7Iji8UPq69uvkpK9qq3dpfb2V7R3b0jHjtVq27Za1dTUqKamRrW1o9dra2u1dOnSUftqa2u1ceNGlZWV5fuPCQAAAJgygm2O/Nu/SX//99N7ju7uLg0MDCgWi41q8XhcJSUhhUL1CoUWKhQ6S6FQaFQzK5F7XO6ueDzTpcs9nrLsVDzeLrMSlZSUKBQKlkELpd03fLsk0SM408NbXZFIRAMDAyktomg0onC4VmVlb1JZ2VtVWhqEUymYHElyubvcJ95OXY9GTZFIqaLRUsVipSopiaikpF9mvZJ65N4m94OKx7vk3qtwuF/hcERlZVGVlcVUVhZVJHJIvb371NOzR9XVJ9XYGNXixfO0cOFCLVy4UIsWLUpZX6WFC187uH/BggXMPAwAAACIYJszN9wgve1t03uO8867RO9///u1ZMmSwbCzcOFCLViwQPX19TntQY3H4+ru7tapU6fU2dmpkydP6uTJk6O2x2q9vb2qqanRvHnzJt1qa2t14sQJvfTSS9q9e7deeuklvfTSS3r55Ze1ePFinXXWep1xxhlav3691q9fr3Xr1qm8vHzGfwalpVJVVdAqKqSSknJJ5ZLmjTo3Eomop6dH3d3dg62vr091dRcP/hmGw/x1BAAAAKbCgh6p4tfU1OTNzc35LiNrotGoKisr1d/fPyuuf4xGozp58qQ6OjrU3t6ujo6OwTZye+S+9vZ2NTQ06KyzztJZZ52l17zmNTrrrLO0ceNGVVdX5/utAQAAAMgCM9vm7k3pjtFFVCTa2tpUX18/K0KtJIXDYS1YsEALFizIdykAAAAAitzsSElzQGtrqxYuXJjvMgAAAACg4BBsi0RbWxu9mwAAAACQBsG2SNBjCwAAAADpEWyLBMEWAAAAANIj2BaJ1tZWhiIDAAAAQBoE2yLR1tZGjy0AAAAApEGwLRIMRQYAAACA9Ai2RYJgCwAAAADpEWyLBLf7AQAAAID0CLZFgh5bAAAAAEiPYFskCLYAAAAAkB7BtkgQbAEAAAAgPYJtEejt7ZW7q7KyMt+lAAAAAEDBIdgWgWRvrZnluxQAAAAAKDgE2yLAMGQAAAAAGBvBtggQbAEAAABgbATbIsA9bAEAAABgbATbIkCPLQAAAACMjWBbBAi2AAAAADA2gm0RINgCAAAAwNgItkWAa2wBAAAAYGwE2yJAjy0AAAAAjI1gWwQItgAAAAAwNoJtESDYAgAAAMDYshpszexKM9tlZnvM7BNpjq8ys8fN7Dkz+4mZNaYcu9XMXkq0W7NZZ6HjGlsAAAAAGFvWgq2ZhSTdJekdkjZJusnMNo047R8kfcXdz5F0p6S/TTx2gaS/knShpM2S/srM5mer1kLm7mpvbyfYAgAAAMAYstlju1nSHnff5+4Dku6XdO2IczZJ+n+J9R+nHH+7pEfdvc3d2yU9KunKLNZasDo7O1VVVaXS0tJ8lwIAAAAABSmbwXa5pFdStg8l9qV6VtJ7E+vvkVRrZgszfOyc0NraSm8tAAAAAIwj35NHfVzSpWa2XdKlklokxTJ9sJndbmbNZtZ8/PjxbNWYV21tbUwcBQAAAADjyGawbZG0ImW7MbFvkLu/6u7vdffzJf1FYl9HJo9NnHu3uze5e1NDQ8MMl18YmBEZAAAAAMaXzWC7VdJ6M1tjZmWSbpT0UOoJZrbIzJI1fFLSlxLrWyRdYWbzE5NGXZHYN+cQbAEAAABgfFkLtu4elfRRBYF0p6QH3H2Hmd1pZtckTnuzpF1mtlvSEkmfSTy2TdLfKAjHWyXdmdg353CrHwAAAAAYXzibT+7uj0h6ZMS+T6WsPyjpwTEe+yUN9eDOWfTYAgAAAMD48j15FCZAsAUAAACA8RFsCxzBFgAAAADGR7AtcFxjCwAAAADjI9gWOHpsAQAAAGB8BNsCR7AFAAAAgPERbAscwRYAAAAAxkewLWCRSETd3d2aN29evksBAAAAgIJFsC1g7e3tmj9/vkpK+GMCAAAAgLGQmAoYw5ABAAAAYGIE2wJGsAUAAACAiRFsCxj3sAUAAACAiRFsCxg9tgAAAAAwMYJtASPYAgAAAMDECLYFrLW1laHIAAAAADABgm0Ba2tro8cWAAAAACZAsC1gDEUGAAAAgIkRbAsYwRYAAAAAJkawLWDc7gcAAAAAJkawLWD02AIAAADAxAi2BYxgCwAAAAATI9gWqJ6eHklSVVVVnisBAAAAgMJGsC1QXF8LAAAAAJkh2BYohiEDAAAAQGYItgWKYAsAAAAAmSHYFiiCLQAAAABkhmBboLjGFgAAAAAyQ7AtUPTYAgAAAEBmCLYFimALAAAAAJkh2BYogi0AAAAAZIZgW6C4xhYAAAAAMkOwLVD02AIAAABAZgi2BYpgCwAAAACZIdgWKIItAAAAAGSGYFuA4vG42tvbNX/+/HyXAgAAAAAFj2BbgE6ePKnq6mqVlpbmuxQAAAAAKHgE2wLEMGQAAAAAyBzBtgC1trZyqx8AAAAAyBDBtgC1tbXRYwsAAAAAGcpqsDWzK81sl5ntMbNPpDm+0sx+bGbbzew5M7sqsb/UzO41s+fNbKeZfTKbdRYahiIDAAAAQOayFmzNLCTpLknvkLRJ0k1mtmnEaX8p6QF3P1/SjZL+b2L/+ySVu/vZkl4n6XfNbHW2ai00BFsAAAAAyFw2e2w3S9rj7vvcfUDS/ZKuHXGOS5qXWK+T9GrK/mozC0uqlDQg6WQWay0obW1tXGMLAAAAABnKZrBdLumVlO1DiX2pPi3pA2Z2SNIjku5I7H9QUrekw5IOSvoHd28b+QJmdruZNZtZ8/Hjx2e4/PyhxxYAAAAAMjdhsDWzq80sWwH4JklfdvdGSVdJ+mritTZLiklaJmmNpI+Z2dqRD3b3u929yd2bGhoaslRi7hFsAQAAACBzmQTWGyS9ZGafNbMNk3juFkkrUrYbE/tSfVDSA5Lk7k9JqpC0SNLNkv7b3SPufkzSzyQ1TeK1ixrBFgAAAAAyN2GwdfcPSDpf0l5JXzazpxJDgGsneOhWSevNbI2ZlSmYHOqhEecclHS5JJnZRgXB9nhi/2WJ/dWSLpL0YsbvqshxjS0AAAAAZC6jIcbuflLBda/3SzpN0nskPW1md4zzmKikj0raImmngtmPd5jZnWZ2TeK0j0n6kJk9K+k+Sbe5uyuYTbnGzHYoCMj/6e7PTekdFiF6bAEAAAAgcxbkyHFOCELob0k6XdJXJN3r7sfMrErSC+6+OutVZqCpqcmbm5vzXcaMqKur04EDB1RfX5/vUgAAAACgIJjZNndPe4lqOIPHXyfpc+7+09Sd7t5jZh+ciQIxJBKJqLu7W3V1dfkuBQAAAACKQibB9tMKbrsjSTKzSklL3H2/uz+ercLmqvb2ds2fP19mlu9SAAAAAKAoZHKN7X9JiqdsxxL7kAVcXwsAAAAAk5NJsA27+0ByI7Felr2S5jaCLQAAAABMTibB9njKLMYys2slncheSXMbwRYAAAAAJieTa2x/T9LXzewLkkzSK5JuyWpVcxj3sAUAAACAyZkw2Lr7XkkXmVlNYrsr61XNYfTYAgAAAMDkZNJjKzN7p6SzJFUkZ+t19zuzWNecRbAFAAAAgMmZ8BpbM/s3STdIukPBUOT3SVqV5brmrNbWVoYiAwAAAMAkZDJ51Ovd/RZJ7e7+15IulnRGdsuau9ra2uixBQAAAIBJyCTY9iWWPWa2TFJE0mnZK2luYygyAAAAAExOJtfYft/M6iX9vaSnJbmkf89mUXMZwRYAAAAAJmfcYGtmJZIed/cOSd8ysx9IqnD3zlwUNxdxux8AAAAAmJxxhyK7e1zSXSnb/YTa7KLHFgAAAAAmJ5NrbB83s+sseZ8fZE1PT48kqaqqKs+VAAAAAEDxyCTY/q6k/5LUb2YnzeyUmZ3Mcl1zEr21AAAAADB5E04e5e61uSgEXF8LAAAAAFMxYbA1szel2+/uP535cuY2emwBAAAAYPIyud3Pn6SsV0jaLGmbpMuyUtEcRrAFAAAAgMnLZCjy1anbZrZC0j9lq6C5jGALAAAAAJOXyeRRIx2StHGmCwHX2AIAAADAVGRyje3nJXlis0TSeZKezmJNc1Zra6tOO+20fJcBAAAAAEUlk2tsm1PWo5Luc/efZameOa21tVWvec1r8l0GAAAAABSVTILtg5L63D0mSWYWMrMqd+/JbmlzD9fYAgAAAMDkZXKN7eOSKlO2KyU9lp1y5jausQUAAACAycsk2Fa4e1dyI7Felb2S5i56bAEAAABg8jIJtt1m9trkhpm9TlJv9kqauwi2AAAAADB5mVxj+4eS/svMXpVkkpZKuiGbRc1F8Xhc7e3tDEUGAAAAgEmaMNi6+1Yz2yDpzMSuXe4eyW5Zc8/JkydVXV2tcDiT7xoAAAAAAEkTDkU2s49Iqnb3X7v7ryXVmNmHs1/a3MIwZAAAAACYmkyusf2Qu3ckN9y9XdKHslbRHEWwBQAAAICpySTYhszMkhtmFpJUlr2S5qbW1laurwUAAACAKcjkgs7/lvRNM/tiYvt3Jf0weyXNTW1tbfTYAgAAAMAUZBJs/0zS7ZJ+L7H9nIKZkTGDGIoMAAAAAFMz4VBkd49L+qWk/ZI2S7pM0s7sljX3EGwBAAAAYGrG7LE1szMk3ZRoJyR9U5Lc/S25KW1uaWtr0/r16/NdBgAAAAAUnfF6bF9U0Dv7Lne/xN0/LymWm7LmHnpsAQAAAGBqxgu275V0WNKPzezfzexySTbO+aOY2ZVmtsvM9pjZJ9IcX2lmPzaz7Wb2nJldlXLsHDN7ysx2mNnzZlYxmdcuNgRbAAAAAJiaMYOtu3/X3W+UtEHSjyX9oaTFZvavZnbFRE+cuC3QXZLeIWmTpJvMbNOI0/5S0gPufr6kGyX938Rjw5K+Jun33P0sSW+WFJncWysuBFsAAAAAmJpMJo/qdvdvuPvVkholbVcwU/JENkva4+773H1A0v2Srh359JLmJdbrJL2aWL9C0nPu/myihlZ3n9XDoNva2riPLQAAAABMwYTBNpW7t7v73e5+eQanL5f0Ssr2ocS+VJ+W9AEzOyTpEUl3JPafIcnNbIuZPW1mfzqZOosRPbYAAAAAMDWTCrZZcJOkL7t7o6SrJH3VzEoUzNZ8iaT3J5bvSVzjO4yZ3W5mzWbWfPz48VzWPaMikYh6enpUV1eX71IAAAAAoOhkM9i2SFqRst2Y2Jfqg5IekCR3f0pShaRFCnp3f+ruJ9y9R0Fv7mtHvkCi97jJ3ZsaGhqy8BZyo62tTfPnz5fZpObmAgAAAAAou8F2q6T1ZrbGzMoUTA710IhzDkq6XJLMbKOCYHtc0hZJZ5tZVWIiqUslvZDFWvOK62sBAAAAYOrC2Xpid4+a2UcVhNSQpC+5+w4zu1NSs7s/JOljkv7dzP5IwURSt7m7S2o3s39UEI5d0iPu/nC2as03rq8FAAAAgKnLWrCVJHd/RMEw4tR9n0pZf0HSG8Z47NcU3PJn1iPYAgAAAMDU5XvyKIhgCwAAAADTQbAtAFxjCwAAAABTR7AtAPTYAgAAAMDUEWwLAMEWAAAAAKaOYFsACLYAAAAAMHUE2wLANbYAAAAAMHUE2wJAjy0AAAAATB3BtgAQbAEAAABg6gi2eebuam1tZSgyAAAAAEwRwTbPent7ZWaqqqrKdykAAAAAUJQItnnGMGQAAAAAmB6CbZ4RbAEAAABgegi2ecatfgAAAABgegi2eUaPLQAAAABMD8E2zwi2AAAAADA9BNs8I9gCAAAAwPQQbPOMa2wBAAAAYHoItnlGjy0AAAAATA/BNs8ItgAAAAAwPQTbPCPYAgAAAMD0EGzzjGtsAQAAAGB6CLZ5Ro8tAAAAAEwPwTaP4vG4Ojo66LEFAAAAgGkg2OZRZ2enqqurFQ6H810KAAAAABQtgm0ecX0tAAAAAEwfwTaPuL4WAAAAAKaPYJtHBFsAAAAAmD6CbR61trYyFBkAAAAApolgm0dtbW302AIAAADANBFs84ihyAAAAAAwfQTbPCLYAgAAAMD0EWzziNv9AAAAAMD0EWzziB5bAAAAAJg+gm0eEWwBAAAAYPoItnlEsAUAAACA6SPY5hHX2AIAAADA9BFs8yQSiainp0d1dXX5LgUAAAAAilpWg62ZXWlmu8xsj5l9Is3xlWb2YzPbbmbPmdlVaY53mdnHs1lnPrS1tWn+/Pkys3yXAgAAAABFLWvB1sxCku6S9A5JmyTdZGabRpz2l5IecPfzJd0o6f+OOP6Pkn6YrRrzietrAQAAAGBmZLPHdrOkPe6+z90HJN0v6doR57ikeYn1OkmvJg+Y2bslvSxpRxZrzBuurwUAAACAmZHNYLtc0isp24cS+1J9WtIHzOyQpEck3SFJZlYj6c8k/XUW68sremwBAAAAYGbke/KomyR92d0bJV0l6atmVqIg8H7O3bvGe7CZ3W5mzWbWfPz48exXO4MItgAAAAAwM8JZfO4WSStSthsT+1J9UNKVkuTuT5lZhaRFki6UdL2ZfVZSvaS4mfW5+xdSH+zud0u6W5Kampo8G28iWwi2AAAAADAzshlst0pab2ZrFATaGyXdPOKcg5Iul/RlM9soqULScXd/Y/IEM/u0pK6RobbYcY0tAAAAAMyMrA1FdveopI9K2iJpp4LZj3eY2Z1mdk3itI9J+pCZPSvpPkm3uXtR9bxOFT22AAAAADAzstljK3d/RMGkUKn7PpWy/oKkN0zwHJ/OSnF5RrAFAAAAgJmR78mj5iyCLQAAAADMDIJtnnCNLQAAAADMDIJtntBjCwAAAAAzg2CbB+5OsAUAAACAGUKwzYOenh6ZmSorK/NdCgAAAAAUPYJtHrS1tdFbCwAAAAAzhGCbBwxDBgAAAICZQ7DNA4ItAAAAAMwcgm0ecKsfAAAAAJg5BNs8oMcWAAAAAGYOwTYPCLYAAAAAMHMItnlAsAUAAACAmUOwzQOusQUAAACAmUOwzQN6bAEAAABg5hBs84BgCwAAAAAzh2CbBwRbAAAAAJg5BNs84BpbAAAAAJg5BNsci8fj6ujoINgCAAAAwAwh2OZYZ2enqqurFQ6H810KAAAAAMwKBNsc4/paAAAAAJhZBNsc4/paAAAAAJhZBNsco8cWAAAAAGYWwTbHCLYAAAAAMLMItjnW2trKUGQAAAAAmEEE2xxra2ujxxYAAAAAZhDBNscYigwAAAAAM4tgm2MEWwAAAACYWQTbHON2PwAAAAAwswi2OUaPLQAAAADMLIJtjhFsAQAAAGBmEWxzjGALAAAAADOLYJtDkUhEvb29mjdvXr5LAQAAAIBZg2CbQ21tbZo/f77MLN+lAAAAAMCsQbDNIYYhAwAAAMDMI9jmEMEWAAAAAGYewTaHuIctAAAAAMw8gm0O0WMLAAAAADMvq8HWzK40s11mtsfMPpHm+Eoz+7GZbTez58zsqsT+t5nZNjN7PrG8LJt15grBFgAAAABmXjhbT2xmIUl3SXqbpEOStprZQ+7+QsppfynpAXf/VzPbJOkRSaslnZB0tbu/amavkbRF0vJs1ZorBFsAAAAAmHnZ7LHdLGmPu+9z9wFJ90u6dsQ5Lil5U9c6Sa9Kkrtvd/dXE/t3SKo0s/Is1poTXGMLAAAAADMvaz22CnpYX0nZPiTpwhHnfFrSj8zsDknVkt6a5nmuk/S0u/dno8hcoscWAAAAAGZeviePuknSl929UdJVkr5qZoM1mdlZkv63pN9N92Azu93Mms2s+fjx4zkpeDoItgAAAAAw87IZbFskrUjZbkzsS/VBSQ9Ikrs/JalC0iJJMrNGSd+RdIu77033Au5+t7s3uXtTQ0PDDJc/8wi2AAAAADDzshlst0pab2ZrzKxM0o2SHhpxzkFJl0uSmW1UEGyPm1m9pIclfcLdf5bFGnOKa2wBAAAAYOZlLdi6e1TSRxXMaLxTwezHO8zsTjO7JnHaxyR9yMyelXSfpNvc3ROPO13Sp8zsmURbnK1ac8Hd6bEFAAAAgCywIEcWv6amJm9ubs53GWPq7u5WQ0ODenp68l0KAAAAABQdM9vm7k3pjuV78qg5o7W1lWHIAAAAAJAFBNscaWtrYxgyAAAAAGQBwTZHuL4WAAAAALKDYJsjBFsAAAAAyA6CbY5wqx8AAAAAyA6CbY7E43GtXLky32UAAAAAwKwTzncBc8WHP/zhfJcAAAAAALMSPbYAAAAAgKJGsAUAAAAAFDWCLQAAAACgqBFsAQAAAABFjWALAAAAAChqBFsAAAAAQFEj2AIAAAAAihrBFgAAAABQ1Ai2AAAAAICiRrAFAAAAABQ1gi0AAAAAoKiZu+e7hhlhZsclHcjiSyySdCKLzw9MBZ9LFCI+lyg0fCZRiPhcohAV+udylbs3pDswa4JttplZs7s35bsOIBWfSxQiPpcoNHwmUYj4XKIQFfPnkqHIAAAAAICiRrAFAAAAABQ1gm3m7s53AUAafC5RiPhcotDwmUQh4nOJQlS0n0uusQUAAAAAFDV6bAEAAAAARY1gmwEzu9LMdpnZHjP7RL7rwdxkZl8ys2Nm9uuUfQvM7FEzeymxnJ/PGjG3mNkKM/uxmb1gZjvM7A8S+/lcIm/MrMLMfmVmzyY+l3+d2L/GzH6Z+F3+TTMry3etmFvMLGRm283sB4ltPpPIKzPbb2bPm9kzZtac2Fe0v8MJthMws5CkuyS9Q9ImSTeZ2ab8VoU56suSrhyx7xOSHnf39ZIeT2wDuRKV9DF33yTpIkkfSfz7yOcS+dQv6TJ3P1fSeZKuNLOLJP1vSZ9z99MltUv6YP5KxBz1B5J2pmzzmUQheIu7n5dyi5+i/R1OsJ3YZkl73H2fuw9Iul/StXmuCXOQu/9UUtuI3ddKujexfq+kd+eyJsxt7n7Y3Z9OrJ9S8B+25eJziTzyQFdiszTRXNJlkh5M7OdziZwys0ZJ75R0T2LbxGcShalof4cTbCe2XNIrKduHEvuAQrDE3Q8n1o9IWpLPYjB3mdlqSedL+qX4XCLPEkM+n5F0TNKjkvZK6nD3aOIUfpcj1/5J0p9Kiie2F4rPJPLPJf3IzLaZ2e2JfUX7Ozyc7wIAzAx3dzNjmnPknJnVSPqWpD9095NBR0SAzyXywd1jks4zs3pJ35G0Ib8VYS4zs3dJOubu28zszXkuB0h1ibu3mNliSY+a2YupB4vtdzg9thNrkbQiZbsxsQ8oBEfN7DRJSiyP5bkezDFmVqog1H7d3b+d2M3nEgXB3Tsk/VjSxZLqzSz5hT6/y5FLb5B0jZntV3BJ22WS/ll8JpFn7t6SWB5T8CXgZhXx73CC7cS2SlqfmLmuTNKNkh7Kc01A0kOSbk2s3yrpe3msBXNM4hqx/5C0093/MeUQn0vkjZk1JHpqZWaVkt6m4PrvH0u6PnEan0vkjLt/0t0b3X21gv9H/j93f7/4TCKPzKzazGqT65KukPRrFfHvcHMvmt7lvDGzqxRcGxGS9CV3/0x+K8JcZGb3SXqzpEWSjkr6K0nflfSApJWSDkj6DXcfOcEUkBVmdomk/5H0vIauG/tzBdfZ8rlEXpjZOQomPAkp+AL/AXe/08zWKugtWyBpu6QPuHt//irFXJQYivxxd38Xn0nkU+Lz953EZljSN9z9M2a2UEX6O5xgCwAAAAAoagxFBgAAAAAUNYItAAAAAKCoEWwBAAAAAEWNYAsAAAAAKGoEWwAAAABAUSPYAgAAAACKGsEWAAAAAFDUCLYAAAAAgKJGsAUAAAAAFDWCLQAAAACgqBFsAQAAAABFjWALAAAAAChqBFsAAAAAQFEj2AIAAAAAihrBFgAAAABQ1Ai2AAAAAICiRrAFAAAAABQ1gi0AAAAAoKgRbAEAAAAARY1gCwAAAAAoagRbAAAAAEBRI9gCAAAAAIoawRYAAAAAUNQItgAAAACAokawBQAAAAAUNYItAAAAAKCoEWwBAAAAAEWNYAsAAAAAKGoEWwAAAABAUSPYAgAAAACKGsEWAAAAAFDUCLYAAAAAgKJGsAUAAAAAFDWCLQAAAACgqBFsAQAAAABFjWALAAAAAChqBFsAAAAAQFEj2AIAAAAAihrBFgAAAABQ1Ai2AAAAAICiFs53ATNl0aJFvnr16nyXAQAAAADIgm3btp1w94Z0x2ZNsF29erWam5vzXQYAAAAAIAvM7MBYxxiKDAAAAAAoagRbAAAAAEBRI9gCAAAAAIoawRYAAAAAUNQItgAAAACAokawBQAAAAAUNYItAAAAAKCoEWwBAAAAAEWNYAsAAAAAKGoEWwAAAABAUSPY5shLL72kX//61/kuAwAAAABmHYJtjjz88MO6++67810GAAAAAMw6BNscWb58uQ4dOpTvMgAAAABg1iHY5khjYyPBFgAAAACygGCbI42NjWppacl3GQAAAAAw6xBsc2Tp0qU6fvy4IpFIvksBAAAAgFmFYJsjpaWlamho0JEjR/JdCgAAAADMKgTbHOI6WwAAAACYeQTbHCLYAgAAAMDMI9jmEMEWAAAAAGYewTaHCLYAAAAAMPMItjlEsAUAAACAmUewzSGCLQAAAADMPIJtDhFsAQAAAGDmEWxzaNmyZTp8+LBisVi+SwEAAACAWYNgm0Pl5eWaP3++jh07lu9SAAAAAGDWyGqwNbMrzWyXme0xs0+kOf4mM3vazKJmdv2IY581sx1mttPM/sXMLJu15grDkQEAAABgZmUt2JpZSNJdkt4haZOkm8xs04jTDkq6TdI3Rjz29ZLeIOkcSa+RdIGkS7NVay4RbAEAAABgZoWz+NybJe1x932SZGb3S7pW0gvJE9x9f+JYfMRjXVKFpDJJJqlU0tEs1pozBFsAAAAAmFnZHIq8XNIrKduHEvsm5O5PSfqxpMOJtsXdd448z8xuN7NmM2s+fvz4DJScfcuXLyfYAgAAAMAMKsjJo8zsdEkbJTUqCMOXmdkbR57n7ne7e5O7NzU0NOS6zClpbGxUS0tLvssAAAAAgFkjm8G2RdKKlO3GxL5MvEfSL9y9y927JP1Q0sUzXF9eMBQZAAAAAGZWNoPtVknrzWyNmZVJulHSQxk+9qCkS80sbGalCiaOGjUUuRgRbAEAAABgZmUt2Lp7VNJHJW1REEofcPcdZnanmV0jSWZ2gZkdkvQ+SV80sx2Jhz8oaa+k5yU9K+lZd/9+tmrNpeQ1tu6e71IAAAAAYFaw2RKwmpqavLm5Od9lZGTBggXavXu3Fi1alO9SAAAAAKAomNk2d29Kd6wgJ4+a7RiODAAAAAAzh2CbBwRbAAAAAJg5BNs8INgCAAAAwMwh2OYBwRYAAAAAZg7BNg8ItgAAAAAwcwi2eUCwBQAAAICZQ7DNA4ItAAAAAMwcgm0eJIPtbLmHMAAAAADkE8E2D+bNm6eSkhJ1dnbmuxQAAAAAKHoE2zxhODIAAAAAzAyCbZ4QbAEAAABgZhBs82T58uUEWwAAAACYAQTbPGlsbFRLS0u+ywAAAACAokewzROGIgMAAADAzCDY5gnBFgAAAABmBsE2Twi2AAAAADAzCLZ5QrAFAAAAgJlBsM2TBQsWqK+vT11dXfkuBQAAAACKGsE2T8yMmZEBAAAAYAYQbPOI4cgAAAAAMH0E2zwi2AIAAADA9BFs84hgCwAAAADTR7DNI4ItAAAAAEwfwTaPCLYAAAAAMH0E2zwi2AIAAADA9BFs84hgCwAAAADTR7DNo4aGBp08eVJ9fX35LgUAAAAAihbBNo9KSkq0bNkytbS05LsUAAAAAChaBNs8W758OcEWAAAAAKaBYJtnXGcLAAAAANOT1WBrZlea2S4z22Nmn0hz/E1m9rSZRc3s+hHHVprZj8xsp5m9YGars1lrvhBsAQAAAGB6shZszSwk6S5J75C0SdJNZrZpxGkHJd0m6RtpnuIrkv7e3TdK2izpWLZqzSeCLQAAAABMTzZ7bDdL2uPu+9x9QNL9kq5NPcHd97v7c5LiqfsTATjs7o8mzuty954s1po3BFsAAAAAmJ5sBtvlkl5J2T6U2JeJMyR1mNm3zWy7mf19ogd4GDO73cyazaz5+PHjM1By7hFsAQAAAGB6CnXyqLCkN0r6uKQLJK1VMGR5GHe/292b3L2poaEhtxXOEIItAAAAAExPNoNti6QVKduNiX2ZOCTpmcQw5qik70p67cyWVxiWLl2qEydOKBKJ5LsUAAAAAChK2Qy2WyWtN7M1ZlYm6UZJD03isfVmluyGvUzSC1moMe/C4bAWL16sw4cP57sUAAAAAChKWQu2iZ7Wj0raImmnpAfcfYeZ3Wlm10iSmV1gZockvU/SF81sR+KxMQXDkB83s+clmaR/z1at+cZwZAAAAACYunA2n9zdH5H0yIh9n0pZ36pgiHK6xz4q6Zxs1lcoCLYAAAAAMHWFOnnUnEKwBQAAAICpI9gWAIItAAAAAEwdwbYAEGwBAAAAYOoItgWAYAsAAAAAU0ewLQAEWwAAAACYOoJtAVi2bJmOHDmiWCyW71IAAAAAoOgQbAtAWVmZFixYoGPHjuW7FAAAAAAoOgTbArF8+XKGIwMAAADAFBBsCwTX2QIAAADA1BBsCwTBFgAAAACmhmBbIAi2AAAAADA1BNsCQbAFAAAAgKkh2BYIgi0AAAAATA3BtkAQbAEAAABgagi2BWL58uVqaWmRu+e7FAAAAAAoKgTbAlFVVaXq6mqdOHEi36UAAAAAQFEh2BYQhiMDAAAAwOQRbAsIwRYAAAAAJo9gW0AItgAAAAAweQTbAkKwBQAAAIDJI9gWEIItAAAAAEzehMHWzEJm9mIuipnrCLYAAAAAMHkTBlt3j0naZWYrc1DPnNbY2KiWlpZ8lwEAAAAARSWc4XnzJe0ws19J6k7udPdrslLVHJXssXV3mVm+ywEAAACAopBpsP3/sloFJEm1tbUqKSlRZ2en6uvr810OAAAAABSFjCaPcvcnJL0oqTbRdib2YYZxnS0AAAAATE5GwdbMfkPSryS9T9JvSPqlmV2fzcLmKoItAAAAAExOpkOR/0LSBe5+TJLMrEHSY5IezFZhcxXBFgAAAAAmJ9P72JYkQ21C6yQei0kg2AIAAADA5GTaY/vfZrZF0n2J7RskPZKdkua2xsZG/epXv8p3GQAAAABQNCbsdbXgvjP/IumLks5JtLvd/c8yeOyVZrbLzPaY2SfSHH+TmT1tZtF01+ya2TwzO2RmX8jo3cwC9NgCAAAAwORM2GPr7m5mj7j72ZK+nekTm1lI0l2S3ibpkKStZvaQu7+QctpBSbdJ+vgYT/M3kn6a6WvOBgRbAAAAAJicTK+TfdrMLpjkc2+WtMfd97n7gKT7JV2beoK773f35yTFRz7YzF4naYmkH03ydYsawRYAAAAAJifTYHuhpKfMbK+ZPWdmz5vZcxM8ZrmkV1K2DyX2TcjMSiT9H43dk5s873Yzazaz5uPHj2fy1AVv/vz5GhgY0KlTp/JdCgAAAAAUhQmHIieusb1d0oHslzPow5IecfdDwcun5+53S7pbkpqamjxHtWWVmamxsVEtLS3asGFDvssBAAAAgIKX6TW2dyWusZ2MFkkrUrYbE/sycbGkN5rZhyXVSCozsy53HzUB1WyUHI5MsAUAAACAiWV6u5+nzewCd986iefeKmm9ma1REGhvlHRzJg909/cn183sNklNcyXUSlxnCwAAAACTkbVrbN09KumjkrZI2inpAXffYWZ3mtk1kmRmF5jZIUnvk/RFM9sx9bcyexBsAQAAACBzmfbYvn0qT+7uj0h6ZMS+T6Wsb1UwRHm85/iypC9P5fWLVWNjo557bqK5uQAAAAAA0gQ9tmZ2mSS5+wFJJe5+INkkvS4XBc5FycmjAAAAAAATm2go8j+krH9rxLG/nOFakMBQZAAAAADI3ETB1sZYT7eNcTz9tPTUU5mdu3z5coItAAAAAGRoomDrY6yn28Y4tm2T/u3fMju3oaFBJ0+eVF9fX3aLAgAAAIBZYKLJo9aa2UMKemeT60psr8lqZbPM5s3S5z6X2bklJSVatmyZWlpatG7duuwWBgAAAABFbqJge23K+j+MODZyG+M46yzp4EHp5Elp3ryJz09eZ0uwBQAAAIDxjRts3f2JXBUy24XD0rnnBkOS3/KWic9nAikAAAAAyMy4wdbMntc419K6+zkzXtEstnmztHUrwRYAAAAAZtJEQ5HflVh+JLH8amL5ATF51KRdcIH07W9ndm5jY6P27NmT3YIAAAAAYBYYd1Zkdz/g7gckvc3d/9Tdn0+0P5N0RW5KnD0uuCDosc0EPbYAAAAAkJmJbveTZGb2hpSN10/isUg4/XTp1Cnp6NGJzyXYAgAAAEBmMg2nH5T0f81sv5ntl/R/Jf121qqapcykpqbMem0JtgAAAACQmYyCrbtvc/dzJZ0r6Vx3P8/dn85uabNTcgKpiSxdulStra0aGBjIflEAAAAAUMQyCrZmtsTM/kPS/e7eaWabzOyDWa5tVrrgAulXv5r4vFAopCVLlujw4cPZLwoAAAAAilimQ5G/LGmLpGWJ7d2S/jAL9cx6yQmkPIM5pRmODAAAAAATyzTYLnL3ByTFJcndo5JiWatqFlu2TKqokPbvn/hcgi0AAAAATCzTYNttZguVuHetmV0kqTNrVc1ymQ5HbmxsVEtLS/YLAgAAAIAilmmw/WNJD0laZ2Y/k/QVSXdkrapZLtMJpOixBQAAAICJTRhszSwk6dJEe72k35V0lrs/l+XaZq3J9NgSbAEAAABgfBMGW3ePSbrJ3aPuvsPdf+3ukRzUNms1NUnbt0uxCa5SXr58OcEWAAAAACaQ6VDkn5nZF8zsjWb22mTLamWzWH19MInUzp3jn0ePLQAAAABMLJzheecllnem7HNJl81oNXNIcjjya14z9jnLli3TkSNHFIvFFAqFclccAAAAABSRjIKtu78l24XMNckJpH77t8c+p6ysTAsWLNDRo0e1bNmysU8EAAAAgDks0x5bmdk7JZ0lqSK5z93vHPsRGM8FF0j33jvxecnhyARbAAAAAEgvo2tszezfJN2g4BY/Jul9klZlsa5Z77zzgmts+/rGP4/rbAEAAABgfJlOHvV6d79FUru7/7WkiyWdkb2yZr/KSmnDBunZZ8c/j2ALAAAAAOPLNNj2JpY9ZrZMUkTSadkpae7I5H62BFsAAAAAGF+mwfYHZlYv6e8lPS1pv6T7slTTnHHBBcEEUuMh2AIAAADA+DIKtu7+N+7e4e7fUnBt7QZ3//+yW9rst3kzPbYAAAAAMF0ZzYpsZrek2Sd3/8rMlzR3bNokHTokdXZKdXXpzyHYAgAAAMD4Mh2KfEFKe6OkT0u6ZqIHmdmVZrbLzPaY2SfSHH+TmT1tZlEzuz5l/3lm9pSZ7TCz58zshgzrLCrhsHT++dK2bWOfs3z5crW0tCgej+euMAAAAAAoIhn12Lr7Hanbiett7x/vMWYWknSXpLdJOiRpq5k95O4vpJx2UNJtkj4+4uE9km5x95cSk1VtM7Mt7t6RSb3FJDmB1GWXpT9eWVmp2tpanThxQosXL85tcQAAAABQBDLtsR2pW9KaCc7ZLGmPu+9z9wEFQfja1BPcfb+7PycpPmL/bnd/KbH+qqRjkhqmWGtBYwIpAAAAAJiejIKtmX3fzB5KtB9I2iXpOxM8bLmkV1K2DyX2TYqZbZZUJmlvmmO3m1mzmTUfP358sk9dEDKdQKqlpSU3BQEAAABAkcloKLKkf0hZj0o64O5Z70I0s9MkfVXSre4+6iJTd79b0t2S1NTU5NmuJxvWrpV6eqTDh6XTxrgzMD22AAAAADC2TK+xfWIKz90iaUXKdmNiX0bMbJ6khyX9hbv/YgqvXxTMhoYjXzPGdFwEWwAAAAAYW6ZDkU+Z2ck07ZSZnRzjYVslrTezNWZWJulGSQ9l+HplCoY6f8XdH8zkMcVsoutsly9fTrAFAAAAgDFkOnnUP0n6hIJrZBsl/Zmkf3L3Wnefl+4B7h6V9FFJWyTtlPSAu+8wszvN7BpJMrMLzOyQpPdJ+qKZ7Ug8/DckvUnSbWb2TKKdN6V3WAQmCrb02AIAAADA2Mx94ktTzexZdz93on351NTU5M3NzfkuY0oOH5Ze8xrpxIlgaPJIO3fu1Lvf/W7t2rUr98UBAAAAQAEws23u3pTuWKY9tt1m9n4zC5lZiZm9X8EtfzADTjtNqqqS9u1LfzzZY5vJlxAAAAAAMNdkGmxvVjA8+KiCe8q+L7EPM+SCC8a+7U9tba3C4bA6OjpyWhMAAAAAFIOMgq2773f3a919UaK92933Z7m2OWXzZq6zBQAAAICpGDfYmtmHzGx9Yt3M7Etm1mlmz5nZa3NT4tzABFIAAAAAMDUT9dj+gaT9ifWbJJ0raa2kP5b0z9kra+5papK2b5ei0fTHCbYAAAAAkN5EwTbq7pHE+rsU3Fe21d0fk1Sd3dLmlro6qbFReuGF9McJtgAAAACQ3kTBNm5mp5lZhaTLJT2Wcqwye2XNTeNNIEWwBQAAAID0Jgq2n5LUrGA48kPuvkOSzOxSSWPcnAZTNd4EUgRbAAAAAEgvPN5Bd/+Bma2SVOvu7SmHmiXdkNXK5qALLpD+8z/THyPYAgAAAEB64wZbSXL3qKR2M3u9pNUjHvOVLNU1J513nvTii1Jvr1Q5YqA3wRYAAAAA0svoPrZm9lVJ/yDpEkkXJFpTFuuakyoqpI0bpWeeGX2svr5e0WhUJ0+ezHldAAAAAFDIJuyxTWiStMndPZvFYGgCqYsvHr7fzHTxxRfr0Ucf1XXXXZef4gAAAACgAGXUYyvp15KWZrMQBMabQOrmm2/WN77xjdwWBAAAAAAFLtNgu0jSC2a2xcweSrZsFjZXXXDB2MH2ve99rx577DF1dHTktCYAAAAAKGSZDkX+dDaLwJBNm6RXX5U6OqT6+uHH6uvrdfnll+s73/mOfuu3fisf5QEAAABAwcmox9bdn0jXsl3cXBQKSeefLzU3pz/OcGQAAAAAGC7TWZEvMrOtZtZlZgNmFjMzpufNkuQEUum8853vVHNzsw4fPpzbogAAAACgQGV6je0XJN0k6SVJlZJ+R9Jd2SpqrhtvAqnKykpde+21+uY3v5nbogAAAACgQGUabOXueySF3D3m7v8p6crslTW3jTeBlMRwZAAAAABIlWmw7TGzMknPmNlnzeyPJvFYTNKaNVJfXzCJVDqXXXaZDh48qJdeeim3hQEAAABAAco0nP5m4tyPSuqWtELSddkqaq4zG7/XNhwO64YbbtB9992X28IAAAAAoABlOivyAUkm6TR3/2t3/+PE0GRkyXgTSElDw5HdPXdFAQAAAEABynRW5KslPSPpvxPb55nZQ1msa84bbwKp4PhmRSIRbd++PXdFAQAAAEABynQo8qclbZbUIUnu/oykNVmpCJKCHtvmZmmsDlkz080336yvf/3ruS0MAAAAAApMpsE24u6dI/YxBjaLliyRamqkPeMM+L755pt1//33KxaL5a4wAAAAACgwmQbbHWZ2s6SQma03s89L+nkW64ImHo68ceNGLVmyRD/96U9zVxQAAAAAFJhMg+0dks6S1C/pPkknJf1hlmpCwkQTSEnc0xYAAAAAMp0Vucfd/8LdL3D3psR6X7aLm+sm6rGVpBtvvFHf/va31d/fn5uiAAAAAKDAhMc7ONHMx+5+zcyWg1Sve5307LNSJCKVlqY/p7GxUWeffbZ++MMf6t3vfndO6wMAAACAQjBusJV0saRXFAw//qWCe9kiR+bNk1askHbskM47b+zzksORCbYAAAAA5qKJhiIvlfTnkl4j6Z8lvU3SCXd/wt2fmOjJzexKM9tlZnvM7BNpjr/JzJ42s6iZXT/i2K1m9lKi3Zr5W5pdMhmOfP3112vLli06efJkbooCAAAAgAIybrB195i7/7e73yrpIkl7JP3EzD460RObWUjSXZLeIWmTpJvMbNOI0w5Kuk3SN0Y8doGkv5J0oYL75/6Vmc3P6B3NMplMILVgwQK9+c1v1ne/+92c1AQAAAAAhWTCyaPMrNzM3ivpa5I+IulfJH0ng+feLGmPu+9z9wFJ90u6NvUEd9/v7s9Jio947NslPerube7eLulRSVdm8JqzTiY9thKzIwMAAACYu8YNtmb2FUlPSXqtpL9OzIr8N+7eksFzL1dwfW7SocS+TGT0WDO73cyazaz5+PHjGT51cTn3XGn3bqmnZ/zzrr76av3iF7/Q0aNHc1MYAAAAABSIiXpsPyBpvaQ/kPRzMzuZaKfMLO8XdLr73YnbDzU1NDTku5ysKC+XNm2Stm8f/7yqqipdffXVeuCBB3JTGAAAAAAUiImusS1x99pEm5fSat193gTP3SJpRcp2Y2JfJqbz2FmH4cgAAAAAMLYJr7Gdhq2S1pvZGjMrk3SjpHHvi5tii6QrzGx+YtKoKxL75qRMJpCSpLe+9a3au3ev9u3bl/2iAAAAAKBAZC3YuntU0kcVBNKdkh5w9x1mdqeZXSNJZnaBmR2S9D5JXzSzHYnHtkn6GwXheKukOxP75qQLLsisx7a0tFS/8Ru/ofvuuy/7RQEAAABAgTB3z3cNM6Kpqcmbm5vzXUZWxGJSfb104IC0YMH45/785z/X7/zO72jHjh0ys5zUBwAAAADZZmbb3L0p3bFsDkXGDAmFpNe+Vsokt1988cXq7e3Vs88+m/3CAAAAAKAAEGyLRKYTSJmZbrrpJiaRAgAAADBnEGyLRKYTSEnB7Mj33Xef4vF4dosCAAAAgAJAsC0SmU4gJUmvec1rNH/+fD355JPZLQoAAAAACgDBtkisXi2ZSU89ldn53NMWAAAAwFxBsC0SZtI//ZP0W78l9fZOfP5NN92kBx98UAMDA1mvDQAAAADyiWBbRN73Pum886S/+IuJz121apU2btyoLVu2ZL0uAAAAAMgngm2Ruesu6f77pZ/+dOJzGY4MAAAAYC4g2BaZhQulL35Ruu02qatr/HPf97736ZFHHlHXRCcCAAAAQBEj2Bahq6+WLr1U+pM/Gf+8RYsW6ZJLLtH3vve93BQGAAAAAHlAsC1S//RP0sMPSz/60fjnMRwZAAAAwGxHsC1SdXXSPfdIv/M7UkfH2Odde+21+tnPfqbjx4/nrDYAAAAAyCWCbRG74grpne+U/uiPxj6npqZGV111lf7rv/4rd4UBAAAAQA4RbIvc3/99MEPyQw+NfQ7DkQEAAADMZgTbIldTI/3nf0q/93tSa2v6c6644gq9+OKL2r9/f05rAwAAAIBcINjOAm96k3TjjdJHPpL+eFlZma6//nrdf//9uS0MAAAAAHKAYDtLfOYz0jPPSA88kP44w5EBAAAAzFYE21mislK6917pf/0v6ejR0ccvueQSdXR06Pnnn899cQAAAACQRQTbWeTCC6UPflC6/XbJffixkpIS3XTTTbrnnnvyUxwAAAAAZAnBdpb51Kekl1+WvvrV0cfuuOMO/eAHP9Cdd94pH5l8AQAAAKBIEWxnmfJy6StfkT7+cenQoeHHGhsb9eSTT+rb3/627rjjDsXj8fwUCQAAAAAziGA7C513nnTHHcGw5JEds6eddpp+8pOf6Pnnn9fNN9+sgYGBvNQIAAAAADOFYDtLfeITUlub9O//PvpYfX29/vu//1t9fX26+uqr1dXVlfsCAQAAAGCGEGxnqdLSYJbkv/iL4JrbkSorK/Xggw+qsbFRl19+uU6cOJH7IgEAAABgBhBsZ7FNm6Q/+zPpt35LSnc5bTgc1j333KO3vOUteuMb36iDBw/mvkgAAAAAmCaC7Sz3R38kRaPSF76Q/riZ6e/+7u/0O7/zO7rkkku0c+fO3BYIAAAAANMUzncByK5QSPryl6WLL5auvFI644z0533sYx9TQ0OD3vKWt+h73/ueLrzwwpzWCQAAAABTRY/tHHD66dJf/ZV0661SLDb2ebfccovuuecevetd79KWLVtyVyAAAAAATAPBdo748Ielqqrgmtvx7vDzrne9S9/5znd0yy236L777stdgQAAAAAwRQTbOaKkRPrqV6UdO6Szz5Z+8IPR97hNuuSSS/TYY4/pT/7kT/T5z38+t4UCAAAAwCRlNdia2ZVmtsvM9pjZJ9IcLzezbyaO/9LMVif2l5rZvWb2vJntNLNPZrPOuWLZMumHP5Q+9znp4x+X3vEO6YUX0p979tln68knn9TnP/95fepTn5KPlYIBAAAAIM+yFmzNLCTpLknvkLRJ0k1mtmnEaR+U1O7up0v6nKT/ndj/Pknl7n62pNdJ+t1k6MX0XXWV9NxzwWRSl14q/a//JbW1jT5v9erVevLJJ/Xwww/r93//9xUb7wJdAAAAAMiTbPbYbpa0x933ufuApPslXTvinGsl3ZtYf1DS5WZmklxStZmFJVVKGpB0Mou1zjllZdIf/mHQYxuJSBs2BLcEikaHn7d48WL9+Mc/1u7du3XjjTeqv78/L/UCAAAAwFiyGWyXS3olZftQYl/ac9w9KqlT0kIFIbdb0mFJByX9g7uP6lM0s9vNrNnMmo8fPz7z72AOaGiQ/vVfpccek77zHencc6Uf/Wj4OfPmzdMjjzwid9dVV12ll19+OT/FAgAAAEAahTp51GZJMUnLJK2R9DEzWzvyJHe/292b3L2poaEh1zXOKuecE4Tbz3xG+v3fl66+Wtq9e+h4RUWFvvnNb+qiiy7SBRdcoLe+9a36xje+od7e3vwVDQAAAADKbrBtkbQiZbsxsS/tOYlhx3WSWiXdLOm/3T3i7sck/UxSUxZrhSQz6d3vDoYnv/GN0utfH0wy1dERHA+FQvrMZz6jQ4cO6UMf+pC+/OUvq7GxUR/5yEe0bds2JpgCAAAAkBfZDLZbJa03szVmVibpRkkPjTjnIUm3Jtavl/T/PEhHByVdJklmVi3pIkkvZrFWpCgvl/70T6Vf/zoItRs2SHffLSXnjqqoqNANN9ygH/3oR3r66ae1ZMkSXX/99Tr//PP1L//yL2ptbc1r/QAAAADmlqwF28Q1sx+VtEXSTkkPuPsOM7vTzK5JnPYfkhaa2R5JfywpeUuguyTVmNkOBQH5P939uWzVivSWLpXuuUd65JHgHrive530k58MP2fVqlX61Kc+pb179+r//J//o1/+8pdat26dbrjhBm3ZsoWZlAEAAABknc2W4aNNTU3e3Nyc7zJmLXfpwQeDntx586TrrpOuv17aNPIGTpLa29t133336Utf+pKOHTum2267TbfddpvWrh11mTQAAAAAZMTMtrl72ktUC3XyKBQYM+l975P27AluC9TaKr397dLGjdJf/qW0fXsQfiVp/vz5+vCHP6zm5mZ9//vfV2dnpzZv3qzLLrtMX//615lwCgAAAMCMoscWUxaPS1u3St/6VtCkoCf3uuukzZuDMJzU39+v733ve/rSl76kJ598Uk1NTbrkkkv0hje8QRdffLHq6+vz8h4AAAAAFIfxemwJtpgR7tIzzwyF3K6uoZD7+tdLodDQuZ2dnXrqqaf0s5/9TE8++aS2bt2qdevW6Q1veMNg2F25cqUsNRkDAAAAmNMItsi5F14Irsn91rekY8ek97wnCLmXXiqFw8PPjUQi2r59u5588snBsFtWVjYYci+55BKdffbZCqWmYwAAAABzCsEWebVnTxBwH3xQ2r9feuc7pbe8JQi5q1ePPt/dtXfvXj355JODYffVV1/VRRddNBh2X//616uioiLXbwUAAABAnhBsUTAOHJAeflh64omglZcHATfZ1q0bfm1u0vHjx/Xzn/9cP/vZz/Q///M/evHFF/WOd7xD119/va688kpVVVXl/s0AAAAAyBmCLQqSu7R791DIfeKJYF9q0D3zzPRB98iRI/rud7+rBx98UFu3btXb3/52XX/99brqqqtUU1OT+zcDAAAAIKsItigK7tK+fdJPfjIUdPv7pTe9aSjobtoklYy4SdXx48f1ve99T9/61rf085//XJdffrmuu+46vetd71JdXV1e3gsAAACAmUWwRdHav394j25nZxB03/AG6cILpde9TqqsHDq/vb1dDz30kB588EE98cQTuvTSS3X99dfrmmuu0fz58/P2PgAAAABMD8EWs8ahQ0HAfeop6Ze/lHbskDZulC66KAi6F14orV8f9Op2dnbqBz/4gb71rW/p8ccf1+tf/3pdf/31uvbaa7Vo0aJ8vxUAAAAAk0CwxazV2ytt3y794hdB0P3lL6WTJ6XNm4fC7ubNUnl5lx555BE9+OCD2rJli84//3w1NTXp3HPP1TnnnKONGzeqrKws328HAAAAwBgItphTjhwJAm4y7DY3S0uXBiH3ooukc8/tU1vbT/TCC8/o2Wef1XPPPad9+/Zp/fr1g0E3uVy6dGm+3w4AAAAAEWwxx8Vi0gsvDPXo/uIXwSRVGzdK550nnXuutHFjv0pLX9C+fUNh99lnn1U4HB4MuvTuAgAAAPlDsAVG6O6Wnn9eeuaZofb880HP7nnnJQOv67TTjurYsaf1/PPPjerdPeuss3TGGWcMa8zCDAAAAGQHwRbIQCwmvfTS8LD77LPSwMBQ2D3vvKB3Nxb7tV56aad27949rNXU1IwKu2eccYbWrVun8vLyfL49AAAAoKgRbIFpOHIkCLipgffAAemMM4L76m7cmFy6qqsP6+WXd2vXrl3DAu+BAwe0bNkynXnmmcMC74YNG9TY2Cgzy/O7BAAAAAobwRaYYd3d0osvBtfuprZDh6S1a4Ogm9rWrIno8OH92r17KPTu2rVLL774orq6urRhwwZt2rRJGzduHGxr165VOBzO91sFAAAACgLBFsiR3l5p925p587hgffll6WVK4eCbrKX94wzpEikXTt37hzVDh8+rHXr1g0Luxs3btSZZ56pysrKfL9VAAAAIKcItkCeDQxIe/aM7uHds0eqrw8C7vr1wTK5ftppPTpwYPeowLtnzx4tW7ZMGzdu1Omnn67TTjtNS5cuHbZctGiRSkpK8v22AQAAgBlDsAUKVDwutbQEvby7dweTVyWXBw5Iy5aNDrxr10YVje7V7t079fLLL+vIkSM6fPjwsGVnZ6caGhrSht7U5dKlS+n9BQAAQFEg2AJFKBKR9u8fCrupwffoUWnNGmndumCI88qV0ooVQ+uLFg2ovf3YqMA7cnnkyBHV1dVpzZo1Wrt27bDlmjVrtGLFCq7zBQAAQEEg2AKzTG+vtHdv0F55RTp4cGh58GAQfBcvTh96k+sLF0rucR09elT79u3Tyy+/PGp59OhRLV++fFjYTQ3AixYtYkZnAAAA5ATBFphjolHp1VdHB97U7b6+IOQ2NgZDnpNt+fKh9QUL+nX06MG0offll19Wf3+/VqxYoYULF2rBggWDy/HavHnzCMMAAACYNIItgFG6uoKA++qrQ62lZfj24cPSvHnDw25q+J03r0vR6CGZnVBXV6va2tqGtdbW0ft6eno0f/78YWF38eLFWrp0qZYsWTJ47W9yvb6+niAMAAAAgi2AqYnHpRMnJg6/J05IFRXSokVBa2hIv1y0SKqvjygc7lAs1qqOjiD8Hj9+fPCa36NHjw6uHzlyRH19fWkDb+r6okWLVFFRofLycpWVlamsrEzl5eUqLS0lFAMAAMwSBFsAWeUunTwZBNzjx4Nl6nq6fV1d0oIFQ4F3wYKgLVw4fFlT0694/IQikSPq63tV7e2vjgrAJ06cUH9/vwYGBgaXAwMDikQiKi0tHRV4063X1NSorq5OdXV1qq+vH7ZMt6+6uprQDAAAkEPjBVumOwUwbWZSXV3Q1q3L7DGRiNTaGgTdtrZgvbV1aH3PnuR2uVpbl6utbblaW1+nUGh48F22THrNa4Zev75+aH3evLiqq6OqrBxQZWW/ysv7FY0OD7/9/f3q7+9Xd3e3Ojs71dHRoc7OTrW3t+vll19WZ2fnsP3J9f7+/sHQmwy8ySHW6Zap63V1ddxnGAAAYAYRbAHkRWmptHRp0DLlLvX0DA/AyWVnZ9AOH5Y6OpLbJersLFNHR5k6O2vU1SVVVw8Pv8n1efOk2tqhtmSJVFMzfF9qC4UiOnXq5GDg7ejoUHt7u9rb29XW1qb29nbt379/2HZy2dXVpXnz5o0Kv3V1daqtrVVtba1qamrSLlPXq6urFQqFxv2ZRaNRdXd3Z9zcXYsWLdKiRYvU0NAwuL5w4UKVlZVN548cAAAgawi2AIqGWRBMq6uDWxZNVjwunTqVGnyD1tERDKU+dSpora1D62O1aLRUNTULVVu7cDDsJmurqRlaP/304dvV1VJFRUxSl2KxTg0MdGhgoF19fSfU19em/v4OdXefUktLi06dOqWuri6dOnVq2Hpy2d3drcrKysGwW11dPdj7nGzRaFTV1dUZN0l6/vnndfz4cZ04cWKwtba2qrq6ejDojgy+ye2FCxequrpa5eXladtMX/fs7opEIoNDz5M98VVVVaqvr58w+AMAgNmBYAtgzigpGeqpna5IZCjkdnUFy+7uYL27e6gljx05kno8pO7uukRbOewx/f1SeblUVSVVVgbLZKuvD4ZeJ49VVrpKSyMKhwdk1qdQqE9VVSWaN69MdXVlqq8v17x5ZaqqssT5wSRfqetlZcEXBhOJx+Pq7OwcDLojg++LL744uN7T0zM4xHtki8Vig9c3jxd+U0PqWOsDAwOKRqMqLS1VaWnp4PXSpaWl6u3tVWdnp2prawdn3x45G3e6ljynvLx8+h8SAACQM1kNtmZ2paR/lhSSdI+7/92I4+WSviLpdZJaJd3g7vsTx86R9EVJ8yTFJV3g7n3ZrBcAMlVaOjTh1UyKx4N7DPf2BsOue3qGrw/fNvX0lKm3t0w9PTXq6ZHa26W9e4NzenuHnivddl+fFIsFATc18JaXB9vJZbBeooqK+Sovn6+KivXDjs+fL5122tBjkq2sLGjJ9fJyKRSKqaQkImlAUr/c++XeJ/d+xeN9GhjoVyQSGRZS060nt8frAU6G8ZG3nEoOCT906JCee+65tMdDoZBCoZBKSkrSNjMb81jyeGlpqSorK1VRUTHmcqJjybCfOtnZWMtwmO+qAQBzV9Z+C5pZSNJdkt4m6ZCkrWb2kLu/kHLaByW1u/vpZnajpP8t6QYzC0v6mqTfdPdnzWyhpEi2agWAQlFSMtRDu3Bh9l8vGg0Cbmrg7e8PWnJ/cn2sfZ2d0tGjw/cNDIxeBushDQyE1N9fMeqcSCT4wiAZgpPrU2mlpVJpaYlKS+cn2rqU/dLixcH9mJPbqS0cdsXj/SopiSkUcoVCrpKS+LBmFlNJSXzwWHI7uYzHY4pGo+rr61Nvb++Ey1OnTo3a19vbO2ySs9TJztLtM7NhveEjZ/4eq010TjgcVigUGlymrqfbN/K4mSkSiYxq0Wg07f5055SVlam6unpwyP1Ey+rqaiZoA4A5Jptf726WtMfd90mSmd0v6VpJqcH2WkmfTqw/KOkLFnz1foWk59z9WUly99Ys1gkAc1Y4HFwDXFOT70qCycFSQ24yDKdr4x1PPj61dXcHvdkj96dvpkikQtGoptxKSoKf7VBYnvx6RUVwTXY4LIVCQUtdH7ldUhKXFJN7VGbBUorIPSYpmtiOKh6PDC7dI4rHI4rHBxSPRxSJDKi/P6J4vF+xWETRaJ/co3KPJJZ9g48JlgPDHu8eUSzWn7JvQPF4VGVlpYM97KktHA6n3V9aWqqqqqrBcwYGBtTd3a3Dhw+ru7tbXV1dg9eRJ9dTlz09PaqoqBgMu1VVVRNe2z3R8XA4POYXB6nL8Y4le+Qn6smvrKxUeXn5lK5Hj8Vig18IpFuO/JknRz+EQiFuYQagqGUz2C6X9ErK9iFJF451jrtHzaxT0kJJZ0hyM9siqUHS/e7+2ZEvYGa3S7pdklZOZSYZAEDBMBsaxlzM3IMh3tFoEJSTy6msx2JDzzXReixWomi0RLFY6eCxdOdOZjnW48daT7cvHh8dxCe7HArvw7dLS4P7YC9ZMvxYSYkngngQ4GOxqEIhl9nQ8WA9WJaUuEpKUvdLZj64LClxSXHF49FEaB9axmJB0I9GI+rvH0jsG0h8MdCfst6rSKQjMdy+T/39fYkvEnoVifSrv79XAwPB/oGBPkUi/SorC6usLKyKijKVl5eppESKxQYUjQbPGTzvgKLR/sRyQFJcpaUlCodLVFoaUihkidELIYXDJYrFIoPPH41GBq9bj8fjw4LuyOCbbKlfGCSX6XrOx9sXiUSGTXKX+kVFujby+MDAgKqrqwcnzkudPT6T9YqKiglDvLtP2NKNPEjdnuiYu6f9kifderp9ZWVlGb0XYK4o1AtywpIukXSBpB5Jjyduxvt46knufrekuyWpqanJc14lAAAjmAWBLNnrOtelBv3pLOPx1BA/fHv0MUsE/FLF45XDzonHM1sfb18mx2Kx4L174n8nZkEQD4WGJoiLx4PjyedKtmCfKxp1xWJxxWLBejzucrfBFo8HLXgdS7yuDXuuSCQYxZBaU3IZhPrklwRBuDeTYjGXuxSJuPr6koF/qAVTn8TlHk+MDhje4vFY4kuFWMqXACcVj7cpFoskvjwoSWk1CofnKRwuSQxhD5ZBQA8pHA4lgnlymLupqyuqkycjg18cJAP/UOtTJNKraLR98MuESKRfAwO9isejo95HMNrBJcUkJY+5pOR7Hr4MvviwREu+H0vUrpT3YQqHTaFQ8GVDajPzRO3RxJcUwZcl0Whk8IuL5P5YLJp4X0NfaEQiA3KPqb6+TvX1dZo/f2g5f369Fiyo14IF87VgQX1ie74WLpyv+vo6lZaGBv/8k3/uyfXxtgspQ8fj8cFJBUtKSgYvnSDoz13ZDLYtklakbDcm9qU751Diuto6BZNIHZL0U3c/IUlm9oik10p6XAAAoGikBn1MhiVadq4VTobuoYCeGohN6UJ/ajhOdyxdG+vxqa+f6Xa6LwIm+2VDPC4NDMQGw3vwJcHQFwPS8GW61039cmCi9z/Wlxcjf5ZjLSc6Fo/HFYnEFI3GdPhwTK+8ElcsFlc0Gh/8UiQWiydeL654vF/ux2VWIrOwzEokBV+UJD9vwXrq/pKUJg0P/0NfAEgaDP/BFwBK2a9h+0Yuh54reL7gEorklw7xxJcmqesxpX7JEIyoCL6gCPalBvLklzaWMjrDEi05QmNoe+jcoWXq+WaWGPlhw1rw5cXwL2KGlmGVloYH18vKwokRFcGoCjPNSBv5JcTILymS6+7Bz3HoS5RoYiRIVL/1Ww1aurQ4v5XN5q+ZrZLWm9kaBQH2Rkk3jzjnIUm3SnpK0vWS/p+7J4cg/6mZVSmYOvNSSZ/LYq0AAABzRup/hOee2XR/62TgLM34EbFYTO3t7WptPa7u7m6VlJSMOQv8yP1BEC6RWUhmJYnREXFFo7HBFolEFYvFFYlEFY3GFIvFBtdTW/KcWMxVWlqmUCisUKhU4fDQetAqRmyHVVISVjBPrY34siWugYHoYItEourvjygSiQ1uB8uh7UgkNvjlQDwefCkQjcYVj3vimvWhLwiSLTgv+aWBJ84JnrOrK5LyOskW0cBA8vr3odeMx13hcKlKS8sVDpeqpCSU+NIh+fMOjVi3wX3Dzwv+XNwtUWNs8MuNoM5Y4j0EtbpLJSXhwZ9l8HMNqaQkrEsuOUtLl67Jzsc1y7IWbBPXzH5U0hYF/4J8yd13mNmdkprd/SFJ/yHpq2a2R1KbgvArd283s39UEI5d0iPu/nC2agUAAADmglAopEWLFmnRokX5LiULSiSVJVrhi8Viw2bEj0ajisWCgB2E0eFtov1mNjhRXXIyu3TrZWVls3LItnny4o8i19TU5M3NzfkuAwAAAACQBYl5l5rSHZuTA1AAAAAAALMHwRYAAAAAUNQItgAAAACAokawBQAAAAAUNYItAAAAAKCoEWwBAAAAAEWNYAsAAAAAKGoEWwAAAABAUSPYAgAAAACKGsEWAAAAAFDUCLYAAAAAgKJm7p7vGmaEmR2XdCCLL7FI0oksPj8wFXwuUYj4XKLQ8JlEIeJziUJU6J/LVe7ekO7ArAm22WZmze7elO86gFR8LlGI+Fyi0PCZRCHic4lCVMyfS4YiAwAAAACKGsEWAAAAAFDUCLaZuzvfBQBp8LlEIeJziULDZxKFiM8lClHRfi65xhYAAAAAUNTosQUAAAAAFDWCbQbM7Eoz22Vme8zsE/muB3OTmX3JzI6Z2a9T9i0ws0fN7KXEcn4+a8TcYmYrzOzHZvaCme0wsz9I7Odzibwxswoz+5WZPZv4XP51Yv8aM/tl4nf5N82sLN+1Ym4xs5CZbTezHyS2+Uwir8xsv5k9b2bPmFlzYl/R/g4n2E7AzEKS7pL0DkmbJN1kZpvyWxXmqC9LunLEvk9Ietzd10t6PLEN5EpU0sfcfZOkiyR9JPHvI59L5FO/pMvc/VxJ50m60swukvS/JX3O3U+X1C7pg/krEXPUH0jambLNZxKF4C3ufl7KLX6K9nc4wXZimyXtcfd97j4g6X5J1+a5JsxB7v5TSW0jdl8r6d7E+r2S3p3LmjC3ufthd386sX5KwX/YlovPJfLIA12JzdJEc0mXSXowsZ/PJXLKzBolvVPSPYltE59JFKai/R1OsJ3YckmvpGwfSuwDCsESdz+cWD8iaUk+i8HcZWarJZ0v6Zfic4k8Swz5fEbSMUmPStorqcPdo4lT+F2OXPsnSX8qKZ7YXig+k8g/l/QjM9tmZrcn9hXt7/BwvgsAMDPc3c2Mac6Rc2ZWI+lbkv7Q3U8GHREBPpfIB3ePSTrPzOolfUfShvxWhLnMzN4l6Zi7bzOzN+e5HCDVJe7eYmaLJT1qZi+mHiy23+H02E6sRdKKlO3GxD6gEBw1s9MkKbE8lud6MMeYWamCUPt1d/92YjefSxQEd++Q9GNJF0uqN7PkF/r8LkcuvUHSNWa2X8ElbZdJ+mfxmUSeuXtLYnlMwZeAm1XEv8MJthPbKml9Yua6Mkk3SnoozzUBSQ9JujWxfquk7+WxFswxiWvE/kPSTnf/x5RDfC6RN2bWkOiplZlVSnqbguu/fyzp+sRpfC6RM+7+SXdvdPfVCv4f+f/c/f3iM4k8MrNqM6tNrku6QtKvVcS/w829aHqX88bMrlJwbURI0pfc/TP5rQhzkZndJ+nNkhZJOirpryR9V9IDklZKOiDpN9x95ARTQFaY2SWS/kfS8xq6buzPFVxny+cSeWFm5yiY8CSk4Av8B9z9TjNbq6C3bIGk7ZI+4O79+asUc1FiKPLH3f1dfCaRT4nP33cSm2FJ33D3z5jZQhXp73CCLQAAAACgqDEUGQAAAABQ1Ai2AAAAAICiRrAFAAAAABQ1gi0AAAAAoKgRbAEAAAAARY1gCwAAAAAoagRbAAAAAEBRI9gCAAAAAIoawRYAAAAAUNQItgAAAACAokawBQAAAAAUNYItAAAAAKCoEWwBAAAAAEWNYAsAAAAAKGoEWwAAAABAUSPYAgAAAACKGsEWAAAAAFDUCLYAAAAAgKJGsAUAAAAAFDWCLQAAAACgqBFsAQAAAABFjWALAAAAAChqBFsAAAAAQFEj2AIAAAAAihrBFgAAAABQ1Ai2AAAAAICiRrAFAAAAABQ1gi0AAAAAoKgRbAEAAAAARY1gCwAAAAAoagRbAAAAAEBRI9gCAAAAAIoawRYAAAAAUNQItgAAAACAokawBQAAAAAUNYItAAAAAKCoEWwBAAAAAEWNYAsAAAAAKGoEWwAAAABAUSPYAgAAAACKWjjfBcyURYsW+erVq/NdBgAAAAAgC7Zt23bC3RvSHZs1wXb16tVqbm7OdxkAAAAAgCwwswNjHWMoMgAAAACgqBFsAQAAAABFjWALAAAAAChqBFsAAAAAQFEj2AIAAAAAihrBFgAAAABQ1Ai2AAAAAICiRrAFAAAAABQ1gi0AAAAAoKgRbAEAAAAARS2c7wIAoNhEo9KuXdIzzwy1PXukxkbpjDOCduaZwXLdOqm8PM8FAwAAzHIEWwB59cgjUmWl9OY3S2b5rma0ri7pueeGAuz27dKOHUGIPe+8oP3RH0nr10stLdLu3UF78slgeeCAtGzZ6MB7xhnSihVSCeNmAAAAps3cPd81zIimpiZvbm7OdxkAJuGZZ6S3vU1askQaGJBuv1267TZp0aLc1+IuHTkyvBd2+3bp0CHprLOCAHv++cHy7LOl2trMnjcSkfbvD0Lurl1DwXf3bqmtLejRTQbd886T3vjGIAgDAABgODPb5u5NaY8RbAHkQ2+v1NQkffKT0vvfLz31lHT33dJ3vyu94x1ByM12L+7hw9LDD0s/+EHw+rHYUHhNtjPPlMJZGtvS1SW99NJQ6H36ael//kdasEB605ukSy8NlqtWFWZvNgAAQC4VXLA1sysl/bOkkKR73P3vRhz/nKS3JDarJC129/rxnpNgCxSXP/zDIFjef//w0NbeLn3ta9IXvzjzvbjxeNAL+/3vB2F23z7p7W+X3vWuIEQuX57/ABmPB0Odf/pT6YkngmVZ2fCge8YZ+a8TAAAg1woq2JpZSNJuSW+TdEjSVkk3ufsLY5x/h6Tz3f23x3tegi1QPB59VPrt35aefTbonUzHfXQv7u/+bhDuJhPqurulxx4LguzDD0vz5gVB9uqrpde/XiotnZG3lDXuQa9uMug+8YTU3z886L7mNbPjWl33oOf6wIGg53zx4nxXBAAACkmhBduLJX3a3d+e2P6kJLn7345x/s8l/ZW7Pzre8xJsgeLQ2iqde650773S5Zdn9pjUXtxIJOjFvfXWsXtxDxwIguwPfiD97GfS5s1BmH3nO4NJnordgQPDe3RPnJAuuSRoS5cG1//W1KRflpXlu/ohkUjQg/7kk8EQ7CeflKqrpTVrgv0LF0oXXSRdeGGwPO+8wqofAADkVqEF2+slXenuv5PY/k1JF7r7R9Ocu0rSLyQ1unsszfHbJd0uSStXrnzdgQMHslo7gOlxl973PmnlSukf/3Fqj0/Xi3vJJdIvfzkUZo8ela66Kgizb3ubVFc342+loBw+HATDn/88+OKgq0s6dWr08tSpoLd7rNBbUxOEyTVrpLVrg7ZmjVRVNTN1dnVJv/jFUJD91a+C10iG8ksuCWaKloIh2bt2BX+uv/hF0F56KfhS5KKLhtqKFQzLBgBgrijmYPtnCkLtHRM9Lz22QOG7917pH/5B2rpVqqiY3nOl9uLu2RNM8vSudwVt82YpFJqZmmeb/v7hgXdk+D1xQnr55eD64337ghmd6+qGB93U9eXLx/5ZHz0a9Jgne2N37gyGGF9ySTD788UXS/PnZ157V5fU3DwUdH/xiyDUpgbdpqag1xcAAMw+hRZsMx6KbGbbJX3E3X8+0fMSbIHC9vLLQeB8/HHpnHNm7nndg5A71rW6mJ54PLgNUjLopobel18OgvDKlUOBd+XK4IuGJ5+Ujh0LrmNOBtmmpul/oZHKXTp4cHjQfe65YLj56acHQ9XHa9XV9PYCAFBMCi3YhhVMHnW5pBYFk0fd7O47Rpy3QdJ/S1rjGRRJsAUKVywWTHT0nvdIH/tYvqvBTOrrC3p1k4H3wAFp9eogzOZjUqv+/uAexAcOBKH7xIlgeHZyPbXFYsOD7sKFQ+vz5wcTi5WUBD3SodDQeqb71qwJ7lMMAABmxnjBNkt3Zxybu0fN7KOStii43c+X3H2Hmd0pqdndH0qceqOk+zMJtQAK22c/G0z680d/lO9KMNMqKqQNG4JWCMrLg8mmLrxw4nN7eoaH3tT1ffukaDTosY7Fhpap6xPt27lTqqyUrrgiuK3UZZcFs3IDAICZl5f72GYDPbZAYdq2LZjkadu2oYmBgLnAXfr1r6UtW6Qf/SiY+Oz884eC7mtfW7jXgsdiwTXX3d1BD3Z5eb4rAgCgwHpsAcwdPT3SBz4g/cu/EGox95hJZ58dtI9/PPj78NOfBkH31luDa5Df+tYg5F5xRTAR10zo7Q2uiz58OOh9Ts6Ina6lzpid2vr7g1myq6qktrbgNlJr1wZDq0e2+vqZqRsAgOmgxxZA1nz0o8HETl//er4rSa+3t1d79uzR7t27tXr1ar32ta+VMZsQcuSVV4Ke3C1bpMcek5YtC0Lu298eTLZVWTl0rrvU0RGE1cOHh4Jruu3e3iCInnZa0NtaWzv5VlU1NLFWNBpM0rVvn7R37+hWVjZ26F22LPfXWQPFxj0YJdHfP9QGBibeHhgI5gKoqAhGVWSyzOcokWg0mFzw+eeD0SzPPx+0gweDusLh4P2kLtPtS7esqQkmkVywIJgjYaz1mZzAEPlRUJNHZQvBFigsP/yh9Hu/Jz37bH57dGKxmF555RXt2rVLu3fv1u7duwfXjxw5orVr1+r000/Xjh07VFFRoVtvvVUf+MAHtGzZsvwVjTknFgtug5UMus89J73udcF/YJPBtawsCKvJlgyvI/ctWJC72Z7dpePHg4CbLvh2dASTia1aFSxT11etCuol+GI2S/4defHF4N7cqe3YsaGgWlIShM/y8uDverr1kdulpVIkEjy+r2/0Mt2+kpIg3CXD7rx5wWiRxsahZer6okWT/zvqLrW0jA6wu3YF/04lR7KcfXYwyeCaNcFjotHg/Ux2GYkEI03a24MRJsnlyPW2tiAIpwu+dXXD3+fIf0NTt8daH/kzmGg93bHU95RufaLjUvAZSW3Jz026lu7Yb/928OdeqAi2AHLq+HHpvPOCnto3vzn7r+fuam1tHRZak+t79+7VokWLdMYZZ+jMM88ctly1apXC4fDgczz55JO699579e1vf1ubN2/Wrbfeqne/+92qTO06A3KgoyO4fVFNzVBgLcb783Z3BzNmHzgQtP37g5Zc7+wMLlNIF3pXrw56fAv1OuRi098f/HmEQkPBqBAHqMRiQRgZOZt56nby2u/Fi6UlS4Jlaqury/176+sLeiNHhtddu4LAdOaZw9uGDcHf7WSwyMXnPBmcUoNuR0cQQltapEOHhpbJ9VOngr+HY4XfpUulV18dCq/JMFtWNhRckyF206bg37R8cQ8uCUkXfDs7h0LmROFzrPWphuHU7XS90eP1VI/cJwUBN9mjn2ypvfzjtf7+4O4VS5dm9jPNB4ItgJxxl9773uBeop/9bLZfy3XPPffoz//8zxWJREYF1zPOOEPr169X9SQTQU9Pj7773e/q3nvv1datW3Xdddfp1ltv1Rve8AaGKgMzqKcnGIaYLvQmb9m0bFkw6iO112qs3qyxjoXDwX/ie3uHWk/P8O10LfUcs6GertShnRO11KGg4fD0biEVjQbXRSdb8jrpifZ1dQX/NldXDw15jUSC/wyn/sySdY7XysqC95EcOpqsceS+8c45dWrs0NrREQTTkbfgSl2vqgrOPXZseDt6dKgXdGTYHRmAKyqGz2Y+cn28Y7HY6CDb0hJ8GZMuwBZy79dEenuD4JoaelPD7+HD6XthFy/Od+WYrQi2AHLmS18KJov65S+zO5Nqe3u7br/9du3evVtf/epXdfbZZ2cldLa0tOhrX/ua7r33Xg0MDOiWW27RLbfcotWrV8/4awEYrr8/uBa5szP9dYYj21jHo9EgyFRWBqGosnJ4S7dv5H5paHjnyGGembbJ3jZq5LFQKLgGuqYmfRvv2Mh/j+PxiX+eY/2MY7HgZ5q6TLdvrHPmzRsdVpPb8+cP9TxNVW9vMHIoGXTThd+BgeALg9QvEtJtj7VeVhZcR54MsGvXBl8UAMgugi2AnNi7V7roIuknP5HOOit7r/Pzn/9cN998s6655hp99rOfVUUOZoNwdzU3N+vee+/VN7/5TZ111lm67bbbdN1116m2tjbrrw8AADDXEWwBZF00GszkeuON0h/8QXZeIxaL6W//9m/1hS98Qf/+7/+uq6++OjsvNIH+/n49/PDDuvfee/XEE0/ommuu0S233KK3vOUtCuXpgsBoNKrt27fr/PPPH7xuGAAAzH7Hjh3Ttm3b1NraqtNOO03Lli3TsmXLNG/evFl3CRX3sQWQdX/7t8FwtzvuyM7zHzp0SB/4wAdkZtq2bZuWz9RNP6egvLxc733ve/Xe975Xx44d03333ac//dM/1dGjR3XzzTfrAx/4gM4555ys/zJxd23fvl1f/epXdd9996m0tFTLli3Tvffeqw0bNmT1tQEAyDd3169+9Sv19/dr7dq1WrZsmUpm+VTrbW1t2rZtm5qbmwfbyZMn1dTUpMWLF+vw4cN69dVX9eqrryoajQ6G3NTAO3K7trZ2VgRgemwBTNuvfiVdfbX09NPBLIkz7Xvf+97/n737DpOyuvswfh+6ICodBBFpAoJUAWtsKGrURI1RUbFXjDGJMcYkbzQx0cQUNSZiR0UFxYIidixIEZAqHZSuIB2pu3veP2ZBQMAtMzszu/fnup5rZ54555nfeo3sfveUh6uuuoobb7yRW265JW2jot9n6tSp9O/fn/79+1O9enV69erFBRdcQOPGjZP6PvPnz6d///489dRTbNy4kQsvvJALL7yQFi1a8OCDD/KHP/yBW2+9lRtvvDFj/1tJklRU06dP3/bztnLlytSsWZPPP/+cFStWcOCBB9K0aVOaNm3KQQcdtMPjfffdN92lF8qaNWv49NNPGTt2LGPGjGHs2LEsW7aMTp060aVLl21Hs2bNdhlM161bt0PQXbx48XeeL168GGBb0O3Xr19G7yPiVGRJKfPNN9CxI9x5J/zkJ8m99oYNG7j55psZMmQIzzzzDIcffnhy3yBF8vLyGDFiBE8//TQvvPACbdu25cILL+Scc85hvyLe1Hf16tW88MILPPXUU0yePJmf/OQnXHTRRRxxxBHf+WE2d+5cLr30UvLy8njiiSdo1qxZEr4rSZLSZ8mSJTz33HP079+fxYsXc/7553PhhRfSoUOHbT8H169fzxdffMHcuXP5/PPPmTt37g6PK1euvMvA27hxY6pXr07VqlWpWrUqFStWLNERzBgj69atY+LEiTuMxC5cuJD27dtvC7CHHXYYLVu2TPqo9Nq1a7eF3K5duxb6bhIlyWArKWWuuSaxA2W/fsm97tSpUznvvPNo3bo1ffv2LXIgTLdNmzYxdOhQ+vfvz1tvvcWJJ57IhRdeyKmnnkrl79k2esuWLbzxxhs89dRTvPnmm5xwwglcdNFFBeqbl5fHfffdx5///Gduv/12rr322pRPz1qxYgVPPvkkVapUoUOHDrRr1y6jfzhKKlkxRqZMmULdunWpV69eustRFli7di0vvfQS/fv355NPPuFHP/oRvXr1KtKeFjFGli1btsvAO3/+fL755hvWr1/PN998A0DVqlWpVq3atrC7u2NrG4CNGzeyYcMGNm7c+J3H3/dalSpVaNu27bYA26VLF1q3bu2+GTsx2EpKiREj4IILYOLExH0HkyHGyMMPP8xtt93GXXfdxWWXXVYq1n0ArFq1ikGDBvH0008zadIkzj77bC688EKOOuqobaFz63qhp59+mgEDBtCyZUsuuugifvKTn1CzZs1Cv+eMGTPo3bs31apV47HHHuPAAw9M9rfFnDlz+Ne//sUzzzzDaaedRqVKlZgwYQLTpk3jwAMPpEOHDjsc/kKrbLFx40aGDRvGa6+9xooVK6hfvz7169enXr162x7Xr1+fOnXqOO1/N3Jzcxk5ciSDBg3ixRdfpFy5cqxcuZIDDzyQE088kR49enD00Uf7R7AsFGNk6dKlVK5cmX333TdpP6u3bNnCW2+9xdNPP83QoUM55phj6NWrF6effvq2AJlqW7ZsYf369Xs8tgbhrY9DCFSpUoUqVaqw1157FepxlSpVSv3a4GQx2EpKiT59YP/94be/Tc71Vq5cyZVXXsns2bN57rnnSvUGSAsWLODZZ5/l6aefZvXq1VxwwQVUrVqVp59+mhgjF110Eb169aJp06bFfq+cnBzuuece/vGPf/DXv/6Vyy+/PCm/gIwYMYJ77rmHjz76iCuvvJI+ffqw//77b3t9y5YtTJ8+nfHjxzNhwoRtR+XKlbeF3I4dO9KhQweaN2/uD/UslpOTw5o1a7Ydq1ev3va4cuXKHHvssUX6w0w6fPnllwwZMoRXX32VYcOGceihh3L66afTsGFDvvrqK7788ku+/PLLHR6vWLGCmjVr7jL4bn3csGHDlEwhzERbtmxh2LBhvPjii7z88svUq1dv24Z7bdu2JTc3lzFjxvD222/zzjvv8Omnn3LYYYdtC7qdO3fO6D8UbNy4kZkzZ1KzZk3q1atHxRTfwDYvL49FixYxe/Zs5syZw5w5c5g9ezYLFy6kSZMmtG/fnvbt29OhQwfq16+fsj8GL126dNsU2a3rPTdt2kROTg65ubk0bNhwj0f9+vV3+98qxsioUaPo378/AwcOpEWLFvTq1Ytzzz2X2rVrp+T7UXYy2EpKurw8OOAAeO+9xM3pi2v48OH06tWLH//4x9x1110lcm/aTDFp0iT69+/Phg0b6NWrF127dk3JLyZTpkzh4osvpn79+jzyyCM7hNCCys3N5aWXXuIf//gHS5cu5aabbuLSSy8t8GhLjJEFCxZsC7lbQ++yZcs49NBD6dChA926dePHP/4x++yzT6HrK4u2jprMmjWLWbNmMXPmzG2Ply9fTqVKlahcufK24/ueb3+uUqVKbNiwYYeguqvHGzduZJ999tl27Lvvvtser1mzhuHDh9OuXTtOOeUUTjnlFDp27JgxAS/GyKRJk3j11Vd59dVXmTlzJieddBI//OEPOfXUU6lVq9b3XiMnJ4evv/56W9DdOfh++eWXzJs3j7Vr13LcccdxwgkncMIJJ+x2w5dstGHDBt566y0GDRrEa6+9xsEHH8xZZ53Fj3/8Y5o3b77HvmvXruXDDz/knXfe4e2332bx4sUcd9xx9OjRgxNPPDHt/52+/vprRowYwccff8zw4cOZMGECjRs3ZvXq1SxbtoxatWrRsGHDHXac3fmoU6fOHj/zmzdvZt68edvC6/Yh9vPPP2e//fajefPmNGvWbNvXRo0a8fnnnzNhwgQmTpzIxIkTKVeu3LaQuzXwtmrVqtDhe+XKldt23t0aYlevXr3DWs8uXbrQuHFjQgisXbuWRYsW7fHY/r/V9seGDRt47rnnqFChAhdeeCEXXHBBUv6oq9LJYCsp6UaOhCuvhClTined3Nxc7rzzTv773//yyCOP8MMf/jA5BWqXtmzZwl/+8hceeOAB/vnPf9KrV68C/cK4bt06HnvsMf7973/ToEEDfvnLX3LmmWcmbVRl1apVTJw4kQkTJjBs2DCGDRvG6aefziWXXJKW+wOvX7++xKa8FdSKFSu+E1y3Pq9YsSItW7akRYsW246WLVtSp04dNm/ezKZNm7Z93f7Y+dyunletWnWXgXX7x9WqVdvj52jjxo18+OGHDB06lKFDh7Jy5Up69uxJz549OemkkwoUHpNp48aNvP/++7z66qu89tprVKhQgdNPP53TTz+do48+mkqVKqXkfRcsWMC777677ahYsSInnHACJ554Iscffzz169dPyftCYtRv2bJlVK5cmX322Scpf1hYs2YNQ4YM4cUXX+Stt96iS5cunHXWWfzoRz8q1i3ZlixZwjvvvLMt6FauXHnbaO7xxx+f0hG8GCNz5sxh+PDh24Ls4sWL6datG0cddRRHHXUUXbt2Ze+99wYSf9RYunTpd3aZXbx4MYsWLdr2ePXq1dSvX3+HsJuTk7MtxC5evJiGDRvuEFy3Pm7atGmB/ngYY2Tx4sU7BN2JEycyf/58WrVq9Z3AW6NGDSDx7/vOO+9++eWXdOzYcYcQ26xZs2J9bnJycvjqq6++E3hjjJxzzjl07ty51PyhR6ljsJWUdL/6FVSrBrffXrzrXHzxxSxcuJCnn366SCOIKppPP/2Uiy++eNstgna37nXx4sXcf//9PPzwwxx33HH88pe/pHv37imvb9myZTz77LM88cQTLFu2jIsvvpjevXvTsmXLlLzfli1b+PjjjxkyZAhDhgxh1qxZ1K1bd9tU6Y4dO9KxY0cOOuiglP7itWHDBmbNmsX06dOZMWPGDkE2Jydnh9C6fYjNlmm+W33++efbQu4HH3zAIYccsm00t3PnzikZzV26dOm2Kcbvvvsu7dq12xZmW7duXeK/UMcYmT59Ou+88w7vvvsuH3zwAQ0bNtw2mvuDH/yg0LcmycnJYf78+TuM+m19PHfuXKpWrcrmzZtZv349NWrUoGbNmt85atWqtdvz++67L8uXL2fw4MG8+OKLfPTRRxxzzDGcddZZnHHGGSkJnDFGpk2btm3a8ocffkijRo1o2LDhtk2o6tWr953HdevWLdAfKLZs2cL48eN3CLKVKlXiqKOO4sgjj+TII4+kXbt2xd7AZ9OmTXz55Zc7hN5y5cptC7EHHnhgyv6g8s033zBlypRtQXfChAlMnjyZ/fbbj7333pt58+bRrl27HUJsq1atMno6uMoug62kpIoRmjaFl1+G9u2Lfp2FCxdy6KGHMn/+/G1//VbJ2bRpE3/84x95/PHHuf/++/nJdvdrmjRpEv/4xz949dVXufDCC/n5z3+etqlhkyZNol+/fvTv35+mTZtyySWXcO655xZ7p+ylS5cydOhQhgwZwttvv02zZs047bTTOO200+jcuTMLFixg/Pjx26ZLjx8/nrVr124b9dgadlu3bl2oX0hjjCxZsmRbeJ0xY8a2x0uWLKFZs2YcfPDBHHzwwdsC7NbR19I4mrFx40Y++uijbUF3+fLlnHzyyZx66ql7HM3dtGkTK1euZNWqVaxatWqPj+fNm8fMmTPp0aMHp59+OqeeemrGrdvLycnh008/3TaaO3r0aNq2bbst6B5++OFUqVKFjRs3Mnfu3B2mq279On/+fOrXr/+dKatbR/2qV68OJMLcypUrWbFixQ7H8uXLv3Nu+2Pt2rVUq1aNk08+mbPOOotTTz21xJcMbNmyhc8++4yvvvpq27F06dLvPF+2bBnVq1ffbehdsGABH3/8MWPGjKFZs2bbguxRRx2V9HuPZ6K8vDw+//xz1q5dyyGHHJLydcJSshhsJSXVp5/CuefCrFlQnN+zb7/9dr766iv++9//Jq84FdqoUaPo3bs3HTt25Pzzz+c///kPU6dO5YYbbuDqq6/eNl0t3bZs2cKbb77JE088wTvvvMOpp55K7969OfHEEws0spCXl8f48eO3jcrOmDGDE044gdNOO41TTjmFBg0afO81vv766x3WBo8fP54vvviCVq1a7TC6e+ihh1KxYkVmzZr1nfA6Y8YMqlSpQqtWrTj44IN3+NqkSZMyf2uHL774YofR3NatW1OzZs3vhNbc3Fxq1KhBjRo12G+//bYdOz/fb7/9aNCgAd27d0/ZiFgqbNy4kREjRvDuu+/yzjvvMHXqVPbbbz+WLVvGgQce+J3g2qxZMw466KDvvRVYceTk5BBjzIoQlJeXx4oVK3YIvds/rl+/PkcddRTdu3fP2tvJSWWRwVZSUv3ud7BlC9x9d9GvkZOTQ5MmTRgyZAjtizPsq6RYv349v/vd7xg5ciTXXnst5513XkaHgOXLl/Pcc8/xxBNPsGTJEi666CJ69+79nZ20165dy9tvv82QIUN4/fXX2WeffbaNyiZrLeX69euZPHnyDqO7U6ZMITc3l4MOOug7Afbggw/OuqnD6bJp0yZGjBjBhg0bvhNcq1SpUipHsHdna7A/4IADnCIqqcwy2EpKqtatoV8/6Nq16NcYPHgwd911FyNGjEheYSqTPvvsM/r168dTTz3FgQceSO/evdm0aRNDhgxh1KhRHH744dvC7Pftzpos2TSyJUlStjDYSkqaqVPh5JNh3jwozv4up556Kj/96U/p3bt38opTmZaTk8Pbb7/N008/TdWqVTnttNM48cQTXb8tSVIpsadgW7YX8kgqtEGD4Kyzihdqv/jiCz755BMGDRqUvMJU5lWoUGHbzrqSJKlsyYy7o0vKGi++CGefXbxrPPzww1x00UXstddeySlKkiRJZZojtpIKbO5cWLwYjjyy6NfYvHkzjz76KMOGDUteYZIkSSrTHLGVVGCDBsGPfgTF2ZDzlVdeoVWrVrRu3TppdUmSJKlsS0uwDSH0DCHMCCHMDiH8Zjdtzg0hTA0hfBZCeKaka5T0XYMGFX8act++fbn66quTU5AkSZJEGqYihxDKAw8APYCFwJgQwuAY49Tt2rQAbgWOjDGuDCHULek6Je1o4UKYNQuOO67o15g5cyaTJk3irLPOSl5hkiRJKvPSMWLbFZgdY5wbY9wMPAecuVObK4EHYowrAWKMS0u4Rkk7eeklOP10KM5tOR966CEuvfRSKleunLzCJEmSVOalY/OohsCC7Z4vBLrt1KYlQAjhY6A88McY4xslU56kXRk0CH7xi6L337hxI/369WPkyJHJK0qSJEkic3dFrgC0AI4FGgEfhhDaxRhXbd8ohHAVcBVA48aNS7hEqexYuhQmTICTTir6NQYNGkTHjh1p3rx50uqSJEmSID1TkRcBB2z3vFH+ue0tBAbHGLfEGD8HZpIIujuIMT4UY+wSY+xSp06dlBUslXWvvAI9e0KVKkW/Rt++fbnmmmuSV5QkSZKULx3BdgzQIoRwUAihEnAeMHinNi+TGK0lhFCbxNTkuSVYo6TtDBoExdnv6bPPPmP27NmcfvrpyStKkiRJylfiwTbGmAP0Ad4EpgEDY4yfhRDuCCGckd/sTWB5CGEqMAy4Oca4vKRrlQQrV8KIEXDqqUW/Rt++fbn88supWJydpyRJkqTdSMsa2xjj68DrO537w3aPI/CL/ENSGr32Ghx/POy9d9H6r1+/nv79+zN+/PjkFiZJkiTlS8dUZElZZNAgOPvsovcfMGAAhx9+uBu8SZIkKWUMtpJ2a906eO89+OEPi36NBx980E2jJEmSlFIGW0m79frrcMQRUKNG0fqPHz+eJUuWcMoppyS3MEmSJGk7BltJu/Xii8Wbhty3b1+uvPJKypcvn7yiJEmSpJ2kZfMoSZlv40Z44w24776i9V+7di0DBw5kypQpyS1MkiRJ2okjtpJ26a23oEMHqFu3aP2feeYZjjvuOPbff/+k1iVJkiTtzGAraZeKsxtyjJEHH3yQq6++OrlFSZIkSbtgsJX0HVu2JO5fe9ZZRes/ZswY1qxZw4knnpjcwiRJkqRdMNhK+o5hw6BlS2jYsGj9t47WlivnPzGSJElKPTePkvQdgwYVfbR21apVvPjii8ycOTO5RUmSJEm7YbCVtIPcXHj5ZRg5smj9n3rqKU455RTqFnXXKUmSJKmQnCcoaQcffwz77w9Nmxa+79ZNo6655prkFyZJkiTthsFW0g6Ksxvy8OHDycvL45hjjkluUZIkSdIeOBVZ0jZ5efDii/Dmm0Xr37dvX66++mpCCMktTJIkSdoDR2wlbTNmDOy9N7RpU/i+X3/9Na+99hoXX3xx8guTJEmS9sBgK2mbF18s+jTkfv36ceaZZ1KzZs3kFiVJkiR9D6ciSwIgxsT62uefL3zfvLw8+vbtS79+/ZJfmCRJkvQ9HLGVBMCkSYlb/XToUPi+w4YNo0qVKnTv3j3pdUmSJEnfx2ArCfh2N+Si7PvUt29frrnmGjeNkiRJUloYbCUBRV9f++WXX/L2229z4YUXJr8oSZIkqQAMtpKYMQNWroRu3Qrf97HHHuOcc85hn332SX5hkiRJUgG4eZQkBg2CH/8YyhXyT125ubk8/PDDvPDCC6kpTJIkSSoAR2wlbVtfW1gffvghNWrUoHPnzskvSpIkSSogg61Uxn3xBcyfD0cfXfi+H330ET169Eh6TZIkSVJhGGylMu7FF+HMM6FCERYmjBw5ksMPPzz5RUmSJEmFYLCVyriiTkPOy8tj9OjRBltJkiSlncFWKsMWL4Zp0+CEEwrfd8aMGey3337Uq1cv+YVJkiRJhZCWYBtC6BlCmBFCmB1C+M0uXr8khLAshDAh/7giHXVKpd3LL8Npp0GlSoXv6zRkSZIkZYoSv91PCKE88ADQA1gIjAkhDI4xTt2p6YAYY5+Srk8qS158Ea6/vmh9R44cSffu3ZNbkCRJklQE6Rix7QrMjjHOjTFuBp4DzkxDHVKZlpsLo0bB8ccXrb8jtpIkScoU6Qi2DYEF2z1fmH9uZ2eHECaFEF4IIRxQMqVJZcf06dCgAey7b+H7rl69mi+++IL27dsnvzBJkiSpkDJ186hXgSYxxkOBt4F+u2oUQrgqhDA2hDB22bJlJVqglO3GjYPOnYvWd/To0XTq1ImKFSsmtyhJkiSpCNIRbBcB24/ANso/t02McXmMcVP+00eAXf76HWN8KMbYJcbYpU6dOikpViqtihNsnYYsSZKkTJKOYDsGaBFCOCiEUAk4Dxi8fYMQQoPtnp4BTCvB+qQywWArSZKk0qLEg22MMQfoA7xJIrAOjDF+FkK4I4RwRn6zn4UQPgshTAR+BlxS0nVKpVluLkycCJ06Fb5vXl4eo0ePNthKkiQpY5T47X4AYoyvA6/vdO4P2z2+Fbi1pOuSyooZM6BePdhvv8L3nT59OjVq1KBevXpJr0uSJEkqikzdPEpSCjkNWZIkSaWJwVYqg8aONdhKkiSp9DDYSmWQI7aSJEkqTQy2UhmTmwsTJhRt46hVq1Yxb948Dj300KTXJUmSJBWVwVYqY7ZuHFWjRuH7jh49ms6dO1OxYsXkFyZJkiQVkcFWKmOchixJkqTSxmArlTEGW0mSJJU2BlupjClqsM3Ly2P06NEGW0mSJGUcg61UhhRn46hp06ZRq1Yt6tatm/S6JEmSpOIw2EplyMyZUKcO1KxZ+L6jRo1ytFaSJEkZyWArlSGur5UkSVJpZLCVyhCDrSRJkkojg61UhhQ12K5atYr58+dz6KGHJr8oSZIkqZgMtlIZkZeX2DiqKMF29OjRdO7cmQoVKiS9LkmSJKm4DLZSGTFzJtSuXbSNo5yGLEmSpExmsJXKCNfXSpIkqbQy2EplRFGDbV5eHqNHj6Z79+7JL0qSJElKAoOtVEYUNdhOmzaN2rVrU7du3eQXJUmSJCWBwVYqA/LyYPx46NSp8H2dhixJkqRMZ7CVyoBZs6BWrcRRWAZbSZIkZTqDrVQGuHGUJEmSSjODrVQGFDXYrly5kgULFtCuXbvkFyVJkiQlicFWKgOKGmxHjx5Nly5dqFChQvKLkiRJkpLEYCuVcls3jipKsHUasiRJkrKBwVYq5WbPhho13DhKkiRJpVexg20I4cgQwtshhJkhhLkhhM9DCHOTUZyk4ivqNOS8vDxGjx5Nt27dkl+UJEmSlETJGLF9FPgncBRwGNAl/+tuhRB6hhBmhBBmhxB+s4d2Z4cQYgihSxLqlMqkogbbqVOnUqdOHerWrZv8oiRJkqQkSkawXR1jHBpjXBpjXL712F3jEEJ54AHgFKANcH4Ioc0u2lUHbgRGJ6FGqcwqarB1GrIkSZKyRTKC7bAQwt9DCIeHEDptPfbQviswO8Y4N8a4GXgOOHMX7f4E3A1sTEKNUpmUlweffmqwlSRJUumWjHt4bF2At/104Qgcv5v2DYEF2z1fuN01AMgPxgfEGIeEEG5OQo1SmTRnDuy3H9SuXfi+I0eO5IYbbkh6TZIkSVKyFTvYxhiPS0YhW4UQypFYs3tJAdpeBVwF0Lhx42SWIZUKY8cWbbR2xYoVLFy4kHbt2iW/KEmSJCnJkrEr8r4hhH+GEMbmH/8IIey7hy6LgAO2e94o/9xW1YG2wPshhC+A7sDgXW0gFWN8KMbYJcbYpU6dOsX9VqRSp6jra0ePHk2XLl2oUCEZkzokSZKk1ErGGtvHgLXAufnHGuDxPbQfA7QIIRwUQqgEnAcM3vpijHF1jLF2jLFJjLEJMAo4I8Y4Ngm1SmWKG0dJkiSpLEhGsG0WY/y//M2g5sYYbwea7q5xjDEH6AO8CUwDBsYYPwsh3BFCOCMJ9UjCjaMkSZJUdiRjnuGGEMJRMcbhACGEI4ENe+oQY3wdeH2nc3/YTdtjk1CjVObMmQP77guFnaWfm5vLJ598Qvfu3VNTmCRJkpRkyQi21wL98tfVBmAFBdj4SVJqFXUa8tSpU6lbty6uW5ckSVK2SMauyBOA9iGEffKfrynuNSUVX1GD7ahRo5yGLEmSpKxS5GAbQrgwxvh0COEXO50HIMb4z2LWJqkYxo2Dm4twF2jX10qSJCnbFGfzqGr5X6vv5pCUJjG6cZQkSZLKjiKP2MYY++Z/vT155UhKhjlzoHp1qFu3cP1WrFjBokWLaNu2bWoKkyRJklKg2Lf7CSH8LYSwTwihYgjh3RDCshDChckoTlLRFHV97ejRo+nSpQsVKiRjXzlJkiSpZCTjPrYn5W8Y9UPgC6A5UISVfZKSpajB1mnIkiRJykbJCLZbh3ZOA56PMa5OwjUlFYPBVpIkSWVJMuYbvhZCmA5sAK4NIdQBNibhupKKoKgbR+Xm5vLJJ5/QvXv31BQmSZIkpUixR2xjjL8BjgC6xBi3AN8AZxb3upKKZu5cqFYN6tUrXL+pU6dSr149ateunZrCJEmSpBQpzn1sj48xvhdCOGu7c9s3ebE4hUkqGqchS5IkqawpzlTkHwDvAafv4rWIwVZKC4OtJEmSypri3Mf2//K/Xpq8ciQV17hxcNNNhe83cuRIfv7znye9HkmSJCnVknEf27+EEPbb7nmNEMKfi3tdSYVX1I2jVqxYweLFi2nbtm1qCpMkSZJSKBm3+zklxrhq65MY40rg1CRcV1Ihff45VK0K9esXrt+oUaM47LDDKF++fGoKkyRJklIoGcG2fAih8tYnIYS9gMp7aC8pRVxfK0mSpLIoGcG2P/BuCOHyEMLlwNtAvyRcV1IhGWwlSZJUFhVnV2QAYox3hxAmAifmn/pTjPHN4l5XUuGNGweF3f8pNzeXMWPG0L1795TUJEmSJKVasYNtvmlATozxnRBC1RBC9Rjj2iRdW1IBxFi0EdvPPvuM+vXrU6tWrdQUJkmSJKVYMnZFvhJ4Aeibf6oh8HJxryupcL74Avbaq/AbRzkNWZIkSdkuGWtsrweOBNYAxBhnAXWTcF1JheD6WkmSJJVVyQi2m2KMm7c+CSFUAGISriupEAy2kiRJKquSEWw/CCH8FtgrhNADeB54NQnXlVQIRQm2y5cvZ8mSJRxyyCGpKUqSJEkqAckItrcAy4DJwNXA68DvknBdSQVU1I2jRo0axWGHHUb58uVTU5gkSZJUAoq1K3IIoTzwWYyxFfBwckqSVFjz5kHlytCgQeH6OQ1ZkiRJpUGxRmxjjLnAjBBC4yTVI6kIxo51fa0kSZLKrmTcx7YG8FkI4RPgm60nY4xnJOHakgqgKNOQc3NzGTNmDN27d09NUZIkSVIJSUaw/X1hO4QQegL3AuWBR2KMd+30+jUkbiOUC6wDrooxTk1CrVKpNG4c3HBD4fpMnTqV+vXrU6tWrdQUJUmSJJWQIgfbEEIV4BqgOYmNox6NMeYUoF954AGgB7AQGBNCGLxTcH0mxvhgfvszgH8CPYtaq1SaFWfjKKchS5IkqTQozhrbfkAXEqH2FOAfBezXFZgdY5ybf//b54Azt28QY1yz3dNqeF9cabfmzYNKlWD//QvXb9SoUU5DliRJUqlQnKnIbWKM7QBCCI8CnxSwX0NgwXbPFwLddm4UQrge+AVQCTi+GHVKpVpRRmshEWz79OmT/IIkSZKkElacEdstWx8UZApyYcUYH4gxNiNxn9xd3hc3hHBVCGFsCGHssmXLkl2ClBWKEmxXrVrFvHnzaNeuXWqKkiRJkkpQcYJt+xDCmvxjLXDo1schhDV76LcIOGC7543yz+3Oc8CPdvVCjPGhGGOXGGOXOnXqFLZ+qVQoSrAdM2YMnTt3pkKFZOwfJ0mSJKVXkX+rjTGWL2LXMUCLEMJBJALtecAF2zcIIbSIMc7Kf3oaMAtJ31GcjaNcXytJkqTSojgjtkWSP225D/AmMA0YGGP8LIRwR/4OyAB9QgifhRAmkFhn27uk65Sywfz5UKGCG0dJkiSpbEvLPMQY4+vA6zud+8N2j28s8aKkLLR1tDaEgveJMTJq1Cgefvjh1BUmSZIklaASH7GVlDxFmYY8e/Zs9t57b/Yv7DCvJEmSlKEMtlIWc32tJEmSZLCVspYbR0mSJEkJBlspSy1YAOXKQcOGhetnsJUkSVJpY7CVslRRNo5av34906dPp2PHjqkrTJIkSSphBlspS40aBd26Fa7PuHHjaNu2LVWqVElNUZIkSVIaGGylLDV8OBx5ZOH6OA1ZkiRJpZHBVspCGzfChAmFH7E12EqSJKk0MthKWWjcOGjVCqpXL3ifGCMjR4402EqSJKnUMdhKWejjjws/DXnhwoXk5ubSpEmTlNQkSZIkpYvBVspCRQm2W6chh8JsoyxJkiRlAYOtlGViLF6wlSRJkkobg62UZWbMgL33hkaNCtfPYCtJkqTSymArZZmijNZu3ryZCRMm0KVLl9QUJUmSJKWRwVbKMkUJthMnTqR58+ZUL8w2ypIkSVKWMNhKWWb4cDjqqML1cRqyJEmSSjODrZRFli5NHIccUrh+BltJkiSVZgZbKYuMGAHdu0P58oXrZ7CVJElSaWawlbJIUdbXLl26lOXLl3PwwQenpihJkiQpzQy2UhYpyvra0aNH061bN8qV8393SZIklU7+pitliQ0bYNIk6Nq1cP2chixJkqTSzmArZYmxY6FNG6hWrXD9DLaSJEkq7Qy2UpYoyvra3NxcxowZQ9fCDvNKkiRJWcRgK2WJoqyvnTp1Kg0aNKBWrVqpKUqSJEnKAAZbKQvk5SVu9VPYEVunIUuSJKksMNhKWWD6dNhvP2jQoHD9DLaSJEkqC9ISbEMIPUMIM0IIs0MIv9nF678IIUwNIUwKIbwbQjgwHXVKmaIo62vBYCtJkqSyocSDbQihPPAAcArQBjg/hNBmp2bjgS4xxkOBF4C/lWyVUmYpyvraVatWMW/ePNq1a5eaoiRJkqQMkY4R267A7Bjj3BjjZuA54MztG8QYh8UY1+c/HQU0KuEapYxSlBHbMWPG0LlzZypUqJCaoiRJkqQMkY5g2xBYsN3zhfnndudyYGhKK5Iy2FdfwfLliXvYFobTkCVJklRWZPTmUSGEC4EuwN938/pVIYSxIYSxy5YtK9nipBLy8cdw+OFQrpD/txpsJUmSVFakI9guAg7Y7nmj/HM7CCGcCNwGnBFj3LSrC8UYH4oxdokxdqlTp05KipXSrSjra2OMjBo1im7duqWmKEmSJCmDpCPYjgFahBAOCiFUAs4DBm/fIITQEehLItQuTUONUsYoyvra2bNns/fee7P//vunpihJkiQpg5R4sI0x5gB9gDeBacDAGONnIYQ7Qghn5Df7O7A38HwIYUIIYfBuLieVauvXw5QpcNhhhevnNGRJkiSVJWnZLjXG+Drw+k7n/rDd4xNLvCgpA40ZA23bQtWqhetnsJUkSVJZktGbR0ll3ccfF359LRhsJUmSVLYYbKUMNnx44dfXrl+/nunTp9OpU6fUFCVJkiRlGIOtlKHy8mDkyMIH23HjxtGuXTsqV66cmsIkSZKkDGOwlTLU1KlQqxbUq1e4fiNHjnQasiRJksoUg62UoVxfK0mSJBWMwVbKUEVZXxtjdMRWkiRJZY7BVspQH39c+GC7YMECYowceOCBqSlKkiRJykAGWykDLVkCq1ZBq1aF67d1GnIIISV1SZIkSZnIYCtloK2jteUK+X+o62slSZJUFhlspQxUlPW1YLCVJElS2WSwlTJQUdbXbtq0iYkTJ9KlS5fUFCVJkiRlKIOtlGG++SZxD9vC5tOJEyfSokUL9t5779QUJkmSJGUog62UYT75BNq3h732Klw/pyFLkiSprDLYShnG9bWSJElS4RhspQxTlPW1YLCVJElS2WWwlTJIbi6MGgVHHFG4fl999RUrV66kZcuWqSlMkiRJymAGWymDfPYZ1KsHdesWrt/o0aPp1q0b5Qp741tJkiSpFPC3YCmDuL5WkiRJKjyDrZRBXF8rSZIkFZ7BVsogRQm2ubm5jB07lq5du6amKEmSJCnDGWylDLFwIaxbBwcfXLh+n332Gfvvvz81a9ZMTWGSJElShjPYShli62htCIXr5zRkSZIklXUGWylDuL5WkiRJKhqDrZQhDLaSJElS0RhspQywbh3MmAFduhSu36pVq1iwYAFt27ZNTWGSJElSFjDYShlg9Gjo0AEqVy5cv08++YTOnTtToUKFlNQlSZIkZQODrZQBhg93GrIkSZJUVGkJtiGEniGEGSGE2SGE3+zi9WNCCJ+GEHJCCOeko0apJLm+VpIkSSq6Eg+2IYTywAPAKUAb4PwQQpudms0HLgGeKdnqpJKXm5uYinzEEYXrF2Nk9OjRdOvWLTWFSZIkSVkiHQvzugKzY4xzAUIIzwFnAlO3NogxfpH/Wl4a6pNK1OTJsP/+ULt24frNmjWL6tWr06BBg9QUJkmSJGWJdExFbggs2O75wvxzUplU1PW177//vtOQJUmSJLJ886gQwlUhhLEhhLHLli1LdzlSkRRlfW1eXh733Xcfl112WWqKkiRJkrJIOoLtIuCA7Z43yj9XaDHGh2KMXWKMXerUqZOU4qSS9vHHcNRRheszZMgQKlWqRI8ePVJTlCRJkpRF0hFsxwAtQggHhRAqAecBg9NQh5R28+fDxo3QvHnh+t19993ccssthBBSU5gkSZKURUo82MYYc4A+wJvANGBgjPGzEMIdIYQzAEIIh4UQFgI/AfqGED4r6TqlkrB1GnJh8unw4cNZsmQJZ599duoKkyRJkrJIOnZFJsb4OvD6Tuf+sN3jMSSmKEulWlHW1959993cfPPNVKiQlv99JUmSpIyT1ZtHSdmusOtrp0yZwtixY+ndu3fqipIkSZKyjMFWSpM1a2DWLOjUqeB9/v73v3PDDTew1157pa4wSZIkKcsYbKU0GTUqEWorVSpY+/nz5/Paa69x3XXXpbYwSZIkKcsYbKU0Kez62n/+859cdtll7LfffimrSZIkScpG7j4jpcnHH8NNNxWs7fLly3nyySeZPHlyaouSJEmSspAjtlIaLFwIY8fC4YcXrP0DDzzAWWedRcOGDVNbmCRJkpSFHLEtIbNmzWL9+vW0b98+3aUozWKEK6+EX/wCatb8/vbffPMN//nPf/joo49SX5wkSZKUhRyxLSETJ07knHPOYd26dekuRWn2+OPw1Vdw660Fa//YY49x1FFHcfDBB6e2MEmSJClLhRhjumtIii5dusSxY8emu4w9uuKKK8jLy+Oxxx5LdylKk/nzoXNneO89aNfu+9tv2bKFFi1aMGDAALp165b6AiVJkqQMFUIYF2PssqvXHLEtQf/+97/5+OOPGTBgQLpLURrECFdcAT//ecFCLcCAAQM46KCDDLWSJEnSHhhsS9Dee+/NM888ww033MC8efPSXY5K2MMPw4oVcMstBWsfY+Rvf/sbtxS0gyRJklRGGWxLWOfOnfn1r39Nr169yMnJSXc5KiFffAG//S306wcVCrhl29ChQylXrhwnn3xySmuTJEmSsp3BNg1+8YtfsNdee/GXv/wl3aWoBOTlweWXw803wyGHFLzfXXfdxS233EIIIXXFSZIkSaWAwTYNypUrx5NPPsn//vc/Pv7443SXoxTr2xe++QZ++cuC9xk5ciQLFy7kJz/5SeoKkyRJkkoJg22aNGjQgL59+3LhhReyevXqdJejFJk7F37/e3jiiYJPQQa4++67+dWvfkWFwnSSJEmSyiiDbRqdccYZnHrqqVx77bWUltsu6Vt5eXDZZfCb30CrVgXvN3XqVEaNGsWll16auuIkSZKkUsRgm2b33HMPEydO5Omnn053KUqyBx6AzZvhppsK1+/vf/87ffr0Ya+99kpNYZIkSVIp4zzHNNtrr7149tlnOeGEEzjiiCNo1qxZuktSEsyeDbffDiNGQPnyBe+3YMECBg8ezOzZs1NXnCRJklTKOGKbAQ499FD+8Ic/cMEFF7Bly5Z0l6NiysuDSy+F3/0OWrYsXN9//etfXHLJJdSoUSM1xUmSJEmlkME2Q/Tp04fatWvzxz/+Md2lqJjuuw9CgJ/9rHD9VqxYwRNPPMFNhZ27LEmSJJVxTkXOECEEHn/8cTp27EiPHj049thj012SimDmTLjzThg5EsoV8s9G//3vf/nRj35Eo0aNUlOcJEmSVEoZbDNI3bp1eeyxx7j44ouZMGECNWvWTHdJKoTcXLjkEvi//4PmzQvXd/369dx///28//77qShNkiRJKtWcipxhTj75ZM455xyuvPJKbwGUZf79b6hcGa67rvB9H3/8cQ4//HBat26d9LokSZKk0s5gm4H++te/MmfOHB599NF0l6ICmj4d7roLHn208FOQc3JyuOeee/jNb36TmuIkSZKkUs5gW0IWLID58wvWtnLlyjz77LPceuutTJ8+PbWFqdhycqB3b7jjDmjatPD9Bw4cSOPGjenevXvyi5MkSZLKAINtCfnwQ+jcGdq3T9wGZtSoxJrM3WndujV33nknF1xwAZs2bSq5QlVo//gHVK8OV19d+L4xRv72t79xyy23JL8wSZIkqYxIS7ANIfQMIcwIIcwOIXxn/mUIoXIIYUD+66NDCE3SUGZS9eoFX34J//tf4j6nV14J+++fuN/poEGwZs13+1x55ZU0adKE2267reQLVoF89hncc0/RpiADvPnmm+Tl5XHKKackvzhJkiSpjCjxYBtCKA88AJwCtAHODyG02anZ5cDKGGNz4F/A3SVbZWqULw9HHAF/+QtMngyjR0OXLvDww9CoEfTokbgH6ty5ifYhBB5++GEGDBjAW2+9ld7i9R05OYldkP/yFzjwwKJd4+677+aWW24hhJDU2iRJkqSyJB0jtl2B2THGuTHGzcBzwJk7tTkT6Jf/+AXghFAKf/Nv0gSuvx7eeAMWL048njQpEX7btIFbboGpU2vx+ONPcemll7J06dJ0l6zt/O1vULMmXHFF0fqPHj2azz//nJ/+9KfJLUySJEkqY9JxH9uGwILtni8Euu2uTYwxJ4SwGqgFfF0iFabB3nvDj36UOPLyYNw4eO01uPFGmDfvWOrXH8KJJ/bjrLOOTW+hAmDdugo8/HAb7r13OK+/vrFI17j33nv51a9+RYUK3k5akiRJKo6s/o06hHAVcBVA48aN01xN8pQrB4cdljhuvx0WLYLBg9vy97/n8N//em/bzJBD8+Z/4/nnRxX5CnXr1uWyyy5LYk2SJElS2ZSOYLsIOGC7543yz+2qzcIQQgVgX2D5zheKMT4EPATQpUuXUpv4GjaEa6+twLXXdkl3KdrBYekuQJIkSRLpWWM7BmgRQjgohFAJOA8YvFObwUDv/MfnAO/FGEttcJUkSZIkFV2Jj9jmr5ntA7wJlAceizF+FkK4AxgbYxwMPAo8FUKYDawgEX4lSZIkSfqOtKyxjTG+Dry+07k/bPd4I/CTkq5LkiRJkpR90jEVWZIkSZKkpDHYSpIkSZKymsFWkiRJkpTVDLaSJEmSpKxmsJUkSZIkZTWDrSRJkiQpqxlsJUmSJElZzWArSZIkScpqIcaY7hqSIoSwDJiXwreoDXydwutLReHnUpnIz6UyjZ9JZSI/l8pEmf65PDDGWGdXL5SaYJtqIYSxMcYu6a5D2p6fS2UiP5fKNH4mlYn8XCoTZfPn0qnIkiRJkqSsZrCVJEmSJGU1g23BPZTuAqRd8HOpTOTnUpnGz6QykZ9LZaKs/Vy6xlaSJEmSlNUcsZUkSZIkZTWDbQGEEHqGEGaEEGaHEH6T7npUNoUQHgshLA0hTNnuXM0QwtshhFn5X2uks0aVLSGEA0IIw0IIU0MIn4UQbsw/7+dSaRNCqBJC+CSEMDH/c3l7/vmDQgij83+WDwghVEp3rSpbQgjlQwjjQwiv5T/3M6m0CiF8EUKYHEKYEEIYm38ua3+GG2y/RwihPPAAcArQBjg/hNAmvVWpjHoC6LnTud8A78YYWwDv5j+XSkoO8MsYYxugO3B9/r+Pfi6VTpuA42OM7YEOQM8QQnfgbuBfMcbmwErg8vSVqDLqRmDads/9TCoTHBdj7LDdLX6y9me4wfb7dQVmxxjnxhg3A88BZ6a5JpVBMcYPgRU7nT4T6Jf/uB/wo5KsSWVbjHFJjPHT/MdrSfzC1hA/l0qjmLAu/2nF/CMCxwMv5J/3c6kSFUJoBJwGPJL/POBnUpkpa3+GG2y/X0NgwXbPF+afkzJBvRjjkvzHXwL10lmMyq4QQhOgIzAaP5dKs/wpnxOApcDbwBxgVYwxJ7+JP8tV0v4N/BrIy39eCz+TSr8IvBVCGBdCuCr/XNb+DK+Q7gIkJUeMMYYQ3OZcJS6EsDcwCPh5jHFNYiAiwc+l0iHGmAt0CCHsB7wEtEpvRSrLQgg/BJbGGMeFEI5NcznS9o6KMS4KIdQF3g4hTN/+xWz7Ge6I7fdbBByw3fNG+eekTPBVCKEBQP7XpWmuR2VMCKEiiVDbP8b4Yv5pP5fKCDHGVcAw4HBgvxDC1j/o+7NcJelI4IwQwhcklrQdD9yLn0mlWYxxUf7XpST+CNiVLP4ZbrD9fmOAFvk711UCzgMGp7kmaavBQO/8x72BV9JYi8qY/DVijwLTYoz/3O4lP5dKmxBCnfyRWkIIewE9SKz/Hgack9/Mz6VKTIzx1hhjoxhjExK/R74XY+yFn0mlUQihWgih+tbHwEnAFLL4Z3iIMWtGl9MmhHAqibUR5YHHYox3prcilUUhhGeBY4HawFfA/wEvAwOBxsA84NwY484bTEkpEUI4CvgImMy368Z+S2KdrZ9LpUUI4VASG56UJ/EH/IExxjtCCE1JjJbVBMYDF8YYN6WvUpVF+VORfxVj/KGfSaVT/ufvpfynFYBnYox3hhBqkaU/ww22kiRJkqSs5lRkSZIkSVJWM9hKkiRJkrKawVaSJEmSlNUMtpIkSZKkrGawlSRJkiRlNYOtJEmSJCmrGWwlSZIkSVnNYCtJkiRJymoGW0mSJElSVjPYSpIkSZKymsFWkiRJkpTVDLaSJEmSpKxmsJUkSZIkZTWDrSRJkiQpqxlsJUmSJElZzWArSZIkScpqBltJkiRJUlYz2EqSJEmSsprBVpIkSZKU1Qy2kiRJkqSsZrCVJEmSJGU1g60kSZIkKasZbCVJkiRJWc1gK0mSJEnKagZbSZIkSVJWM9hKkiRJkrKawVaSJEmSlNUMtpIkSZKkrGawlSRJkiRlNYOtJEmSJCmrGWwlSZIkSVnNYCtJkiRJymoGW0mSJElSVjPYSpIkSZKymsFWkiRJkpTVDLaSJEmSpKxmsJUkSZIkZTWDrSRJkiQpqxlsJUmSJElZzWArSZIkScpqFdJdQLLUrl07NmnSJN1lSJIkSZJSYNy4cV/HGOvs6rVSE2ybNGnC2LFj012GJEmSJCkFQgjzdveaU5ElSZIkSVnNYCtJkiRJymoGW0mSJElSVjPYSpIkSZKymsFWkiRJkpTVDLaSJEmSpKxmsJUkSZIkZTWDrSRJkiQpqxlsJUmSJElZzWArSZIkScpqFdJdgCRJkiSp5OTkwNKlsGTJjsd110Ht2umurmgMtpIkSZLKnJUrYfZsWLgwEfRiLNoBUL48VKmy41G58u7PVa6c6JNsmzZ9N6zu6li+HGrWhAYNdjxyc5NfU0kx2EqSJEkqdWKEL7+EOXMSAXbnrzk50Lw5HHAAVKwIIRT9yMmBjRsTwXLjxm+PnZ9vf65ChR2DbrliLhJdty5x1Kv33cDarduOz+vWTXzPpYnBVpIkSVJWys2FBQt2HVznzoWqVaFZs0SAbdYMTjvt2+e1aydCaTrECFu27Bh2t47+FtVee0GtWsUPyNnKYCtJkiQpa8yfD2++mTjefReqV98xvHbrlvjarBnss0+6q921EKBSpcSRqTVmG4OtJEmSpEJbvjwRylI9pXXjRvjwQ3jjjUSYXboUTjoJzjwT/vMfqF8/te+v7GCwlSRJkrRbmzfD9OkwaRJMnJg4Jk2C9esT02mbNoU2beCQQxJf27SBFi0S60aLIkaYOfPbIDt8OBx6KPTsCU88AZ06pWbjJWU3g60kSZIkIDEaujW4bg2xM2dCkyaJcNm+Pdx4Y+Jxo0aJ0dSZM2Hq1MQxYEDi6+efJ/psDbpbQ+/BByc2S9rZ2rXw3nuJMPvGG4nNmHr2hMsug/79oUaNkv4voWwTYnFXKWeILl26xLFjx6a7DEmSJCnjffNNYnOlnUPspk2J8Lo1xLZvnwikVasW7vqbN+8YeLcec+YkAvHWwFutGrzzDowbB4cfDiefnAi0bdqkb2MnZa4QwrgYY5ddveaIrSRJklQKxAirVhXsPqZbtsCBB34bYm+4IfH1gAOSEygrVYK2bRPH9rZsSexYvDXorloFv/41/OAHiZCrgosxMn/+fCZPnszkyZPZvHkzXbp04bDDDqNu3brpLq/EOWIrSZIkZYm1a2HMGBg7NrE78PZh9csvExs57XwP0/33/+65ffd1RDSbrFy5cluA3XpMmTKFatWq0a5dO9q1a0fFihUZO3YsY8eOZZ999uGwww7bdnTu3Jl999033d9Gse1pxNZgK0mSJGWgvLzEpk2jRn17zJkDHTpA166JNaw7B1ZHPbPbpk2bmDZt2ndC7OrVq2nbtu22ELv1qFWr1neukZeXx5w5cxgzZsy2Y8KECTRq1GiHsNuhQwf22muvNHyXRWewlSRJkjLc11/D6NHfhtgxY6B2bejePXF065aYOlypUrorVXHFGFmyZAkTJkxg4sSJTJgwgcmTJ/P555/TtGnT7wTYAw88kHLlyhX5/XJycpg6deoOYXfatGkcfPDBO4Tdtm3bUqFC5q5WNdhKkiRJGWTLlsSmTduPxi5dCocdtmOQrVMn3ZWquLZs2cL06dO3BditXwE6dOhAhw4daN++Pe3ataNVq1ZULup9kgpp48aNTJw4cYew+8orr9CiRYsSef+iMNhKkiRJabJhA0yZAuPHw4QJia+TJ8NBB30bYrt3h1atvD9rtlu1ahUTJ07cIcROmzaNxo0b0759+20htkOHDjRo0IDgQudCcVdkSZIkqQQsX54Ir1sD7IQJiXWxBx8MHTsm1seed17i6z77pLXUMikvL4/ly5ezZMmSbcfXX39NTk4OMUby8vJ2e+zp9YULFzJx4kSWLVvGoYceSvv27enWrRtXX301bdu2pZqLn1POEVtJkiSpkGJM7Eo8fvyOI7GrViXWwW4NsR07Ju7JWkKzS8usTZs28dVXX20Lq19++eUO4XXr86VLl7LPPvtQv359GjRoQIMGDahTpw4VKlSgXLlyuz1CCHt8rV69enTs2JFmzZoVay2s9swRW0mSJInETsPr1sH69fDNN989CnJ+5crEVOIqVb4NrxdeCP/4BzRtCuaawtm8eTNr1qzZ47F27dpdnl+9ejXLli1j7dq11K1blwYNGuwQWjt16rTDuXr16pXYGlaVLIOtJEmSSr01a+CRR+DeexPThatV2/GoWnX352rV2vHcvvtC27ZQr166v6vssmHDBiZOnMi4ceP49NNPGTduHDNmzCAnJ4d99tnne4+mTZtSvXr1Hc5Vr16dOnXqULt2bUdKyziDrSRJkkqthQvhvvvg0UehRw944YXEzsNKrW+++YYJEyZsC7Djxo1jzpw5tGrVis6dO3PYYYdx9dVXc8ghh1C1alU3UVKxGWwlSZJU6kyYkJgaPGQI9O4N48ZBkybprqp0Wrt2LRMmTNhhJPbzzz/nkEMOoXPnzhxxxBHccMMNtG3b1mnAShmDrSRJkkqFGOHNNxOBdupU+NnPEqO1NWqku7LskZubu23t6urVq1m1atUOX7d/vHLlSqZMmcKCBQto27YtnTt35gc/+AE33XQThxxyCJUqVUr3t6MyxGArSZKkrLZpEzz7bCLQhgC/+lXiljqlIVfl5OTw9ddfs2zZMpYuXcqyZctYtmwZX3/9NVu2bCE3N5e8vDxyc3N3eFyQcxs3bvxOWP3mm2+oXr06++67L/vuuy/77bffdx7Xq1ePli1bst9++/GHP/yB1q1bU7FixXT/p1IZZ7CVJElSVlq5Eh58EO6/H9q1g3/+E048MRFuM9m6deuYM2fOtpC6c2jd/vmaNWuoVasWderU2eGoXbs21apVo3z58pQvX55y5crt8LUg5ypXrvyd4Fq9enU3YVJWMthKkiQpZdavh4kTE2tcx49P3G6nZs3E9OCtX7d/XLMm7LcflC//3WstWrSIWbNmUbdudx58sApPPw1nnAFvvAGHHlri31qBrV+/no8//pj333+fYcOGMWnSJA466CDq1q27Q1jt0KHDd87VrFnToCkVQEqDbQihJ3AvUB54JMZ4106v/wK4AsgBlgGXxRjn5b+WC0zObzo/xnhGKmuVJElS8WzYAJMmwdixiWPcOJg9G1q3hs6dE7sRV6yYGGldsQIWLfr28cqV3z5esyZxa50aNfKoUGEdmzd/xerVc9m06Uv22qsxq1d/Q7Nmr3HLLRvo1es4GjVqlO5vfQcbNmxg5MiRDBs2jPfff5/x48fToUMHjj32WP70pz9x+OGHU7Vq1XSXKZUqIcaYmguHUB6YCfQAFgJjgPNjjFO3a3McMDrGuD6EcC1wbIzxp/mvrYsx7l3Q9+vSpUscO3ZsUr8HSZIk7drGjd+G2HHjEsfMmdCqVSLEdumS+NquHRR0I9wYI1OmTOGNN97i9deH88kns2nZsivt2x9Hy5bdqVGjKZUrl+OEE1YwYsSbvPbaa7zxxhs0btyY0047jR/+8IccdthhlN/VcG8Kbdy4kdGjRzNs2DCGDRvGuHHjaNeuHccddxzHHnssRx55JNWqVSvRmqTSKIQwLsbYZZevpTDYHg78McZ4cv7zWwFijH/dTfuOwH9ijEfmPzfYSpIkZYiZM+G99xIBduxYmDEDDj74uyG2SpXCXXfZsmW88847vPnmm7z11ltUqVKFk08+mZNPPpnjjjuOfffdd4/9c3JyGDVqFK+99hqvvfYaS5cu5dRTT+W0007jpJNO+t7+RbF582Y++eSTbUH2k08+oU2bNhx33HEcd9xxHHnkkVSvXj3p7yuVdekKtucAPWOMV+Q/vwjoFmPss5v2/wG+jDH+Of95DjCBxDTlu2KML+/p/Qy2kiRJybNlC3z0Ebz2WuJYvx569ICuXRMh9tBDCx9iIREKR44cuS3Izpo1i2OPPZaTTz6Zk046iebNmxer7i+++IIhQ4bw2muvMXz4cLp27bptNLdly5Z77Lthwwa++uorvvzyy23Hzs+3nmvduvW2IHv00Uezzz77FKtuSd8v44NtCOFCoA/wgxjjpvxzDWOMi0IITYH3gBNijHN26ncVcBVA48aNO8+bNy8l34skSVJZ8PXXMHRoIsi+9Ra0bAk//GHi6NCh6LsNb9iwgddee41nn32Wd999l5YtW3LSSSdx8skn071795Td7/Sbb77h3Xff3Taau/fee3PaaadRp06dXYbWjRs3Ur9+/W1HvXr1dni+9VyDBg1cIyulQUZPRQ4hnAjcTyLULt3NtZ4AXosxvrC793PEVpIkqXBihM8+g1dfTYTZKVPghBMSQfbUU6F+/aJfOzc3l2HDhtG/f39efvllOnfuzAUXXMDpp59OnTp1kvdNFFCMkfHjx/P666/zzTff7DK07rvvvoRMv1eQVIalK9hWILF51AnAIhKbR10QY/xsuzYdgRdIjOzO2u58DWB9jHFTCKE2MBI4c/uNp3ZmsJUkSZlq1SrYZx/IhLu2bNwI77//7RTjEBJB9vTT4Qc/KPhGT7sSY+TTTz+lf//+PPfcczRo0IBevXpx3nnnsf/++yfte5BUNu0p2Kbsdj8xxpwQQh/gTRK3+3ksxvhZCOEOYGyMcTDwd2Bv4Pn8v45tva1Pa6BvCCEPKEdije1uQ60kSVIm2bQpsT516FB4/XWYPz9x/9YmTaBpU2jW7NujaVM46CDYa6/k1/DVV/Dll4mv8+fDO+8kNoA69NBEmB0yBNq0KfoU463mzJnDM888Q//+/dm8eTO9evXi3XffpXXr1sn5ZiTpe6RsxLakOWIrSZLS6YsvEkF26FD44INEYDzllMTRuXNipHTu3MQxZ863x9y5MG8e1K6969DbrBnUqpUIn7m5sHx5IqxuPZYsiSxYsJkvvtjE4sW5LF1ajhUrKrNpUwUqV15DhQrLiHEJW7YsoGbNybRuPY+WLWvSpEkTDjroIJo0aUKTJk2oU6dOoabhLlu2jAEDBtC/f3/mzJnDueeeS69evejevbvTeSWlRFqmIpc0g60kSSpJ24/KDh2a2Hjp5JMTa1NPOikRRgsqNxcWLvw26G4ffOfMSaxXLV9+E2vXVqZSpQ1UqrSCEJaSm7uIjRu/oFy5r6hRYxN16uTRsGF5DjywMk2a7EPDhg1o0CBx1KlTh6+//prPP/+cL7744jvHhg0bOPDAA78TeLcetWvXZv369bz88sv079+fESNGcNppp9GrVy969OhBxYoVU/cfW5Iw2EqSJCXFvHnfTi/e1ahsMtfQzps3j+eee45nn32Wr77awuGHn0DTpnvTqFHdbWG1fv36NGjQICn3TF2zZg3z5s37TuDdGoQ3b95M+fLlOfLII+nVqxdnnnkme++9dxK+U0kqGIOtJElSAeXlwYoVsHRp4vjqK/jkk+KPyhbE0qVLef7553nmmWeYMWMGZ511FhdccAFHH3005cuXT+6bFdLq1avJy8ujRo0aaa1DUtmVls2jJEmSMsW6dd8G1e87li9P7GBct27iqFMnsdlSv37JH5WFRGB86aWXePbZZxk9ejSnnXYat956KyeddFLK7u9aFPvuu2+6S5Ck3TLYSpKkjPPFF/D00/Dss7B48bfndzXRbOdzOz/fsiWx8VK9et+G1a3HgQfCYYdtH2Ij1atvYsOGNaxdu5Y1a9awZs0aqlSpQoMGDcnNrUe5csVfS7phwwZee+01nn32Wd59912OPfZYLrvsMl588UWqVatW7OtLUlljsJUkSRlh5Up4/nl46imYPh3OPRceeQRatfpu25033d3d840bN/LKK4NYt27ptpC6NbBOnLhjeN36Wrly5ahevTr77LMP++yzD9WrV2fjxo0sWrSIZcuWUbNmTRo2bMj+++/P/vvvv8vHtWrVotxOQ7tbtmzhnXfe4dlnn+XVV1+lc+fOnH/++Tz66KNO75WkYjLYSpKktNm8ObF29amnEvdY7dEDbr4ZevaE4s7CnTt3Lueccw61atWibdu27LPPPjRo0ICDDz54W2DdGl63Pq9evTqVK1fe7TVzc3P56quvWLx4MYsXL2bRokUsXryYESNG7HBu3bp1NGjQYFvg3WuvvXjjjTdo3rw5559/Pn/729+oX79+8b5BSdI2BltJklSiYoTRoxNhduBAaN0aLrooMTq7337JeY/BgwdzxRVX8Pvf/54+ffok7b6q5cuX3xZW92Tjxo0sWbJkW/BdtWoVt99+O02bNk1KHZKkHRlsJUlSiZgzJ7Fu9umnoXz5RJgdMwaaNEnee+Tk5HDbbbfx3HPPMXjwYLp37568ixdClSpVOOiggzjooIPS8v6SVNYYbCVJUsqsWJEYlX3qKZg1C847D555Brp0+e662OJasmQJ5513HlWqVGHcuHHUrl07uW8gScpYSd6wXpIklXUxwqhR0KsXNG0Kw4bBrbfCokVw332JXYiTHWrff/99unTpwgknnMDrr79uqJWkMsYRW0mSlBQbN8KAAXD//bBqFVx/PTzwQPLWze5KXl4ef/vb37j33nt58skn6dGjR+reTJKUsQy2kiSpWBYsgAcfTGz+1LEj3H47nHIKlEvxvLCVK1dy8cUXs3z5csaMGUOjRo1S+4aSpIzlVGRJklRoMcIHH8A550D79rBuHXz4IbzxBpx2WupD7dixY+ncuTMtWrTggw8+MNRKUhnniK0kSfqO3Nxc3njjDZYtW7btXIyRTZsqMGpUU959tzW5ueU4/vhp/OlPs6hSZQvDh8NHH8UdrlO9enW6du1KkyZNknLLnRgjDz30EL///e/53//+x9lnn13sa0qSsp/BVpIkbbNp0yaeeuop/va3v1GjRg1at24NwNq1dZg+/XjmzDmKunVn06bNo+y//zRCgHHj2CG0bv94xYoV3HTTTeTl5XHEEUdw+OGHc/jhh9O5c2f22muvQtX2zTffcM011zBx4kSGDx9Oy5Ytk/NNS5KynsFWkiSxbt06HnroIf75z3/Stm1bHn74YY4++hjefTdw//0wYgRceilcdx0cdFBHoGOBrx1jZP78+YwcOZIRI0YwcOBApk6dyiGHHLJD2D3ggAN2O6o7ffp0zjnnHLp06cKoUaOoWrVqkr5zSVJpEGKM398qC3Tp0iWOHTs23WVIkpRVli9fzv33388DDzzAcccdx/XX38bKle15++3Eetlq1eCGGxK37klmlly/fj3jxo1jxIgRjBw5kpEjR1KhQoUdgm6nTp2oXLkyAwYMoE+fPtx1111cdtllSZnSLEnKPiGEcTHGLrt8zWArSVLZs3DhQv75z3/y+OP9OeKIX3DggVcwfnwtpkyBI46AE0+EHj0SG0OVRI6MMTJ37txtIXfkyJHMmDGD5s2b88033/D888/TsWPBR4klSaWPwVaSJAEwY8ZMfvvbpxk6NIf69S9k2bJWHHxwOXr0SITZI4+EKlXSXWXCunXrmDBhAu3atWPfffdNdzmSpDTbU7B1ja0kSaXc4sXwyCNf8Oij81i48GD22ecmzjlnL04/vQrHHw+1aqW7wl3be++9Oeqoo9JdhiQpCxhsJUnKYps3w8qVsGrVjsfKlTBtWmTw4PUsXJhHhQqf8cMfVuHVV6tz6KHV0lu0JElJZrCVJCnDbNwIL70EixZ9G1J3Dq1bH2/ZAjVqwH777Xhs3vwVn376EvAO//3vqVx8cS8qV66cpu9IkqTUMthKkpQhNm2Cxx6Dv/wF2rSBtm0TobV16x1D6/ZBtmrVHTd3ys3N5a677uL+++/nn//8Jz/96QDKly+flu9HkqSSYrCVJCnNtmyBJ56AP/85EWgHDYKuXQt/nUWLFnHRRReRl5fH2LFjadSoUdJrlSQpE5VLdwGSJJVVOTnw+ONw8MEwcCA8+ywMHVq0UPvqq6/SuXNnjj/+eN59911DrSSpTHHEVpKkEpaTA888A3fcAY0bQ79+cPTRRbvWxo0b+fWvf83gwYMZNGgQRx55ZHKLlSQpCxhsJUkqIbm5MGAA3H471KsHDz8Mxx1X9OtNmzaN8847j5YtWzJ+/Hhq1KiRvGIlScoiTkWWJCnF8vISgbZdO3jggcTxwQdFD7UxRh599FGOOeYY+vTpw8CBAw21kqQyzRFbSZJSJC8PXnwR/vhH2Htv+Ne/4KSTdtzFuLBWrVrF1VdfzbRp0/jggw9o06ZN0uqVJClbOWIrSVKSxQgvvwwdO8Jdd8Hf/gYjR8LJJxcv1I4YMYKOHTtSt25dPvnkE0OtJEn5HLGVJKmIVq+GWbMSx8yZW7/mMXVqDrVqreOmm1ZxzTUNqVKlcrHeZ/t70/bt25czzzwzSd+BJEmlg8FWkqQ9+OYbmD17x/C69fH69dCiBTRrlkuFCnNZsGAY06YNon37KhxwQFUeeWQyt946h6ZNm3LooYfSrl27bV8bN25MKMDwrfemlSTp+xlsJUll2oYNsHAhLFjw7TF//rcBdvlyaNYsEWBbtIAjj4RLLoEmTbbw2Wfv8PzzA3nllVc45JBDOO+8c3n++cfZf//9t11/06ZNTJ8+nUmTJjF58mT+85//MHnyZL755hvatm27Q+Bt27Yt++6777a+r776KldeeSV9+vTh1ltvpXz58mn4LyRJUuYLMcZ015AUXbp0iWPHjk13GZKkDLJ5MyxatGNo3f5YuBDWroWGDaFRIzjggMTRuDE0bw4tWybOb82TOTk5vPfeewwcOJCXX36Zli1b8tOf/pSzzz670COpy5cvZ/LkydsC76RJk/jss8+oXbs27dq1o2rVqowePZr+/ft7b1pJkoAQwrgYY5ddvmawlSSVBjHCiBHQrx+MH58IrStWQP363wbWAw7YMcAecADUqQPl9rCVYk5ODh988AEDBw7kxRdfpFmzZpx77rmcc845NG7cOKnfQ15eHnPnzmXy5MnMmzeP3r17exsfSZLy7SnYOhVZkpTVFi2CJ5+EJ55IBNRLL4XLL08E2Pr1vx1tLYzc3Fw++ugjBg4cyKBBg2jcuDHnnnsuY8aMoUmTJsn+FrYpV64czZs3p3nz5il7D0mSSqOUBtsQQk/gXqA88EiM8a6dXv8FcAWQAywDLosxzst/rTfwu/ymf44x9ktlrZKk7LFpE7z6Kjz2WOI2Oj/5SSLYdu9evNvpALzyyitcd9111KtXj3PPPZcRI0bQrFmzpNQtSZJSI2XBNoRQHngA6AEsBMaEEAbHGKdu12w80CXGuD6EcC3wN+CnIYSawP8BXYAIjMvvuzJV9UqSMt+ECYkw++yz0K5dYnT2+eehWrXiX3vjxo3cfPPNvPrqqwwYMICjjjqq+BeVJEklIpUjtl2B2THGuQAhhOeAM4FtwTbGOGy79qOAC/Mfnwy8HWNckd/3baAn8GwK65UkZaDly6F/f3j88cSa2UsugdGjoWnT5L3HtGnTOO+882jZsiUTJkxgv/32S97FJUlSyu1hu4xiawgs2O75wvxzu3M5MLQwfUMIV4UQxoYQxi5btqyY5UqSMkVuLgwdmphi3KxZIsj+/e/w+edw++3JC7UxRh555BGOPvpo+vTpw8CBAw21kiRloYzYPCqEcCGJacc/KEy/GONDwEOQ2BU5BaVJkkrQihVwzz2JnY0bNUpMNX74YUhF1ly1ahVXX30106ZN48MPP6RNmzbJfxNJklQiUjliuwg4YLvnjfLP7SCEcCJwG3BGjHFTYfpKkkqPd96B9u0TU4/feisxSnvNNakJtSNHjqRjx47Url2b0aNHG2olScpyqRyxHQO0CCEcRCKUngdcsH2DEEJHoC/QM8a4dLuX3gT+EkLYevO+k4BbU1irJClNNmyAW2+FQYMSG0P16JG698rNzeXuu+/m3nvvpW/fvvzoRz9K3ZtJkqQSk7JgG2PMCSH0IRFSywOPxRg/CyHcAYyNMQ4G/g7sDTwfEvdnmB9jPCPGuCKE8CcS4Rjgjq0bSUmSSo8JE6BXL2jbFiZOhJo1U/deixcv5qKLLmLLli2MHTuWAw444Ps7SZKkrBBiLB1LU7t06RLHjh2b7jIkSQWQm5tYS3vPPfCvfyXCbXHvP7snQ4YM4fLLL+faa6/ld7/7HeXLl0/dm0mSpJQIIYyLMXbZ1WsZsXmUJKns+OILuPhiKFcOxo6FAw9M3Xtt2rSJW265hRdffJHnn3+eo48+OnVvJkmS0iaVm0dJkrRNjPDUU9C1K5x+Orz7bmpD7cyZMzn88MOZN28eEyZMMNRKklSKOWIrSUq5FSsSOxxPnZrY8bhDh9S9V4yRfv36cfPNN3PHHXdwzTXXEFI5z1mSJKWdwVaSlFJvvw2XXQbnnANPPglVqiT/PdauXcuoUaMYMWIE7733Hl9//TXvvfce7dq1S/6bSZKkjGOwlSSlxIYN8JvfwIsvwuOPw4knJu/a8+fP5+OPP952zJw5k06dOnHEEUfwy1/+kh49erDXXnsl7w0lSVJGM9hKkpJu/Hi48EJo1674t/HJyclh0qRJOwTZTZs2ceSRR3LkkUfSq1cvOnXqROXKlZP3DUiSpKxisJUkJU1uLvz97/DPfyZu43PBBYW/jc+aNWsYNWrUthD7ySef0KhRI4488kh69uzJn/70J5o3b+66WUmStI3BVpJUbCtWwIAB8MgjUL164jY+jRt/f7+8vDxmzJjBqFGjGDlyJKNGjWLu3Ll07tyZI488kptuuonDDz+cmsUZ8pUkSaWewVaSVCSbNsHrrydu4fPuu9CzJ9x+O5xyCpQvv+s+K1euZPTo0YwaNYpRo0YxevRoatSoweGHH0737t258sorad++PZUqVSrZb0aSJGU1g60kqcBihJEjE2H2+eehbVu46KLE5lD77rtj29zcXD777LMdRmMXLlxIly5d6N69O9dddx39+vWjXr166flmJElSqWGwlSR9r9mz4emnE0fFiokwO24cHHjgju1Gjx7N4MGDGTlyJGPHjqVBgwbbRmNvvPFG2rZtS4UK/uiRJEnJ5W8XkqRd2rpu9qmnEsH2/PPhueegc+ddbwg1ePBgrrzySq6++mpuvvlmunbtSq1atUq+cEmSVOYYbCVJ22zaBEOGJMLse+8l1s3+9rdw8smJkdrdeeONN7jiiisYMmQIhx12WMkVLEmShMFWkgTk5cH998Of/vTtutknnvjuutldeffdd7n44ot5+eWXDbWSJCktDLaSVMYtWgSXXALffAMffwwHH1zwvh999BHnn38+L7zwAkcccUTKapQkSdqTcukuQJKUPs8/D506wTHHwIcfFi7Ujho1irPPPptnnnmGY445JnVFSpIkfQ9HbCWpDFq9Gm64AUaNgldfha5dC9d/3LhxnHHGGTzxxBOceOKJqSlSkiSpgByxlaQy5qOPoEMHqFYNxo8vfKidOHEip556Kg8//DCnnnpqSmqUJEkqDEdsJamM2LwZ/u//oF8/eOgh+OEPC3+NqVOn0rNnT/7zn/9w5plnJr9ISZKkIjDYSlIZMHUqXHghNGoEEyZA3bqFv8bMmTPp0aMH99xzDz/5yU+SXqMkSVJRORVZkkqxGBO38fnBD+Daa+GVV4oWaufMmcMJJ5zAn/70J3r16pX8QiVJkorBEVtJKqUWL4bLLoNVq2DECGjRomjXmTdvHieccAK//e1vueyyy5JaoyRJUjI4YitJpdCgQYnb+Bx+OAwfXvRQu2jRIk444QRuuukmrr322uQWKUmSlCSO2EpSKbJmDdx4YyLMvvIKdOtW9Gt9+eWXnHDCCVx11VXceOONyStSkiQpyRyxlaRSYvjwxG18KlZM3ManOKF22bJlnHjiiVxwwQX8+te/TlqNkiRJqeCIrSRluTVr4Lbb4IUXoG9fOOOM4l1vxYoV9OjRgzPPPJPf//73ySlSkiQphRyxlaQs9sorcMghsGEDfPZZ8UPt6tWrOfnkkznhhBP485//TAghOYVKkiSlkCO2kpSFFi+Gn/0MJk+Gp56CY48t/jW//vprzjjjDLp3784999xjqJUkSVnDEVtJyiJ5efDgg9C+PbRuDRMnFj/Ujh8/nssvv5wWLVpw5JFHcu+99xpqJUlSVnHEVpKyxNSpcNVViXA7bBi0bVv0a23ZsoVBgwbxn//8h3nz5nHttdcyc+ZM6tSpk7yCJUmSSojBVpIy3MaN8Ne/wn//C3fcAVdfDeWKON/myy+/pG/fvvTt25eDDz6Ym266iTPPPJMKFfxxIEmSspe/yUhSBvvww8QobZs2MGECNGxY+GvEGBk9ejT3338/r7/+Oueeey5vvvkm7dq1S3q9kiRJ6WCwlaQMtHIl3HILDB0K990HP/5x4a+xceNGBgwYwP3338/KlSu5/vrr+c9//kONGjWSX7AkSVIaGWwlKYPECM8/Dz//eSLMTpkC++5buGssWLCA//3vfzzyyCN06tSJ22+/nVNOOYVyRZ2/LEmSlOEMtpKUIebPh+uvh88/hxdegCOOKFz/4cOH869//Ythw4Zx0UUXMXz4cFq2bJmaYiVJkjKIf76XpDSLEf73P+jUCbp3h08/LVyoXb16NZdffjm9evXihBNOYN68edx7772GWkmSVGakNNiGEHqGEGaEEGaHEH6zi9ePCSF8GkLICSGcs9NruSGECfnH4FTWKUnpsnQpnHEGPPooDB8Ot90GlSoVvP/WTaAqVqzIlClTuO6666hevXrqCpYkScpAe5yKHEJYC8RdvQTEGOM+e+hbHngA6AEsBMaEEAbHGKdu12w+cAnwq11cYkOMscMeq5ekLPbGG3DZZdC7NwwaVLhAu2bNGn75y1/y1ltv8eijj9KjR4/UFSpJkpTh9hhsY4zF+bN/V2B2jHEuQAjhOeBMYFuwjTF+kf9aXjHeR5KyysaN8JvfwIsvwjPPwLHHFq7/22+/zRVXXMHJJ5/M5MmT2Wef3f6NUZIkqUz4vhHbmnt6Pca4Yg8vNwQWbPd8IdCt4KVRJYQwFsgB7ooxvlyIvpKUkaZMgfPPh9atYeJEKMydd9asWcPNN9/M0KFDefjhhzn55JNTV6gkSVIW+b5dkceRmIocdvFaBJomvaJvHRhjXBRCaAq8F0KYHGOcs32DEMJVwFUAjRs3TmEpklQ8McJ//gN33AF//3ti+nHY1b+su/HOO+9wxRVXcOKJJzJ58mT2Lew9gCRJkkqx75uKfFAxrr0IOGC7543yzxVIjHFR/te5IYT3gY7AnJ3aPAQ8BNClS5ddrQWWpLT78ku49FJYsQJGjoTmzQved+3atfz6179myJAhPPTQQ/Ts2TN1hUqSJGWpAu+KHEKoEULomr+T8TEhhGO+p8sYoEUI4aAQQiXgPKBAuxvnv1fl/Me1gSPZbm2uJGWL116Djh2hS5fErseFCbXvvfcehx56KJs2bWLSpEmGWkmSpN34vqnIAIQQrgBuJDHqOgHoDowEjt9dnxhjTgihD/AmUB54LMb4WQjhDmBsjHFwCOEw4CWgBnB6COH2GOMhQGugb/6mUuVIrLE12ErKGhs2wM03J4LtwIFw9NEF77tu3TpuueUWBg8eTN++fTn11FNTV6gkSVIpUKBgSyLUHgaMijEeF0JoBfzl+zrFGF8HXt/p3B+2ezyGRFjeud8IoF0Ba5OkjDJxYmKDqA4dYMIE2G+/gvd9//33ueyyy/jBD37A5MmT2a8wnSVJksqoggbbjTHGjSEEQgiVY4zTQwgHp7QyScoyeXnw73/DX/8K//oX9OpV8A2ivvnmG2655RZefvll+vbty2mnnZbSWiVJkkqTggbbhSGE/YCXgbdDCCuBeakqSpKyzeLFcMklsG4dfPIJHFSIrffWrVtHz549OeCAA5g8eTI1CnMPIEmSJBVs86gY449jjKtijH8Efg88CvwohXVJUtaYNg06d4Yjj4QPPyxcqN2wYQNnnHEGrVq1on///oZaSZKkIijo5lHdgc9ijGtjjB+EEPYhcfud0SmtTpIy3OLFcMopcPfdcPHFheu7adMmfvzjH7P//vvTt29fypUr8Eb1kiRJ2k5Bf4v6H7Buu+fr8s9JUpm1di2cdhpceWXhQ+2WLVs499xzqV69Ok888QTly5dPTZGSJEllQEGDbYgxxq1PYox5FHx9riSVOlu2wDnnQNeu8NvfFq5vTk4OvXr1IsZI//79qVDBf04lSZKKo6DBdm4I4WchhIr5x43A3FQWJkmZKka46iqoVAkeeKDgOx8D5OXlcdlll7Fq1SoGDhxIpUqVUleoJElSGVHQYHsNcASwCFgIdAOuSlVRkpTJ/u//4LPP4LnnoDCDrTFGrrnmGubPn8/LL79MlSpVUlekJElSGVKgX8lijEuB81JciyRlvIcfhmeegREjoFq1gveLMXLjjTcyZcoU3nrrLapWrZq6IiVJksqYAo3YhhBahhDeDSFMyX9+aAjhd6ktTZIyy5Ah8Pvfw9ChULduwfvFGLnlllsYMWIEQ4cOZe+9905dkZIkSWVQQaciPwzcCmwBiDFOwhFcSWXI2LFwySXw8svQokXh+v7xj3/kzTff5M0332TfffdNRXmSJEllWkFXh1WNMX4SdtwhJScF9UhSxpk7F844Ax59FLp3L1zfv/71rzz//PO8//771KpVKzUFSpIklXEFDbZfhxCaAREghHAOsCRlVUlShvj6a+jZMzEF+YwzCtf33//+N4899hgffPABdQszd1mSJEmFUtBgez3wENAqhLAI+BzolbKqJCkDrF8Pp58OZ58N115buL4PPvgg9957Lx988AH7779/agqUJEkSUPBdkecCJ4YQqpFYl7uexBrbeSmsTZLSJjcXevWCZs3gzjsL1/fxxx/nzjvv5P3336dx48apKVCSJEnb7HHzqBDCPiGEW0MI/wkh9CARaHsDs4FzS6JASSppMcKNN8KaNfDYY1CuoNvsAc8++yy33XYb77zzDs2aNUtdkZIkSdrm+0ZsnwJWAiOBK4HbgAD8OMY4IbWlSVJ63HMPfPghfPQRVKpU8H4vvvgiN910E++88w4HH3xw6gqUJEnSDr4v2DaNMbYDCCE8QmLDqMYxxo0pr0yS0uDZZ+H++2HECCjMnXlef/11rr32WoYOHUrbtm1TV6AkSZK+4/sm2G3Z+iDGmAssNNRKKq2GDUtMQX79dWjUqOD9nnzySS655BIGDx5Mp06dUlegJEmSdun7RmzbhxDW5D8OwF75zwMQY4z7pLQ6SSohkyfDT38KAwZAQQdcc3JyuOWWW3jllVd4//33adOmTWqLlCRJ0i7tMdjGGMuXVCGSlC4LF8Jpp8G998JxxxWsz8qVKzn//PPJzc3lk08+oWbNmqktUpIkSbtViL0+Jan0WbUKTj0V+vSB888vWJ/p06fTrVs3WrVqxdChQw21kiRJaWawlVRmbdoEP/4xHHss3Hxzwfq8/vrrHHPMMfzmN7/h3//+NxUqFOh24JIkSUohfyOTVCbl5UHv3lCrFvzrXxDCntvHGPn73//Ovffey8svv8wRRxxRMoVKkiTpexlsJZVJN98MixfDW29B+e/ZTWDDhg1ceeWVTJ8+nVGjRnHAAQeUTJGSJEkqEKciSypz/vUveOMNeOUVqFJlz20XLVrEMcccQ25uLh9++KGhVpIkKQMZbCWVKQMGwD//CUOHQo0ae247atQounbtytlnn80zzzxD1apVS6ZISZIkFYpTkSWVGe+/DzfcAO+8A40b77ltv379uPnmm3nsscf44Q9/WCL1SZIkqWgMtiVkzJgxbNq0iaOOOirdpUhl0uTJcO658NxzcOihu2+Xk5PDr3/9a1599VXef/992rRpU3JFSpIkqUgMtiVk3LhxjB071mArpcGCBXDaaXDvvXD88btvt3LlSn76058SY2T06NHen1aSJClLuMa2hLRu3Zpp06aluwypzFm5Ek45BW68Ec4/f/ftpk2bRteuXTnkkEMYOnSooVaSJCmLGGxLyNZgG2NMdylSmbFxI/zoR9CjB/ziF7tv9+KLL/KDH/yA3/72t/zrX/+iQgUns0iSJGUTf3srIXXq1KFcuXJ89dVX1K9fP93lSKVeXh5cdBHUrw//+AeE8N02Cxcu5IYbbmDq1KkMHjyY7t27l3yhkiRJKjZHbEtICMHpyFIJiRFuugmWLYN+/aDcTv/S5ebmct9999GxY0c6dOjApEmTDLWSJElZzBHbErQ12B533HHpLkUq1f7xD3j3XRg+HKpU2fG18ePHc9VVV1GtWjU++ugjWrVqlZ4iJUmSlDSO2JagNm3aMHXq1HSXIZVqzz4L990HQ4fCfvt9e37dunX88pe/pGfPnlx33XUMGzbMUCtJklRKGGxLkFORpdR6773E7sdDhsABB3x7/rXXXuOQQw7h66+/ZsqUKVx66aWEXS26lSRJUlZKabANIfQMIcwIIcwOIfxmF68fE0L4NISQE0I4Z6fXeocQZuUfvVNZZ0kx2EqpM3EinHceDBwI7dolzi1evJhzzjmHX/ziFzz++OP069ePOnXqpLdQSZIkJV3Kgm0IoTzwAHAK0AY4P4TQZqdm84FLgGd26lsT+D+gG9AV+L8QQo1U1VpSDjjgANauXcuqVavSXYpUqsybB6edBv/5Dxx7bGJzqAceeID27dvTpk0bJk2axPHHH5/uMiVJkpQiqdw8qiswO8Y4FyCE8BxwJrBtkWmM8Yv81/J26nsy8HaMcUX+628DPYFnU1hvyoUQaNWqFdOmTePwww9PdzlSqbByJZxyCvzqV3DuuTBx4kSuuuoqKlWqxAcffECbNjv/PU2SJEmlTSqnIjcEFmz3fGH+uaT1DSFcFUIYG0IYu2zZsiIXWpJat27tBlJSkmzeDGedBSedBFde+Q2//vWv6dGjB1dddZWhVpIkqQzJ6s2jYowPxRi7xBi7ZMu6uTZt2rjOVkqCGOHqq2GffeCEE16nbdu2LF68mClTpnD55ZdTbueb10qSJKnUSuVvfouA7fYlpVH+uVT3zWhuICUlx1//CpMmwfnnv8p1113NQw89xNNPP03dunXTXZokSZJKWCrX2I4BWoQQDiIRSs8DLihg3zeBv2y3YdRJwK3JL7HkORVZKr4BA+DBB+H111dw0klX8/zzz3PkkUemuyxJkiSlScpGbGOMOUAfEiF1GjAwxvhZCOGOEMIZACGEw0IIC4GfAH1DCJ/l910B/IlEOB4D3LF1I6ls17RpU7788kvWr1+f7lKkrDRiBPTpA6++CnfeeT3nnXeeoVaSJKmMS+WILTHG14HXdzr3h+0ejyExzXhXfR8DHktlfelQoUIFmjdvzowZM+jYsWO6y5Gyyty5cPbZ0K8fzJnzIp9++injx49Pd1mSJElKs5QGW+1amzZtmDp1qsFWKoSVK+HUU+F3v4PDDlvGoYdez6BBg6hatWq6S5MkSVKaGWzTwA2kpMLZvDkxUtuzJ1x/Pfz0p33o1asXRxxxRLpLkyRJUgbwfhhpYLCVCi5GuOYa2Htv+Mc/4IUXXmDixIn86U9/SndpkiRJyhCO2KbB1qnIkr7fX/8KEybAhx/CihXLuOGGG3jppZfYa6+90l2aJEmSMoTBNg1atmzJ559/zpYtW6hYsWK6y5Ey1tbb+owcmRixveyy67nooovo3r17ukuTJElSBnEqchpUrlyZAw44gNmzZ6e7FCljjRz57W19GjaEgQMHMnnyZO644450lyZJkqQMY7BNE6cjS7s3dy6cdVbitj7t28PSpUv52c9+xhNPPEGVKlXSXZ4kSZIyjME2TdxAStq1lSvhtNPgttsSt/eJMXLddddxySWX0K1bt3SXJ0mSpAzkGts0ad26NW+++Wa6y5AyyubNcM45cPLJiWnIAAMGDGDatGk8/fTT6S1OkiRJGcsR2zRp06aNI7bSdmKEa6+FatUSt/UB+Oqrr7jxxht5/PHHnYIsSZKk3XLENk1atWrFjBkzyM3NpXz58ukuR0q7u+6C8eMTt/UpXz4xBfnaa6/lsssuo2vXrukuT5IkSRnMYJsm1atXp1atWsybN4+mTZumuxwprQYOhP/+F0aNStzWB+C5555j5syZPPvss+ktTpIkSRnPqchp5HRkKRFmr7/+29v6AHz55Zf8/Oc/54knnqBy5crpLVCSJEkZz2CbRu6MrLJu3Dg480x44gno0CFxLsbINddcwxVXXEGXLl3SWZ4kSZKyhFOR06h169aMHj063WVIaTFmDPzwh9C3b+L2Pls988wzzJkzhwEDBqSvOEmSJGUVR2zTyKnIKqtGj06E2UcegR/96NvzS5Ys4aabbnIKsiRJkgrFYJtGW6cixxjTXYpUYkaOhNNPh8cfT3zdKsbI1VdfzdVXX03nzp3TV6AkSZKyjlOR06h27dpUrFiRJUuWsP/++6e7HCnlRoxIjND26wennLLja08//TRffPEFL7zwQlpqkyRJUvZyxDbNnI6ssmL48ESofeqp74baxYsX88tf/pInnniCSpUqpaU+SZIkZS+DbZq5M7LKgg8/hLPOgmeegZNP3vG1GCNXXXUV1157LZ06dUpPgZIkScpqTkVOszZt2jB16tR0lyGlzPvvw7nnwnPPwfHHf/f1J598koULF/Liiy+WeG2SJEkqHRyxTTNHbFWavftuItQOHLjrULtgwQJ+9atfOQVZkiRJxWKwTTODrUqrt9+G88+HF16AY4/97ut5eXlccskl3HTTTXTo0KGky5MkSVIpYrBNs4YNG7J+/XpWrFiR7lKkpHnzTejVC158EY45Ztdt7r//fjZs2MCvf/3rki1OkiRJpY7BNs1CCI7aqlQZOhQuughefhmOOmrXbaZNm8af/vQnnnzySSpUcKm/JEmSisdgmwFat27tBlIqFV57DXr3hsGD4Ygjdt1my5YtXHTRRfz5z3+mefPmJVugJEmSSiWDbQbwXrYqDQYPhssvT4Tb7t133+7OO++kTp06XH311SVXnCRJkko15wBmgNatWzNs2LB0lyEV2csvw9VXw5Ah0KXL7tt98skn/O9//2P8+PGEEEqsPkmSJJVujthmAKciK5sNGpQItUOH7jnUrl+/nosvvpj77ruP/fffv+QKlCRJUqlnsM0ABx10EMuWLWPdunXpLkUqlGHD4Prr4Y03oFOnPbf9zW9+Q6dOnfjpT39aMsVJkiSpzHAqcgYoX748LVq0YMaMGXTu3Dnd5UgF8s03iTW1jz0GHTvuue0777zDSy+9xKRJk0qmOEmSJJUpjthmCKcjK9vcdlvidj6nnrrndqtWreKyyy7j0UcfpUaNGiVTnCRJksoUR2wzhDsjK5t8/DEMHAhTpnx/2z59+nDGGWdw0kknpb4wSZIklUkG2wzRunVrnnnmmXSXIX2vjRsTU5Dvvx9q1txz2+eff55PPvmE8ePHl0xxkiRJKpMMthnCqcjKFrffDu3awdln77ndkiVL6NOnD4MHD6ZatWolU5wkSZLKJINthmjRogXz5s1j8+bNVKpUKd3lSLs0blxis6jv2wMqxsgVV1zBVVddRbdu3UqmOEmSJJVZKd08KoTQM4QwI4QwO4Twm128XjmEMCD/9dEhhCb555uEEDaEECbkHw+mss5MULlyZQ488EBmzZqV7lKkXdq8GS67DP7xD6hXb89tH374Yb788kt+//vfl0xxkiRJKtNSNmIbQigPPAD0ABYCY0IIg2OM28+3vRxYGWNsHkI4D7gb2HqTyzkxxg6pqi8TbZ2OfMghh6S7FOk77roLGjWCXr323G7OnDn89re/5cMPP3T2gSRJkkpEKkdsuwKzY4xzY4ybgeeAM3dqcybQL//xC8AJIYSQwpoymjsjK1NNmZLYLKpvX9jT/6G5ubn07t2b2267jTZt2pRcgZIkSSrTUhlsGwILtnu+MP/cLtvEGHOA1UCt/NcOCiGMDyF8EEI4OoV1Zgw3kFImyslJTEG+887EiO2e3HPPPVSsWJEbb7yxZIqTJEmSSPEa22JYAjSOMXYEfgE8E0LYZ+dGIYSrQghjQwhjly1bVuJFJpsjtspE//43VK8OV16553YTJ07knnvu4YknnqBcuUz9p0WSJEmlUSp/+1wEHLDd80b553bZJoRQAdgXWB5j3BRjXA4QYxwHzAFa7vwGMcaHYoxdYoxd6tSpk4JvoWS1atWKWbNmkZubm+5SJABmzkysrX344T1PQd60aRMXXXQR99xzDwceeGDJFShJkiSR2mA7BmgRQjgohFAJOA8YvFObwUDv/MfnAO/FGGMIoU7+5lOEEJoCLYC5Kaw1I1SrVo06derw+eefp7sUibw8uOIK+P3voWnTPbf9wx/+QLNmzbj44otLpjhJkiRpOynbFTnGmBNC6AO8CZQHHosxfhZCuAMYG2McDDwKPBVCmA2sIBF+AY4B7gghbAHygGtijCtSVWsm2ToduXnz5ukuRWXc//6XWF/bp8+e23300Uc8+eSTTJo0iTK895skSZLSKGXBFiDG+Drw+k7n/rDd443AT3bRbxAwKJW1ZarWrVszbdo0Tj/99HSXojLsiy/gj3+Ejz6C8uV3327t2rX07t2bvn37UhqWA0iSJCk7ucNLhnFnZKVbjHDVVfCLX0CrVntu+/Of/5zjjjuOM844o2SKkyRJknYhpSO2Krw2bdrwyCOPpLsMlWFPPAHLl8OvfrXndi+99BLvv/8+EyZMKImyJEmSpN0y2GaYrVORY4yuV1SJW7wYbrkF3n4bKlbcfbslS5Zw7bXX8tJLL1G9evWSK1CSJEnaBaciZ5iaNWuy1157sWjRzndGklIrRrj2WrjmGmjffk/tIpdddhlXXXUVhx9+eMkVKEmSJO2GwTYDbd0ZWSpJAwbA7Nlw2217bvff//6X5cuX8/vf/75kCpMkSZK+h8E2A22djiyVlGXL4Oc/h8ceg8qVd99u2rRp/N///R9PP/00Ffc0V1mSJEkqQQbbDOTOyCppP/sZXHghdOu2+zabN2/mwgsv5M4776Rly5YlV5wkSZL0PQy2GcipyCpJr7wCY8fCHXfsud0f//hH9t9/f6666qqSKUySJEkqIHdFzkBORVZJWbUKrr8ennkGqlbdfbuPPvqIxx9/nAkTJrhbtyRJkjKOI7YZqEGDBmzatImvv/463aWoFIsxsa72zDPhmGN232716tVcfPHFPPTQQ9SrV6/E6pMkSZIKymCbgUIITkdWyv3znzBuHNx1157b/exnP+Pkk0/m9NNPL5nCJEmSpEJyKnKG2rqB1NFHH53uUlQKvfAC/PvfMGIEVK+++3bPP/88I0eOZPz48SVWmyRJklRYBtsM5TpbpcqIEXDddfDWW3DAAbtvt2jRIvr06cOrr75KtWrVSq5ASZIkqZCcipyhnIqsVJg1C846C558Ejp02H27vLw8LrnkEq6//nq6du1aYvVJkiRJRWGwzVDey1bJtmwZnHIK/PnP0LPnntved999rFu3jt/+9rclU5wkSZJUDE5FzlAHHnggK1asYO3atVTf0yJIqQA2bIAzzoCf/hSuuGLPbadMmcKdd97JqFGjqFDBfyIkSZKU+RyxzVDly5enZcuWTJ8+Pd2lKMvl5sKFF0LTponR2j3ZtGkTvXr14q677qJZs2YlU6AkSZJUTAbbDOZ0ZCXDzTfDihXw2GMQwp7b/u53v6Np06ZcdtllJVOcJEmSlATOM8xgbiCl4rrvPnjjDfj4Y6hcec9thw0bxjPPPMPEiRMJ35eAJUmSpAziiG0G85Y/Ko6XX4a774bXX4caNfbcdtWqVfTu3ZtHHnmE2rVrl0h9kiRJUrI4YpvBnIqsovrkE7jqqkSobdLk+9tff/31nHHGGZxyyikpr02SJElKNoNtBmvevDkLFixg48aNVKlSJd3lKEvMnQs/+lFiTW2XLt/f/plnnuHTTz9l3LhxKa9NkiRJSgWnImewSpUqcdBBBzFr1qx0l6IssXw5nHoq/O538MMffn/7+fPn8/Of/5z+/ftTtWrV1BcoSZIkpYDBNsM5HVkFtXFjYqT2jDPguuu+v/3SpUv56U9/yk033USnTp1SXp8kSZKUKgbbDOfOyCqIvDy45BLYf3+4667vb//hhx/SqVMnjj32WH7961+nvD5JkiQplVxjm+Fat27N4MGD012GMtxvfwsLF8I770C5Pfy5Ki8vj7vvvpt7772Xxx9/3M2iJEmSVCoYbDNc69atuasgQ3Aqsx58EF56CUaMgD3tMbZs2TIuvvhi1q5dy9ixY2nUqFHJFSlJkiSlkFORM1yrVq2YPXs2OTk56S5FGWjIELj99sRtfWrV2n274cOH06lTJw499FCGDRtmqJUkSVKpYrDNcFWrVqV+/frMnTs33aUog8SYCLWXXpoYrW3WbNfttk49Pvvss3nwwQe5++67qVixYskWK0mSJKWYU5GzQOvWrZk2bRotW7ZMdynKACNGwK23wtKl8Oyz0L37rtt9/fXX9O7dm5UrVzJmzBgaN25csoVKkiRJJcQR2yzgzsgCmDQJTj8dzj8/sQPy5Mlwwgm7bvvxxx/TqVMnDjnkED744ANDrSRJkko1g20W8F62ZducOdCrF5x0Epx4IsycmZiCXGEX8y3y8vL4+9//zllnncUDDzzA3/72N6ceS5IkqdQz2GaBrVORVbYsXgzXXgvdukGrVjBrFtx4I1SuvOv2y5cv58wzz2TQoEF88sknnH766SVbsCRJkpQmBtss0Lp1a6ZPn06MMd2lqASsWAG33ALt2sHee8OMGfD730P1/2/v3oM8u+s64b8/09OTmcxkcp0MuRIjyUK8RRgCArqRAjcolbAszxJdfXDFCrpYiz76UF6qFtddF62t6IKyYsQ8oCUii1ziKihkEeIFyATCNURDBJJIZibJ3O/T/X3++P2609PpnunMTPf5nZnXq+rUOed7Tp/fZ6ZO8uv3fL/ne86Y/2c+8YlP5JnPfGauvPLKfPzjH89Tn/rUpSsYAAA6Jtj2wNlnn501a9bkgQce6LoUFtGuXcmv/mpy5ZXJtm2DZ2r/+38/8mt8Wmu5+eabc8MNN+TNb35zbr755qxYsWLJagYAgFFgVuSemJpAyiRAJ5/9+5Nbbkn+239Lrr02+fu/T6644ug/d//99+dnfuZn8vDDD+eTn/xkLrvsssUuFQAARpIe257wnO3JZ2Iiecc7kn/xL5IPfSj54AcHr++ZL9Q+8sgjefe7352bbropl19+eZ7//OfnGc94Ru644w6hFgCAU9qi9thW1XVJ3pRkLMnbWmu/Nuv4aUn+IMmzkjya5JWtta8Oj/1CklcnmUjyH1trf7mYtY66ZzzjGfnsZz/bdRkco9aShx8evKJnavmbv0kuuCD5wz9Mvvu7n/gze/bsyR133JGPfOQj+chHPpL7778/3/M935MXvehFed3rXperrroqVbX0fxgAABgxixZsq2osyVuSvDjJg0nurKrbWmsz31vz6iRbW2tPq6obk/x6kldW1VVJbkzyLUkuTPKRqrqytTaxWPWOuquuuirvete7ui6DBdi5M/niFw8PsZ///ODYt33bYHn+85Of+qnkWc9KprLpoUOHsnHjxnzkIx/J7bffnjvvvDPPfOYz86IXvShvectb8uxnP9urewAAYA6L2WN7TZL7Wmv3J0lVvSvJDUlmBtsbkvzycPs9SX67Bl1QNyR5V2ttf5J/qqr7htf7+0Wsd1H97d8mH/tYsmpVcvrpg/VCtleuTJYte/xdtq01vXQj4uDBwSt4ZgfYhx9OnvGMx0PsS186WD/lKY+H2GQw8dOXv/zl6SD713/913nqU5+aF73oRXn961+f7/7u786aNWu6+wMCAEBPLGawvSjJzGl8H0zynPnOaa0dqqrtSc4dtn9i1s9etHilLr6JiUFP3ubNyZ49yd69g2W+7an9/fsH7y1dtWp9tm//QsbHv9H1H4UkSWVy8qyMjX0j4+NfzvLlX874+GA555yvZfPmydx+e3L77fNfYe/evVm9enVe/OIX55WvfGV+93d/N+vXr1+6PwIAAJwkej0rclXdlOSmJCM/W/D3fM9gebImJwfhds+eyje+sSq7d+858cVxDFrOPvvRrFo1nuTbhsuTMz4+nvXr1+uBBwCA47SYwfahJJfM2L942DbXOQ9W1fIkZ2YwidRCfjattVuS3JIkGzZsaCes8hGybNnjw5PPPfesJGd1XBEAAMBoWczX/dyZ5Iqq+qaqWpHBZFC3zTrntiSvGm6/Isn/aa21YfuNVXVaVX1TkiuSfGoRawUAAKCnFq3HdvjM7E8l+csMXvdza2vti1X1K0k2ttZuS/L7Sf5wODnUYxmE3wzPe3cGE00dSvLaU3lGZAAAAOZXgw7S/tuwYUPbuHFj12UAAACwCKrqrtbahrmOLeZQZAAAAFh0gi0AAAC9JtgCAADQa4ItAAAAvSbYAgAA0GuCLQAAAL0m2AIAANBrJ817bKtqS5KvLeJHnJfkkUW8PhwL9yWjyH3JqHFPMorcl4yiUb8vn9paWzfXgZMm2C62qto438uAoSvuS0aR+5JR455kFLkvGUV9vi8NRQYAAKDXBFsAAAB6TbBduFu6LgDm4L5kFLkvGTXuSUaR+5JR1Nv70jO2AAAA9JoeWwAAAHpNsAUAAKDXBNsFqKrrqureqrqvqn6+63o4NVXVrVW1uaq+MKPtnKr6cFX943B9dpc1cmqpqkuq6qNV9aWq+mJVvW7Y7r6kM1W1sqo+VVWfHd6X/3nY/k1V9cnhd/mfVNWKrmvl1FJVY1X1mar638N99ySdqqqvVtXnq+ruqto4bOvtd7hgexRVNZbkLUlekuSqJD9YVVd1WxWnqLcnuW5W288nub21dkWS24f7sFQOJfnZ1tpVSZ6b5LXD/z+6L+nS/iQvbK19R5Krk1xXVc9N8utJfrO19rQkW5O8ursSOUW9Lsk9M/bdk4yC722tXT3j3bW9/Q4XbI/umiT3tdbub60dSPKuJDd0XBOnoNbax5M8Nqv5hiTvGG6/I8nLlrImTm2ttW+01j493N6ZwS9sF8V9SYfawK7h7vhwaUlemOQ9w3b3JUuqqi5O8gNJ3jbcr7gnGU29/Q4XbI/uoiQPzNh/cNgGo2B9a+0bw+2Hk6zvshhOXVV1WZLvTPLJuC/p2HDI591JNif5cJKvJNnWWjs0PMV3OUvtfyR5fZLJ4f65cU/SvZbkr6rqrqq6adjW2+/w5V0XAJwYrbVWVd7fxZKrqjVJ/jTJT7fWdgw6Igbcl3ShtTaR5OqqOivJ+5I8vduKOJVV1UuTbG6t3VVV13ZcDsz0gtbaQ1V1fpIPV9WXZx7s23e4HtujeyjJJTP2Lx62wSjYVFUXJMlwvbnjejjFVNV4BqH2j1pr7x02uy8ZCa21bUk+muS7kpxVVVP/oO+7nKX0/CTXV9VXM3ik7YVJ3hT3JB1rrT00XG/O4B8Br0mPv8MF26O7M8kVw5nrViS5McltHdcEU25L8qrh9quSfKDDWjjFDJ8R+/0k97TWfmPGIfclnamqdcOe2lTVqiQvzuD5748mecXwNPclS6a19guttYtba5dl8Hvk/2mt/bu4J+lQVa2uqjOmtpN8X5IvpMff4dVab3qXO1NV35/BsxFjSW5trf1qtxVxKqqqP05ybZLzkmxK8oYk70/y7iSXJvlakn/bWps9wRQsiqp6QZI7knw+jz839osZPGfrvqQTVfXtGUx4MpbBP+C/u7X2K1V1eQa9Zeck+UySH26t7e+uUk5Fw6HIP9dae6l7ki4N77/3DXeXJ3lna+1Xq+rc9PQ7XLAFAACg1wxFBgAAoNcEWwAAAHpNsAUAAKDXBFsAAAB6TbAFAACg1wRbAAAAek2wBQAAoNcEWwAAAHpNsAUAAKDXBFsAAAB6TbAFAACg1wRbAAAAek2wBQAAoNcEWwAAAHpNsAUAAKDXBFsAAAB6TbAFAACg1wRbAAAAek2wBQAAoNcEWwAAAHpNsAUAAKDXBFsAAAB6TbAFAACg1wRbAAAAek2wBQAAoNcEWwAAAHpNsAUAAKDXBFsAAAB6TbAFAACg1wRbAAAAek2wBQAAoNcEWwAAAHpNsAUAAKDXBFsAAAB6TbAFAACg1wRbAAAAek2wBQAAoNcEWwAAAHpNsAUAAKDXBFsAAAB6bXnXBZwo5513Xrvsssu6LgMAAIBFcNdddz3SWls317GTJthedtll2bhxY9dlAAAAsAiq6mvzHTMUGQAAgF4TbAEAAOg1wRYAAIBeW/JgW1WXVNVHq+pLVfXFqnrdHOdcW1Xbq+ru4fKflrpOAAAA+qGLyaMOJfnZ1tqnq+qMJHdV1Ydba1+add4drbWXdlAfAAAAPbLkPbattW+01j493N6Z5J4kFy11HQAAAJwcOn3GtqouS/KdST45x+HvqqrPVtUHq+pb5vn5m6pqY1Vt3LJly2KWCgAAwIjqLNhW1Zokf5rkp1trO2Yd/nSSp7bWviPJbyV5/1zXaK3d0lrb0FrbsG7dnO/pBQAA4CTXSbCtqvEMQu0ftdbeO/t4a21Ha23XcPsvkoxX1XlLXCYAAAA90MWsyJXk95Pc01r7jXnOecrwvFTVNRnU+ejSVXni7du3L9u2beu6DAAAgJNOFz22z0/yI0leOON1Pt9fVT9RVT8xPOcVSb5QVZ9N8uYkN7bWWge1njC/93u/l5e//OU5ePBg16UAAACcVKrneXHahg0b2saNG7suY14TExN52ctelgsvvDBvfetbM+yQBgAAYAGq6q7W2oa5jnU6K/KpZGxsLO985zvzd3/3d3nzm9/cdTkAAAAnjeVdF3AqOeOMM/Jnf/Zned7znpcrr7wyL3nJS7ouCQAAoPf02C6xyy67LO95z3vyqle9Kl/84he7LgcAAKD3BNsOPO95z8vNN9+c66+/Plu2bOm6HAAAgF4TbDvyIz/yI3nlK1+Zl7/85dm/f3/X5QAAAPSWYNuh//pf/2vOO++8/ORP/mROltmpAQAAlppg26Fly5blD//wD/OZz3wmN998c9flAAAA9JJZkTu2Zs2a3HbbbXnuc5+bK6+8Mtdff33XJQEAAPSKHtsRcMkll+R973tfXv3qV+dzn/tc1+UAAAD0imA7Iq655pr81m/9Vq6//vps2rSp63IAAAB6Q7AdITfeeGNe9apX5V//63+dffv2dV0OAABALwi2I+YNb3hDLr744vz4j/+4mZIBAAAWQLAdMcuWLcvb3/723HvvvXnjG9/YdTkAAAAjz6zII+j000/PBz7wgTznOc/J05/+9Lz85S/vuiQAAICRpcd2RF144YV5//vfn9e85jX59Kc/3XU5AAAAI0uwHWHPetaz8ju/8zt52cteln/+53/uuhwAAICRJNiOuFe84hW56aabcsMNN2TPnj1dlwMAADByBNse+KVf+qVcccUV+dEf/VGvAQIAAJhFsO2Bqsqtt96aQ4cO5Zu/+Ztz8803Z9euXV2XBQAAMBIE255YuXJl3vve9+bP//zP86lPfSqXX355fvmXfzmPPvpo16UBAAB0SrDtmauvvjp/8id/kr/5m7/Jgw8+mCuuuCI/93M/Z3IpAADglCXY9tSVV16Zt73tbfnc5z6XiYmJfOu3fmte85rX5Ctf+UrXpQEAACwpwbbnLr744vzmb/5m7r333px//vl5znOekx/6oR/K5z//+a5LAwAAWBKC7Uli3bp1+S//5b/k/vvvz9VXX53v+77vy/XXX59PfOITXZcGAACwqATbk8zatWvz+te/Pvfff3+uu+663Hjjjfne7/3efPjDH05rrevyAAAATjjB9iS1atWq/If/8B/yj//4j/mxH/uxvO51r8s111yTW2+9Nffcc08mJye7LhEAAOCEqJOlF2/Dhg1t48aNXZcxsiYnJ/OBD3wg73rXu3LnnXfm0UcfzTOf+cw8I0WgWQAAHPRJREFU+9nPzrOf/excc801ufTSS1NVXZcKAADwBFV1V2ttw5zHBNtT0yOPPJKNGzfmzjvvnF4mJiamg+7Ucv7553ddKgAAgGDL0bXW8tBDD+VTn/rUdNDduHFjzjzzzFxzzTXTQfdZz3pW1q5d23W5AADAKUaw5ZhMTk7mvvvuO6xX9+67787atWtzySWX5JJLLsmll176hPX69euzbJnHtwEAgBNHsOWEmZiYyMMPP5yvf/3reeCBB/LAAw9Mb0+tt23blgsvvHA67M4OvhdddFHOPvts4RcAAFgwwZYltW/fvjz44IOHhd2Z2w8++GB2796dc845J+vWrVvQcu6552b58uVd/9EAAICOHCnYSgqccCtXrszTnva0PO1pT5v3nAMHDuTRRx/Nli1bnrB84QtfeELb1q1bc+aZZ04H3fPPPz/r16/P+vXrD9ueWtasWWOGZwAAOEUItnRixYoVueCCC3LBBRcs6PyJiYk89thj00F38+bN2bRpUzZv3py77747mzZtml42b96cycnJOQPvVBA+//zzc8455+Sss87K2WefnTPOOEMQBgCAnhJs6YWxsbHp3tqF2L179xPC7qZNm3Lvvffm4x//eDZv3pxt27Zl69at2bZtW/bs2ZMzzzwzZ5111nTYPdr6zDPPzJlnnpm1a9dm9erVgjEAAHREsOWktHr16lx++eW5/PLLF3T+oUOHsn379mzdunU67M5ef/3rX5/e37p1a7Zv354dO3Zk+/btOXDgQNauXZu1a9dOh9251kc6tnbt2qxYsWKR/2YAAODkI9hCkuXLl+fcc8/Nueeee0w/f/DgwezYsWM66M63fuihh454fPny5UcNxzPXU73Fa9asyZo1a6a3V69enfHx8RP8twQAAKNJsIUTYHx8/LiCcZK01rJv3745A+/M7U2bNuUf/uEfpoP07t27s2vXruzatWt6e/fu3RkbGzss7B5p+4wzzjjqsmbNmoyNjZ3AvzUAADgxBFsYEVWVVatWZdWqVXnKU55yXNdqrWX//v2Hhd25tnfu3Jldu3blkUceyT/90z9l165d2blz55zL7t27s3LlyumQOzP0nn766Vm9enVOP/30w7bnapvr+OrVq73OCQCAY+Y3STgJVVVWrlyZlStX5rzzzjsh12ytZc+ePU8IvLt27cqePXuye/fu7NmzZ3r7scceywMPPHDYsdnnTK13796d5cuXHzaUevb20Y5N/aPAypUrp7dn7q9cuTLLli07IX8XAACMFsEWWJCqmg6Rx9ujPNvsHuapZeb+7GOPPvroYft79+6dXvbt2/eE/f3792fFihXzBt/TTz/9Cesn2zazV1qIBgBYOoIt0LnF6GGebXJyMvv3758z+E61TfUm792797D1o48++oS22efN7I3eu3dvTjvttOl/CJhv+PXUemYgnh2ejxSqDd8GABjwWxFwSli2bNl0L+1im5yczL59+54w3HquYdlTx7Zu3ZoHH3xwOmjPDs9zbY+NjT0h8J522mlZsWLFvMvRjq9cufKwgD07dM/cN5kYADAqBFuAE2zZsmXT4W+xtNZy4MCBJwTeAwcOzLns379/3mMHDhzIrl27cuDAgelgPbMHevbz0VPL1HPRs4PvzKHex7KsXLkyp5122nQIn7m9fPnyVNWi/b0CAP0k2AL0UFVNB76zzz57yT9/6rnouSYDmz3Me/aybdu2Ix6feiZ6KpDP3J6cnHxC2J1re6oHeyFDweebyXvVqlU57bTTPC8NAD0g2ALwpM18Lvqcc85Zss+dmJiYM/DO3p5vKPjOnTuzadOmIw4Nn1pPXWd8fHz6zzpzOe20047aPldv9ELaVq1alRUrVuidBoAFEmwB6I2xsbEle1Y6eXzI91Qv8r59++ZcZh/bu3fv9GRlO3fuzObNm+ecrftI+wcPHpx3Fu+FBOOZr7qaWs8M3LMD+VTb+Pi4QA1A7yx5sK2qS5L8QZL1SVqSW1prb5p1TiV5U5LvT7InyY+21j691LUCcGqbOeR7qU1NQnakIdvzHdu5c2e2bNkyfd7MwD07lM9uO3To0LwBeOb+0cLyzP3ZQ8XnWmYPKxeuAXgyuuixPZTkZ1trn66qM5LcVVUfbq19acY5L0lyxXB5TpLfGa4B4JSwFJOQzWViYmLOnuejBeKptl27duWRRx55wjVmDhWfb5k65+DBgxkfH58OulOzdU/1Rs+1fbTjs4eJz7V92mmnCdQAPbXkwba19o0k3xhu76yqe5JclGRmsL0hyR+01lqST1TVWVV1wfBnAYBFMjY2Nv1+5a5MDQGf+cz07HdHz3639My2xx577AnHZg4Xn2/7wIED06+9mu856tnDvxfStpAh5V6fBXB8On3GtqouS/KdST4569BFSR6Ysf/gsE2wBYCTXFdDwCcnJ4/6TPXMYeAzh4NPbW/fvn3eY0caSr58+fIFPTc917Duo+3PbDvSM9ZepwX0WWfBtqrWJPnTJD/dWttxjNe4KclNSXLppZeewOoAgFPNsmXLpoPeUprqoT7aM9RzvQpran/qVVrzHZ/Z+z3fkPLJyckFTTI21xDwha6nXqM1Pj6eFStWZHx8XKAGTohOgm1VjWcQav+otfbeOU55KMklM/YvHrYdprV2S5JbkmTDhg1tEUoFAFhUM3uozzrrrM7qOHTo0JzPVM/cn28Y+NTs30caKj713usDBw7k4MGD0+uJiYmMj48fFnbn216xYkVWrFhxWGCevb2QY1O931OL2cCh/7qYFbmS/H6Se1prvzHPabcl+amqelcGk0Zt93wtAMDiWb58eZYvX77kz1dPTk7m4MGD08vs4Dt7eyp8zw7MU5OXbd68+QntM7en3lM91YO9f//+TExMTD9ffaTZu+cb4j17Vu+FrOd7F/ZUDZ67hienix7b5yf5kSSfr6q7h22/mOTSJGmtvTXJX2Twqp/7Mnjdz79f+jIBAFhsy5Yt6+y1WlMmJyfnnKl7KvgebSbvmeupZ63nOz61Ptr7scfGxuYNvlPLzJ7o1atXH3E9V9vU0HDDwTkZdDEr8t8kOeJ/OcPZkF+7NBUBAHAqW7Zs2fQzwKOgtZaDBw/OOXv37GHhu3fvzp49ew5bb9269Qltc62nhoZPTk4ecfKxo72L+kiv0Dpa28zrT33G2NiYoM2T1umsyAAAwOGqajrorV27dtE/b2JiYs7JxuabhGy+3u19+/Zl27Zt875Wa75XbU0NMT9w4MB00J45dHtm6J1r+2g920eaEO1owduQ8P4QbAEA4BQ2NjY2PVx5FExMTEyH3JnDt+fan2/G76ll+/bt2bRp07zHp3q/Zw8/nzkkfHbonSsEz/e+6yOtZ7fNvO7UtmC9cIItAAAwMsbGxkZiaHhrbXq28Ll6nmf3QM/1nuupcH2k918fKVxPPYM+O/DO1zZ7BvHZy9GOXXvttUsySmAxCLYAAACzVNX0K6fWrFmz5J8/V7A+2vbULOIzZxefWvbt25cdO3bMeWxqufrqqwVbAAAAToyug3XfLOu6AAAAADgegi0AAAC9JtgCAADQa4ItAAAAvSbYAgAA0GuCLQAAAL0m2AIAANBrgi0AAAC9JtgCAADQa4ItAAAAvSbYAgAA0GuCLQAAAL0m2AIAANBrgi0AAAC9JtgCAADQa4ItAAAAvSbYAgAA0GuCLQAAAL0m2AIAANBrgi0AAAC9JtgCAADQa4ItAAAAvSbYAgAA0GuCLQAAAL0m2AIAANBrgi0AAAC9JtgCAADQa4ItAAAAvSbYAgAA0GuCLQAAAL0m2AIAANBrgi0AAAC9JtgCAADQa4ItAAAAvSbYAgAA0GuCLQAAAL0m2AIAANBrgi0AAAC9JtgCAADQa4ItAAAAvSbYAgAA0GuCLQAAAL0m2AIAANBrnQTbqrq1qjZX1RfmOX5tVW2vqruHy39a6hoBAADoh+Udfe7bk/x2kj84wjl3tNZeujTlAAAA0Fed9Ni21j6e5LEuPhsAAICTyyg/Y/tdVfXZqvpgVX1L18UAAAAwmroainw0n07y1Nbarqr6/iTvT3LF7JOq6qYkNyXJpZdeuqQFAgAAMBpGsse2tbajtbZruP0XScar6rw5zrultbahtbZh3bp1S14nAAAA3RvJYFtVT6mqGm5fk0Gdj3ZbFQAAAKOok6HIVfXHSa5Ncl5VPZjkDUnGk6S19tYkr0jyk1V1KMneJDe21loXtQIAADDaOgm2rbUfPMrx387gdUAAAABwRCM5FBkAAAAWSrAFAACg1wRbAAAAek2wBQAAoNeOK9hW1eqqWjbcvrKqrq+q8RNTGgAAABzd8fbYfjzJyqq6KMlfJfmRJG8/3qIAAABgoY432FZrbU+Slyf5n621/yvJtxx/WQAAALAwxx1sq+q7kvy7JH8+bBs7zmsCAADAgh1vsP3pJL+Q5H2ttS9W1eVJPnrcVQEAAMACLT+eH26tfSzJx5JkOInUI621/3giCgMAAICFON5Zkd9ZVWuranWSLyT5UlX9vyemNAAAADi64x2KfFVrbUeSlyX5YJJvymBmZAAAAFgSxxtsx4fvrX1ZkttaaweTtOOuCgAAABboeIPt7yb5apLVST5eVU9NsuN4iwIAAICFOt7Jo96c5M0zmr5WVd97fCUBAADAwh3v5FFnVtVvVNXG4XJzBr23AAAAsCSOdyjyrUl2Jvm3w2VHkv/veIsCAACAhTquochJvrm19m9m7P/nqrr7OK8JAAAAC3a8PbZ7q+oFUztV9fwke4/zmgAAALBgx9tj+xNJ/qCqzhzub03yquO8JgAAACzY8c6K/Nkk31FVa4f7O6rqp5N87gTUBgAAAEd1vEORkwwCbWtt6v21/8+JuCYAAAAsxAkJtrPUIlwTAAAA5rQYwbYtwjUBAABgTsf0jG1V7czcAbaSrDquigAAAOBJOKZg21o740QXAgAAAMdiMYYiAwAAwJIRbAEAAOg1wRYAAIBeE2wBAADoNcEWAACAXhNsAQAA6DXBFgAAgF4TbAEAAOg1wRYAAIBeE2wBAADoNcEWAACAXhNsAQAA6DXBFgAAgF4TbAEAAOg1wRYAAIBeE2wBAADoNcEWAACAXhNsAQAA6DXBFgAAgF4TbAEAAOg1wRYAAIBeE2wBAADotU6CbVXdWlWbq+oL8xyvqnpzVd1XVZ+rqmcudY0AAAD0Q1c9tm9Pct0Rjr8kyRXD5aYkv7MENQEAANBDnQTb1trHkzx2hFNuSPIHbeATSc6qqguWpjoAAAD6ZFSfsb0oyQMz9h8ctgEAAMBhRjXYLkhV3VRVG6tq45YtW7ouBwAAgA6MarB9KMklM/YvHrYdprV2S2ttQ2ttw7p165asOAAAAEbHqAbb25L838PZkZ+bZHtr7RtdFwUAAMDoWd7Fh1bVHye5Nsl5VfVgkjckGU+S1tpbk/xFku9Pcl+SPUn+fRd1AgAAMPo6CbattR88yvGW5LVLVA4AAAA9NqpDkQEAAGBBBFsAAAB6TbAFAACg1wRbAAAAek2wBQAAoNcEWwAAAHpNsAUAAKDXBFsAAAB6TbAFAACg1wRbAAAAek2wBQAAoNcEWwAAAHpNsAUAAKDXBFsAAAB6TbAFAACg1wRbAAAAek2wBQAAoNcEWwAAAHpNsAUAAKDXBFsAAAB6TbAFAACg1wRbAAAAek2wBQAAoNcEWwAAAHpNsAUAAKDXBFsAAAB6TbAFAACg1wRbAAAAek2wBQAAoNcEWwAAAHpNsAUAAKDXBFsAAAB6TbAFAACg1wRbAAAAek2wBQAAoNcEWwAAAHpNsAUAAKDXBFsAAAB6TbAFAACg1wRbAAAAek2wBQAAoNcEWwAAAHpNsAUAAKDXBFsAAAB6TbAFAACg1wRbAAAAek2wBQAAoNc6CbZVdV1V3VtV91XVz89x/EeraktV3T1cfryLOgEAABh9y5f6A6tqLMlbkrw4yYNJ7qyq21prX5p16p+01n5qqesDAACgX7rosb0myX2ttftbaweSvCvJDR3UAQAAwEmgi2B7UZIHZuw/OGyb7d9U1eeq6j1VdcnSlAYAAEDfjOrkUX+W5LLW2rcn+XCSd8x1UlXdVFUbq2rjli1blrRAAAAARkMXwfahJDN7YC8etk1rrT3aWts/3H1bkmfNdaHW2i2ttQ2ttQ3r1q1blGIBAAAYbV0E2zuTXFFV31RVK5LcmOS2mSdU1QUzdq9Pcs8S1gcAAECPLPmsyK21Q1X1U0n+MslYkltba1+sql9JsrG1dluS/1hV1yc5lOSxJD+61HUCAADQD9Va67qGE2LDhg1t48aNXZcxr+3bk/Hx5PTTu64EAACgf6rqrtbahrmOjerkUSedd74zeeYzk7vu6roSAACAk4tgu0R+8ieTX/7l5CUvSX7t15KJia4rAgAAODkItkvoxhuTjRuTD30oeeELk69/veuKAAAA+k+wXWKXXprcfnvyAz+QbNiQ/PEfd10RAABAvwm2HRgbS17/+kHP7a/8SvLDPzyYXAoAAIAnT7Dt0NRkUmeemVx9dXLHHV1XBAAA0D+CbcdOPz15y1uS3/7t5JWvTH7pl5IDB7quCgAAoD8E2xHxAz+QfOYzyWc/mzzvecm993ZdEQAAQD8ItiNk/frkz/4sefWrkxe8IPnd301a67oqAACA0SbYjpiqwTtv77gjueWW5GUvS7Zs6boqAACA0SXYjqinPz35+79PrrpqMLHUBz/YdUUAAACjSbAdYStWJG98Y/LOdyY/8RPJa1+b3HOP4ckAAAAzCbY98C//5WBSqeXLk+uuSy67LHnNa5L3vS/ZsaPr6gAAALol2PbEWWclb3pT8tWvJh/60GCo8lvfmlx00SD4vvGNg1mVJye7rhQAAGBpLe+6AJ6cquQZzxgsP/MzyZ49ycc+Ngi7N96YbN+e/Kt/lbzkJcmLX5yce27XFQMAACyuaifJA5sbNmxoGzdu7LqMzt1/f/KXfzmYbOpjHxsE4OuuGyzPfnYyNtZ1hQAAAE9eVd3VWtsw5zHB9uS1f3/yt3876M390IeShx5Knve85GlPS775m5PLLx+sL7ssOe20rqsFAACYn2BLkkGw/eQnk698ZdCzO7V+4IFk/fpByJ0ZeKfW55zTdeUAAMCp7kjB1jO2p5CLLkpe/vInth86lHz964+H3a98Jflf/+vx7bGxw8PuhRcm69Yl55//+Pq88wazNgMAACw1UYQsXz4IrJdfnrzoRYcfay159NHDe3n/4R8GQ5w3b062bBmsH3ssOfPMJwbeqfXsEHzWWYY/AwAAJ4ZgyxFVDYLoeeclz3nO/OdNTCRbtx4edqfW99wzmMhqav+RR5Jt2waB+qyz5l/OPnv+Y2eemaxYsZh/cgAAoC8EW06IsbHHA/BCtJbs3TsIuNu2DULx1PbUsnlzcu+9T2zfujXZsSNZtiw544xk7drD1wttW706WbNmsF692ozRAADQV4ItnahKTj99sFx44ZP/+dYGsz7v2JHs3DlY5tt+6KG523fvTnbtGqx37x70AM8MuzND7+y2NWser3/Vqiduz9U2Pn7i/x4BAADBlp6qSlauHCznn3/812st2bfv8KA7tT1X244dycMPJ3v2DHqe9+yZf3vv3sHPJIeH3all5crH1zOXhbatWDF4Xnlqmb0/1bZixeDvDQAATjaCLWQQ+KaC5rp1i/MZBw8eHnz37BmE6fmWvXsf396zZzBB11zH9u9PDhwYrGcus9sOHnw89M4Mv1Ohd2oZHz9628z98fEjL8uXL/yc5csPX2a3GS4OAMBcBFtYIlMhbu3abj5/cnIQducKvFPtU8uTadu7d9CDffDg3MuhQ/Mfm3nOzGW+tmT+8Ds2dngAnr29kLbZy3ztRzu2bNnR2450zrGsj7R9pLapRW8+ANBngi2cIpYte3z4cl9NTs4dfg8eHMzMPTHxeNtc2/Mdn9qfaznasQMHDm+bnHziebPbjnTOsaxnXm9qf/Yy37GJicFQ/Kr5Q+9cIfjJnDPf9kLOO571sbQdaftEtM23PJlzu1qSJ3/sRP7MXO2z247lnJn7x3ru0baP5fjMNQBHJ9gCvbFs2eNDoDlxWhssM4PufAF5apnvnJk/P/Oc+bbnOjZ1jdnXWeh6anu+azyZtrn252ubWfd85xxpebLnL+Uy8z5Z6LET+TNztc9uO5ZzZu4f67lH2z6W43M5llB8vD9zpLaFHDvWcxd6vRNZy7F+5vGsl+qax1pDl5/TZc1HaztR1zyWzzveehay/ZrXLN5jeYtNsAU4xU39krds2WCINTAaFhKAT1SQfjJtCzl2rOcu9HonspZj/czjWS/VNY+1hi4/p8uaj9Z2oq55LJ93PPU8mWtMTqa3/AoDADCC5up5AWBuy7ouAAAAAI6HYAsAAECvCbYAAAD0mmALAABArwm2AAAA9JpgCwAAQK8JtgAAAPSaYAsAAECvCbYAAAD0mmALAABArwm2AAAA9Fq11rqu4YSoqi1JvraIH3FekkcW8fpwLNyXjCL3JaPGPckocl8yikb9vnxqa23dXAdOmmC72KpqY2ttQ9d1wEzuS0aR+5JR455kFLkvGUV9vi8NRQYAAKDXBFsAAAB6TbBduFu6LgDm4L5kFLkvGTXuSUaR+5JR1Nv70jO2AAAA9JoeWwAAAHpNsF2Aqrququ6tqvuq6ue7rodTU1XdWlWbq+oLM9rOqaoPV9U/Dtdnd1kjp5aquqSqPlpVX6qqL1bV64bt7ks6U1Urq+pTVfXZ4X35n4ft31RVnxx+l/9JVa3oulZOLVU1VlWfqar/Pdx3T9KpqvpqVX2+qu6uqo3Dtt5+hwu2R1FVY0nekuQlSa5K8oNVdVW3VXGKenuS62a1/XyS21trVyS5fbgPS+VQkp9trV2V5LlJXjv8/6P7ki7tT/LC1tp3JLk6yXVV9dwkv57kN1trT0uyNcmruyuRU9TrktwzY989ySj43tba1TNe8dPb73DB9uiuSXJfa+3+1tqBJO9KckPHNXEKaq19PMljs5pvSPKO4fY7krxsKWvi1NZa+0Zr7dPD7Z0Z/MJ2UdyXdKgN7Brujg+XluSFSd4zbHdfsqSq6uIkP5DkbcP9inuS0dTb73DB9uguSvLAjP0Hh20wCta31r4x3H44yfoui+HUVVWXJfnOJJ+M+5KODYd83p1kc5IPJ/lKkm2ttUPDU3yXs9T+R5LXJ5kc7p8b9yTda0n+qqruqqqbhm29/Q5f3nUBwInRWmtVZZpzllxVrUnyp0l+urW2Y9ARMeC+pAuttYkkV1fVWUnel+Tp3VbEqayqXppkc2vtrqq6tuNyYKYXtNYeqqrzk3y4qr4882DfvsP12B7dQ0kumbF/8bANRsGmqrogSYbrzR3XwymmqsYzCLV/1Fp777DZfclIaK1tS/LRJN+V5KyqmvoHfd/lLKXnJ7m+qr6awSNtL0zyprgn6Vhr7aHhenMG/wh4TXr8HS7YHt2dSa4Yzly3IsmNSW7ruCaYcluSVw23X5XkAx3Wwilm+IzY7ye5p7X2GzMOuS/pTFWtG/bUpqpWJXlxBs9/fzTJK4anuS9ZMq21X2itXdxauyyD3yP/T2vt38U9SYeqanVVnTG1neT7knwhPf4Or9Z607vcmar6/gyejRhLcmtr7Ve7rYhTUVX9cZJrk5yXZFOSNyR5f5J3J7k0ydeS/NvW2uwJpmBRVNULktyR5PN5/LmxX8zgOVv3JZ2oqm/PYMKTsQz+Af/drbVfqarLM+gtOyfJZ5L8cGttf3eVcioaDkX+udbaS92TdGl4/71vuLs8yTtba79aVeemp9/hgi0AAAC9ZigyAAAAvSbYAgAA0GuCLQAAAL0m2AIAANBrgi0AAAC9JtgCAADQa4ItAAAAvSbYAgAA0Gv/P4NSnx4f+p4nAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 1152x2160 with 5 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "fig = plt.gcf()\n",
        "fig.set_size_inches(16, 30)\n",
        "plt.figure(1)\n",
        "\n",
        "nplot = len(train_history.keys()) // 2\n",
        "print(nplot)\n",
        "for idx, m in enumerate(metrics, 0):\n",
        "    plt.subplot(nplot*100+10+idx+1)\n",
        "    plt.plot([i+1 for i in range(len(train_history[m['name']]))], train_history[m['name']], color='black', linewidth=1)\n",
        "    plt.plot([i+1 for i in range(len(train_history[m['name']]))], train_history['val_{}'.format(m['name'])], color='blue', linewidth=1)\n",
        "    ax = plt.gca()\n",
        "    ax.set_ylabel(m['name'])\n",
        "plt.subplot(nplot*100+10+nplot)\n",
        "plt.plot([i+1 for i in range(len(train_history['Loss']))], train_history['Loss'], color='black', linewidth=1)\n",
        "plt.plot([i+1 for i in range(len(train_history['val_Loss']))], train_history['val_Loss'], color='blue', linewidth=1)\n",
        "ax = plt.gca()\n",
        "ax.set_ylabel('Loss')\n"
      ]
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.7.12"
    },
    "papermill": {
      "default_parameters": {},
      "duration": 423.669638,
      "end_time": "2022-01-27T11:22:40.427480",
      "environment_variables": {},
      "exception": null,
      "input_path": "__notebook__.ipynb",
      "output_path": "__notebook__.ipynb",
      "parameters": {},
      "start_time": "2022-01-27T11:15:36.757842",
      "version": "2.3.3"
    },
    "colab": {
      "name": "Patient Survival Predictor (using Pytorch)",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    }
  },
  "nbformat": 4,
  "nbformat_minor": 5
}