myDNN_with_MNIST_3layer.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "myDNN_with_MNIST.ipynb",
      "provenance": [],
      "authorship_tag": "ABX9TyMCS8Xv/RGIwD0Z5jAI438q",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/khodid/2020_SAI_MONING2/blob/master/myDNN_with_MNIST_3layer.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XDwnKZW3-eGT",
        "colab_type": "text"
      },
      "source": [
        "# 참고 소스:\n",
        "\n",
        "- 모두의 딥러닝 시즌 2 - lab 07-2 mnist introduction\n",
        "https://github.com/deeplearningzerotoall/PyTorch/blob/master/lab-07_2_mnist_introduction.ipynb\n",
        "\n",
        "- 모두의 딥러닝 시즌 2 - 프로젝트 A\n",
        "(내가 작성했던 것.)\n",
        "https://github.com/khodid/2020_SAI_MONING2/blob/master/projectA_DNN.ipynb\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LWIepOUNDDiV",
        "colab_type": "text"
      },
      "source": [
        "# 패키지 받아오기.\n",
        "- torch\n",
        "- torch.nn : Neural Network를 위한 모델\n",
        "- torchvision: datasets(Torch에서 제공되는 dataset 받아오기)/transforms( 이미지 파일 가공하기)\n",
        "- matplotlib: 데이터들의 시각적 표현을 위해서.(그래프 그리기, 등등.)\n",
        "- numpy: 수학 함수들 불러오기\n",
        "\n",
        "밑에 DataLoader 패키지를 import해오는 부분도 있긴 한데, 예시에서 따로 중간에 import를 해온 것은, 아무래도 일회성으로 쓰는 것이기 때문에 읽는 사람이 이해하기 편하도록 그 패키지가 사용되는 블럭 근처에 선언을 해준 게 아닌가 싶다."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iqy-XdCn-dCa",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import torch\n",
        "import torch.nn as nn\n",
        "import torchvision.datasets as dset\n",
        "import torchvision.transforms as transforms\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HolPuQhAjRLv",
        "colab_type": "text"
      },
      "source": [
        "GPU를 사용하거나 CPU를 사용하도록 하는 코드...\n",
        "이걸 왜 써줘야 하는 지는 예전에 설명을 들었던 것 같은데 까먹었음.\n",
        "\n",
        "궁금해지면 'cuda'나 'gpu와 cpu' 등등의 키워드로 알아보면 될 듯."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "34TlgPSbjUZu",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "device = \"cuda\" if torch.cuda.is_available() else \"cpu\""
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OYz9WuTokc9t",
        "colab_type": "text"
      },
      "source": [
        "# 하이퍼 파라미터 설정.\n",
        "- 배치 크기(미니배치 방법을 쓰기 때문)\n",
        "- 학습률(learning rate): 너무 크면 overshoot이나 gradient exploding 일어나고, 작으면 학습이 더뎌서 원하는 결과가 나올 때까지 epoch 수를 늘려야 함.\n",
        "- epoch: 1회의 학습 과정(cost 계산, GD에 따라 weight update) 끝나는 걸 1 epoch라고 했던 것 같다.\n",
        "\n",
        "이 파라미터는 전부 모델을 설계하는 사람이 알아서, 감과 경험을 동원해 다르게 설정해주는 것으로 알고 있다."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HafU2jOdkdL8",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "batch_size = 50\n",
        "learning_rate = 1e-3\n",
        "num_of_epoch = 5"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fbEyku1jlgSo",
        "colab_type": "text"
      },
      "source": [
        "# 데이터 불러오기.\n",
        "\n",
        "위의 참고 소스의 데이터 불러오기 파트 따라치기로 구현했음.\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "DnIIejsspwmz",
        "colab_type": "text"
      },
      "source": [
        "root : 데이터를 어디서 받아올 것인지,\n",
        "\n",
        "\n",
        "trasnsform : 데이터를 어떻게 저장할 것인지.\n",
        "\n",
        "[torchvision.transforms 라이브러리 설명](https://pytorch.org/docs/stable/torchvision/transforms.html)\n",
        "- transforms.Compose 함수는 여러 가지 전처리 함수를 묶어주는 역할을 한다.\n",
        "- ToTensor는 이미지 파일을 텐서로 바꾸어주는 역할\n",
        "- Normalizaze 함수는 데이터를 정규분포로 바꾸어줌. \n",
        "이 정규화 전처리의 유용성은 [4월 6일 한땀한땀](https://www.notion.so/20-04-06-8c08ab4a451d4930b47d8f7191422dac#32766507a8db49ce94961f974cd9c439)에서 정리한 바 있음.\n",
        "\n",
        "[torch.utils.data.DataLoader 파라미터 설명](https://pytorch.org/docs/stable/data.html#torch.utils.data.DataLoader)\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PCXoArkqlkA5",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "root = 'MNIST_data/'\n",
        "from torch.utils.data import DataLoader\n",
        "\n",
        "# data 가공방식 설정하기 - 프로젝트 A에서 한 것과 같이 Normalization으로 전처리 시킬 것임.\n",
        "transform   = transforms.Compose([transforms.ToTensor(), transforms.Normalize(mean = (0.5,), std=(0.5,))])\n",
        "\n",
        "# data 받아오기\n",
        "train_data  = dset.MNIST(root = root, train = True,   transform = transform,  download = True)\n",
        "test_data   = dset.MNIST(root = root, train = False,  transform = transform, download = True)\n",
        "\n",
        "# data를 실제 학습과 테스트에 적용하기 위해 미니배치로 나누어주기. (매개변수로는 데이터덩어리 / 배치 크기 / 섞을지 말지)\n",
        "train_loader  = DataLoader(train_data, batch_size = batch_size, shuffle = True, drop_last= True)   # 트레이닝할 때는 데이터를 섞어주지 않으면 순서 그 자체를 학습할 수도 있음.\n",
        "test_loader   = DataLoader(test_data, batch_size = batch_size, shuffle = False, drop_last= True)   # 테스트 할 때는 데이터를 섞어주지 않아도 되므로."
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ql8qw-c_s_t0",
        "colab_type": "text"
      },
      "source": [
        "# 데이터 샘플 시각화.\n",
        "\n",
        "matplotlib과 numpy 쪽은 생소해서 그냥 태그랑 정렬 방식(열, 행)만 바꾸고 소스를 프로젝트 A에서 그대로 복붙했음."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TYo_uQQYq6Q5",
        "colab_type": "code",
        "outputId": "627f5703-da26-4302-d03a-c38e8a2002e0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 435
        }
      },
      "source": [
        "labels_map = {0 : '0', 1 : '1', 2 : '2', 3 : '3', 4 : '4', 5 : '5', 6 : '6',\n",
        "              7 : '7', 8 : '8', 9 : '9'}\n",
        "columns = 8\n",
        "rows = 4\n",
        "fig = plt.figure(figsize=(8,8))\n",
        "\n",
        "for i in range(1, columns*rows+1):\n",
        "    data_idx = np.random.randint(len(train_data))\n",
        "    img = train_data[data_idx][0][0,:,:].numpy() # numpy()를 통해 torch Tensor를 numpy array로 변환\n",
        "    label = labels_map[train_data[data_idx][1]] # item()을 통해 torch Tensor를 숫자로 변환\n",
        "    \n",
        "    fig.add_subplot(rows, columns, i)\n",
        "    plt.title(label)\n",
        "    plt.imshow(img, cmap='gray')\n",
        "    plt.axis('off')\n",
        "plt.show()"
      ],
      "execution_count": 48,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x576 with 32 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YnyUzMIEt_vc",
        "colab_type": "text"
      },
      "source": [
        "# 네트워크 설계\n",
        "\n",
        "층층이 얽힌 그래프 구조를 만들어야 할 때인데...\n",
        "\n",
        "기존에 프로젝트 A에선 layer를 2개 쌓아준 걸\n",
        "3개로 늘려서 정의해보도록 하겠음.\n",
        " \n",
        " \n",
        " -> 이렇게 하니깐 밑에 학습 과정중 outputs = model(imgs) 라인에서 자꾸 tuple에는 dim이 없다는 오류가 뜸.\n",
        "\n",
        "Module의  linear 함수에서 이런 현상이 일어나고, \n",
        "\n",
        "AttributeError: 'tuple' object has no attribute 'dim'\n",
        "\n",
        "이 부분은 따라치기를 많이 했음.\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ndzt0GMagdNE",
        "colab_type": "text"
      },
      "source": [
        "! 위의 문제는\n",
        "\n",
        "Batch Normalization을 적용한 [DNN 코드를 검색해서](https://towardsdatascience.com/batch-normalization-and-dropout-in-neural-networks-explained-with-pytorch-47d7a8459bcd) 이 구조를 참고했더니 해결됐다!\n",
        "\n",
        "(layer를 각각 다른 self.layer1, 2 이렇게 나누지 않고 그냥 하나로 통합하니깐 해결된다.)\n",
        "\n",
        "다만 왜 그런 문제가 생긴 건지, 왜 해결된 건지는 아직 잘 모르겠다."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FZ_fsXEFusC9",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "## 3 layer Model 짜려고 했으나 오류가 나서 포기한 부분. -> 해결완료\n",
        "\n",
        "class DNN(nn.Module):\n",
        "  def __init__(self, num_classes=10):\n",
        "    super(DNN, self).__init__()\n",
        "    self.layer = nn.Sequential(\n",
        "        #layer 1(input layer)\n",
        "        nn.Linear(784, 300),     \n",
        "        nn.BatchNorm1d(300),\n",
        "        nn.ReLU(),\n",
        "        # layer 2 \n",
        "        nn.Linear(300, 64),        \n",
        "        nn.BatchNorm1d(64),\n",
        "        nn.ReLU(),\n",
        "        # layer 3 (output layer)\n",
        "        nn.Linear(64, 10)\n",
        "    )\n",
        "  def forward(self, x):\n",
        "    x = x.view(x.size(0), -1)\n",
        "    x = self.layer(x)\n",
        "    return x"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "V8FGk7OoDX4U",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# class DNN(nn.Module):\n",
        "#     def __init__(self, num_classes=10):\n",
        "#         super(DNN, self).__init__()\n",
        "#         self.layer1 = nn.Sequential(\n",
        "#             ## 코드 시작 ##\n",
        "#             nn.Linear(784, 64),    # Linear_1 해당하는 층\n",
        "#             torch.nn.BatchNorm1d(64),    # BatchNorm_1 해당하는 층\n",
        "#             torch.nn.ReLU()     # ReLU_1 해당하는 층\n",
        "#             ## 코드 종료 ##\n",
        "#         )\n",
        "#         self.layer2 = nn.Sequential(\n",
        "#             ## 코드 시작 ##\n",
        "#             nn.Linear(64, 10)    # Linear_2 해당하는 층 \n",
        "#             ## 코드 종료 ##\n",
        "#         )\n",
        "    \n",
        "#     def forward(self, x):\n",
        "#         x = x.view(x.size(0), -1) # flatten\n",
        "#         x_out = self.layer1(x)\n",
        "#         x_out = self.layer2(x_out)\n",
        "#         return x_out"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ByroRwpExLZT",
        "colab_type": "text"
      },
      "source": [
        "# Weight 초기화\n",
        "\n",
        "DNN 에서의 Weight 초기화를 이론적으로 다룬 부분은 \n",
        "\n",
        "[한땀한땀 4월 20일](https://www.notion.so/20-04-20-edc672e58455447a989f7f669b7beca2#edb720ded0304d13bdbb3503d6c9dffb)\n",
        "에 있다.\n",
        "\n",
        "요약하면 weight initialization은 gradient vanishing/exploding 현상을 방지하기 위해 필수적인데, 주로 쓰는 건 He나 Xavier의 방법이라는 것이다.\n",
        "\n",
        "코드는 프로젝트 A에서 긁어 왔다.\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lOkF59J_xxh3",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def weights_init(m):\n",
        "    if isinstance(m, nn.Linear): # 모델의 모든 MLP 레이어에 대해서\n",
        "        nn.init.xavier_normal_(m.weight) # Weight를 xavier_normal로 초기화\n",
        "        # print(m.weight)         # 프로젝트 A에 있던 line인데, 굳이 print해야 할 이유를 모르겠어서 주석처리 하겠음."
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oEU4YaeUztGx",
        "colab_type": "text"
      },
      "source": [
        "# 모델 생성\n",
        "\n",
        "이제 지금까지 만든 레이어 구조에 weight 초기화를 넣어 학습 전의 모델 기틀을 만들어 준다.\n",
        "\n",
        "\n",
        ".apply()는 torch.nn.Module 에 있는 함수? 기능? 이라고 한다.\n",
        "\n",
        "[설명](https://pytorch.org/docs/stable/nn.html?highlight=apply#torch.nn.Module.apply) 봐도 잘 모르겠다. 객체지향 얼렁 공부해야 할 듯."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZMB56EPm0vAK",
        "colab_type": "code",
        "outputId": "cabdcf81-40eb-4fd1-a793-02fdea7e264b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 210
        }
      },
      "source": [
        "torch.manual_seed(7777)\n",
        "model = DNN().to(device)\n",
        "model.apply(weights_init)"
      ],
      "execution_count": 52,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "DNN(\n",
              "  (layer): Sequential(\n",
              "    (0): Linear(in_features=784, out_features=300, bias=True)\n",
              "    (1): BatchNorm1d(300, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "    (2): ReLU()\n",
              "    (3): Linear(in_features=300, out_features=64, bias=True)\n",
              "    (4): BatchNorm1d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "    (5): ReLU()\n",
              "    (6): Linear(in_features=64, out_features=10, bias=True)\n",
              "  )\n",
              ")"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 52
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "doEyr9uu3muw",
        "colab_type": "text"
      },
      "source": [
        "# 학습시킬 준비 - Cost function과 Optimizer \n",
        "\n",
        "cost 계산하는 함수(?)와 optimizer를 설정해준다.\n",
        "\n",
        "이쪽도 코드 베끼기...\n",
        "\n",
        "나중에 개인 프로젝트를 한다면 여기서 다른 cost function을 쓰거나 optimizer를 쓰는 시도를 할 수도 있을 것 같다.\n",
        "\n",
        "Optimizer 관련 내용은 [4/20 스터디](https://www.notion.so/20-04-20-edc672e58455447a989f7f669b7beca2#933466115ddb4f98aabdb794f481e04f)에서 다룬 바 있다.\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_wV3_8_y4DY9",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "criterion = nn.CrossEntropyLoss()\n",
        "optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lFMCElu94v4q",
        "colab_type": "text"
      },
      "source": [
        "# 학습시키기\n",
        "\n",
        "프로젝트 A에서 참고 자료로 주는 것은 [이 문서](https://pytorch.org/docs/stable/optim.html#taking-an-optimization-step)다.\n",
        "\n",
        "많이 공부해서 익숙하지만, \n",
        "\n",
        "**예측 -> cost 계산 -> back propagation(gradient 계산) ->gradient 업데이트**\n",
        "\n",
        "순으로 써주면 되는 것이다."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "D3dXC-4r5RLl",
        "colab_type": "code",
        "outputId": "fe6c2501-2326-47f5-81ac-3c4908464d2d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 369
        }
      },
      "source": [
        "for epoch in range(num_of_epoch):\n",
        "    for i, (imgs, labels) in enumerate(train_loader): # 여기 for 문은 직접 작성할 엄두가 안 난다... \n",
        "        imgs = imgs.to(device) \n",
        "        labels = labels.to(device)\n",
        "        \n",
        "        outputs = model(imgs)\n",
        "        loss = criterion(outputs, labels)  \n",
        "        \n",
        "        optimizer.zero_grad()       \n",
        "        loss.backward()\n",
        "        optimizer.step()           \n",
        "        \n",
        "        _, argmax = torch.max(outputs, 1)\n",
        "        accuracy = (labels == argmax).float().mean()\n",
        "        \n",
        "        if (i+1) % 300 == 0:\n",
        "            print('Epoch [{}/{}], Step [{}/{}], Loss: {:.4f}, Accuracy: {:.2f}%'.format(\n",
        "                epoch+1, num_of_epoch, i+1, len(train_loader), loss.item(), accuracy.item() * 100))"
      ],
      "execution_count": 54,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch [1/5], Step [300/1200], Loss: 0.4047, Accuracy: 90.00%\n",
            "Epoch [1/5], Step [600/1200], Loss: 0.2892, Accuracy: 90.00%\n",
            "Epoch [1/5], Step [900/1200], Loss: 0.0847, Accuracy: 98.00%\n",
            "Epoch [1/5], Step [1200/1200], Loss: 0.1094, Accuracy: 98.00%\n",
            "Epoch [2/5], Step [300/1200], Loss: 0.0248, Accuracy: 100.00%\n",
            "Epoch [2/5], Step [600/1200], Loss: 0.3223, Accuracy: 92.00%\n",
            "Epoch [2/5], Step [900/1200], Loss: 0.0793, Accuracy: 98.00%\n",
            "Epoch [2/5], Step [1200/1200], Loss: 0.0178, Accuracy: 100.00%\n",
            "Epoch [3/5], Step [300/1200], Loss: 0.0605, Accuracy: 98.00%\n",
            "Epoch [3/5], Step [600/1200], Loss: 0.1643, Accuracy: 94.00%\n",
            "Epoch [3/5], Step [900/1200], Loss: 0.0127, Accuracy: 100.00%\n",
            "Epoch [3/5], Step [1200/1200], Loss: 0.0384, Accuracy: 100.00%\n",
            "Epoch [4/5], Step [300/1200], Loss: 0.0158, Accuracy: 100.00%\n",
            "Epoch [4/5], Step [600/1200], Loss: 0.0299, Accuracy: 100.00%\n",
            "Epoch [4/5], Step [900/1200], Loss: 0.0304, Accuracy: 100.00%\n",
            "Epoch [4/5], Step [1200/1200], Loss: 0.0219, Accuracy: 100.00%\n",
            "Epoch [5/5], Step [300/1200], Loss: 0.0350, Accuracy: 100.00%\n",
            "Epoch [5/5], Step [600/1200], Loss: 0.0042, Accuracy: 100.00%\n",
            "Epoch [5/5], Step [900/1200], Loss: 0.0352, Accuracy: 98.00%\n",
            "Epoch [5/5], Step [1200/1200], Loss: 0.1031, Accuracy: 98.00%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LbAq7CT9Fv-U",
        "colab_type": "text"
      },
      "source": [
        "코드를 자꾸 수정하면서 계속 돌리니면 학습이 쌓여서 Accuracy가 95% 이상으로 처음부터 높게 나오는 것 같다...\n",
        "\n",
        "실제 모델 설계할 때도 주의해야 할 점인 듯하다."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nFHxTrkqFpQn",
        "colab_type": "text"
      },
      "source": [
        "# Test하기 \n",
        "\n",
        "이쪽도 솔직히 아직은 직접 만들 엄두는 안 나서...\n",
        "따라치기 정도만 해야겠다.\n",
        "\n",
        "Gradient 업데이트를 하면 안 되기 때문에 torch\\_no\\_grad를 써주고\n",
        "\n",
        "training 과정에서 예측을 하던 부분과\n",
        "\n",
        "Training의 진척도를 출력하는 부분을 똑같이 가져오면 되는 듯 하다."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZcC-DBpHGRTg",
        "colab_type": "code",
        "outputId": "b1fbcfbe-431d-4c23-b99f-80cfeed7f63c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "model.eval()\n",
        "with torch.no_grad():\n",
        "  correct = 0\n",
        "  total = 0\n",
        "  for i, (imgs, labels) in enumerate(test_loader):\n",
        "    imgs, labels = imgs.to(device), labels.to(device)\n",
        "    outputs = model(imgs)\n",
        "\n",
        "    _, argmax = torch.max(outputs, 1)\n",
        "    total += imgs.size(0)\n",
        "    correct += (labels == argmax).sum().item()\n",
        "\n",
        "  print(\"{}개 이미지에 대한 테스트 정확도: {:.3f}%\".format(total, correct/total*100))"
      ],
      "execution_count": 55,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "10000개 이미지에 대한 테스트 정확도: 97.750%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Yhokw7SuHpAq",
        "colab_type": "text"
      },
      "source": [
        "Overfitting 된 것일지 걱정을 좀 했는데 생각보다 좋은 모델을 만든 것 같다.\n",
        "\n",
        "이제 이걸 시각적으로 표현하는 것만 남았는데...\n",
        "솔직히 이건 따라쳐도 이해할 자신은 없다..."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2sawPOvcHkq5",
        "colab_type": "code",
        "outputId": "bf2b3de9-d40f-4fc3-a12b-aeb6d420c1e6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 435
        }
      },
      "source": [
        "columns = 8\n",
        "rows = 4\n",
        "fig = plt.figure(figsize=(8,8))\n",
        "\n",
        "model.eval()\n",
        "for i in range(1, columns*rows+1):\n",
        "    data_idx = np.random.randint(len(test_data))\n",
        "    input_img = test_data[data_idx][0].unsqueeze(dim=0).to(device) \n",
        "  \n",
        "    output = model(input_img)\n",
        "    _, argmax = torch.max(output, 1)\n",
        "    pred = labels_map[argmax.item()]\n",
        "    label = labels_map[test_data[data_idx][1]]\n",
        "    \n",
        "    fig.add_subplot(rows, columns, i)\n",
        "    if pred == label:\n",
        "        plt.title(pred + '(O)')\n",
        "    else:\n",
        "        plt.title(pred + '(X)' + ' / ' + label)\n",
        "    plot_img = test_data[data_idx][0][0,:,:]\n",
        "    plt.imshow(plot_img, cmap='gray')\n",
        "    plt.axis('off')\n",
        "model.train()\n",
        "plt.show()"
      ],
      "execution_count": 56,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAcwAAAGiCAYAAABqJfasAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAgAElEQVR4nOy9eXBc133n+zm972g01kZj34iFIAkuAndSomRKY1myIymWlMx4UvHLxLGfU+Vnj132myrXjF9lUknepPyc5CUTz4sVy7YsW5G12VqohRIXkeIGggCx7xuxNoBuoIHuvu+Pxr1qgBQFSgC7AZ5PFUpC33Ob54tz7/md8zu/8ztCURQkEolEIpHcHF2iKyCRSCQSyXpAGkyJRCKRSFaANJgSiUQikawAaTAlEolEIlkB0mBKJBKJRLICpMGUSCQSiWQF3FaDKYQwCyEahRDeW7hnixDi5FrWazWRGj/ynvWmMUMIcVUIYb2Fez4nhHhmLeu1WtwJbQh3hs6NrjGZ9K2JwRRCPC6EaBJCBIQQ7UKIA4uX/gQ4rijKYFzZvUKIN4UQ00IIvxDiRSFElXpdUZR6YFII8bm1qOsnRWpc/xqFEB4hxL8t6usWQjwZd/k7wL8oijIbV/5BIcSZxfJjQoinhRC56nVFUV4EqoUQW26jjI9ECPE1IcQHQoiQEOJfll3eKG1oFkL8eLH9poUQF4UQD8QVWfc67xCNlYv19Qsh2oQQX4i7nDz6FEVZ1R/gPqAb2E3MIPsA3+K1K8C+uLJ7gBngzwEn4AF+AEwAxXHl/gB4abXrKjXe8Rp/DjwDOID9gB+oBszAKJAbV/ZRYAp4ErAC2cD/ArqA1Lhy3wN+lGhti3X5PeDzwD8QM/7x1zZKG9qB7wOFi8/pg8A0ULhRdG50jYABaAG+AeiBe4AAUJ5s+tZC/Engj2/weT4wCxjiPnsX+PsblP0t8FTc777Fe82JblypcWNoXOyE5tWXcvGzfwX+O3AQaIv7XBAbPPznZd+hAxqA/xr32T6gM9H6ltXzB8QZzI3ShjfRWw88spF1biSNwGZiBlDEffYa8N+STd+qumSFEHpgJ5CxOK3uE0L8aHEdqAboUBQlvFjWBuwFnr3BV/2S2AwHAEVR+oEFYNNq1veTIDVuDI1AORBWFKUl7rNLxGaYNUBz3OebiL24SzQqihIFfk2cRqAJKBRCuNai0qvERmnD6xBCZBFr2ytsUJ13gkZig9TNJJm+1V7DzAKMxNxXB4BtQC3wfwJuYm4EFc/ivz/I9QwC6cs+m178jkQjNW4MjQ5iLtZ4/MRcPMs1qhpWolG9Lxk0fhQbpQ2XIIQwAk8DP1EU5SobUOcG1dgMXAO+JYQwCiE+AxwCbCSZvtU2mGqAxP+jKMqgoiijwP8N/Dti/mVnXNkJIArcKPLJS2wNKR4nMLm61f1ESI0bQ+MMsHwW6CL2gi3XqGpYiUb1vmTQ+FFslDbUEELoiLnU54GvLX68oXRuVI2KoiwQW2v/LDAE/B/EZot9JJm+VTWYiqJMEBMZfwSK+v/1QJEQwrBYNgCcAh67wVf9PnBM/UUI4QNMLHWTJQSpcWNoJBZkYBBClMV9tpWYm6uemMtLpZnY32OJxsUO7BHiNAKVQJeiKMtnr8nERmlDAIQQAvgxMc/II4sdMGwgnRtdo6Io9YqiHFIUJU1RlKNAMXCGZNO3Bgu4/xU4C2QCqcQWaP+b8uFC9d64svuJRUN9ndhIIJVYgMIkUBZX7knglUQuTEuNG1LjL4hFytqJBeuoUbImYITFqODFsl/kwyhZCx9GyfYAaXHlvssNAhISpM+wWNe/IDYzsbAYPLFR2nCxTv8vcBpw3ODahtC50TUCWxafTxvwTaCTxWCdZNK3FsKNwN8vVn4I+CFgWbz2VeAflpXfD7xNzEU2BbwMbF5W5mXgoUQ3qtS44TR6gOcXX74e4Mm4a38FfHtZ+YeJDSICwDgxY5u3rMxlYGuitS3W5fvEPAPxP9/fYG1YsKhrbrHe6s8fbBSdd4jGvyLmbp0hFu1aGnctafTd7j+KGWgEvLdwzxbgVKIbVGq84zRmAFcB6y3c8zngl4muu2zDO0vnRteYTPrE4pdLJBKJRCK5CTL5ukQikUgkK0AaTIlEIpFIVoA0mBKJRCKRrABpMCUSiUQiWQGGm10UQnziiKC0tDTcbjelpaXodDpGRkYYGhqir6/vk37lLaMoivi4Mp9GYzIgNcaQGpMfqTGG1Jj8fJTGmxrMT0NlZSU7d+7kG9/4BjabjVdffZWXXnqJn//852v1T0okEolEsmasmcEEWL5lRaeTHuCNhMViISUlhSeeeAKPx4Pb7ebUqVNcvnyZtrY25ubmEl1FiUQiWTXWzGDq9XqMRiMQM5zRaJRoNLpW/1zSo9Pp0Ov16PV6hBBEo1EWFhbiN9quOxwOB16vl8cee4z8/Hxyc3Ox2WwEAgF6e3ulwZRIJBuKNTOYdXV1PPHEE6SmpjIxMUFjYyNDQ0Nr9c8lNampqWzatImDBw9SU1NDWloa/f39/M3f/A2jo6OMji5PsJ/cmEwmrFYr/+k//Sfq6uqorq7GarUSiUSw2+1kZmZqgyWJRCLZKKyZwXQ6nXi9XkwmE/Pz8wwODuL3+9fqn0s6jEYjRqOR1NRUvF4v27Zto7a2lq1bt5KamkpaWhqFhYWEw+F1ZzCtVisZGRmUl5dTVVWFw+FgdnaWwcFBhoaGmJycJBwOJ7qaq4LL5cJiseBwODCZTNjtdu1aOBwmFAoxPj5OMBhkZmYmgTX95GRlZeFwOLBYLAQCAfr7+4lEIuveIySEwGazYbfbsdlsGAwGHA4Her3+urKhUIhr166t63a8ERaLBbPZjNPpxGQy4XA4NG8XxLx/8/PzzM3NMTExQSAQkJ6hm7BmBtNgMGAwGBBCEAgEOHPmDMPDw2v1zyUd6enpZGZm8uijj7Jp0yYOHz6M3W7HYrEAMRfto48+yquvvkpbW1uCa3tr5OXlsX//frZu3UpxcTFCCJqbm/nbv/1bPvjgA1pbW4lEIomu5qdCCIFer2fPnj2UlZWxZ88e8vPz2bZtG3q9HkVRGB0dpaenh6effpr6+nref/99otHounOxP/HEE9TV1VFZWcmZM2f4/ve/j9/vJxAIJLpqnxghBBaLhcrKSurq6tiyZQs5OTns3bsXk8kEsMRo9PT08MMf/pBLly6t23a8EQUFBRQVFXHo0CHy8/PZvXs3DocDpzN2xGQ4HGZoaIjGxkZ+9atfcfbsWZqbE37aV9KypkE/8WyUB/DjMJvN2O12Dh8+TGVlJbt379ZG8EajEUVR6OjooKuri9OnT9Pd3Z3oKq8YvV6Py+WirKyMffv2kZaWBsDs7CwTExN0dXXh9/vXvbGEDw1mdnY2xcXFlJeXk5GRgc1mI3Y0YczVDnDvvfdiNpu5du0ag4OD687QuFwubYDn9XrJy8sjEomsOx1qm5WUlJCRkUF1dTWFhYWUl5fj8/lwu904HA4MBoNWXo0nyMjI4MiRI6Snp2Oz2bh48eK68/xA7B3Nzc0lJSUFr9dLVVUVxcXFVFVV4fF4SElJWbJcIoTA5XJRVFTEZz7zGfR6PQaDgfb2djnTvAG3zWDeKdhsNm1muX//ftLS0rToYEVRWFhY4OLFi9TX1/PSSy8xPT2d4BqvHKPRSFZWFjU1NTzwwAM4nU4URWFmZobR0VE6OjqYmkrmc5NXjuq2ysnJoaysjIqKCqxW65JBn+rq8/l8GAwGOjs7mZmZWVeGRgiB0+nE4/Hg8XjIysqitLSUiYmJdecR0ul0GI1Gdu7cyebNm3n88cdJSUnB5XItKRffhur/u91uPv/5z1NWVobP5+PatWvr0mCaTCaqqqooLCxkz549bNmyheLiYiwWC0II5ufnmZ+fZ3p6Onb6hhCYTCZtYCGEQKfTMTQ0lNQGUx20qnzUZEwtt1qTNWkwV5m0tDS2bt2K1+slJSVlyVaakydPcuHCBf7t3/6N3t5eJiYm1tVan81mo7a2ltLSUpxOJ0ajkUAgwG9+8xvOnDnD+Pg4CwsLH/9F64hr167R29v7set5u3btIisri5/85CdcuHCBCxcuMD8/f5tqKYHYjD8zM5PHHnuMyspK0tPTl8ymFEVhdnaWhYUFgsEgOp0OnU5HamqqNuv0eDxUVFTg8Xi0+Iv1wrZt2ygpKeHP/uzPSEtLw+VyMTk5yfnz57l8+TITExOMjIwwPj7O6OgoiqJgtVo5ePAgFRUVfOYzn2HHjh2kpKRw/PhxxsbGEi0JIQRms5n09HSys7PJy8sjPT2dAwcOYDQaCYVCjIyMMDk5ecP7HQ4H4XCY3/72twwNDdHR0fGp6iMN5iqh0+mw2WyaC8/tdmtrJeFwmEAgQHNzM++//z6XLl1ifHw8wTW+NfR6PXa7ndLSUrKysjRtCwsLNDQ0bLh9lxaLBbfbrbnRP46srCyysrI4ceIEw8PD1NfX34ZaSuIxm824XC5KSkq0tfV4otEoIyMjBAIBJiYmNC+C0WjEZrNp0d9ZWVmkpKRgt9u1rV/rgaysLIqLi6msrMRqtRIMBhkbG6O9vZ3Tp09rSwajo6PajgU1ql0d3KempuLz+RIa5a7Ocq1WKyaTSQucLC4upqysjJycHB544AFMJhNzc3P09fUxMjKy9NzKRXe7y+Vibm6OhoYGQqHQp66bNJirRFpaGr/3e7/Hnj17OHLkiLa+B9Dc3MzTTz/Nu+++S319/bpy2UHsAc7JyWHr1q386Z/+KW63G4D5+XkmJiZ4+eWX6e/vT3AtV5d9+/bxh3/4h+zatQuv14vNZlvRfS6XC4/Hc11nLVl7wuEwc3NzBINBZmdnr2szv9/PD37wA1pbW2lubkan02GxWPgv/+W/sHnzZrZv305KSgoOh4M9e/YQjUZ5/fXX181AUF2jfe211xgdHeWDDz6gqamJ7u5u5ufniUQi10U/6/V60tPTtXe6paWFc+fOJTRSOCUlhZSUFB544AFKS0s5cuQIKSkpuN1u9Hq9ZkyFENqySEFBAeFwmIWFBUKhEGazGavVyvT0NGNjY2RlZa2Ki10azFVADV8vKysjLy9Pc+dEIhEGBgZoa2vj0qVL9Pf3EwgE1l24vmow8/LycLvdWqTv6OgoAwMDTE9Pr5tO5eOwWCzk5+dTUVFBRUUFmZmZOBwOAObm5picnGR8fJxAIEBFRQV2u33JGnV+fj5+vx+PxwNAMBhMmJaVoCgKfr+f8fHxdfdcLkdtn3PnzjE+Pq5tBTKbzXg8HmZmZujp6dFmJKq7r6mpCbPZTG1tLTqdDoPBgM/no6SkhLfeeivRslZMf38/Op2OwcFBJiYmaG5upr+//yPdlSpqMpWFhQXGxsbo7e1N6NJKeno6hYWFbNu2jeLiYvLy8rBarVitVvr6+picnGRkZER7Xufm5lhYWCAcDhMOh5mdncXhcJCamsrg4CBjY2P09PSsiotZGsxVQK/X43Q6qauro6CgAKvVCkAgEODtt9/m9OnTHDt2jHA4vC47Jb1ez/bt29mxYwcmk0kzEFevXqW+vp65ubl147b6OFJTU3nkkUc4cOAAO3bsAD4MGPD7/Zw/f54TJ07Q2trK97//fUpKSjCbzUBsYHHgwAHKysp45ZVX6OzspKurK1FSVkx3dzdXr16lrq4u0VX5VPj9fqanp/nRj36Ey+XC7XaTmppKdnY2dXV1mEwmurq6tJmGugfx2LFjjI+P88gjj2hbTbZu3YrVauWpp55KpKRb4r333vvE96rLRj09PTQ2NjI7O7uKNbs1KisrOXjwIA8//DBZWVlArK3C4TBvv/02ly5d4tixY1odQ6HQdQYzMzOTgoICmpqaGB8fx2AwrErfu+YGc3p6mqmpKc0lsNEQQmjrBgUFBde5K59//nlaW1sJh8Pr2qioG6DjuXbtGn19fRuiXYUQlJWVUV5ezpEjRygsLATQ1kRmZ2dpa2vjZz/7GZ2dnVr2KoCKigptEGE0GrXEDskQNLESZmdnl0RNqj/rkWg0yuDgIOPj45hMJiwWCzabjcbGRnQ6HaOjo9palhACg8FAZWUllZWVSzRvlFn3zXC5XGRmZrJ161Zyc3OZnp6mu7ubpqamhHiM1CjnkpIS7rrrLux2O4FAgMbGRlpaWqivr+fy5cvaqVdqwKTqZo5Go9rWt7vuuouHH36YM2fO0N3dzYkTJ1bFzbzmyddnZ2cJBoPrdnb1ceh0Onw+H4WFhWRmZmrBMKFQiMnJSU6ePLnuwvOXo24CV0PTIda24+PjDA8Pr/t2VTvO4uJibS3LZrNpxjISiTA1NUVXVxdvvPEGMzMzRCIR2tvbcblcWjg+xGbjJpOJlJSUFa97Jpr5+XlCodC6HtDFMzExsaJy8e1eXFy8xLU+MzPD+Pj4hvmb3IiUlBSys7MpLS3F5XIxPT3N0NAQPT09CamPajBzcnLYtGkTZrOZyclJLl++zDvvvMNLL73EzMzMTd3F6oEQ1dXVfO5zn8PtdnP16lUuXry4KrPmNZ9hqhl/NupJJTqdju3bt7Nt27YlGuvr67ly5cqqRGYlGp1Ox65du7Q1noWFBebm5rQHcb1vJUlLSyMrK4s//dM/paqqCrvdrrnmgsEgo6Oj/I//8T9oaGhgYmKCSCSCXq/nzJkzhMNhDh8+fMN0a+sFn89HcXHxutbwSTAYDNhsNo4cOcKWLVvQ6/Xas/3OO+/w3nvvJdQ1udY8/vjj7N+/n6KiIoaGhnj77bcTGrwXjUa1dJPqtp+uri6efvppuru7mZqa+tjBudvt5uGHH9YGvXV1deTm5vLzn/98VdIerrnBNBqNmsFcr26ej0J94Xw+H16vd4nB7Ovr01yxKqr++HLqSS7JijoKd7lcuFwuhBBEIhEt9+TExERS1//j0Ol0ZGdnU1ZWRklJCT6fTwuCgJhrbmhoiIaGBrq7u7X2VGfYG2EWYrPZcLlcWsDL8mwwGxWPx4PP5yMjI0NbSgmHwwSDQYaGhhgYGNgQyw3LUb0g+fn5lJaWYjQaCQaDSZN4RPXsLCwsEAgEGBwcZHJy8iPbQs3w5HK58Pl8VFZWav2x3W7X9mKuRluuqcFUs4ioL+BGG8F6PB68Xi979+6lqqpqiUvnzTff5NVXX10yQlX3fBmNRq1sKBRK6lGsmuovvv1mZ2cZGxujv7+fgYGBdWsw9Ho9VquV++67j4ceeoi8vLzr1mnVrEwXLlxYcniAoihMTk6u+wMF1L1qHo8HnU5HWloa+/bto7W1lYaGhkRXb025++67uf/++7XAEkDroDs7O+ns7Fy3z/bNcLlcZGVlUVhYSE5ODn6/n46ODo4dO5bQGaa6/9JoNGIymbTZ/szMzE0TSKh91Gc+8xm2bNnCQw89pC2HTE1NMTo6qkXLflrWfIapWn+Hw0EoFEKn02nrO4WFhVqGjZsxMDCA3++nv79fS+2UDHg8HvLy8nA6nVpHGwgEmJqaYmJigpmZGRRFwW634/V6KSkpIScnh7S0NG0E39nZqc1ekjFNntvtJicnZ4lG+DBy9GYpqYQQSX3ep8fjYdu2bVRXV1NQUIDZbNZmlmpquJMnT3Lx4sXrIoF1Oh0VFRVs2rRpifdkvXlRFEXR1q22b9+urVdvtMFtPCkpKRQVFbF161btaDp1RtPZ2cmJEye0jfAbkYKCAg4fPkxOTg6RSIR3332XDz74gGvXriV08K66ZBsaGnj99dc5dOgQXq+X++67j6GhIYaHh7UofXWgajAYyM7OJiMjg3vuuYfCwkJsNht+v5/h4WEaGhro6OhYNV23ZVuJXq/XNp2qiX6zs7O5//77tS0YH4WiKJw8eZKOjg7NGCWLwUxPT1+SJg7QIs3GxsYIBAIoioLb7Wbr1q088MAD1NXVUVJSou1lfPPNN3n22WeZnp5OSoOZnp5OcXExKSkpWp0/DtWNqxrMaDSalK6trKws7r//frZv305BQcGSa8PDw5w7d47f/va3XLx48bp7dTodO3bs0NZ11zO9vb00NzcTDocxGAxYLJYVDWTXK+np6Rw8eJA9e/ZQW1sLfOiKbWpq4rnnnmNwcDDBtVw7KisrefLJJ8nPz2dhYYHnn3+epqamhJ9XrG7zef/99xkZGWHfvn0UFhby+OOP09XVRUtLC263G51OR2dnJxA7alBNMF9XV6flDR4eHubEiRO8+uqrtLa2rtp+6NvyVmRnZ/Pnf/7nQGxNU51hZmZmrqizKSgoYGpqirq6Ourr63nmmWe0zaqJZvko9Nq1a5r7zmaz8aUvfYmysjLq6uq09RI1khagpqYGt9tNMBjk7NmztLW1JVV+2ezsbCorK5cYy9HRURoaGrBYLFRUVHDo0KElEbRCCLZs2YLVaiUUCtHS0sL58+c5d+4c165dS5QUDSEEKSkp5OXlsXPnTjIyMrRr4XCY6elpLl68yDPPPHPLHed6n5Wo6Q/T09OxWCzrNnrW6XRSW1urzVqam5uZnZ3lwIEDbN26lccee4z8/Hyt/NjYGL/61a84ceIETU1NG+pMTIgZlrS0NA4dOsQ999xDaWkpbW1tdHZ2cu7cuaSK5B8aGiIYDPLTn/6UiooKDhw4QGlpqXY0m3pkJHyYstNkMtHb20soFGJ6eprh4WECgQB+v5+pqankT74eH8xis9m0Te/q9oRbGZVnZGSwsLCAwWBAURR+97vfEY1GE24wVY3xjaEGDFitVnw+H7t376asrEw7R1EIQSgU0lxfalqq3NxcOjo6PnVy4NXG6XSSmZm5JAhEHY2rOWX37Nmz5NgrvV5PXV2ddrB0VlYW0WiU3t5eZmZmmJ2dTWgnrNfrtchYn8+Hw+HQ6jM/P8/AwAAdHR1cuXJlxbN+9f715pKF2Dq66nI2mUx4PB5t3Xp+fn7dGEyz2awd8p2WlkZNTQ3RaJT5+XlmZmaYmZlh8+bN1NTUUFlZqS0xzM3NMTY2xvnz52ltbV3xtpREI4TQgiqNRqO2DKKmjYt/Ju12O9nZ2ezatYvy8nLcbrd2wtC1a9eSIthHJRAIMD8/z/nz51lYWKCurg6bzUZKSopWxu12azZAbWP1IPdgMMjc3BwGg0FbB016gxkMBvH7/dpJ35mZmcCHa1u3itFoZPPmzYTDYe69915Onz7N1atXV7vat0QgEGB8fHyJu3F2dpbx8XG+/e1vU1JSQl5eHiaTCaPRSEdHBwMDAwwPD+NwOLjvvvu08+dsNhsOhyPpOlx1nTbeYObl5eFwODhy5Ag6nQ6Xy3Xdmpe6HuhwODh48CC7du0iPT2ds2fP8pvf/CahayU2m40nnniC2tpa8vLyrotu/u53v0tbW9uKTinZCJw/f57x8XG++c1vatlxbDbbkmjhZEcIwa5du9i+fTtf+tKX8Hq9mkFUz6ANBAIUFxfjcDiw2WzodDoikQgnTpzQThFK9lSGKurMSk22UVNTg9FoxG63c/To0euOolNPZlETmgshaGlp4eLFi1gsFi0iNRmIz8B08uRJXnrpJbKzs8nNzQVia51qjtjW1lYg1t9861vforq6mv3792t980svvbSqCSjWzGCOjIzQ3t5Oenr6knRqKpFIhN7eXubm5j6yoZxOJ263Wzv5Q3XnlpSU0NTUtFZVXzE3mmGqro+cnBx8Ph8Wi4WZmRltnai3t5fp6WmysrKWjADVwJFk66DUEx3i66VuzDeZTFrkbzQaJRQKMTY2pq0zWywWNm3ahNVqxWw2k5OTQ25ubsIDSgwGA0VFReTk5GhrdYqiMDY2pkVIjo6OfqJ1V/XoqOHh4Y/N4ZksLE9ckKzP4kehnhJUU1PDrl27yM/P13L5quTl5REKhfB4PNpsTI1QP3fuHBcvXtQSUiQz6vtWXFxMZmamprekpASdTofZbNYSqKhG8kantoRCIbxeL+Xl5XR1dWkuzGRCXXZTE0mos+BoNMrs7Cx+v5/BwUGysrJwu91kZmaSlpaG2WxmaGiI1tZW/H7/qg5618xgXrlyBYvFQmVl5XUHuEJsJvbaa6/R399PV1fXDafMFRUV7Ny5k9raWm2dyePxsHv3bj744IO1qvqnIj09nZqaGrxeL06nk2AwSGtrK08//TTNzc309fXhdDqprq7mj/7ojxJd3U+EyWRasg4LMbfe6OgoJ06coL6+ntHRUXw+H9/97ne1bTSFhYUEAoGE7/EzGo1s2bJlSaBPJBKhqamJixcv0tHR8YlTg83MzDAyMsLFixcTHkRxK8RHM68XF6xKZmYmR48e5dFHH+XQoUPA9Rri16nV61NTUwwPD/Ov//qvWprDZMdiseByuXjkkUeoqanh4Ycf1gZ9CwsL2mkd8/PzGI1GzGbzdQFcoVCIQCDA4cOH2bFjB+FwmIsXLyZl3uNIJMLk5CSTk5MfWb/t27ezf/9+9u3bR2ZmJvPz81y8eJGf/OQn9Pb2rmp91sxgXrt2jba2Nm22sbyDNZlMbN68Wdtioc7WmpqamJiY0DYOd3R0YDAYqK6uJiMjA4vFoq07JRPhcJiJiQmmpqa0Y3Smp6d57rnnaGxs5J133mFiYoJgMEhubu6S47/WM3Nzc7z11lv09PRw+fJlbbS6sLBAeXk5o6OjSZUmLi0tDa/Xi91u19zGqrFQ995FIpGbGg2n04nH46Gqquq6bSXqMzs3N5dUwVsbFbU/OHz4sOayU495Wj5jCoVCnD9/Xmubq1ev0tHRsS4iYlNTU9m0aRN79uxh8+bNVFVV4Xa7tX2j7e3tdHV1MTExwcDAAB6Ph5qaGrZv305ZWZmW2/r48eN0d3fT2tqqbaW5cuXKusl7rGI2m/F6vezevZu7776b7du343Q6NTfsqVOnaGhoWPW12TUzmJOTk1qQx40MpsFgoKSkBLvdrq0bhMNhWlpamJ2dpb+/n/HxcQYHBzlw4AA+n4+0tDRMJhPp6ekfux3ldqDuMRVCaAZzenpac/VEIhFef/11mpubuXTpEvBhlPB6OTPxZvso1YjSd999l8bGRo4fP04wGCQUCmG327FarUxNTWkJsJNhT6bqurFarUtmutFolL6+vhUlk3c4HGRkZFBcXKx10iojIyP09/cTCtSLIZoAACAASURBVIXuiPXPRKIe2p6Zmcm2bdu0WaT6/g0PDy953qanpzl+/DhTU1OEQiHOnj1Lc3Nzoqr/scRv5FeTpD/44IMcPnxYm0mOjo7S0tLCqVOnOH/+PAMDA3R1dZGfn8/8/Dw+n4+ioiJmZmYYGBjg2LFjXLp0iTNnziRa3idC/ZvY7XYKCws5evQodXV1VFRUEA6H6e/v5/XXX6exsXFNZsxrZjDHx8e1yKWZmRnsdvsSA6FmFXE6neTl5QGxzvnw4cNMTExw9uxZnE4nGRkZVFVVkZ6ejsFgSJo9mBDbz3TvvfficrlQFIVQKER9fT3PP/88v/vd79DpdLz55ptapKXdbtcORlVzVyY7qtsqPiI5HA4zPz/Pc889R319Pc8++6wWvq2eGHDvvfdSW1uLz+fTBjfqOXaJXCcqKiqiuroah8OhGczx8XFGRkZ47733uHLlyscaupqaGvbs2XPdUoOiKPT399PR0ZH0a2Efh9pRJ+ugTo38/NznPsddd91Fbm4uRqORcDjMc889R1NTE6+88sqSWb6iKFouYEVRkm7NbjkFBQXk5uby9a9/ndTUVCwWC263m/b2dp5//nk6Ozu5cOECk5OTTExMoCgKNpuNr33ta1RUVHD33XczPDzM7373O5566im6urro7u5eN4FNNyItLY309HS++MUvUllZyZEjR7Db7Vq7X7p0aUmfu9qsmcFUzyUbHBzUMjEsNxBqYvZ4d11aWhqZmZksLCxgt9tJTU3VFnJVkmGmAiwJawa0MPyxsTH8fj/hcJjx8XHN2KgRpwUFBWRnZ2szUzXZcDAYTLpZyfT0NCMjI0s6nkgkQigUor29ncuXLzM8PKyFbrtcLlJSUqiqqqKiogKr1apFI46MjDA4OJhQjRkZGVrnqtPptIi8QCDA2NgYk5OTH/lsqUZEPeHBbDZfV3ZmZobJycmka8dbxWazkZqaqj3HyYYa4KJuWlfbIhKJMDw8TG9vL729vdrARQ0adDqd2iAgPjBIvXd2dnaJIVWXV27nQF3dJlJYWMimTZuoqalBURTtLM+5uTnOnz9PV1cXDQ0N2lmQeXl55OTksHXrVoqKikhNTaWhoYFLly5x6dIlBgYG1q2xVGeWmZmZ5OfnU1NTQ0lJCWlpaVoQUEdHB21tbUxMTKzZlsM1TVwQiUR48cUX6evro6SkBKvVuqIMIna7nV27dt0w3Zi65yYZOqRz587h9/vZu3cvTqcTr9erDQ4uX77M6Ojoknru27eP++67j5qaGm0N0+/3MzAwwNWrV69L1p4MdHR0oNPpeOSRR7REE6FQSPMCHD9+fMk+p7vuuovdu3fz5S9/mZycnCUnQLz33nu8+eabCdtSoiZU2Ldv35IB2NzcHFNTU0xNTd10w7rVasXr9bJjxw4OHz6M0+m8royat3K9zzDLy8s5ePAgY2NjSXnijtFoxO128/jjj5OdnQ3Egl7U4wQBzdAA5Ofnk5mZyZ49e5a0vbqGrS4vXL58WXNXRiIRZmZm8Pv9qx48cjNSUlLwer088cQT7NixA4/Hw6lTp/je977HyMiINohR4z7MZjNOp5OHH36YnTt38uCDD7KwsEBLSwvPP/88L774IuPj40nXt9wKBoMBq9WqHex+3333Ybfbgdj2PrU/unz5shZZuyb1WJNvXSQajdLa2sr8/Dy/+MUvqKqqYtu2bSvKVXmjxAbz8/P09PTw5ptvJsUG/4mJCbq6uujt7cXhcJCenk5BQQH79+/X8sgGg0EyMjKora3l4MGD1NTULMmKMzAwwHvvvUdfX592TzIxPj5Od3c3AwMD2vqfuq3k0KFDuFwurly5gtFoxOv1sn//frZt24bb7dZcnkNDQ7S0tNDT05PwQ3nVbTLLudmpMWqC8qKiIu6//342b96M0+lc8j29vb1awvLu7u51ZTDVGXZvby96vV4bGCXzkoE64zAYDEvqqdPp2LZtG/n5+Utmih6PB4fDQVFRkVY+vo9RvSZpaWna4eHRaFRL3t3Y2EhfXx+jo6M3PTnj02A2m/H5fGzZsoW9e/eydetWUlNTOXPmDOfPn2d4eJiZmZkliU98Ph9lZWVs3ryZvXv3kpOTQ2dnJz09Pbz11ltcunSJ6enpdfU83oiSkhJ27drF/v37KSsr0/rQhYUFGhoatPdOdU2vFWtuMBsbG+no6KCzs5MvfOELFBYWkpqaqmWjuJXvCgaDNDc38z//5/+kr69vDWu+MsbGxpiZmaG1tVU7Wqa8vByr1aqdCD40NMTmzZv52te+RllZGT6fb0kWjo6ODl566SU6OzuTMpfsyMgIU1NTdHZ24na7tZRpFouFRx55hN27d/Ozn/0Ml8tFXV0dW7dupaSkZMl3dHd38/rrr9Pe3p7UWVQ+KvuUXq8nPT2d2tpavvKVr5CamqqNbtV2bG1t5dlnn+X06dMJO4D3k6KmE2tra8NgMGhJRmD9ZS7S6/UcOnQInU53Qw9APPHa1HZUc8uqn01NTdHX18fp06c5fvw4ly5dIhgMromXxGq1UlNTw0MPPcR//I//EYgNyl9//XXNY6WiHrm3Y8cO7r//fr74xS9iNBqZm5vj17/+NSdPnuSf/umfVr2OiWLLli38yZ/8CZWVlZorXV3OOn36NC+++CItLS1rntLwtuSSnZ+fp6+vj2effZYPPviAuro6vF4vDoeD7Oxstm3bhsPhuC6598DAAO3t7dqorr6+Xksf90n3ya024XCY1157jZGREaqrq7VcnF/96lfx+/3Mzc3hdrspLS1dEvgUiUSYmprStmMkU2qq5UQiEV544QXa29u1DjUzM5Ps7Gzcbjdf+cpXNBeZmr5qamqKwcFBXnjhBerr6zl9+nTCw/fVjC+ZmZmUl5dr65hOp5OsrCzy8vK0teT09HRycnIoLS0lKyuLvXv34vV6SU9P14Jh1ECvrq4uzp8/z/HjxxkfH0+oxk+Cekp9VVWVll91bm4uqWcm6kkj8e5V9Qi61ZphqIFF+fn5OJ1OqqqqGBoa4h//8R9pa2uju7t7VWczbrebz372s1RXVxONRnnllVe4fPkyb7zxBuPj49jtdi172MGDB8nJyWHTpk2kpaVp7+iVK1d49dVXkyo37CfFbDaTlpbGwYMHOXLkCJWVlTidTiKRCH19fXR1dfHuu+/yzjvvcPny5duy1HNbDGY0GmVmZoa2tjba2tqYn58nPz+flJQU7b9ut1sbtauoQSUDAwOMj4/z/vvvMzo6mlTGRVEU2tvbsVqtDA8Pk5aWpp1Ool6/UdYUNWfp0NAQo6OjSRX9u5xoNMrVq1eJRCLU1dUBMReX1WrFZrNdFzwRDocZHh6mo6OD9957j7a2Ntrb25PC3Tw4OEhXVxfz8/NaQJK6BpSbm0soFGJmZgafz0dxcTFbtmzROqj4VGuRSISFhQWmpqZob2+ns7OTvr6+pBnI3QrqvtL09HQt4jvZDaY6wIl3x6rbvNT2UYPobiXuQc2ko9fr0el0mEwmHA6HFrGvbk1ZWFhYElS0GpjNZoqKinC73YRCIVpbW6mvr9cOKc/NzWXTpk2Ul5dz9913k5WVRUpKCnNzcwwPD3P+/HnOnj3LpUuXknLdeaWo7ZiVlUVubi47d+6kvLxc62fm5+cZGhqira2NU6dO0dbWdtv2kSbkDJ/Tp09z9uxZ7Riov/7rv76hUVE7pUgkklTBPvFEo1FaWloYGRnh29/+tuYesdls2hFXN6K3t5dvfetbtLW1JX0nG41GaWtr49q1a0xPT3PkyBEeffRRcnNzr0sgEQgEmJyc5C/+4i+4fPkyjY2NSZXA+7XXXuPKlSt8/vOfRwihuVfNZjPf+c53tOhsi8WiBampJ9Sr+P1+zYXZ0tLCD3/4Q0ZHRwkGg0mj81a499572bVrFy6XS3NLd3V1cebMmaTdeqEm477R++X3+xkfH+fcuXOMj48zOjqqrQF+HEVFRVRWVpKdnY3H42HXrl04nU5cLhcWiwWTycS3vvUtLly4wKVLlwgEAqsekTkxMUEgEODatWtEo1GeeOIJiouLueeee7SJhcViYWRkhKeeeorGxkYtQ5Xf71/XxhJiQZ8ej4e/+Zu/obq6Gp/Pt8STMDs7y/PPP099fT1vvfXWbT2EIyEGc7036HLC4bC2lpmTk0NBQQFlZWV4PB5cLpcW9qxGuKmJ2NXMHOuBcDhMIBCgp6eHixcvaqeYLE8gMTs7y8zMDE1NTfT39yf8ZJLlqDMnv99PWloaqamp2sAtMzNT28umJsWHD9e35ubmmJyc5OrVq3R3d9PZ2Ulvby9DQ0NJp/NWyMzMxOfzLclYpK4PJasmdQYYbzAnJibw+/3U19czPDzMlStXmJqawu/309zcvKLcvnNzcwSDQTweDykpKfj9frxeLxUVFXg8Hmw2G6Ojo1qu4bX4+6hJ1NV/Mzc3F6/XS1ZWFqFQSNs3PDAwwLlz5+js7KSrq2vdG0t1ZllYWEhZWRnFxcXa4fXqNsXu7m76+/tpbGykp6fntk82Nu4psbeZ2dlZGhoa8Pv9XL16lT/8wz9ky5YtbN68mVAoxMDAAG+++SYXLlzQjtZRz29bL4RCIW0N+de//nWiq/OJUffqud1ufD6f9nn88UE3mrlMTk5y7tw5fv7zn/PWW28xNjaW8CPmPi1CCHJzcyktLU14jt9PS0tLC/X19fzd3/0dvb29n+gcxKGhIe3AcJ1OR0pKCjU1NXzpS1/irrvuoqSkhJdffpnz588zOzu7Jls1UlJSKCoqorS0VPtMTfXX3NxMc3Mzb7/9Nr29vdoRWMnqOr8V9Ho9FouFI0eO8NnPfpbS0lItcGt2dpaJiQl++ctf8sEHH/DGG28kxDMnDeYqMzExQXNzMz/96U9JT0/H4/EQjUa1sP3R0VHMZrOWQm4jPOjrjWAwyNNPP011dTXT09MUFhbi9XpvWLanp0fLpNLe3s6rr75KY2OjljN4vaMoCs3NzaSmplJeXk4wGKSrq4uBgYGknmHW19czPz/PE088wcLCAm1tbfz2t7/lxIkTDA4OrsqMX31v29vb+cUvfsHx48dxu928//77a5KxanJykpdffpmKigpqamq0OszPz3Pt2jXq6+vp6upicHBQ24YWnzhlPWMymcjJyWHPnj3s3LmTwsJCTCaT1gb19fW8//77WkxEogaq0mCuMupBtcmw7UVyY+bm5jh27BhDQ0PauYnxs0v4cIap7kHt7e2lqamJ1157jbm5uaQO0rpVuru7cTqdTE1NEQgEuHLlCiMjI0kZM6DS0dHB/Py8drhDfX09p06d4vjx46tqyOKD89T17Lm5uTX5u8zMzHDy5En8fr+mQTUYnZ2dvPHGG9o2r42Emhs2NzeXQ4cOUVFRQVZWFkIIZmdnGRsbo6WlhWPHjlFfX8+1a9cSVldxs1GYECI5h5crRFGUj91EJjUmP2uhUT1M1+PxaEEUNyIYDDI/P69lK1I3Rq/2zCuR7eh0OrFarWRnZxOJRLTDd1e7Y15NjWoEa2lpKYqiMD09rR1+sFaogYk3M5afRqNOp9OCzeLTharnV05PTyfFjHK1n1WXy8VXv/pVampquPfee7WDEY4dO0ZLSwu//vWvuXbtGkNDQwSDwduSseijNMoZpuSORB25J2sU6O1kenqa6enphI7cb5VoNMrc3BwNDQ237d9c6xzWanKWYDC47o7b+qR4PB4yMzO1/NpWq5VAIMDMzIwWZa/usUwGr440mBKJRCJJCEePHmXz5s2kpqYSCoVoamri7NmzNDQ08MILL2hLA8mCNJgSiUQiSQitra3MzMzQ0tKC2WzGZrPR3d3N4OAgfr8/6aLQ5Rqm1Jj0SI0xpMbkR2qMsVE13jjbtEQikUgkkiVIgymRSCQSyQq4qUtWIpFIJBJJDDnDlEgkEolkBUiDKZFIJBLJCpAGUyKRSCSSFSANpkQikUgkK0AaTIlEIpFIVoA0mBKJRCKRrABpMCUSiUQiWQHSYEokEolEsgKkwZRIJBKJZAVIgymRSCQSyQqQBlMikUgkkhUgDaZEIpFIJCtAGkyJRCKRSFaANJgSiUQikawAaTAlEolEIlkB0mBKJBKJRLICpMGUSCQSiWQFSIMpkUgkEskKkAZTIpFIJJIVIA2mRCKRSCQrQBpMiUQikUhWgDSYEolEIpGsAGkwJRKJRCJZAdJgSiQSiUSyAqTBlEgkEolkBUiDKZFIJBLJCpAGUyKRSCSSFSANpkQikUgkK0AaTIlEIpFIVoA0mBKJRCKRrABpMCUSiUQiWQHSYEokEolEsgKkwZRIJBKJZAVIgymRSCQSyQqQBlMikUgkkhUgDaZEIpFIJCtAGkyJRCKRSFaANJgSiUQikawAaTAlEolEIlkB0mBKJBKJRLICpMGUSCQSiWQFSIMpkUgkEskKkAZTIpFIJJIVIA2mRCKRSCQrQBpMiUQikUhWgDSYEolEIpGsAGkwJRKJRCJZAdJgSiQSiUSyAqTBlEgkEolkBUiDKZFIJBLJCrjtBlMIcUIIUXsL5bOEEE1CCPNa1ms1EEKYhRCNQgjvLdyzRQhxci3rtdoIITKEEFeFENZbuOdzQohn1rJeq8kdonHDP68bXeNG16eSLO/jqhtMIUShEOIVIcSEEGJICPEjIYRh8drngGlFUS7Ela8SQrwghPALIaaFEG8JIfaq1xVFGQbeAv5ktev6SRBCfE0I8YEQIiSE+Jdll/8EOK4oymBc+b1CiDcXtfmFEC8KIarU64qi1AOTi3+bpEEI4RFC/JsQIiCE6BZCPBl3+TvAvyiKMhtX/kEhxJnF8mNCiKeFELnqdUVRXgSqhRBbbqOMmyKEeFsIMSeEmFn8aY67vFE0/lQIMSiEmBJCtAghvhx3eUM8rxtd4x3U5yT/+6goyqr+AK8A/wJYgGzgMvD1xWsvA38QV7YEmAD+L8ADOIGvAzPAnrhy+4CG1a7rJ9T3e8DngX9YbMD4a1eAfXG/71nU8ueL2jzADxY1F8eV+wPgpURrW6bl58AzgAPYD/iBasAMjAK5cWUfBaaAJwHrYrv/L6ALSI0r9z3gR4nWFleft4Ev3+DzjaSxGjAv/n8FMATs2EjP60bXeAf1OUn/Pq6F6Cbg38X9/lfAPwImYHaZ6H8FXrnBd/wDsVGT+rsBCAIFiW7UuDr9IP7hBfIX9RniPnsX+Psb3Ptb4Km4332L95oTrWuxPnZgHihf1lb/HTgItMV9LoBu4D8v+w4d0AD817jP9gGdidYXV5+PekE3jMZl9d0EDAK/v5Ge1ztF40bucxbrlPTv41qsYf4t8LgQwiaE8AEPAL8DyoCooih9cWXvA569wXf8Etin+qsVRQkDbcDWNajvalEDdCzWFSGEDdjLR+u7T/1FUZR+YIHYy54MlANhRVFa4j67RGwkXwPEu0o2EXtxl+hUFCUK/Jo4ncQGU4VCCNdaVPoT8hdCiFERW1s/vPjZhtIohPh7IUQQuErMmLzCxnpe7wiNN2Aj6kvq93EtDOZxYh3rFNAHfAA8D7iB6WVl04k93MsZXKybJ+6z6cXvSFaW6/MQ0/BR+tKXfZZM+hzE2i8ePzEXz3Kdqo6V6FTvSxad3waKiY22/wl4UQhRwsbSiKIof0as7Q4AzwEhNtbzekdovAEbTV/Sv4+rajCFEDpis8nniLn10oFU4C+J+dCdy24ZBW4U3eUFoov3qDiBydWs7yqzXN8EMQ0fpW902WfJpG8GWD4icxF7+JbrVHWsRKd6X1LoVBTlfUVRphVFCSmK8hPgBPDv2EAaVRRFiSiK8h6QC3yFjfW8AneGxmVsKH3r4X1c7Rmmh9hU+UeLoseA/4+Y6DZALLppVd4AHrvB9/w+cEpRlCCxmwxAKTG3YLJSDxSpEcGKogSAU3y0vmPqL4t/ExNL3Q6JpAUwCCHK4j7bSizAoJ6Yy1almZgnYYnOxcHTI8TpBCqBLkVRls9ekwWF2PrIRtZoIBZst5Ge1+XcCRph4+tLvvdxDRZuO4iFABuITYP/DfjZ4rUXgCfjypYRs/zxUbL/OxBgaeTXXqAx0YvSi3UxEIsA/gtigTAWFhfdiTXs3riy+xe1fH1RWyqxhftJoCyu3JPcIPgpwTp/QSxS1k5s4VyNkjUBI4AvruwX+TBiTY2O/l9AD5AWV+673CAgIUH63MBRtf2IRQ0GiL2YG0VjJvA4MRe7flFvAHhoozyvd4jGDd/nrJf3cS2EbyMW7TRBbGr8SyBr8dpngd8uK78ZeGlR/MzivfuXlfk7FremJPoH+D6xkU/8z/cXr30V+Idl5fcvappZ1PgysHlZmZfVFzxZfogNYJ5ffGh7WDrQ+Svg28vKPwycXSw/TszY5i0rcxnYmmhti3XJWKzv9GJnchq4bwNqfGdR39Ri3f63uOvr/nm9QzRu+D5nvbyPifjDnABqb6F8JrFIJ0uiG3UFdTUDjYD3Fu7ZQsz9nPD630KdM4hFI1pv4Z7PAb9MdN2lxiX13fDP60bXuNH1xdU5Kd5HsfjFEolEIpFIboJMvi6RSCQSyQqQBlMikUgkkhUgDaZEIpFIJCvAcLOLQoh1vcCpKIr4uDJSY/IjNcaQGpMfqTHGRtUoZ5gSiUQikawAaTAlEolEIlkB0mBKJBKJRLICpMGUSCQSiWQFSIMpkUgkEskKuGmUrOTTodfryc3NJRwOMzk5SSgUIhwOJ7patwW73Y7ZbMbj8WA2m7FYLIyOjjIzM8PExASKoqDX67HZbJjNZiYmJu6Yv00yotPpSE9Px2azkZaWtuSa+vwGg0FmZmaYn58nGo0is4RJ7jSkwVxDXC4X3/zmN5mYmODll1+mq6uL4eHhRFdrzdHpdFRWVlJaWsrv//7vU1xcTHl5OT/+8Y85e/Ysv/rVrwiHwzidTmprayksLOT5559ndHT5cX2S24XNZuOxxx5j586dfPGLXwTQDOLo6Cgvv/wyFy9e5OTJkwwMDGiGUyK5k1h1g6nX6zGbzZhMJgwGA4qiEA6H8fv9WpmUlBScTifbt2/H5XKRlpaGELFtLwsLC8zMzNDU1MTo6ChDQ0OEQiEikchqV/W2YDKZ8Pl8PPjgg7z77rtcunSJsbGxDTebslqtZGZmkpeXR3Z2NrW1tfh8PjZt2oTD4WB2dpaBgQF6e3tJT08nOzubAwcOUFhYSEZGBoFAgM7OTs6cOUM0Gk20nDsCvV6PyWRi69atFBYWcujQIYqLi7FYLEvKpaamUltbS0ZGBiUlJdTX19Pf38+pU6cIhUIJqbvL5cJut1NUVITBcH03ZrVaKSoqYnx8nKGhISA2kEtNTSUQCNDT08Pw8DATExPX3bvR0Ov13HPPPaSlpeFwOGhqauLs2bMsLCwkpZegsLAQn89HVVUVNpttybVoNEooFKKhoUEbsM3MzBAIBBgaGmJhYWFN67bqBtNoNGoG0WazEQ6HmZubY3p6WusIMzIyyMvL48tf/jJFRUVUV1drBjMYDNLf389TTz1FfX09s7OzjI+Pr1uDqSgKeXl53HvvvUQiESYnJ5ment5wBtPpdFJZWck999zDzp07qampwel0YjKZtE6rsbGRq1evkpeXx4EDB/je976H1WrVXLPnz5/n3Llz0mDeJkwmEw6HgwceeIDdu3dz4MABzGbzdZ2o3W7nrrvu4q677gLgjTfe4NKlS1y4cCFhBtPj8ZCTk8MDDzyA3W6/7np6ejoPPvggV69e5fTp00Csb6qoqKC/v5833niDs2fP3hEG02g08vjjj1NTU4PP5+Opp57i8uXLRKPRpOyHqqurOXjwIP/hP/wHsrKyllxTlwf++Z//menpaXQ6HT09PQwODuL3+wmHw2s6CFg1g2k0Gtm0aRObNm3iwQcfxOPx4HK5iEQiBINBmpub6evro7Ozk9raWgoKCti2bRspKSnAh+4fk8mE1+vlySef5O6776a7u5uf/OQnNDQ0MDExse46UyEEFouFjIwMqqqqGBsbo7Ozk2AwmOiqrQpCCBwOBxUVFfz7f//vqaioID8/H5fLRSAQ4O2336a+vp5Tp05x/vx57SE3mUzY7Xb0ej0A+fn5jIyMoNPJOLS1RqfTYbPZ2L9/Pw899BC7d+/G5/NhNBpXdH9VVRUWiwWTybTGNf1o0tLSKCgo4IEHHtD6kHhMJhM2m43y8nIyMjKAD5/VsrIyNm/ezCuvvMKpU6e4ePEifr9/w7yT8TidTjweDwUFBeTl5ZGSkkJubi6VlZU0NzczOTmZ6CpquN1uSktLOXr0KEePHr1hu+p0OlwuF4888og2iQoEAgSDQUZHRxkZGeG9995jbGyMsbExWlpaVlXjqhhMnU6H2WymuLiYLVu2cOjQITweD06nk2g0SjAYxOfz0dbWRkpKCtu3b8fn8+FyuYhGo/T392OxWDAajdjtdmw2G5s2bcLr9VJQUMDJkycZHR1lampqXRlMddas1+uxWq14vV5KS0uvc3mtV1TXu9frpaioiC1btuD1eklNTSUYDDI2NsaVK1f44IMPeO+995a45XU6ndZBR6NRjEbjijvs9YDBYECn02EwGDAajej1evR6PUIIotEos7OzBAKBhNRNr9fjcrkoLS3lwIED5Ofn43A4ALRZh3r+n9FoRKfTac8yxNyhqampCR3cuFwuMjMzKSoqwu12a58vn12kpqaSmpp63bXc3FwGBgYIBoOMjIyg1+uZnZ1NShflp8HhcJCRkUFqaioulwuLxYLdbsftdifd+2axWLQ+v6SkBCHEde0hhMBoNFJaWqo9k5FIhGg0yvz8PCMjI4RCIYaGhhgcHGR8fJz5+flVa9tVMZjp6enk5eXxjW98g/z8fHJzc7WXSafTYbfbqa6upry8nKNHj2I0GllYWODs2bOcPXuWf/7nf2bfvn3U1tby+OOPk5mZiRACp9OJ3W7nj//4j9m7dy/fJLiSgwAAIABJREFU+c53GBsbWzfuWbXTiUajRKNRduzYQWlpKT/+8Y/p7e1NdPU+NXV1dZSXl/OFL3yBvLw8Nm3ahKIoBINBnnnmGZqamnj99dcZGRlhbGyMaDR6w1lJJBLh+PHjnDt3bt207c0wGAwUFxfj8XgoLCyksrKSgoICsrOzMRgMBAIBXnjhBX784x8npH4pKSkcPXqUvXv3Ul5ers3yAfx+P83NzQQCARYWFqisrCQlJWWJUbpy5QpXrlxJaNDPtm3b2L9//6ea5R45coS9e/eyc+dOLl68yF/+5V+u+RrY7Wb79u0cOHAAr9eLxWIhEoloBiTZJh9CCPR6/XUDtI9DvUfdlfBHf/RHzM/PEwqFePrppzl//jy/+c1vmJ2d/dR1XBWDmZOTQ0lJiTa7EEIscXEYDAYcDocWCAQxkW63G7PZjN/vp7W1lWg0ypYtW5idnaWgoED7A3q9Xubm5nA4HMzMzKxL14kQArPZjN1uX/duR6fTSWpqKtu2bWPz5s2UlJRoM47BwUGGhoa4cOECra2tDA8PEwgEbvpyKoqiLdwnG+oMMTU1FYPBoP0uhKCnp4dwOIzL5cJms+FwOHC5XJrbLyUlhfT0dHw+H+np6QDMzs5q22sShRACk8mE0WjEYDAQjUZZWFhgeHiY7u5ujh8/jtvtJiUlRdv+oxKNRmlvb6exsTGhxsXhcJCSkrLkXVpYWGBhYYGRkZEldRNC4Ha7sVgsWK1WrTNW3crFxcUEg0EKCgoYGxvbUOuaTqeTzMxMjEYjiqIwOzvL1NQUo6OjCVt/vhk3MpTqst709PSSz61WKx6PZ8m9er0eu92O1WolEolQVFSE3++/YWDYJ2FVvqW2tpa6ujq8Xi9Wq5WFhQXa29vp6uoCYg93VVUVLpdL80sbjUa2b9/OxMQE6enpXL16lXPnzmnRs1/60pc0kXl5eZhMJjIzMwkGg+vGYAohEEIsGTHdysgpWfH5fNTV1fHkk0+ydetWTCaTFg197tw5Tpw4wa9+9asVbaFRFOX/Z++9o+O6zkPf35legMFgMBgAg95JAATAApoCRYqSVWxVW3KLIyfXsZ9XXuJnZ92s5L6855Qbp64beyW5iRU7jp9si7IdR7JsySomKVaJnSAK0XtvA0zBDAaY9v4AzjFAsYAiSMxA+7cWF4E5Zwb7m3PO/vb+KpFIhPn5eYLBYNyZxMxmMxaLhV27dmGxWJR7WKfT8fzzz+Pz+aisrKSgoIDy8nKqq6txOp2UlJQAKA/63NwcJ0+eZGRkhMuXL9PX17dhMsViMYLBoBLwEQ6H8fl8HD9+nDNnzvAf//EfPProo8oOzmg0AiiK9dSpUxw5cmRdVuzvB9kXebXCnJ+fx+Px8PbbbytKT34Gd+3ahdPpJC8vb9UCQKVSUVFRgdls5qGHHqKhoUEJEtoMpKSkkJWVhVarJRwOMzs7y9DQEK2trRs9tDUhuy+Ghobo7OxcdczpdPKhD33omu9TqVSoVCqqqqoU19F6sO5RsvLkd/r0aQ4fPkw4HEan05GRkUFFRQU7d+6ktLRUsaE7HA4efvhhjh8/TktLC6dOncLtdrNnzx4cDofisE9EVppk5Z/jTSGsFY1Gg8lkorCwkH379vHEE0+Qn5+PRqNRdieXLl3irbfe4syZM+9ZDcpotVpFwcCSQvF6vVy6dImmpqa4MxPV1NRQWlrK448/rlhJZN9Kamoq4XCYjIwMdDodOp2OYDBIe3s7b7/9Nl6vl8nJSYLBIMFgUMlfnJqawuv1bphMc3NznD17lnA4jNfrxePx4PV6OX/+PB6Ph7y8PHbu3Ml9991HSkqKomDGx8fp6+ujv7+fqampDbtWsViMY8eOMTIyQl5eHvPz80xMTLCwsMDi4iKDg4OrzMWSJHHq1ClSUlLIy8sjNTWV3NxcamtrKSoqUqK0y8rKGBkZ2RCZ1huz2UxmZibl5eVs2bIFg8FALBZLqOIpi4uLTE9P8+abb9LZ2UlLS8uq4+np6Zw6dQr4tX9a3nDp9Xr0ej39/f2MjIysm6tnXRRmNBolEokoymFxcZH29naOHTummNlMJhN79+5Vopx0Oh3JyclKEFBbWxvRaJTW1lbC4TADAwPodLqEVpgrSXSFqdfrsVqtVFRUsHv3bh588EFg6drPzc0xPDzMO++8w5kzZ7h8+fI1P0M2BRYXF+N0OgEIBoPMzMzQ2dlJb2/vXZNnLUiSRGlpqSKv2WxWfLTBYJCSkhKi0Sh6vR6v14vL5aK1tZXh4WHeeustJicnGR4eZnFxMa4mqWAwSEdHB4uLi8zMzOByufB6vfT19ZGamkpVVRVbt26lpqZGUZbhcJipqSmam5sZGRlZFcC1EVy+fJne3l7S0tKYmZmhvb39pu9Rq9Xk5uaSnZ3Njh07lLxSOSgvJydnla82kTGZTBQUFFBQUEBeXh6SJDE/P8/8/HzC+GnlIJ4jR47Q2tpKU1PTquNWq5WWlpZVVrv09HTFBWg2mxkdHWVsbCy+FGZLSwvhcJiPfvSj2O12bDYbjzzyCElJSfzgBz9gYmKCQCDA6dOnaW1t5e2336a0tJSvfOUruFwuOjo6FBOKHKUnB8okMitNsiqVilgslpD+S61Wy2/8xm9QVVXFo48+is1mQ5IkRkZGGB0d5W//9m8ZHh5maGjohjunjIwMcnJylMhMgNOnT3PixAlcLtfdEmdNmEwmLBYLTqeTzMxMJf/L5XIxODjI+Pg4/f39TE9P09bWhtvtxu12EwwGWVxcxOv1Kj61eLyPY7EYw8PDShENrVbLvn37qKmp4VOf+tQq06XL5eKtt97i3Xff5ciRI3GxC5uenmZmZobh4eE1T4aRSITR0VG8Xi+zs7Ps3r07oRexN6KoqIgvf/nLVFdXK/PQzMwMBw8e5Pz58xs9vJsSi8UYGxujs7OTo0ePXnNe8fl871kodXV10dbWpvgz5WdwveIj1kVhulwuRkZGGB8fR6vV4nQ6yc3NZdu2bWRkZCh+HK/Xi9frxWg0EggEaG5uZnp6mv7+fmXFKvsmkpKS0Ov1wNLqVl6lx+Pkcy3k8G25VmqiYjKZSElJobKykm3btlFUVKQofZfLRX9/P83NzUxMTNw0kEWOpnY4HIov2+Px3JUKHbeCWq0mPT2d0tJScnNzsdlsijm1r6+P4eFhJiYmGBwcVFJn5ubmrmuGjlcikQjBYBCj0UhKSgo1NTVUV1dTWlqq+C1DoRBer5empiba29sZGBiIi2slj+FWAldUKhXJyclYrVYyMjIwmUyKspSPmUwmNBqNYjFLROTiMQUFBVgsFmBJuUxNTdHe3p6Q5Tk1Go2Sjif7sFcWu1npk7+T7o51UZgDAwO43W5+9atfsX37dpxOJxUVFeTl5XHx4kVsNhsnTpxQlF1XVxd9fX20tbURCoWUVBG1Wk1tba3yT67g4Xa7mZycxOfzbVigwa2Sl5dHUVERlZWVOJ1OxY+ZKApfpqKigm3btvHYY49RUFCwaod89uxZzpw5w/j4+JpWcHV1dezevZusrCwl7y/eUKvVJCUl8ZGPfIQ/+IM/ICsri3A4zDe+8Q2am5s5dOjQKt80kLCFyOUIyoqKCoqLi/mjP/ojJRoYllb5U1NTdHR0cPDgQTweT1xGVq4Vg8HAww8/TE1NDU899RQOh0M5ZjKZ2L59O83NzaSmpuLxeBKyVq5arcbhcOBwOEhLS1NSSS5fvsylS5c2NFjrVpAkiYyMDEpKSti/fz8TExOKotfr9XzoQx9SNiJyRP7MzMwdd32si8KUncmNjY1KKbisrCwsFgv79+/HZrPR29uL2+3G6/UqEZWzs7NEIhFltajRaMjPz6ewsBCdTodKpVIKG/T29uL3+xPmJpZ3ZikpKZhMpoSLkk1OTiYvL4+9e/eye/dubDabMpFeuXKFs2fPcuzYMXp6em56TfR6PSaTiaKiIsrLy9Fqtfj9fkZHR+nq6qK7uzsuJmLZv15XV8eWLVuwWq1K2a1Lly7R398fF+O8XTQaDampqZSXl7N9+3YqKirIzs4mOTl5Vfi9vJJ3OBzcc889tLe3c+XKlQ0c+a0jF4+oq6ujoKCAD3/4w+Tl5WG325VdNCztwI4ePUp7ezvBYDDhFrYyOp2OnTt3UlFRQXJyslKcQHYZxKuLAFD0guyOkzMjHn/8cbxer2LB0Wg0FBQUKPdqTk4O/f39Ss733NzcHTO1r1uU7OLiImfPnsXtdpOWlsb+/fsVYUtLSzl27Bh9fX3KdlnOvVuJWq2mvLycsrIy9Ho9kiQRiUTo7e2lpaUloVa4cjUNq9Wq7KZkX0IiYLPZuOeee/joRz/Kfffdtyos++zZs3z9619nenp6TQnQRqOR9PR0Kisrqa6uRqvVMj09zaVLl2hoaKCxsZFgMHinRbopGo2GtLQ0HnnkEaqrq0lKSuLtt99WunQkmsn1esgNAfbs2cMzzzxDWVnZdYNdLBYL+fn5PPnkk+h0OiU4L55Z+YxptVoMBgNPPPEEe/fupbq6Gr1ej0ajUc6LxWK4XC5++tOfKkUb4l3G66HX67nvvvuorq5W3B6y/11uqxevyGlL4XCYSCSCXq8nOzubZ5999prny9dvcnKS8fFxhoaGVqVMxbXCjEQiSpRgIBBQulfYbDays7P57//9v3P27FkOHz68KshH5v7776euro6PfvSjZGdno1KpFN/oK6+8wrlz5xLClCCze/duJY8tFoshSRKNjY1KgEi8otFoKC4upq6uji984Qvk5+crSc/T09O88cYbHD9+XEl8vtHEIiug8vJydu7cSWFhIUajkcHBQc6fP8+//uu/0tfXx8LCwoY+yGVlZRQUFPCxj32M3NxctmzZQlJSErFYTAkQkRsJJNI9uBK5pNj+/fspKSnh0UcfJScnh/z8fMX1EQqFGBkZ4d1331VqAlutViwWCwcOHCASiTA7O8ulS5eYnJzcYIney4MPPkhdXR1VVVXKIlUOuCstLcVms2E0GvF4PPT29rK4uEgwGOTIkSP09fVx+vTpVU0iEg05ab+uro7CwkIApUTckSNHaGtri+tKWm63m0uXLnHlyhXKysqu2TnnWqSkpKDX6/na175GW1sbf//3f8/09PQdCSRc1zzMhYUFXC4XgUCA7u5uBgcHMZlMGAwGdu3axeLiImNjY0xPTxMMBhW/pdFopKKigvr6eoqKirBYLCwuLiqFytvb29+TtBrPSJJEbm4uxcXFSiUVgNHRUVpbW+NiN3U95PJSJSUlStKvSqVibm6OyclJzp8/T09PD4FA4KZKTqvVkp6eTkFBAVVVVdhsNtRqtZLPJ+8sN3qCcjgcVFZW8sADD5CdnY3RaCQcDrOwsKDkn+bk5OByuRgbG1NWwImEHDQhB2/V19crAWmhUIhAIMDs7Cx9fX1cuHBBMWWaTCZF/qKiIrZs2UJnZ2dcKszc3Fx27drFgQMHlB3zyt2mfL/KecMej4fZ2VmOHTvG0NAQY2Njcb0DuxkWiwW73Y7T6VQq4LjdboaHh+np6VF2YPHK4uIik5OTDA0N0d/fT1ZWluKauxFyYYLdu3djtVopKChQFnfrPbese+GCcDjM3NwcBw8e5OjRo3z9619ny5YtZGVl8fDDD7Nv3z6ee+45GhoaGBoaIi8vj4985CPU1dVRXl6OXq/H7/crZrA333xzQ6uivF/y8/PZunUrGo2GhYUFZmdnaWxs5NixYxtaFu1mmEwmnn76aaqqqhTfazgc5s0336ShoYGDBw+uSVlKkoTdbufTn/40dXV17N27F4PBgM/n42c/+xmNjY2Kr2Gjqaur4zOf+QxZWVloNBrcbrfiA/n4xz/Oxz/+ccLhMOfPn+eb3/wmXV1dDA8Pb/Swb4nMzEzy8vL43Oc+R3l5+aoScS0tLfT09PD8888zOjpKT08PR44cobCwkG984xsUFxcDS76i+++/n3fffTfucmavh2zdWXmfyT9fuHCBCxcu0NbWhs/ni4t78XZ49tlnqa+vJyMjQ9mZdXZ2cvz4cXp6epiYmEgIGV9//XWampr467/+a6UJx1pdWdnZ2fzlX/4lr7zyCs8//zwzMzPr6sZbd4UJSzekHNBz9uxZ/H4/DzzwAHq9HpvNRk1NDWazmampKSVKT77Ibreb0dFR3n33XRobGxkaGkpIM5hcc1QOiJqammJiYiLuUihWkpycTHp6Onl5eUoB/MXFRfx+vxKc4/P5brq7UqlUZGdnU1JSQnV1NXl5eZjNZjweDxMTE/T398fVwytHgspReCt9PampqSQnJ1NVVYXT6eT+++9X6sX29/fHtbVgJXJfWjm61+/3Mzc3h8fj4ezZs3R2dtLT08PMzAxzc3PKdzE+Po7NZlOiZw0GQ9znEq8M+FipLOX/9Xo9mZmZbNmyZdXzOT4+npDWAzmorqCggOLiYqUaFSzlq/b29sZlsfXrIacYHj9+nOHhYbZu3YrBYLhmep7cPtFgMJCamoper1fal+3atYtz586taxrNHVGY8OseZc899xxVVVWUl5eTnp5OWloaTzzxxDXfE41G6e7u5vLly3zzm9/E5/MlzIR0LeQH1+v10tHRQUdHB93d3Rs9rOuSn5+v9AqUQ+5lJXfixAlaW1tv+tDJvrK9e/eyY8cOPvKRjyiRen19fXR2dsadD+zkyZO0t7fjdruVzu3yziQrK4u8vDz+7u/+jvz8fL72ta9x6NAhTp8+zXe/+11GR0cTYiKSzVNDQ0PK7qOnp4fLly/z4x//mK6urlWKYmpqCp/PR2NjoxJlulmwWq3s2LGDmpoaFhYWeP7552lububnP/95QjZ3kEv+VVdXs23btlVtu/r6+jh37lxcNja4HnNzc8zNzfH1r3+dtLQ0Hn74YRwOx3uaScvIWRl1dXUkJyeTlZXFfffdR2ZmJn/2Z3+WGAoTlhSG3InkW9/6Ftu2bWP37t0UFhYqCbUybrebmZkZfvSjHymJ4PG6E1sr8ipvfn6egYGBDa0feiPk6N1t27axc+dOLBaL8tA1Nzdz7tw5+vv7bxhlJ1c0Ki4uJj8/n4997GOUlpaiVqvxeDxMTk7y85//nMuXL+NyueJqITQzM8P8/DwLCwuEQqFVOxKPx8PAwAA//OEPqa2t5ZlnnlH6R54/fx69Xk9vb2/c7Javh1yA4N///d8V/57X61UsH1crfbVarZSvNJlMwNJupaWlJW7v41tFrVaj1+vZt28fxcXF2O12Ll68yPHjx+M6/eJqbDYbW7duxWq1KqkWw8PDnDx5kqamJrxeb8LtmmX8fj8XL17EaDSuSgNaidx+LhQKUVRUpOyy5SwFi8Wybib3O6owYakKw+joKL/4xS/weDzY7XYyMjLeozDn5uYYHx/n8OHDdHZ2Jky+5c2QJIlgMMjw8HDc+i5VKhU6nY6ysjJqamowmUyo1WoikQhdXV2cPHmSsbGxG45fDt4qKipi+/bt7N27l+zsbKLRKLOzs/T29nLixAnOnz+P3+/fUAVztU/L5/NdN2XE7/cTDAY5dOgQgUCAffv2kZOTQ2FhIUVFRbjdbvr6+uJCYarVaqVxNSwFUcgmWLkpwhtvvLGmz5LTMZKSkhRftsfjobu7O253YHKg1vz8vJIGda2gH9kKIn9f1dXVFBUVkZqaSiQS4cKFCwkTLSu3LisuLiY5OVkpZzg5OcmhQ4fo6upKqN3l1QSDwZsGfMpzT21tLQAFBQVotVqlVrnFYsHv96/LouGOK0xY8uc5HA4yMzPJyMi4pi06HA4rq/t4mHzWA1mWoaEhfvzjH8dtHl95eTkf+tCHePDBB6murkan0+Hz+RgcHOTcuXO88847N1X2lZWV/O7v/i5VVVUUFBTgcDjw+/10d3fz+uuvc/DgQYaHh9cUMHQnkVtzBQIBQqHQmhdmKpWKmZkZTpw4wf79+yktLWVmZiauctv279+v1IGNRqN861vfUqLMbxW5XVl5eTk5OTnA0sJidHSUWCyG0WiMu3Zsv/zlLzl58iRJSUk37H9YXl7OM888w/bt2ykuLiYWi2EymaiqqmJhYQG9Xs9PfvKTuI/MNxgMSnGRZ555BqfTqeS3j42N0dDQwNTU1EYP844j98v89re/TXV1NSUlJYqZes+ePUiSFF8NpG+ErOlLSkrIycnBZrMp5r5wOKwUydXpdJjNZnJycvD7/UpLlkRY5V0PeScjh0vHK+np6UpJw+TkZAAl0Gd8fPy6Jrjk5GQMBgNpaWlUV1dTW1tLfn4+6enpeDweXC4XDQ0NtLS00NXVdceSiW+FrKwsrFarklM5OTmJRqNBrVYrpqtoNKqYqY1GIyaTifLycqUvq9/vZ2JiQgmQiRccDge1tbWKwqyoqFAaPt/qs5SZmUlZWZmS4ybXcw4EAsoOLZ7M6rBkWp+ZmbnpeYFAgMLCQmX3IVc4MpvNZGdnU1NTw9tvv83Q0FDcLQpWIm9EsrKyyM7OVlKEurq66O3tVdL3EhWVSoVarcZgMLC4uHjDaNdoNMro6ChWq1XJmzYYDDidTqUN4XpwRxWmJEmkp6dTUVHBH/7hH5KVlaWECEejUdxut1IoOCsri/T0dL761a/S1NTEP//zP+PxeBLanACsah4dr1RVVfGFL3xh1c6/v7+f7373uzdsNLt9+3a2bNnCb//2b5OZmUlOTg6RSAS/38/hw4dpbm7mueeew+/3x40/+qmnnmLPnj3k5eUxODjIm2++id1uJykpiWPHjin+VY1Gg16vV2oiP/3001gsFsxmMxcvXuT06dO0tLQwPDwcNxOqzWajsrISg8FANBrl2Wef5cSJE1y+fBmv13tLK+y6ujo++9nPkpqaqjTxDQQC+P1+dDodSUlJcZMWdKt0dXXxjW98g/7+fu6//34eeOAB7HY7ycnJFBYWkpeXx/Hjx/F4PLS3t8dVa7aVJCUlUVdXp1Rqkmv//s//+T/p6upSrAGJSlJSEsnJyZSWljI8PHzTgMnFxUV8Ph+9vb1KJPS2bdvQ6XR897vfXZcx3TGFqdVqMRqNPProo1RWViq1ZSVJ4sqVK4yNjdHS0oLVaqW2tpacnBzS0tIoKipCo9HwxBNP0NjYyPnz5xOquLXD4SA3N1cpS5UIJmY5+EGlUhGJRJQyUysDlVJTU7FYLBQVFZGSkkJqairV1dVkZ2eTm5uL2WxW3jsxMcHx48dpb29nbm4uriYcOZ0iNTUVnU7Hgw8+qHTGSUlJIRAIEI1GlcR9+b6NRqMMDg7S09NDc3MzXV1dSl3keEFekctWm4yMDAoKCqipqaGtrY2hoaEbvs9kMpGamkpZWZkSKS2nKGi1WqxWK3l5eUxPT2+4H/pGyFaS67lAYrEYoVCItrY2wuEwMzMz5OXlcf/99ys+2z179qDRaOjv748rK4KMwWDAarVSXl6uRLT7/X5mZ2cZGhpieno6bq/PWlCr1ezcuZOcnBx27NhBe3s7SUlJ9Pb23jDoLBQKMTY2pvTbTUlJISMjg+zsbIDbrv5zxxSmnBfzuc99jm3btikKJBqNcubMGc6fP8+rr75Kfn4+n/70pzlw4AB2u53S0lKys7OxWq385Cc/4fLly0rwQiKQk5PDgQMHSEtLS5im0fKECUtm8v7+fqWjTCgUQqVSKZPvU089RWFhIWVlZdjtdkwmEyqVSuk519vbS2trK6+99tp1J+iNZHp6mvHxcZKSknA6nVRWVirH6uvrAVbVzZXz9C5evEhTUxMHDx6kv7+f8fHxuz72tSCbXeVrVlpayt69e/F4PNe9HlqtVmnWvnXrVj7xiU9QV1dHZmamco5eryc9PZ3y8nIuXbq04Q2kr4ckSaSmpgLcdAfc2NhIU1MT58+fp6KigrKyMjIzMzGZTDz00EOUlpZy8ODBuFSYZrMZu91OdXW1ohw8Hg9TU1OMjIwwPT29wSO8PdRqNQcOHGD79u0cOHCAM2fOoNfrlQYe12NhYYHBwUGKioqApYX+4uIixcXFSvnW2+GOKcyMjAyKi4ux2WxKrcrBwUHa29v55S9/SUNDA7Ozs0oeVDQaRavVUlRUhF6vp6SkhMrKSmpra69ZezZekVfoK/u1xbtJdiVyx5iqqiruvfdeHA4HdrudXbt2kZmZSXFxsdLnUzb9yaWsZPNfV1dX3AYbnDhxgs7OTsLhME6nk8LCQqU4gWx6lFexMzMzjIyMMDMzQ0dHh1LbOF6jRAcHBzly5Ai7d+9WctYcDgePPPIImZmZHDhw4Jrvs9lsJCUlUVxcjNVqVfy88Otdy+nTp5XKW/HQQPp6yHnP8s83Q6VSKSX/5I4tsLRoMhqNSnu3tfhG7xZqtZp77rmHmpoaSkpKlFKOJ0+e5OLFiwlZ6OVqJElS0kV0Op1SvKC9vZ2ZmZlrpomYzWaysrJ45JFHFIUpWzrNZvOa6tLejDuiMCVJIi0tjfz8/FURa3NzcwwPDzM8PMzY2JgSpejz+ejo6KCkpITMzEylLVZmZiaFhYUMDQ0ljMLU6XSKuU8mkRSmSqVSfMpVVVXk5eXhdDqpq6sjNTV1lal5cXGR+fl5pbfpmTNnaG1tZWBgYIOluD6jo6PMzs5y7tw5cnNz8fv9OBwORUHMzc1x5coVpqenmZycpLu7m+npaQYHB+MiaOlGyM2s5d2/XIy7sLAQlUql7ESuJj09XQnMk59Vubm0vGO5dOkSTU1NcR85KgdqyZW2IpHIe1wCsrldo9Gg0+koLi4mLy+P5ORkpUuSHIhoNBrXZaJdT1QqFbm5uRQWFir37cLCAt3d3bS2tsaVC+R20Ov16PV61Go1NpsNvV6vxB5MT08rQWgGg0GZd+Ua3mlpacCvrWdarXZdAn/WXWHKN9q+fft4+umnlYEDSjJpcXExwWCQjo4OwuEw4XBY2XX+zd/8DVu3biUrK4uKigo+//nP09fXF9er2quRc9+ubjScCMhqlk2AAAAgAElEQVQ9EGtraykvL0etVivXVM7vk3Pdmpub6enp4R//8R+ZmppidnY2boJ7bkQwGOTNN99clbcoyyb7t+SefHKptERI/G5paWF8fJyCggKSkpLIzMxEp9Nhs9lISUm5rgyy33PlhDI4OMjg4CDf//736e7uprm5OSFa62m1Wr761a9SXFxMUlIS4+Pj9Pf3rzpHbllWWFhIVlaW0u5rZX1diN+FriRJFBUVKQuhyclJ+vr6OHr0KKdPn07oyNjrISvFL3/5y3zqU59idHSUtrY2XnrpJT784Q+zY8cOnE4naWlp5ObmrltU7NWs+6fKCmJqaoqenh4KCgrQ6XRotVqSkpLIzs5WTEZ5eXlKFRi5Z9vY2BhpaWmKL+F6eZuJQCwWw+12x6UPZCU+n4+hoSHS0tKUSUOn063aJctKQzbBdnR00NTURH9/PyMjI/h8voQpNhGLxTaF2epqgsGgksqj0WiUIBY5oGtlybRrIZukZ2ZmlHSg9vZ2RkdHE6o4uRxoJpvbZZ+mjNlsJiMjQ+nqcS25fD4fLpeLubm5uLtX5MWNrBTcbjc9PT3Mzs5uGmUZi8WYmJhgYmKCSCSiVBLLyMjAbDaTkpKC2WwmEAiwfft2SktLSUtLUyyaK3udyoU74rJwQTQaZWFhgePHj9PX16eUKUpNTSUzM5PMzEx27NihtNjp7e3l0KFDNDY2Mjg4SFtbGyqViqqqKuXGTjSFKV/ccDhMb28vY2NjGz2kGzI0NMTRo0c5cOAAubm51zxncXFRqaRy4sQJvvvd79LZ2RlXvp0POouLiywuLvKDH/yAo0ePkpeXR15e3qrgnZu9f3x8nNOnT/OjH/2It99+e1WpwERAbvgQCoV46qmnKC0tBa7tz7xWYXaZgYEBWltbGRkZiXt30PDwMMePH99Uz2IkEqGhoQFYKsghm9kdDgcOh4OioiLq6ur4xCc+ccPAStkkL89ft8sdC/pxuVwsLi7yT//0TxQVFfHoo49it9ux2+2YzWZ0Oh12u11xytbX1+Pz+ZRAIbVarZgaEikXU1Y+VqsVh8PBiy++SEtLy0YP64a0tbXxwx/+kMnJScrKyqivr1f8Pz09PUo/PY/Hw/j4OD09PfT39yfUdfkg4fP5GB4e5jvf+Y7S1is7O1vp+6nT6ZTnMxgMMjAwgMvlUlpAyf1n491ney3koB+Xy8Xo6Khijr66HOL1mJ+fx+fz8Ytf/IJz587FbYAXLKVQTE5O0tXVxenTpzeVwoxGozQ2NuL1esnOzqaiooIdO3asKv24ls+4cOECnZ2dirvidrljClOuz/nSSy+Rn5+P0+mkoKCAaDRKRkYGJpNJqaKSnZ39Hn+BXNigt7c3oSbmqakpLl68SEVFBR6Ph1/96ldxm4IgMzQ0xNDQEBqNhomJCYqKijAajYRCIZqammhsbOTixYtMTk7S09OTcJPoBw25kPyrr76KzWajuLiYqqoqSktLycrKwmQyMTo6quSkXrhwgf7+fs6fP680yU5k5MheuWep2WxWKjddyy8pLwyi0Sher5fx8XHeeecdTpw4cbeHvmZkX7vL5WJoaOiGBUYSkVgsRn9/P16vl4yMDKLRqFIvdy1NpeX4g9bWVs6fP79uGy/pRpOfJEnrMjPq9XocDgcWi0VJfk5PT6euro709HSKioqw2+1KBGYgEKC9vZ0jR47w05/+lO7u7veV9xWLxW7qtV8vGWXkyLu0tDS0Wi2jo6OEQqE7FjSynjJaLBaMRiNpaWmoVCqlLmUgECAQCCgm2bvNRlzHu82dkFEu+afX6zEajej1eiwWC2q1moWFBSWwye/3K31PI5HIHQvculvX0WAwYDKZlFqiDz30kJLbXVBQoPj+QqGQ4j6anJxUoqKbm5vp7u7G7Xbf8t++GzJqtVq+8pWvUFJSQkFBAa+88grf/va3b+cjb4m7+Tyq1WosFgs2m4309HR+7/d+j507d1JSUoJWq1UsB1frsfHxcQYGBvibv/kbzpw5w8zMzC0FX15PxrtSfH1hYYGhoSHlwZ2dncVms6FSqXA4HMzOzr5HYXZ0dHDlyhX6+/vj2ixyNXLUbyKNWcbr9eL1ete1f5xg45DvxYWFhU3TkmstBINBxdxsNptJTU1lfHyc1NRUXC6XojBln6/cn3VycpL+/n7a29vjOnhGbuggL3bi3YJ1O0QiEWZnZ/H5fIyPjyt+TY/Hc0OFKTeq7+/vX9ciDndlh3mNz1VynVaaSq6ObJK31e8XsTNZQsgY/wgZl1hvGVeWC7zWPANLk/LKqly3kwZ2t2RUq9VKneq7nfa0kfeqVqtV5L5R2o98HcPh8Pu6nhu6w7yaRMxPFAgEiYe88N5sJEpu8Hqz0Xneaws3EggEAoHgA45QmAKBQCAQrAGhMAUCgUAgWAM3DPoRCAQCgUCwhNhhCgQCgUCwBoTCFAgEAoFgDQiFKRAIBALBGhAKUyAQCASCNSAUpkAgEAgEa0AoTIFAIBAI1oBQmAKBQCAQrAGhMAUCgUAgWANCYQoEAoFAsAaEwhQIBAKBYA0IhSkQCAQCwRoQClMgEAgEgjUgFKZAIBAIBGtAKEyBQCAQCNaAUJgCgUAgEKwBoTAFAoFAIFgDQmEKBAKBQLAGhMIUCAQCgWANCIUpEAgEAsEaEApTIBAIBII1IBSmQCAQCARrQChMgUAgEAjWgFCYAoFAIBCsAaEwBQKBQCBYA0JhCgQCgUCwBoTCFAgEAoFgDQiFKRAIBALBGhAKUyAQCASCNSAUpkAgEAgEa0AoTIFAIBAI1oBQmAKBQCAQrAGhMAUCgUAgWANCYQoEAoFAsAaEwhQIBAKBYA0IhSkQCAQCwRoQClMgEAgEgjUgFKZAIBAIBGtAKEyBQCAQCNaAUJgCgUAgEKwBoTAFAoFAIFgDQmEKBAKBQLAGhMIUCAQCgWANCIUpEAgEAsEaEApTIBAIBII1IBSmQCAQCARrQChMgUAgEAjWgFCYAoFAIBCsAaEwBQKBQCBYA0JhCgQCgUCwBu6awpQkSS9JUqskSVm38J5qSZLevZPjWk8+CDICSJKULklSuyRJxlt4zxOSJP3kTo5rPdns13KzyyfzQZDzgyAjgCRJ70iStP0Wzs+QJKlNkiT9eo1hXRWmJElfliTpgiRJC5IkPX/V4S8BJ2Kx2NiK8+slSXpbkiSfJEkeSZJelSSpQj4ei8WaALckSU+s5zhvhw+CjACSJB2TJCkoSdLc8r+OFYf/b+D5WCw2v+L8xyVJOidJkl+SJJckSQclScqRj8disVeBSkmSqu+iGDdEkqQXJEkakyTJK0lSpyRJX1xxOOGv5QflXgWQJOkzy5OjX5KkHkmS9i0fSng5lxXif0iSNLA85suSJH10xSkJLyOAJEkFkiS9LknSrCRJ45Ik/YskSZrlY08Avlgs1rDi/ApJkn6xLJ9PkqSjkiTVy8djsdgEcJSl72d9iMVi6/YPeBr4GPAcSxPqymNXgL0rfr8HmAO+CiQDNuCvgFmgaMV5vwm8tp7jFDKuSc5jwBev8boemAZyVrz2CcALfBYwApnA94B+IHXFef8v8C8bLduK8VQC+uWftwDjwM7Nci0/QPfqQ8AAsIelTUA2kL1Z5ATMwF8ABcvyPQ74gILNIuPyeF4HngcMy3NIM/CV5WO/BH5zxbnFy/L89bJ8ycBXluW+Z8V5e4GWdRvjHRL8r1Y+oEAeMA9oVrx2EvjWNd77BvCDFb9nL79Xv9EX9IMkI9dXmPuB7hW/S8uT1R9fdZ4KaAH+csVre4G+jZbtOvKWA2PApzbhtdzs9+q7wBeu8fqmkvOq8TYBz2wmGYE24NEVv/8v4NuAbnmcKxfpPwRev8ZnPMfSblv+XQMEgPz1GOPd8mFuA3pjsVgYQJIkE1AP/PQa5/4nSytGAGKx2AgQYmlCi2c2o4x/K0nS9LLv4MDya9uAlebZcpYe2lVyxmKxKPASK+Rk6YEokCTJcueGfGtIkvQtSZICQDtLCvN1Nue1XMmmkU+SJDWwC0iXJKlbkqThZVOekU0k50okScoAyljaWW4mGf8R+IwkSSZJkrKBjwJvAqVANBaLDa849yGuL+Pe5evP8vfSDdSsxwDvlsK0smRCkLEt/+2xa5w7Btives23/BnxzGaT8X8ARSytQr8DvCpJUjHvlVOWYy1yyu+LGzljsdjvsWTO2Qe8DCyw+a7l1Wwm+TIALUtugX1ALbAd+BqbS04AJEnSAgeB78disXY2l4wnWHKTeIFh4ALwCu+VEZbkuJ6MKpa+B5l1k/FuKcxZliallb9HgWtFdWWx5CNbSTLgvjNDWzc2lYyxWOxsLBbzxWKxhVgs9n3gHeBR3iunLMda5JTfFzdyAsRisUgsFjsF5AD/J5vsWl6DzSSfHHj2v2Ox2FgsFpsGvsm179VElhNJklQsmSIXgS8vv7wpZFyW7U2WFq1mlhRiKvD3vFdGWJLjejJGl98js24y3i2F2QQUyhFPsVjMD5wGPnmNcz8FHJF/Wd6a61htBoxHNruMMZb8lU0smYNkOlhaDa6Sc/kBeIYVcgJbgf5YLOa9s0N932hYCibY7Ndy08gXi8VmWbr/YitfXv5/08gpSZIE/AdLO+pnYrFYaPnQZpHRxpJr51+WF+ku4P9jaeHTzdJXkL3i/MNcX8bTsVgswNKbNEAJ0Lguo1xnp62GpQinv2VpJWRg2RnN0oWtX3HuvYCfpcimZJZWE3/F0kqgdMV5n+Uazt0NdEx/EGS0Ao/IsrEUTednSVHqgCmWoxCXz/80v46SlSPcvgcMAmkrzvt/uEYwwgbJ6AA+AyQB6mV5/cCTm+VafhDu1eUx/SVwfvmaprIU9PL1zSQn8G/AGSDpGsc2i4y9LKWsaZbnoJ8BLy4f+wXw2RXnli7LszJK9v9alntlxHA90LpuY1xngf+CpdXdyn9/sXzs94Hnrjr/XpaiMedYmnB/CVRddc4v5UksHv59QGRMX56AfMs35RngoRXH/xfwP656z1PL7/EDM8CPgNyrzmkGajZavhUyHl+Wz7s8tv9jxfGEv5YfhHt1eUxa4FvL13Ic+GfAsFnkBPKXr11weczyv9/cLDIuj6d2ecyzLJlc/xPIWD72GPDGVedXAa8tyze3/N57rzrnX1lOTVmXMd7FL0MPtAJZt/Ceapa21xt+MYWMq8aczlJUqfEW3vME8J8bPXZxLT8Y8n2Q5PwgyLg85neA7bdwvoOlyHzDeo1BWv5ggUAgEAgEN0AUXxcIBAKBYA0IhSkQCAQCwRoQClMgEAgEgjUgFKZAIBAIBGtAc6ODkiQldERQLBaTbnaOkDH+ETIuIWSMf4SMS2xWGcUOUyAQCASCNXDDHaZAIBAkMgUFBWzdupXS0lJ0Oh2vvvoqLpeL6emrS6oKBDdHKEyBQLApUalUZGdnU19fz3333YfJZKKpqYlIJLKpFKZGo0Gn0wFLhWjC4TDRaJRIJLLBI9t8CIUpEAg2HSaTiS1btrB//36eeuop9Ho9fr+f4uJiFhcX6e7u3ugh3jaSJGEwGPjUpz7FV77yFQB8Ph8vvPAC3d3dXLhwgYWFBUKh0E0+SbBW7orCVKvVJCcnY7VaSUtLo7+/H4/HQzgcvht/XiB436hUKjQaDenp6RgMBsxmMyqVCrVafc3zXS4Xfr8ft9tNJBIhGo3e5RGvP2q1mqSkJKxWK3a7nfn5eYLBIAMDA3G5i1GpVCQlJbFt2zacTieSJDE4OMjU1BQ+n49gMLjRQ1wXJElCpVKRkZFBbW0tAHNzc7S2tmI2m5mbm2NsbIyZmRmCwSCiqtvtc1cUZnJyMgcOHOCxxx7js5/9LF/60pd46623mJmZ2RQTimDzYjKZSEtL44tf/CKlpaXU1dVhsVgwmUzA0qQlE4vFOHjwIJcuXeLVV1/F7Xbj9/s3aujrRlJSEnv27OGpp57iv/23/0ZrayudnZ18+ctfZmZmZqOH9x4MBgNFRUX8+Z//OYODgxw+fJgXXniB9vZ2wuFwXCr522GlIjSbzfzO7/wOLpeLjo4OXn75ZY4dO0ZPTw/z8/M3+BTBWrgrClOSJNRqNVqtFqPRSGZmJjk5OXg8nk2tMPV6Pffccw92u52cnBxCoRDBYJBf/vKXTE9Pb/odtkqlwmQyYTabSU5OxmazkZycTFlZGdFoFLfbzdDQEJOTk8zPz7OwsBB3vqXMzEwKCwvZsWMHOTk5pKWlYTAYFJ8RLN3f8qRVU1OD2Wymra2N4eFhenp6Nmro64bBYKCiooKcnBwMBgPRaJTFxcW43LGoVCpKS0spKytDo9EwMTHBuXPnmJiYYH5+Pi7H/H6JxWJEo1GCwSCzs7OYzWa0Wi0ajQaLxUJ+fj733XcfdrudF198kdHRUebm5jZ62NdErVaj0WjQarXvMSNrNBrlWFlZGenp6RQXFxONRgkEAvT39zM+Pk44HGZhYYGJiYk7Zt25az7MaDSq3Kw5OTmUlJTQ0dGxqe3rJpOJZ555hm3btnHfffcRCASYmZnhypUreL3eTa8w1Wo1VqsVp9NJQUEB5eXl5Obm8pnPfIbFxUW6uro4evQoDQ0NTE5OMjs7G3dWh8LCQmpqarjnnntISUlRXl858a78edeuXZSUlHDmzBkkSdoUCtNsNisLhlgsxtzcHB6PJy6Vj0qlYvv27VRVVRGJRBgYGODo0aPMzs7G1X21XkQiEebm5picnMTpdKLVaoGluaegoACn08n+/fu5cOEC8/Pz+P3+uLtukiSh0+kwGAwkJyczOzur6AWVSoVOp8NoNJKUlMQDDzxAbW0tTz75JLFYjLGxMd58803OnTtHIBDA7Xbj8XhYWFhgcXFx3ce6IUE/99xzD1arlePHj7O4uLjpTCSwdKG1Wi1Wq5Xk5GRisRg6nQ6LxUJlZSWhUIiGhoZN+RDLGAwGSktL2bdvH4899hjJyckYjUb0er2yWkxNTeXAgQNcvHiR9vZ2Wltb1+07MRgMALfls9q+fTv33nsver1eeU1+GCORiOKfX4nRaOTpp5/G4XDQ1NSE3+9nYWHhfY9ho5AkCYfDQV5eHkVFRaSlpQFw6tQpTpw4EZcmPo1GwwMPPEB+fj6nTp2ipaVl1QS8mZAjYjs6Oviv//ovdu7cidPpJDs7G51Oh16vR5IkjEYjpaWl+P1+RkZG4kph6nQ6UlJS+M3f/E3y8vKoqKigq6uL/v5+Tp48STgcpqqqiq1bt7Jjxw6cTicpKSno9XpisRhZWVk89thj1NfXE41GmZ6e5o033lD87HNzc/j9fpqbm/H7/be9w94QhZmbm0soFEKv16NSqTatwtRoNKSkpGAymYjFYquUqMVi2egh3hCVSrVKSVgsFjQajfKQhkIh5ubmCIVCqFQqRTa1Wo1arcZoNGKz2SguLqaiooKdO3cq/r5YLIZarSY1NRWr1UooFMLr9eLz+VCp1q+Wxu0qTEmSlGA1WUEuLi7i8/kIBAKEQiG0Wi1ZWVno9Xrl72m1WsrLyxkdHcVkMrG4uJiQChNQgn3S0tIwGo2EQiGGhobo7u6OSwuJSqUiLy8Ph8PBqVOnmJycTNjvfi1Eo1EmJydpbGzEaDTi9XoJhUIYjUZSUlJISUlBp9ORnZ3N+Pg4KpUqrhbp8iairq6OsrIytm/frij93t5eAoEATqeT4uJiampqMBqNqFQq5ZpqtVry8vIoLCwEUHJsA4EACwsLeDwe3G43Xq+X6elpQqEQoVDofX8HG6IwMzIyCIfD6HS6dZ0g4wm9Xo/FYqG6upqsrCwAwuEw8/PzjI6OMjY2FlcrvatxOBzU1dUBSzf1H//xH5Obm0ssFqO3t5crV67wrW99i5aWFtLT00lLS6OoqIjc3FycTidPPvkk6enpiiKRJGlVgIz8sxzpd/Xx20WlUlFSUkIsFnvfO/lYLMa7776Ly+ViYmICj8fD5cuX6e/vZ2xsjGAwiNVq5cEHH+See+7hoYceUv52eno66enpWK1WAoHAusl1N5EkibS0NDIzM8nOziYYDNLb28vMzAyBQCDu7l+dTofZbEan0xGNRhkYGIjLoKT1pr+/n5GREQ4fPqz4AtPT06murua3fuu3ePDBB3n88cdxOp289tprBAKBuNmkZGZmKlao9PR0AMrKysjNzeXkyZO0trby4x//mF/96lf84Ac/4OGHHyY/P59YLIbBYMBms5Gbm4vD4QAgNTWVj3/848q9GY1GCYfDDA8P09DQwPe+9z16enoYGxt7X+PdEIUp777Wc4KMR2T7u0az9DVHIhFCoZCy4ok3kpKSMBqNZGdnU1BQQH19PbBk5srPz8fhcBCLxQiFQszPz1NTU4PJZCIjIwO73U5ZWRkZGRmkpaWRm5tLSkrKql1lV1cXbrdbuZnlYJlIJEJHRweDg4PrtvqNxWLKd3w7E/vQ0BCRSIRYLIbf76erq4vx8XFmZ2fRaDTo9XqMRuOqIKBoNKqsbOfm5uJyJ3Yz1Go1er2eoqIi8vPzCYVCDAwMcOnSJWWxEG8K02KxkJGRgVarJRQKMTo6isfj2ehh3XHC4bCyGJdZXFzEYDAwMjLCzMwMFouF9PR0LBYL0Wg0bqK3o9EooVAIv9+vWLSSkpIwGAyo1Wqi0Sizs7OKedVqtTIwMAAsLZDkGAnZXaDX63E4HNjtdkUBy3OKx+OhsrISt9udWArzg4qsaIaGhhgdHd3o4byH7OxscnNz+Y3f+A1KS0vZu3cvsVhMMSfLP1utVsrLy3n22Wfxer2KwiwtLVV2ivK5MtFolJdffpmLFy++5+9GIhGmpqZwu93rqjA7Oztv+3NaWlpoaWnhV7/61arXJUkiPz8fu92uBMTIyH6ljo4OhoeH42Y1fyvo9XpSUlJ4+OGHKS0tZWpqiiNHjvDcc88xOTkZl7vmnJwcKisrMZlMeDweGhsbPxA7zGvh9Xq5cuUKV65coby8nOrqajIzM8nNzQWIG4U5NzfH9PQ07e3tSlBdaWkpdrsdWJo35CCz8fHxVc+0vKgzm82KsrXb7TzxxBPU19crFh+1Wo3NZmPLli18/OMfZ2xsjJaWlvc13g1TmFqtltraWoxG4/sefDwjh957PB6SkpIwm80bPaSbkpSUpJhynE7nKn/H3NycstNSqVTKDbiwsIDZbFZMYPPz88zPz3Pp0iVcLhfj4+PKZ1y8eJGpqalr/m05rSSe/CvXIzs7m4yMDO6//35KSkooKytbFUG7uLjIxYsX6ezsXBUdnkgUFRWxdetWtmzZQlJSEm+88QYXLlxgZmbmjkQfrgfFxcXU19djMBiYnp5mbm4ubsd6p9FoNCQnJ5OZmUleXp4SabplyxYWFxcZGRnZ6CECS5WJhoeH+c53vkNVVRVPPvmkYvL/yEc+wu7du9m7dy9NTU2cPn0at9ut+C/lOVZOL4GleIU33niDiYkJpqenqa+vx+l0KgFQt8tdUZhyvpA8eUiShFarVSK3NqPClGX2+/0Eg8GEUJhyoEBRURGpqanAr82mcqRnOBzGarVitVoxm82KidblcjE4OIjb7cbtdvPzn/+cwcFBWltbE1JhXI28c9ZoNGRlZbF161YeeOABCgoKyMnJUR7GcDhMIBCgra2Nvr6+hFgArETOmc7Pz2f79u0UFBTg9/s5e/YsnZ2dcelKkHE6nVRWVqLRaFhYWGB+fn5N0bGyD31l9aaVPjD5WU4kNBoNJpMJm82Gw+FQ8hizs7MVk2Y8IC+w33jjDXw+H4899hgzMzNIksTOnTvRaDTU1tby2muv0dfXRzAYVHKA5QDElS4Pv9/P9PS0EuhXUFCAzWZT/Nq3E/ADd0lh+v1+Ll68SHl5OVNTU1itVlQqFXa7fdXKfDMh+2ktFktCKEtAuQlX7opkH6Mc6n3o0CEeeughPve5zxGNRvF6vfzLv/wLAwMDdHR04Ha78fl8ykowUZWlHIyk1+vRaDSYzWby8/N56KGH2L17NxUVFcqDKC8qZLNzY2Mjhw4dSihzoEqlIjU1FbvdztatW3n66afZv38/Npstbsx3N8NgMJCUlERHRwdtbW1rmhhVKhVlZWVkZWWxe/duJQgxGAwyPz9Pe3s74+PjdHR0JNS9rNVqSU1NxWQyKfEigUCAS5cuMTQ0tNHDuyaXL1/mS1/6EnV1dVRWViqLc7PZTF1dHTk5Obz44ou0tLTQ399/w9iArKws9uzZQ1dXF0NDQ+zZs4fOzk4OHjxIe3v7+x7jXVGYkUhEmUhls5u8y5QDYjYjckLuZpBxfn6emZkZJSp269atxGIxPB4Ply5dYnh4mMHBQXw+X1zm560VuQ6p2WwmJSUFq9WK0WjEbDZTUFDAjh072Lp1KwUFBcDSgkIOX5+enqapqYnGxkYltD1RkBVmXl4eO3fupKSkhMzMTBYWFvB6vczNzcVteoY8l8hVpTo6OhgfH7+hgpMkSUnvqq2tJTs7m6qqKsUNIacOydHuY2NjzM/PJ4SJV6fTkZqaSnl5OampqUiShN/vZ3Z2lomJibit9jM3N0dnZydWq1WpCpeRkUFVVRUWiwWj0ci2bdsUvXF1XqWc1jU/P4/BYCAcDivV5MxmM93d3XR3d9+WleSuzOSy41Y2kSRiEMStIpt4EklhymbHq9M85J+DwSDt7e1cuXKFf/u3fwNQTLLy7jSRVuHXIikpibq6Onbs2MG+ffuoqanBbrcr34FarV6VCrW4uMjQ0BBHjx7lZz/7GY2NjUxNTSXcPa7T6di+fTv19fX87u/+rnLPtra20tTURFNTE7Ozsxs8ymuj0+mU9IL8/HxeeOEFLl++fN0dpmxaP3DgAPfeey+f/OQnsVqtuFwuFhYWlBKN4XCYz372szQ2NhIIBGhvb2d4ePguS3drqNVqMjIy2Lt3L3/+53+O3W4nGo3S1NREQ0MDHR0dcb+gvXz5MleuXOH111+nrB4KumsAACAASURBVKyMv/qrvyI5ORmTycTv//7vI0kSp06doq+vj3feeQdYkru0tJRoNEpbWxtqtZq3335bia/43ve+x+zsLNPT0/FvkpWZmpqioaGB+vp6LBYLdrs97hP43y9ZWVnk5+cnjLIEVik92dcsK0C73U5xcTEPP/wwnZ2dt2XWiGfkiLrs7GxKS0tJTU1VVq7Xwu1288Ybb3Dp0iV6enrw+XwJpyxhSens2bOHqqoqdDqd4os+fPgwra2tSrmxeERezMj/AoHADfNEHQ4HO3bs4N5772XHjh10dXXh8/lobm5Wqjj5/X5UKhWTk5NEIhEeeughFhcXmZycVBaI8YbJZCIlJYXHH3+c7du3K3WPI5EIzc3NNDc337YP724g9/OMRqMMDw/z2muvkZqaSmpqKg8//DBOp5OysjLS0tJIS0tT5iqbzYbL5eLSpUtK4QKNRkM0GmVqaor5+fnblv2uzuYjIyOcPHmSLVu2KJOSw+FYNTFvFoqKiqiqqkoohQms8mGu9GU6nU6MRiOf+9zneO211+jo6Fj1ns2CVqvFbreTm5tLWVnZTc93uVy88MILjIyMvO/crnjAaDTy4IMPKmkHU1NT9Pb28tJLL9HV1ZUwfkyAQCBw3V2USqVS6hnX1dVRWFjIP/zDP9DY2Mjrr7/OwsKCsuAxGAx0dXVRV1fH7/zO79DX18fly5cJh8Nxec9bLBZyc3P5/Oc/T25uLlarFVj6Ps6ePUtDQ0NC5ATLylIuOPD973+ftLQ0srKy2LZtG4WFhVRUVBCLxbj33nuV983Pz9PW1sbs7CyTk5NMTU0p3WnWa5FwV2fzcDhMMBgkEomg1WqprKzE5XKxdetWxsbG4tbk834oKyujtrZWKYacCLS1tTE9Pc3Bgwepqqriwx/+MPDrIupy15Hi4mKefvppYCnf64UXXmBsbIyRkRElKviDQnp6Ol/84hc5dOgQL7300kYP55aQi11//vOfZ9euXeTk5ChtyyYmJmhra8Ptdsf99dTr9aSnp2M0GolGowwODjI4OPgepWY2m/nEJz5BZWUl27dv5/z58/zwhz/krbfeYnx8/D1pTaFQiObmZtLT09FoNEp91kAgEDe7NNkXm5aWxrPPPktNTQ1FRUVK39bJyUnGx8cZHBxkcnIyLhX9jYhEIng8Hubn55menubkyZNEIhF27969qnRnNBqlu7ubK1eu0N/fj9/vV1JO1lPmu6ow5TqcsuM8LS0Nh8NBVlYWXq930yhMSZKw2+04nc5VO0y5/Uy83rRyo9mLFy+iVqv50Ic+pESJ6vV6JUk4OztbaVg7OztLU1OTYlqfnp5WbvB4mVRuBbldksfjWZUzKkc9G41GNBqN4sc0mUzU1tbS1ta2UUN+32i1WsxmM7t37+bAgQMkJycjSRKRSISZmRlGR0eZn5+PexOz3OBaXpz6fD7m5uZWPWeSJGEwGKitraWwsBCNRkNPTw/Hjx+ns7PzmoEwkUiE6elp3G43sGS2losixAOyLzYtLY3y8nLF956SkqIEL01OTtLX18fs7Gxcdiq5GbFYbJXic7vdzMzMvEcOuYuOfO3v1Dx7VxVmb28vw8PDPPHEExQXF2O1WrHZbOzevRuPxxP3DvVbIT8/ny1btqxSmP39/Vy5ciWuzVvz8/O8/PLLdHV1odFoqK+vp6qqSjm+sqxdNBolOTmZP/mTP8HlcikBQV1dXbzyyiu4XK6NEuN94/V6OXz4MA0NDfzkJz8Bfl2EvaSkhEcffZSioiIyMzOBpd1NXl4eNpttI4f9vtDpdCQlJSlF8OXALpfLxcWLFzl06FBc513KyFGhN/I1y2Xz6uvrGR8f50//9E9paWmhr6/vhrmacpTw8PAwKSkp1NTU4Ha74yJaVqfTkZOTwyc/+Un+4A/+QFk0SJLEwsICPp+PF198kZdffpnBwcG4txTciJKSEnbt2sVTTz1FbW3tqt0l/Lp2tN/vx2q1Kov29eaum2QjkciqrucWi4WqqqprlkxLZOTE6ZWrHL/fz8zMTFz7EWKxGIFAgLGxMc6ePYvf76enp0fpuuJ0OrFYLMoEq1KpsFqtqNVqQqGQ0oWkt7eXgYEB+vr6EmpVK5uAZGsILClMk8lEKBQiLy9PmXzlKOKVO85Ewm63s23btlU1f91uNw0NDfT19Sk+oERgZRUXg8GAXq9ftWtMT08nLy+P1NRUxsbG6OnpweVyrUnxyXOWnOoQD9faYDDw/7P35sFxXeeB7+/23o3uxtYAGvsOECAAAtxJcbMsmaJobbYl23LsaJKqN5Nx4jeT58Su915NebZKpfLq+U0mVuKZjB05tiwvUmwpjlZTFFeR4gKSIPZ9b6zdDfSC3u77A7jXaBCUIAlEN6Dzq0KJ6Htv63y4957vO9/5FofDwX333UddXZ1aS1XB4/Fw8+ZN+vr6GB8fT9qG3++H3W7HZDKRlpZGQ0MD+/btw+l0YjAYcLlcav/LlJQUtV+m3W4nOzubWCy2+RWmwvKkeKfTyaOPPsqZM2cSMZR7xsTEBMPDw+Tm5qoVRDweD2NjY5uiN9/g4CDPPfccsOjy2r59O2VlZTz55JPU1NTQ2NgYl3pis9morq5m27ZthMNhTCYTV69e5dlnn900ky4sTo5erxev14vL5Yo7Nj4+jt/vJysri+rq6k0X0LWS2tpannnmGTXQB2BgYIC///u/59atW5smiEmp+KLkdyttrZa77urq6ti5c6eaZtHe3v6h38NkUZjKlk9dXR3f/va31SLjyxkYGOAHP/gBV69eTRoX8odBo9FQUlKC0+lk586dajcgvV5PKBTi2rVr9Pf3c+XKFcrKyigqKuKzn/0sWVlZbN++HUmSmJiYWPdxJeSNX231tdU6lyhtkZaX25qZmWFwcDAp3Dkfhmg0ysjICPPz8/j9fvbs2cPs7Cx1dXVqjuLyHEytVsvu3btJTU2lvb2d3t5euru7EyzFx0NxvX7uc5+jsrISrVa7aZ9ZjUajBnBZLBa0Wi2RSESt1qRUbNosKF4Ql8uFz+ejoaFBrW2seLIUA1aWZdW9Pjo6+oHVmJQ+ryaTSV21JHpvXqvVsmvXLhobG3E4HGqgllKK87XXXqO5ufl9azcnMw6HA4fDwcmTJykoKCA3N5eCggKi0Sg3b95kaGiIF198EZfLxdjYGN3d3WRmZmI0GrHZbBw+fJiSkhKqqqo4f/78uja6SIjCDAaDap7UZp10Pgi73U5GRkacNbqZVpgrmZ6eZnp6mt7eXubn50lJScHhcGA2m+/odylJEtXV1aSkpLBr1y7C4fCmVpgajQabzUZxcTFHjhwhJydHva9KpZ/NtIpWgmRSUlKwWq1otVqi0Sjj4+MMDw8zMjKStDmXq6H0mJ2dnSUQCFBZWUkgEIhrTj8zM6NGiVosFvLz8/F6vR+oMPV6vRrwFo1GE64wlUCf2tpatm/fTmpqqmqUKxWn3nrrLW7fvr1p3zmHw0FFRQWHDx+moKBALRwfDAa5ffs2N2/eVPfXw+EwRqMRi8VCaWkpFRUV7Nixg4KCAkpKSujq6sLlcq1b4NqGK0xZlrl06RLRaJQ//MM/xGw2b/QQEobL5aKjo2NTTUarcePGDfr7+/nRj35Ebm4u3/nOdygqKqKoqCjOALJYLNTU1NDf35+4wX5MjEYjmZmZfPvb36ampobi4mI1GjMcDjM4OMizzz7LlStXEjzStaOUGKutraWiogKTyYQsyxQWFlJSUkJJSQljY2ObphZuNBrF7/czOzvL7OwsTU1NGI1G9Hq9mjMZCATiikoo1azeD7PZzMmTJ9m1axfp6elMTk6qSfGJIj8/n4KCAo4dO8a2bdviDPLf/OY3XL9+nddff31TZhwo7bq+/OUv8/TTT+NwONRaza2trVy6dImf/exntLe3x7UCDIVCRCIRfvazn1FRUUFRUREFBQXs27eP8fFxcnNzefvtt9cl6CkhK8zR0VH6+/vVvS6NRqN2ylDaSG1FQqFQUuVwfVR8Ph8+n4/R0VFcLhc3b94kHA6rbhFFoSid35WG1BvdR1Gj0aDVarFYLGrz7rVWOlE615eUlFBUVERDQwNFRUWqcolGowwNDdHV1cWtW7eSsr/p3dBqterqcrk7z2AwqAEzy7cSNgNKo+GRkRHq6urIzs6mqKiIiYkJNV1qbm6O6elpIpEIJSUlzM3NxZV1hEUjaGFhAb1ej91up76+HqfTyfDwMBMTE3g8noTOT5mZmZSUlJCTk6MG3ilMT08zNjampo+kpqZumvq38Lvmz8XFxZSXl6vFUwKBgDrPDAwM3LE3qbyPs7OzeL1eteKR0kjA7/dz7ty5zaswW1tbCQQC+Hw+tflnVVUVBw4c4MKFC5silF2wiNfr5a/+6q/Yt28fFotFfZllWcZkMlFTU4PVaqW+vp7vfve7NDc3b9jYzGazGmE3OzvL+Pj4mpW2zWbD4XDwp3/6p+zbt4/q6mrVEFBaeH3/+9/nxo0bnDlzZlO5ZFdDcS0Hg8G4ajebiatXrxIKhaiurqa8vJx/+2//Lb/97W/59a9/rZa0e+2116ioqOBb3/oW7777Lm1tbXE5e6Ojo/T19ZGVlYXT6eQP/uAP6Ozs5Dvf+Q5tbW1q5HSiqK+v58SJE+Tl5a3aBUmr1ZKfn49OpyM/P19N5N8MOJ1OTp48SUlJiXo/AoEAra2tnD59mueff/6uka9KsFd+fj779u3DbDYjyzKPPfYYu3bt4rnnnluXe5cQhan0LGtvbycUClFWVkZGRgZFRUWbyrW1VpRG0subMG8VYrEYbrebtrY2XnjhBR544AF2795NRkYGWq1Wbd9msVgoLi5mdHT0YxdAXguSJGG1WsnPz+fRRx9lfn6eyclJLl26xMjICMPDw2ojAI1GoxYm0Ov1pKWlUVtby969e6mrq8PpdKLT6dT72N3dTX9/P83NzfT19SVtqbS7kZKSwv79+9WOK7AYV/Dmm2/S3NzM9PT0pszZGx8fR6vV0tXVRX5+Pjt27GB+fp7p6Wl6enpYWFjg7NmzDA0N4fF41MIMSkyB8rxWV1er6VNvvfUWnZ2d9PT0JEW0qZKatppBU1dXh81mo6ysDL1eT0ZGBk1NTWq0t7ISGx8fZ2BggKGhoTsKPCQSZX92uZtZiVqfn59Xq8SthiJbOBzG7/erlZmUBhgOh4NAIPCx72FCFKYyeV29epVoNEp5eTnZ2dmUl5ffkZC6FQgEAgwPD6ub1MnygK4Hsizj9XppaWmhpaUFrVZLdnY2NptNbZGVmppKQUEBVVVVDA8PMzs7uyFuaaUZ9u///u8TDodxu93YbDaam5txu92qAaMoSqWNV0VFBcePH+erX/0qNptNXVkGAgEGBwd58803uXDhApcuXUr4iuOjkJaWxkMPPaQWX4DFifiFF16gp6eH8fHxBI7uo6O4TZubm5EkicOHD6vu5V//+td0d3fz+uuvk5WVRV9fn5q/p+QT63Q60tPTqaqqori4GIDvfve7DA4O0tvbm2DpFpmdnWVoaGhVN+u+ffvYs2cP4XAYjUaDyWSKK1UZi8VYWFjg0qVLvP766/j9/qSr/rOy45HSosvn831gsKRSetXj8aDX6zEYDMDi1lBeXh5+v39zKkz43Z6DMuFUVVWRkpLCyy+/rPZe3CoEAgGGhoZwu91bTmGu5NatW+rKTslZU16CPXv2oNVq6ezs3PBIYaWk3VNPPcUDDzzAiRMn8Hg8TExMkJGRgd1uJzc3V+2pqCh9JdcyFArR0dHB3/7t39LW1kZ/f3/St0laiVarpampicbGRpxOJ1arFYD29na10e5mDBZZTiQS4cc//jEXLlxgdnYWp9PJ448/Tm1trapMQ6GQuppRalpnZmZSWlqK1+tlamqKc+fO0dvbmxRu2OV0d3cTDAZ55JFHMJvNcRWmlNZzSsqTEhuiKA5YXLHdd999VFZWEgwG0Wq19Pf3J4ULfmJigjfffJNdu3Z9pOuVEp42m02VeWxsjP7+/nUzBBOqMGdmZvB4PITDYbVaTGZmJjabjdnZ2U2pWJQXcTnhcBiv13tHcefNhNFoXDVhW5blOPed0p1+bm5OjXJTcDqdFBUVbVhASSQSIRKJEAqF0Ol0mEwmKioqiEajOBwOvF4vo6OjZGVlkZaWRlFR0R0eDqXKy/T0NAMDA7z33nu4XK5Np1iUCMTKykqqqqqwWCzo9XpkWVb37bxe76YzAlai9EN0u93U1taya9cuqqurqampoaioCFisNevxeNR958zMTHJycsjNzVVzjru6umhtbVWDhJIFZV4cGxtT50pFUa5M7YJFJbLyfVMaMytG0/L0m0Ti9/vp7+9ndnZWDbxSVsqpqalkZWWpVbhWolQYy8jIUAMPldqzk5OTuN3udQk4TJjCDAQC/PM//zNer5eDBw+Sk5OD3W5n+/bthEIhhoeHk+ImfliUPp9paWlbxr2sFGJfrXdpMBjk7NmzaqrM2NgYkUiElpYWZFlm27ZtqzalvtcoisBut3Pr1i011FyRp7i4mFgsxrZt21SLfLXKPSMjI4yOjvLcc8/R0dFBd3f3pnwu09PTycnJ4Rvf+Aa1tbWYzWYikQiBQIALFy6owXbJpBw+KtFolLGxMb73ve+Rn59PaWkpe/bsoaCggOrqalJTU3E6nfT29qoR+9euXaOzs5OBgQG6u7tVF2Cy/T38fj/hcJif/exntLS08OSTT5KVlYXD4VjzdyjR+sreYLIY8Up6UEdHBxcvXlSLn9x///1UV1dz4sQJ/uN//I9cunQp7jqlJvKf/dmfsWPHDmw2m1q0vaWlRfUqrAcJU5jK3tfo6CgtLS3o9Xp1w9rtdm/aggaKH10p1bUV0Gg0NDU1kZ+fH1d8XclvCwaD6ipTsfIyMzPjcmxX7k1sBOFwmNnZWd5++222b99OLBZTK6NotVq0Wu0d7deUF03pntPc3MzAwACtra2Mjo4m3QS6VpxOJ7W1tWRlZamu2IWFBWZmZhgeHmZ4eHjTyrYa0WiUubk5xsfH1ZSZsbExpqensVqtOBwOxsbGmJycZHR0FI/Ho9bPnZ6eTlrvllIGUAk2U9K2iouL1Q4l09PTajUnJagpOzsbi8VCRkaGaiit1tUl0ciyTE9PD+fPn1d7YGZmZqoGQWNjI+FwmOnpaXV+zcvLw+l0qsFaynafy+WipaWF27dvr9sWUEIVpt/vp7e3lxdeeAGr1Up5eTmHDx/GYrHwgx/8YFO+wD6fD7/fz/z8/Kas6LMaer2er371qzQ2NqrVmZRuJYFAgKamJtWVl5aWRmpqKrW1tWrYu6IsN1ppRqNRBgYG+A//4T/wwAMP8KUvfYkHHnhAzT282zVut5vbt2/z7rvv8sorr9DV1YXH49nUBtDOnTt5+OGH1abCsFhovaenh1u3bnH79u0Eju7e4Xa7cbvddHZ2Jnoo60YsFuPatWvcuHGDd955h4qKCpqamjAYDIRCIc6ePYvJZKK0tJSRkRFmZmY4fvw4paWlHDx4EEmSiEQiah51snHq1CkuXbqkts47cuQIdrsdu93OV77yFQ4ePMi5c+dYWFhAkiSOHj3Kzp07KS8vx2QyqavUM2fO8OKLL9Le3r5uY0t49ei5uTk6OjrUIB+9Xr+pmi5/EohEIrzyyisMDw9z4sQJdDqd6mY1Go3U1dWp6TImk0lttKso1uVKNlGeg87OTn7xi1/g9/vZtm0b+/fvx2QyodVqVau9paWFsbExLl++TH9/P11dXfT392+JYhOTk5N0dXVx+PBhQqEQExMTXLx4kVdffZWRkZFED0/wEYjFYvj9fgYHB/H7/WqAncvlQqvVqi7XhYUF3nnnHZqbm3n33XdVY7enpyfRIqyKkuf8m9/8hp6eHmw2G7m5ueTl5VFaWorD4SA/P1/dGikoKCArKwuj0aiWBrx69SpnzpxZ94jvhCtM5YbPzMyo/vnNuEe0EiXYRHHNrnfn740kGo1y7tw5fD4fn/rUp9RyVUpEXmlpKcBdawMrCkn5SQTDw8O4XC5sNhtut5uqqirsdjt6vZ5oNEooFOL27du0t7fzyiuvMDk5mZTW90dlenqawcFBvF4vOp2OwcFBmpubeeONN7ZURPonCWX7YGJiYtXOHFNTU+q/kyGHdK0sT3+ZmJhg7969wGKLNiU4q6qq6o7rlGo/Fy5c4OrVq3fsda4H0vtN4pIkbcgMr9FoyM/PJzMzk3A4jM/nY2Bg4GMrGFmWP3A5c69kfPLJJ6mvr+fQoUNMTk5y+vRpzp49S0tLy7r+fzZCRkmSKCoqIicnh/vuu08Nkjly5IjaFUCn093hroXFe9vW1sZ7773HxYsX6e3t5cyZMx+qnu56yahUA1H2cpZ3HInFYni9XjWPa6MDPu71fbRYLFgsFnJyctBqtWqhbiUKdCOMuUS+jxuFkHGR9ZBRiZAtLCxk586dHD16lNraWnJycigpKYmL/lW8YC0tLTz33HPMzs5+LCPhbjImfIUJi5PV0NAQQ0NDiR7KutHX10csFsNiseB2u+nv70+qfK4PgyzL6tivX7+uri6tViszMzPMzc2pChN+FxCk/Lu9vZ1r167R0dHB+Ph4wjwISpi52+3eVLVf1wO/34/f749bdQgEyYwSI9Hf34/ZbMZmsxEMBsnOzmZqaiouqj0SiaiRzhMTE/esfm5SrDDvFYm09pQ0CmWP7F65ZTdSxpV5XkpJuQ/am4zFYkSjUVX+D/s3EFb7IkLG5EfIuMh6y6jMNcvnm+VzjlIaT5lrPi5JvcLciiguya2wH6uwUtltJdkEAkHyoiw4Es2dpVsEAoFAIBDcgVCYAoFAIBCsAaEwBQKBQCBYA+8b9CMQCAQCgWARscIUCAQCgWANCIUpEAgEAsEaEApTIBAIBII1IBSmQCAQCARrQChMgUAgEAjWgFCYAoFAIBCsAaEwBQKBQCBYA0JhCgQCgUCwBoTCFAgEAoFgDQiFKRAIBALBGhAKUyAQCASCNSAUpkAgEAgEa0AoTIFAIBAI1oBQmAKBQCAQrAGhMAUCgUAgWANCYQoEAoFAsAaEwhQIBAKBYA0IhSkQCAQCwRoQClMgEAgEgjUgFKZAIBAIBGtAKEyBQCAQCNaAUJgCgUAgEKwBoTAFAoFAIFgDQmEKBAKBQLAGhMIUCAQCgWANCIUpEAgEAsEaEApTIBAIBII1IBSmQCAQCARrQChMgUAgEAjWgFCYAoFAIBCsAaEwBQKBQCBYA0JhCgQCgUCwBoTCFAgEAoFgDQiFKRAIBALBGhAKUyAQCASCNSAUpkAgEAgEa0AoTIFAIBAI1oBQmAKBQCAQrAGhMAUCgUAgWANCYQoEAoFAsAaEwhQIBAKBYA0IhSkQCAQCwRoQClMgEAgEgjUgFKZAIBAIBGtAKEyBQCAQCNaAUJgCgUAgEKwBoTAFAoFAIFgDQmEKBAKBQLAGhMIUCAQCgWANCIUpEAgEAsEa2FCFKUmSUZKkVkmScj/ENQ2SJF24l+NaT4SMd71m08i41eUDkCQpS5KkdkmSzB/imkckSfrZvRzXevMJuZdbXkYASZLOS5LU9CHOz5EkqU2SJON6jWFdFebSjftfkiQNSJI0J0lSsyRJJ5ad8r8BZ2RZHlt2zUFJkk4tne+RJOkVSZJqleOyLN8E3JIkPbKeY/2oSJI0v+InKknSf192yqaXEUCSpJqlMXskSeqWJOmJZYc3vYySJP2xJElXJElakCTpH1Yc3vTyLUeSpEpJkoKSJP142cffBv5BluXAsvM+K0nSZUmSfJIkTUuS9BNJkgqU47IsvwJslySpYQOHvyYkSfrS0uTokySpR5Kkw0uHNv29/CTMqwCSJJVIkvQvkiTNSpI0LknS30iSpFs69ggwJ8vy9WXn10qS9PKSfHOSJL0tSdJB5bgsyy7gbRb/PuuDLMvr9gOkAN8BSlhUxp8F5oCSpeO3gfuWnX8AmAf+d8AGZAD/BZgFypad9xXgn9dzrOskr3Vp/EeWfbbpZQR0QCfwp4AWuB/wAVVbSMbPAY8Df8ui4lh+bNPLt0KeN4CzwI+XfjcCU0DBsnO+AHiBpwEz4AR+APQD6cvO+7+Av0m0TCvkexAYAPYvzTv5QP5WuZd8QuZV4F+AfwBMS8/fLeAbS8d+A3xl2bnlS/L81yX5bMA3luQ+sOy8+4CWdRvjBvwRbgKfB4qAAKBbduws8Owq17wK/GjZ7/lL1xoTfVNXjPP3gV5AWvp9S8gI1C09eNKyz94A/vNWkXHZmP4LyxTmFpTvS8DPlyZcRWEeAbqXnSOxqHD+fMW1GqAF+E/LPrsP6Eu0XCvGeQH4w1U+31L3csV4t9y8CrQBDy/7/a+A7wOGpXEuN/D+EfiXVb7jb1lcbSu/6wA/ULweY7yne5iSJOUAVSxaQPVAryzLkaVjFuAg8ItVLv05i1YjALIsjwBhoPpejvcj8PssPoDy0u9bUUYFiUVFupVlhC0knyRJduA/segpWE490LHs92oWJ944GWVZjgEvskxGFie1kqXvTjiSJGmB3UDW0tbB8JIrz8wWupfL2cLz6v8HfEmSJIskSfnACeA1oBKIybI8vOzcB7m7jPct3X+W/i7dwI71GOA9U5iSJOmBnwDPybLcDqSx6EZQyFj6/4+tcvkY4Fjx2dzSdyQFkiQVA0eB55Z9vFVk7AAmgD+TJEkvSdJnWJTVwtaR8W5sJfn+M/C/Vkw0cKeMigxrkVG5LllkzAH0LLqUDwONQBPwf7O17iWw5efVM8B2FrcGhoErwK+4U0ZYlONuMmpY/DsorJuM90RhSpKkYXHJHAL+eOnjWRb9zCz7PQasFtmVy+Iey3JsgHt9R/qx+CpwTpblvmWfbQkZZVkOs7i/dxIYB/4PFi23YbaIjO/DlpBPkqRG4AHgu6scXimjIsNaZFSuS7iMSyhBS/9d3sYBtwAAIABJREFUluUxWZangP8XeJgtci8VtvK8uiTba8BLLO7ZOoB04C+5U0ZYlONuMsaWrlFYNxnXXWFKkiQB/4tFy+/zS5MvLPrcS5WoJ1mWfcBF4MlVvuYp4LfLvjOfRT92xyrnJoqvEb+6hC0koyzLN2VZPirLcqYsy8eBMuAyW0jGu7BV5DvGYpDIoCRJ48A3gc9LknSNRRmrlp3bwaIxFCfj0iT2eZbJCNQA/bIse+/ZyD8EsizPsjh2efnHS//dKvfykzCvZrC4LfA3siwvyLI8DfyQRcOnm8U/Qf6y89/i7jJelGXZz+JFOqACuLEuo7wHG7d/B7wLWFc5dhM4uOz3QyxGX36DRSsgncUgDDdQuey8p1llgzdRPyzuEfgA2xaWsYHFaDULi5NtH0vBAVtBRhaDAUzAX7BotZtYCpzYIvJZWIw0VH7+H+CXQBaLk+QkS5GkS+d/kd9FySpRij8ABoHMZef9n6wSUJJgWf8T8B6QvXRvzgL/eavcy6XxfBLm1V4W0510LLpQ/wl4funYy8DTy86tXJJneZTsnyzJvTxi+CDQum5jXGeBi1m07oIsRlkqP19ZOv514G9XXHMIOL10npfF8OG6Fef8Bng00Td02Xi+D/zjXY5tFRn/ikW3xjyL0XUVW0lGFqNG5RU/39kq8t1F3h+vuL/fWnHOYywqHh8wA/wUKFxxzi1gR6LlWTEmPfDs0gQ6Dvw1YNoq9/ITNK82Lo15lkWX68+BnKVjJ4FXV5xfB/zzknzzS9ceWnHO91hKTVmXMW7wH8QItAK5H+KaBhaX2Am/oULGT4aMW12+pfFmAe2A+UNc8wjw80SPXdzLT56MS2M+DzR9iPOzWYzqNq3XGJT8QYFAIBAIBO+DKL4uEAgEAsEaEApTIBAIBII1IBSmQCAQCARrQPd+ByVJ2tQbnLIsSx90jpAx+REyLiJkTH6EjItsVRnFClMgEAgEgjUgFKZAIBAIBGvgfV2yAoFAsJlJSUkhLS2Nz3/+8xQXF9PW1sbw8DDXrl1jbm6OQCDwwV8iECwhFKZAINiSSJKEzWajoKCAL3zhC+zevZu33nqL5uZmhoeHiUQiQmEKPhRCYQoEgi2HTqcjPT2dBx54gKeffpqqqioMBgOHDh2ivLyckpIS/vEf/5E333wz0UMVbCI2XGFKkkRaWhpmsxmHw4FOp0OSJKLRKJFIhOHhYQKBAAsLCxs9NIHgQ6PVaklNTcVut5OZmQlAOBymu7ubhYUFotFogkf4ycRoNFJcXExlZSXbtm0jJSUFSZJITU0lEolQWFiI1WpN9DDvCRqNBqPRSGpqKtFoFFmWcTqd6HQ6QqEQOp0OvV4PQDQaZWxsDL/fz9zcypaTgpVsqMLUaDTo9XqOHTtGQ0MDzzzzDKmpqZhMJubn55mdneVb3/oWLS0t9PT0IMr2CZKd1NRUPv3pT/PQQw/x1FNPodVqcblcPP744wwODjI7O/vBXyJYVzQaDU6nkz/6oz+ivr6eoqIiFrtjbX0kSSIlJYXS0lIefPBB5ubmCIfDfPOb38ThcNDX10dmZiZ5eXkAeDwe/vIv/5KbN2/y9ttvJ3j0yc+GKsz09HRyc3PZv38/O3bsID09HYvFgl6vR5IktFotDz74ILm5ufh8Pubm5pifn9/IIQoEayYjI4OSkhIeeOABtm/fjsViQZIkjEYjkiRtmUlap9ORmZlJSUkJ1dXVnDp1iuHh4UQP666kpqaSk5NDbW0tubm5d9yHWCxGOBwmFoslaIT3llgshsFgwOl0Ul1dTWpqKtnZ2VitVvLz87FarZjNZgAikQhlZWVMTa3sK51c6PV6TCYT27dvx+l0UlBQgFarveO8+fl5WlpaGBsbY3BwcN3HsaEKMysri6amJj796U/T2NgILFpEyg3W6XR8/vOfZ/v27Vy/fp3h4eGkVJgrX8BP2kr4/RTBJ+VvodFoyM7Oprq6mi984QtYLBZV9q32N1Dcm8ePH+fLX/4yQ0NDSaswJUkiMzOTwsJCduzYgdFovOOcWCy2pd3l0WhUVZi7d++mqqpqsdOGJKkrS+UZ1el0VFdXMz4+nsghfyAmk4n09HROnjzJnj17OHr0KEaj8Y53bXR0lL//+7/nwoULm1dh6nQ6MjIyOHDgAF/72tcoKipa9TxJkrDb7RQWFnLy5EnOnTuXVC+m3W7n61//Onl5eaqbJxKJ8MYbb9DT08OpU6e23EtoNBoxGAzo9XrS0tLYs2cPNTU17Ny5c1XFcO3aNdrb27lw4QKzs7NJafCsB1qtlk996lPs3LkTs9mMTrf4Ks3Pz+N2uwmHw0n3LEiSRFZWFllZWezfv5/u7m6uXbtGIBAgEonc9brU1FSOHz9OTU1NUhsDBoOBlJQU/uAP/oDGxkb1niwnEokwODjIiy++SHd3dwJG+fHRaDTs3r0bu90eZ7xqtVq0Wi12u52qqirq6+vp7OzknXfeobu7m5SUFE6cOEFubi4FBQXA74yHUCiUKHHeF0mS0Ol07N27l8cff5wDBw5QUFCATqdb9VnMyMjgi1/8IjU1NdTU1PDyyy/T39+/buPZEIWp1WqxWq3k5eVRX1+PwWAgEAiobpFYLIbZbMZsNmMwGLDZbJSUlNDa2roRw1szBoOBvXv3UlVVRU1NDZIkEQ6H8Xg8mM1mbty4QTAYJBKJEIlEkGUZrVZLNBolHA7f8X06nY5YLLa8f1vSoOw3Z2ZmYrfbsVgsOBwOdu3axZ49ezh27Niq47bZbKSkpDA+Ps7IyAgjIyNEIpFV5d/MaDQaSkpKKCkpUQPXAKanpxkfH0+6FYxer8doNFJUVERhYSH79u0jFovR0tJCMBh83+tsNhs1NTU4nc4NHPGHR6fTYbFY2LFjB3V1dWg08XVZYrEYXq+X8fFxWlpamJmZSdBIPx4ajYa8vDyys7Ox2+1xn2s0GsxmM7m5uciyTH9/P83Nzdy8eZP09HQaGxtJSUlRr5FlGY/Hg8/nS4Qoq6LIoRjrNpuNqqoq9u/fT1lZGampqcDvejkvNxqMRiNVVVWq5/LChQsMDQ2t27u4IQpTo9FgsViwWq3Y7XZu3rzJyMgI169fZ2ZmhpGREZ555hlOnDixEcP5WCgKXlEUWq2WRx55hIMHD7Jjxw56e3vp6Oigr68Pn89Hbm4uU1NTNDc3x32P0WgkPz+fQCCA1+slGAwm1QSbnp5OXV0dTzzxBEeOHMHhcGAwGDCZTBgMhrsq+f3797Nz506eeOIJent7efbZZ+np6aGzs5NwOJx0hsFHQa/XYzabycvLIycnR31hZVnmf/7P/8k777zD8PBwUkV619fXU1tby9e//nUKCgowGo3Mz8/z6quv3qFYFDQaDdXV1TQ1NfHQQw8RCoWSWsnYbDaysrIoLCzE6XTesXXg9/v5yU9+wpUrV7h58+b7rqqTHbPZTHl5Of/qX/0rdSUdDAbx+XxcvXqVjo4O/viP/5jh4WEmJyeJRqOUlpaysLCg7t1Go1Hm5ub4xS9+QVdXVyLFiSMjI4OMjAwOHz5MWVkZDz74IFlZWWqkrzL3RCIRFhYWMBqNatSvQl5eHmazmZdffpmRkRFcLte67FlviMLU6/Xk5eXh8/k4d+4cN27cYHR0lM7OTubn5/F4PJvCdSfLMj6fT7XIlcnfarUiSRI1NTWkpaXhcDgoLCzE5/PhdDqZmZlRUw4UlD2GYDDI/Pw8PT09zMzMMDExkdBgBJ1OR3FxMSUlJRw6dIj6+noKCwux2+3qJnsoFMLj8RAIBIhGo6olaLVaMRqNGI1G9dxDhw7hcDgwmUwMDw/j9XoJBAKbWnHa7XaysrJwOBykpaXFHZuYmGBwcDCpjANJksjPz6ehoYH8/HxsNht9fX1MT0/HTaCrYTQaVe9PssizEq1Wi9FopLKykvr6+rhnVcHn8zE1NUV7ezv9/f1J64JcC7IsMzo6it1up6urS1WYHo+Hubk5mpub6e3tZWBgALfbjd/vx2azYbVacTqd6qq0r6+P3t5ehoaGksIQUlJhGhsbqa6upqGhAafTSVFRESaTCQCv18vCwgJTU1PMzc0xNjZGXV0dpaWlcffcYDBgt9tVj5dGo9k8CtNqtbJv3z6Gh4f59re/TU9PDx6Ph0gkgtFoJD09fVMozGg0yujoKFlZWXGfK/t7u3fvVj9T9rFSUlIIBoNMTEzEXaPVaklJSSEajbKwsMBPfvITrl+/zptvvvm+LrJ7jdls5tFHH2XPnj089dRTSJJ0x0Tp9XoZHR1lZGSEYDBITk4ODoeDqqoq9RyLxUJpaSn//t//e1paWnj33Xf51a9+RXt7+7q6SBJBYWEh27dvp7KyUg2iUHC5XEm1767sATU2NvLoo4+SnZ3N5OQkv/rVr7h06dIHpr3odDq0Wi0ajSZpo36NRiPZ2dk89thjPP744+Tk5NxxzujoKD09PZw5c4axsbEEjHL9iEajnDt3jtbWVmZmZlRFMTo6yszMDB0dHXFbIJIkkZOTQ0VFBXv37lUDoV577TUuXrxIR0dHUlQ8UpTlH/3RH/Hoo4/GzTsLCwv4fD76+/uZmJjg/Pnz9PT0cP78eb797W/zla98BYvFonpLlLiLzMxMHA7HugUAbYjC9Hg8vPrqq2qupc/nW3XCTNYXUiEQCPD666/T1dXF2NgY+/btY9u2bauea7FYiMVi6PV6tFrtHS+xskcYi8WIRqM0NTWh1+t55513EqYwd+7cSXl5OQ899NCquWuzs7OcOnWK3t5e2tvbcbvdhEIh1ZJT3JMajYa6ujpyc3PZt28feXl5HDx4kIWFBUpLS/mXf/kXZmdn8Xq9CZHz47Jjxw5OnjxJWlqa+jdyu91MTk4m1V4QLE5CFRUVlJWVkZ2djd/vZ3x8nMuXL69pElme5O7z+ZiYmEgqVzOAw+Hg2LFjVFdXq8VQVtLc3My1a9eYmprC7/cnYJTrSywWY25ujuvXr6uf+f3+VffOtVot9913Hzt37oxbhVVXVxMMBnn11VcTqjCVEoYVFRU88cQTlJeXq5/Pz8/T2tqK2+1mamqKCxcuMDw8zNjYGAaDgaamJnJzc9XURL/fT2dnJ1arlbS0NNWw31R7mD6fjytXrqx6TJlgl++jJKviDIVCvPfeewwODuLxeMjMzKSsrEy1vpfLsNynbjAYMBgM6u+ryVdeXk4kEln1Zd8INBoNNTU17N69mz179mCz2dSNc1mWCYfDTE9Pc+rUKW7fvk1zczOBQIBYLKa6YZVgAp1Ox8mTJ2loaGDPnj1kZmaSkZFBIBAgIyODGzduIEkSc3NzSevmWw1ltVZVVcWBAwfUSjFK4ERvb2/STcZ2u53a2loKCgpIS0tjZGSEiYkJOjo6mJ6eft9rJUlCr9erz/LCwgLT09NJ58602+3s2LFD3TpYSSwWo7Ozk+vXr+PxeJJO4X8UZFkmGAx+YKSvcg8bGhqor6+Pm6Nyc3PxeDyr5jNuJBqNBrvdTlFREYcOHSI3NxdYjGj2er3cuHGDiYkJxsbGePXVV9Wo14qKCg4fPozD4VCfUb/fT2trK5mZmeTn5+NyuZiamlq3ba6E15LNyMhg3759ZGdnq58l6yQqyzJ+v5+xsTG1fN+5c+eoq6sjOzubmpqauwZQKCih/Ss3qRNJSkoKqampHD16lEOHDqmuDUmSmJmZYXJykpdeeon29nbefPNN/H4/fr9ffQgXFhYIh8OqstBqtdy4cQONRqPeS0mSqK2tpbS0lJKSEt577z2effZZJicn8Xg8CZP9w1BeXs7nPvc5Dh48SEZGBlqtlnA4jNvt5rXXXuPv/u7v1jWE/eNiMBgoLy/nmWeeobKykmg0yqlTp7h27Zoayft+aDQa9uzZo65MZmZmaG1tTartE41Gg8lkIjMzU93nWo7P58Pj8dDV1UVnZ+emDvT5KOTk5JCXl8f+/fvj5idZlvnlL3/JO++8k/D7aTKZeOCBB9i7dy9lZWWq5+3MmTPcunWLv/mbvyEYDKoZCfC7uJhPfepTFBUVodFocLlctLa28td//ddqtsLo6CjBYHDddErCFabVamXbtm1kZGSoQkUiEdxud1L41Vei5C3Nzs7S1dWlukGysrLw+/3vqzCVakaHDh26Q2EmMrVEr9djtVpxOBxkZ2erFqdS27ezs5Nr167R19e3qrUmyzLRaFR1e0SjUebn5/H5fHEypaSkYDab2bZtG263m4KCAgKBwKZQmHq9HofDwc6dO3E6naoLKBwOMzIyQn9/P11dXUmzetFqtWRlZZGXl0dpaSlWq5VgMEh/fz8DAwMfmPaiBNIUFBSQn59PNBrF7XYzNDSUVO+lojCzs7PV6jXLCYVCeL1ePB4PXq93y1b3uRtOp5OamhqysrLiPCJKDdmBgYGExxNotVpyc3PJzs7GZDKxsLCA3++nra2N27dvMzg4GGfoaDQaUlNT1VWk0WgkEAgwNDREX18f/f39zM/P35OtrYQrzKKiIp555pm4QJr5+XnOnj2btInF0WiUQCBAR0cHXV1dvPPOO3e4ZFdDp9Nhs9l49dVXaWhoiDuWyFxFk8mkhnIrUZ+hUIj5+Xl+/vOf88ILLzA+Pk4oFPpYE46SM5Wenk5lZSUnT57kV7/6FSMjI+slyj1Bq9WSnZ3N9u3beeKJJ+ICYGZmZnjppZd47733kkqRmEwmHn74Yfbt20dxcTEejweXy8WFCxdoaWn5wPtotVrJysri0KFDbNu2DZ/PR3t7O7/5zW8+0JW7UShlCPPz87n//vtXff+8Xi+9vb24XK5PZF3fBx98kKeffprc3FzVEA6Hw6qh6vF4Em5EaLVaioqKyM7ORpZlBgcH6e3t5Uc/+hHd3d13eAUMBgONjY00NTVRXV2Ny+Wiu7ubn/70p7S2tjI7O3vPjICEKUydTkdJSQnl5eVkZGSoZY56e3tpa2ujt7c3KUKd3w/FUlsrSsmq1VaS7e3tXLlyJSH7Q2lpaWzbtg2bzaZ+5vV6aWtrY2hoiNnZ2Q9Ve1OWZdxuN9PT00xPT5OamqrWWQXUzfne3t6kX10qk/LOnTuprq5W95hlWWZubo6JiQlaW1txuVwJHunv0Gq1WCwW6urqKCsrQ5IkhoeHaW9vx+Vy4fV6P9CbYbfbycnJwWw2o9VqCQQCqis+0SsSBSXoo7q6+o59OGUvq62tjatXr8bdHyXgzmazYbFY4q4zmUxYLBZyc3MxmUy43W5mZ2cZGBhQvUmbASXxPyUlBYvFouZ5GwwGuru7aW5uZmhoiIWFhaTYAltee1lR5vPz83cYoUra3vHjx9XiFEoMxezsLDMzM/dUnoQpTIPBwM6dO6mtrSUjIwNYtHxu3bpFc3MznZ2dSWWx30tkWebKlSu8/fbbCXkhc3JyaGxsVCtowGJE7OXLl+nv7//QSk2WZaamphgbG1ND+JdPTIpCvX79OpOTk+sjxD1CKbpx5MiROK+ALMtq0Y0bN24kzaoLFt3HKSkp7N69m5KSEgC6uro4c+YMIyMja2rjlJGREZf/pijLD8rb3CgkScJsNvOpT32KpqamO4673W5++9vf8t5773H27Nm4iGydTkdKSgoFBQVxsRMAmZmZOJ1O9u3bh8PhUL1Ir732Gi6Xa9MoTJ1OpxZZNxgMeL1eNBoNNpuN5uZmfvjDH9Ld3Z2Uc2wwGFTzvFf+vbOzsykvL+dLX/qS6pVUDAPFSN+SCtNsNvPII49QW1urfhaLxbhx4wbNzc3qpu1WQqmyv1pwQiJxuVxcuXKFQ4cOqZ85HA6OHDlCZ2cnvb29TE1NrdllrNFoqKiooLKykpSUlLgI4Wg0ytDQEL29vfT09CQ84OCDUFYwDz74YFxpuGg0yuXLl2lubmZycjKpJp5HH31UTRFSXOzKangt75QS7PPII4+QkZGBXq9XO1/s3buXW7duJdxA0Gg0WK1WTpw4oRoFy4lEIoyNjTE5OYnX6yUrKwubzUZBQQGFhYXs37+fvLy8OwqKKF0x0tLSMBgMlJWVsX//fg4fPsypU6e4evUqt2/fxufzJYXhcDcsFgvFxcVq68TOzk5kWaaiooLu7m5u3bqVNGld4XCYlpYWLBYLBw4coKSkhJSUFJ588kna2tp46623VA9XZWUljY2NWCwWAoEAo6OjzM3N4Xa7cbvd99ygSYjCTElJITMzk23btlFQUKA2kA6Hw4yNjTE+Pq7WYt0qKN0tKioq4jooKPsJoVAoYdVh5ufnGRkZYX5+nmg0ilarxWw2U1hYSHFxMcXFxcRiMfx+P5FIRA3wUfZtlxsAOp1OnWhKSkowmUxxtVaVep5ut1stXpGMKLIVFxezbds2iouL1WLXoVAIv99Pd3c3PT09SeWmhMVo3h07dpCWlqY+axqNBoPBQGpqqpompOQAr3zulET3yspKjEajqpyU9nzJUEZNSfEpLi5etVCBIqNGo1FXk1lZWVRWVlJZWcnRo0fJzs6O86qshsPhUIuVezwe/H4/g4ODhEKhpF5tKi5nWPQOzM7OqtsISk5jshCNRhkZGWF8fJxAIKAa2Tt27ECn06kNAsLhMAUFBZSWlmIwGAgGg0xPT+NyuZiYmMDr9d7zOTQhCvOpp55i7969VFRUqPl+Ho+HiYkJhoaGGB8f31LKUin59OUvf5kvfvGLcW6g27dvc/r0aS5evEhvb29CJl4lh7Cvr4/8/HzVMs3JyeGZZ57h5MmTvPPOO4yNjdHT08PIyAiDg4OkpaVht9s5evSoqjRLS0vJy8tT90SVslTwu/xTpTh9MpWPW4nZbMZut3PkyBEOHjyI2WxW91k6Oztpb2/nhRdeSNg9uxuSJKn7PMuDYB566CH27t3L5cuXmZiYoK2tjcnJSUZGRuju7sbtdsd9j9/vVw0aWZYJhUJIkoTVak1YrvCHwWQyUVVVRVVVFU8++SS7du3C6XSSmpqqFqJfa/6hwWAgIyODJ598kuPHj6u5gcna5F6n06ml4Pr7+zEajZSXl6vRpMmWR7uwsMDp06fx+Xxs376dmpoaioqK1O4kWVlZjI2NMTIywpe+9CUaGhowm8243W5cLhfPP/88b7/99l0L4qwnG/rkK1ah0+mkuLhYddWFw2G6urq4desWLpfrjnSEzY7VaqWqqoqCggK1EoksyywsLDA6OsqtW7eYmZkhFAolRO5QKMTc3JyaO6nX60lNTSU1NVVdpdTX15OXl6cWkx8fH8dqtWK1WtmxY4d6L3Nzc3E4HGRlZcW5YhVisRgTExOqxZusKM12lXD35cpneHhY3bdMJlcsLK6surq6SE9PJy8vD7vdjtVqxWazodfr1RQDjUbDyMgIOp1OTeNSUoY0Gg3BYJCpqSnVm6AYtErwS7JjMpmorKxUXayFhYWkpaWphs9KlPdRMeSsVmtc6peyl60UrvD5fPT19SXMWFJWzisDniRJwmAwqB68srIynE4nVquVQCDA9evXGRoaSsiY70YsFsPn8zE8PMyZM2cIBoMEg0H1ntXX16uGfH5+PikpKYRCIaanp7lx4wZDQ0N3GHz3ig1VmDqdDrPZTGlpKZWVlerLGgwGefnll/npT3/K6Oho0llAHxen08mjjz6qurgAtXJOW1sbr7/+OjMzMwl7+RYWFlhYWODHP/4xb731ljrZ7NixA4vFgtls5tixY3HXrFZjVvlcYbXjoVCIGzduJG3KkEJ2djZ79uyhsrKS/Px84He5slevXuXFF1/csJf0w/KLX/yCc+fOUVRURFlZmfrcGY1Gdu3aRSAQIC8vj76+PhwOBxUVFbjdbk6fPq26Gd1uN+3t7ezZswe9Xs/Q0BDNzc288cYbCZZubaSmpvLwww+v+fxIJMLMzIy6XVBdXU16evod52m1Wj7zmc/gcDg4ffp0XOeijUSn01FYWMif//mfx23xSJKk5qXW19cDiwqpr6+Pmzdv8v3vfz/h+8+rEQqFaG1tpbW1lePHj3Pw4EF+7/d+D6fTyac//em4c2OxGB6Ph7a2Nv7hH/5hQ9OFNlRhZmdnU1lZSVFREZmZmWg0Gvx+Py6Xi8nJSaanp5N2T+ujohQFyM3NjetDFwgEaGtrY2BgIGn28ubn5xkbG+OHP/whBQUF7Nixg+rqaoqLi9WWUEBcesjKyULprbe8g4mCkr/a29ubtLmXRqORwsJCDhw4wGOPPRYX6DMxMcHNmzdpb29namoqKe7ZagQCASYnJ3nhhRdoampSXeNKGySDwUBeXh5Wq5WCggKCwSALCwscOnRINdpyc3PVQJlgMEhHRwfj4+MJlmz9icViXL9+neHhYc6ePYvJZCI1NRW73Y7ZbL4jQE8JaPN6vWi12rsajvcKpTetz+fDYrGwffv2uDEqXrzlRRyi0Si//OUvuXHjBlNTUwlt7rAW2tra1M4rVVVVPPXUU5hMJnXFr8RTKPuxG7mXvGEKU5IkHA4H27dvV1vMyLJMIBBgbGyM6enpu4a7K4niybRXtBaU0HelT9/yh3hhYYH+/n5cLlfSuLgUV8hrr72Gw+Ggp6eH+++/n0gkorr2PoipqSm1nZnFYiElJUVVsErHgeHh4aQKOliOwWBQu5EcPHgwbjKamZnh4sWLaredZH0eQ6GQmlYRCAQ4dOgQGRkZat6hEsSjNHVX3i9l8lcanyufeTweBgYGtlTivyzLatWurq4uWltbeeWVV3A6nVRWVnLfffcRCoUwGo1xXhMlIConJ+cDC5WsJxqNBp1OR0ZGBpmZmYyOjmKz2dTAuuXjWy4jLCrM8+fPc/36debn55M6uhdgcHCQoaEhpqamaGpq4rOf/SxarTYuB1oJuopEIuo2kvL8ruxZvJ5siMI0GAzk5ORw7Ngx/vW//tdqqa2RkREuXbrEf/tv/42+vr67Xt/Q0IDVauXy5cubxl2r0+mw2+08/fTTNDU1sW/fvjiFqfjtkzXSzu12c+XKBSgnAAAP0ElEQVTKFbq7u9VOJKvtSa5kYWEBSZL45je/SWVlZdyx06dPc/PmTZqbm5N28lVWX5mZmZjN5rhJsb+/n//xP/5HXDBMshKNRhkfH+eNN97gxo0bpKamYrPZ1ICu5dTW1qoGjlL2sb6+nrKyMjIyMtQV5soWdZsVRcaWlhYGBwcpLCzk05/+NEeOHCE1NZWMjAyys7PjjD2FaDTK22+/zbVr1z525au1ovTa/cpXvsL27dspKiqis7MTi8Wyak3q5Q3NYdGNfOTIEdLS0njjjTfUiHxl/MsVkaJoEq1UZVlmfHycsbGxVbuv5OXl8ZnPfAabzcbg4CBTU1OkpaXh9/vVTjz3oq7zPVeYFosFu91OVVUVZWVlFBYWqlV9FEsgHA7jcDhITU3FYDDc0bmkqamJlJQU+vr6cLvdSddCaTUsFgsZGRnU19dTWVkZ10Vhfn6e6elptYpOMhKJRJibm8Pv96PVanG5XGuKKrTb7WRmZqLVatWXWbH8RkdH6e7uVpvAJhsWi4X09HTKysriokyVcPzp6WlGR0cTPMq1Ew6HmZ2dZXZ2Vq3jOzU1dYfC9Pv9OBwOLBYLoVCI2dlZnE4nZWVlRKNRtZ9rsuTtrQfK6lmj0ZCRkUFqaqqa6G+1WuPKH8ZiMbUmrdfrpaWlhc7Ozg1TKjqdjvT0dOrr66mpqVFzuddSjhMWZa2oqECWZaanp9UiFIFAQO1JrGyXzM7OMjc3l3CFKUkSKSkppKSkxEVlKzWQY7EYNpuNpqYmMjMzmZqaQqfT4Xa7GRgYwOv1otfr192wvacKU6PRsG3bNiorK3nmmWcoLS1VV1myLJOdnc3u3bv5d//u36klnIqLi+M2sWGxHJLP58Pn89Ha2sqlS5fu5bDXhbKyMmpqanjsscfikqNjsZi6yvrhD3+Y9MpfiZKcnp5eU9u1AwcOcPToURoaGsjOzo67pr29nevXryesZu77odFoqKqqorGxkX/zb/5NnIETDAa5fPkybW1tCRzhx0N5f1YrN3nlypW4fWmtVktDQwOf/exn1Rq0zc3NSRvk9GFRGk4fO3YMWZbjlONqz7jP52NoaIjTp09z9epVTp8+zczMzIY8x0qh8by8PDW3Vq/Xq96bu0X8Lken0/HII48Qi8X4kz/5E7V36+joqNpyz+Px0NHRwWuvvcbly5fVxUyi0Gq1HDt2jJ07d6oyw2KEutvtRpIkUlNTqampYdu2bUSjUa5cuYJOp6OgoID5+XlcLhczMzPrapzfM4Wp1+sxm83s3r1bbeuUkZERd4MNBgPp6enU1dWpHbKVyiJxg9Tp8Pl8eL3epFcwJpMJq9XK/v37aWpqiusCrvSwO3XqFM3NzWohgM3CWiw1k8mE3W6PK1YQDodZWFhgamoKl8uVlHt/Wq2W48eP09jYiN1ux2AwIEkSkUiEQCCwKSJ718Jqk+Dyz8xmMw6HQzVsZ2dn1VVJMhk6SvzDqVOnqKmpWbU83vuhGAbvRyQSoaenh8HBQS5evMjt27dVL9e9TCdSin9kZ2djtVopLS2lvLycmZkZTCbTqq5iWFTsoVCIQCDA3Nwco6OjZGZmkpaWpnbYiUajeL1eBgYG1BKKXq+XsbEx2tvbmZycTHh+tBL3sWfPHhoaGtBqtYyMjNDT08OVK1fUusAOh4O6ujp1b15p+XXkyBEcDgdGo5F33313XYPV7pnCVMpL3X///dTX11NRUXHH5rnSMikrK+t921stLCwQCoVwuVxJb+WmpKSo/nUl4V0hGo2ysLDASy+9xI0bNxI4ynuHojCX98JcWFjA7XbH1ZZNNgwGA1/84hdpamqKexaVHNWLFy8mRYWbe43i5bHZbMiyzOTkJBMTE6r7LlmIRqPMzc3x0ksvcejQIRobG9e18bySl3n16lWuXr3K888/z9zc3Ibk3SoVmRT3qxLpPDo6qra1Wg2Px6PWUx0cHOT8+fPU1dVRUVFBWloaGo1GXS1fu3aNhoYGUlNT6enpobOzk/PnzzMxMZHwOJGUlBSys7M5evQotbW1aDQaent7+fWvfx3XQDo3N5e9e/dSVFREbm4uhw8fpri4mIMHD1JSUkJ6ejq9vb3JrTC1Wi0mk4kvf/nLHD9+nN27d6v1LFdDmZhWU5YLCwsEg0FcLhcDAwNq5FQys3fvXr72ta+pxcyX7zFcuXIlqQNePg5Ki7CGhgYOHDgQF1H73nvv8ctf/jLpFI4SeXjkyBF27dp1h+EmyzJvvfUWN2/e5Pr160nfPWc9MBqNOJ3OuBSoZGVhYYErV64QDocpLCyksLCQ7Oxs8vLy1hSgtpy5uTmuXr3KxMQEo6OjdHR0MDk5yeDgoFqndKMMhrq6Oo4fP86uXbvUQiBKOtD73RdF8f3TP/2TWjLu0qVLpKamsnPnTnQ6HX19fUxOTjI2Nsbrr7+udvnwer1MTk4mRWyBEhGrpEHJsozX62VwcJD5+Xn1PkxMTHD+/HmuXbuGyWTi9OnTFBQU8PDDD5Odnc2JEyeYmJigpKSEN954Y13SadZdYRqNRrKysmhsbOTYsWPYbDbV9bG8QbQSjaUEFaymMP1+P/Pz8wwODqr5ismaQ6S0mSkuLubAgQNkZmbGvbSxWIzBwUHVFbvVMJvNFBUVUVBQQF5eHvC79mfDw8O8++67SWco6PV60tLSqKmp4cCBA6SkpMQ9o6FQiM7OTpqbm3G5XEn77K0nSiePzVD+TokEttlsXLt2TW1abrFYsFqtd7TuUlACDsPhMOFwmGAwyOTkJC0tLWoTYqUlmNL+aqNclJIkkZ2dTWNjIw0NDWRlZakBPitX0Mr7pew3jo+P09fXx5UrV/B4PIRCIUZGRjAYDOpWVnd3t5o+pgRYJttzrbjLtVotGo2GaDSK3+9neno6rltOIBCIW/GPj49TUFBAUVER+/bto6SkhG3btjE/P8+pU6fWZWzr/lZs27aNb3zjG+zatUt1za1EyekKBAL09fXx+uuvr/pdXq+X2dlZRkdH1a7pyUpqair3338/e/bswel0xk04oVAIj8fD5cuXeeWVV5LerfxR2L59O3/xF39BWVmZ+pnywg4ODjI4OJh0ZeRKSkp46qmneOCBB9i5c2ec+3xkZISOjg5Onz7N9evXE+6mEqxOOByms7OT733ve1itVux2O7/3e79HXV0dJ0+eXHWfcmFhgenpaZqbm2lvb+ett95ifHycoaEhtc6xknKxkXt5Srm7rKwsSktL0ev1hEIhgsFgXDF1+N0e7vDwMB6Ph5mZGd566y1aW1vxer3q86o0S1C2gJZHjSb7M600B/B4PPT19fHuu+++7yp/dnaWYDDI888/j06n49ixY+Tk5FBcXLxuBuC6K0yTyURubi42my0u2CUcDjMxMcHw8DCtra1qxYnR0VHa29tX/a7lkX3J4Cq4G4o7srGxkcLCQjXpW2F+fp6uri7Gx8eTOuH9o6B0YSksLKSoqEgtpq9YhZ2dnYyOjhIMBhMeqr4SJVpSaXANqC9oT08PFy9eVNsHJXPO5XqiBJtYLBa1b2myG3hKCpRSGOPatWvMzMwQiUTuqjC9Xi9dXV0MDg7S3d3N7Owsbrc74fd5ucKORqOqwlu+8JBlWe3QcfbsWebm5pibm1PnmJVKRZl/NwsZGRlx6YfLV/jLi9goVY2UrZXCwkIyMjKorq6msLBQDbJcz9rk98wlu1qS/vnz5/npT3/KjRs3mJiYSJpu3x8HSZLIyMigvLycz33uc3cU6oZFV8H/3979vCSzRnEA/2qlMUpJZBBEQbWoSEjoHwiC/ob+maBNq6BW7doELSMIoUVQiwhCUvtd8sjYZNMvJRVz1EnHu4jnuXbf93btvr42+p7PMlo4zOh55jnnOcfn8yEcDpvuLetX2Ww2TE5Owuv1ilFtwPs9f3l5gc/nM+02tM1m++FZzefzYIxhZ2cHq6uryGQypl+J15LL5YLX64Xb7YZhGJBlGbIsm26x8zO6rkPXdWxubgIAlpaWvvkTfY1hGMjlcqLRPU9n8W42lf8XCAQQDAaxsLDQdM8nH7/GF98Wi0VMfNJ1HcViEZqmobW1VfS6djgcmJ2dxejoKKanp2G328U2taIoNcs/1zxgqqqKjY0NzMzMwOl0QlVVqKqKra0tMMZwfn4uzjA1erAE3lc8Ho8HExMTcLvdH5LypVJJFBD4/X5RDt1M2traPqzogL+7dPC3NDP3IP1nXiiTyeDg4ADhcBjZbNZUlaH10NLSInKYVqsVfX19eH5+rmkFKvlcIpHA2dmZGNzd29srzqDHYjE8PT0hFApBVdWmfD4TiQSi0SgKhYLYpp6amkJHRweSySSy2Szu7+8hSRL6+/shSRIkScLIyAi6urrQ3t4uBqafnp7i8vKyZm/YNQ+YiUQCe3t7GBoawvDwMGRZxsXFBdbX1+tWll1PVqsVg4ODGB4eRmdn54e9ch4wo9Eorq+vTZ2D/b9+NsSXH0eIxWK4vLw07QqYHx3I5XLIZrOwWCx4eXnB8fExFEUxXTFEPfDxUDyt0N3d/cP5afJ7pVIpMMbEGLaenh4kk0kcHR0hFAqBMYZYLAZN0xrizf+rUqkUVFUVxaBtbW3weDzweDx4eHhAOp1GJBKBy+XC2NiYGCzAt24Nw0AqlUI0GkUkEqnpGLaaB8x0Oo2TkxPMz89jeXkZmqYhn89/6/iq78BH0KytrYmcbTNef0tLiyh950qlEvx+P4LBoKm/0FdXV5ibm0N7e7toF6frupj8/ifSdR0PDw8YGBiAYRgIh8NgjJn6PjYbWZbx9PSE3d1d2Gw20bKQt6rk59Kb9Z7w5iaBQADJZBKSJMEwDBSLRdze3kLTNLjdbkiShLe3N9GxKBqNIp1OQ1VVMMZwenoKRVGQyWTMGzBLpZIo1vkTlMtlUc5eucUcj8dxd3cHxhju7u4aKun+FZU9gYH3o0B8uG4sFjP1trumab+lQXMj4wVqvAPM9fU1bm9vTX0fmw0/9mHGuZX1wPO4PH3ncDhEwHx8fEShUEBfXx+cTifi8biINzc3NyJgKoqCSCRS825q5j9sZXKGYeDi4gJ2ux1vb2+iJdz29jb8fj/8fj9eX1+/+2P+Nrx9mNPpxPj4OG5ubsAYw/b2NmRZbsq36mbGGMPi4uKHakRe3k9IPWSzWWiahsXFxQ/nTyurZfnfK+eR8mNAlVuztUYB8xfxEu+zszOsrKyIgHl4eAhFUZqiEvgzhUIB+/v7Ik/Ly93j8bhpc5fk3zXaEQTSnMz6HFo++zG3WCwN/UtfLpf/s1KBrtH86Brf0TWaH13ju2a9xvqNDCeEEEIaGAVMQgghpAqfbskSQggh5B29YRJCCCFVoIBJCCGEVIECJiGEEFIFCpiEEEJIFShgEkIIIVWggEkIIYRU4S9eAMSaszIJ+wAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 576x576 with 32 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}