model_training.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "cs329s-ml-model-deployment-model-training.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyNpc09/rMn9nAU9Op2f/zuH",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/mrdbourke/cs329s-ml-deployment-tutorial/blob/main/model_training.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UTWetPM7AWfY"
      },
      "source": [
        "# Let's train some models we can deploy...\n",
        "\n",
        "Model code to go along with [CS329s machine learning model deployment tutorial](https://github.com/mrdbourke/cs329s-ml-deployment-tutorial). \n",
        "\n",
        "* **Want:** [`SavedModel`](https://www.tensorflow.org/guide/saved_model) format to upload to Google Storage (GS).\n",
        "* **Data:** Different slices of [Food101 dataset](https://www.kaggle.com/dansbecker/food-101).\n",
        "* **Model(s):** [EfficientNetB0](https://www.tensorflow.org/api_docs/python/tf/keras/applications/EfficientNetB0) backbone's with different output layers (e.g. 10 classes, 11 classes, 12 classes).\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Nq4kxIpQMpZT",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "ab7de6e3-17fb-4334-9619-8f9437c8c798"
      },
      "source": [
        "# Are we using a GPU? If not: go to Runtime -> Change runtime type -> Hardware accelerator: GPU\n",
        "!nvidia-smi"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Sun Feb 14 05:54:43 2021       \n",
            "+-----------------------------------------------------------------------------+\n",
            "| NVIDIA-SMI 460.39       Driver Version: 460.32.03    CUDA Version: 11.2     |\n",
            "|-------------------------------+----------------------+----------------------+\n",
            "| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\n",
            "| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n",
            "|                               |                      |               MIG M. |\n",
            "|===============================+======================+======================|\n",
            "|   0  Tesla T4            Off  | 00000000:00:04.0 Off |                    0 |\n",
            "| N/A   54C    P8    10W /  70W |      0MiB / 15109MiB |      0%      Default |\n",
            "|                               |                      |                  N/A |\n",
            "+-------------------------------+----------------------+----------------------+\n",
            "                                                                               \n",
            "+-----------------------------------------------------------------------------+\n",
            "| Processes:                                                                  |\n",
            "|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |\n",
            "|        ID   ID                                                   Usage      |\n",
            "|=============================================================================|\n",
            "|  No running processes found                                                 |\n",
            "+-----------------------------------------------------------------------------+\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ol3NDTVlRLSv"
      },
      "source": [
        "If the cell above doesn't output something which looks like:\n",
        "\n",
        "```\n",
        "Fri Sep  4 03:35:21 2020       \n",
        "+-----------------------------------------------------------------------------+\n",
        "| NVIDIA-SMI 450.66       Driver Version: 418.67       CUDA Version: 10.1     |\n",
        "|-------------------------------+----------------------+----------------------+\n",
        "| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\n",
        "| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n",
        "|                               |                      |               MIG M. |\n",
        "|===============================+======================+======================|\n",
        "|   0  Tesla P100-PCIE...  Off  | 00000000:00:04.0 Off |                    0 |\n",
        "| N/A   35C    P0    26W / 250W |      0MiB / 16280MiB |      0%      Default |\n",
        "|                               |                      |                 ERR! |\n",
        "+-------------------------------+----------------------+----------------------+\n",
        "                                                                               \n",
        "+-----------------------------------------------------------------------------+\n",
        "| Processes:                                                                  |\n",
        "|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |\n",
        "|        ID   ID                                                   Usage      |\n",
        "|=============================================================================|\n",
        "|  No running processes found                                                 |\n",
        "+-----------------------------------------------------------------------------+\n",
        "```\n",
        "\n",
        "Go to Runtime -> Change Runtime Type -> Hardware Accelerator and select \"GPU\", then rerun the cell above."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Bk-UIzd_SA5T"
      },
      "source": [
        "## Setup helper functions"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "r9IdsOGuLYVK"
      },
      "source": [
        "import zipfile\n",
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "from tensorflow.keras import layers\n",
        "from tensorflow.keras.layers.experimental import preprocessing\n",
        "\n",
        "# Unzip the downloaded file\n",
        "def unzip_data(filename):\n",
        "  \"\"\"\n",
        "  Utility function to unzip a zipped file.\n",
        "  \"\"\"\n",
        "  zip_ref = zipfile.ZipFile(filename, \"r\")\n",
        "  zip_ref.extractall()\n",
        "  zip_ref.close()"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "C_1Ol5LZXrb8"
      },
      "source": [
        "# Setup data inputs\n",
        "IMG_SIZE = (224, 224)\n",
        "\n",
        "def create_data_loaders(train_dir, test_dir, image_size=IMG_SIZE):\n",
        "  \"\"\"\n",
        "  Creates a training and test image BatchDataset from train_dir and test_dir.\n",
        "  \"\"\"\n",
        "  train_data = tf.keras.preprocessing.image_dataset_from_directory(train_dir,\n",
        "                                                                  label_mode=\"categorical\",\n",
        "                                                                  image_size=image_size)\n",
        "  # Note: the test data is the same as the previous experiment, we could\n",
        "  # skip creating this, but we'll leave this here to practice.\n",
        "  test_data = tf.keras.preprocessing.image_dataset_from_directory(test_dir,\n",
        "                                                                  label_mode=\"categorical\",\n",
        "                                                                  image_size=image_size)\n",
        "  \n",
        "  return train_data, test_data"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "s1Icsoq0YYRb"
      },
      "source": [
        "# Create a data augmentation stage with horizontal flipping, rotations, zooms\n",
        "data_augmentation = keras.Sequential([\n",
        "  preprocessing.RandomFlip(\"horizontal\"),\n",
        "  preprocessing.RandomRotation(0.2),\n",
        "  preprocessing.RandomZoom(0.2),\n",
        "  preprocessing.RandomHeight(0.2),\n",
        "  preprocessing.RandomWidth(0.2),\n",
        "  # preprocessing.Rescaling(1./255) # keep for ResNet50V2, remove for EfficientNetB0\n",
        "], name =\"data_augmentation\")"
      ],
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kn5qDRF8XqLs",
        "outputId": "ef83fb12-64a5-4401-f5da-b1485fb91d3a"
      },
      "source": [
        "# Setup input shape and base model, freezing the base model layers\n",
        "INPUT_SHAPE = (224, 224, 3)\n",
        "BASE_MODEL = tf.keras.applications.EfficientNetB0(include_top=False)\n",
        "\n",
        "def create_model(input_shape=INPUT_SHAPE, base_model=BASE_MODEL, num_classes=10):\n",
        "  # Fine-tune?\n",
        "  base_model.trainable = False\n",
        "\n",
        "  # Create input layer\n",
        "  inputs = layers.Input(shape=input_shape, name=\"input_layer\")\n",
        "\n",
        "  # Add in data augmentation Sequential model as a layer\n",
        "  x = data_augmentation(inputs)\n",
        "\n",
        "  # Give base_model inputs (after augmentation) and don't train it\n",
        "  x = base_model(x, training=False)\n",
        "\n",
        "  # Pool output features of base model\n",
        "  x = layers.GlobalAveragePooling2D(name=\"global_average_pooling_layer\")(x)\n",
        "\n",
        "  # Put a dense layer on as the output\n",
        "  outputs = layers.Dense(num_classes, activation=\"softmax\", name=\"output_layer\")(x)\n",
        "\n",
        "  # Make a model with inputs and outputs\n",
        "  model = keras.Model(inputs, outputs)\n",
        "\n",
        "  # Compile the model\n",
        "  model.compile(loss=\"categorical_crossentropy\",\n",
        "                optimizer=tf.keras.optimizers.Adam(),\n",
        "                metrics=[\"accuracy\"])\n",
        "  \n",
        "  return model"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://storage.googleapis.com/keras-applications/efficientnetb0_notop.h5\n",
            "16711680/16705208 [==============================] - 0s 0us/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "h5w0PI1pOl7I"
      },
      "source": [
        "# Create a function to import an image and resize it to be able to be used with our model\n",
        "def load_and_prep_image(filename, img_shape=224, scale=False):\n",
        "  \"\"\"\n",
        "  Reads in an image from filename, turns it into a tensor and reshapes into\n",
        "  (224, 224, 3).\n",
        "  \"\"\"\n",
        "  # Read in the image\n",
        "  img = tf.io.read_file(filename)\n",
        "  # Decode it into a tensor\n",
        "  img = tf.image.decode_jpeg(img)\n",
        "  # Resize the image\n",
        "  img = tf.image.resize(img, [img_shape, img_shape])\n",
        "  # Rescale the image (get all values between 0 and 1)\n",
        "  if scale:\n",
        "    return img/255.\n",
        "  else:\n",
        "    return img"
      ],
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UIwVrX6vXb4z"
      },
      "source": [
        "## Download data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qwWwP657Szfv",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "004ed027-ab11-4541-c95d-f912779a8c81"
      },
      "source": [
        "# Get data\n",
        "import zipfile\n",
        "\n",
        "# Download data (10 class subset of Food101 - https://www.kaggle.com/dansbecker/food-101)\n",
        "# Already formatted in standard image classification directory style\n",
        "!wget https://storage.googleapis.com/ztm_tf_course/food_vision/10_food_classes_all_data.zip\n",
        "\n",
        "unzip_data(\"10_food_classes_all_data.zip\")"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2021-02-14 05:54:52--  https://storage.googleapis.com/ztm_tf_course/food_vision/10_food_classes_all_data.zip\n",
            "Resolving storage.googleapis.com (storage.googleapis.com)... 74.125.20.128, 74.125.142.128, 74.125.195.128, ...\n",
            "Connecting to storage.googleapis.com (storage.googleapis.com)|74.125.20.128|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 519184959 (495M) [application/zip]\n",
            "Saving to: ‘10_food_classes_all_data.zip’\n",
            "\n",
            "10_food_classes_all 100%[===================>] 495.13M  93.4MB/s    in 5.3s    \n",
            "\n",
            "2021-02-14 05:54:58 (93.4 MB/s) - ‘10_food_classes_all_data.zip’ saved [519184959/519184959]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "agzJYtfFBl6I",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "da62030b-260b-479d-e71c-14c20b65b39b"
      },
      "source": [
        "# How many images in each folder?\n",
        "import os\n",
        "\n",
        "# Walk through 10 percent data directory and list number of files\n",
        "for dirpath, dirnames, filenames in os.walk(\"10_food_classes_all_data\"):\n",
        "  print(f\"There are {len(dirnames)} directories and {len(filenames)} images in '{dirpath}'.\")"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "There are 2 directories and 1 images in '10_food_classes_all_data'.\n",
            "There are 10 directories and 0 images in '10_food_classes_all_data/train'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/ice_cream'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/ramen'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/chicken_wings'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/pizza'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/steak'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/fried_rice'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/hamburger'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/grilled_salmon'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/sushi'.\n",
            "There are 0 directories and 750 images in '10_food_classes_all_data/train/chicken_curry'.\n",
            "There are 10 directories and 0 images in '10_food_classes_all_data/test'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/ice_cream'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/ramen'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/chicken_wings'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/pizza'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/steak'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/fried_rice'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/hamburger'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/grilled_salmon'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/sushi'.\n",
            "There are 0 directories and 250 images in '10_food_classes_all_data/test/chicken_curry'.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "34VT7QiuL7Il",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "4df6609b-3511-466e-9810-3e84a6847d6e"
      },
      "source": [
        "# Check the file in 10_food_classes_10_percent\n",
        "!ls -la 10_food_classes_all_data"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "total 28\n",
            "drwxr-xr-x  4 root root 4096 Feb 14 05:54 .\n",
            "drwxr-xr-x  1 root root 4096 Feb 14 05:54 ..\n",
            "-rw-r--r--  1 root root 8196 Feb 14 05:54 .DS_Store\n",
            "drwxr-xr-x 12 root root 4096 Feb 14 05:54 test\n",
            "drwxr-xr-x 12 root root 4096 Feb 14 05:54 train\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TwxIlyH2B8PG"
      },
      "source": [
        "## That one pesky file that always gets stuck\n",
        "#!rm 10_food_classes_10_percent/.DS_Store"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "F0r-zyagV7Qa"
      },
      "source": [
        "Notice how each of the training directories now has 75 images rather than 750 images. This is key to demonstrating how well transfer learning can perform with less labelled images.\n",
        "\n",
        "The test directories still have the same amount of images. This means we'll be training on less data but evaluating our models on the same amount of test data."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2yamhJ8xJA5x"
      },
      "source": [
        "# Create tensorboard callback (functionized because need to create a new one for each model)\n",
        "import datetime\n",
        "def create_tensorboard_callback(dir_name, experiment_name):\n",
        "  log_dir = dir_name + \"/\" + experiment_name + \"/\" + datetime.datetime.now().strftime(\"%Y%m%d-%H%M%S\")\n",
        "  tensorboard_callback = tf.keras.callbacks.TensorBoard(\n",
        "      log_dir=log_dir\n",
        "  )\n",
        "  print(f\"Saving TensorBoard log files to: {log_dir}\")\n",
        "  return tensorboard_callback"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "11TjBJQXdCyZ"
      },
      "source": [
        "Because you're likely to run multiple experiments, it's a good idea to be able to track them in some way.\n",
        "\n",
        "In our case, our function saves a model's performance logs to a directory named `[dir_name]/[experiment_name]/[current_timestamp]`, where:\n",
        "* `dir_name` is the overall logs directory\n",
        "* `experiment_name` is the particular experiment\n",
        "* `current_timestamp` is the time the experiment started based on Python's [`datetime.datetime().now()`](https://docs.python.org/3/library/datetime.html#datetime.datetime.now)\n",
        "\n",
        "> 🔑 **Note:** Depending on your use case, the above experimenting tracking naming method may work or you might require something more specific. The good news is, the TensorBoard callback makes it easy to track modelling logs as long as you specify where to track them. So you can get as creative as you like with how you name your experiments, just make sure you or your team can understand them.\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2mXArGOXXn6P"
      },
      "source": [
        "## Model 1 (10 classes)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cwrIl-rsLdz8",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "015faef0-2406-4f24-85bc-6a0bf64f8191"
      },
      "source": [
        "# Create BatchDataset\n",
        "train_data, test_data = create_data_loaders(train_dir=\"10_food_classes_all_data/train/\",\n",
        "                                            test_dir=\"10_food_classes_all_data/test/\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 7500 files belonging to 10 classes.\n",
            "Found 2500 files belonging to 10 classes.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "U_h12Aozqge7",
        "outputId": "530c6a46-be30-4fa3-ad6d-fc76e840eb7c"
      },
      "source": [
        "# What size is our data?\n",
        "train_data"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<BatchDataset shapes: ((None, 224, 224, 3), (None, 10)), types: (tf.float32, tf.float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ZSo9wTmBy6zr",
        "outputId": "f2b95a59-16f7-4efa-9cda-4b6aced9e961"
      },
      "source": [
        "# Create model\n",
        "model_1 = create_model(num_classes=len(train_data.class_names))\n",
        "\n",
        "# Fit the model\n",
        "history_1_percent = model_1.fit(train_data,\n",
        "                    epochs=5,\n",
        "                    steps_per_epoch=len(train_data),\n",
        "                    validation_data=test_data,\n",
        "                    validation_steps=int(0.25 * len(test_data)), # validate for less steps\n",
        "                    # Track model training logs\n",
        "                    callbacks=[create_tensorboard_callback(\"transfer_learning\", \"all_data_aug\")])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Saving TensorBoard log files to: transfer_learning/all_data_aug/20210210-231226\n",
            "Epoch 1/5\n",
            "235/235 [==============================] - 71s 255ms/step - loss: 1.4620 - accuracy: 0.5484 - val_loss: 0.5066 - val_accuracy: 0.8536\n",
            "Epoch 2/5\n",
            "235/235 [==============================] - 50s 211ms/step - loss: 0.7201 - accuracy: 0.7819 - val_loss: 0.4265 - val_accuracy: 0.8569\n",
            "Epoch 3/5\n",
            "235/235 [==============================] - 47s 199ms/step - loss: 0.6176 - accuracy: 0.8028 - val_loss: 0.4116 - val_accuracy: 0.8618\n",
            "Epoch 4/5\n",
            "235/235 [==============================] - 47s 199ms/step - loss: 0.5606 - accuracy: 0.8227 - val_loss: 0.4060 - val_accuracy: 0.8602\n",
            "Epoch 5/5\n",
            "235/235 [==============================] - 45s 189ms/step - loss: 0.5435 - accuracy: 0.8308 - val_loss: 0.3677 - val_accuracy: 0.8750\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "wUKJxKVaY3aA",
        "outputId": "2aa249cd-9223-41b7-b9ee-0490b5f49ab0"
      },
      "source": [
        "# Get an image Tensor\n",
        "!wget https://raw.githubusercontent.com/mrdbourke/tensorflow-deep-learning/main/images/03-pizza-dad.jpeg"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2021-02-10 23:18:11--  https://raw.githubusercontent.com/mrdbourke/tensorflow-deep-learning/main/images/03-pizza-dad.jpeg\n",
            "Resolving raw.githubusercontent.com (raw.githubusercontent.com)... 185.199.108.133, 185.199.109.133, 185.199.110.133, ...\n",
            "Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|185.199.108.133|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 2874848 (2.7M) [image/jpeg]\n",
            "Saving to: ‘03-pizza-dad.jpeg’\n",
            "\n",
            "03-pizza-dad.jpeg   100%[===================>]   2.74M  --.-KB/s    in 0.1s    \n",
            "\n",
            "2021-02-10 23:18:11 (27.7 MB/s) - ‘03-pizza-dad.jpeg’ saved [2874848/2874848]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "a3SwO_LAqHAM",
        "outputId": "61055710-bec2-4832-a753-1cb576444c27"
      },
      "source": [
        "# Classes our model is trained on\n",
        "class_names = train_data.class_names\n",
        "class_names"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['chicken_curry',\n",
              " 'chicken_wings',\n",
              " 'fried_rice',\n",
              " 'grilled_salmon',\n",
              " 'hamburger',\n",
              " 'ice_cream',\n",
              " 'pizza',\n",
              " 'ramen',\n",
              " 'steak',\n",
              " 'sushi']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "p7gCXuavaDAF",
        "outputId": "2c7775a9-8c87-4409-dcb5-589226d1912d"
      },
      "source": [
        "# Preprocess image\n",
        "pizza_img = load_and_prep_image(\"03-pizza-dad.jpeg\")\n",
        "pizza_img"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(224, 224, 3), dtype=float32, numpy=\n",
              "array([[[ 73.625,  76.75 ,  67.125],\n",
              "        [114.   , 122.   , 101.   ],\n",
              "        [146.875, 151.875, 129.875],\n",
              "        ...,\n",
              "        [ 14.5  ,  17.5  ,  10.5  ],\n",
              "        [ 14.25 ,  19.25 ,  12.25 ],\n",
              "        [ 19.75 ,  22.75 ,  15.75 ]],\n",
              "\n",
              "       [[239.125, 243.625, 246.125],\n",
              "        [225.375, 232.125, 234.875],\n",
              "        [240.   , 245.   , 244.5  ],\n",
              "        ...,\n",
              "        [ 11.   ,  14.   ,   7.   ],\n",
              "        [ 20.   ,  23.   ,  16.   ],\n",
              "        [ 20.875,  25.875,  18.875]],\n",
              "\n",
              "       [[ 32.5  ,  34.5  ,  31.5  ],\n",
              "        [ 44.625,  44.5  ,  42.375],\n",
              "        [ 33.   ,  38.   ,  34.   ],\n",
              "        ...,\n",
              "        [  8.75 ,  13.25 ,   6.25 ],\n",
              "        [ 14.875,  17.875,  10.875],\n",
              "        [ 13.625,  20.625,  12.625]],\n",
              "\n",
              "       ...,\n",
              "\n",
              "       [[ 61.875,  40.875,  19.875],\n",
              "        [ 60.   ,  42.   ,  22.   ],\n",
              "        [ 61.   ,  43.   ,  21.   ],\n",
              "        ...,\n",
              "        [134.5  ,  96.125,  60.75 ],\n",
              "        [104.875,  69.375,  43.125],\n",
              "        [106.25 ,  75.25 ,  46.25 ]],\n",
              "\n",
              "       [[ 62.75 ,  44.75 ,  24.75 ],\n",
              "        [ 61.125,  43.125,  23.125],\n",
              "        [ 58.125,  40.125,  20.125],\n",
              "        ...,\n",
              "        [159.125, 111.125,  64.625],\n",
              "        [160.375, 112.375,  66.375],\n",
              "        [134.   ,  94.125,  56.75 ]],\n",
              "\n",
              "       [[ 59.625,  42.625,  22.625],\n",
              "        [ 61.75 ,  43.75 ,  22.5  ],\n",
              "        [ 59.25 ,  41.25 ,  21.25 ],\n",
              "        ...,\n",
              "        [161.5  , 113.5  ,  64.5  ],\n",
              "        [159.   , 108.   ,  63.   ],\n",
              "        [155.875, 104.875,  59.875]]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "81z-ynOlabav",
        "outputId": "4e9ed977-5b2a-4e6d-85e5-9fab2680f8ef"
      },
      "source": [
        "# Make predictions\n",
        "pizza_expanded = tf.expand_dims(pizza_img, axis=0) # expand image dimensions (224, 224, 3) -> (1, 224, 224, 3)\n",
        "pred = model_1.predict(pizza_expanded)\n",
        "pred"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[8.2119545e-03, 3.3304928e-05, 6.2837714e-04, 1.6367142e-04,\n",
              "        3.0865319e-06, 5.1294103e-07, 9.9020046e-01, 1.2048778e-04,\n",
              "        5.5204531e-05, 5.8289926e-04]], dtype=float32)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 16
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jHIoie0Faoin",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "b6a85e7b-b9f1-42c5-a860-f79b127b8c04"
      },
      "source": [
        "# Check the predicted class\n",
        "class_names[tf.argmax(pred[0])]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'pizza'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 17
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "z6mNxcPDqwQk"
      },
      "source": [
        "## Save and upload Model 1 to Google Storage\n",
        "\n",
        "We want to save our model in the TensorFlow [`SavedModel`](https://www.tensorflow.org/guide/saved_model) format because:\n",
        "  * We're going to upload our `SavedModel` to Google Storage.\n",
        "  * Then we're link our `SavedModel` to AI platform to serve it ([AI platform requires `SavedModel`](https://cloud.google.com/ai-platform/prediction/docs/deploying-models) format for model serving).\n",
        "\n",
        "We can do this using: \n",
        "1. `model.save(\"PATH_TO_SAVED_MODEL\")` to create a `SavedModel` object (our trained model).\n",
        "2. Upload the model to Google Storage using the following guide: https://cloud.google.com/storage/docs/uploading-objects#gsutil "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "h0jm4Hzdc85u",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "5b0e6b61-4624-4e41-bb95-afd58e94f49b"
      },
      "source": [
        "# Save model_1\n",
        "model_1.save(\"efficientnet_model_1_10_classes\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Assets written to: efficientnet_model_1_10_classes/assets\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "o4zRwctvsQrj"
      },
      "source": [
        "The next few cells will allow you to upload files directly from Google Colab to Google Storage."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YDsYbY64bFMp"
      },
      "source": [
        "# Authentic Colab \n",
        "# (running this will give you a code to authentic your Google Colab instance, keep the code secret)\n",
        "from google.colab import auth\n",
        "auth.authenticate_user()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "19sjvgD2G6xF"
      },
      "source": [
        "When you run the cell below, it'll ask you a bunch of questions about setting up `gcloud`, here's an example and the answers I chose (you'll want to change the project to your specific Google Cloud project):\n",
        "\n",
        "```\n",
        "#  [1] Re-initialize this configuration [default] with new settings \n",
        "#  [2] Create a new configuration\n",
        "# Please enter your numeric choice:  1\n",
        "\n",
        "# Choose the account you would like to use to perform operations for \n",
        "# this configuration:\n",
        "#  [1] daniel@mrdbourke.com\n",
        "#  [2] Log in with a new account\n",
        "# Please enter your numeric choice:  1\n",
        "\n",
        "# Pick cloud project to use: \n",
        "#  [1] airbnb-amenity-detection\n",
        "#  [2] auto-ml-dog-vision\n",
        "#  [3] auto-ml-dog-vision-test\n",
        "#  [4] daniels-dl-playground\n",
        "#  [5] ml-kit-codelab-demo\n",
        "#  [6] nutrify-v0\n",
        "#  [7] tf-course\n",
        "#  [8] Create a new project\n",
        "# Please enter numeric choice or text value (must exactly match list \n",
        "# item):  4\n",
        "\n",
        "# Do you want to configure a default Compute Region and Zone? (Y/n)?  n\n",
        "```"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vo7ZFsIQdqTJ"
      },
      "source": [
        "# Install and initialize the gcloud command line tool - https://cloud.google.com/sdk/docs/install\n",
        "# (this will give you access to gcloud tools within Colab)\n",
        "!curl https://sdk.cloud.google.com | bash 1> /dev/null\n",
        "!gcloud init"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZKZKFF1YuPJw"
      },
      "source": [
        "Once you've gone through the steps in the above cell, you'll be able to use `gcloud` utilities from Colab.\n",
        "\n",
        "The main one we're after is copying our model from Colab storage to Google Storage.\n",
        "\n",
        "For example, running:\n",
        "\n",
        "```\n",
        "# Copy local model folder to bucket named \"ml_deployment_lec_bucket_0\"\n",
        "!gsutil cp -r efficientnet_model_1_10_classes gs://ml_deploymet_lec_bucket_0\n",
        "```\n",
        "\n",
        "But the above line uses my Google Storage bucket/model, yours will look more like:\n",
        "\n",
        "```\n",
        "!gsutil cp -r YOUR_MODEL_NAME gs://YOUR_BUCKET_NAME\n",
        "```\n",
        "\n",
        "**Note:** To upload anything to a bucket, you will need an existing bucket. See the documentation on [how to create a Google Storage bucket](https://cloud.google.com/storage/docs/creating-buckets)."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "z9Xmp96WdkJD",
        "outputId": "77c962ea-9df1-45bf-b8c1-973affe473d0"
      },
      "source": [
        "# Copy model to bucket - https://cloud.google.com/storage/docs/uploading-objects#gsutil \n",
        "# Use \"-r\" for folders (r stands for recursive)\n",
        "!gsutil cp -r efficientnet_model_1_10_classes gs://ml_deploymet_lec_bucket_0"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Copying file://efficientnet_model_1_10_classes/saved_model.pb [Content-Type=application/octet-stream]...\n",
            "Copying file://efficientnet_model_1_10_classes/variables/variables.data-00000-of-00001 [Content-Type=application/octet-stream]...\n",
            "Copying file://efficientnet_model_1_10_classes/variables/variables.index [Content-Type=application/octet-stream]...\n",
            "\\\n",
            "Operation completed over 3 objects/22.1 MiB.                                     \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "EPzjTa_MwAHV",
        "outputId": "07dce2af-501b-4209-9ff1-30a225733f6c"
      },
      "source": [
        "# Copy model to bucket\n",
        "!gsutil cp -r efficientnet_model_1_10_classes gs://cs329s_live_bucket_creation"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Copying file://efficientnet_model_1_10_classes/saved_model.pb [Content-Type=application/octet-stream]...\n",
            "Copying file://efficientnet_model_1_10_classes/variables/variables.data-00000-of-00001 [Content-Type=application/octet-stream]...\n",
            "Copying file://efficientnet_model_1_10_classes/variables/variables.index [Content-Type=application/octet-stream]...\n",
            "- [3 files][ 22.1 MiB/ 22.1 MiB]                                                \n",
            "Operation completed over 3 objects/22.1 MiB.                                     \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-dZUgNMuc3Cw"
      },
      "source": [
        "# # Download model (to upload to GS, you could probably use gcloud cli here but whatever)\n",
        "# !zip -r /content/efficientnet_model_0_10_classes.zip /content/efficientnet_model_0_10_classes\n",
        "# from google.colab import files\n",
        "# files.download(\"/content/efficientnet_model_0_10_classes.zip\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "f8I-_NPchy3S"
      },
      "source": [
        "## Model 2 (11 food classes)\n",
        "\n",
        "Same as model 1 but we've added the donut class."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fOe3rxUuzg_c",
        "outputId": "9394dfd6-ccbf-4968-ebae-dbe1c5a10757"
      },
      "source": [
        "# Get data\n",
        "import zipfile\n",
        "\n",
        "# Download data\n",
        "!wget https://storage.googleapis.com/ml_deploymet_lec_bucket_0/11_food_classes_all_data.zip # note: this may be broken if using well beyond Feb 10 2021\n",
        "unzip_data(\"11_food_classes_all_data.zip\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2021-02-02 07:20:31--  https://storage.googleapis.com/ml_deploymet_lec_bucket_0/11_food_classes_all_data.zip\n",
            "Resolving storage.googleapis.com (storage.googleapis.com)... 74.125.20.128, 74.125.195.128, 142.250.99.128, ...\n",
            "Connecting to storage.googleapis.com (storage.googleapis.com)|74.125.20.128|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 570914387 (544M) [application/zip]\n",
            "Saving to: ‘11_food_classes_all_data.zip.1’\n",
            "\n",
            "11_food_classes_all 100%[===================>] 544.47M   283MB/s    in 1.9s    \n",
            "\n",
            "2021-02-02 07:20:33 (283 MB/s) - ‘11_food_classes_all_data.zip.1’ saved [570914387/570914387]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "grxb1zD00vcj",
        "outputId": "f26ec76f-e526-4ed8-df7a-6aff9be5bfd1"
      },
      "source": [
        "# Create data loaders\n",
        "train_data, test_data = create_data_loaders(train_dir=\"11_food_classes_all_data/train/\",\n",
        "                                            test_dir=\"11_food_classes_all_data/test/\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 8250 files belonging to 11 classes.\n",
            "Found 2750 files belonging to 11 classes.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "lhMu44ulnJPb",
        "outputId": "8adfa5d0-32d4-4f36-8835-6457abc1e85a"
      },
      "source": [
        "train_data.class_names"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['chicken_curry',\n",
              " 'chicken_wings',\n",
              " 'donuts',\n",
              " 'fried_rice',\n",
              " 'grilled_salmon',\n",
              " 'hamburger',\n",
              " 'ice_cream',\n",
              " 'pizza',\n",
              " 'ramen',\n",
              " 'steak',\n",
              " 'sushi']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 12
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bimrIF94ypNL",
        "outputId": "c5ab28ee-847b-43a6-a38e-b6896c28c062"
      },
      "source": [
        "# Create model\n",
        "model_2 = create_model(num_classes=len(train_data.class_names))\n",
        "\n",
        "# Fit model\n",
        "history_2 = model_2.fit(train_data,\n",
        "                        epochs=5,\n",
        "                        validation_data=test_data,\n",
        "                        validation_steps=int(0.25*len(test_data)))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/5\n",
            "258/258 [==============================] - 76s 249ms/step - loss: 1.4867 - accuracy: 0.5498 - val_loss: 0.5164 - val_accuracy: 0.8497\n",
            "Epoch 2/5\n",
            "258/258 [==============================] - 56s 214ms/step - loss: 0.7410 - accuracy: 0.7740 - val_loss: 0.4438 - val_accuracy: 0.8601\n",
            "Epoch 3/5\n",
            "258/258 [==============================] - 52s 199ms/step - loss: 0.6339 - accuracy: 0.8005 - val_loss: 0.3787 - val_accuracy: 0.8765\n",
            "Epoch 4/5\n",
            "258/258 [==============================] - 50s 192ms/step - loss: 0.5804 - accuracy: 0.8139 - val_loss: 0.3722 - val_accuracy: 0.8720\n",
            "Epoch 5/5\n",
            "258/258 [==============================] - 48s 185ms/step - loss: 0.5594 - accuracy: 0.8187 - val_loss: 0.3605 - val_accuracy: 0.8780\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "AKIg0TGqnm8f",
        "outputId": "f3b1cdad-dd01-441a-9154-ce35f37826df"
      },
      "source": [
        "pizza_expanded = tf.expand_dims(pizza_img, axis=0)\n",
        "pred = model_2.predict(pizza_expanded)\n",
        "pred"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[9.0466682e-03, 4.5336943e-05, 9.9710169e-06, 6.5794133e-04,\n",
              "        3.1481599e-04, 3.2332186e-06, 2.0485681e-06, 9.8864961e-01,\n",
              "        2.1775642e-04, 4.8929705e-05, 1.0036530e-03]], dtype=float32)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LlT5jOIb3RM4",
        "outputId": "5bf81651-5457-4df8-f9f3-cd948b72f65c"
      },
      "source": [
        "# Save model_2...\n",
        "model_2.save(\"efficientnet_model_2_11_classes\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Assets written to: efficientnet_model_2_11_classes/assets\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2ubWlHH6HZ1e"
      },
      "source": [
        "## Upload Model 2 to Google Storage"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KIzFU2H34tgF"
      },
      "source": [
        "# Auth Colab\n",
        "from google.colab import auth\n",
        "auth.authenticate_user()\n",
        "!curl https://sdk.cloud.google.com | bash 1> /dev/null\n",
        "!gcloud init"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "U4vbfR7c48LG",
        "outputId": "f25c7bd4-84e1-46b6-9f19-0b00b9605ed7"
      },
      "source": [
        "# Upload to GS (requires auth)\n",
        "!gsutil cp -r efficientnet_model_2_11_classes gs://ml_deploymet_lec_bucket_0"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Copying file://efficientnet_model_2_11_classes/saved_model.pb [Content-Type=application/octet-stream]...\n",
            "Copying file://efficientnet_model_2_11_classes/variables/variables.data-00000-of-00001 [Content-Type=application/octet-stream]...\n",
            "Copying file://efficientnet_model_2_11_classes/variables/variables.index [Content-Type=application/octet-stream]...\n",
            "- [3 files][ 22.1 MiB/ 22.1 MiB]                                                \n",
            "Operation completed over 3 objects/22.1 MiB.                                     \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "f89CHkkT3Ujd"
      },
      "source": [
        "## Model 3 (11 food classes + 5000 non-food images, 12 classes total)\n",
        "\n",
        "* 11 food classes are same as model 2 above.\n",
        "* 1 `not_food` class is comprised of ~5000 randomly downloaded images from ImageNet (500 random classes and 10 random images within each of those classes).\n",
        "  * To do this, I used the script at: https://github.com/mf1024/ImageNet-Datasets-Downloader "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "m2ZEO1A03ZsV"
      },
      "source": [
        "# Get data\n",
        "import zipfile\n",
        "\n",
        "# Download data\n",
        "!wget https://storage.googleapis.com/ml_deploymet_lec_bucket_0/11_food_classes_plus_not_food_class_all_data.zip # note: this may be broken if using well beyond Feb 10 2021\n",
        "unzip_data(\"11_food_classes_plus_not_food_class_all_data.zip\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-h_tQU40__4B",
        "outputId": "e9774a6b-4930-4d1c-ce51-5aaf7bbf4c47"
      },
      "source": [
        "# Create data loaders\n",
        "train_data, test_data = create_data_loaders(train_dir=\"11_food_classes_plus_not_food_class_all_data/train/\",\n",
        "                                            test_dir=\"11_food_classes_plus_not_food_class_all_data/test/\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 14343 files belonging to 12 classes.\n",
            "Found 4273 files belonging to 12 classes.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "iYuvw9xnACbC",
        "outputId": "f4c35d94-dfd0-4ddc-a7d2-900412e487e4"
      },
      "source": [
        "# Create model\n",
        "model_3 = create_model(num_classes=len(train_data.class_names))\n",
        "\n",
        "# Fit model\n",
        "history_3 = model_3.fit(train_data,\n",
        "                        epochs=5,\n",
        "                        validation_data=test_data,\n",
        "                        validation_steps=int(0.25*len(test_data)))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/5\n",
            "449/449 [==============================] - 117s 235ms/step - loss: 1.2646 - accuracy: 0.6447 - val_loss: 0.5087 - val_accuracy: 0.8778\n",
            "Epoch 2/5\n",
            "449/449 [==============================] - 88s 195ms/step - loss: 0.6192 - accuracy: 0.8225 - val_loss: 0.4408 - val_accuracy: 0.8873\n",
            "Epoch 3/5\n",
            "449/449 [==============================] - 84s 186ms/step - loss: 0.5418 - accuracy: 0.8443 - val_loss: 0.4145 - val_accuracy: 0.8864\n",
            "Epoch 4/5\n",
            "449/449 [==============================] - 78s 173ms/step - loss: 0.5053 - accuracy: 0.8512 - val_loss: 0.4024 - val_accuracy: 0.8864\n",
            "Epoch 5/5\n",
            "449/449 [==============================] - 77s 170ms/step - loss: 0.4768 - accuracy: 0.8578 - val_loss: 0.3706 - val_accuracy: 0.8873\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "FhJf4JqnAEgk",
        "outputId": "be56b233-f112-48bf-9ce5-11c361bc48e8"
      },
      "source": [
        "# Save model_3...\n",
        "model_3.save(\"efficientnet_model_3_12_classes\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Assets written to: efficientnet_model_3_12_classes/assets\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "X3rKYI4cNmgZ",
        "outputId": "cdb49b46-901d-4869-f032-aadfd902dba4"
      },
      "source": [
        "model_3.evaluate(test_data)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "134/134 [==============================] - 17s 124ms/step - loss: 0.4300 - accuracy: 0.8781\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.4300059676170349, 0.8780716061592102]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IthAoLz7Hf9r"
      },
      "source": [
        "## Upload Model 3 to Google Storage"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "H-w8VfVyMAob"
      },
      "source": [
        "# Auth Colab\n",
        "from google.colab import auth\n",
        "auth.authenticate_user()\n",
        "!curl https://sdk.cloud.google.com | bash 1> /dev/null\n",
        "!gcloud init"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "XlG4n046MB0t",
        "outputId": "2c6bdeb7-42a4-4d32-d2f9-d430835dc81c"
      },
      "source": [
        "# Upload to GS (requires auth)\n",
        "!gsutil cp -r efficientnet_model_3_12_classes gs://ml_deploymet_lec_bucket_0"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Copying file://efficientnet_model_3_12_classes/saved_model.pb [Content-Type=application/octet-stream]...\n",
            "Copying file://efficientnet_model_3_12_classes/variables/variables.data-00000-of-00001 [Content-Type=application/octet-stream]...\n",
            "Copying file://efficientnet_model_3_12_classes/variables/variables.index [Content-Type=application/octet-stream]...\n",
            "- [3 files][ 22.1 MiB/ 22.1 MiB]                                                \n",
            "Operation completed over 3 objects/22.1 MiB.                                     \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yivB8oTIIebA"
      },
      "source": [
        "## Get test non-food image and make predictions"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "0elI0uUSIiVa",
        "outputId": "88f75ec3-ade3-4de2-eded-aeca099ea26e"
      },
      "source": [
        "!wget https://raw.githubusercontent.com/mrdbourke/cs329s-ml-deployment-tutorial/main/food-images/plant.jpeg"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2021-02-14 06:02:51--  https://raw.githubusercontent.com/mrdbourke/cs329s-ml-deployment-tutorial/main/food-images/plant.jpeg\n",
            "Resolving raw.githubusercontent.com (raw.githubusercontent.com)... 185.199.110.133, 185.199.109.133, 185.199.111.133, ...\n",
            "Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|185.199.110.133|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 542790 (530K) [image/jpeg]\n",
            "Saving to: ‘plant.jpeg’\n",
            "\n",
            "\rplant.jpeg            0%[                    ]       0  --.-KB/s               \rplant.jpeg          100%[===================>] 530.07K  --.-KB/s    in 0.03s   \n",
            "\n",
            "2021-02-14 06:02:51 (17.9 MB/s) - ‘plant.jpeg’ saved [542790/542790]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "sGcZLtMLOJXa",
        "outputId": "970af992-bab7-4747-8b08-68be7a3f119e"
      },
      "source": [
        "plant = load_and_prep_image(\"plant.jpeg\")\n",
        "plant = tf.cast(tf.round(plant), dtype=tf.int32)\n",
        "plant"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(224, 224, 3), dtype=int32, numpy=\n",
              "array([[[110,  96,  95],\n",
              "        [107,  93,  92],\n",
              "        [ 88,  74,  71],\n",
              "        ...,\n",
              "        [141, 155, 164],\n",
              "        [207, 210, 209],\n",
              "        [242, 241, 237]],\n",
              "\n",
              "       [[117, 106, 104],\n",
              "        [117, 106, 103],\n",
              "        [116, 105, 101],\n",
              "        ...,\n",
              "        [132, 144, 152],\n",
              "        [207, 207, 208],\n",
              "        [202, 199, 201]],\n",
              "\n",
              "       [[107,  96,  92],\n",
              "        [119, 108, 104],\n",
              "        [102,  91,  87],\n",
              "        ...,\n",
              "        [177, 187, 189],\n",
              "        [206, 204, 202],\n",
              "        [204, 200, 197]],\n",
              "\n",
              "       ...,\n",
              "\n",
              "       [[ 20,  70,  49],\n",
              "        [ 37,  79,  57],\n",
              "        [111, 159, 133],\n",
              "        ...,\n",
              "        [122, 175, 142],\n",
              "        [ 84, 143, 110],\n",
              "        [ 49, 116,  80]],\n",
              "\n",
              "       [[120, 154, 131],\n",
              "        [137, 164, 147],\n",
              "        [108, 127, 111],\n",
              "        ...,\n",
              "        [103, 169, 133],\n",
              "        [ 46,  95,  66],\n",
              "        [ 83, 143, 110]],\n",
              "\n",
              "       [[ 43,  83,  62],\n",
              "        [ 20,  49,  31],\n",
              "        [ 33,  74,  54],\n",
              "        ...,\n",
              "        [ 53, 117,  80],\n",
              "        [ 22,  69,  35],\n",
              "        [ 63, 129,  93]]], dtype=int32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 29
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "X28-g67qQAaL",
        "outputId": "1cf5a8b2-7963-49eb-ce38-90b1496b2ded"
      },
      "source": [
        "model_3.predict(tf.expand_dims(plant, axis=0))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[1.14653085e-04, 3.54328367e-04, 1.86755587e-05, 5.58937667e-04,\n",
              "        2.95464328e-04, 4.18224517e-04, 1.86513625e-02, 9.65816736e-01,\n",
              "        2.96051439e-05, 6.32123483e-05, 6.92888675e-03, 6.74987910e-03]],\n",
              "      dtype=float32)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 30
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Q2hqgWyEHl4R"
      },
      "source": [
        "# Make and plot test predictions"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aw2rajQyOr7s"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "def pred_and_plot(model, image, class_names):\n",
        "  pred_probs = model.predict(tf.expand_dims(image, axis=0))\n",
        "  pred_class = class_names[tf.argmax(pred_probs[0])]\n",
        "  \n",
        "  plt.figure()\n",
        "  plt.imshow(image)\n",
        "  plt.title(f\"{pred_class}, prob: {tf.reduce_max(pred_probs):.2f}\")\n",
        "  plt.axis(False);"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 264
        },
        "id": "iGFwjYKIP0NL",
        "outputId": "27df69b7-6f3a-4ad0-a6d2-f495a5183f79"
      },
      "source": [
        "pred_and_plot(model=model_3,\n",
        "              image=plant,\n",
        "              class_names=train_data.class_names)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAOcAAAD3CAYAAADmIkO7AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOy9eaxlyX3f96nl7Hd/+3u9d0/3LBxq6BEpiRIlSrFly5ZiGVEc27K8JLDhBFkQyBASAYYpJIgD/5EVcAIDDuhEspUohqOYFh3JoVaTHFIckRzOwlm6p/e33/2steSPczloDWZGHIYSaaq/wMW7p6pOVZ1761u/te4T3nse4iEe4psP8hs9gYd4iId4azwk50M8xDcpHpLzIR7imxQPyfkQD/FNiofkfIiH+CbFQ3I+xEN8k+IhOb9OEEL8+0KIAyHEQgix9nXs98NCiDtfr/7e5dgfFUL8l9+IsR/iITnfEUIIL4S48lW0C4D/BvhB733He3/y+z+7b34IIZ4SQnxOCJGv/j71Dm0fE0J8QggxFUK8KoT4Mw/U/fhq0/vKK199N0//wTzJNwYPyfn1wRYQA89/oyfybiCEUL+PfYfALwI/CwyBfwj84qr8zW31qu3HgBHw14GfFUJcBfDe/9xq0+t47zvAfwBcB579/Zr/NwO+5ckphHhdCPE3hRBfXO3K/7sQIn6g/q+tdupTIcT/LYTYXZX/xqrJF1a79b/zNv1fBb68upwIIT6xKv+gEOKzqzE/K4T44AP37K7GOl2N/dceqEtW6uRYCPEC8P538awfFkLcEUL8tBDiePXsP/5A/UeFEP+TEOKXhBBL4PtXEuvXhBATIcTzQoh/803drgshfkUIMRdC/LoQ4vxXOZ0PAxr477z3lff+fwAE8ANv0fZRYBf4b7331nv/CeBfAT/xNn3/ZeB/9d/q6W3e+2/pF/A68BnaL38EvAj8jVXdDwDHwB8BIuB/BH7jgXs9cOWrGOPCqq1eXY+AMe3i0sCfX12vrep/A/h7tNL2KeAI+IFV3X8N/Oaqj7PAl4A7X+WzfhgwtCp2BHwfsASureo/CkyB76bdmLvAq8BPA+Hq85i/qf0c+N5Vf/898FsPjPcx4D97m7n8p8DH31T2MeAn36Lte4AFIB4o+xXgn75F2/OABS5+o9fW7/va/UZP4Pf9AVty/sUHrv8u8D+v3v8D4O8+UNcBGuDC6vprJedPAJ95U5tPAX9lRTgLdB+o+zvAR1fvrwN/4oG6v/41kDN7oOz/AP7W6v1HaSXOV+o+BOwD8oGyfwx85IH2P/+mz8cCZ7+KufytB+9dlf3cV/p+U3mweu6fWr3/QaAG/p+36ffXvtHr6g/i9S2v1q6w/8D7nHaRQStNb36lwnu/AE6Avf+f4/2ufle4uep3Fzj13s/fou4r995+U927wdh7v3zT/bsPXD/Y9y5w23vv3mYuv6v96vM5fVN/b4cF0HtTWY9WEv8ueO8b4EeBP0X7Xf0k7abyVl7qv0Rrv37L4w8LOd8O92jVJACEEBmwBtz9eva7wrlVv/eAkRCi+xZ1APdppeuDde8Gw9VzPHj/vQeuH7TT7gFnhRDyTe0ffP435iKE6NCq2w/293Z4HnivEEI8UPZe3sZp5r3/ovf++7z3a977Pw5cojVH3oAQ4rtpN4b/86sY/197/GEn5z8G/urK5R8B/xXwjPf+9VX9Ae0iebf4JeCqEOIvCCH0ypn0OPAx7/1t4JPA3xFCxEKI9wL/Hq1XE1qJ8Z8LIYZCiDPAf/Rgxyunzkd/j/F/RggRCiE+BPww8Atv0+4ZWk3ip4QQgRDiw8CPAD//QJs/KYT4npWX9b8APr16ht8Lv0arAv/HQohICPEfrso/8VaNhRDvXX0eqRDibwI7tGr1g/jLwD95k9bxLYs/1OT03v9LWhvmn9BKrMvAn3ugyUeAf7jyZP7Zd9HvCS0pfpJWTf4p4Ie998erJn+e1k69B/xT4G+v5gLwM7Sq5Q3gl4H/7U3dn6X1ZL4d9mmdT/dobby/4b1/6W3mWdOS8YdoHWN/D/hLb2r/j4C/TavOPg38xa9UCCE+LoT46Xfo+0dp1dAJ8O8CP7oqZ+VR/vgDt/wE7XdwCPwbwB/z3lcPjBUDf5Y/JCotrLxjD/GvB1bS6wvAe1d22pvrPwz8rPf+zB/03B7i6w/9jZ7AQ3z1WEmdx77R83iIPxj8oVZr3w1WatjiLV4f/73vfoiHePd4qNY+xEN8k+Kh5HyIh/gmxTvanP/JX/0r3hjD6ekx58+fZzJfcHoyZjAcIAQEQUBVVQgJQkJZlgQywhhDksYoJSmLEokgSUKEkORFRaebUDYN48JQmYayrulIyVam6fcStFZ4D6enpyilUEKilaKua9I0xTnH4eExo/V1TF3ivWe+mBMlCb1ej/l8jvcQxQnj8RgpQeuAJEnRSlFVBXXVYIzFrcJ+WkkWi/kb/QvvicIIpRTGGIwxOOfodHrMljOms4IoTBACnDdUVUOaZcznC5RS7O3tcuPeMaenJ2yuDdBa0RjDdNEwW5Z0swiBZ94YcidASAQQKcnasE+v24ZBO50OYRTz+GOPsbOzzcHRIRcvnOXlL3wOV4zZWhtxcnKCDiAONXHapzES4eZUVUWWZQRByMHhIR/+vu/lzp07lGVBURREcUqnMyTtD3jh1iFfupvzpRdeYnp8gKxLAh0wOdrn3/7jH+IDTz2OlBItFVEUczTPWd/Y4bkXXuP20Zg7h8eEOuDKuW2eevIazz33RZI4IUh6PPPsczhvWF9bxzpYLOe8/+mnefZzz2CdwxlHv9vl6W9/mt/65Ccpy5Ikjul1u/R6Aw6PjwFPEoVEYcil83tkkWJZ5Dz53vfx5Zdepa5zNoZdqrJke2eLXqeHdY7prCLOUi6e3+Wzz/w2u3t7XLtyltODu4xnOc99+TU+9D3fQ9JJuH39Jsu8IEl71HWDqWsuXDzPdDomTVPwiqPjYxCS2XTGfDbh3Pk9zp8/T1F5gkCxXOR88refRff7vHbnDifjI9579TGe/eJzLJ3gvd/+ASaTJZvdmB/4jie5fP4Mvf6aeCv+qY985CNvS85P/PK/+EgYB3R7PRZ5ztbWNsuiQkiJwOK9pyxLoijGNJYszVqiCrDGUhYF1joCLfF48I5uNyNNYhbzJU4IlNY4D7YuGXRimipHCkFZNLz6yg22NtbxHpbLnH5/QJ4viOMY7z23bt6k280QAgb9PnEUkxc5Wmu01kghSOKYMAgxtiVX3dSUZYEHpJBIKQmjEGMaOp0OcrUR4MEYgxC0bcKIPM+p6wbQGOMII42UHqU0znmaxqCUwgvBdL7gxp37IBSDTor1iroxlGWOUhKtVbswhcQLiUcggEArwiAgiVP6/QGdbo/haEBZFiAFYRhSFxW7O7t00h7DtS0uXHoEATTGMJvMkEqzs7WBFAol2/03SWLyosCjiaKMKAoZDgaURcEyX3LzzgGv3TmmKCtG3Q4Xzp6lk2W8/31P8cjFCwyykGE/oywqpAx58bVbHE/nJFmH2lguXDjH449epcrnTCZHKBUSRTF1WRBpz1OPXuTzz36e7/7gB7l95zb79+9y9fIV6qahPxwghOT69etEcYQAvHNcuXyZc+fO4m3N+lqP4aDH2b1dLl86z9HBITdu3KIsSnZ2dnnf+95HFCjWN0eEQYD3ot2Ywpjx+JA0jQmUwHjHr/7mp3n02mN0BgOWy5pACOq84NzlR9k/OmJttMbG2jo6lCgtOT6ac+PGvXY9O08Sxzxy+SISh20sUgjy5Yys3yPKYt7/1BOc3VyjP8hYH4w4u7fH559/ntI6up0Oy2VOLOGRc3sc7R+xd+bMz7xrcv7iL/z8R8IwYjabIQRM5xMQkiSN2d0egWtYXxvinUUHAeAIQ02v2wHvCaMQpQXeeYq8Ii8KdJQwXSzJG0teN0itKE0NOHqxJgs0+WJOEoWcPbOLtQZrDVmaoJVECUVdN5RVwfnz51aLOcRYi7MOby2B1njvMdZiTINpahpjmc1zJpMJWktMYynLCu8tcRwhZavhSw9KKsqyxDmH1gq8xxqDVrrtq6lRWiEAnGv/CrDG4b3DO0+oFMP+gE4cYE3NPK+YzecM+x2kEnjAOU9tLJV1NMa2WoIUZEmEszWnp8c0dYk3dbt5ZRHzkwNSDcJZqipHhxolBUVZMp9NaMolSaQJwxDnBJPJlOGwj44CjIP7BwcIrel3eygpCSONkuBVxNRoFkXFcjYllDDq9xj1M157+SW6WUIUR3gZcOvufcIgYDoeE0Qh3nlMXXL35h2khIODfSpjmc4XNE7grGdnfcB3fOBpPv+F3+GJJ96DbQzzfI7Hk8QpH/6+D6OE4PjomKqq2NjcYp4XzOYLTo4PsU3BhbN7rI2GlHlOEiWsb2yQxgLlKz73zG+xzA2T8ZiqLtjd2aOuDWmaslzO0Kt1WDYN/UEf4Rxh3OH6SYGplqx1OvzCx3+J7/jeH+A3PvGrdLOUPM85PJnxyc88w/d++ENMZ3M+9elP0+1kYC2maegNBhhriaOYZ5+/zo17BfsHM77whS8ihMY2FZ/57Gd47fZtnv7ODxLFKd5Znn7yGt1QIbDsnjn77sn5cx/92Y+cjmesbwzJ0pimaZBCMZtOGJ+cIKSmrhryosQZgzUNgdYURUkUtQt+Nl8glCIIQ4IowlhDUVWMF0ssnrquQSpsWbLWiQmlQAUK6yzOOaQQaKUQQhBFEbP5jCgKCYKAsixRslV3nXfEccxisURqhdYaZ8wbRrUUkjDQpGlEHCXoIEBqgQ50S+66QfhWclnbSsxAa9RqbGttS0gpgDYpWQDOe4xzBFqRpBE4i2nsiqQG6UGroB07iamaGmccUkiUUiDACYmTEu88wjniMCTSYataB4pQCRItoMnZHGac3LvBYnKAa5YU8zHNcsZyeox1hs2tLUaDIYeHh0wmcwaDIVVVEwQhRVHQ63Y5vHcPKQRKS05PJjjrOZnlPPP8a0ymM4ZJyM6oS7FcsDHs8MRjj7HWieilmrNndpnMluxsbwKO6WxKHCcsy5LRaMj+/n12zpzn+s07vH77LvPFgmtXr/LiCy+15ooxnJxOOHfxPEEYs5zM+OD7P8BLLz3P4dEhy6Kg3+0Rh4qmzOnGmisXL7A+7OPqgvl8zun4lEW+4Orli/S6HZI4JstSbGOxjWNzc5d5UbA+6uFdyeOPP8qXX3yFw4NDHnnkCjeu3+LytUf51V//FNJ5Nroxs8WUy2f2CLzn4iOXODg4xQpBYRRPPfk4aSi5d+8eQko++MHvwjtHECkGowFaKubLOTtbm+BqBDU6EFy+dAkVaf7Un/4zfOxX/l/irEvVSJSq+KHv/k5e/MIN+qPO1yY5f/mff+wjQsD29gb79++jgxBnHXEYYExDlmSAoJPGK2nUkOc5xljKsmQxX1CXFf1+F60gTRJ0GOCsIwpDjHVY21p9Uko6UUCkJXEYgAep9BsSTQiBM4YgCJhOp0RhiHeOpqkIw4AwCLDW4Zwj0JqmsVgvMc6hlcRbg7UWIcUbKq4UCmvbTUAg8N5hrUUrRdM0BEFAFEUIIdBaU9c1SrXED1YbTlXX4FuTwVpPbQxCCKRUCClpmrp9Zm9wtsFZixACpYNWnbaW0npUGKKVJlAaSasqmbqhkyQkUcCw12HUz1BA09QMh0PSJGbY71EUeTu+c+zs7HB4cEC/38e6hni1SeI93V6Psiyp64o0izCmIU17CBRF41j6hKoxRL7miYtn2BoNsFZwOinYHnVJA8/45AitFN4Zsl6Pu/cO8CgO7+9TVznj6RSLom4sQRCwtbXFxcuXmeUloRIM19d56ZXreNvgjWE+nXJ0coT3lkBJOt0M29Q458iXS3Y2R3jX0NSG82f3OHd2j+31AYESnJ5OmE5mnJ6OSbIug2Gf23fucfbsBW7dvIE3NUmcsMhzpJSc29ujKRvmi8Wq3SXWttYp8xkXLl+gLEoCpZlPp/Q6GWWRc25nA+Ud45Mxvf4QJwOe+Z2X6A/X2NwYcvf2Teb5nN5gQABknZSqqemsr/OP/tnHuXTlGn//f/koDZILV64wWBthl2POrq/T7SbsHx7x6KPX3pKc7+gQquuSbrdDURQIIaiqGiUEQRAQ9PtUpiIKArzwCCEJo4gkTSmKEikl3W6HMAwRQlAURSsx5wsMLWmqumlt1k6GjmPuHBzSP79L0zQ0xhEEviWRlASyJYzQgq2tDYwxhKFu1VZjCXSA9Q1RGGCsw8mIl2/eYW3UZ72j0VohlaAxFtPAfFFgTUWvm6JXKqqQEr8a7yt2bVVVK+JZ1GqR25XqbKzBe4uWurW1lUPqgOWiwjsQSrbqpXfEccJ0uqDxmixNWS7nNE2DdQ4lFZW14Hw7Fy8IlCRAY+oaZwKauiAvYDat6fe7BEFAlsYcHtxvzQel6Kcdjg72QTiW+ZwkDsnzBU1j2NxapyxzOllCJ90l0BFVXdLUNcvFEgg5Pj5mOpmyvtlD4MAbkihke2uDTjdhWcwItEZYCIKIRVmjlWBj1KcxDc5Lvv+pp3HWcHhwwMuvvcLuWp8vfPbT9IdrPP74k3zppZc4s7PDY5d2OV3WmKYiUIpQyVbFXy7YGo1wQrCzsw3eM5kvsU3N+qiHMRWmKTnYP+Li+UsEUYBUrRnjnOPM7hqdpOEHf/A7eP3lV0nikNpCWZTUjWFra5urj14jjjssFoY7d17l3O4W/d6I61++RVl5giAkCODi+T0O9u+jdUySpfSGXbbPrfGB73ofd2/e5dXX9xl0B3QzxXK+xNqGxWKOUpIoCvgLP/ZjPPfay+gspacD0iRjUc5Z73ZRwGitz+b6xtvy7x1DKXEUUFclp6cTjBWkaYxxBocjimOkkCzzHOvh5dfvczStcVJhvaVpGuq6YT5fMB5PqSuLQBBIQSA8wlmcacg6HaIwwjuPDBMQreT0woP3xEGANw1VVeG9Rwgoy6p17lQlwoGilapatKqhNZb55IS1XkpVFMyWBQIwdYNEYKqKJAjY3lgjTROCIEAqCcKjtKYxpj2cqTXOOpqmwXuL1K23y9SttLXGEIhW4qog5Pt/+FHe9507RHGMFzCdzlabXEVV1XihmM1KFosKKVSrxrbuD8qqom4aBB4pfOtRrkvSNCEKNU3TtOdJBIRRjLU1tikYDIZESUKaZjjb4D0cnY5ZX19HKUUn69HJemRZgreW5WJOGAaYpsKZGqjZ2BhSlTnYhtlkgsKzszFke32D2/sT7h4tmRcN82XJjRuvU1c1VVly9+59ttfXyELJxrBLUy/44u/8Nkf7d8giyYfe/0fY3RgSasnBwQEf++e/xHDQQ2nBvcMxxwdHxEryRz/0AR65tEccaIS1bA4zzo4y+hpm4xOiQPHUe67x+KNXSZMIhSONFFpa1kcD0iShPxwwHKR82xNXEQhe/tIrOGKOjybki5zd3V12NjY5vLPP7PSUk8N7zE9vcX5nncVszqsvv8KZc9skWcTFq1cZbexw6/oNkjAijUOUcBzvT7j96j1Obr9OJAyumtDUM6w1OO84d+kK5y9dpdsbECtJsX+LJ3bWmR4dMxgMcbQb+1NXr/LFF65TNZKj0/235d87kjOKIhaLGWmaEoSau/cOECKgKBtu3b5NVZdorShrS2U8ZWWxTSthlJKEocY5gxCe/qBDL0vY3lhj0M2IA0mWRFTFkqpsvbq1E8gwozIerQMUAusFk/kSHYXUtgEvmE5nFEXVejhX9mBVlXg8UnnCEDbXMnbXYs5tZGz2UoxtQzbGWuJYk2XBijQV1hikbG1AKUUbOmmaVloqSRi23j/hJcdHY8qyXHlyPUmakCUx0ZbkyR/b47g+oalrAAaDPoPhgCgK0NoTBZYz2x06iUdIC2LlQILWgRYExGFIKHUbPgoUy+UU09Ro2W4CZ3b3qIoSpTRHJ1OKoqKua2aLGcY6mtoQRZrpbEaDYF5UWOkQKiKKE9IkZXI6xlmLlArTOO7eu08QtKGuOIpQ0iO8BV/zXe+7xpOXNzFlSa8/ZHdvD9MU4AwSh5KSO/cPmEwWnN/ZIbIlpig4muTc3j8hlI7HL51hfZCRZCE3X/ky270udTHnzOaQF194jvv37nF8cESkBBfP7tJPQ04O7wEVSRxiG8PNmzf59Kc+yc3r16nriu2tLaSAuqpZTGdkgaBZlHzit7/Ipz73Itdvn5D21rh5+y5KCF59/stcf+UVhsM+Vy4+gnQOJT3Hx0dkWcL6aEi+WCC9Y3a4z81XX+bchXPESYKxjgtXrnDh8iUuXH4MEQwI44StrR2EiomyHrNFzvPPfYk0TZFxB6sU+IayrLDO0ZiGo+NDYuGIQsmZMxs0eY0n/VrJGbSqY1MhBHTStFVrdUCv06OpTOshdQ27mwPWhynW1AgkWdoBD4HSOGuYzaYcnxyTpSlayVbqKEW3k6Fka/s1zlNZx3y2RMuAqioBT5LEeG+JwgAJbK6tMxtP0EKgtURrSRSFSAGmNngvcE4gnEfhkLINtQwGfYSUOAenp1OOjybgwK9UVrvywMVRRBxHOGfxzqKAMNRICVkSMxqNyPO8jV02DXWzQGjFLFyjFCl1XeFca9cuFwuElzS1xTmwpsE1Bm8c3rmVBG7tUL/yCltnEPhWzZWytbNnM8Ig4PDwAGjDN0KGSKXpdTqEQURjFYcnM4aDDZyHNB1iXUCRN8ymE4p88Ya9LoWjqUucqwm0wJiaKI7o9XskoaKXhAzShO1uQmBrOnFIoBQCyTIv6HYy3vPYNSIl2Vpfo6wbXnj5VR578knCOGY+m5IkMVVjyPOKbppx7fJlut0MYwou7m1xfiPjx/+tH6ETSXxdsD7qsrHWJ4xiZJASRz3W1wZ0U8mZnRGXzu/y+LWrDPuDlQBQLOZzDg8OWUwnJFHA+64+wnuubvLI+RHUjisXrmBtxYUL26RJhPCe4/1bTManbGxsEQYhdVUiBcxnC+anMw7379LvdRhPxqvwE7z84ovMxifMJmOaoiEMIsIo5M6t2+zf2ydNEtbX1nj91j2ee+ElgjhlfWeXg/GUaZEzWB/R73bweYkSQes/UCWDYfK1kVNIQZqmGFOTxCFJGBAFCluXLbG6PYSUbIwGxIGjyic424Yc6rqiqWu01iRxSqADtA5wDhrjQGqkgCRs41pBEBJHEUWRk3U65EVBnEQILHGgccZiqoaqKpkXBUEUIpUkz5etnSZAaUkcB0gBHof1nsY6iqqiKEqE9zjTgPAMRz129zaIogBjzMoRpNGBxnnPclngbGtLV1XrjRY4er2U2XyychQp4jjBNorDOwe89sqXOfPYNdb3zqHjEITE2PY+KdrwSWMsxvnWieQFwntCJQmDEOc9Rd2q7N47hBAIJZlOpwxHQ5rG4LwlSWLKskQIi7E18+UcISSdbo9ub8h0VrKcF4Qa4qhid7uDbSrCMCLrZEjdbkRBoMmyhI3NIWWVU1YlVZWzmM8QwpImmrLOSdIYLT3W1Bhr2N3bo64bjg8PERg6sWJvo8elc7ss53M6SUjkSzqBJ9QBTzzxOMKUbA0yNoYdyuUc15R4BM98+lnCMCbJYrytqYqCl6/fZrooKJZjQmrObG6S6ASswDUNURCws7ODtRYpoN/vMlssOJ0fUkwPSVWCsIoqP6EsTpASjo4mZMkAnOXw8B4bawOyKGZnaxOPpayXJFlMfzCiv7bOZDaj2+mxe2aPMAzJOhnOlmDm9LuSxXyMaQqeePwRhv0OWRbjcSgsl89sEfmG2XzBC69eRynFdDqlLnLObfdZ3xxhK0MYJcxnxddGzvl8ijEVaRKjlKDTSQkCtVr4re2ppKAqS8IgJI5itA6QShAnEVEUAbTq4so2u3vvbhvwzxKEkBhnyYscaw0gkEqyXM6ZzuY4597wbgqlaQAnQz7/5ZtMK0dZG5bLHBDUdYNSASoIsc5ijUFpjVSydd4YQ1VVRGGEVBohFEHQEvMrWUDWOqwFvGPZKGZFaxEa0xK3KiqkEPR6HYTwCCEoqwKdKOIwIgg0T/6xb+OxP/EhssEaYLCuVVuFkMRhSByGbbxWghTiDW90+22skiKCNoTT2u01nU6Gs46qqlhf28RaS11VaCUQAqIoQQeKKp+yNgoZdjXdLGI+PQFrOT2ZkMQJTdPgnKWuK4wD5wXFMmc+nbae9KBdhEJIprOc0mp82GeR14DHuYbFckbZFCSdGJCMp0vCMKCbdanLkjhQDDspj1y+xMnJhPF0xuee/SyChnp2QhYI1vodtBYs65zv/6Pfz3w+Zndrg1AqRFNy9dwOV89usjXs0E9jOmnMYNAjjiPquiaOY/b3j1E6AQICHSGFQhIQxpq0kyLwVFVOGCYURcG583ss8ilhHNAfrtHp91kWOV5IlAw4c+483W6frNNBoLGNJc8L5osFHo8KQuIkAeEoipw0yZhOFhwdnTKf5eBAItk7u0dlHTbKePH2Pp9/4QUuXrpEFEV425BqzXKxQAYhx9M5cRLzdnjnI2PWEShNqDV13axCFRYpJc5anHeEWnM6mTFb5GRZhm8aZrMZSRIRBgF1VZMkbUZKURZo3YYgjHNUUnJ0dEJLS7DesVjmbKcpnW4XpWA2XdDvZjghuH94wMUzZ7h0doeNUQdTLt9It5NS0tQ11nuMMdSNIQsCpJR4798gYGPauGzreY0oyqoNy3jQGsIwQHhLN2u9zIKabrf7hm3blonVj005vLNY4wiSNoZrdc77f+QRpJF89hf/BaZc0ul2mc1mREGId1BVNcYZpGw9n6ZpsKKV2EEatZ5H68iyFKW+ou42+ECxzBfgDW1OUeugOtg/IkvatiK0OFPjfJuyGEUhOtAsFnMaaxHAcDigWBYEOiSME5IkxkaCJFpy/eR1NtZjokBTzsaIsKQqSsJAcOmRKxihGY42Obx/m8Y6tre3uXv/EFSKR1PXhjgw2Lrg8qWz3Lh5D61DpBJYJwnjBKUrNtfXcF4wm02oyoY0SVHKs77W2uj97hrjyRRjLIvZlPl0RlFWDPo9gjDkZDzGIwl1zMbakCyLkN5R1UtOjo+JoogsHnL73hG7Z3aw3pOkGVXjSaOd0zIAACAASURBVNMOprbkyxwdRngveOlLL1FVnrTTgVX04bXX7/L00+/BI4iSlCQIqcuaebmgNoZur4P3MBlPCbVmbX2ErUqiOOW5165z9/AIpRXrG5tM8xzbVGytDfHeYpxbCS/3tvR7R8mpVJtpkuc5bmXUVnXrEQzCCLxgmVdUxrEsKypj3lgQSZy0vyIGFEXOYrnAWIsQkiAMWk9sFFEWBXL1MzNKChZFifUg8XgHtTEgBEoILpw9S2MbzmwNKRdTnBfUTcNymdMYQ14UrY3god/voaRCStnGFZUijmOSOKLbSZGyDasEYYSxrWPDe08SRSACEumIqKnr5iu/+vaGpMU5kjhpY6pBgNYSbzxNbjBVxc7oEb7th54g21xHB4K19TXqumY6m70RmpFCgOCNJIdWOiucdxjnqZ2nMg4VRSAkdWNRgUaHmrIq6Q/alLfGGDrdHrPFDCQsFgukClY2uEdLAd6zzJdopelkGU1V4t0qDBS2m8JkOicvCwbDEZvbO0RxxKDfQWEpqgKlAyZHJ0hbU8xPMMZR1xW9fkoQhLi64Olve5SLZ3d44tplsiTEmxrvHN20y8baBknWxSHRYYDAUyxzbF0T6JAs61DXhqI0TOYVReWZlwYjA5I0a2PfqzxojydNI0bDPlGkqOol+/f3GY9PWSyW4B1ZmuK8o24aJpMFs1nBZJFz+cn3YBEcHo3pDgfcu3fQmmFVhVKKvFhw9tw2GxvrbK5vUhQlp6djJsenLGZLhFTEaUK/3yNN4nZNpQlCgRQe7yxJFLK3vcbx0T5CSsq6jfXHWpMXho3NM2gdk6YdvHvLtNrfW3KGQUi32+X2ndt0e32aoqIsa5IoptPpMJsXjCdH7O5uE2r1BqFlFOHxaKUhgCBsVU1T1tR5hfEevKXXH3F2dxsZRBR1ReM8Re0giIg0lEW7U+ZltUo89jjnKcochwTf2m9StpImiiKEtSglsMZhjHkj/tWSFKSEUGsm5Qwp2mT0UGtc0NqbyzyndbgohA/I8zlRGONcq8aCx636zMuSOF5JBWtpTMPB8jXOrj3N0p2w+cQGL915nddfv4UQLfGc8DTe0ziPdZbGOcrGYB04EVBY16pFzuEWOZsCjNSkcQwV1AvH6WlOrRbYpkFoRTWZEqoQkxvKIierHN40zCYzup2MKI1ZLAs6tWDUV8SRxmtB5UCLgNLCwil0nHL2/AipSorGYV1DVTuybo/aWo7HU7a31tjfP2QwWGPQ6yCsoSkWXDq3h8TgXMV4NsULhZYS1xRkgeXC3gisR4YRt+7n5I2ntgbnINSSV157DSE1Iu4QSIEUtk0ZnS0Jge2NNbJel2VeYJ0j62RMlw2b6+vMpxOGa+tEquF0MmZtbUicdjg9OAVhef3WHa49fhUP3LvxCkcH+1grUMozXy64FJ+h19/h13/zM1w8v8fh7fuMT0/Z3N7k+HRMVddIa8lnM6yAOE0wTU0QpTROM55XaLnA2oYk7VAucgglpmqI45T5fEHazdjY3KTTHXF6OqPX6/LyK69w8eLb/0TVO5JTBQGHxyc0jScvKoxxDAZ9vPecnE6IooQsS6nKHO8ahJDUq/hZHEdIBEGo38iwCcOQk5MZcZKgtOb4ZEyaZO1pFqWw3lN6x2SR0+lECN3uKs7SOnSkQyqJX0kxpQRSauqmxjtPYw1aqlWYo5WWAN63RNVar6S3ZzTo412bEeSdf0OaCSFxzmCNRSLpZNmqH09dV2gd0NQVFs9stiAM10E40AqpJZPyVV6Z/F+clif0H+syeGWX/VePMSisFJSmpqgNxoFxHusFBgkCjHEYPFKKNulISsZ5zcxMSJKYbuPgaEoYRxzuL1FKUVYFzjtSJYnTiIP9fdbW1mnKgtlkSicrW8cJnrW1NdytEzbWW2JVVUPTzKisY1w4yqrmdDzhbjknXyzpJRHW05orUtKUJemd+1R5zsVLgrooCbVmVpR88ZUb3Dsak6YdLl2+wMuv3uLM3h4n8yVXsgwZtrHZZVFwfDrl/rTgsct73Lt1l+2tTU7zE7zyyMmc89ubVKYBqYh0jZSevKqpTEPSSel0O5xOF9w5nRN3+2yOBrimZmNjg7Kp2T/cJ+1bbt8/ZWe7x2SxYDDqsd7N2L9zhziKmExPUcJz8eI2XnhefuU1msaSdXrklWG2WCIP72GEIApCdNBqL6/duEmnk3Lp0kXyukSFGbP5gk4aMlzbpK5KrK2YHtd479na2WV9ewucIw0z0jDh5o2b7OzuIoUnid/+P2K8IznzYo6QgiAKEELRSVO8dywXS+azJbt7bYJ7szpSJaWgTaJpJYvSmrqqQQiiOCbQgiRWaK3oDXtMbt5t1dtQY4SgdhYdRUzmc3aygDSNqcqKJI5Y5kuiSBMELdnLIkc5tZKQFqkkTdNQ2bpNJTTNyjklmc+mK9I5iqJsv/QgJIwi6qaiLEtArubeqqpN3ZDXFUIHVCtjvzAO4S37xxN0GNHomHFl0VpTmZIgTpibgtPxy6gooLeX8OSffJLDn32W8c17ONt6jz0a5IqEXqAc4DxKCjwC61vvrnMeh2hDL1QIlbfqGkCg6Q6G9IRgOp3Q6XRIswwfxtR1Q5r12Lt8Da0DqqpGrwyY+WJBGaZMVcqknmKMpKoMx4fH7O/vY51DIljkJUK1qY7dbpe1zU2WiwW2qRiNRty9sU+Vl9y7fbvVQJqaqqy5dPEid4uXMcYyOzjhpDTcmOa8vH9MJ4solgXjWcnh8QEyjLhx8w4v7R/x+uEJo26fMs85OT5iXjvCKMJVU07mM7q91hzY0gEENdPKoNc2MFFM7Q1JlqHDDK1jJuMps/wQgeTXf/PT/OAP/TkmJ2OW4xknpeWJxy5QNxVV49g5d5F/9anPYoTk6uPXkFJydHxM0RQkjUApiZeS/eMxp9MZly6dY2tjhPBw9/CYk/EttrfW2dvdQUtB3lQ423B8csTdO3e5tr2LDkJsueDKhfNMTsckYYgxNXs725T58m3Y93uQ0zYGgSSOAqxzWOOI4hBrGvZ2t1kupkRBiPUWpUPCQNLN2nzbxXyB8K2NiQDT1Hhr6fc61HWJpEeWhJR5waC/jgHyul6pjAKnAlxjEB6kcCRRgHOujUPGMUEYMpstCQOF0qo9VyratMBABrAKXYDAOteqtEJirSeQHqUFUku0160NjMKtzrtVBua142iywOv26BPWY7zDOMPSBrhlQxBGzEoDAmQgcDJGyBivFN1knZ3OHkWnw3f96W/nmX/2LMe37iEbB0JgjV19yhInPaX1eAFSqTbEAmAtWniiULG+NiBNY4a9LmGgGQwGdHpdFouc9117ildfv8Vw0CPRnkEvw9SGfr+3SkywlFVJt9dnURRopeh0u6RZB6UCjLGsrW/QGwyJowSEIOt0sTiOT4+Jg5A0SxkM16jqku3tbbI05fRkQnd9qw20NzVKSYaDIQjIi5Kw18UvSn7ntVvEccT9L71M0zQMBgPm8zmLF1+jWC4o8yNEEHCan/L863cZ9PoIKeh1U4SrW9uynqG15qA5JApjxoucWgeMJycM44hlUXJme0mZz5hMa+oqZzpd8LmXbrL16AtIb/ns577AcOssw7PnGC9KpuMZ5nTGMukyLS3lfMHINNw6PmV8fMiTj16hG4R4L1jkOYPhCBWu1ruzDNKQO7cnHE8nVMYwzFJCpej3+8STOXXdsMxzkrwk8Y5YS26dTtnZ3cULSRCGHB0cfq2Ss25T25yj3+9TrBKIldJ4LEGoieOYoqypmrpV+RpLHIcEocY2rYMI0Z7xdM6hhCAKNPPJmE4SI5wnyzocz6aruJXEKc2yrHE0yFV4Adp0Om9qnLUoqVanQiCQ7YL2UhBFMU3TxvDyoiROIrJOl7puT2boMMZ5GC8WTBbjth/ZngpxtIee87KmtJ5FafDCo0vbJgrINiMpUAFh2IaKvHdtsngUUtcGH4aUlUN3UzrpgFCmPP195wltxL/8uY+RnxQorTC+TUBQSrfqrfPU3gGqdU45y6gTcWa9Sz+NWB/16WQRG6MBVb5k0PckSc29yRGXOpuIvmV9TXG3yrmwuc4in4Of0jiH1WADwbDvWSjXxpBrTycSBIGnrGq2tjMeP/MYngBrPVm3g3GWsjnbJjiYmizLODo6pNfrIXWAq2t2d3eZjE8Zn47ZPXOWqrHs7O5Qlg06CqidZjgaUeY5o60LVGVOVddcuLo6mxtFTMdjnG+dKbVpCMOQpqxAS0ZrQ6ypOa0rhGmYe8VyfgqydahNpwHdbg8VBtx+9QbdbsZkVjAaDKit4LHv+h6+fDzG2Zq628NEkufv3OJkMicpHWVRsHvmLN6V3M+X7C9zZmXBtCjR94+I44hRr0OQdphax8Hd+4QHnmo24+R4zOmypEBz43jGmfU1Igl7e9t89ktfYrC7y2jnLHGc8tSlq0wnE86c22Z7e4vp9BSlIMnePgnhHcm5rKBZ5uxtjfDOAoLZfI4OQvKiIo7C9siXd6RxzHAwoMiX5HnrMUVKsk5GFEdoLSnLirIo22B8YxASpBRUVUlTV2gpaLwDGVCYVsLVVUmaREglkErha0exWOK8Z2t9jfGs3aG8lERhjEeSNzXFsqTIS/w8p6gqhFKoIMR5MI723GJpEcIBHtTq8LP3GNueWJFBSFM1OGcIA43EE4UBoZakYdAmXEhJKNpsHoFBi4CibMc0SYNWIRtbV7HfITl87S4v/NqnMNYQCIlzsl2UeLT3NKL1rDoPmIa90YBzo5RQawJf0QkyOqEnEgG2yTHKMOyFHN59ja1uSqqWrCceVc8ITI4xDaNOKx3rqib2c7LEMatz+p3WSx0Entws0EGAp6SuDJ0sZa0fUjdN6zlXsCwrhv2UYSOBOYiAaXXE7kbChTMZ6ZV1CCLuH03pqZrDxX3e+8hTlAeKs2fXaJo+nf4a11+/SRhFbKxt4pQiyHqMj2dILTkZj4mTlLqukTpA6YBeN8VUSw7u30IqSZb1sI3DWEOeLwFBkmYkWYZpDEJY1rcvAQ1V2eCFoigLkjCkOyiIk4BaSrqjbYb9Psenp8zyEiU8SafHYjZFJQnrF86zbAzjecnhsmQ0WqcxDYt8QagDxkcT9u/eR4UBUmvun55wOJ1SFUt2D4+4czqjs7aJd4bp+JgXXl4SYhn0e9w4vIMxlk6nw2Qy4wPf+Z3vnpzWS4xldSKlDXFY7wkCTVUWGGOx1mKtQzSWZbAkDDXT6ZQgCIjjGOd9e8qk8ZimjZEWZdkeK2salnnBPC+orQUd4J1F6ICTRY6KJZGQ4HybIOAERkiCrMf9/4+zN+m1LEvP857V7P7053Zxb7QZERmZWW1Wzw4uEhIHAizQgIeeeOJf4IlnNfJfsAcGLMA2YIuyBJAyIdoiiybFTlT1rKrMyi7aG7c7/e73Xmt5sE6lJqoyijGMwA1E3HO/vdd6v/d93qtrcpuz3hW0XY/QmthqmtbQWkeVe4dP1/d0xr/lRW/9ztAahNAEiUZKBc76nS1+OJyUCOcFm7b1lr6DccY4CemaGoUh1ZrOWYZZStVa2tYgpSVQEXmzpTE9Vb8jFUMiNeLW6Tt8/pslz3/0EfnyEoE3RrT7gLaUAi0kRgicMcRKME0j0ihEa43QgrI1lHVDlsZsNlskEh1otPZv8rqqGI2GNF2DlgFFvfNvObW3N2qFNY40S9Daq9dKOCbjIWVZ0LUtsVYENNT5tb/fW4ttLSmC8nqHbju0lmRpRjt23J2CsR1ZEtHZAiu36LIgTWpG5Us+eyhR3Sv/UDMtgyPBaBSBWRINRlxVNySpQ2lHUBRMJhF9L4gHE0Q8YVPVdBKssSiliJKUPoQQgYhStqsloyzj5PiQYldydHIIQmDrgqaoGAwzbpZLDg8Puby85ODogKuba7JsyHR2wOnZGU3jtwJ+9dLTmx6lJG3bUlcdUil6a9ntdkxNT9f2hNmc6ck9mqqkrj29wzsuHbUTHN19079QhKM3PR+fXzAYpHyyWOCsZZwNMOevKctf7BD6pcOZRpphFqKF3+sVVU0UBl4ajkOs7QmCkMEgoigKbxq3xh85pMA4R1FVe7XU+OSI1t6ehjeYx1HEarOj7g3JIERJBcKRtz35riAQjiSOaHuDkxKpNEmmWHWCtsz9A8NItJCUtcE6h1QBIhRIIHKQKfXzQAfOOqzzQ6qlZxWFWrFeLZiMhkgHUkucsXTOEwaMdByOUoaRYtV4RhDOUu52zMcjqtYSRoIw0jRO0HbQ9Iaqr4jjBkPHNDtjenLOwd3bmGpDU9feoodDCgBH33eIIEAJwWiQEkcBDkvd1WTxgL//yU/56tfeoSsdSkU4BE1v2BYtYRhgrSFLMr7zt/+Bb3zjq2w2jjiKieKYPN+ilECrEGugLCuEM2RpgpIBcRSSxhFxEtM0lU/oKIGUAqsUOIkU3qccxZowkNw+O2KQhSwWOa8WS7/PLTcIHfLgwW1uri+4czj3JoZY40xOFGu64pIs1rgyJyk6RNcynR/g9AaV74jDgKBreHHznB988BSlQ/q2RjrHPFSEfe/3iUpxdwR3hh271z9Cowhe3hAPY/LlhrmGWTzjdCZQcsGtY0UY18xGhiRpGEQbwiBi1W05zmLKqoBYkw1HnL96jQgVtesoqxYrBHcOUoz0OoM5vYUKIoyzOClxQlCXDUoo4igg323prUMrxWa7AyexrmM4zPamF4fSktV68w8bzjJfMxql4BJ606EUKCMIw8CvJWxPECmCSKNahXVQ5hVt53ePxlY8ePAA6xzvv/8zoiBCaY3pfPbPIjg8OKTsLPlmQ9t3SCWoyoqmbWnbHpwjFZqu7QijCO2gKmpEmJDEEiHlp6sT8APn95mxf4L2HXqf07TWu5qk0GAsSoASEucMRJJMeatd39ZIQOuIAOutVoslwTTzbyohUFIQBoq+a2jLmmyWEQ0DVpXxCra11H1JbzuM8BbDyWTK8b1Trt9/z4fJhQQ8NUEi0EpgAaElSRx7k0KgOL51i4NbCWePNEcP5yTjB5hcsz6/4Ob8OSpUdM4yGg4wfc9sNqEst96KJqDtW4qiYJYk1GVDqDWxDiiKrXd6SelXR4Gm67zO4GxPFEf+IbI3jxhrEBKUClFSsy1y4sSvNjrjcEIyngz9SUoH/Mmf/hm/91/8HkEcklcFSiqCOKTalcymB7Sd4WQ+YLPbMIyhSjWtkaSRpqxXNNuSZntDbRXj8ZBBIAm2VxzEAZKeSMZkCrKyxZQLhAuZpGOCesNmfUE0SLn68DUnZyckg4y8KOiV5N7hAXmxorteYpUm6ix21VIsrhjNxog6oXv1jDhMiaTP1TZNQ0KGCBL6bU7dWbJBgkEQJAOPgYlBGUssJWIiKOoKh+CdN6YUu4rVKmc+HBGlCdkopiWgvf2Le6p+6XCOhhmxVkRhRG8NxjicMxwfzVkulkgBWmnKqiSMIjpjfG7ROq8ACcXNckNeFGyKCkSDDrT/MzQIaFZLNk1FJxwaRxSnNL0lkoJsMKDrOuI4Qo2UX3PoAKkVzjlCpQmjwAPByoo4DFESNusFo3QE1tDYCqwhCr1RvW16L7707afEAussoyxBKQk4lPEK8LqsAEvdWFrtFdZoD+Da7XYcHszBOSSCqq5QkcQWfg0ipPKDaTuEqLCiYTgYc3AyQyj2KRRPgFBSooDASX/vRKI0PHnzjEefDRjeVchxhbGKp6+e0qUHuGTITiraJGEQhZzcfUR3VXLz0+c8ePAGgRIY22Otoa4rv4NuWhw+GSOVZjQafpqEaZoGrdWnjqWuMxRFhRQS4Ty3yQt7Aq0DqrJCCs3lxQUnx4cMs5RdWZCmiWc8tS133rhP3dekYcJoNObl6wvSrifNMhAeRdP1Bik9UkZJCKX//bauUbYnUpK8aujbaL8TbtFRxjANfWRNmL3/VTDJMpK0p9lVe+KhY7UtCbXEtDV9W3F4covzVy8Zz0ZI409UVVOSmASkZJCOuHx1ThRGDMdD+s4iwpCy0vRGUuY7XLXFlTmhniJkxPr6gizVpIOMJIrRQtF0hqGzBIHm/vEpN7LmMM6wlPRdjt5ptIi4uqrgwcmvPpyDQYI1vY/9VDWogLpukLqgs5Y4STBCcbnaooKAPN/hjCUIQmosQikW1RVN22Jk4P2tzgeMhXDY3vgjmg5IpSKIYqyFW6dn3lCuNU1dk0QRgQTbdx6TEofUdUUUOGyTk6YpzpSEfU8UBag0wDYF+W5HFGlP0+t9xjIQgDUIZ1FB4AfU7VEoeOKBDjRKKwKt0NIf7dq+Zzqbslnc0DrnCX9Kkec7wijChD1N33rUSJQgZODXQq7F2DVG3yBViAgkTeON5EpJrPHeSoknziEdgYIvfOYR7359xuzJmmK4YdXXFLuc81VJv/gBrVEk6ZhwrHm5eUV5vWGz3jA+G3P30RNW712iNoZO/dy76fZXis4nfJz/vv+cSGhtT9e1e8ulA9dDY8FJmrrd76k9X6muK9q+81yoUJDnBVGcMh6N2O4K4jRjsV7x+S98lqqqya8X6CAiDMM9GTEgLxu2ecHx8SGb661nq+KIAkeSDnBS8/TiFU3TEceJTwUFATpKaNHIeEi+XnJ2ekBZdby+WPD4ZMh6ufBikq2wMuT2vfvUnUNIR9NZrhcrZBASxRnL7YbTsxMEcHOzAAfr9YbhdEpT1bRty9XViuF0ShDFHkPTtqRJhFaQ73KisCVWDtO2fPAzj5AdTzzE6+BgTl3XLK+v6ZuGrmo5PjnhxcUL7hy+Qdta2vUn/7A3Z9sZkiSldZa861kt14RhyKpe+B9CIzCbgtYAtsFJjQwUKgwRgQbnoO+JoxilvXlASIXSwX4UHFL62FUgHW3X0fc9wyRklCicMbSuRokaYSxCwmq9QYsBuinRLqRtG7JRSkOPbTrKxhGGkbfbOcet4yMOjw54+skzmrrdv7X9P80rgv6N/HPagTE9VliKpkIQEGlF1XukSl76RESxyxmOBpSlN953xtCElrY3RPEQKT2TtreC3hnK/pzSVjT9iOFkgooShDV0bYP7OeNICaTpCFXEwzcPuf+lBdXJlhfaUDaOunX0RJi05+b8HElAVW2YzQ7pTcf54hwrd+jxivddzTou+PIbv8364yVDlyCU9yGbtiVQkiSNML1ls95ycDjHZolnJGm/jlJK+u+FdajAPyTbznFwcOj5TBgCpRmkQwKlePbsKQdHx/SdIYz9Plor/wBbFhVBBEp4/7B2cHV5xWw+Iy8K2KNvWgv5rvB3PCPorWO5zonHCq1DlNY0vWN4eMarmxu0ldzkLcaFiOEhXRByVSiUdjx6dI8PPviQo1nGZDhis17QC8V6uSAdztk8y7FtheAKoQRJnLHZblgslqSDKfl2y62jQ45OIvKyJhaCfHsDvQMhGSURlTBkaYzZ0xNPjiYcnR1hTIutDGVe0lmLdf5zHmQpVnjH1fnFBcPhmMko+4Xz90uH8/VuxwAoysLL1GGIVRIpQ3pn0VqiQs1AZx4hIr2yKoTwdDflI1txFBGFgafaAcZY0jDE9o2PP9Vb4kgxCCVFnTOwgmfPnzMcZIyHGWmScH1zgxOSh/d9ZjANM9quIxwk9H3NfDZlu9n4HyYpkFHIemPZ5gVd37Ne74jjhCxLaKqaOE7YFQVV2fi9WtvtuUQhCEfX7g33UgMtBuc9lrJnMBjsq8ENSkjqrqETPVpGtDiSLMDZFmsNQjkauaHvK8quYvbkjH/8X/9XfPt//z/oLi+Q+xWOMxblwAjDF/7xDPHoPT42lqw5IowHOBydE+hM42SO6VvKYotxW5zsuHX6mLwIubl6SidygqOYn4g/ZRt2fP7Ob7L7Ucs8VFgt9gQJiZAwHA3o+w4pFVGUkMSxvwI4S55viGMP4r6pGuIoIokTwtDtIeINO9OjlSAIA1arJUEUkyYRRSG9pZOe48MZKEWgJBcXF7wxe8MbC7QmjCOmkwmr1QqMQwnF5eUlqAiEou17Uqk8i8oaxoMhTblGmBIVSFCauvaIVasTdLxEC/ibv3tBks7pyHh9c83BbARW0HY96BEXS0sgBpyOpwjbUxQ5STYFEfDixTWTyYC8tZRlzS7PMdYQhAqJJcsihDCEowE68AJpnMTEmeLq9SVnt0999rjrmEynWOMNJ0VeEGWaW0czFjcb8u2K8Be79/5/jO9pStdbwiAhigRBHNB1ds9E9fe/MPCRpCgKkQhsbyiLHc70hMoRKQNtTrkucHg7WhRG9KXF2QatNcVmw2A2xlrDMNHYvuD0eEJVNWjpl9NxGBAlCc52lMWW0WjEfD5FByGvz1/TiNqjK62/u7x8eU1eG6TccTAHpYP92sfnN7e7HVIrYh17yG8c03U51rk9IjOgR1LtckCCUFiLX5JnXqnWUlFVlacxoNE6xLY+2J3GPjxujMG2lkCGRFJwVf6Ak3dP+Wr1W/zlP/83tKsKrPTYDwWf/51j1Ml7bNw5o+AO1gZY5/OwxtTEWUSQ+PVGrASTQ0fZb1isnnLvzhf53t89J9CCkW1Z6Bfow5Tn8q+o72fcufdb1H/zishJlnu86ePHj9HaXxP6ruNqsyVQmtPbJ4wmU27df8ji6opkVyCEZLVckKQJgfCDMRwNWSxvCKOQ2cEBRVkRhgHXl9dEYYqS/uh/cnrKZr1mkGUcHx/tCf41Xd1QFRVlVZBECUdHh5R1xUcvXnO1dQil6E1HVcKXv/RFZJ1T9x1N12KsQNclu9zH366Wgr4x5EaTjGdkqebDT97j4f0Tyqql6y0GzSzV9H3JOBsilOLpi3Ou17Cs/Inu7sEQqUMur7csFiuGwyHbXc1ufYMEwuCYpvHH8myQMhqP/IM8COl3BbvVGktPGMdUZbEPYOBda9axWqx5+eIFb7/9GezVzS+cv18aGbNCkY0njGcz5vNDsnTA8cEB4zRlEGkGGs7mA0aqR+YrXmsnoAAAIABJREFUuuUF1dVLZLFhYBvcdkF+c0lf7hhGmhBLvlqzulnQNg0CRxIFnBzNOTk+IggC8l3BKBuhpSbUAVqH3Fzf4IxFOodwliRJqOuaui7ZbbaUeQFOMJ3OaJqevuuZzwe889Zt3nnnDZq6RAqPGqmq2qvCWiOVH66fe4G1DtBKf7rXKqqaum2Q0r9plFYEgcQZw2BvU5TS+3gDLbHGIaUkihLiKMFa7001tsF0gkjHVOVLlsV3uPPuCV/87a9isPTGK6KzO5LDd65JpjlJOGI2PkTpiKqqaRpvMo8iDbInHUUc3prStob5LObuG46X598FF9FVcHmZkwQJZb5hVbzm7luOH2z/b96b/hT9eIiIQg6P5yTpmCKvSdIhYRChtUIHgqau6JqG808+ZHN15WsYAn9MrcqSIt9RVTldUxMqTb7NfYh4H3L46te/QlHl7IqC0WhEFEqSJGYyHrNerHDWEwwXNzeslktmkxmd6WjbijSNOT48YL3ZIJViOBySxBE31+fYvqVpGqqu99G4MOLzn/88YRRg+4bxbMTs+JBNvgPb8+DBI7SKCaKY1XpFlg64urxChwGt67m6WdBaS9m1bNcbMtUThzVlvSIvNozmA9KBpiyWnN05IhuGtH1JmHiiY9v2rNZbttucpu1I0oy8KFFC0e3B5Nt9uFqHES/Pr6jqljTL+P4Pf8DFxS+27/1Sbu2ff/vffqurSoaJZhxrTLllc3lOjEF1Ndr2NLs10rTkqwV92/HqcsWL8xviUBMoCEPl1xh9z/HhAWkS0jYlk1FGmvqA7GA0Yr3esVpvCcOEosgJI4lzdp/za4mikN1uhzGW0dBXL+x2O6T08bAwCj0jVkpUEPgBbuq98thT17U/PjpLmmbeCoij7w049qpw+ulO1gB5XhJGic8gKsXp0QzbFMRRjOkNZVkShtojQweKW79zm871DNIEIRxt35CEAWHoTexJcMg4e4QwCU3dEA8lzaZgc71l9iDiye8aTt7KGWQjkuiIfGeIoohQx9RNhw48WcI5QV20BIHCdCFlUaCiivFMsV4UTIYn1HWLEz3FrqEuOwajHktBaa5Zmkuq0HBn+hahNZRtSd92xIn3LEut6dt+r277t5uUeAELODo6xjq3h3tLgkBz9+49rDXsdlsuXp/TNTVRHDGdHvHoyVssVgtWNwu2G0+4mM097DpJIsI4REpFXhboQCNwZMMhl6uSq01FbyxtXTOIA0Tbertm71Xot5884emzl4xGI+7cvsPF5SWz8YRhqgm14/TWES9fPGe33TCZTDg4OmIwOeIHf/9jvvDZJ6SJB4EHumeU1Nw+iJiPUtqq4fr6kiSOWd6sSaOMyXhCEGmM86JnVbdIqRiOR0ihqJuOxWpN3xs603NwMKdqGpyD+cGMi/Nz/vqv38N0PW8+vs8nH3+EVpLHn/3Krw6V/pN//fvfqlY3UBXcPplD39D3LaNRRhxqtPBMHq28yCL26ZMsjTg+GPPGPc9f2eU5s9kEKSymqwkDyWw8ZDAY0LYN6+WauvFujLrxvtjBIKOpW+qmBuDJW2/5jhPr/KU9HdD3rfd/NvU+QqbQQUCSpqxWay4uLzHW0jaepJckyd7t1BCEAdb9R+fJz+NlQaDpjKGqW0ajIb0xfocnFZiGLNjXMOA8BMxZdKhxGZz89h2UilDKUuxyelcR6YBQemzGIJ3S1kd8+NFTPnn+HXq1ZTqfgqx583cUd78UESdjsCGBmjDIbuFshLGCm8WCMI6o+5okCqhz5R98RnBxvsMaR5Q1zA5Cljdb2soHed988zHLxRKhIUtCwkigBx3BpKNRBf2lYhxPvCrfes5QmqZ704bHoY6GI4bDEVIEICzpcOTTRvgluzE9Qaiom5rZ/ABrHaPJBGcsSZLRNA1pktDUFdnAc4IHg4z1es1sNuXZ04959fIVo+GQ2WRCGEQUteGHH77icl2gAu3XeqEgjTSnp7cYj0cEgd6XVGmE86Dy1WKB7VsO52NWqxuM7Tg7PWU0GpJmGZdXV1xeLjg9OaJYvGa5uKTOa/qu4Xg2Is40u11FlKaMZ6dstpaDk1s0XUXXtWAFUijPmLKOKE7onfeHR2HIo8cPKKsKpQK6rmeYZdRNhbHGF01FipOjEbvNguFgSmsEj9754q8+nH/2f/3+t1zXMhnPKMuaxc2CvrNsVju6xkvvQagJwwjnJLttSddWnB1PcbZjl29x1nH77JRAeff7YDhgNBrR946PP37qUYxKcnJyRBxr0lRhTMf8cEJRFKzXG46Pj7m+vt4TGXrmszm73Zaub1E6BKCqGrbbHX1v6Hvj733ZiDAMvQAihY9A9Z7/6tm0Yl9A1O2hWg7nHFJ6dm3XNfTG0vSWvGoYpzGh7UiTiCzL/NdYn3DJbo+ZfGWG7aVfQzjQuiUUAWEQIYQl1gPqIuZ7//6nvPrwiq+/83V+73f/CbffUSzU3yF1TxTPgSHOpoThgNZ4BvB6u8IKSzoYIIXl5nXNZDyhbUuUjFhfG8qiIpt2NGaN7SOETdksdggHVvRstmt2m5bV9ZKOS1xSEs4T7CJENZAkKVmaeoxG19LusTJ1VVOVDW3bczj3iZLlzQ1KQpYklLm/GoRRyOuLa+qu5403HvPd732XOI4YjwcU+Y6ua6nrmrt3b3/KNM7znNPTM6bTiceQxiHXNyu+++MPqUiZHp8xno7o25qxgrPJiCQIiIRgtVxwfOuYXZ7TdBXWtNw6PaG3hulsCsKRZRkvnj2jqipWmw3OWWIlWVxec3R0TN9bnOh588l9qrpgkE3RShHGAokiigMGqeNkPme13HJxcUEQeIh3nMTIIAAp2e18H0vZlIRRys3lFaFWTCYTbp2doqRkNBzg+orDwzlXl0vOL1/zjV/7OoPZ2a8+nP/r//Q/fMsKQV4XYC2h0iRJRBBGJHHIaOiZm8YaL95oycNHd7wDJY7J0hjbd4SBJo5iloslzvR7yviWg8MZSFBas93tqGuvor733nskcYLp/P0S4XxesGkRUrLLc16+Oufg4Iiu85ftMAw9JS0beENE1xFHoUd/mB5rzN4YLTzNfY/B1EHg/bUInBD0ndmra5Y0TXBWkDcdTob0dcnt2eA/AqetpWkrdvmO+NaA8ecmBHHqmT3Fa0LVEusRYRCjAgMWQjUhDm4RxzFnh0f89d/8azbix8iJQ7gRg/AAqVJ0mBHo2NseXU9VFT5hk0VIpaiLlt2i3N+ZFV2lqTchVdHTGRhMQrCCfONwxiCVQaoeiaZrDEenEeO5JJ3Aweg29iZiNprSdy27/e5WSUVbNwyGAw4PD5DSeSavtXseb+yBbFXBdDZjsVoShZosTdisVtw6Oeb46JC6rnjx7BmzgyOK3APB4jji5uaGw8Mj6rqm2oclBsMRbdOxKht+/PElVimqfMc4DplFIdvNBqk188NDxpMpl4s1f/YXf0cYptw7OyXfbRkNBpyfn9M2LW1d8+bjxzgr+dl7T5kOZ3zzm7/F9dU5Wlruv/GAq6sV49GEvvdkv970PH3+kvv3bxMGEhVowszXM5zcOkHogNW2oio6bh0f0TclAsfR6S3veVYhw1TzmXff5eb6CgesVzvWmzV15fjBD37M2d17fO1rv87Hn3zI2YN3fvXh/OM//D+/1ZuWg8mYUGuapmI4SKirgqapCOMIb6pxHB4eMBhmLBYLVqstu11JHIYM9m1NTdOQJKn/QPfg5qZrmUwmJElKWVbkeeFLf8KILBugA0nX9YBjs9mRJBlhGJOmGQ8fPmS1Wn3aZ6KUIssyyqL09jEp9qJRRZL4h4hSCqk8U+jw8JAoimi7jq7ztRACvDdYeNU53dPetNI0Fqp8x72TOeBYLBZEsVeAR+MB/cBx9+u/jpCHFOUFRXXFMB2SRgdkcQjWYo0l1Bkn83cpumvOX36bLzx5m1/7wj8lVBmVaRA6BgK0itA6wJgOISxlXtC3HUkWY2SPViGvnl2DVQRaokSAaTVNqVAio6y3HN8Z+re49cffKAoJI4ntvRbQtDkiKHlxfs5p+gS7cxRlTt3U9J0h2DesKaVwbm8zVBKLZbvdUrcN84NDlPQlU0JK/zB1Pm00noz55ONPwMHp2RmjyZT5wQH/7J/9z7z99hOEELRt46n1YUSaJOzyNcPBiGVp+Kvvvw977vGDwwmfu3+bJFKMBhmb9RJnDCdHB4xHQ46Pj+jbgqbI6TAI0TPKJrz7xXd5+cJ3kp6czEgzzbMXT7l35x4SwQ9/9Pcs12u+9KV3ubi48Ob/MGA0Hnobad0S6ZiL19fAgOtlzu179zm+dcxycUUUecRrlnmMTpnn7BaXSGsodjuWNyvefPtt/tW/+gPeePCAfHeNlCFHx8f8y9//I+7eOeLs4Wd+9a6U48mI65uK8WDAq/NLprMRxjQ8eHCfqu7YbDc+I6kctm/RWnE4G7MrS7QMiCNNkiQYY/xRaLnGmP7TTs226XzVQ+coihKtIy4uroiiwIstkabvW3Z5TRimWOPY1b4UdrFYMBqOKMrCd1a2LXme03c99Z459HMx6MWLF6RpwtvvvE1Vd7z/3gckcYa1PavFAq0D5tMpRVVj+pZ33nzCe+9/QNW0xLFGGlg3FUkSYYTF9u2ngymlpOscTVnRtSGjdEJRGrJkQpzeARlT1QtCHQEli+K7DKKcNBzzZ9/vKS++w/LmQ161zwhmMQcHh7S2o2pz2k4ShBrXCTCeYjA9nJFXJZEeILXCNIZsOuZ8tYZAglMoGzKJ3uLq/DW//luP+Jtv/4S6sMSpJh4HrJc5y5UhqGuEbbhz7z5JW5Nfd3TGMRlPkBbAMZpPqYuCuqrYbDY457zxomkYzw5ZrTZUxZbNZoPUan/31zgHn3z8lOPjW7j93b5vCzarLb/2ta9RlyXD4Yiu8zWMu92O4+M56eCAtnY45xNCGIMVGgkkUYBzGYMkYT5K2Wy3vHz+AXfu3ONmtaDqHa7tOZyfUOQ5URTz+//iX/DWm485PJrx4QfPKRvB2595jBWW9XbFk7ce8/zFC8JQs1wu+PJXvsJuvWK33tHWhuurBaNsx+QgpawXHM2GxLqhyHcczQdMxhlNU9PUDXVdMZtOuexaiBJm0wGvXn6EsTm/8Y0v0m5vuH10yuyzQ1693vLNb75FU/9iEsIv7+f85//Lt4TUbNZrzk5v0XUNk+mcsqrYrlekqeeljkYDtJYslkuqpvEFqMKh9mfxvjdorbB9R5IkHBwfY8GT7pxDKs1yuUYqjQoCoiT2naAoT1iIQvI89y3UDtI0QSB48eKcZM/H3W59Uj4IQqazCVEcAY4oDMmyDGMMRV6xWG4oq96DyuJwX7Tki4vapiZLM25ubpBSkMYRpvf+z8oKemM4HkXeCyt9nUPfd0RhyMlbd1CPDnlw+pCPXvwHJpMxs8EU7A5B7z3BEpyrUbLncHaPd9/+XZqy4fjwDr/9jX9Kb3q2zQ06UjRNvW/4dpjOIoxgt8sZjEd+TYMkUgnltuLi8oo7909YrW44mB4ghGCzKSh3htcvL8mGIWC8WDdSdI1hMJUMRp47ZG3P8rwiq4+YZGO6tkMrh3MWrZQnFHbdpw1rcRyTFzuiPUUvSmKiOKSuGq6vb+hNz3K14M0nDxESdpstcZiw264IAs3tu7epqopBltA2LVVZIKSgqhuQAQbJ93/8IR++WjKcTDg8mPLw5JDvfedHvDy/5vr6ijhOODo+QkvJdr1hNplxfHCA1orF1dW+PFjy5jvv8Oz5M+bTIW3TcPvshI8/+imH8xlVsaPtena7irKoGA/HGCv46INPmE7GRMOUtz73GZSEm8UNYeQ7WcqmodgVdG3HZr2grCt0GCKVwhrD2Z37PHr3G/zwhx8yP5yTJjHvvfchQsacnM745OlTimJFHKW0fc+dx/8AQehv/+yPv7XZrjk5Pma329C2NVGcoJTyO6rFDWEYMJtOWa3XfmXivMlgMhmz2+W0naU3DuMsSiustaw3G+qmJhtkdH1HVZUEQcDJyRHj8YDVarEnxft1SBiHJEnmQ9XW8O677/Lq1XMev/mAvKhYb/wKJo5jf18UfjUihK/L8x2fwZ7LA5GWJKE/Whtn6buWPN/h0Yse5PVzD6npOtI0ZZ3X9MZyPB2ihU/CDAcjmqbDApeba578o7f46OPvIhODkgFKSNp6R+96lA5wtvt5bo1QDzmavs1qZfm3/8+/4fnzv6YJLumlAecfBJ4i4Y/hfevY7UruP3hAXuQURU4cZiwutwR76Fhd1bR9j3EtQQJSWkZTyXBi2OZLhHDewqgM43EA0qCUZrVd0lWWU/0Y0WjyfEtdlb4X1VoUgiLPCYOAQZZhbE+SxPhsuGN+cMDBwZy+N3z88cccH3uMyavz12Rp5sHjjf+MR6MRo/GItm2I44g4DpnNppRlQS9i5qdv8/TqitfLkvc+uSRNE+bjIQdpjLKOe3fPePLmfY4O55i2Id9sGAz96uoP/+APiKO9B7jruHh5Tt1sePX0ClBI1TIeDrl1cka+KxAuQNAzzGJM1zAcpGglWCwWTA/nnN65y0c/fZ9d0QGKONSkmQ+/P//4E6Io5OGjh9y+e5vOdMRpjNQBl69fs76+5PzZa7aLJYNBRBJHRElM33sU7BtvvMFyuUQjOX30hV99OP/oX/5v3xqNRmw3a05OTlitl4zHk/2Sf8ZwOEQKSd3UvH59TpZmJGlKEARsN7t9wsHjSfJih7GG0WTMnbPb1GVF2zZsd1tM2zOfTSnynPV6w2CQ8eTNt1htNoyHA3CG+/fvc319RRRFnL96TV23nyquFxdX5LuS+cGUQGufmFGSOE4YDHyESUqJsR3O9r6NWSr61uyTKF7gAcA6mrrxP4R7O59SEqcjqsYwiiN++qOfcPvOGWWVAx6Pol3A6VdvU4mcQEuSaEwQBmw2KwLt7yXWNGilUCpAio6AiFuHb/DknQfceesMkbQU5Q0oQRB6UUtrTRRFmNawXq1JByFFuUNpAVKxuFgBiuvLS45vHdK0JenQEcQ1xlUIepyFJNMEYYt1HXFmcapEiI6y3mFdwzg45MQ+QtW+KXw6mTEZj2mahs1uS5amSCkpyp0X1/Yq8nQ6YbPd8OrlK4RSzOeHrNcr7t+/Sxh40SzPCyaTEcZY8tzfaV88f0FZVEipqJuWtm4JhKWtN7x8/prv/fhjVmVLlqUkWvD5B2ccj0KkM0Q6QEpNnrdUVYexsFxtefszn0GHijiJ2G62lGVOEA55+PAOfdfw/gcv9gC5DZgO2zfkxY66KgiDhI8/ec7R4R3KskYHiquLS27WLR8+K9nlHWkIWRyy2S6ZT8dEcUg8SKmLgqYocbbH9o7xYEpfrgkiw3yakaUJSZBxeDDFdLUX1EZjlosFcRRx9OCzv/pw/vkf/+G3qrLwxTx9SxCFGNNT1SW7fEdd+SW/sw4hFEIq6v1RTwjHcOAHq65K0iTi1skRprP87P2PaLserSRpnDKdzUAqyqomTVPqqmG9XtP1LdvdhuFgzPMXL3j85DGb9Zq6LlFaIqRisdoShCFJEuxLhbp98j8iihKeP79kW/RMZ3MQnm+b5znWeOiXktL7Y6uGMIhQge97bFqfgG8NfOVrX+Vnz5/SG8dsMODB3VPatt47ihTFZue7OE8mRCcDEAGdaTC2o+0alA4+BWn/vAhKKUFn1/RmxScf/T0/+/4POcmOCNOAvFl7Ur31vZ5CCKqyYb3akQ5iOlOSxhFlXaO1xlQ1QjkQLU60DIcZw7Fju10xGg1IMkgGkKYRVhQEUUuc+l6ZpqnBKXQ/Ytw/ZPu6JJReWa57zzjKSx85cziSbMhmVyKxvphXayyG23fv780fmd8fW8dw7L2nt05PWaw3e9/y/m07n5PEIf0ebi1QmM4ym6QsdyV//O9+QG99yfDpfMyTWweAhCRhs17z4YcfogPFZrNmmIUMhilP3nyTpm24uHjJ0fEZKsr47DtPuHj1jCCKeHD/HrHWDFIfdv7pTz/h7p0TTO+NpTjJ9c0Vw+kALTQXl0vqxvCFd+4xGQiEsBjTkcbh3rCvub6+osy3tE3LYDCit46b5ZrJNKNvW+LxgKvrG4bZgDDStHXO6aNHfPThx5imwwnHrX/Im/NP/+j3vxVHIVJKbp0cs1isUEpyeXHBbDbDGM+D7fZ1eeH+DlnVJVEUU3eGq8WWy+ucpnP7Zuee4+NDZrMxbz5+iHOW9f7NvNluePrJU9I0JY5j4ihms93QNh1B6Bu2rDXoICSKY8IgxBjLeDggSRLS1DtvoiikbVqWi7WHUSvNzWqN7TvapiIMQyaTsRc22p7dziu8WZbSNDUCi+k7wjCgLAsuX78ijFLqzvKFz77F6uqlx5nogKaHTgtwAa+ul7zx65+harY+zooEAqIoJYolzu7NEsJHxYRzJHHKnVufZ3W94WbxE/rohl62n7aNOWcJwgQnSqQNKbcdh4djlqsFQSgReyKDpSWMBMNRRBAIqqpkMArIhoIw6VBxw666ZjqPiRJBbzuiKGaxWFDVDZtlwL//C8tffW/N//vjV/z5997no5c3lC5k1zim8xOuFxtGkxkHR7d4frllNBrjupI7d+7w6uXlp+kMITzj5+L1a9IkYXlzg3T205Jb0/sGtiiOcMaipUQKh9LKc3Ovt5wvO2QUE8cR949nnE0ykihkMsoYD2Lm0zFpEnJwMMNaSzYY0PcNbd3x+PEjoijENBXXN1cgBeuba9Iw4fz8kulkyGQy4OzsFi9eXHJzsyIIJdPJmJevX3Hv3l2c63n41kPK7Yqry3MmkyGj0YBhNqDfd6U2+74aKQUI551vkwFhCFkS4Jw36p+d3Nob6AtenedUu4KTWzPKsmA8njG7/Q9otnbOiwht13B+fs7BbMZ6veHRg/vs8i3T+RHWQNv6lcjh4SH/7i//kuFowM1iwyDLGKUR40FKmsVk+wawuu5YLFas12vatiMbDnn58pzZdMZ8MmG93dL1DXnekKZDxsMRebnjjYf3efXqkrKoaRrDbpd/+uHM53PyPCdJErrO19k5HG27497tu2it+fiT55RlTxA4lCw867bvCQLNbpdTlrmvAex7RsMhRVnyxsOHPH/+lLZufX/Gi3N2haXvLS+vL5BJhnDGVw4u4Us3OW1giaOMrusYDkf0rmazrUlDSekaetkRi4QoVpTtBhUu+dyXfgPpvsqzm7/i6eI7GCpm8wlKGqryCoRFxzXLmw3lLmG3bJgfR/5qUOzQgSRJAozbkQyGiLbm6CSjKNZkWYwI8IVI1rFeFaggYrnc0fcgZUDbxah4zvGdIaESLFcrXhc1T//2PcqyYH78HCEl059cM5tO2W23pB8tefbBT2nzNb/5tS9xfHTIm2/c5XA85vz5J5zce5OLy9e4vudgPiPfbfYURL/+qsqKoigpq5LZfEpdVMwPDwmuGl68fM3w8Ih6s4b+kCiQ7DYrNotr0iykqzvmB0fcbAtM31GUOVXrcZjDccri6oq7Z7d5df4CGQ1w2Rmv85bxLKGsa8I440c/+DHDQch4PObO7VOeP3vGl77wefq6JQoDnn/ylM40DEeDT+ehqta0Vc1mk4NUTKdjdCAYjQe0TY1rW9I4oG17kuEIJQM++ugDz+JyMZ/9zOf42U9+zDKKiYcHLFZrHv6C+RM/7wH5T/36H//7/9ZdXb3GWkOel9RVyWx+wOHhAV/+8hf5k29/m9fn19y+fYZzPXfv3WOz2bJabnl9fknXNhzOPWDKWd+BGUUxXW8oipKmrXBW0BmBMZ6zivXxmqbvsAayQUrXm087RpqqZjCc8vT5cz73uXf4+OlzdpstB7Op38vhUzMePuYrGrqu2/ecGIqy8nu7vfATJwlFWYHw/lkhHMFedTMO6s7QGMHlusBISV6VxIEvoJFS+hazMPAGgzDgwT8648F/dhcXS9JQEgxSktCw290QhpJxJBjGGukUoYyJI8/wiZhx89E1t4Zjfrb8PtfVc2YHU9CGtq0ZZmOMaXn96ordpmE4GlJWG3rrOH+5YDAQpIlFBTXp0DIY+Tt2VVWkgwwnNHUJ203BZDwmL0oMkiiSOKe5/ugWz//iAW/ee8xgOPZOqz15oihyytIXMlljkdKR5xXj6ZDzVxeYtiUdJDhjqcqStm2ZTqY0neF6cUm5XlAs1xzPhnzzt36d3/jGFxiGkCWal89fMMq8vfHV62vOlwXf+eCClRnyuS99lQ9/8h0ejCRffXBK3/XIIEUpjZaW4XTOZpczGSa0bU2YpHzywUfM58fs1muGQ8UgG7Jc7+iEpHeCti757Dtv8/GHH/KlL3+e85cvqYsaGYbURU2UJnRNx73HD3j67BnvvPMWV5cbfvaT95nPhgxHKZPxmLZtWa1WCCUYjocEzlHXJSIIUSJkm1fEASgBZVuTjoaMBiMWl1tGo4yb5Q1pOmCYDXn8m//5f7Iw5ZcO53/33/yXrjf+Uu6cb7sKgoC6rnj48A3e/+lPfZ28sBjbk6YjRpPhp3V6Al9Ma633PdZV41EX0v9dfd/zk5/8jMdPnvgsZaBo6ppdWRCFCcPhEB1ImrZnuVxhjGUyGWH3rNttviHfVQgBB/MZ15fXKCXIssE+aeL/b8b4Ord0X24TBAHWGByCum1oOkNZeSiyEBAnCXnd8+pmgdnjVqzz91OBIwpjhKddY60hkAolHcejGck44Ph3jwkfzRDCEmYhwyygqdYo7ZhkMEk0prHEOuNgckRrGhwwi27zyd//jBerH1GJDePpBB319H1HHMQIadis1pTbFie8r7RuKuqmYTysmM0FMrDUTUleFIwnQ8rK0TWa5Xr3qWe5bQ3D8YjxIP3/OHuzH8u2+77vs9baa8/7jHVq6rn7TuTlcEVK0ERIlmwLhiw5QIIgBgwECRAgMIL8BXkhkMc8BspbEsB+8YMDJLFhCLAlUaKGajVHAAAgAElEQVQoi6ZIkZfDvbdv3x6qu2uuM+15WjsP67CTAJYAsoFGF7qB6lOnau211u/3/X0+LNcrgmBK+vhLfPpH8Btfus9bxzPyrCYMLWk/7QYG5REmY04vlxSNYW/vkCxbkxcVl8uU+WLG8+cvAMlkOmW5XlNVGVJKJuMp61WKGQzbdYryXaosxw8iqqLE1YK+LamKnG2ak8z2+a3f+X2CKOb0xRNuTwMmLmy3KR8+ecbFekO2TdleX/MPf+s3eXh7n+lsgTtbIOqMq5cnOEpx/9Y+XVujlF2YUTLm5fMTOywdRkDNfDahKmre/cWv8N0//xbHt28jB8Vnnz1lbzanaiuq3bTJYn9OHAb0XUO2WVNkOdp1yPINdd3heCPG8/t8+/tP6estD44cHj68Rdd1bNOMwQzUVcV4kjAIg0EyHU95+Ev/yc++OL/+3/+ToWkrsizl4aO3+eijT7hz5xbz6YQiK3j27DnjyYj9/T2yLOPi/BrPc/E86+j8aVDd9wLLqm1aq2xQir431LUFFYudi/Kn1Hbr22yQUtq/U4LxeMZqteLRowdcXFyx2TklbejdYTD2ntO3u6N43RBFMXVdY4x9EDCA2O12wwBlZ6Nu3TCwzApWaYZQjlU3DBZaDCAHdgZthed5+EoxmI5JbE3fN9fXxGFAECiKPKMaCb703/4dVKDwQkHb5QSeA7Il8jscCmLPIw5CXB0AEkc6OHKESEc8ffIR1/kTwpkkmSo8T3F1dcN0bMPi25uMxfyYbXHFyemPeeudI+KwI9umFMVANJ7Qti1Z2lCWhrruODreZ7Ve4wUefd8APZ52GM0WsLrLD/7NnCGNeDgV/MoX71DWgOmZjUI2mxWL2YS2qSjKgqPju1wsN3i+wfV8fvj4NZ9/7z2QEj+e8PzVOV40oipblqtr2kFyfPcR3/jGn/POW29ztVkjHU1ZV6TbnCCI6PqOdL2hLDL25vNdhT9nm9lRrN70CAxKWBgZAnoBSZTguYrjW7etf6bIuD674L/8J/8Ytpf0dUGZFyTTOd3gMBqNOTs5QXQF7751C1P3nLx4wXx/D4ng9OKCg+OHfPYq5fatA0JVEAQeTddg+oGbqxvm8zGLxYT16grHEbw+vWA+P2KdNSyLiItVxru3IvKrp/zd3/v7pMsVJyevaLqWzXaFkIL9+R6+56HDkC/9xn/xH12cf7vISAmUUQSuz831Ncf7e1RZxrPlCuUqvvabX+PJp0+4Wa4p84zJZLQDRTmMkxFtGHNxcb4LkyvL2ukb2romTUtmiwOQsFxeMZvt0XU2KrZerdlb7HFxcU7gacIw5ObqGu0HaK3fsGhsD9NiPrwgpKoadOQwiAGhwI99Bgmb1Za6aUBIBunQDIKsabnapHQGK1DVCun6MBiEUBYgrS3Ma+gNe9Mpe3HA8uqcg/mY5fIG3drA9/XpCYvPvUdb1gRuQFsW6FKTtjVikGzLAjkb4WpBUTa4akArg9cblNO/sVi7osab5Oztj8hfxwS+QpKxWW1oqoKrNkNJies5VO0VWfmCd95ZUNUb8iFkm9m7cHGxRqiIzbpBa4cgdMmKgiCMiOOAuslYL1Nme3sE7ZzNszsMm5a6LelNwDYrCIKAO/cecvv4kNPnTzl99QrlQDIeU7Ut8WxGnWc0Zc801FTplU3LXK95Z5GQpmfoyONX37pLXbVcXL3gv/6dzyGFS2UiusGhlg4XyxzTgxe4tC00bcv1aksrNLfv3me9SdluUhAKqRV1WVBXGXllMa1RGFLXtkOgTI+azBhLzf/+L/9vdOCglU1/BZuOy4sLVssNh7MZJ59+xBeevYVjaj73hffw9o+5PDvFCcdsq95SJ/uaoik4v75COXYEcjKJ0FqSVwVtZyiKltCP2SyvrA5DrDgIeqJYkug5xWbNR58+ZTKaM50pothnPB5zc3VJEFob9t/0628nvqcZyg0pa8N0ZsMBQhj8yMfzQr7/1z+wGncgTiYW6Nx2KGX7R55vc7BdZ9EeVV2jdUTaVfSi5uL8As9zmM9nDKbDkYbpZEJT5WBaxqNoh4+E2WzGxdUNn376lKFvuXf3LmdnFzhOSBiGFmMyGOoWkNJ6HrM1Vdtzs8nJ6gYjJUJpemNDEUZIpCNxlSLwXeLAp+86HCFxlSQJPBLfZbPd4DqGcn3FJNTkWYqS0h6PjeHgYP//N7TtCcWT/+svuf21L5AjkI5GDA5COFRVRasMyhnwhx7V19ABqgdjaFXD4b1DVts12eYZSkHoBwymQQDr9ZZBtJRFihGGTdpzcZ4zTjyMCakq27oZho6jowPOz8/xg4iyKtlfJFxeXSOlxA99slPBrHkH92pO1ZzgBwFdb/uiAkNXZTz+8Yc0dYkQhrqqicKQIs8xosBxBGYwHM4njOKxpWH0NcW2IEnGnJ9fEXghm80N2nUpsxuOju4wlhYLI6jZX9j3++r61PJ5w4QvTCNeX6xoNs849kd86c49xtM5bhyxKVvqFtabDIOglwqzSxc1ua2Sb9ZrC6QzHVW++155AW7Z8Yuf/yJaa0ScUGlNlqasnq/419/5lDi2xrC2rjmczzBVjjC9NVoPPV9+/3M40qWuO0xR0pUFWkmauiYMQlAu/kgj+o75bI+XF9fk+Zp33r7H8ydPOdx/QBAottuMqqzJUxth/LkW53Q2ZZXX3Hn4iO3yAiEtg8VD4Loel1dbRklI6DtcX1+TFTVKwmx6bGNRNxswrV18g4Ghpyga8rwkCkPCwLc6wRsbBQyCgCzLGI1GtG1HWdRMZ1OKvKDpbOhgmxWYpubFyWuu1+nuyDvhcp0RRjFN2/Hi7BQcn06K3XFIoHWA7wfUdckwNLhKopVrv1FmIHIcVNeQaEVf18i+J0tv8GZTurLk08dP+Py7bzGIAWNsA77rOnzf5/j4mOVySeAHpFnKeG/GF/ff44d/+GMmX3uA93aCFK4tWPmGrskYDLw5N++US2boEU7HVf4xD9+9x8uTAsffkGVrRD9wc9Xy5PGKMA5w3ADpKLrQR3SSIh+I4gBDT+QHGLMr2ozHaNehThuqQjJ0mmjmobqAUfN59PItsmaF9j2EEhhaENam9vrFM7J0y/7+ggFDGCdU3YAbRvz5t/6Cd99+mzgMkK7mxavX3L17m5vVhqIoOFB7eJ7P5eUVSZJQVhl37tzn3/3Jn/H3/u7f4+mzzzg8PGA0Snj25Amj8RStfXzPRSrDrcOIuhvo+xRxeUl7MxBNZ4x62KzWHEnFumyJ5gfs336IfzTm4xcFYTzhSZEzOjzGcXxOr2+IZ3OOF3usN0tqY/jRj37IV37tN7m+vLKBDyG5907P++9/jo9/+CHr9QrP88mygsjXrFc3dHXJj15f8+1PX9I0LY6SaAacvmWaeNw63GMmA77xrb/kq19+l0kt+KvvPuad2xOODuakWc6Tp08IkzF9N/Dg0Xt88vGnnF1ueO83fo7FuVzf4IYj/sE/+Pv8m3/1f1KmMIoT+r6HvmeUhFRlgels0zkOfRjsLOVyW1o9QKQ5PT8j9P03js8wjFFSYvqOsiyZjEdUdUXbtLRdtyMhxNQt1iHZNcSjCdfbnKvLG24fH9Hg0bjw0dNnSLW00b1VacPuxkENAgeFpzXuzg0qHYsQGToPjcBzFNJ0JIGmzLb4WhN6Ht2OK9v5Lvv7+4R5CsLguRozVPSmI4ojGKDrO26ur1DSoW0siWGoGv7DD75PoFxWf/WSB3e+Qt/ZAZE0LQlch763FWjTDvhuiGLA9B1DL3G1y6p6QjTzWK9SuibDkQnf+dZHlFuPKHYYTX20X5O7Fa7nMmBsiiUe44Uu6XrX45WSvpf4yqPvW6bzEFdOWPA+QfYWwgm5ufmMx58+ZTaJOXq4hxCG7XqDrzXj0ZSm7hmPJzS9JfzfOjzk93/vdymykrOzU5RS3Lt/jzwvEFJw6/YtfN/j9auP+eCDD/jjP/5jtNb0HezP98EYbt8+pMgqPru6QSpL1lMaHEdzfnGOEAOeF9g50rJkmoysoLjObbrKDVFVCdWS6qJB+R4P6GnXL/icVyGrFUe37nDPk/zVx9/l3bd+A28ask5b3vvVD+gclyd9j/Y9bt25R9FBUeZ8+Zd/nfVqiRSK5WqD7AXj+YY832D6Bp1uMWagLnMEA+PRGDMYnmxbuusbuskt/uTjS5S6ZugFNxc1wXZJkxWMInj98mO6uubB/Tt88IUv0Db1z7dzlmXFarnln/1v/ytHRwfUUuK6rp1AkJI41oxGEzCGNCvpjNmhIhscOeBqh21a4mpJFNgEvxmso7JvW6u8kxLTW55M17VILLXvq1/9Cn/57e9Q5gXG0Xx8csarqw1ZOfCT731CsKsgS+milLM7ogrbrvGtucn3PeazMfQtgfYRpkWYlq6saKoKuXNipOsVebZlSEb4vmdFqbVtNL94+YrxJEFKRZ7VGNMilMLzQy4vzxmNEoIgoMhtuOGnraLry2uOD49Iz665+MFzRBazf2+OoxVNB7ozVG1PMzS42p4ghFJIqRgQeF6MH9s86bOLj/nhh+d0tZ3v7BpNvpGIzCEaByhpEKrHEYJsW9hhAunS1A5lbXEmj966xc31JfW655b6Cm95X6P0My6zDKkctOPgCIVWGt8LQTqEvh0qH4aBvh9wXZ+ub1mvblCOw+XVOfPFgq6uiOOI09NT+t5QFi4vT04IAn/XIhtzeHhIUZZcXl6g9ZdZvl5R1z1BFOH7nqUiVjX7+/sYM1BVJaY3FNmN5RVHMWlWMGD1BxeXlziOJNCSvi6pe8Fmu2Q8SZhOfbK0oM0v8TrDW0lBe25HwwYDXVUTxQHv0BMHe1x9+pT1pmV/ccxheJvjiWZTNcSzAOW4FBMPoW7b13+zopcKBDZv7Gq0I3GkQCuJpz2MMVxcnNF31j63Wi1xowle4LEIYjablJOy5Ponn1EWBf/5P/45FqfWHnvTPYqi4PXr1/ja5eDggOVySVlV1iPStwS+T9PUbzTsSjks9kZsNhm11rhakOU5QuR4Xsh6k+L7nrWHCUHT9gjHIatsZTVvOv7su9/nYpWyLRvq3pB3hry2lVUdRkhH4yhJFIaAIQx8fK3tiFXfo4VgbzriYD7js8cfI6qSoe8YpMFzNDrwqOuaJPSgr9CjEVp7NlM79JagYAYuLtfkRYtWDmDQrsurV+e4Wu98nz1ZltHUNnEjhD3y7y8W3CxXSDVw+pcfk1QTbr01QzgjqnpD3fZQNrhOvzNyS8Qw0A8Kho7AizGm42D/bR5/r2d9ZekGRZ5h2hIxKLQnaUos/FoL+qZHOwG+M6YsC8q6xY8CHNfw+vQlgUmY3bxHt55z4j4j9GPSdYoarPZB+65VbSiBVnZGtu/7NxZx4Ti0pSUYwMDx4QFRMuLy4pxRHHP/3l1Wqw2j0YgwDFivl6xWS8bjMUKIXaqsZ7lacnR0izyvuLq+wRjrjBmPx2y2Kdp1dx4ZuHvnNkIM+J4lXrRdQzKOOTp0WK1WSEfg+QGBF/P85NmOyOCglMXNFHnGKHC5ubhgGMD3A9bphq5vcaTGFRX+sOH+JKDePGVZvebW0R7xMDAom/sdlOLiek12k3J2siI8vM/x3XvE/tT2dMcJcRjiOC5hGHBxs2Sy2KfqekzbIoNzDLbdRz+wXq+pm4bxZERVFj/fzulq/SYL6WqbR/zss8/Y31+w2W5p256mbtDKIQh8/NA2otO8oK4bi+eQw24g24bfzQBBaNsHm9y6JNpe0DBwsynJ6xqpA6rNkvU2ZRhAOZLBDPRdgx+EDK1hMZ0iHMFsElHkOaF2OZhPEMZQlwWi7xDZmmKoOZiO6PqeIi/eBNn7rkcODl1Tk8QRSlksYtf1Nm8qBqIoZD4fobUmLzocqXn67BkPHzyAwRD4PqEfIqeWir7dpsznM5q6tmpEYRGbYz/keA/G/hludJ/XpxUtA01bEfmKXJcIY9COg9YC09fUdYrnjdi0S37113+DF8/XPHv+ExzlsV6VuMrD8336tkcOHkPfkKcpQeSTbTq6QbI4mNL0lrk6D26hL47ZPvaYLzSBG3Dy4iU1hul8wnQyZzoKWUwm+BK04zAgqSobORQSTN8xSsb0XUdRFGy3W4q6wnUVr16+xHFd2rZGCIExhrt377Fer5hMp3RdZ09Hbcv5+TnT6ZQkHuN7dqdpm45Xr05pmmbHonLxtctkMqWuS9KiQGIZyPSSsioRg0E7AWVRIYXk9u1j+wDV2kY56wpjGpRyyYsMX3u42uX9L7wPEq4urpBIpqMxZ5drsqzi1eNn3LvzdyiLDNdziXyHvK5JEgc3nFD0gg9fnvD48U+I4xH745jbj26zJ0KaMiee7nP++hXR4ghTWwXhL737iE1ZwzBQVyV7d+/gej5d39HsCqo/8+LsmhonipBS4kjF9fUVx8fHbNIt8705eVHgOuHONzLgB1Z57vtWKuv7Dl1r8R9IjdMN5J3l7mRVTd12dsBZKAYhaTuD0j7DTvDjaocgiJGyRwBSiF3bw2OxN0W0OcfTmKu6gK7CaXKquibyA/I0tyDo2lZ7fxrGr8qSrjP4vtUDMJhd+8fgeT6bzYbpdEqe292wKktevXrFfG+OFJJ7d+/Rtg2m7/D8wBbCB0FRlPS9zcJKAQM9rlY4SnFwZ8Y/+q++zGc330DKnFkY0TCmaK1eYptZIXAS+XS9sjBq7DC3UiVDeMaXfuFtri9fUmQpTZWRZxVR5BJoRbrNSByH0WxKayoQPYgOP9T4w4ThOmS2eRe1mqBHNYMUVrmAwVWaNEspi5w+9qxhTQiEUohBEIYhZVna477n0Tb2dMMwYHpwlf3el21FIHcImMHO7/6UcBHFCScnJ3Rti+d5TEYT6qphfz9g0iUAeJ7lD99cXRHF8e5EJCmLkm26pus6xqMRxvTUVY2UkmSU0HQtw9CyXq8sQdDz6Do70qaUy958n81mQ+B7DMNAELpcX5yzf3BAGPiESUyx3XJ4sEeeNBwc7lEWKQJB2zQUVUFedugwJI4SFocho02PUFZJr0VHv70hmYW0rqK+esUdb6BeviBoWjzHZy+Bq8+eM0p86qLi4J33EKbm9OzyzRXsZ16cQRBR7YLkVVXSD4b9gwM+efyYrr3Ecx2qqiYvK8ZJQlvZXqKQmrbruF7n9IOwpuq6ZZu3FHVL3eU0vbFKB8fFDwJ7HDXgOArXsU/IzdpO3se+IPA8zLDP9z5+jnAUabrhMJSIpmTs2ye2dhSD0VRFxjBYSkHfd9Y4PQy4rkIpF2NaC3YaLB3ecewxqK5roijEmJ44TuwOimAynlpGUqSJ45Dlcom782KeX9zgaltwGo+TN6Kfzhj6zjCYnu064/HHL0nuLVAIHh0rnryqyNYa7deAHR4Q0sDQopXCc/WOgCdpyfnKrx7z8Y+O2W6fIqQkyyr29iYMBrQn3xyxlRZ0Q4N0erbLjnF+xEH6EFUn7E/3eF1fkGY5DIbxKES6HrnZ4jiKMPCZjEcEgbbIlraxs7s7zpIQ0hIPu85iSQRIxM5PE5BlFs5tjBUSp+mG1WqJHwZMJmOkEBRFQRiGLF+9Yrvd4CiFFKCVZDyOKYoxN8sV+4sFg7GTQ3EUo5SwV4vh/3XaFEVJXVckcUy6uWY8GiOlfb1lWVLXDeFuhHFvb05ZlrRdC9KKp4LAZ728pi4bXD9EyoHz1+cE7jFd3+OHHghpE0ZNw1W6ZFN7KEdSVQXjZESo7WxzXZV0VYHradqmQxnYP1qwvNxS5GumoUCLjtkiZuLUXJw+J6kL7ixu/3yLs+ntiwgDTRiFeH7Aq5cnSGNwHYcgTKjaFdsyp1c1Xd9TdwNIh6pud1JcA9oFaf8rxw3wtCHY6fomkxHjcYjve/aILMChZxRFDPsxZ2dnjEKXcTLG0S4fPX3Jtu1Iy4G3FnsMXcMkCWg7TVlVSCktl2Y83v2A5NysMsLQtwJdIewb2Hb0BpsWgl12d8DzvB2P1t0pClyEaKnqCs/zcJSy9APPtXfqqsN1BKbvCIMJbduR5Tm+79M7A1mWcXWW8p1/9ZTf/qf36MSGeaBwe8XVa008c5FOR9e3+L5h6EtcRyKVQmtLme+6nlpu+eqvPOL16RWev6XKSsqiI0x2eEh3wMiOui7RniA0EZwkjNv7KDx60XJ1vaarC7SUzKcTtpsV/WDwPW2HgV0XJYSlsbcNRVVjhLQu1r4n227RnmdJFGbnl1Fq55uEwPetanA22+ktGnrTWSRmFNDU7Y5Sp5jNpghshdR3LWOp7zpGo5iDg32kUvZztFbWqx0H09m2WNXUBEHAeDJlu17jaM1ib4/ReMzl5RXa1TiORkrJaDyia3u2aWbna7cbPM/DmJ6msVcJ49m+LkPLnTsLpCOQOPRG0vaG8XjENit49fo1H51XDI6/EzO3VJ0CV3N+cYEDLA73LC2wFziBz/Y6Jd3YguOt23ehb3j99DFuYB8GQnY/3+J8uS2Qw8C6rnBlSRT4hIGLjiKusoLV2TVSB+TCJysMeVFZDKXqGQaJ9GI8KUBCFNqo23w2pWtq2qrElYKDvQlJ7NF3La2jwBgwgnq7ZHGwoB1bBCVCkGUp8yQkW9Wssgq8kM3NBZ62CvVhGKxX0rWl+a7rUI5FGAolGYS0KjusuFQA2nUpsnyXXurIspxhR5Vv284GvYVgMhrRdy2vX1v1QJpm1o7tSoSwSI+yLG3ksCoxpsd1PTzf4ky2zwyffbPg8BcEeljhugGBtyBbu/ihAEqWyxrtZISBQiiJEJrA1RjTUGK48/YY5fVoT1BkkiI3THvBgKHHYESHFAq/TwgvFrjnY/w7AcubNfFowrbcMBtFuFrS9RbUjXQQWWWHwIVEO3YHHgYIoxiEoO06At/fvR8dXhTaKvpuQbVNw2AERZ6jHcfmZV0Xx/OZ7+3RdR1mN3Rw9+4dlJI24VU3qJ01Hdd7k8f2XE1V12RZilaK2XSKQOySVHbOtW0aAt/D9XYP0djD0Q6z+Xx3YuoxYkdw8KyrVSqFozSBH1pFAjby6XkebdPhOBrPd2n7gSDyrCt1B4AzwsHIAOFAuoPIeUHIpur48LMTPn80ZTYbM/SG1WqNdALa9ZY4kigZwNASJwH5psb3Y7reto3SrPr5Fuf5Nrdza1WJdiTTBIJqQEnFzbpkk5Uop8NxLZvV8SMcR6O1g1IK11HEoQZjjVOmazlINI7QFLnAVRLTlgwNVHmxq1YaXFfjeh5pmpIkMZu1lby2TcujOwd88vJHlL3ks7NroqEjLAsUtoBTZBlRZMe1yrIiCH2C3ZHLUuFsT7bvrfHKHp1sWAHsv0dRQtPYoo7juNZTYgx5nqG1xtGaru1JkhgnzxE7c1TTWAuZ53mUVfVGcd/UNctXhr/6lw37L0KOv2y49chjb1zx4096S2Sf9FRFgdY1o7GwwOTd9IwjNf0Q0js1i+OQl68cnI2iKjraxsUTVrdCC+Nmzry6TftS8ODuQ5tRFoK+bVivlhzM79N2tTWjFSVRMra6CQaEMPZqIMWu+OKitUNZ5W+qtkhh22nLlaUZSrvQsrQgSRKrH9jamkTXNriuS11WVEWB7/pUdcFmtUY5CkdJjLFGsihIkNRURYW7M4o1TQ2OvRNLaY+zrrYsprou2aaDhWm7djzQHmXtvbaqKgI/4GZ5w2I/YDafsd1s7UNDawYBaZ7jas3F2YUd/o4jgjDi5uYaqRz8MLQ/K22Ddlyk9nF9GDnWMTubz4mCALYrgjDADXzicUKZlSyv1iSjiCBQ1EXJ/YcP8DyXPMuZjeacvHzKeDZC++HPtziLqiQOY7TnMQDbqmNT27scg4P2fFzPIwhDPN+zNuih52g+QYoB09aM44CqsKYvN/LRdISuixY+bd2AVLvRrZDLy2uiyCdOErabDUpp4jCiqWs60yGVQ1uWCNPRd4IXpxc8WCSYXeV0OplwdXnFsOuVCmFxmHVTY/p2J4eVGNPveoH24yD0ybOM8XhsHZlY8vswWO3gsHOeIHY7bV4gEDiOZDJJdlRBO31TFQW+51m1n6spazt6pg2UNy0vvpFz/Vhy+faa2jM4/T43Fxtc7eNGmryuQHQEYY3rlvYBonscx8dxfd774jt8+nhLvh6o0p6+E/Z4WQd0V2PS6xn+4DNyPerO0JQVSoAYOrSCTz99zrvvPsD0DaUxlhlb95RVvYM+l0yjGINNQg2DwpEOTd1YsvrOV+p6Lr5n+7NlWduvvaqI49iSD4Gzs1OSZGLvp55dFEXeM94bcXV1xXyxR1NXeNqzKg3XfTOPK6UkiWJ830c7DlYXaYs0YO/12tH0oqduKqgHtKtpmobe2Nfu+p7tSYcBm80G17Mw6LZtrWG978kbKxLWnkvTtgyl5WT1pqet6p3lu2e93HBxs8IMPr4fIB1Jb3Y0iNC3d3IUYZiwv3+MMI8JIvsQK7OctsHa8pZbbt17yB15mzzP6Xfe2J95cTrKIQjDNx9rra1Hsa7QDAgp6IXC8Tz6YaAuchzRMnI6+q6hkwa6GkdCkIQ4WmN6Y9/AvreYksjeU6QcSJKQwHcJfJeycNDao+06etMTRQmVasjOrwl9l05AVTestiUHSYjbGdZphrc76jRNgxkM6/X6zW6ntf1ylbQtAiFsFbcqKwLfJ/B9NpstRZGTJDFxHNtyvJAI7NymUoooDqnKiuXSkiGUUqRpak1lTbvrWyrEgB0vG1x7/0LQ1Q6n39vy+MMMOVkzfs/BJJKLlyvuvzdi7/YCYwq6vqOoWlzdIITG0z0DDmnWoYMEHVZ0VQdSIwYPlT9CZnNcJyajxR3Peb5J6esONRgYGkbjMaqDrKiIApfxbG6ltEOJI8WuwqkwnT0em2GgLEpLSXQFjqOAgXyHeWlb2wbodn9Wu2O9dBR5UTCbzRiMwNWaoqne8JqGwd7tlVQwWB9oUZR0nWXlKqXQrn5D//N9H7CnnW2ZITHE90cAACAASURBVLFFmkFb9lNdl+R5TjIaM53N2KYprudTFoU9NXQtRZ7jTqdI7WB2ra7F3oLVasnh4aEVAAe+je2lqaVHOuBoTdsZ1mnFalsgQpe2L1FaY4TgZrXi7iTGcQfKvOLi5DX+o/tAT1/VBK6H9AXt0KMcyeuX53z+/RKhLIRO9T/nnfPo4Ajlukgx4ClF4Gn6tmO1znhw/xbDYLjYZGzyjNbYHo5wYLlNiVxFEAZ4ns823bLdbAl8C5RuW+uD9D1bVdNa47kuwhiqumSztkeU1oBWgsXeHmVZE+zUClEcIULF5eUVNQ6tcOgRrFfrN41zGwawnpbADywHabe7dV23q6o6SCGtutDs+pvYsn6/MymrXWFiMhmzXF0zGGN5tcq2SYpih2TZFaMUggE7eN0Y2BY1WVHTGkOLraS2WCCYUzr4jYP0Q05Pr4hHksXdfaR2UdK2M5quQ9QVnhvw5MkJP/7xM5rGgNaIQHJzkXIrOWS8ndFWmmTqk8xuUQ0GP4rZbjdUTUvb1lSdh0PP2eulJfErh7bK8MZ7TPf28IIQqTzatiPyXcxgLMjL9AgpcHajeVVRIARvIoiOUrbnujv6ljuY961bR1xfLW3LCvADj+l0zKvXL5lMJqTbre0h5zlhEJLn+a7HGVD9dLi+tjthEGh7Ahp6zGB3ybbv0EpZSa/WeEFImmU0VUVhKhZ7c9s3HQam0zFlZaFyXdO9oXeEYWgnjLSH6Qea3a6tHYtpXS7X+GHM4CbU5pz85prZfM5oPCKMIkTdsL1akow93MhDez4nz04YMJRpwXQypRIVQjnUTcXB3oSrszOMGWyRVf/NS/BvZQj9uz/6t18fupJZ5DIJNF2+YRp4xJ4k8hViaBn6Fikk3SC5c+8erhpIAockjlgt19bj6dp2ibAsRdu62MXsLi8vkUKx3eZ2d1OKm5trhsF6LqIoIE1T8qLierVmces+T1+cs60qWmPhw1HgEwcBXdvQNi1KyN0saGOt2tKhbhsEA3JHPHj48AGB57Pd+TMsEMz28YQQDP1A01SIHV3wl3/ll7i8ukQKaREVBsbjGCEEeVFwsH9gRd4GgumMB5//gG9//0ds65a86+mFgxFWZ3HraME49NAM+Eqgk4BGwtXZFUJ1zA4i/NBBSkPdVAwFfPbRGU+fvCZvSoau4/VlQeyPyDcVv/fWr3HkjqmWawIBh3FMdn1D0LXcHseIukEIxXQ8R3kBvXAIRnNyI8i6gaIXVJ1hu814fb3m2dmSZ2dX5E1HPzi0rdVdBL5LXRU4SlKVDUHgE8fhG43FMAiePHlCGIQkSUhdDjz55CWLxZTFwrYy/CBgu9kCIJTE93zEMOyOiS1lXhInMddXS5Y3KyaT0c5Z09K2DQPgeb7VCoYeXddTVTXGQDIZWbbTAAOCMArodgMKSinyLN+xpmwk0fd9us7Yq9kOpFbVJVrZ1lSapqzWa1bbjJfLhtN1DlIw25vjBz5v3z7m7iQk7EvmIw/lCsqiIoki6q5CKoe+7zh+9BanZ1f4nsdkFttOQd+zv5hxfXnN4cMv/uwMoZGsCR1DLA3H+3M+WV/iqZDZ/vRNSXxvHOGXLdAwCzV5rUnXS7bXa4oiZ5SEHB0dEUQheZZihh0vtt8p7dua3gQMDCxvlsRRaHmtKIq0pG87uq5HKE3TQ3F1TVO3NHVD3w/kpuJstWboKqauwhHWvdK1DQzsigUWHj6YAT/0EUC63VrmrDEYY8iLDNfT9F1NHCcUeYnvB1RlgxTw6SefMHQ9fTdQ5Dnz+RRMj0Ki3ZC86Tm53NBLRdCn/PLfSfj9//Q3+ef//A8ZT+fESUzou4wDl/XVFY4UqLahuVoS74+ZzCZsrq548r0rZtOE/paDNDWRGbP+4UBzHjE0hnDWsrjvU/YB5san9FtkMKbNtyxmE37yyWMOxmPMdokrJhwFBySDJK9bljdLHOXgDT2zSCOPbvHiwsEEHo0c6JuOomoolKRVijp3OClKBnI8Z4seDH25QfUtd48OOVQhY6XJC8v8NWZgPt+j63q224w7d+5x6+4Brufy+PFnSGVFVo6jGIx9H22RqMaUOXEcYzzbfgp9zWJ+Fy9wGYaIy6sL6qa2D9qqRQBhFKKkw3xvwXq9pshzkjgm9Hwur69J1xvGoxF5absIs9kMIeyDpixLfvSjH/Hw0dtIKQnDkNV6Y8n0aqBqKqLxCOEonp1c8Op0iac9XDEwmc64vT8jGUoC03L/3UdkmxX5dsPrF5e0hy2LoxnSKKToOD17TdtWeN6Y69UVd27f54cffh/PlcRJ/PPtnD/+7h99vc5zJqMRRZkSBj5lmZPEEV1ndqZew+XFBb6WFJslnjDInQPT9WxWUyqFGYzdYbuOYRDUVU1dV0RhYCFdvT0u7s3nuDtFvb2MD7tCjkI5Eqk9Tq9XNEDdtCjsULgjBu4fH+K7irapkRZ/t/M92iOYdhxcbYsG6/UK3/MQQnJ9c4MQAq1dRvGYNM2t7o2Bru9R0mG93oIZ7OfTHjpIuNykXGxK1p3hYlOStT0oRV73PH39Kb/3373P4u05p59ec2u+x8jTlNvtGwph33dIqZC+hrFP1TYUqxzHSOajORc/aik+jiie5IxMwNuzh4ybBH+tWezNuWwFjpAsX1/zhdv3KdIMISVt1XKw2CcrDRhDEHjk25Surbl7dExXFEx9xcPDfYqraw7GY+5OptyLR6is5O5kzvuP3ubo+A6+H2CkpB0ErXRYli3poHi1Knm1qXhytuLx6Yqn51d8+8NPOFtXfPj4BB2OGSdje9w3Dr6rbG+wsUiWOJmwWa2ZjBNOlxu++/iUtjWEnnzjWEUM1E1L1w+EcYLnWjGW52ocrd7kmpMkwXG1DakXJek2pevbHTI1oigrPN/nRz/8IYeHB7Rth+M4dmoHe33oh4FB2NSP54e4QYDjauqmoTGKx+cbirohHEU4UpKojqPEh7ZmvVnR9729mglrnQtHY54/fYHnOigpGIc+l+cXzOZ7fPMb/55f+OArLG+uODw+Il7c/9nRmP/2X/+Lr3dtayuxOzh0nudUVcVgDNttShiEeJ5mFIe4jk3mu65jG8dao1yfsqpoq5K2aUjTFCFAuw5NW3F4eIwjbQB+m24IQxv5apoaz7NN4735nDRNOT444NXZBSJwuViuLSvX1WgzEPgOngJPKvIiIwoD+l3PU2uN2eEYu75DScuOLaua5fUNSRwTRSFV3XF+uabvDL6n3uBMEILGQKdcTk4vWdWwaTsutwV1JzAMtIPACwI8z2OxP+d4b8rerxwjxxGHd97j5Ic/we0sBKptSnxP47qaYRioygI9T5CxR7FKadYtzalLdx1R3vTcvDrlYDJBDhDpEFkphpU9sq8VXJ9c8KuP3kcKwWLvAFc6uFHI1fUN2lPs7x9w9vol8z0rvT1YTOn7jqfPTxgMbFYpq6sb9scjXnz6KcU6pSlK+jwnrBuCosRpDCMkcdvx61/+KpNwhtEuXjLFSyY4YUKlIvRkHyeZkRqXj15e8PSm4DsfP+UHj09IS8XeKOI66/nT732Xu8d3mE8SrtOOH5xlLBtBqCRj1wZJhsEG3r3QVkjLLKfveuI4RoiBLN/S9y3GWOeq6Qf+4A/+GV/60udJRgnzxQFd2yKlJNtmSEe9eeiWRcM3v/lt9hYL0m2Bdl2iUUxZ1iyOH3K9vOLs9Su+853vM96/x8tNQ1bk+H5IEvrc3RuzPj9nu1kytDV78z0Oj44oy4KqyhHCcHS4YLtNCTxNsUmpy57JLGE2nvD8xUvu3LmLoySjw4c/++L8kz/8P77OINlsNoAkCkJbtdTWMh1GAY52bGFFSuqq4mD/CN8PyLOcumkRjibwfRvLc73/D3hLsr84pG17Xp68ZDRO8H0btG+7jiRJiJMEz/cZGIijhJcvX+IojRYWablJMxu+7wy+q7m7P+VgNnpDJG/age3W9rKqqiCKQqQU1FVFVdVUVU0QRcz2ZtRVZecxXU2a7aQ8bkzZS15crllWhpuyoUbSD9iTA4IwTthb7Ftmr2nxPc2DW/t8/8c/4Z3f/kVEtODu57/Kgy9/nvntPf7iG99kHiXszfap25a6a3CFYny4z5B4VEVBmzUMuWI+2icKImahh2w7fvd3/yGfffaCKAi4Ob9BBwGbQJFtUt7fv4UcBiQDeZ4xnca42qGpK9Y31/RCEEQx+4s9JAZj4PTskqbtqOqeMIzpG3tf8gKfTZahpMI0HdkmIw4jDsZjKHLKm0vkdovfGkZGMkdy4EcMRcPv/NZvoQdw/QDiBOXF1IPCWxxwsSlYFuDPA/7n/+Vf8OnZhv/ww4+5rgT+3iHRaMxiFLGfRGgkeVUilEQOkG83dF2NFAPadanrhiyvODw8oiwz2rqlrGveeeshWZYyCDCdIUli6t3E1MH+IU3bUtcNDx69RRhGHN+7xSRJKOvSFryQ5JsV6+UN8WiCcj1OVzWvVwV+ELA42CdQglAY3nl4l1Hk4boOSRJR5RnrzQrXczk+OiQIXKSW+H7E8mpDZwyzgxlx6CMcwWRvzGazZf/ef1wB+LfeOdc3N/hBQpFXDEJR5gWOo3YjUtaCXA0NZVXZrKl2OT09JQwCZtMZ19dXyM4+uTzPI8sLRvGIzTZleb2m2Ja73QM225R0m+IHAfuLBb4f8dFPPkZIi7B89OgtVquUHgijBE8OzGcjXr5agiPwlSJbrlgZC2SWjrszDFvzshBWCagcjecHCNWCqNGO5PnTJywODgmimLKBXjac3NT062uUo+iRGAYE0j5otKZqewLt4jqKW5OQ1XVGuD/C91zSy0uOHu4RJAknN9eU/WuObz0iHE/47f/hv+FgPmEsFCd/9iFP//1HdE3P6rNXNA/GKN+ldUroBzabNYu9BfEooV9v+LM//Sbacew8rBF064raTekCKKTBb2t6M/DBB1/iW3/xTe7ee0TfVkihEELxR9/4UxZ7e/zaL3+F/b05H33ymDu3b1M2HWEYE3t2AL6obPZ4GgZ0dc3D999nmxZslilNqwg8Byd08KWmMR0vPnvOL33wAW6WUz97xn6dM5xfQtOQzGa8O51y/4MvcKZb/qc/+B+pv1WxLXuWZcfdL36Ami8o6posr7i8qPi+KxmagsiBX/mFLyMGl6rq8V2PfrBtk2g8stjSVhEkE6oix1MORZ7tPC7SQsPynzpM2VVuJQ8fvsXLV6/Yv3PMdrmiyHPqvsf1I3stqHMWi0Omh4dsio4f//WPab2AcTQBIejQrBtIswJT2F601j6nF6/wA58gClgtb0g3tlJ+dPcu8WxMut5SlTWib5kvxkynE4pt9jeuv7915/zut/7467eOb3F6do7jeIjBpnC22y1BGJOnJX3fUxUVph9Yrza8fn3KYCDeJWaqXVLG7paKpmmRgO959G1rPRJhwF9//wc8ePCQ7XpNus0QQnJxccG9e/fIshwz2BbMdDKlNwNDP7De5KR1h0Ryb3/K5x/ctoLewSZJurZCOwJHSVs5M3bH2Gy3NFVt+2ZZRT04vLje8HKVsswKOiHsb2PQrsdoNObW/j5tljGLY8ZRSJrmdEbgSYPXlQjTIAdD17V0XcXX/ul/xlXd4HtTzm+eo4MRea7IG8Hs+BEHb7+LOgpoI7h8fU273BLuj6l1T5XnRG6ApwKEUFBmjFxNuvl/OHuzIFvy/K7vk3vmycyzL1Wn1rtv3T3d0z0zas2AQBIgRraEwDiMZDAmCGxshx0OBzjCfmkvIPOEHQEyEcgGSwojgSRkWQIJbTOj6dm7e2533/3eurXX2Zfc9/RDlubJKGL0Wm9Vp/4n//n7fb+fzwpNk0nCCMu2sFSVceGwziPyWYjmZXzyU2/x4MFDiiyhZjUAheH2LhQh165cwVBlhv0eWZazMejS7XaJIxdZEHGcJYuVw3juYphNZEVBVyWu7A2gyBiPxhiahFCE3Li6gyjm6Er1ZDs/Pub6lR02Gg2Mep3j58+4sbdDuJqz1lQ+Upb8P7/xq/gjj/GLMUKSc6XT4U/9iR/g1htvEUQhSRoxGO6QIpMIKrGg8fRszsPzOQ/OFzy6WPHgcMTEi7E7mzw893k582k224SrRTVcKTKGG5sUuYCsVQoJRalscWEQoOg1/DAmTiJESoqyQNU1dnd2Kl9rkqHqFq1ul9V8TphLPBs7hElEp9vGtGxkEUy5QIw9WvUmXpAjSBK6pjPc3SNLIrI4QJckBMWoYHGqzPHhAbquc+3uHZ4+eMhoPMF3fPbuvPndX2t//p/81Dvr1bKC+CqVHj4MwkqWquo460q7l+cZcVz9s/d6PQShemcQRLH6ZeQqaB7HCWFYXScDz6PRqBMlCaqq43tJ1cAnpVYz8L1KSW5aJqIoEYYR/V4PQzNw1w7tpoWha5zNVhQIpL5D19TZ6PVwPZc8z5AVmbwsSdIq1WTbFpZhUEFoFdZhwpnjscpKCkmpfqwaaIbJcHOb3Y0hR0+fUUQBrZrGejnD1CTyLCYpIUfAMnSieEWjXoGagjDkyXjOzX/nTZxsRZj6BO6aWq1Jp7VFw96kZjaJspiTxTllW+fVt78Hs1AYj0ZIDY0oDCEtkUoFTdG4sjEgDQLsmkmRF+iKxNZwC2e1IjUUZnmAN17wudt3efzkETXdwKq3GE0WhGGIrErESYLrhlh2E9eLmM+XJGnB4csTwiBE11RUVSNPyyrYLlT7TNuy8T2f9XxGr93ErOncuXOHKIqYz2bkedXtjJOCG7ev8+TxY4Io5drWVY6dCd+WFfS3PsUv/9Iv0G+2+OBLH7AerYndiJam8Mb1a7x3/z3OJxNURaajm2w1GuwP+ly7ehNVt0hLGUEzEY0amaYTSTVejtdM/JhIUvjo+QkvZi7f+ug5YSHT7A1ZxTm/9ptfxdRqxP6aK7fv8I0HL5jS4dvPTxlYMlkGhqoiKyrzxZowjkmiCEU1WK6WLFcOUydh4iYEYUS/P6CuKWwaAnbm061bFLKKVLMpo4R2s86Xvvj7yJJIGPiUgkivv0GRZWwPN6vXjWadJw8+otNt0x8MWM4X7N/91Hd/OL/4G7/8TrfbqYZAYUSWl4iCgKHrhFFMo95gOp2iKPJ3qkJFUYGnXbcaHFUA6SrPKolVvaYsK3GMXW9UPCKxZGOzi+c5yLJSQYhXa5qNBsOtLdbrFZJYuSE/fvwCL05oNqr7elKA44dQZuwPuuSXevo/kOeKooAiy7ieT5xlLPyQg/Gco/kSH5kcEVGUadRtet0WN3e3ECKP5fkRhA6WJlE3dcyagSyVtBo2nu8RxSlBWtLvdVDFAl2sdrSFovDX/sH/yJP5C9xVgGbUSL0M3agzGa/Z37/FfOVyfHKCWFY4FMHSufP666xfnDFdjsjVksD1aJotyATStcOw26XM0yqAXxToqkacF1XBedBkcnDBZ/buUmSQBDG5ICBSsNFtgVCVr81alTleOQ5W3cCsmWR5TK/bqyS2RRXgDzwX2zJo1C1kSWLtecRxhCQJBIHP4fEpH338iDxN6XY7GJZNgch6NuPqtau4iyUbn7zOP3j3V/n+n/gb/PTf+UmibMVHX/k2uZvjLj26mwO8OKO9tY1QU5k7LmIh0JFlyuUMi5JoOiJbjOmqEqLr4o4ntM0Gd++9xnD3GsPtKxh2E02vg2pj9DdY5yIvpi7Pxx6+UuPB8RmHM5f7xw6BPkBvtBFrDQrF5uZun7OjQ1RZgCIlSSq7XGswYDy64Ph8yoOTJYWiIyoq9WaTtm2ybSkMGnVWyxWmYZCHPuvI5fDogH6nQ7fXQ5AkoiQj9AIkEZIsrTq7us1qsUZWVabjCaqssnP7k9/94fyln/vH78RxgqZqqJrBhw+esre7U4WkqVoiZVnSajUZDockSXIZ6RJoNOqoqowoXbYOKEnTrPKYaBqCICAI+eVkV+fxwwPWKxfLNFmvHRaLxeXOLCVNUkbjMaIoYGgyn3rzTebzGU3bRtUNxksHRZLY3exTpAnSpaErjCKWbsDBaMY6KRh7CU5SkokqmmHSaTToddq8euc2chojBB7xakGRhLSbNlEUglBimFVipWbUGI/GmIbN3HFA1pA1hY12E0ES8MOIg/WM9RUdvaZhKw3OTg7w3II0F1Flm7OzM/q9bVrNFo7rMJmO8dwVjpBw/5v3CVcLZFsmi1PkUkEXddqGjgpc3dthcjFG1w0Gw010o0aSp5ypCcvTCW9s7hEEHq1GG/UyJLLd7xN4HsvlnDxJ6bSaqLJIGEaVuNb3KXOBZsOuZDxJjKJqaJqOKMioWpUtjbNq7TOdThEEiU67za1bV5lOp3h+zMODAwaDIaIicDi74P+8/+v8yF//r/mFn/lVtva3efbhfUo/xVl4CIqE3qmxcfcaP/af/U0enT7B9RYYqkm/ppM6a9IsIS8KOu0mhqaQxTHteo1bO31kf8Xk0UdMHt2nJWSoSUi6XnP7+iu8/sqbDLd2KGSJa9dvIehN6ptXERUVIc/w/JA4STg5Oebr95/w7OwCN8zp9zepuHUlgVvtR+NCZZYodAabXLt1C93QCOYjlNDBNjQEYO2saPV6uHMHyzARpII4Dun3B4zHY0DkzquvoYgC5+cXLGYr2t02e/tb1AwTXTNo79z+7g/nb/zyP3unLAWu3dhnuVyS5SXddp35fAJlFSpvt1sIgsDGxgar9RqEKsplWRZFnqPrlyEEw2C5WpOXVX1MEIqKSyNUpmpRgla7SUm1m1Q1Ddu2OTs9oW43Lp+6Vc5yNp+zchwkWeHG7ds8OjgiLgROxmOCrCQVJeZ+xNgJGDs+pWpQiCqyqtFqN9na3kZTdNIwIHcdnPGIYLVEEaWqMC0KOCsHQRCrhFAUkWcZ8eX7cwWxNgnyAlVX6LQb+KslrU6P//x//R84WM8ZDAZcjC7Y3x3S7W3hByGfffuPoahwfHKIptRJY59Ws85qtcBZL/mTP/gn0aOc6fKCpMyQC5GN1gaWpJH4Pu1mG1nVqJsWpxeneCuX8WLCQbnAW3p8dv9exdo1azx7ccDGcIPxaMLaDTHNTjW4kEAWBZIk58reLmVW4HoVkqNqc8SMx2M6nTaKKuF5DttbfbI4xFmtsEwTP/QY7m4xGk2omBUllm2SxCVx5vHbo0ekvTZvf/qHeOv1t/jnP/NTXBycICERhymSrtIYtjF6bb7nT/0wL44eU8oFvW4f3c8hL+j2u2RJguN4eK6H5zsossRyMafTaWNZdTRVo9du4jsrbu7voiQuj77xNfyTU3TPpytJnD98SOY6vHL3Dq++/ga6UUev1cklFd3uUZod5M42TyYeHxyMeDb2effbT/jitx7xxW98hKTqNBp1FElEyDJqiAhBiESOooqIgkDkOsSRx2wd06y3yPOIMI64snOFKAwp85L1eoasyfS6fU6Oz+j2O8xXK6KsZPPKve/+cP7Ov/rFdw5eHnD96h5R6HNlb4/Do5eoinJZBC5Yrx1arRZQBaCrilFEmqaXenKP0XiM63r4Qcpq6Vet/TSj0ejg+wlRGH+H/ep5Hm+8+Sb/0V/9q7z7lXdp1O3vTN8sy+bB46dV/lLVSDN4+eIlo/GSKMmJs4I4K5mtfJwwRdB1dMvGbNQZbg741BuvMex3+Oj99zk7OoE4pK5JmLpK3TLxfRffD8izCjWiaVo1upclwiC4DIaLVXBAlfHTnAJomgZaVnA0OWfj858kKyvte55lzGZLEFQcx+NidIxtbdHv7jHoNlDlkmfPnpInGTvDbVahx8dfe580CUAv8JZLhFhkp7uFXkrErkuZ5XiOix+FWDWTdqfFEWsoZfaVFlKUotZ0ms0WYZJSMyziOEESMnZ3tihLiMOMPI+o1w3SNEbVZFxnTRRVy/pOq4O79phM55QlyFJlpDZMkzCOaDZtVqs5ZSmzs73DcrGkYdaJowB1q85jJty6/Vl+5p/8S37p//7HXBy9oG33WM3WFbTs5pB2v0kmSGzu3QZSPnrwAbPJCTf6Q7pGjcX0oqJvFAKOE1KWGYN+hyAIOHh5wGq5RFYU1us1glB1ScejGaZdR9ZUnNBFN3W67Qa7nRab9ToXDx/y8r2vkS0vSOcjwvGIe9s7kJXcuXMXu9HCaHXQ6h2am7t0Nrcxmw1KUWI0HnF8dsbReMQiCInyklKSMUwby27guAFnq4SnkyUNQ6dj2nzpa9/k+rWrTMcjup0B127eZHR6wv7+VaIoYTxfMR0vufX6p7/7w/nl3/qVd3Z2dknikCLLGU8WnBydsL2zja4p1ULYtsjyFNu2cD0Hzwuo2w3KooJsxHEVJBZFEU0R6HZtBv02aZbw/vsf0mg18D2P9WpNrWYQBiF5kfNzP/dz7OzssFxO6XY75HnJZDxHVhWKIqPebDNdrhk5PjMvqHROQl65TGoqG5sDhhsDBu0WpiCwODpicfyCxekJNUWiXa/RbjRwXZc0SaovgZpW5XH1Kqcry+J3cpn1ev2yeWIQ+gHIEmFeUlAlWsI0RG3ViLdrHJ+e0GhUkb00KZEkFU3V6LS20PUmq9WUIFzx4uBjuAxyx0lCkUW0hhsEzpogXJLEPq16lzQTWS4WCECcJdiWzcpxoSgoBIEzJSQvBT7RHaIkKbOlx9agjxcG7O1Uot/V3PmOF1M3dQRkVisPUKpB0eX7pQREl0p4VZEoBQFFN4izjJXnE8cpeZaysbGJYmvMvJiLixFbG31yveTnP/wS11/d596bn+fP/8iP8oXf/BXqZoMoLfAWM+7duorUMascrGLw5/69v0waOkSux9HRIY7v0VFrDPsDOv0evu+SpAWWZQE5O7t7fPoz30MQhKxWHjt7e4iihN1oIOsKVsOm0Wqw0euRhgk7O1d5+OgR7mpBEq7Z299jvZwyHLSJ1iuUPKKlSzTImDx7RKdIMaI1ubPg6t4Vtrb2uHH9Frt71+h3N+j0+khmk1xpx5TpfAAAIABJREFUMIslns89nkwdjp2CWZyiNpq4aYoTpNy7cZsozwjjhJcvRlycjKjVVEI/wlmt2du/QhQn7N567Y9yrf25d5I4JssFXD8iiCL6/R5ZlpIkCQcvTxlN5jRbbTzPx7YqJGK32+ViNKIoCmpGxZoxDL0qYYsSvl+VYrd3toAC3VDZ2hoCAlpNuwwsqMTxpZpPVBmNRqiawjrMcKOMbzx4yvOzCamQI6sqVr3Jzt4unabNZqvJlX6b00cfYYk5SpEglzmiCJqusVquiaMIq1bxUmWp8nO6rnv5ZBTQVJmt4RDfq9Y6YRijKCpRFNPp9nG9AFFR8aMcs25jiyXrPKf5yU0adgdyGduyiWKfOHExdBPPXzIaneB7CUIpISJWwfu8ZD6dsVoucT2H3rADUowXLEj8BEPpIel15o7P2g+xOk3qdgu9JtNv9XhSjsgLaCU6QlyQxwnj2Yy6YZNkGUkU027UUVQVTVex6g3iKGY8nZIWGTVdwzZM0qQCjlm2BaJAmRfUrDprP0QSSkxD5403XyeJQxRd5/79x2gC9HpNHCfgOF5xVq549dU3+Ml3fprX3nibX/in/zvdzQGnT16gWwY3336D+XKBJIu88faf4cmTI7zlku3BBs+efljto2cenZpFkSdYtsH29oAr+3ukWbXOCbygYkOlMRejC1RNpShK0qzEdTzOTs+4ev0KYejj+i43b91kPh0DOYHrkkQRzsrhtVfv8ezFEbplcO36TVwvYDweszkYIOUpZ88eooUrli+ekk/PWB0+5dEHH9BtNNjs9+l32+zs7lBvd2j1ewyGW/Q2tslllYmfMgpTPjqecRpCqGo8OTwiLmUSQeJovODByzNu3rlFf3jlj5AQ+n9/4Z0kCdE0nQKV9XJJq9lk7awu6z/NCl+R5ZxdzFAUhSDwmM1mZJd9zRKh8muGlevD9wOyLK+6j1lOmuQoqkYcJSRxQpmXLBcLsjRlvVqh6QZrP+Rk7vD+i2PO5mucrKTUdJBl6vUOw+EOW/0Oe00T7+QFlpSTBQ51U0cSq9pRGMUUl1CxmlGpH0RRwrJrFTIjqXqjeZpd5m9TlisHUaqUCYqisLGxQRzH1Jt10iQlFwS8KKLV7WIqIl6e8vqPvo0i6ty9/SrPD56zWi2gFFjMppg1A01TyPMYTbaJIwHHXSEIBUbNwHddwsDDC1YUJCRpQBYn3Ohsc2t7h972BkqrSywrHLkBo/WKB+ML5G2T0PNQlznhOuDOrds4UYCpVZnawHe598pdJvMlolTBzBYrh7wQEBWVmmmiaQqO6yBJMoFfleMVVWXh+Dx8foQmSQwHfY6OjhmNp5yfjdjf26PXaaIqGqcXF7xUIswNlaZxF2u4ybd+7zcZ9puMzsfMTse0hj3cJMSyVJwLhx/8sb/BVq/Hr//Kz3FxfkDoOWi1GlqesdfepN9tkaU5s+mUKAxYLtcMBkOiOCYMI7q97iUVPiJJI1RVx7BM3nzzTS4uKq9sxYuKydMKieK7HmEY8Se+7/sYT8aohkEQxARRSBAEtDo9Tk5OqFsmlmlQSgI126RmVU4eU9MwFZFiPcNMXOLROc7JCVIUUHorhi2Lve0dCkliuHuNdr9HKQpkoozS7jKJCqZxyelyTSZqfOXbT/ihP/knvvvD+S//+U+/s1757OzsUZQBh4fPqddtFEVFEEQ6nQaWVYMyo9WoY1nGJWCrQkTadp3l2ud8PGftOOhaxQ2tVH8lRZkjyRKGoVMzdOp2Hc/3yUWZdVpwunB5erHidOkQlSW5JCJrVZm6aVp84tZ1enWDH/vTf4zJsw/JAxdVVpDESi5j1WpQgOcFlznZimwABXmRIysySZxXqBIxR5JEdLUaHOWFxMHJBYIoYGgq0+m0wpaoCp7nIQmQFAIZFfbTFEtiYm784Bs8eviQ09MzZEWjLKqGfG9gkuVhRRaQNOIwwTAM8jxBlAS2t3fQVZX1qpqqKrJEkSYEyzXDRpcNScTOQ6wyZFg32b96hUG3h7kxYB6OGU9GWEqHWNV5NBpxunTwRYFYrNT2nutzejGlvzlgvXLYHA6RS4Gt4YBOu854NmU0n1bprkYTRdEoRRAVuDYcUlNkptMxtmWhGzUEUcCqW5yOT3j17c/ywcUZ92dPqOkmZ4uMT1x9lV//5f8Dozfk7Okz6ht1tq5fQxIEJFnh83/xr3Hz1hs8++jr/Kuf/UdobZumadEfDrCEnIZoUtMkFoslSRLR7XaI4pS0kHH8hCD0EQqRre0hWRZSMy1effUe56dnTMdzNFXHdX1m83lVP9RqLOZzaqbNaDSjKEta7SZXr1/DbtRZLBc0mzatZhOhrErcQRhc7n5T1ssVs+mcRrOJgoh52eF13TWqJLAandGzJUp/yfLFc9LxMe7olMn5iLoEr9+6TrvV5OrVawwGAwzDYDoac/3aNd589f9/IPSHE98lhVa7xbPnT2g0Gty5dbeioqU5vu/jByaOt0aVZTRVJM0SojQlSHNUSUbTdTY2NSzboswLJLGk0WxgJjWyLKXVbHN2esZ65bIQAx68eI6bFEg1q4I2qTVsrcJVJEXGvm2zt7VBWYjk/pLjpw8wDZX7X/19TF0nCHzCyGNzcxPfFwnCgNXKQZLlqvdp1oijmNW6oiOYhkGGgON71C2Dmm6iqjJ7V/ZZrdZ0Oy0eP35IadewbZtarUYUJ5xNXQyjRhyuibQarFfkdh9xHfPo8VN816ex1WW1HNHu7FAUVaN/c3PAfD7n2eMndNubCIqAVtNJo4CHD+6TphGKqrKcXRBFa+ymjNow+fDiEUHh8dbwBp2yRuQsUfIMoxTJnBWIMUkQU982uDm8RiJplKJEEIY46zXjMCRdT7BqEmcff8gbn3iDk/WaQctm5Tisjpa02k269SaqJOOHYcXTkVWUvCLf5+SIskQQBuzsbuGsx4hlnWarxTe/8XXqW016uY3VucuXvnKfTn+PuR/RWs1J0pS9q7ew6zaRLzOZr7n7+vfx5S//Fi8//Bp////6Rf7hP/y7aIrMyycPKGcu8iCm134DTZfpmE2ePHlGt9cjTZbIEgiKzHg6w/UXtJoNgjDgww/vo8gqnu+zYXUochWKFMswWS4dolxEAT755iucn5/jeh5hmhOGIbduXufw6IjV4iXzRUWt7w/6xHFOmkTEaUC91YKyxK7XWM5n1C2LdmNIGLqQG0xGMwpKlssFOzs7bNZ17rSbvHzxgrNvjRBKkfJSPykadd6+eYOXRy/+refvD2cIBdHlHrNVAbOiiMFggOfNWSwWmKaNgEC90WA+XyDLGp4fI5TQbDUoqGjviiIRI1QTziCsPCmUjBcOB5MFKz+gVDUkrU6jpkAJzXaTMI549ZU7XNsY8K/+9a9RrEJSKbq81gRoioimqawWS7IiR5arie5iviLPC8xajb3dFmES47red3irnU5V/M3zHFEQaDfqCGLlhtR0lQ/ee58sL9B0nV6/S6/X5/DokPySgiAIBYYmcWXnCs/HS6KypFm3KeIVfbvL2cEBx4cFmq4yHR/Sag04uZggiCWWadFsNpjNz1E9l3Zzq0I3Oh5JHGCbFSqj1bCR1JRk7ZAXOe8fHvKlbz/kr/zxP8O2aSKrUMQx127s8tHxglIsUTKHfjilvbmJn6RkhkbY3kSp2bwcL4jTnKJf8ni84PTslDIKaOgGr928wTKMGWwMOT8/pWFb5EXCi2cn9PpDbNsGqfpMfcfl4OCQemsDQZS5uXedX/vt3ybdbbFY5+zf22d3e8WH3/wqu1f3WM4nyHWDza09Tg6PkGT4e3//n5KUKV/4zV8gCXx+8u/+90hZRnIyYvvOLhcTh7OLMXGcsLu9xWg8RlIkyjJDlytSQ5oldNo2rXaNLI2ZzFx2dnYutwd1QOLifM5qvaZmNZlMJ+zt7SJJBStnTa1m4ngetcuS9d7VqyyXS3wvpN/rE4VRhXgpMpI0omWb3Lp7mxJ4eXCA5/uYZo1aTeXgxZiiKGi1G4RJRKO+g1UzWC3mTMbnRFFGp90HCnRbp6dpTMcrnMkhn/30vX/r+ftDr7Uffv333jk9Pa2ucaKCqikEQYQkSTRbHZI0udRu+yxWDkmaUbftCkWJRLvVRpQEJkuX48ma8WxBkme4ac6L0YzzxZpYEFBNC00z0A0D3dC4dWOfP/8j/y5pHPLxN76GJeTE7oqmbSFKErPZpNLUqxrz+Yy8KLFNk9VqSVEU1OsV5t8PfEoq61mlm1crnGOekyRVsz7PM5I4Ir/sk/qujyxrOGv3EmpVdVEBHMcjy3Ou7O8gCgWr5ZIIhSQvGW71EJOSH/3LP8H944f8qe//88RJymJxjrcSSOLsUsz7B6CqHFFUmUzGlyIfi1rNIPB9osBjMp6QZhGyXBI7LnlUUOYaL0Yznh8fcu/6TYbdHtPxmNjQeDGecKu1yd12j6mzRs4TDDIEd0EyOeF7b+5hJx63Xr1ORImk6WiWTSRKPDg+4N1HH/Pe4SEXns/Fak0qy2jNJkGSoWoKtlkDRcFNUtIkw250Ob844f6j5xxNVryMHQpdZ2f7OpvDDu9/8ysYlsZyNOHem28gqwZllvE//b2f5uODZ/yzn/4p/PWSP/cXfpyTwxeUuUSt1cDs1jk/O2N9OkPIC8QS9na3GA43KYsS265g36oqs3bdCgBu2tRMg2azjWlaHB0f02y1uH33LnlRMJ/PsSwThJzNzT5pnLK9s0MURfzWb/02O7u7vPvu75MmKf1ejySMeOuTb/D08cdESYAowtXrVwGJjz5+SJalKIrEcj5jMZ9Tb9ZpNm0GG33OTk7Js5Iok/jSNz7m5tV9iiLiyrUd0qwgdlPm8znzxZI0Stjo9hhevfVHSAj97E+/U683CILw0v6UM55MUTUDz3UZT6Yoioooy3S7XfZ29yiK7DvUu/lixmg6JZNkgihhHfi4cUqUlchald7XNYPh5hbdVpO333yd82cP8c4OmBy+YH52Qq/Z4E//0J/h8OULajWrSrAoGut1pWJrNdr0e10++7m3OTw8RJdVwjDi5PQMP4yo1UyiKEJVVHw/oNlqXRLeFZrNNnleVb9kqeIJxUmV/TVNi6xIaTZazKYLUkHlbDLDNgzWqxVmTScvJQoRSllDlVRIU/w8ZKUnUAYU8oqN3ibeOmBnt4PnhrTbbW5cu4WzWpHnIZtXGly7cQWp0FjOF4hCSRKFOIsVSZJgWipJENCst8gykSTKMbo9vvrgQ1qyiLt20Hpt3jt4ziuNIQ1DJfICiiLn9t3XyLMYw6xV1/f1nKaYU84vKJ0FO1vbfOpTn8FutvGiGMVQQZbJVJ2XswXfevSIJyfHPHpxgFpv4CUZT54/443X7jGZzzA0lc3hJs1Ohy88vY/V2uITb72Fu065//6XGQ66eEnMnbuf5Oz8jB/6kb+EE0SE6ykff/AeoeMQlyXPHn+EXW/T29pnPptTpjGiG3F1uMOV/e0qyxwnqKqGs16TJNU6KcsqIn12ydVdLWYs1x6DjU2mkzmWaVYPClHg6rUr3Lh+jSzLWa5dfumX/gXXrl3FtC329vb51tff5+aNWyRpSG8w4NHjh+hGjeHGLmsn5PDlMS+OjpivVvS6leG9YdvYdh3d0Ol0W2iqShpnWHadf/N7X+GH/+wPoCkK44sJV/auE6c5R+dn9De2sRstPvPZz/Hld9/nre99+7s/nL/+L372HYB+v49lWSR5giSrFAi0mm0QRay6jSLKbG5s4IcBQRSSpRnTtcPJbMncTXCChFKSqVkWlmVjmhamZbKxuU2v1cEUcwRnRjQ5R8pjbty6gbteYxk1zscLnj97xnA4wHd9uEQ23rp1G1VRWDsrev0uH374MUVZfVBQfucKrqpqBaJKErJLsHFelARBVFnARJFut0uapqhqFf6WZalau2gKQRyh6RqT6QyjZpIm1XuhKAg4jkecFRiNBoqs8ODZc775/n1+7L/4y5wdH3LyaMHRwSmFkCIrsLd7ncePnzIdzwi8NZPRiNlkge8HREmALFXOlSypWDr7+zcxbR3fnUMm4LgR7jpkuV7j+z7ddoc3r9zgNFxw6q+5UmuzWbdpddpIqsnFeIznrskKePz4OWIJ7tpDkXVMQ6HdqPPi+VPe/frXCaOYXs3gjav73Lt5lbptodUsOoMBtU6To+mUpxfnTIKA3/nWN/jmh/d57VNv4ZUw91wezV5itXvs7V/DdROePf8IRZPZuXoTSbL4D/7SX6XV3+B/+Tt/i2Cx4OjlM37iP/5PeO/rX2N3s4eh1vmzP/oXefH8ARvdLkYmEHserYZNv9elVjNwnDW6pmMYlaTo4MUBe1eucHExYjgcsrG5yYOPHrA13CDPU95/7z0+/8Of53e/8LvcunWdf/Nb/4bjk2N+/D/8KzSsGl7g8wPf//2cnJxw9+5dVFUhjjJEsUQ3VMq8YL12UZTKvt5pt1FVlShNsOo27srl6OiUXr9F3TaZjuccHBySppVVfHe7z3CjT6vVZO2uOT+9YLr0mC8cJMVk4WY0ehvcunPjuz+cv/aLP/OOIlcqhvPzc+qNDhejGc1WF88PyIqS6WxGo9lmPF/S29rl4cExL87HOElJWkioukHNMjENnYZpcOfaPp9+7R5lEJA6K67tbOBMLogDj1anw3K9ZjZdohk1ClHgm+89uRTjlMiyAJQEQcBysaBuW2z2N1jO53T7PVzPw7YtJFlma2tInmUsF3PipIJaC6KErqkIVPwaTVWRZQnHcYiiiKIo8Fyvel81a7iOC0CZ5lhGjSs7W1DEqIaJrGukpUQsyMydNWVWoNcttto9hjd2eHz0At2SEMQCXTcwaxrrlYcolEzGY8q8wK7VkUqRwUaPja0+aZbS7Q+I/ZA0SbHrHfIypCxcAsdnfu6gKyZFkqEUCo+PDnn77j0Ks+DB6UtuNTbYrDc4PxsjUnFzG606ilZjs98iTWManQ2OL1Z86nu/h/OLEXGW4mYFi9WahliwqUnE4xFS6BHPp1hJzPVug2G3w6DdptusV/1Q0+DZ6RnvPXrC1FuyyHymY4eL0xEbwy1c74yyzEDU+NEf/fd58OAhP/W//SQNVWM+HmE1eqzWc8YXRwilyLVrn+DJs4+ZTg6ZjWbMXpzTazS5srdFw7Kp1TSMmkbdarBcrKmZJrfv3MHQDaIwoN1qcTGeYVtNRAE6nS6LxYr1ekmzZVMWJYqs8LnPfZaP7n8bioLlak7oe9y9c5v4co7x9PkLkFSazTYlJZ1uncmkynXrukYYBDTrdYYbG+i6iqYqlSslSfDDkPU6QBRVdF3F0DTOTo7QNJ1+f4Nnz17QbdW5sjvkyYP7NG2dwJvx2luf+e4P59HjD97RjGr/tVgsKUqQNQPXd4iznHZ3QJSXnMzXjN2Ag7MRaS5gNxrouk6tZtLvdri5s8X1rQ1yb8F6fMz07JjZ+Rmr+Zgs8CizFPkP2ipFTr1hM5vPaNQtui2bnZ0NkjiiYdmEYUhRCDRabcLQY7126Q26ZGnMerWiUW/g+wHKZU0tz6t3zTSvqGyGUfFSK1q5ULViqHCYmqHS6XYoxWpiqSgiQimQFyW+HyBrGlGW4sYZp5MVuawgKAJbw21eu3ebqxt9zg9e8vijRyQt0AyDNKzYMhIK0+mYoqwymUZNJQ5D4jDB9R0EOSNJUqajOXmSkWQZiq7hex5lERK6Ad3GBkmQk6VphfbUahyPL7hxdYsPz17yyb3bGIXARr9fAbOzyj7uLFcEQYBes2h0+zx4/Jg0jchKgbP5kpPJCqNWowg9SGJkSaBp1Bi0m7SaDcoo5OLwmAY5WzWR3Vad21d32O1vcHW4wVwMOZ6ekgs1rt/7JEUZE7hzXhwfc/PWW5RZybvv/h69RoMwcPjv/vb/zPH4nCcPv03DtGg0O/zxz/85vvY7v8p8uaDwIsq1TxpHZFnC1mavUjQECb4fVmKhLKXb76NpOkmSEASVmEiRNJarFaUgsFytgZLXXn2Fb3ztG1zZ2+fk8BBVkXn06AH7exU1Ib88uJPphOOzCwb9FpQZRVlwenKGadYq1UO7SXnpCRIFWK7XtLsdihIEQYZCwPECGg2bosi5OL+g02lQCPDVr3yNwaDDcNjHtAx29/cYDLp8+/2P+dwP/OB3v0o5PDokLVRmMx+tVkOUYL7ykXST2XLKNMnJSgFZ0WnXjUrLnudoioihaugSWJrA4uQZgSJCCZZeQ1FEmnWL09MLDMNA1zQcz6OgJMkyDFHEUHXGZ2MWa48oiWjYJjVD5+XxMft7V5FEEUksmHpTnj5d4fuVam50+S3nBf5lw0LF8yNKKtTlahlhmmYFmpZlBCElL0pESaYsRcaTOaulQ6fTQlMN4jKpqOSNFudrHy+I0QyT1mafZt2mYZrs9Tq8PHiCWJb0WzaxJHDr09/P8+PnOMGSLE9YzVeUQolSxtSMBntXd/j6179A6IpsNfbY2dzlYnbBfLysQNSCRFkIbA+38YICd7pARabd7xIcRRRlSZHlHHsuD10HQZEQyhJNk5EUmWs7uzx58pTJZMpsuqDf7xIGHs3M5y/82A/x8vCItRdRXE6f4zii327TMlVUQFElZFUhKxViJePeKzchijBUqVqZzecsZjO2egOC8TkCMrdu3+bm9XuEkcujh+/x1//Tv4Xjpjz64KsMmhazucNf+vH/it/4/S9y+OIZu/tXqYsK3//5v8Dv/t7v4Hke7W6HqXNElsQUugolnJycsZrN6HQ6NJpNbNtiuVzx4YcPsK16RaBXZTqdFqfHF+iqzNbWBnkeIYoCH3/0PsONLq4zx3N9nLXD5nCTJI3wvBVZmnDv3j1cd06/VaPbsqmZFu9++V0oRBxnDaKI44W8cvcuk9GY2XxOq93FshqM/XOQS5I85drVLVRFptVuMZlOOD87w/d9ev0W167vM1s42E2T3d1dRmcXDDea/9bz94f7OVF5+PIM3aqjFgrffnyIbrdRswA0i7yAumXRajdRROi3WugifHz/W9VhQKSwLBqWSZam+GFF4NY0Gy+IkTWd+WJBmefkZUmWZbRaLWqGRahF1OtNev2crMjJ0pgkTbHtOr7nkucFURKTpiWm1SDLueSQVpU13w+QJJnpZEaSljRbNpZpoygajrusXBqGgSTLyH9gMA5jBFlE0VTiPMdZxwiyzsSLKKOqSD5oNbENg+1Bn9OXz8mSkGNngVhmCGKlLqypKuYoRJEEVNNGkRTqlsl6PUcUCxRB4MWDY6KVim6YuK7Lu1/8ffrbTdr9GrMLn1bHRpMUHj98SrNTohkaYlriOQ4yEnEWI1LQ39hhkaSkeU4ax7QGW3hRzLe+9T62bWOaNSRRQpQ15tMLGr7HZDIlzAoOjkdEssrFaEy7O0A3bXJS/DhEUTRsVSTLYjSjRpBkKIgsZkvq9SalpqDqCeMw5thZUYgSpQh7V3Z5/uwhGQV1e4OXT79CmYcsVhl/42/+N/zu736BD77yBfobQ2I/5K/87f+Wr33198ljl3uvvM4yWKKlIg+fjnDlhNnKZb8QKAUVz4tQ1BBRrD5rMysQhBwQWK88DFPmEirPR/c/QBQV2q0WEiXNZovJ5IL9/auMJzOaDYuz8zNM02K5dPjwwwfIssqV69tMpjPqUcKNG9dYTKc0u20kWWc0mnP48hhJEojihPPRFE11UBUBxw1I4piT0zOu7O+jhimCoPC9n/0cruPw4sUBeQ6+F5BlF7x8eYiq1fjEm5/6ox3Op5MFuabgpAFFCrVGmyKvJLZJliIIYNdNXrm2w/P7H3B+cYiiKBiqxvH5EqtmYtclfM+lpHo6uU6EpuWXuP0SUZYwbLOyV8mVSlxTJDRNRlUlcgRUQSHPK0yhIIBlWwRRzNJxSTKZ02eHXL+6W5m4oWr/yzJlWYGL27aJrivEUYTnBZVbpVZJf1VNp8gKiiJFkGVKUSQTJVZuRJSlaLpAsz9A0xTu3rhBTZP41he/gOqvqRsaThAglgpZXgl0HMchLXIe/qOf5Y3/8keI05jlbAXkmFYdd+2yCKo1Tc3QEKVKCZGmOcvxmu6WRJq71K0WZZojiQpJnKJbNZwThyQq8eMQURAQNIVULFjMV8iFgNw0OT0/JylKijglzxMa9W1kU+T4dESz0UJVdBoNFSsv0fQaD0/GiJfE/SjN6BgyzUYPSZQRRBFV01n7IW4YogmQpSXn4zmiotLsdDmKHXJZQlVKTs7OmM3GTKdHvP7JzxL7CVHgcfOV7+H69df49X/9y3z8wX0G3R7Ti3P6+3d4OXL53S/9NpsbuzS3d6hFc/o33+Tbv/kuczdGnq7pLz1aG1tYjQZpmeFGOZJQXLKrFKIowjZtzkdTmvU6ZVnQ6bQpc5GT02MU1cCyDRRF4uz8DC8IcL3KsZrFMbpmslr6XLm29R19YRzF2KaFo6wu/SwagRcTGSGttk2WyoRhiSqKbF3dR1QKpqMz9JpJnMOz50f4nsf1WzeIFy5INT64/4jNQQ9n5XLj5m20WoM0y/9ohzPKcjS90mOrqoJeMymp5KTz+QwEgTjJePzgEVngoSoS7VaTLK86eVGcIlxG5fI8pyCjP2hSZAmiqGNbFs12i+l0TAnkRcHm5uZlKTtlsViSlQW6XolikjiBUuL09ALNMCqM5nzJ1rBHFIdE8R9cWWWKoiRLS2qGWWVn0wxZUciDStjTqJsVn6csyYFEEMmRWCwdsgJM26bXqCOSc2VrC9KY9ckBnlDS77Yoipzw0mQllJXJ+g+UDIausX91jzudXXK5oN2oEwUx08msMpQVUfX3yErCoKrQlTmkQYEkiGxtbeKtUvyVjypLhL5Pu61TsKIoBYqsAEnCNi1Uo0aj2SBxxjwMXLobNxCOjxAEAcsyq5xzlrKzvY3rOliWyfnFGSISW70+QQZffvC40jOUOTIiYlExb8qymnrLqoSWKkR0qQJoAAAgAElEQVRhUBUBJIVSVdCGbQInI01yCkHiM298GkWpcXz2jB/6/N/kxbMP2Lv6CmVZ8pu/9vNIUoqm5JxdXHDt+m3ufe/neO+bX2R3d4/PfPYHiaOIyVRFDDPqloYoysiGysj1CZ6+wDJrGLpO0zbZ3xoiGAJRFCLKCuskot/uEPoBqiJiGibL1RJN03D9mOlsxWDQBlFmOlsQeD7DzT6GpbE72OCDDz4mcD3EsqBp14jjmChwMQwDVZPpduuosogfhCgKyHJOTROpaTC5OEdQNYxaj2sDk9HZGZpi0O9ucnQ0Zj53efbsjJ3tDnkJu/t7LNcrLEFhc7j5RzucW8Mt7JqGrqiYuoqb5KSFQJRWxIP5csHK9ZAMhZqsUatpZHlOzdAZ9Nu4ro+iqri+j+979Lpt0jQljlNa3T5+MObi4oIg8Bj0BwhlgaZrFEWJXa+TlyXr2RJZ1mm2TPwgoG5ZlxZrmZppkqcZSRKR5dVUNAhDkiikVjOr91JZZr5YUKsZ6IaKrisIIqQF5BS4UUJWCqy8EEGSMesNDE2j26qzuzHg4PFDZofP0P4/zt4kyLLzPNN7zjyfc6ecp5pRqAIKLAAESZAUyBYlUVJrMrtD4bbb9q43bS8c9sJeyd7Y7qUjrOhwLzrCCnnoplqtpkSKpCgOIMUBAAliKFShppzz3rzzmefjxUnTG4kR4jIjM2pxM7/6///73u95VJksz35mTXYchyTLMWSRIi8xTZMgCMiyDE93KNKUN/7ka6z97l32Tx6hohNHIZLcgrI1RSHJsxbvUjYURYo/jXh6P2J1r4cg66i6TOQLJH6KttkFGQxTx1+2WyGKZbC6vYnVVZieQIJG79WPU9UZhGkriBJA04328zJN0iSmKitKIM4qgjBDN22qpiYOfTLJRnLdtpl2gSEJ4pjAD5FUiY2tTTJZoryyiby1wf6Xf0xvZ52jo2PKumT/6D53X/w4eRZy6dIz5LHPk3vvYXUGJLNj5LJC12xe+eXf4OjoCcPjJ6RRyl9/7S+58+JHuXvnZd76s39Lz1ARRLi+0ePG1Q0s28Du9MhErc1elyXj+YK8LNFEEdswUJAZTpfsbqxzeD5HlSTSUkRSLXTTIgoDRudTBCQc16LX81jf2kBRNSzXQdFk5osF/cEK4/N2kUNVZKqiJPZ9PMeAJiNPQyxDZ5lFGFaHsirwZ3P8uUjeWwdZR1YVNNNhOp5QSQa37t6hKnx2Ll3lfHhOHGdk1Zw0y7h0829PCf3c4ryxvYapiqiCRF1XhMEMXVZQxAqlY+MvlyRpgi/CoN8lXbanaZrEeJ7H+voKfhBTlA2maePYDvP5guncpwF0TcG2zHZeuQxai7TUFkCeF4iSdOFRaBidjSjKklRJL2aXJUeHB2iq2n6dtkjMrMiRL8DWRVlRFxmGoSLJYrtWVDckJaRphZ+kZFWFppusb25BldN1LBxdYz4ecuxPcQyVqi4pi6LlGEkKSTLHdp3WgiXLLJc+itoCvkSpdbSohgZxQT2NKLICUzfQdQ1RUnAsh+n5FFFsyIsMUZBaFUSdU0QVlAKGpbEYtzM2Q9Wx3C6aYSAXKt2BR1Y0uD0XtzMgbaZUCniSycPZBFeWcYr2CSCIEpIkE4YhmmkwWF0hiSJkVUWSRIS6QGigzDJM22XgdRAbsY0zlinzyRRN11nrdQirlFAVyAce9Exmp0eUQsVgrcfJ9BjbUnn86A2eu/1pzseHSH5GViW89Oqn+T/+1R/iuhZ3X/oUYmfAyfERTVXze//4P+edn7xJFPl0vAEHH37I977yJeqywLQ0qsSn8kdsru6xvaoTl6CYJt1+j5nf4WwRIykSuu0ynC9ZajqP4wxd1RCyHFU1SNOCrtyexFE6QdMlBms98jxjNp1SZA0fffkV3nzje8Rpyngyoygq+v0edVWSXjheFEnGtV3SNKXIy1ao1VTYpo4itjIsmZDZeIrXW0URdO585Dmqpub07JS66DKaLMgqAVHzqGSN5uLv5u9dnMePH9LzLO7f/5C9y1fp9jyODw/Z3FjD1U2OJIFCEknzktPJHIcSQxGRRKl1cmYZeV7guSa2aZAmEUkSY5sGpmHQ6bhkWcZ4PEEUJcqqwg8C+r0OZdGQ5w39fgfLclgsShaLORuba0RRyN7uJX7w/R+gazq9Xp/JZNrqF2rpQl6bk+U5kiS3aR8EllHCLM5J8rr1kNgu6x0HS1PY6Hc5O9inWE6JYhldEkjSlCSJsGwbEBEE2jUz28I0DaqqIggCXNehqiokSWo/VFWjyAuavObh196k8+o1Hj54gqobdLqt5rCpa4q0lcuKktgSGOqKJq043T9lfbdBVgWioKAWoC4FNMugKkXkpCCnphHAdnsIVY4gCkT+kgcPPuCTN28QT95g1elQVCVRkLKy2iNPU/YfPWnlQaKIaxp0bANDNwAoGpH6QjAsXGyPhEFIBSRKybkYcy4LSLWCMhnx4N13CYMx0wUISolQxFy+us7aygBV0Xnja1/l5dd+jX/zx/+armvy8md+BUN3ePLgPlmZ8swzt0mWM+LpkMZot5pO3nnEcP8Q2zCoS0jSnCQryMuKLI2wVRWl9OlUCpqcosghg8EK69s9nkwUFgMbQTGIs4qCmvPplErTmEk6JwcH7O7sURYJjesyG08IZyF1UaONx6iGTafXqkHapFiC0DQsA7/1d4oiS3/Z9lWMdm95NpsRJyVJnFIWJZZlEflzVld6LEdPsJUcwzLJFkdQVnRdFz+LEWQTze2wd3n3FyvOKk8wjB69fh9VkhmdnuA4FuPzEartYWgyWVmSlCVBI+B5No4hIzQVYZSCUCOIAnkak0stFcFzHARBouO5CIhUdYOkKBiaRl7krK6sUZcl+8dP0XQDXVNZXbV4+PABqqbwdP8U21LpdD36/f7FhkyCdpGlLYuKrEgIs4JKEGmoqMmJ8pIwSdEti82NFSxNRROh71gIdY4/OqJIAmzLpipLdMskTto0kCCIVHVFluWtvco0UVUVXddJkoxut8toNGrxKZqGpCgITUXVCNi1ghI1eFaHg8MjirQk9lJs10JTNdKk1eXR1KiKTJaWRMuY5WKMojnEuY+oCQyfniLrMrVYUlCxtr1JmMZ4nR7DJ0+oy4ZRNOMGuxynSyJSLH2NxWKBKImUCMhye+OomgapbkiTiLIskBUZBJlalMirGiEOW7u4YUJeESkix2LGfhFQ+WAj0TQL0iQg8OcIikq328H3pyznCa++ssWP33ibO7/0S/z7//BHRIHP7/yzf07daPzoe9/gB3/+Zf7bf/EvGA9HvPfejxgdHbBz8wW+/fV/y8kPvk+/08GwdKosp6H1pUymC1RZ5tLeFrIsIqkqUlnQ75h4tkbqT+mVObZSUokFtaXgdXrMBx5+UZNXkKV99G6Po9MTpEIhxGB6coprWhzce4RuqFzpDBgOh+3vpqgwNRVBUiiqtitsW3ZrJbhgMiuaju8v29uSrKKrOr1ul2Dp4zg2FDknB6OWglhX1EgsJiF1GmIaMtlyDPztBSr+vOK0HYflMsAyDSyrPQ0UufV/m7qCIYnoqoIoCBQNlIKM1+tTNQKIUgvK8gOqqiEvSgTAti0M02j9ikXecnAliaoukUQRatA0C9OyaOo2OzkajS7MYwJpmrG9vcNP3/5pC6oWRPwwaJtOTU3RNMzChFlacjQPOQ0SZlmFYlpcuXSJ3fU1tvoum55FGcw5Pz4gXMyJ/AChARrIsgJFUVH1FjJWFAWyotAf9NvoniRyenpKFIYkScx4MgHaheymabAudOVNUyMLIsnpHH8aocsmZVITzAIEGmRFoKgyqqqkLgvyNEOoG+RGZnE+Q5YL1rZcsizh6MNDVF2nrFpUh+XZZGnBeDpmPpnQ1BWb65s4nsPZ6ZC03+VenHLS6IT2Ckdpw/2Jz3GQ8uB4xDzK8LOS82XAfD6nyDKoCmgqRLFBoEEXBERD5HG95KCMUM0OgiizsbVNFC8RhIKyyVENlZXBZYz1Af3OTdKkwLBlvvxXXyRpKn7l9/4JurHC+fCUOAr4/O9+gWvXn2ERzDkYjfnt/+Kfce36DeQk4uzhEzTNQFFVqrq9jdRVRZGXNKLK+TxmvIzIG6gljSxvKIoaWWjf1KoiowgNqlih1hk7jshavWRQLbnR19gwG66sOPR0kd2NVbauXmdRwUmUEik696chR3HFWVxx6Kc8DXN8QWES50Rlg2K7qFqbCc/THMdy8LxeC7VzbBpK+l2PKPBZ7fexbZOqzFnMJ214Ik8pkehuXiYr4PDpo1/s5NRVDdPQOT45YrlsoVphEJBlKUeHB0iqjSqLyJJIVVaEecXET4mDhJVBn6IqieMWdFWVJZIIjQBJmqDELdv28ZMnXLlyiW63w2I+Zzgeg6TjZzWb/T5xFKNpGqqqoSgK12+0aJMgDNFUk7JKQRA4my2oEMkagWUpkpQVSCqyIrOxtsp6z2tpBcGSaHxGJohQVRRFQdxAXlTt+IVW2BT4wc+sZDGQ51kLuJYkBGrK+Zw8a0+dMAzpdjqkWdbCzyQJQRBR5RZxOT8558l8jKypdDwbUZEY5iWWo2PbBrkiUhYK/iKgrmrStEAqRDRRxN3q4o9TYjnEcbuM6xmKIrGIAp6/+1HiJCIKFwgIFGHM+XxCGvj0uiYIBau6Q1YVTLKGRdpg2Rp5UbKYp1TTlLNFjqIZ6KZBJdb4SYLV6yDrGkGd8yA6odzeJJ/EpMsFeVmgSh5JlJAVKbKmIComW1uX0SSFXmeHk7N7/M2b32bn8g16nR1sp8vh/iOKsuTFFz9GFCd89S+/xNvv/oTf+Idf4Mr1ZzkXHvD6o2M0SSPLU7Isbj2ZokhR1eRVzWzuE8QFrqujn54jXThY61pkdUXCtA26mkecJCArQHs7oCg4PzyiM+jhSAK9vsT59Jy9vetc3/D4pr+gWBmgKipV3ZAJIbUgkwkKZZQjA4rQ4Ikwnid4hoYpqyRRSCqluLZLPgtYDidc3l3HX8zpdDxOzs7o9rt0Oh3OhmMMV8HuDqgmJV5nhTB4ymw6/MWKczGf4Xl7LcWdBllqRxyGYRBGMRUiuigjJgVFXTH1I7I4YkUTiNKc6TykKgt0tR1t6LpJFCbUTcNwNKYoa7oXd/wwiEmLCj9IaMiJ45TctYjjGFEUcV2XMAxZLn2iKCLJCjRDZugnFFVDmFdUgKSqOP0VtjouYRASX1iTLVUh92dUSUyZ5yDLrUZeFKjrdu5pGDpFnmNaOkmSYJpm6/BsGkRRYj6dX4TeQTctZElCVVROTk4wdYOqLCgRCMPwoqmVU4sgaiKmYTJf+uRJSn9tgCBWpHGAoopYrt4SGsQGzVSxOzaj4yEnDyfItw0UrX1v2paDKIs0Qk4ahrz06c8QL5dUzTlPfjpGKkFsJC7vXeVvfvLX3FK71JMDTEVnc2MP1xaIm4yd69eZhRGT0QTR8dh2OmRZzPHwlLwwaYwas8h5FBzwIJnTj1wM3WD/8B6feOUfcj46Zz47pywz3E6fazfv4kdTntv+JOfDp/hnEVeu3KbjbqKrBm++9T0+/olPUxQNSpUxPj3kwfE+v/kbv8OdO69wdPSU73/pz7j/5lsYqoIkymiKhCwLiKJM3QhUTQNCg2NqNIJIEKVoSkuuMGyLQlIoRYXZdIwsgqmozKdTJFmhrisc16FualzTYjKd0tEktNpnY7BJ5c8RVgzGwyHdQZ/uSg/HtjFkmTKKqAWJShCIshRBsVgiM88KKnSWSYNZZ0SNSo6OnknM/Jx+x8GfLtFtF7e7ysp2D820CJYBti5SRFN0TSKM8l+sOJeLOYu5cwHoMtANk+ViiaZq2JbL0fmEvAJV09r9yLJGdVxW1j3Ojk6ZTENMrb5oCNkX+gYFVdUYjcaYjs3u7jb+YkZdVxRF0xaSoXH18ga+P6fX77XNCVFqJTNlwzItCUqRaBGyLASqBhTLY3XQxzYM4ihAKnNUAWJgvvSRyxQpj6Fu552S2AL7kguNhKoqdHsek8mUXq/beh6jkDwvqOuGjY0N4igiz9J2LimApGoEQdg2AaIYyzIoyhyplJktlgiizCwIyE0Ve80lTHJEQSRJSiRVRURhMpoTBDGarrS+yKZBkWUUWeXs6RRBEeisdlDUhiRvN2QEoaa/tolhdtl/dJ9P/9Ivc/jhT5hMzljfXqWoCrzuGpJm00xP0MWKeHpAU5d0bJdVIUKVM6y+QYyCX0BZ2MgIVMBB2lDnc06yObde+jin50fMZiOoC15+8ZN85/v/gapO0AwbWde4fPkqgljy5hvfoq5ynn3mVVYG1/nrb3wFsan5pc+8hqHpfHjvDTxHJkqWvPDcXZ5/5hn+4ov/J2ePHvDul7+ELIoICG1nXGiQJBlRkpBVFdXQ0UyDsmnoOh5VlaHqBpsbq4gCqIZJsPQJfL8VKEsyeV6gSSKiLLG5s0maZpRlzWCwztnwlIPDY+apyNnSp0pLaMDxHFY2NhHKmCtWg9dRCaKMk/MF65ZNtydR1DWK6RJkFfOkJhEVtNVNRC9nKQiU3S1GWcbm7nOkqoIsOEwWQ5jOKaKQzcs7aLbD8mxOmgm/WHHu7O6QZe0sbjabt3drTQdRZLqYE8QJtWpiGjp5kVMlGUXdkNcirtvOGRWxYdAfIImtZ6UsCzZW1mmaViqkKQrUDa7jkOclZtcm9H0URabT6VJVNWfDc/orGyzP55znIWFS0ogKiqBy5Wa73pRGEWfHh1RFjlpmZHFIVTYkYQtuanKTVUtBlyXqpsEwDIq8JklaD6imaYRBSF7kRFFM3TRMJhM0rQ3wh2FIlmVYlkUaJyAKLBYLVFVF03WKvCIvMhAkxsuQ07GPrJtkWYmtaJiVSOJ1yLN2zlsvYzRFo9ftkqUp01OfzCmJIx9FbccnmqQzOZri9XVUD85GI1AEwijkc7/1T1ku5kwXZ1TFTQRF5uDpPmsfeYEgjrh27SU2Oh2y6bfoyTJRlrDSWWGxWBKMz1raQRFRZCVdRcO0DDZEm7EfMMkDRvWYWJZAblm9XbdLXUaoikpRtFZzSZW4deujnA0/pN9zOTy8x6//2n/C2TCmKI/Yu7TDG99/nbvP3eVP/+yLPD24z43rtwnigs8+e5s//td/yPs/+iHF4Rl1FNFUNUhQNzVFmWPbOjQ1oti+h05Ph3iex8bWJrKkoasSiqRwenaG4+XkacJKv8doOCRNU7q9Phs7W8xmc7Ikw/cDtFWDIA6QFRVZVtGdHnp3QBCF9PormLZJOR+zqiW4tYiJSF0luNUcV2zooTIPF3iKR09W6agVhu2BpOCnMufTBEnWsJwOZ+Mpkiizu2dS6xYHB+eYto0mGgyPRhztH3P50s4vVpyyrFFVsFwGLVe1FpBVlSTLKcsCTWzY7ttgddpxZDEjK3POpzNWDYWOrRFFEXEcU+Q5kiwhyiJJHLfk9smMs7MCWVJQNY2qygiDgiwvePz0iE5vwCKKOV8kzMsZYSmSFSW65fDaa+3/xsf7T8nCJURLstmYwvG4em2Pg/2YvCyQBUjqhkqQ0E0bIYtoqqKFRRUNumEg/EwVUWLoBnVdIwitT1TT2rduURQURUGapnQ6HZaB3zJydZ3pwkeWVLK8Yrz0qUSZUpSp84zVnotrqJi6TtGInIsgJRWqoiMJEMwj6rImiwvSeI4kQ10JFEnYhgWSkpMPzxjsDYjjGMs28SOf1fV13r73Fv3uKpbdgUbAsV06gwHnkxH9TZfdy7f56de+TtaI6EYHQdQxzIblMmw/7yxFKys0uUHOcwwRCjVnIkREecKdFz7FbD7m3nvvsLV+hbpsKJOUOIzQTZMoK+h4A/aPv8/O+hXWBtdBkJhNztBkBcvweOnux/nq1/6E88ljPvHqazx8MORXPvcbfO/bX+WDd95g9mQfLS5beNpFMFbTVHRNRhSE1snZ1KiKSr/bwbEMiiTC8GwCP0IUpsRpiu06SJKEIElsbG4SRhGaqvHo0RM6vR5pVlKWEKYZs8mYy5euUjUCp9MJAiK9Tg/XcxiYIpdUkTII0HDQVZOQiH7HxDBlTEMiSSTC2ZhOz8MpS07e3yeIa2rFQpQ0Lu1tUzYhsiNh2h0sXaERGrauXiUMYkRRA0Hn6nN32djb+sWKM0xyTF1DTFPqulV/7x+csr2ziaII1EWBaWgcnh3RtTrEcRv1y+oa2+vgqhBEIVEYMp/PkGUZTdNYliUgomsyZV2ztr5FFAcoikJWVCS1yDiumGYzchpq2UASFJ776B3W1zdYTEZMjg/Iw5DT41OaKiWNA9ZWVtje3GQ2beemitygKxWNJCFIMoKsUSQhdQ1iXSNS/Swoneft2posq+iaznK5xHM9hPYxSNmUKBfjniLL4CLQEC9izpcxOUnblBEkRFlms2+gSw2qLKGJIk0cY8kSlmEiIdI0DVGcEPoRuqojSTIN9UVQQaAscoqqQBQkylikKWoUQ6bb67E8XfDwgx9y/9473Llzix/86CtIooBuWCyXC9IsQVME8qpkZXUND/D91lMpiiKdbo/h6JyVlVV0XSNYLjB0Az9OiGSVua9x8/KrCI3F/fe+SZqELP0ZedGw8GMWywhBFvj0K59jMjsgXOacTxZcu/EMB/sHXL/6DJf2bvLue+8g6xmPD095+aOfxXM3cK0r/M3ffJPre5scmjbTIKMWxVYbD9RN2zKXROEiguhhWA5ZltH3bBzLwL7Ytc3SnDzP6Pc7ZFmO7TitsCqDMArxg4zpYs5mWWEaNrbnYeoG3atXkUSFd3/6LqFgMJqcs7ezjSjAuq2hlzGD7S2C5Zwsacl/RVFTVS1pUXFWGZ2OKBclnqmwud5n6QdUNOi6QBkco1su3ZUuuiGhmw35Yoij6EgrFuubG8wMHYSC6XL+d9bfzx2lNI3A/fsPWF9fw3ZMgnCJZepsbqyRZ1k77E9ShKpEBlRVpakhLxryqkHTDaqioqlBQCTLstY3UleIVDRNRV2kjIZHHJ0MmcUVR/OIsyAlF1UKSaG/scNv/kf/mDsv3mV7dYOn772NmERU/pIPfvo2hiph6lqLmSxL9vcPuHfvIVHUWpr7HQ9NVIiihMkiQNbMn/GCOp6L6zrIsoznuQiCAILEfjhh3hfY++wdyoFCIOfUpsLR+RBRUQjLkhKRsZ+wPw3wGwlB1eiu9Hn21g0UQaRjGmiShGnoyGLLAVKSGKnKESSxNYKrKoqmgAJ1nlIXOYIstVcuRcV2HSzHRhBlpsMFrm1h2TqiWHO0/4Cd9U1O9g95/a++Tl1XpGVGnsd0ey5lHnF08pg4zRkNzxEE6HQ8FssFILS6RUEkyXIUwyKvKirNQd+7y62PfY7L127Sc1zWequUaUVTC4iCjq6Z9LtbIKjkRUkYBTSNQCMU+MGSfm8Hw3BZBqd8+Pg7FGXB53/tP+Zgf46kODx/5xaf/Ow/IDqdcvSTB1DWNHWbKa6qhrwsCMIAaFCVVqycZS2WRJYV6qpGUWR6/T57e5fwvA6ra+tEccL5+ZiiqFFVhd29PQRZZGd3hyIvWNlYw/UsZtMZiury9OAM03Y5mkwRJImyapABtw6w1XYFrW4a4jhBECQkSWF8PiZOEySry+6NO7id9u0LObreYJsiji3juRqykGGrOYacotZz1pyCjpoQnT4lmpyRR6fsDkSU5ZNf7ORUpIb19RVkRWJ1bY3T41NEGg6fPqEBwqoBAQa9PpMoJU9blULaiDw6OSPxNZosIc/zi0VnG1ESUdT2HZNlOXHTcDZNiIsKQRGRNRfNkPnsr3wOVVP54bdf54O3fkQWLlnoGmcnx0S2SRzF6IaOLAukcU2WlkhCRZ5nrKz0sR2LqsqxDZNlUoAkYbgustSQJGfUmkactqBrSZKJ44Q0TbFMmasv32KohTxuptTPeHzkuVd544c/xrq0h980eJZHU4tctnoUWYWEwOjkBM9yCEdTkAqyMkYVIckiKiElEgpUQUSuRCRRoagrREVqY35Ng96xkCSR/uoqeZYRBtXPwglSWTBw+5RpRZYnyKrM+eScdcPj/GTI+emIXm8bSZapGoGmEjg7e8JyvGT72GfL1ln6PmXVekpbBSMUZclwOKTf7SKpGsM44OHkLabhnKbOqcuGOAvor1xC0Va5cXmXb73+NQRR5uql5zg4eMDW1mV0RcNfpnS9PSxT5/0PvkV/sMbnf+13eHDvCUkSc+f55wn9Mf/3//NNXrl1k//rD/8l6XyBqmrkaYYiC2iKjCrJqEobk/z/5sZlWRJFEa5jkqQZw7Mhlm3R6/bQdA3D0JFpEC6STXN/yvXnbnI+m9DpD5icj1ttfVVx+6WXGR0OWd/aodFU/viv32Dr6jOYtkPPMfGMnDIOSVORQb/PYrFgNr2IMK73yPIYT43JMp9ZeI6pazRii8LpDNYRgeFoiKpdZKnjCFVR6HY7fPhkzDKJsaMZqpKRRxPW+tbfWX8/n/j+l1/8A8e2CYOIJEowTYOG1o+Spjm1AJ1ul7woyMsaTTeIs7KV0Eoyqiji6BqqphFGYatwLysEWWORFZwuMlIkBMPA7PRY27nEa5/7Vbr9Ltki4PW/+Crp8hxbk4mCJfPJjDLLOD0dMZtN8bo98jTHti0c16HIi4s8qQCCQFW3rNW8rgnyjLIsEZsahQaRhqauaWpIs4IwipBlFUGS6NxaJ7VVLj/7LPN0SlSMqZUK1RVoNJPelXUqR+a13/xlfCmmsQXWb62grrvEZoi1K1EMZK7+0keIBznOCybO7Q729XW6O9sMx3MkSWuJBlLTLn1LIpqhI0gCZVVR1TUdz0HVNcqqJo8SwiTBdA3SKCENCyQkZtMxuiFjmDLhtMbZ3CGvclRBIpgv2MyBC8h3lmW4ThuZRBAoqoo0y5hMZiiawSjL+OnjDxidHRAlQftOt13W1nYYdDfpen3uvf8TLl3bQfhuYBIAACAASURBVFShrCKGwymOMUBRHA6P7+FHcyyjz872JU5OTzg5PeXm1Zt869v/BsWo+czzL/K//PP/hpMn+xSVQF5WpEVJXlXIooSiKhcjKgnHc0CEuiwp8oymrrFtk8Ggh9A07aGxssLo7IymrinLHMPQkCWZxXRKv9NFaOCNN95gbW2NPEvZf/IESRRQVYVc1vj+T95nc/Mqluey51Z4Ysrq2hqL+QzbtmgE6HQHjM6GjMdTVEXDNmQifwpViSiIFxR9BdtxSLOWoGFZDpbtEEYlSDoncxjPY7av7LCx+wx5OCaLfHTTort+7e9PQvDD+P8HZKUZrm3jOC5BGLGyukKSZKiKShiGlFnM+XSJYnlIsgqCQGdlDU+q8Odj1ja3yGuRo9GYk1mArJtItovldvjoJz6F53i895Mf892//AquLpEnEYvZIVub6yRxQFNXIMCtW7d497336PS20DSdOEpwHBc/WFyMRPQ2DKDqZHlOnhXUVYHQtJS9qq6pqwJFFKmbtoPcEuhFyrqmTGM02cbqqjw82KepJV588depCpP7j79FEiWIio0sKgRJyv7wCd3eLo0k8sKnNnhFfxb/3OPe/Z+wFDOElTUkr8DyUvx5RT5NuHxji6MnUyzJbndji6J1norQ0CBIDbt7W8RhO8oRlIaVjsdofIYqKCiqxHIa4s9nBGHE7//T3+frX/8iXqfH9tYm+/sPUbt9Nrd2CE/uoeoWtqGzXM7Ji5QsLSiqkjTPsSwLr9NhEYccT6YkSXaB3JDQDRvb8rDtHobuEkUJlSCS5RZPD98EoaDXuUKUBJSBjyq5XL10jfsP7hMn64xGM1b7Nq9/54/Y6q3xxf/5f+XPG5PDB/s0QhvdrKlBEClrqLKSsG7QFJG8EhCUmK7n0F3pYhsymqqga/oF17hHmmbMZu3ifLfjIYg1/V6PBojCiP0nDxkMVnn2meuokkAaZTimju2aHB8dIdirxHFOVGTYeYgnCFCVxBUIqsFiNqeoappGoN/38ByLoqyZnJ/jBxGO5UDdYNgGUZQi1FDnNXEG07ii8gv6vR0kt8+TH/yQj730DEmWE6clldrjB9/5kDCJ+a8/8vm//8n5na//+R+4rofvB3ieh6K12yNJktJc/Mzp2RkVECc5juOCpJDlZfuPSzJBnLI/GjNcxCzyglxW0O0Oz3/kJV791GfodXus2Cbf/PM/ZXz4mO1Bl+nZKWmSo8j6hYJPuQiZh9iOQZbnGIaJaToIgkCSJm1rX2oXhEVRpKwq4rQtvLqBom6omoamKpGpL2KCOnUNkqwgihJBECII8PRshLkxQO+aXLt6l4PTY5KqxLBtZosxnZU1Ot11jg6fMplGbG1eYX39CqNRhOWqaE7Nyuo6jx4dkRcZsuy1ZIBunyQuEVBoKpmqqJBEiaqqKPIcTddogLJqvS11VTFfzBAamExmVHmO5ZnIqkQSZgio9DbWuX77Wd575y021nZoDI2O65LnYCoK3jJheHpG13XRNJWqar0sZdXgOG47P5YEjsZTXv/pewiaRl1XeJ0uK6vbaJqOouqtLrmGz3z2s8yCA8LwlDT1UWQLWVKQRJPR5CFJusQyXMoyZHb+iJsbm3zrj77O5N0DykXGaDilrNuCFCSx1XGoErLcAsIlpcWs1KJIUJQs04xFlBDnFZUgoqoqkijRNALL5YLJeIymyfS6Hv5yQVm2MiZZlnA7LstFK7W1LJNev0cjCkxnc1Rvla989w0ks4PpuNjlkhe2HBzPYT4cUmcplmNR5jllXpBnJU+fDAmChqzUiDMZP26YRA2zWCYqLRaxTCbayOYAGhFHqxHqhCeHQ67srvL4w0eozgBZqGkqv+UL5Uuee/HVv//JKSKiyBqdbp8iy1EtC1HN2FpbZzaZ0Qg1rtdpsYqrqxSNQOYnWJZJWdUcjyftm0mxUGyN/mCFm7efa2VCssSPv/115uMReZpCXVGmGWEYEKUxkiBi6i2G0A8zsixjY2OL0dkYVVM5OR6jGyGq3GBbJrUIYRhi2iZpXlLVNYZhtielKJInBYsw4u7HX2bV0nh4731qaqq6leKKkoygaMyDlCguCL/3Dluf2MOye0iajh8v6Lk9+oNL5EXD0emPyWKRT37q80wmR7zz3rdQ5ZKH92cIksQrrz3LK5/bZXJeMjxKERUN5BC7O6AxPGDOdPguRZYjKS3h0DANkjTBkE2qskKWW5mvaVjM5hOKtObeux/wwkdvIQgll67tMC9TfvCjv0ZpIClTzDwkbgw+ffdTvPHd7zN/csjm2irH51PKLOHG1StEYYIkSzRVRdO0YQpVt0jyHPKMssoIQ58rpkFW1vjLOapmY5k2aVEgiTWyZKOKAvPJmLJp2NjYodtd48aNW3z3O99BKnOqJwu+/LV36QoKUZyQpBmNULf09qJBllqsSN+xUKUKWVaQNQNBUUnzgkYUUHUVNA2/lghnCSfxGZJwiue6vHBlkxeu7aHUKf58wfbmJaI4ZLmYoak6brfH5avXOTg4wigFvvet7/CFf/QFNHPKvFBZ5AWuayErCrsdnXAx5fw4xOtYZFXFbLrENHQMS4Oq5tLOgCjN0YwGr7fGkyfjtotsaBgWqGq7frdMhnQNiySrORrGvPyZ3yAaPqB7+xri4BKLw/s4nkSRNOxsbv6d9fdzi9PpdjgdDlnf2GgTK8MhjaAxnbdhg9D3EQUJ23E5OT7BcD1msyWLKKHT76GrKrphcvWZZ9ja3ePB+x8QTsZMjvdZnI8oiwpBFMmSgqIoqeuSs7Phz6DUaZridDyyvMS0DBaLBb1elyxPWF3ro2oqRZ6xDNscbEXD0g9wnS7dbpeDw33yoqSoampFRxQV7n34hKGp0OQlRdqS3qO0IExjkrJGNkwcS6Xj6jRMODtOWFl7HkXUSYoQPyy5dOkaqyu7xHHK6dmI9fVL7O3doS5T7t97iy/89n/K2/d+gGgskeUp/fUEVe5hmH3qtEDRFFYH65wePWZyUFLFCagqZZFj2DaW7SJJClmaMZ9NiLIcu+ui4LFYSmiaiWZo5E2BKMm88MLzfO3DB5R1hYDA9tY2hydjbly9zf5JwGgZ40gSvY0+4+USTRYQKoUkCVhb30Q2DWaPnqI0kCU5a+srmK5D3ZQ0YgOChCgI1E3JOx+8zsP9e3iDVV569fO8+MKrLIOA4+EjxuNzUD2u3f4YtimTfSxkc28XvRLoyQqL/X3+9//pX+JPSmpJhDKHoiIJS8yuQ1nEZGmM7Zh0DBNJ1RFUlcH6FqrpUAGmZbTLFiLcn8Z8cHoPihTb0rhzveDa5gq6YvH06WO2y5qyPsfUdXRVZHt9nbPDI85GY/aDmuHEZ+/SVTTD4Np2h/GT93C6XSbnE2RV5/LeJpIskRU5iA1RmmDZNpZrURYRnY6CourYjoZpmBRZzWg4YuD2iCuNXICdbQtZapj6JaKi0k8jLMuht7LOPDrh8ekpn/xFinM2W9LrrTAcjaFuyIqcphZBFOl4HnlpMppMiIOIcZjSRDmKYbB7+QrdXp9uv8/W1hZS3VD5AXK0YHQ8ptPtMq4gyUo0RaWqGuIkYW1thShcsLa6xsH+PmIjo5sl165d4cGDB6RZTBgprWfSENEUlcVsQV03VBeyXkM3SZIMUfYpq7rdZtcEcmRKSaJpauIwoslaztAiSgmzEkFR6XVM7ty4hlTlHE8PabIen//V3+Kr336d5+58EttaZ3dDhjqj39/GMOAHP/wuutZh/+CAW7ducv36q5zPajqdS8xOh2RlA8oJk/ljxEah6+2y3b+JKe2w8U8+zdMP3+EbX/kTgtmy3assGxaLaUvXqysEpUGVVLIiIc0yQj9hNltAA8eHB7z267/Og/ffp8xztre3ycqS4ckJV6/cJvrWmzyzPkC40eN8tmS8XFCkAk0YkwdjttcGTBdL5DRjb3uXF4qCv/ngXQ4eLej1Vlnb2OPmndskZUlWV3iDDqZ3nduvfpamkZGEljzw3vs/JYkSfvkzn+HDh/dZWx0QxnMEsSYKIl79B79DVcX0r0n86a//D0SPdH7/C/89QZAy6DpoIjRVQ1nQLtcXJUUdQhrT73s4xZyOJGF7LoUokGs27touZZlQ5Ql1kSFR884w4N4wZKXjEUUGQtzw8u1nOHj8AGE2Rbdd3n54wOd+5fN8+X/7V5RpzPlkzs1djcX5CFVrgeKDlVXqRuDk7JS6qpBVGVlR0U0by3bI84Isz3jv/ffo93rsXdphOZ9RVzWGNWAYazQaXFk3mS6WVGmGH4JqCrz7nZ/gWBK36oauUfNz4HsITdP8nd/8H/+7/7JpaKiqCk3VODg+ZHUwYD5dEsQRqDqnkym1LKGoGpKosrK2yUfu3sVQZN598412kyGNkCUR3bSYTqdoqkKeZgRBwmLhs742AKGCCyfJSy/e5Rvf+Ca7O5d48uQRg5UeW9vbHB+2tLRW4Zbx4ME+2zureF6L1VBVjTRNcF2X6XRGWdZYloXj2mQVpKLIMgxJ/YA8TxEVFVUx8ByLq3ubLM/PKPIAsoIojujcvc7Kpz7K5s2XEJuSIDzhzu1fRVNt3r33Jkt/yu1bL/Lho4eYps7NG89zPj7gg/vvsbe7S8feIM2mWLbCm29/g73tVb75V39BFmUMelusb1zl8uVdjp8e8qPX36KMSxoaJF1FUmUMQyMKA5I0RZAlDEVjMjoCKWdtfY3FJGLz+jM0RJy+f5+bL75Gb2uVfn+NvcZk+d030S2DxTLEcjqojoezvsH5MmD/ZMhwdE4eBvRsB9fWOIum/Hj/AaZp8uIrn8bZ2GLvxlUqoSZKEs7OD7EtE0EQGB2fEscFv/vbX+Dtt99G0wXyPOfRo31+6wv/qHWx6KDbNlKZoQoKq7rMb924i1L3COIUiZKBoWOqIkWdY1s2ddMgy2BoOoauY2oSq/0OvV4Xz7aRZQnTcZBkhdToMBdUVLuLrpgkSUBRldiGhlBVzBdL6izDszTOz89RtDbxVSHy8P5D1jY3MVyP22sGV9Ula7u71ElAJbRraIogoBk6AKpuUBQNs8mE1dUBadbOP8syJ05i5suSMNG4cfdVzk5P6Bg5HdfleBhweLZke/cKsiygCgteeH4Xua55/50PcTsWN179wt8asP25DaGvf+Xf/UESJSiK0pqwpkvivOLhyQkpIknZYFoetumydekqr772GVRJ4IO3fsTh/Q8YHR5QlRmSJNDpughNTbfTpcgLtrZ2CaP4gqIdUJUloqCQpjmKIrNYLHBdm7W1VXzf5/Gjx6ytrlM1Ja987CUeP33ErdvX2NjYZDQao17Q22VZIs+LliF0cTWuq4owjvGjmDTNKRtQdZOO4/DRF57hhWcvMT07wLoYPpd5QdnUKLnA7/1n/xVO7zqaLBJnfepoyZs/fovnn3+Z0WhMLSwpiwRFUZGlmigsuXHtRQxDZjqdsrZ6CUkw2dm6jWUP8LwuVV5wuL+PP59ydnZCmodcvrHDS594iSTJmJ5PkCUJJBHFkOn1XGRRIIx8puMZntvB6TqE8wCj6+G5GidnZ6wONsn0Bi3K6ZwnVFGMahoINWRFTrqc0YRL/ONDtjoeK/0esmES1TXLKmMSL0AR2Li0Q39ng0U8Zzw9Y3x+RhxOqBKf0cEBy+mC3qD9A53OxpyeHHJ0NORXP/97OL0eVlcjzIcsgkMs0+Hs7ARVkvjSv/tjkmDJoLfC+OwMS1JRNKgoUYV2oVlVZWRZRpUvqBKSiCqL9DseqgSGJuFZGo6u01UaVhURTwK752K7Lp7rcmlvB1VVWF0dMNjYoRJlDMtC0RT8KOLk8IQbzz/H1Wef5erlS3iywKoOcpUyWy6YT/9fzt4sVrPsPM979l57Hv/pzOfUXF1dQ4/VTTabFC2JhKzIEGI7lh3EToIkQJDACJCLBAiQm75KkItcJUgCJDGQSGZMS7YFRhJEU6I4mGx2N8lmj9U1njpVZz7/vOdp7VzsMq9kBeRdV9/9B1h7rfWt932emLA3pFZU+v1VZtM5i2VE3TT4oyGWF9CiYZg+Dx/POJ1ZuKNrhJvnaaRKND6mrUrGJ1N0UyMM+vR7AceHh7z+uduktcaytmgNm5PxkgtX/3I/51+5OP/kG//8rUWUgG7y0b17nCYZ86LEcAP64QphMOTVL77J1qULpOMJh3fuko1PsNSWVmnQDB1DM/A8j9OTU0zbJs0LzsbdE0BdN5RV1RENyorZbI5lmc+kpB1nqKlryrLg8uUr7O/v43ohn3xyh42NbabTJVXVMB5P8DwP6JyiG+sbLBYLGlnT64XYjss8TmgAvxdy6cI5vvT522wMPMZPd4mnY0zdYHw6eXbvc1EAA4W3v/lNhlee58P77/I7v/1VjqIZt19+gwe7d5BS4farb+LaAYf793n/wz/lyZNH6LpCkiaU7RT5DJs5Gm0yW8RUVNz+3G1o4HDvCctpTJqUNK0kyqZYAfS2XG7cfp6g53F2fMh8mYACyArH0UjThPWNddIo5cXXX2U+PeLk8JjhcBWjlfSOMsqjMdvnd2iblkePHqO1YNkWSRbh+y7RYkKVRuSFJE9LXnvxBbYvnef+8S6ta9BfHWJZJg8fPKQpSyzL5MneExynj+P3OpZJ22AYgoqGF157Gc1RiaoxZ7M9egOL6SRjZ+sSq/45zvYf8udf/0PyIqM0IjavDRmu9Zg+naJJE3Ro2xZd1+ievBpUIAhcVoado2S+TCjyHF10vB5ZN0gyyuQMr1ywqtSIyQnTux+yZTQwPeHq1ibb57cxQ5/R5nmGq9usbG5jmBplmbJcLPhs9ynSW2f3LCGNc67duIyugYLGdD6laipoQVF1hGJT1ha7BxG5tLHcAKnWFFVB6K9w9OAuTTphY8Xh4oUhiygly3Vc3+bSxU2ixZwPP7zP0XE3Zb5w4TzBYO0XX5x/8Adff2u8iJgkKUnTEoQhjuMxGq3x5d/4Ck7PZ/x4j4+//R1mx09p2oI0T8myksDt/qhJmjGdzMnSEstyUBRBFEXdfW+2wLQs2rbm3t27uK5DELgkSczt27dZLpesjEacnJxi2za6ZrBYLjocoqrStl0x2nGc7t+yYTqbUhRF1zpvmq5Bs4iJiwq/P+D5a1fRmpTDex8xPdynrksWyyVploLa0tQNZV7hey5ZXeAp8NOfvserv/432bjwHJdXz/PpvXuE3gbD0QDf6/Ptb/8z1kc3ee32m4xnD3G8DFnYfPzpxzhOj8lszGQ85uzklMODfba2d+j1Ap48ekCZdXzU2emU2dmClfU1gl4IWstoO+D67XOsbAbcuPEc7/3wHeJZzM72NuEoYDGdskhjXM/i9NEhF5+/QX9RcUWEVArIpmFyOkHTDRoayqokdAOoBWXd4LoOu8cT3F5AFE04a2LUocul8+fRRcudO58wGg7IioTJ8UmnP3B9bMfm448/5OL5yyyiBMu30W2BYpToFly+eI3lvCbKSmSVcnP7Ov/tf/WfUBcSRaqM1oY4rof0G7ZeWmfnxjaLccRysXwGSCswNR1N6LiejWwlaV5Q1Aq26+G7Fp7nULU1SJXQCmnbjqGcJSmDvk9dJDQyR0nH3FyzaQ93aU4OsQ0VqarsXHmOlbVtWs3A9gNyNGrdJg82Wco+UaUTJ51JzDRc0hymC4W41MnKhl5gMxo4HJ+csVjkBL0eWXJK343pBQphz8MNfAzbZLKADz78lMV0iaImDPsWN66OWB0M+Po//WM+/6Uv/+KL83/5R//oLdW0qVqF0WCVz3/xV7hw4Tw/e+dHvPudvyCbTti79xmq0mVVfd/HdWySOGEyW7K7t08LXX9S74Qv/X6PMAyxXZuwF/D6a68ymUx47rkr3LpxHddx0DWThw8e0esFjCczdMPk8PCIvCjoBX1kI7l44TynJ8f4nst8Mcfz3K7EfXiCYVo0aMyzkuNFTKOpDHo+ok4RRURgqESLBYqq4jhu95YnNJAtitKBvPKqY+SWeUOdLNBpCM5f5uBswXSyIC1TgiDk3v2PQfG5ev089+/tcny8xPNWqZE4rs2Nm9d4cHeXd9/9Cc9du8DTg49pKRgNL1C3S4RecXw0RlM1hCIQusbB3j7xMuKdt39EsszI64i8njNcD/kH/+Hf54+//v+iKxa6bhHFM5zQ4OjpGa9dfJGrWo9hr8/qyipnp2coqkKv38fQtI4gZ9mUbQ2qRNZQGzZiFOBdXeXhZJ/joyP29vbIy4peGCKlpG1VNNNBM22SNGU6nfL8zVvUomb70gYXnzuP6xvs7j7Ac30mZ1M2189xcPCIN157if/sb/8tenjEcURRFiyXS5pWoukuluOg2irD51ZZv71O/2rAytUhmqdxfHqMbCS+6RL4HqFvY+sKrmuDovHGG1+gKCvyumZ1cx3dMInTBMdxKcuCSxcvoQmVu/fuk+Y5m1uriDJlZMAn3/smN7d65PNTNlZX2Nw+x2B1nQaFRZayKCWFEfLh3pgHRxGHccPDoxN6gcHWSo87d/c5eDpHaVQGftcZNUTNShhgmjZRUnF0mPD2j+9z7sJFrl/t8crNTY6PH+I4fX783gPe//FPeeXVl9i5/Pwv4ef85p+8tXnuHK+/9jm+9OYX+fBfvc3+g7sYSsvA9ymSnEE/RNMElmUym81pZIPrejz//HX8wGc0GgKQ5x1rNooTev0+utEFC5aLJacnJ53v8+SE8+fOkec5iqJiOxa7u4+5detFiqrixs1bHB8e4Xlul1oqCk7HZ4xGIyzbZhHFmIFP1irM8xLF1NjeWmNnpU8TTVjxbQxVIU4SDKOb+gpV7aJtZUlW5GRZgaSllWAYDkXToCsap/cek5mCf/LP/wXL7BChq5S1JPBt6ibi5OiA/cOHCA2i7D5PdsfoIiSOFmzvbHDh/EXKvGI+m7H76B6mZeP5QxqR0Co5rmuTJRmHTw5ZLhacO3eO4XBAnme4VkgVVRzvnfDDt7/Ptdd32Lq6wdnxjMUkQjM10qOIf+vKaxw+2kUIQVk1nJ6cIIRgPJnQypY4SXj65AmqUFGUlvks47TKqUcaP/n0Z2RpxqDfJxz0eeX2bXShczaeEA5G+EHA1tYWZVni9VyCkY9iNPg9j0UUMZ1M0DSVMAgRQmW8PGAYDPjv/tN/iChM3HMhf+8f/F1+/Te/wtPjA9zQRbZl513JC1qpoqsudVmTVSWaZ3Ph1fNsvrKGsGBxMEM0EAQOmxsboAr29w+J0pzVjU0UVJK8pCgqer0eZVHw5OkBhmljWjZ7+wcEvQHhYMjkbMx8OmW9H7Dl6QRKyZOPf4bdpKz6Br5t4XkhXtjDcl1q08AJ+3h2SBqrzJcNUm2xeh6Xb+wwHDoEzhBDCBTDYFnoLHOB6wWE/pAH9+4QDgzOoobHT2vSrOL561tcuTygKiVbl2784ouzbJS3DEXhj77+dT5++23my3Fn7Y1yJtMxTVXh+i79QZ+w1+ONN97gyZOnVHXFZHqCpguiZ+IY3dDJ06JjARkaZVmwubWO/mwAo2k69+8/pGoa5sslvcGQyXTK5UtX+eCDD9jY3GTv8VO80OPg6JAoibhx6yYKCss4Ic4r5klJWkvQDLbXV3jpyg5qvoA8xhQqpmWTlxWN7KbCTV2TZWkHkZItrVRwXPfZ7qlS5CV5WiA1A882mHx0h//xH3+Ng8mCtY1Vfvd3/w8W8zOavOJk/AlB2FAXBk3qIFSFJD7j8cM9FtMli/kJh/v3USkYDQc4TsjO9mXyPMUOFTzfYjFeYOkWtmETz5fMzmZUSUlVNWRpyRfe/FVWhytUdYznWuTLTvXgeT7Zacy1tS3uRk85LSPqpKKtGzQhWEYRtC15XjEcjJBtx+0dxwl7+RmpKVnf2cJ2HSzPxbUdltGSuqnRNQ1JS103xEmMG1ioVovQczY21hmfTRGi5ZVXP0eUpNRtQdlEaI1g//ufsTZa5fztS7RRy727n/IXf/bnDHt9dFVFFQZ/62/+HW688BKzxRLTNFjfPMfVmzforfndgO3BgtMfHxO4PbYvbCPULhkkm+63TWZzWgWkrHFsB8u2GU8XHJ2cddhQoXPh0hWEZjIcrbH35Cn9QY/BoNf5U2YT8iRlfTgAWePQYhQ5A62mGD/FExYUNUcP9xj0R3iOxXw2RgY96tDn04MxH909Y/9wwniaUqJhG5LQgmgecXR6RiJ1bH/EdK5QNwU7qyY7A5Mkzbj7aMytVz/3iy/O//6t/+Yt8oIyyTrI7iikF/Y4HU+I4phbN6+TJhFrayuUZc0nH3/G9s4WQRBgmhazSWdw+upXv8qwP+D46ISirAjCgLzs2gVVmtHK7v5TFDk7Ozu4rs3Dhw8p8orZLML1POaLGcvlEt3UuXrxImmccnoyZpplLIuaVIJqapiGxmposeoLFmfHNFUFgCoEWdaJaXVNR9e6OpJ49v+F0LvYX92VrlVVx7I6vmtdFChAUuS8/SffICmXnL95hdWeS28loK4kSqujqy5JVPHq7RucHBwyOZlg2waqVrGcTrB0i6ZWODudsff4LgcHdzg9nrGcx2huywuvv0wWp+w/3sP3fJbzBbqmE7g+jrB47wc/5PHuXWazGbPpFMcNiBYZnhmynC24ffkWn+3vshfPOC0yLMXCxqBFoW0VomhJXdfkRclZFDOJI/yr6xR6Q1NWOI6NH4QdBPnwgEZKhKFjWAaGpeP0LHqrHpalcvnyc8wXEQiFz7/0Oe48+Ag3bInjCVe2X2F5lDCZPeb49BBZNQw3R6gKXH/pFnldsXVuhyvXnuPTO7tolstwtMPnP//XuPzc80gyHr3/Lgc/fMrssymGMDA0lcnZGYrScYMMy8BxTDyrgz0fHh1huzbj0zM0ITBdF0UIdveecP/uPbI8IYrmhL2Q733n+/RHIxoEvd4IoSoMV1dYXV1ntlyiKJ34qCoadu8/5PqFHVyjYXF2ik7FxRtXP0/aqwAAIABJREFU6G+tg26iKFr3oTEt7K0N5mhUZsC9JxNqPUCzXHrDgJdfucnh8REjX2NzqBFHKegqZ+OaW6/e/ksX51/5zvnv/PZfb7OkwNAtWiHRhYKqaOzv77O9vcXa6gpZ0eVa0zQnCEIW8yXQ8vk3PoeqCH7yk59gOzaz6YI0zzpfSdNF0+qqwLFMijTrakwtPNp7wsVL51jMExRUVNHiuS7T2Yzz57aZTWeUsiWnZRZFCFVHMyx0TWW9ZxPYXSRMpSVPc4qiQBOCppEUeYllWZ2nREpM00RKSZqmNE2DZVkslzGapjGbLej3B5RlhdBUkihCVbv2gVQUvvxf/kd8tL+HVAXHJ0eMRh6O1ycvcsLQo8wlVZmhm0Aj8Byf+w/uEfZC0iRiPj1lMOzh2N2AJVwTjKMD6qTkdHfM8d4ZTdGgayqa3kHImqakN+pRlAn9lR6u3+PpoxNevvUKf/Evv8+vbr+G8Cz+7IO3UZWWZlnwWy+8ThklVLJF1h1kO+wP2Tx/jh8c3yUNNZzAA1UliiL6/QESqGWDKjQs26KqS/y+Q3/FJc6XVBWsre383G1q+waqUjOdLWhSB6dWmZ88QTcsijzHNgxsz+Px3hPaWrK+ss7FnfO8+PLrmKpG2zTsHe3RiJK9J+9z8N49vvO7P2Lo+WiGQq/v4RoWdZ6ztbXGKy+/gKY23Lv3GS/cfJEwCMiqEl0zSNKYjc1tGgQ/++gez129Qp4sGISdxNlzfbJMcjbtrOSvvnaDpw/ucfXKRRSh8/jxY3r9IUVdUzVQo/N475Sh47Gx4SNESVWUzJYxy7zkwpXrLOOG1nBwR0Pm8ymT2RLHtmjShOVkimrbeI7B8eN9zm0EXFzvszpwONzfpSwN3vytf+8vfef8KxfnpXPn21vPX2Q4CAjDAacnE0zL6LZcTbBYLhGqwLQsomjJysoKUdTR7uq6o1/btk0URUwm0y7Q3MrO5FQUKErLZHzGfDpF03Rcz8PzXCzbYrGIWC4jbr14ix/88B22trZZLmcsipKmVdF0HcPSURSNvqOx2nNI4zmz8ZThcIiCSpnnOLaNoipEy+gZS7bjBymKQpIkzzT1oCgqZVlR189yuZZDVdYdyV1TOwKf0CiKgriGZVnzO//1f857jz5GEYJ/93f+A/7X/+1/wA8GHJ5NsQxBL+yOt7PJEss0qeuqY/SWFU1dopkao/URedyAWhJuWyhqhVI0fPCjD4lOO/u25ZjEaYRpdjR7pQXFbNBtiyppSZYRN29cZ728yv2zY37ynT/F3RhR5SVf2rjK9WvPs/d4l7ooUE0Dazjk5X/7q3zjvT/F1C2UViWvSzRdB6GiqgJVqBimTisabE/HCbpam+f1qJoWtAbP8SiLAsuzGU/mbPWukp7WtGXGcnFEEhf0woBrV6/jOS6qaBFVTd+yOXj0iPW1IQcHT9i4uMOHD3/Kg598ynvf+Aglk4BEqCquaxG6DpoA37Noaolp6OxsruJ7Lq5lURYFQtcYDAddrrttsd0+y7Rmc3OVO598yMXz58nyjNX1NVRhoQgToZuk2YK+a+M7NmeTM1QEUkoMw+DBo8eUqsurb36Z3U/epeeYVGVJ4DsIFfIsRxUmmumjaBqeZ3I2n2N563xy9zFu4PPeOz/i9q9/hXDYZ3Z4hGbrLOYpVRUh6oaXLm/x5he++ouHEN757rff6o55DQpdIXk+XyCetQmStHOiqKrWJTcch+2dbcazKbNF3FnCqoq6brBtu+sQFiVxtEDKjoTQ1J0E6PLly7iuy2K5pJGSxaKDUY/Ppii6zuHpGa1hIIWK59j4puA3fuXzOJokXkZYpkBTlO54qqgIVUVKief5RMuo0w2qnQMkjmPSNKNteSbQhabp7qG62UGjDd3AMDtCuioUZvMlUnc4mqdITTDwfR6981MCyyY1Wu49/BBZVai6jqxLLEPv7kBGBxU2LJPN7W1kVTMbn1EkGUWUdpNuKtK05PjuCevhDk7oUlUJWZJjWTa2Z+O6NnVZYhomUpHPcsktfhiymMQYK3Oq04KkVhht7pAnCTRArXA2nqFqgqzIKFA4aSQ/uvszhpsr3Hj5BbwwpMgLXN+lbRtqWaGbBo5roRo5wcgE0SL0rj7o+iZrq6vESYWqCyy9x3ObbzDen7I+WOPi+Wusrmxz4eJlPNfBBCaHj3GFw8N/9R5GlHDy5BBndch7H7/L1/6nr/Htf/pDHr1/SFO2WJbVfYRoEUJFyqbTVqgaqtARQidNEtIsww/7eJ5PVRZ4vkdZlowGKzRVies6xFFEGARs7WyTpRm60FguJsi2oa4r8jilUWC0MiKJM+azOYahMVsuSfOSyxfPI8slhmUzGK3hug5N1ZX0FVWA0lnLDFNFVbp18uj+Y9b7FtfP9VjpOcTxEs+20DQTvzcgaiAcjOitb3C4rHj58sVf/Fj77/+d326lbOn3+piW0Q1r5nPalmch9GGnJEhzsjzrYnJFgev5FHlNHMWoSstoNCLPn2kHnv3B54sZZVkSBn63mJ4BhF3P58c/fp/ByipZUaAYBlXbIkwDz3VwLBOfgjZbItuWXm9AVjc0dU5d1gjdRCgKrWxJkwRNiE7BLiVl2U0HFaWjuUnZ/txxImXbCXUdD6F2i1VRBIskoZBweLbEcVwG/YCmqmibCkc3EJpC781r2BdW2N7ZIKsKnh4cMRmfcfBoj2ISIXQdw/OxLZsizZ75UfTOHZOnuH6nI1QrHVUR2AOfG7evMB7vcu/9T8kXKWVZYnkmhqZQ1Q2a0FAM6A37HO2dsH5V4dXnv4Bev8Hi6Al//q0/oc5LXrnwPGorODo5ZTmdElxe59qvv87h/JhKFtiOhy40qqrAtAw0sxsAtaLB6wkcX8P1fUzTIY66JFSeZ7RKxfXnv8CTvRNeuPASBw+PGJ+MuXbtAgfHj3i6v8fxwRMuXrjEpfPXqOuKWbZksreL4bSYwwA7h/d+7/vEswVx1BAlGU3b7ZhK26KrLZ5lYpkCy9A7b6shWFsdoUqF4hlsLQxcXMdACAVD72BxtmVimDaXrzzHyekJ3adXwXcd6qpC1y1QFKIkwe2FBGHYSYbbFtFWHJ7NKVULx7FwLJM8y9CEoB/6NEWGZeqgqIzHMwzToGxqzs4i8lJy69o2TbbANCzKpkR3hyiGSZk33Skjl9RKh+GRqPzmS7d+8Z3zj/7wX7yFVHEcj1aB+WLKV77yFT7++CM0TSdJYjShYZgGG5ubjCcTVFVA09LKFl0I0qwj7UGLKlTCXkAURZi6jqIoxFFnF2tp0U2TrGrwhyuMkxSpa7SaYDTssdn3+LXPvcT5FZ/xySFCU7EdD1NoXaM/SdE0jSzvyID/uiepaxq2YyFlQ113RIC6ls/+m64qhkKe593ifcbfzWtJqRrsHk3JCknfc9gc9ZBNTp1l3W9UFdoWVne2OGlzPvrgA5JlxGI6oakqwjBA03WkaKmKEuqWuqyoipL1lTUGfsjJo33KtELTDfKiIgh7KFLh048fcP7yRV58/QW2L2wzOT5FU9pn2VOdPCkoipIgcCnLCt/1macH9OzLCFXwzrvvYAEblk2TzJFljj0MeOE3v8Sd/c+6v59lQwtpniJ0lbAf4PomdZvjBBLDkThOSJnVTCcz4jjFsp2OE6yZbA6fQ8kMvvmNb7H3+CmTyZjF8oAPPniPYW8dx7YZT0+olZy4nDFPxkyTCYWMWev1We5HPL7/mGG/z+zoCEOoeJaFaei0CqAKWqFRtipx2dCqGkKzOD2bUtUVQRBQSolE4ez4BM8PWMyfOTWFYDo+4ejggLaVCKFi6BrRoju1aUKlNwjRdJ08y+gHAfF8gWEajEYrxHFEulygqAq2ZVJkGfFyialrFHlO00DdNNRNTV1XlIUkzeHcziqeq9PUoBsahaJSZhKpNGhtQxOdcPncJtXsiFAt8DSFtbXtX2Ig9Dd+q9WEjpQ1585v83jvIefOneP09AzLsvAcG9M0SJLs5/lW3TSRDaRpgqZp6LpBlqUYpt4ZpW2bssgpshzbcSjygqqpKRtJowiivELRBEIojAY9qjTir/+1z3H85DHHx6fYtvnz6apt21TPprHLKEYIQZHnGKbZoTigq5I1DaZpEUfdHTPLctoWHKfre5ZV08mKmpK6kuStYH+yQNcNLE3gmx3lHaXbZRWpUpYltmd1wOsLq/zqf/z3+drv/h7bG+ucnh100Kiqosgy0qS7O5Z5537RhNYVw4XAtGzqpkFI0FHJ8py6bUBXGfT6pNmUm1+8xWAzwFI0fvDH3+HJ3V0UXUNtW1AkWuBTZhlXXxDI5SWe/nTMaRYTVhW3d85DA0ldcWpJzK0BzihA1TRsx+4ksUVG0PNRVAVFr7H74PdMZK1RZgq6aVE1BUITgMBxfVbCC7TLlpP9BS++9Dnu372D4xjYnuDTO++DUqHSkhU5a+sraJqKJjQePLiDbQm8TOGTP/qMqugYw7KuUWSLgoqiKDS0aIbBaH0dhIEEWkWhqVvaukKRFesrK2RVReDZ+FpDzzYY9HrMZ1OuXDpHWeUMBgPaRmF1bY3JbMzG2hoHT/ZwbIeN7R2qRqIaeicAbiS6aVBkGSgKmmEi1Q72LRtJnuUowHBl1BnhpKRFUtUtR0czXM/HD22aqsC1DTRVQWiC2WJJ2Ovh2DZVUaArEkNTyLKEppHcfPNv/+I75+/9X//7W5qusHNuk9lsjKZZjM+mDIcjnuztEQYBaZr+nHqu6Z0iPU0TbNtCaCpZnuK4DgBJEqMoKlnWyW7TrCDOcvJWIaoaCtniBS7DQcjQNfjqF28jkymL6RnHx4e4no8muv6o47gdk6iVVE2NaXaTQaHraJqGEIK6rhFC0Mq2+whUFaraXfhVVaAoCrWsyfKCoqgpERwtEmZxjmHo7KytYGkacbSk3+shm5osq4D2GXFBsFxENAb0r11hbeMiUbwkizKaSpJnnf2srip0IaAFQzMQQiPLcrIkBwmyarqKGAqBHyAUgWUaxHFC05oc7+5z+dplzJ7HpauXOdk/ZDGdUVUSKVuEopKnFeHAYGdzlaF5g+ODfSzbxFd0alrSYUh4bhPhdLSFoBdQlF2JXRgamqFhewLbV1C0Btvx0XSXupbdB0CoOE6AKj2ubL7K7p2nzI4XyFLh8GifzY1tXNcjihb0+gF5kRMnGRsb64TBgPF4QdtUZIsFZ+8dcPgg4+/9F/+Q2oJpHKGZBpplIIVEMzUsz0G3DNJoTpMuWfEdtlZXcB0XN+ihOR5Z3aLoFlWr0RoOhm1TljmBZzMajoiigqdP9rlw4TwHh0+Rzz6WTV0TBAGWZVPLmiiKKfIchS6zGy2XNHXFcDjo+px5QV02zOcLJrMp4aCPZprPlPEKk2WKqlmUeYLrewyGQxRNEMUxZVnjegF13ZAkMXWRoQuVsig6zUQDa+ev/xIJoa997a2mLhn0QxbzBYtFTEuLbZusrq5QlhXNs87k1tYmR0eHrK+tkmUZhmFSVV36Q1FUTMPEMroUz9liSdZCWlc0QqNEwXIc1oY9tvoeL1/ZponHLGZnyKpGti1B2Ove6PKSoqowTJ2jo2OaViJld0x1HJcs76h8ut5lI6uyoiorZNP5TkBB07QulgaUlSQpJOM4Z54U+H6ApWm0ZYEiG1RVwfVc4ijqkB6KgqJ2CJbxZNq9l1YN1k6fq8+/QN2C5wfUTUtTd4MbpEQT3Yer49N2UC/ZSFrZ/lxR0bYtUZwQxynxIsHxhjhOj+uXb+K5HoarU8iSay9c5+77n3alAr9HEPZJ5im94ZAsyfjRt37G5RdvADUrlk+sq6zevsWySmmakrapELrA9Vxs18YPPZxAY7Tpo9mAVAEdVeioQiBUHUU6aE2P2WGD3vYZDdeYT084m+6ziOaMx/u0NAS9AM8P2Nq8wGg05LO7b6Mpkqd3nmCnKt/6/T+nET6/9jt/l7u7j1FDjyuvv875V57DWvPwtocML++wce0aw3PnSdoaDJWyzomjCdHkiDqLWRv1WF8b0Bv2WNvapBGCcVwxjkqyGsqyYf/kCHuwyqJWiOZLdFXh0eM9zu1sczY+ZTgakmYZdVWjCY3pdEoYBJRFieP6tLWkzFMGvZCqyFlGcy5fvoTjOOhaV4Xu9Qesb2x2baa6gLYLR7SNRDaSIAioqorFYo7vORhG9zdNspLh6gbzKGXrwi8R3/vDP/jaW5ubm4DSEdGbGtd1SJIE1+neHnXDoK5r5vM5ZVnie273VZCStu2Uf6raaQMtzyMtK6KiQAqBVBVUTXlW33K5dWkDrYxA1pyenmIYFkEQQtud7yeTGULt8qeu5yFUlbIoO2mp7T6j6XUDpzzPSZOs+zCYBoahUz0LGKB2Q5VplHCyyBgnBcK2CHse165dIk073EmrtKi6oJINmmNR09AqLWVZ4touDRJFVRn4IZqlc2+8z1d+9TeYzOcoQsfv9dF1HfPZ3a6VLU1To+kdeNq0ut8nZUec1zSDwWCI5/looqOEW5aO65h89sl9NjbXafWW2XzGhecvIGydwcoKnh1ydnhMb7iOpgXkcca1azd48uQxrm0RuzqtgLZt2NhYYf/gKYOVVbzA4fTsjKrJMb0GPzTx/ZBGdkdQTbUQbYBrrKGWIQN3C8eweOfd73N8fIBj+5i6TVG29AYWGxtb/PjH3+Pqcy8yWtlEbRQ+ee9HKPOa7//+t3mwu0u4MuL2l77ElWvXsX2f9c3z+EHAYNhj9+SIC9evYwyG2GtrXH/tNYaXRtQDjcGta1ibAY3oRFjCUEnOjqnmU7ZHAxzDYDDos3PhPFgO00JylLQ8Gc+oTZ/7u08IfQ/XdXFtC99zOTs9xbI9NKGhaxpC2JiWxWQyYTgYdD1aRen8NYZA1wX9fo9Br08cLTr1ZVmSLiM82+z4VKoCssX3HGgbsiRH1w2iJCEMwo4i2MDe06esrq9hOw6DlXO/xM75T/7xW1mW4XkeZ2dnCLVD5Gdp2lW5io6+lyZphwhUVKI4pUWhlg1pklI1LQ2wyHJO5nPSukYRGqqqoKoqjmPTC3wubK3iCYmpi2eWLo2qKMmLgizLaGl/nuixbQspJcPBkOl0+ixnW/Kv/Z1xFFFW3S6lG90xrijyn9MSlknGNC6Z5g2JaNH6NsNz64SbI07jOVFT4G+tUdoqsVoTazWFIVjKnNYzKVqJMHWiLO3I7XnK9Ok+t379Ta4+f53Hj/dpZIMi1O7jY5uYtvWM/Ce7E0ctaRrZHd9M7RkO0uTFl17F9zwmZ2eomkFRlyi6xcbmNnv3HnH56hXQdYoi7/CeheSj9z9C0QTHJ6cEXo/J0YRPPvwUy3eQnkWSZxRVBUiSLEHVNLbOned0fISkZbjh4Q/MZ1AxBUOzEdg0sYFVuRw/2OXRnY9xTIWzsz3ybIHrahRlCgq88uoX+MLnf4Veb5UfvvMdPv/qy3z9//yf2f/sDj/8xvf57NP7aD0Tx3N59c03uPnKi/zLb3+TX/nVL9MbDrj74B6qpuP5ISjQNBV5njGdjYnTDNkaZHmDYkhGV7e5/IXb9C6sc7qYYOiCbD4lOTuCdIYuK3q+z2hlxMbGJnEcM5vNQDM5mkVERY1uecRZTlYUHS5k3r3R51lGXXenJc91cFyL1bURUtZYpoFQYblYdMIlWZNEEUrbUhYJyBpVBcfUydMYoSjEcUQv8LEdk8EgpMiT7rTUNORZtyM3dc1o49IvDvgyjC5wMJ1MkVIy6IdEcYTndkVmyzIxTIM0TToRbtbd+VBU5ssllZRoVo3QDbKioN8LcQyBoCGtGpIKVEUQuA7PXz7Pw49+wtrKiJPj465XaduUdZfmMVQVz/O6MX7bNSqSJPl55E7TjGcLuEVoOoqqEPghURzTNJJWESRZRlZK5nlBIiUEJmZoorkmbd8itwwUo48T+CxlS2kaRGnCaG0TFBWZxIQrfY4PjkkNk6RNGPghi/EY29SZPTllkYGkZTDsMZuccDoboyothqnjBBZSKbAajWSZkKUZ43GJ63lUeYHaCh49vk+0jJktFhiui6KrWLrKg4efkS2WnDw6ZvW5NcRwA1Oz+fTpz1jdHBDNEtI0YzoeY9oGRlFhuw5n4zGO53F6ckS40ufqznNMFhOKOkGzVHp9B7/n0TYSx3apliVNJZF1xdHeUz756AN6/R4Iiw8++ggvcGmqmhdu3qYoJXlZ8WTvKS4QJznpJOLo48e894c/pNQrVEdg6haD1VX8wSo7126g2C7nrlzm29//Hm9+8Utsn9sgTSNcz8K2POLYRtNtdF0jy1KiaMFyMWY+O+Q0a9jZuMbB6WMWrkLv+kWmn+3jZ4KqkixOThkfHWG5DqrTJ1os0G2HixfOoYpOy7iXVTSVZHE2YaYeEc8XHEwX2Cr0goCV0QA/8EnTiKeP9zD0TsOhaV3S7ORwH1VVEbpJVpaoKoyXc8KwxyTLcR0bXWkY+S5SNswn4w4/Oh7juw6e5yHWV8iSCNNx/o3r7/9357Qsuyu2StnlVNuWPO/efHRTRwiVNM06SpqqopkmeV2j6hboOmgqhqEz8B0ubW+w3ndwRbeASqmg6gZ1VZPOzrA1yf7TPbRnOde1tTWStGPnlmWJZdtomkLTSKbTaRetEwJVCAzTZDga0kpJXhR0R/HurjlZxiySiriUxLVEDVxaT0f1dVq1parKTslet1R1ixf2aGTXX53PI85duIQX9DibzXE8H9N22bpwiULWrO2s01oGIgw4fXLAq1/+CrVseHj/Y9J43D2g6wbLaEmWp8i26epaSo1mdPdXRZVohkAIwXw2p5IVrWxRhMAJXe5+9BF10yWZjvYP2Lm4jR96aIZN09Q4rk4aJyRxgqrqOJ6LZbmoancKkW2L7dusba/RtDXhMEQ1W5xQ4PoueVohG4UyLXj3T7/Pu9/6IbuPHiFUnTIrWF3Zot9bwXZDLNtnZW2VJEnphT1W+32+8f/83zz8+AOK8YQPv/s2P/judxChiWYa+GHAcGMDf9hn7cIW48Wc7fPniJKY56+9hKGZPLh/lyLL0LSGg4M9Do6eUpQls9mMvIgpipSqypFSsjpYZ7mcsFyOaaqCOE7RhyGFUneDuyDE0jvpcSN0krJ7BbBoWLVsUASXnr+O6/tsb22TSTDDAZOkJJUquuOwSDNUVaMoa06PTzB1nThJKUqJ53iApG5qTo5PqPIcXe1ctq5jodIy6IfIqqQsEubzGSvDPovphFG/x2I+ZmVlAHR+HlUIVrau/uI7Z56X1HWDY1uYpk7bgGmaqKqKaZrkZY5t2d3TRdMySVJqOi+K4zr4moGja4SejUBSRVMqW6OpCixVoLY1adVih30kDVnayYA8z6WquiC263a4+i5b2r1NxnFMr9fj5PS0m9rqncLPdV0m0wmqqlBWNVleEhcN86RCCBPbtdFMaHydZTRDEzpN0yKB5XKB56tYlkPoe0zHY3zPY2Nri/lizutvvMHj/X0Mw6c/cFnd3MQPh7z9gz/nwuWrnD/3AtPxLqcPPuV73/szXn/zVT744Igg6JHECb7vc3qcdO+/ukZ/MEJS0jQl0IBUkJXA9RzSLO92P8NiFAYsPRs3CKBtkW3DvU92uf3miwjRTV1tQ+Gpttc9hmc5aq0SeCPC4YDT2RhF6SRIdVshWhBGgxMIWiHIM0mTKzRlwf079zg9OEV3OuW77dpo1hZnsylDwyBeTMmShKsXL7OMMs4ePOHs4JjTDz/jqWy473yCZgoM3cF0LBzHRrM0rl6/QoVkuLnOzVsv8d57P0U2LYYG73/wA1oJ/XDI9vZlVM3GDSI8p3PzeL5HWWScnEyBlMFgmyib0+YRlt+SLnNcf4OzJOLqr7zM/r1HbCsOolLI6+4ur5oCNcuZnN5hGkWo0YzxbMrN69fYEAqGZ7P1yq/x4MlT0rJGKJL3n04IbIOyEhTTmCrLUBUNwzQZDX2yNMH3bITSOVW2NjfQNJW6bjC0bp5Sy4rhqI9hiGdBCRtF36R+FojphwGzxfzfuP7+ysWZpgme56GgkEQxiiI6vothoBsG4/mcChWpCqZxgtQMXNsidGyELNDaBstQcE2d2WyK2pToqkPbSpANliLJmpZlnKA7gq3AwzR0qqpE13WOj4+I0xzX7fqbWZSjPoNGp2lGVVakaoatwHxeMJlOaKQkLUqWSUFWqZStYLQ2wjFV0CWfHO7StF1Er5YVlZSYto2m62iGwtbONmmaM1pZIYqXNElKFCUoaLz08iscHewR9s5x9eoLfPdPf5+skGxtP8fq2gVW19cZ9nqoeYpnWuycvwSKQi0limzoDUKgxXW7j09ZZzSNoKXAtjXSKCOJpxiWjVRVQJKlKf5gSNgPOXjyFN2wONw7QlQ5t7/yMsvFlNB1aVDRDIN4PsczfI73DuiPRigqqBromqBVakxPx+2DbrXMpyUGKk/u7FLFBdOTYyzLxjAtDNOlLBqapqZYLllWEEfdm20kQn78ne9So6CZFqrn4KkKUijYrkklG3qDHpajk9YZK+e2UXUHP1gnTUBRDISoOJkc4AYeN6+/ztpwyMd3f8J4ckKWLigKk93HDxitDLEtm1apaJB8+vADdEMi1G7CXWgp4+kxiq5g9n1YczhICwaVxQc/+ojeYAsHg9nRAUOlpWcIPv7ut+j3+jyYnWI5JqOVNQaayuLhLhsXL6F7Pv6FS5yOJ4jeGosoRjVKRr7FrEwoJ0tsTUUzLGRVoendmkifzWIODw8ZDofMZnN8z6Ouiu4NVRO4rs/R/iHLaEGv51OW+S+3OONoyaDfQ0qJphuEYY/pdIZhWcRVTY5KnKTPHrQtttZWMAVYakuWSGh1+sNR17mcNpiGhq6pmJZLVTdkVUSpqESyYZnWXFxfQy7PqGVD3QiapvvhWZZ17YikO9p5nsd8ERH2eqRlhxmxTBtF6B2qs5RIpQvSq1mKb8IovZ7VAAAgAElEQVQ8nrIUDdI2aZUWKWsQKsIQ6KaKQkucLFhZGzEezzg7HSN0wf7RAb4XsL65he15ZNmSg6MjLl5O8NyQzZ0trl1/kaoSLOKaR4cRL9z+GySLjCQqccOAre1LlFVMXeYsFjM0TaAaAl0KsgxaFISuYHstipZhCIXZJOkme3mLogsW0QJVFwxGA4os5f333sdfM3F9E003GA7WKOKapplgaganh/cRgYnStqgqVHWOG4YMNjqb2vwo4+lnT6jTlqO9fUzTRGgCoZkI3cW2ByiVRhtlKLOC4/sPqaSGYuh8MPuA2jTY3tlmuYjIywxFKLRSgqagSqhkgarA1qXLrG9dpZIt+3sPEcJgbXUVw7AQWkWDTpbPOZsmOK7A87rJfFosWN9eg7b5/zh7z19b0/O87/f2vvrae6/dzi7nnDltzlQOZzjUsI1EWRBpipE7LBtwhAQOHAGWkw/5EE8CON8iwEGABLEFKLDVi61CiRIlUqTE4XB6Ob3ts+vaq7e313x4t4h8sAiMzp+wz3uv53nu67p+F344L9vVLYfAEyAHQZDJ0pSllWVcz0VE4tbdm5ClKLrKUMuQN1VMU8fBYvLwHpohU6vUSKKQZrNeNmEvZhRxxuNHewiiiOovqCwtEwz7DN2AeqPJzs455oFPnORIRo2eN0OOYkxZII1jWpqESXkLm8wm6LpBtd4oDw8/oGk2mXgzJC0gc100TaWltX7g7/7r/v3QN+fv/PqvvWYYBlmWIQhQCDIT1yWTFUa+j6goaJrCE7ubVDSBuqGhCBm6IqHIMqquUalYqLJIGpcplCSNMUyz9NJSvj2ToqQKqIpEEfvkSYhqmBTkyIrGdDJFNw0Qy7V2mmXEaUqcpGiqRpRkTNyQWQyTIMW0KzQqFRxNxp+N8QKPaRExLWJysQApR1BEJEVCNzREWUQQBfIiw65UmEzHjCd9arUGgqKQUFBv1AnjmFZrjTBy6R7v8fxLP8b6xjbvvv0Gnc4qjUadR4/2MHWbd97+Sw4fPqDeWGL34gUm8yGCVC7AcnIkWSlbv6TyNpKmOYbpIEkigpIjqgWmpZCkEbJSGikcp4qqq8RZhF6xCN2I3d1Vur0h7XqL026PwPVRUpHlxjKLLCDLUxRbYXm7TX2pRrQI6d8fMno84uDOHnmakws5VsXg8tNX2L18EcUwCEZzgsMB4ckEJRQICpAdi0IWSIqYtY01NENjMp8gKgKWZSLJAoZjojsmmq3z1AsvUig6aZFRqRl0j/d57rln8bwQ07SpN6r0eseMRl2SJGJ3Z5v7D25h2jb1RgsRSCKPJPRZzCbEQYC/8BBzgSRKWMwXKLIGuYi38PHdAF03ieOyC0av6whGxnKlhtSbMxr3qVer6LqKLIGmqAhnurcgQqtVJwoToiDk4Z27DE+6CKFHTS6IpiPUAurVGqplI9k1xEqTRS7j5SJunOOlOWGWk8QpuqGR5wmT6YKt7S36vR6GbpTAdFXBsk1ESUIQRJY2/gYmhF/91f/4mgBUazUWfsDUD4gRKGQZ0zZ5+ZPPIiUuG0sNFHLyPCXP8zJVoOtlv6HnsZhNS06sJKCqatmilebIooCiaqSCiB+nJHnB9rkNJsNeuQARBJK4TONHcUScJPi+X25Oi4IgjBjOfCZhxiTIEBWVSxfPoxYZRRAQBwG6bVHYOsPEI1dAUsvTUtUVRFkqCX95hqLJCIpQem5JyfOINI1xqjU03eb46DGGpSGKJlKWs+g9JpVirly+zuHxXXrDhwhqRn98wu72Fpphc+36k1QrBo/u3sb3XM5tbeN6M8LYL5vYxDJBE0cxYRQjCkrp2UxzBAE0Q0aQEuyKThzHIEgouoIoQ0LCslPj+OERWhXkQqZ/OiD0A9yJi63Z+GJIe7VJJiUoukwyjzm8d8i0P2Exc1FUHdMxaJ1rsbTb5NLz5zl3tYNkxWhmQmu1Sn2zRft8h7nnkgsFmZBi2gaqoTEY9ZFlEUWR0E0NURFw2hXqKzX0qs7OhYtYFYcgWZCmKXFWsH3+MrOZi2WapBkUec7S8gp+EIMocnR0SJ7nhH5AHAWkcZnHlcSz9jFJwtR18jTDXfi4swDPDZlO5yRRjqbpJHFCmuQgyMxdn1xJUb0IfzAvtUhFQhZl8rxAVRUkScK0LRzHQdV0XNej4jhUbAshS0kDH28yJZzPiOYTLEBKUrKodKBptRZaawkMB8moEaAw8gJGbnjG0JKYTz2mc48kz7Atk9D3sG2HVNRp/02kFFEUCdOUkevh55DLMlXL5Orlizy6dwM19ajbOvPpGFlREQSBPMtIk4QkjlnM54iiQJbEZ63REpIolQNWAIKAoGRogookQBBlTLwIZI1cgDiOmYxmNJo1oigkSVPyvCDNcsaLkNE8JExBNQw++YmrPL53C7NICJIQocjRTRO9WWU4OUE0JDRVRFAkZEVGUiQQyu5gSZKwK1VEWSRNUjRdxbDq7D8+Yf3cBeIINA12z+9y1H2IbAnsXr+A0TZ5cHqDz3/5c3izCE2V2drd4eR4wKs/8VWmbpfjg9t03zoi9Hzimc9kekp7a4mEFFkxsCpVVMtA9uaEno+qqJimSZqGBKGLbpZNZEghhZAhKTL1egvveER/cMzeR0d8evk5eu4pnZU2s9GMaTHlYHpEfatKtaWRTT3m/TGTuGB5vVOW/+QSS6stmqt1BCdHtUVm4TFRvw9GQeepBppcI/RFosDlcmeHurrJ9//0ewSux2w2pshTFFWmEAoUQ8LQdXRHJRdTFFVnNOmTFzKiAvP5CFmRuPfgPcgzgjhFkgTWNlaoVlssL0MQhiwtnWO+OGU2HVKtOUiFikCOKhTkeQG6SJ5myIJIFZswiDAMCytRzzTuGSCgaQZZXCCicjyasOKIqFapVZKDYdl0T7pQlL5rPY5J05xqtV6mT2oVgjBEVcoEkSJShhZcj3F0SJxkmJUqimmQzee4mk5SCDi1Ckub5ylkicl4TBHMeTCOEM0l9vf22HE6dOpbHLz/JqLi8GgccvmZ//L8/dDhTIBcLE8XxbZ44foVZsNTCm9Ew1Dod0+o1qoMhyMcW0JRFALPZz6bk+UpSRyjKjKapmBbJQs2iiJEQcCpOERxjOt5aJqIpUq4qcD+cY+GIqEUCVmSkRdZ+XHmBZKo4MUxJ/MFizDFsGz+zhd/jG/9yR+SzfoQhvSOjykKiNOMIIko5JhZvEBQBZBFBEkAsfxtEEWQZQlVLzOS2ZkAnaQRy6vLRAnYNYcklkiyBbF8ylMvrnLan2Dqyyh6FS8MkDSbpdU29++/i2HK3H38PpevvsDe/j00RedTn/kUX//d3+P2+x8gCSKz8ZzaSp3tK5eo1NcwbYvZtMts0icKPCAnzwVU3UBEJHADTNMmCF2SdEqvNyGJR0ixTREUnN4fcfW5LW7dvYtYFCi6wsbWCs6qThFnRIuUtc4aWV5g5CbBMKCxbNG8YGLURLIsRLMKCnJ8H0RJQRDU0gOaBSiWQeeSgylrPOGv8fAvHrJw5yRZCkKKbpkopkmlXsOuVuie9vjktWcZjCcoisKgP8IwddIsosgKKjWHJEuJ4gRZK3GYVafK/skDzIpEtbXGUlSlyHO8xQLfXRCFIUmSUGTld1XkGaqqoBsOcRyXmqGuoigykqzg+wGmWqEoQEkEZoZIbcVBD2RMRSdOy9ifbtmEUYiGweHRIUVRUKnUcCy7zH+KMr7nIysylmGgafrZ8EvoskTsugh+QIZAkmSEY51s0CNRVFJBYHl9E9WyySSRDbtJIYo8nMSM5Sb9wwVGtfXXzt8PHc5CVTE0mU+/+AKPHtwlHJ8SLSYohoEIyIrMfD7HNE0m0ymyfCYFuB4SAlatUr6hAP//Rx3gbEh93yNJCyQC8ignxySXFQRNQQhDPC/AsR0WbkCQCPTGM2JZRdUsvvTqK3z9d3+bkwd3kLOE8aBfLg88nyTNEU2NceYThwGFJoJcXl8FRIq8lE9WV9dJ0pS0yEnzAt0w0FSRuevieT5OTWIye4RmyaxtrzCLH1DLrhDnMRc211FEk7k/J4kF4mCOZW3SbtVYX59z5/538P2Iwq6DorC0uYKQ5wwOBwyHIybdId7xAkH4AFWVCRZTkjBk5cIaja0VdMc5k48SilwhDkNEcryZTxy6IKUsJjO8yRQhzkjDOYKaMhmOypKjICOaJwwOxgTjmCkz5rM5QiKydWmZzU8uIdWkcvOpKDSWKgi5jJCrnA4GBKFPlo9IshTDWkaSJMbzHs9/+kVWWit87Zf/GD8QsCs1dNsiTQUWboRuKXQ6Wzy4v0cURayunUOTKzx97Vn+5Ou/iya0mEcSta0lkHSqVofRYEgcpUiigm7IxLHL0lKL6XwAsoJTazIaFizGCWQCsijiLlz8ODz7MU3QNBVdVUsnWJagaSKSmJHkIaJQkOsSwVoFKXOY3+8RzsvUkiwJVB0bWRTQZAlVlQmjgP39fbSzjtcsL8jiFFXJkeIERSnTVnFU2vuEIqNZrRJ4LlnswyxCl1U0zUDpHREgctIbsnHhAka9hqDZ7Dy1TPekiyQrf+38/dA35+OHN14rghnLFZNxr4tpGsRhiO04GIZJGEZouk7gBz/QPiVRIokiNFWlXq2SJjG6qiKKIkkSn51WMltbW2RZwdHxEZqikOYiuaSRF2J5UksSpAlRXvDoZMwiFShkhU++8AJ56FEEHvPxEG82RVUU4jBBACgyvCIhMCAyRARDRlIlEEq0iqZqZYJdFBERcX2PMIgQRRlVVVjudMjJqC4VGHUfyYhoLOusrDdIcp8o9tGNjKwYMZ49JIxiCmHC7XtvsL15nYq5Qau1QqO9wqPDdznp3qUANjZ2SLOEYD6HDNIowZu7jAZ9Jic9vMMJi/0xwiIj6c8JewtYpLhdl70P79N71MXvL/DHAaICUipg6Cph4NJcafHw5D6f+MxTvP2H72A2KkyHCxqVGo8/OCBbBIgIdA8GUIRc/fHzyA2Bas0izSJEQUISdQIvp0BD0yoIaCiyztbmeURBIfBdEESqtSZjd49KRcF3C4pcIhc0nKpDr9tDUTQM02FtdYP79+6S5RmtZoscmXPru1w/fwUhlqjVO3jenOFwiCQYZFFIt3eAbTtIksHp6RhBVBFFlbwQUVUZ3RBRVZGiSCnKQEips6sqRZ7j+x4IOXEcYpomQeiTJDHVao0kzYiLHLFVIZ+HzPpjNFlhOBpSrzVonrVlx0mMIMmEUYJ85seWFRXTMrEs62z4szJ1JUnlhlsQCX0P0zLLWhBRxNRULE3FnUyY9XtE0yli6CEFPv7xY/zpANXQWV3psN6sf/w35+DwIXKecHCwR5TELLXbZGlaRqBUBUmW8D0f5Sw4LUkSaZKiqipJHNLtdhGFgvisHVuWS73TsiwGgwF+GFKvNUiTiCSMEU2dMEmYLwL6WQRJRFJAYdo899RTPLp5k8nRY9LFhJG/YL2zymQyosgpg9WKQphkCI6Kr2ekYoFIjmHYqGqZAxUlCUVVCaOIOE1QZLm8qhQSsqgxng1pdDIKdYbl2HRWV5nPZ5z297DsKkUuUWCRJhJTb0StUgM5Z2l1HcVKuXfwF0wXHyGKIot5j2ZjCVGImSyOcdoqu09tsX93n8hVydMcNZRIgwxRK28qmRfiH6SMiz6BlBMWBV4QIBYZqg5f/uefZ+D2yYsM07CQKwXTWY/DO112fu4cVaOCqRvM+hPiRUK2iDHrdabzADUr2LzSwVxRcCoWlYqNEigoioUkqSiyAIJIkqZUKjpJknJ8fFAmj1SZeqXDaNYnTFOef+VVHOdDbr/dZzSbsJiLLLVWcQyb2+++x2B5mRdf/AyaafHhO2+y3Frhw/feY/enf5pIOOX9d95mbW2VLHU56B1TqTU5OnjIbDrFMG1cb4GumyRpjOXoVCtV5osp9Xody17Gd6Z0jw4wTI0kSonyAFVSUBSFJEiIgxhRFFFViLMpgR9iaBX2eyfYToGWJ0iajZoYhGnCeDolS5KyU0XRKMSCKEmRRAlFzdA0lSAom+nks28IQWA2n6EqJc1RFCRkRSEIfCrVGn7gY1oOumFhGDOKLGXw8C6WIfD2d45ZXu3Q/OR1uLjz8U/Or//er7+WF6AZBqqmwhlzR1Zkwig8q2yPOT4+pV6v/oBqB0X5PsgzVEUpBfc4Isty+v0BaZqSphlBEJY/f0UBBQxnC+KsNKcXCBhOjd2d86hZTDgdMR8PmU1n6Fo57MGZ8TxJSvRIFEUoskocRaUoXTPJZEpSQhwThAFxHBFGMUmSk2YZrueRpAl5ESEbKXYTFLtAEAta7ToIMWk2RVHKtbcgReRZRJoFbK1foTf8CH8RcPXyS2iGRX+8RyHktFpL7B/eQjbGKJLKYLhPmnkUJEhyVpoDVLDrJoajkogpiZKRKjleGDCPwrL9uYBoFiJLGX/nf/wi5jkNWRcwaxaKIbN1bRml4XP+2XM0N1tkGjQ2HZZ3l2hvrbB7fZurn3yC1fUGSxt1Vp/s4KxWqFVbFIVAGAbEUYpl17Asm8lkTJIkaJqJqpqkqYCmO1hmlcBPcaxVVlcu4/kRiqGCVNBy1ukdH2MZFsvL61x44jK3P/gQw3BoNdfY3tymd9qjXrNZ21zj+2/8Be2VJsdHD6noNu5iwGQxwDSr+IHPeNxHFgs0VSsDEHHExuoVWvVt5ospsiLTbJ9HkyuIYoLruVAUSIWAWACFiCrLxHFCGMVohkq1YhKGPmKekwopSphhZxKyahBnOYuFi6YoZGlKIUpUaw02Ns8RBD5plpRgNy/Acqo/+I6SKCJNE5xKFdt2yPKijCtqGlGUMJ7MKUQJ2zZZWmoSByGj+Yws9LEtA1E1uHv7Aa988cc/vpTy7T/7/ddu3bqPKJTRL/eMNiAIQqkPSTJJknLaO2VjY500K5MqRVYOZZqlP6j1kySRwPdRFBnNNCmKkgKfJjGFIJKJKpNFgKCo7Oxe5Ke+/BPs3f6AdDHFnZSV69EPtmelAblScRgOR4D4AwN8GEQoSGReRDj3qDUqqLZMGPgICIiSiKJr2LYFuUAUp6iGSHVFo1Bj7KqOoilAQRiEpGlGmgg8cfEai/kIx9HxwgG2oeB5J8zdAT/y3FeRBY3h8AQkl+XlVTx/QrUmYVoGmi6i6xmGLlFvVFE0iV7vGEHIidKQrEiR5fJtHmUZKeVyKpwGhPMAipT//hf+O8zNmKKQcCo2ZbNOjiiLPP3Mc9QbbboHXdprHZIiptquYTcNGmsNVi82yGwffSXHWTap1AzazQ7zWUycjkGQCHwdRWkRRypkAlD+mGZFRsVuoalN1jpXsIwGzeoGzfYG+8cf4kUCr3zqR7j/+odcunaN733zW6xuXeDCxUss2RWOel0u7OxiWnVEWeX0sItpq7hhRCGmHB7d5dz6DoE3Iy1iNN2k6iyxtNziYP8+YiHhWC2KQkZTTCpWmyQJGQwOMc0qMzcsZSkpp8hiwkWMN12gSDK6qiOlBUIhgiQRxQGOrpNJEU7TYbrvlZyfOCbNhRJhUwikeUlBGE+mjIZjLMvEsh1ERSZMUo5OeiRJylK7SeWMgRWGIX4QsJjP8BZeiXPJc0SxoN/rksYxw34XXZXOAiUFSZ6zdekqV59+5uNfa4u84OLF3dKyp2gkcbnQmU1nZFlJRNN1jbW1VZI0pSgKEMo0iyxJJHGEQAnOSpIY1/fJ8oK6oqGpKpmqECQKR8MFQRGhmFWeevIKt999k7fjCbk7I5EUZEmk6lTwPQ9d11FVlTAMmU5L2nkcx8RxeY0JgjmOU0XKJLR5jvd+j5AEVZPJENh5+jrbn3yGm49usXBjrj71JBP3MW58yBPndzk5PWA6i1E1tdwSU6AZEm8Gb2CaAlmuYBsSFy5uMZnINBpbjBdDLMtGN5YZjCJOj+9Qq2t4i4yl9jlu3HyHwJtRJCALC/qnI6qNGmkcEwyGaIpJ4Kd4C48kzEniDDErKKKMNCmQdXDWKkjNhNPTHrkogZhQqTlsb2/x9jtvY5s1Vld32N8/RjdVao16eYKoKokQIekF25sXmLtT7IoEsselJ65x74GApFqoSo0sTdjpXKNZXWXqHyNKFXJERCnh9OQBvrCgUV/BdT3q1Qq7y5+D7H2W1jcpbIcPPviInfOXcCcLajtb1OtNzA689+FtVLHg4vXn6R0/RhEt5Din0zlHEgcc9h5x+OgxrXYT1Y6JlIDbtx5R5An5md6tKAaWaTHo9jl//glMo8n9vbd54blXScIMQ5X46IPXGah3sByD3sEESYoxFQ0/9HAMhaVWk/l4Tr1aJfJd6s93GNycoQkirUYF4owiB1lSS8WBgixNmc1doiQjL3Lq9SZPXLhAHIUMRzMcx8K0TLIsJwxjbNtE0zXGkzFhGCGL0G7UiYM5mxsdBr0+vckUvbrMNMy40tn6a+fvh56cb3/vW68lZ2hL27bwPA9JOpNMggDd0ErLFvwgdK1qKkmS4LoL8jQtsfmyjGEYNBp1TEMjiCIGM4/DkUvfT8kkDceyefXllzi4/QG2JjMZDciK8oDQdb2ktCsKeV7geT71ep1e7xQQyLIS3lWiL8XyCp3niKKILCsUWYGYigiJRO+wz/GNPdzumItPXEE1RdzoGFnNCEIf13O5cGET3QipN6FSK1haVpCVgKefepLVtS1OT8c49nn29t9i7t9hNOlyfPqIGzfeJY6HDMfvI0ohsmDTOxmQhHMUWaHdWOVor8ukN0EUZWq1OlGQ4piVM3NGglOziMKIIIjJ/QjynKdevMiTX7rAIspwozmaLJZZxUXMhfOXefBgn6X2Bqe9PsfdU1bWVplOPHSzxsIfl7eKTMOulC4hXZfRDJG602R3/UkspcpktiAJQVcqTOc+s3mAY3c4PjrGnc9Y39jBW0Q0a8sIKEynU9p2k4qziW7YeNMpK60OL3/684iyxEn3mNP9Azrr59BNm9/8xX/Lk1cvs7Te5ORkgKrKKLpEmoDvR2imQa97iiZLKFJBvdHCMCtkaYZt6UwmPeaLBeur57l68Tq+F1Gx1+j273HSe8C1yy/y9NOvUogFQT5A06FIyw+oWq/iVC2KrCjND1FWZoMNgd0L1+l0LpJpKl4h0h8PQBAxDB3TNInTjCiMWGq3yYuSYySekSKTMx5VekZqFESRzuoa09mc+XxGHJe9OydHBwS+i2GY3Ds4IkImERQ+/+qP8e3vfIfPfO4zH/9a+6d//J9fm0wnKKpCfzCivdQkjEJEqcxWxknKbD4v+a5pSpImhIFPEoaYmo5hljm4MIwI/BDP9ShklYPhgtNFQiqZGIbFJ568AtGCxw9uU7FNsiSjXm8wGo5IkhKk5LourVaL+XxBkiQkScJ8vgAEdF2nXq8ThjFZXr4l07OTXBBKpIgfpcxcjzwHVdbIwojJwz3G7j5SXSTKPao1ie3zTZJkgee5bJzbJMlCVs9ZOM4Kk1HMae8EVVOQ5Yz5Yg9TfB5VfoZLF5/mo5t/zN/9yf+ah4/vULVX6B8PODi6jeFI9I6HzLpzTg5PEBSBzfObxDH4bkjohqRRSpLmeKGPWTXIs4Q0CkHMeemnP4mxadNoLDGZ7FOtXERWLKJAZbEIUBQDPwlYXW+R5QXt1i6uNyEq+ljWBhvrz2OpS/jRvPRzJiZJtMCqZChKk8f7j+ifdkuyv6ix1N5ivX2JNEiwTR0vmJaxv8Yq4/GISrWCZdoM5sc8fPw9tjfXUEWJB69/yFs33+edj77H6OgRtlHl4NEDDh/f5J/+7L/kT974MwzLwtQ1Ft6M09MTDMPE9QPSNOXc1jYTd8Le7VuIZEiyjK6aLOYuRZ6T+Av2H98hiTzWzu3wl999nSuXP4FlVfnO27/Lg3sPqTfaHHX3ieOU85fPISoas9EIMuh3pyRJiqmoFKnAdDRHkaChSDxz6Sp6u05rY4fE0HHThJPuKaos0l5uEXgeRQGaphJnCVEaY1sWeVFGv8LAJytyppMJ4/EIRAFRlrhw/glu3f6Aa9euECsmXlKwvrZOnoYMu/voecwLn/38xx/O3/9Pv/Karutnp6XK97//fTqd1ZJMcObYXV5eZjgYoOsa5zY3EYviTNf08V2fwHVRpAJRU9kfzHnQWxALJf7wxeuXMIjpPn6ApSll4Nd0ODo6QdcNbNtmaWkZ0zRKnW08BQR6vR6iKKHrBkEQYBgGvh+UvFMRVE2lyIvSXB8lTBYBSQFOpUbNsSCNKaKEKHTRNyxWLi6TCwGimBJFKYIosLS8glWxCSKfolDoHi+o1gx2zp1HFFPu3r2Ppq2hKedY6ywxd48ocgXdNnj86JjBYMJTzz5LmkgM+x7eLGQ4GHD16jXsis58sUCSZS5e2GX9/Bpm28RuGdTXK1SXNDavrPAjX32FpWdaXPrUMlee/CJrrS2COOJx912mkx7Ljcv0+0MOurdotBWGs5B27QJVW2WRPaRduc7gdIwqtznqf8jUfUgUhuhGHctyUAyROExQWGEyGjEYdomCBfuPb9Pr9qlX6ziVCqph02w22Xu8h6wV5da0scJ3v/s7bK+tc+fxn/D0+Rf4f17739m7dZfZ/hEvv/Ayr/7ET/HjP/ll/uyPf5t33/wOX/jCV4jChMBf8ODRbR7v7VGtV1FUhSIvGE2nqIrKxnKD61e3OTg6IAhiNNVg0CtrBWs1kzff/ja3P7rDcNyjUrMIvJSnrr7CS5/6Eb7/9u/Tbl7iylOv8oXP/yPa6wrD6UOmIx9HtXAsi4d7x4CMbVn4oUc6maEuPCaHj9GlmLVWg92Ll6htbGIsrxCkOeP5FMSyoUBCQBFkTgdDiqKgVquQ5TmCIFCp1LDs8iZUtqvnfHjrHk899yM8OBlQW9ti6dwuer3N7uWrGNUWFy9d/PjD+c1v/N5ri8UCVVXJ84L19TXy/AxvGcfoqka328W2bYoiZzIeE/kBeUwZZpMAACAASURBVFbS3FutJqKmcTjyORxFLOKy6evlF56jaSpMB33ms3EJKo5STro9Ws0mnU6H09PuGV3Bp9stc5Gu62EYxg+WUn/17p1Op7TbLQI/ZDKbEsYJaSEQ5wVeFGNYFvVqBbKU0POQpQKyDKtToX65hZuEeNMUUpMkFojSgE9/9rOc9sZQyPT7YxqNdcJwgufvc37jE+hqh+cv/33eu/WHDCc38VyPvID3Pnidy5fPoygyo8kJURRz8fIlnrh2lUaniWwYLKIBqi6jmwqngxO6vRMmszGCCOd3thAlibk/wjBSPvjDW7R3amTknE6PGU7eZ2PdYTL0EXKd/YMDWqsKrfomDW2LO7c/wgtGWM4G60uXODoYkAseQXYCscaLT3+FWqNGo9rmdLCPqou0G1uASRhkCGRoWkGz3mA4HFGIKU6lhSBIQMFyq4m3mBOGGTvbT/L9977J88+/yvG9I7779b8k8CPIBB7d2+f1P/lj/uA//jvc8YjrL7xIc2UDzbBptzsEQUyjXmE2n5JkMZoioykKYRgxmM7pDiZlQkaWmE6nnB4dMhz1EEi4dvXTHL9zi8d3b5GLGVka8Rd//m1ODsc899SXOTw45unLT/HGB6+zsn6Jre1PMJxNCFKX0WhMw3GoOS0OD4+oOg3IBbZWVsmyhDwOSb05uTcinpyysdzmlR97FReZ2somcqXC1HeZuzM0VSRNI+I4wdB11jc2St94lpAlKZqiMg9DNnc26Z885v6Nd1mxJA7u3+YLn/ssv/1rv4KtyVx//hN/s5NzOp2iKAqWZWPbTgliFqUS5uU4Z4kVgSIrqDpVBHLc+ZRcEDkeu9w/nbFIJSRV49rFHaTIY9I9xFQkTL1cuoRRgK6rVCoOaZqwv/+YlZVl5vMFURSi6Qq+H5Km8Hjv8ZlDQyaOY2RRLsXiOCFN0tLNIcgMJnMKQcSxDBxDJfQ9dEVGloVy2yyIODs1iprI/fv7zKc+Syt1mssWtbbGeOwyn7tE8YKNtUsoakGczdF1lXZ1mSQeMVx8SL2xwbPXvsIbb/wprbbC6maLZrNBr3cfTQ0RlRwKkfFgxtVLz+AYFpaj0Vx1MJyCKA+QdQ1V0pnNprizOe5owWKacHLgsrS+xqVnNlBsg0q1hpq08P0F26svMhosWFrr4PouYqJw7607/A///P9gFAzZ7Fzh+OiQ8WTBs8++gpS16bReIBKmIMgkiY+kyNgVm/lsTM3ZZOJOCcMZopAxnfcZz45QZJ16bZXZ1Gc+O+UvXv9zFM1g0D/AUKtkYkAQePza//lLTMYTJFNFNRUKIcV3p0RJRuym3H/nQ97902/y3jf+lC999aeYuhOOTx4SpT55VLqdKAoCf0K70mBjdakM1i/c0q+dp8RxiqJWiBMRrVLjM59/hdAPsA2dOMz47Cuf4/TgIUISctC9xyI84ntvfYOD4zmfffXvEwsy56+eJxcS7n90D1NSkVQFhIJwOCHzQrI0QZWV8klEhirkTA4PSOZDomCOYVpU20s0N8+RGjZBVjAcj5iOJ8zGYwpRQJJF3PmCOAg53N+nYhpEvkujajHxImazMcPeKSurG6SSwjPP/Zdbxn64lPLNr73W6XRot5fOFjELBAFcd0Gvd1pyYAUB13VJk5j79+6w8H3mccYwKDidB2RIrLSbKFlMMB5Sr1TQdZ2Fe5ZUSBKevP4kp6d9qtUqvV4PXde5c+cu1WqtpAbUGkRRzGm3R6vdJI6jEt6lKCw8F13TiJMMP4PRbEEYx1SqNsutGkWWIApl2kwSRCgKhCzHk2JcMyHOBeqNBlu7q0hmwvJalTDK6A3GaLqM4zgsLduc9O5RsVYxtDqn/XskcY0nL/0MNx++Q7d7m0pdp9PZJU5Tcj8kKIac9gLChcywP6MqNXn4nQf8v//zL/Lmb71DeJRTr3aIshhZzBGBRqOG5ThcvfI8K2sVFospjx8csrplUa2fQ6DGp65+Admscf/+CWGyoNKosdy+xqjncuv2u1jtVdTKMoogcHR8kyxUWW5tc/PG61y8vMNg/BDXv8NoehtVaTGde3Q6bSyjiiBmeMGILJVwA5fR+ITxbI+F94C522d77Xl2dq4Tpz6ngwMW8xkIM072bvLW114n8UMaqyvUl5qIlo611MBZaWC0K2xe2WX32nm2L25x8OgO4+MDOq0q7VWL1prOxhM1ksBlfrSg4miMxwOOjk9QVQNBkLGdCoZhYDkVWo0mvV6XwanL0nKHK1efRZQE/uBrv0lWeHzr29/gcH+P2LX50o//Y65feZ6l1jo7567w/XdukQsC2xdrIEaMu2MKWSMWBC6tLCGkBYqhkhYZy+02UDAZD0kCHzXPmZ92UTKP8+tLdJot7HqD9vYF9NYyPgWP7t6FOAGhQJAhDnwS30VAwLQq1DSJdqvN3b0TWhu77J+M+NxnP/3xh/M//NL/9Vqj0eC73/0uYRhh21bJw5nNkEQR5ezEkkWJNM9B0fAyid48QlAM1paXEZOIRb/PVqeNJgoUecruhV1cN+D0tEueFezvH7C9s02v10NTNYpC4Ny5c2ctYBn9/gBZljHPOjwN3Trz6RZsb28znC3oj+d4Z3KKoSpUDJ0k9M56U0qfQxzH+H5IJGSce+kC2pJNlAbUlkySLC5bxvIMQdTIigLLMmg0KvT6xywtrRHGU6I4YK3zEofdI9y4S6ezSWflIgfdW4xmJ2ysbeL6FXZXv4ITrvOb/8t/4Pbv3+HDb9zg5tu3zxZVAidHI26+fgfmIq3lVfwiJMtiAnfG4eEDppMI33dZX29BUsNprFFv7nDj3i0sc5m5e0Ah5Xi+z9NXrvLWd7/F+uoquSLTWT7H9voGb7/1Bv/kv/qf+OXf+F8J0hPsisXcnZHGIc9c/TKP9h9TrdSJoilFlrDRfhrTXMKqLLHW2aFWbeAtZizmQwpmHPfex/XHTCYj7tx4D6eSo0ggBgq9Yw+z3sCPY7zIx6nYxHFEQUaQRqiWwTRycbOIsTcjkaFQdKbBgvF4QH9wSKtlsNHc4tt/8E0CP8KuOGXSxPPLRaTIWS1jhLeYMZqe0qg3uHPvBg/uPOT89jr3Ht7GDzy2Nrb4B//op/nzb/8Rv/WffoUkE2g2V3j5xU/x9DOfpt05T6HnVGomp3uPMEyLeZKwVqlBmpEHEUlY9rLW6zUqjkUcBSwv1anaCovxgGQxRs8CGpbG9u55ljZ3ae5eRmutkObQ63W5cHH3jFUs8uJnf5SwsoGzdp31J67gBTGf+8KrrK40Pn4dw3/zT79cSJJEGIaoioamq7gLl3arRHioqkIYhBRIPDg8ZTBZoJo2tZrDeqOGkiX43hxZVkiSMl4UJzG+HyAiMpvNMO1y2RAnMf3+gFqtThxFZZOy6+I4FSaTCaqqIkkiIJag6CTBqThMph4n0wmyrqEhoorFGcZQJC9yEITSMSQICEJpXkgsgXRZoL29gl2rsIjmSLKIXVVw6nUUU6VSK/m8zz13nQeP30GQYtI0RFMqXLx4nocPDlFlB9Os0tS32XYu8bN/+2fo752iKioJGXGWQ5ohiOJZ4r2g7O+TkAqxbDdDQBTglS99mqXLSxyOHyJrAnmRsph6LCYRQZjzD3/25zl38TKmYTCZHPLWO7+FUUlZX30RU2nwtd/4FksbCbmm8MzVLzHqP+D+vRuMhgesb9X5xAtfobW8zYP7HzGfDVAMmXqjhmXLCEWEpbZZa18gTmzywkRTDd57/7vcufsGqpYSJT1kScBwNBBFhkcuia+i6wJv/NFN2uYSJ91jVFUjDD1MWSbOc6rVCgvXQ9UM7CQjd0wmsznLnRUG8yGqJqErEnEeIKoiF7d2efny8/zZ61/n5t5jVNXGspySBk+B6wU4lSrrqxt0T3tEkU/oJojkNBp1BuMeSZbw4ksvcrrfZbIYoZkVVN3kwqVtXnrpb7F/MOKFl15mpdVhOOlSCF2+/Ye/xI2/vElHstittrGRmfoeuqigWzpZWuaKZUkqaYlxXBpqNI00jgjDkFRQaK3tUr9wjVGQMh7OkLIULc/Io5gr15+i9+g2036f0eCI3Seu8t7tQ/7N//avP34/58/+zE8WRXGmI8YRrVaT2XwGgCwr+GHIYOpx2Bth2BUajRq7G2vIWcL+g/vYpkWeFwSBT6ezQiEUzOZzHLtKGIV0Vjrcv/+wdAqlKRsb67juAkEsODo8od1eJgwjdnd3OTk5YTabomk6YZQS5XBwcoKm6UhCwfr6KoE7J4nK0xPKnKYolr0mpV6bIwoi89glaSlsPXuBqAgYTMZce/oK2xfP0RvPsCsGCBFe0KfSzJGlhCLOcBp1UrHg5Qs/Tdpr8fVf+1V+/Rd/HWoKRAJ5WpAWKRQZCBLkZ3/bnLPOS4kszc7yasLZrJ6JuXmBKIsUWcbOszt0nqrRudDkze/eZvLY4+d//l+zeu1J2hurdI8f8cu/8xprKzt8/jP/gG9841v8k7/9r9h79C5/9NYvcHHnq8xnU7Y2N/nOG/8eTbPYvfgyqm6ys9rhxu2bxMUjstzDsVZQhV3WV1ZZW+lw3N/n+LSHotbY2HwC21jj9u2PmIwekKczfvTlr/DR7Xc4cW8gqCkf/OcbXN14mb9863Vq9SphHDMcjIj9GNKMyPeRJJVMBFNVyfOMJI6pNptsnT/Prds30FWZSqPKabeLLIuIEtiOhiLLnJ72kVUF3TIRipIAqRk6sixTCGVIenV9izwEu1bn5js3+Md/7x/yB9/4A3ruKWJRkMUxSeQjChnkBZXmCrX2CtsbG7Rba9y8cYP59Ihqpc5g/yE7To0nKisEUYApq0hS6eqRRAFZls46gMrKj+ef/wSD4ZD9vUdYpkkuFkiKSZAW1HafxFre5mBvD1UxeHD/NlVFAM/FFgW2t9Y5POnxz/7Fz3/84fy5//bvFkEQsVgscCyzJJWLInGSMnVDbj86QLUryLLGZqfNK598jg++/yaTYZ80T5EkBceuEMcRoigQhCE5RRmRCiLyjLKvIssJghBREBCEgigOaDWXGI0maKpBf9Cn0WgQRQlxGtOfeSQFWJbF009eYu/uPVRRYDAeIMsqhmGQpulZ4a2IKEmIlMQBQRCZhR5xXWL9mR00SyIjRrUNchGMikScTnEqJksrDr474smLL9NWL/J//5t/x3B0zGA+QEAhlwRsp8X4dIwoCGRJeVKf5cgRKPtIRUkskzB5QZFk5btXlCjSgiLLEXIoipwiT6AQkQqJ9mYTqS4S5CVywzJtvvgzX+Qnv/ov6Z3u8/Vv/ls211c5OTllpfEy68vnWW/s8Btf+wUuXH8KWGCYFo/3H5KnBVs7z7Nw7yGIImk6BXpUnWUOH8/R1AaVao2nrr3MYHrKxO2iSgaqIuIvIqrWBba3nyFOQ+7deYPDvYf81I/+DHf232SQfETQ8/nOb33IsOsjSAXVhoVhlD06g9MuhmYgSSqqqiIUAooo0u/1cL05eZ6iSRKyaSGoMlmeUJCf/d/B+uoyQpIyPD5FlZWSwJhlCLKI0dRRTYU0yAinCTE59dUVdNsi6E9xu2PadgNHVtBVkSiN0QWRhu3gezNsTafTOUe/bjDUHCotjXDQ5ei99/hbT34COcuoOg5plhH5ARQFtm1iWSWkrigKJElGlhXOndtkOOjjeT7rW6u4nsvxTOLYFVk7dw4vLbhycRdHgt5iQRglBP0e0ajHv/i5f/Xxi4zefP3PXvsro3qjXkPRNMIk5cHhKccTD0m32F1f50tf+Cz9vUecPH5MnoMfRVhmWSmQJKW1LgwD4iTH98OyyFaQqVYrWJbJ8ckJ9VoDQRAQRYnFvCyASZMcTVdQdY3BdE6ci8zCEGQZRRGxZchDv7xeZym6pp+VFYkUxV91b5bdJGmSIggCURhTFKDaOrPYI0xDVL2Mxy8tt4hit3QkSSLPXf8i/x9n7xFsW3be9/3W2nnvk889N7/7cu4ckNFgglgECUKUSKpEh1KVqSrT5YkHLld5hOLAVVYpDDx1uWzTpAhKoiQTpkxStECAiM0OeN39+sV+6eZ78s5pLQ/2RY8MuKDhGZ5z9rfXt77v///9bfsl/O4Od569w9nXVzE3bLobAdvXd/BX2rhDm62Lm3z6i5/iaLnHcL0DNnRHbdy2RXe1Q60VluegBUjHxHRsMASGKWgszhUChRAmwjagIxGBRLoWZy9dQAi48MIG7751i0X2N+yc/ST3797hxetfQolnbK9d4cOPvktqVFy+8AppNmcafkCrO2A0eJ5OZ8jO2Q2e7P4ZJ5OY4aDLqH+d2SRlZbBNlk9YZvcpS1hfu0q3O8QwHTx3g07rHE/23uLhg28yn53Q7mzx6us/x5P5Yx7d/4DPXf11QjXj4ssdDo6OGPS3mUzG1DXYtkFZFwgpqZViOY+o6holBNKy6PYGrG5vISyHV1//DBubm5ycHKPrGseyQUnCJCPKClZWNjFTKI9inMzEyAyMuUZOwI5sjESx1t6gU3ukTyf0rS4XN8/hugF5VRIvE3zLOhUglLiuj1awyJYMsFi5fI7WyhoijkmmEzrSZtBqU6uaOIobJ5Zpwsdy1PLj7swwTA4O9vACn063gy4r4qQgNAK01QyxDMPCk4ry6fsE0RIjr7j80sv4G2e4tLnx0w+E/ugP/9evuq7dfBHD4u33HvDkeEFaS0a9Pl/4xGuU4Zx7799CVSWGaTShpZZFHEVIKVlZGdLv95hN56RZ0bw9EU2Gou2QJE2WxHy+QANK1U20g+Fg2yZRmhFmJWFegAGO7/LijRv4lsR1G6lgWVXI03Rq07SbFlIamKaFPD3JlIKyah4MrTWGZ6N9geVZBK0A05OkRQQCWu4qv/mr/xDXbfPk2R5b61e4eumzxFlCq+uxtbXDcG3AcNRnsNLF7zpIp2K0NWS0PaI/6uJ3XbyOjd+2aQ09+itdvJaL7VkIUyNN0SSLqRqkRNgm0jewOy7+wG9UzwaUYcX41j7nVq9z49pz+O2UzXNXeOMzv8FfvfWvGPWvoW2T8fz7JOGMje1tDg4/Ynf/fV6++YtM50uODmesbQxJ8/tYlgRjgesEULnMxlPyWNHuOFh2xUb/GllasMyPCFptxuNnXDzzOtsbN1gsDjncO8D1HFZGm3R6K9x+9B/Yf3qMGwRc/bRNXWs+uj3FNSz6wxHSctg5c4Famly5chUhTOIoJkkSpDC4evkageex9+wRT3ef4fk+w5V1Bt2VJldTNLEdtdDY/Q799RVE2eBChGVimAZK1RjCoDNYwbEcsmWELioW0xnpMqLf6XPl8lWUaZIWNfPlkk7QpswzLNEEEC0OD+ie2cYPXB5+eJvldMZav0eexGgag4FpNXGAUkoM00RI4xQMLfB8H0Fz3TMch8MwJ5dt+uvrmG5A12+RHe+jkWD5uH5AGS3RScylS/8RUGmtBAjBLE65+/ghhhNAXXDj3BnycMaH77zJdDzl3LkthsMV9vb3mabpqWhBURQ5k8mEoihOnQOC8+fPcnx80rBghWA6nWPb3mkCWFNghtEwVgsFx/MQYTooKQm6PYb9DkcHu3imJFwsCIIWdZ1jWTZJkuE47mmIUtOuK6XRCCzLpq5zUDWGNKmLCsf2GQxbpFVI1/EwTRvLVPzqL/8Sd/f+b0ad1wmCmK2VDe589CYXtq7xww/GmG4CwmDlwjaT6QnLZQRKsbo6ZGfnPEcHR9y7dx9TmBztH2EXJVWhCKSNFgVB0MU0HA4PDrDbBr7nNfdk2RSslBJdKrKioNaaMq34yz/4U7SpcM745JnDl3/rCn/rc/8Fe4ePqDmhLGo6w5RvfeffcvHsq1zb+k3Ojj5PUb7PnTtPKGuHKDRJyn02g0/TaV2iinfp+h1mZcJiMcW2+yhT0uuukc1y0rjk2sU3sFWXvYOHnN1+A8e6hZD7xNESx7vAuStf5Mz5Zzy89z3UYo1zNyK+8sW/zz/67/8n0qTkYPcI32vTao2ohU3QGzDa3mG0usrh3h5P9w8Z9jss0sYUkeYpdZ1j1grPcnCCAL/dJk5TyqokcRz8K1uEOkdGdWMRk5KyVrSDRgwjDYM4aiBi3W6XxWxGEkWMxydoBM+98CLSMalMk6rM8ITFGjb73/4+q68+D9LE6fsUtcbWgrqosVwDy5DNXr4oUKp5Th3HaWYHNHZKQwKmS//MJlmYgZS0gg6EM8hSVjc30MLAME1s16Xd7f7Y+vvJKWP/+mtfffODB8xShWEZrAQuL167zP5HD3Eti3anw6UrFzFMg/FkTFlUCClxXZflctnYNLWmqipWV0f0+23G4zFxnGJIA07b2JPxFMexcVybPMswTYuorJklGRgmWhrYgc/29hZZlkFVQF1RY7AMIxzXoygrkuRHKdogRBPHUJ6KJJRSlHV9qrtt2utrL91kuL5BLVIqlVBnU27cPMdwdZOV/mvE0ZI4e8TG5iXiRDCPn3Dx4vNMxkuyPCGMFpgWuA6UVUK341HrjOPpPo4rgRKlSkzDIlxEDb5FSJQoafU9hptdWgMHzArbt3B9G8NqJsqqUmRpQR5m6KgEodBDG+XDnW/c5Vd+69fx5AjpwQ8/+CN2H59wZv0NsPd54eaXuP3BOxyMv4XXkmxubfKDt/6YK5cu8mzvmOHwErt7B0yPDrl3/y1a/TZr6zdw7VWkWTPsrdNx16i0R1EIlFAIqZAWtIMzVEpz78F3Odh/RJakXDj3KZ4ePMNQDtPlLo+Pf8D1Gzt89/95jxde+ARZNKXT7XHzhVeIoxjbcen1R6gKPvW5n6WzskKZlxiGxvUdhGEQtHtM4yWlqkDqRoReFtSqmSW4oy4rm6sUaU4R5ji2R9DqcHJ8RLfdoi4ralWzXIaneaySlt/C81zGJwdMTiZ4dpvtS1cZjjZIXQMWGa2VFZ4+fMBsOSeQBrKo0VrTagUsFotTM0WzY59MJtiWRavVZAcBOLbDNArR7RFBq8/h7j7HB2NWHPDNiun4hKqsyPKCPE442dvn+gsv/PRt7e/+k3/2Vcf1EFXBFz//WZbHB1RZxqDfP52AFiyXU+azJXUtkIZBVdeYlonj2TiOg6orTNMgTVOyLKeuGiG7EIKqKE+DbiWVqqmVYhomTKIE0/GwHJv19Q0wBEVVMZlOGfa6JGFMO2hjSonnuqAba4/rOB8XInCq8qBRDakarUUz3bVM1jc3iVTO8mRGW5s8fusJn/q5TV7+zCeZLKbk1YxOZxVTrtLyAr7/ve9z8dKI/eO/IC/2QIzx/AzLzNE6w7RqHBd2dx9RVwVpEpPnCyZHU6pCUZYpK2sD+qMehmcCgsl4yrA/pN3poQ2T0eoarU4LjJpWzyJLEuJJQr+zQiZKsCxcb8CLn/0kRx/s4q5r7h5/jXZwnr/3C/8da+smh5O3cKxNTBs+/OhPiJKn7B68Q7vTQQiXMJxhmT6j9ja7T36I3zEJ2h0Cr08cJfQ667humyhdniI3TKpSNyzf+ZjN9U0ePP4W1298ho2NFU5OPkDJAs/d4sZzn+bx41tsrK2hPY8b1y9y+HDJ7vgBJwe7dNprLOKMmzdusFjOyaOI9/76u+xcucaLr36Kbn8VKQUHB08pqhLX9bBMg1bQwjBMsjzHOI2TrLVinob0z20QxxFFnCJqTbSYEUUhlmWB1jiOdUozrJoVntEYJXzHIcsi4tPoyo31HTItSNOE48kJ3U5AqWrOjFYwkU1CnWiYQ3VdA42uvNNqg1IIwPNcNJpYe5RGgNfus33+IuvDFjcuX8BxXBbzOe998CGm4dAb9Gj3u5y7cOmnL86vfe3/+Oq1C+cpliG+JamrmvMXzrG3u8fR0RFB0KbX7TGfL0jTlMViSZ7nKKVpBR3SJD11hkgMw6SoaqRhojGoVdXEoA2HLMIILItpGCNdF9N3uXntCpfP7lBUBft7h1SqOYF7gwGB47CYzah144H8kb7W87yPg3HrU6saQqAQVLVCAWVV4bsBRVaQlAmVVhw+OuArv3OJn/nKl9C5Q51aXDv/c7iOw3jxNov4ARcuDgmXCyzT4cbV53j69D5VnZFmKVrVXLx4npOTY6qqxkAShQsW04SqrDh7cRtVm4yPE7IMArfH7uMDJAbSsJhPFx87b8q6xG87rG52uPrcVWZP5hzvH6ANyWjrAl/42V/kww9uc/vBfd79+l+zftbnues/w9Pwf2Ecvsl660tsb30eJaaEyR4WLeIwRFcGO2euIaWPbw159vAOutJcuHSTohT0uqt84vWfA+FwfHxCni6oCJEm6EKjK8HRyRNs3+TVG5/k/rO/ZB7mXL3+HLvPPqTfcShyj/XVNW7f+3OieErBktt//SZnz7yIKBL+wX/+D9ifzJCmw0svfZKTRcRrn3mDdrvLsNfm/qO7RMuQcLakvbLKuYsXiU4DmuIoJs1SAt8DFEo3fOS8rmitDdi5dpbJ7gHhbIFtNSwrXdXUp9cby7JwHJu6bl7cjmVRVTVlntNtd5gcHjI9PqbX6dMJfNI8I0xzFnFJluYMel10XWGbFo5lUlYFlmGRFwXLMCRJkwa8bnnU7U3M9oh2t0sSzXjwg2/B9IjJ4RE75y5z7tw5lG504UkUce3mcz99cT744K2vHu/uMuh1CZOIsiiakNy8SasO2gH9wYCDg0OkYTaR4IF/6q0UCNFQ9rIsQxoGUZxQlBVSGk1CdqtFqTXzNGeWZDi+z+svv8jQdXn28D6LyYQ0DpkvI3rDEXGaYNkOG6ur5ElCWeR4TsO6sZ0mfKcoCpI0xTBMaq3RQhKnGUpper0+nh8wPj4hTTJMz2S00+Hv/pev0N0xmC3GDILn2T2YMo7fJ1dj1lcvsn94zHx5zMqwD0LTa3dYRIdUVcNPAou8KDk8PCSOMo6OTugPNrFsF1W4PL0/J59LxvsxhuGCMglnMUHQQho2Spn0h+v4QRvTtuj1+nQG57k0epl7b90lsp0gSgAAIABJREFULiqk1+bLf/vXmU6OScMFqjT59Buf589/7y3+/R//SzqDNc4N/w4v3/wFjqdP+cGtr/ObX/pHnN16lWeH7wCSWta0fId4FjM+mrGyNuLp/i5ZlmI5Do7fZ7EI6bZ9/E5MUhwReD28lkteVlw8+xxZHrNIYjx3leOjXc6vPI8w+jx7+hbdYUB/cJmqkqSRiy5jXvzCS/zNt95ndbTJs71DWq0ey2VEVlW89NqncNyAt999k/sPbjMePyFKZqxtrTNYWSPNSoQUTZ7J6gi/HZClOVmaNU4p08I2DGpdcjzZp5IF6WJBnaqGYFErwjCk2+2Q5ylnzpxhuVw299D5DK013U6DRql1g0WN50tM3yZVOYvlkkpKLl16EcsbogwLxwsaVI9sGMwAlm01QGrb5oMnx/gb57E9j7XBCru3bvHmt/+ab33n+2xsn2fvySOqIuSTr93EFiXh4oQrz7320yuEfvPLv6B/5P7I85wwjuj3unTbLRZhhHsalV7XNa7n4jou4/EJWdbYuKIootvtNsf5Ysl4MiHwAxzXoVSaRVYwjxJarTbxcsqvffmXuP3OOzTCHsFkOqPTaVMISWU4PN0/RkvB2e1G6BDPThBKYxgGRVU2O0alm+QwDXmtGhCT5mPcRFnmoDVaC9o7Ab/yX79C91xKWRuEUYHr9fC9TXx3wMHh2xT1GNNSjEYDwuUCVRXUWhGnFbbVpigVGotlmJAkEYPeBtGs4v77M4525+hCki5KHMujVJrLN65wcnxMlsZ0um2SKGZlfYONnR1MyyEKE6Rhc+XCNf7d732Nh3fvYPoepmXiuQ6LRYiTVQzOXSCeh9SGpm06zPMlajbln/zJ73Pr9l+RseC5a5+k29nGMU2++cEfYbsGn7vya/z+v/hnZEnN5tkNxuME24XNM+fo9HdIs5qt1fN4NlT6GbbX4ugwxTRdiiJhbfUCWjnYrs14dpfbD/6cl298mdlkj6KOMYw+62s7fPToHZ7cu0WrozGL51mTG3zzG3/FL3/5K/zw1i02N86i3C6t/hqeAx/c+j7h4gRdxzx8cAfL8Wh3BjiOzXw+bdAfrkO3122uQXVOHieMT47QdUaVJ2RhxJl+jyfv7mOULnmRAzWWaTEY9HEdBykkvV4XJZorTydokabNrKKh6UlCkVN5sIhDhBacG2wzam9gegGb25tUaYTOl6wPulRZQjgf0+u2Wd/aIm1t4Y7OUte6Eczs3UbODliEM4ajDc6e2eLJ06cIrbEsC9d1+blf++2ffs/5f/7x1746mUyaqO4gAKkRCNI8Y3W0SlmV5KfG54ODAyzLOk1v1g2iwbJpQB/648AW1/dRwuRovgTLxDYMPvHKi3Qcg4f37mBaFpubW8wWi+bNaDf8n/X1dfK8IslTiqLg8qWLzMbjJmVYyqZtVYpagTBMytPPVy5f5bVXX+Pu3btUZYXERAvBYLPF3/md13jxjeeJ84i0CFkZncE02+TFCWFyi1Y7w7YqpMxp5kwFmgrDBMOEVjcgjEPSNGew0iNeWOzdLXlyK+Kj2wdYyofCwDZd8qKBYG9ubjE5HuPYLi2/RZZm2I6LlBa2aREuxuw9fsC3/68/52B3H2E0XYZSJdIwKSvN6tYZFIL8ZEJ32GNydEg4HeOWFpdfvsI//8N/Qx4F5HHG17/+J0yXY/qrbZaLHHTJ+OiI6eQEyzI4OpqxtjECY45ldhiubLMyWqUsp826oQrZWL1Mv7eF5ZTk1ZS8TNkY7ZDlMVFc8Onnf5H3b99GGDVlGfHR43fYXD+LZkl/tMLO2Ytc23qVqrC4ef0Kf/i1/41PfOpzlBoOj3fJ0gWj1Q36o1W0FMymU1zHpt/vEEUL4jiiN+jT7XVxXZeqLFnMF4ThHK1LqCuKMKNvtbl69jqP7u2isWh3u2xubFEWFYYQzdpF15RV2eT0CPExQUMKgWs7SNMgz3Jm0YJ5FGGYJnVZMrBcRJ4TzqcErQ4b5y9hdUY4g1WkG4DtkSqLcSboD0c4rkuZpaT7j6jziOViDrqJmhz0+yRJwmA4IE4SLj/3iZ++rf0Xf/C/f9XzPBzXpaprbMdlESfMF8tm/G+YTZak4+L5AWnaLGeFFORFgWXZ2E5TsAjJdJlwtAhZJAmWIbl5+TyUOeF0chrKm9Ht9UizDCkkSZpi2jYrgz5RlGKYNhVQqRrfdSiyHF3VZHmObTsIJJWCJG+ofpbVhO6+/fbbqEohkUhD0l/3+aXffp7Lb7gs8kdUOkEYJkHLJc+maKYEnkbpDGkoXNcBalrtACE0htRESYgSJvNpRbK0mBwoHrydsX8n4sm9Z4wGQ9I4xXMDFvMQN/A4e+4s8XJBGEWsbaxTFSV5lqLrmvl0zP6zx4TTKcePD1lOEtASaTQTbSkkVVGiqpr10Yjp+JjNXkOmiKI5VRYzHGyyfnWDuM5YzOZYpoklW7zyyqewZAchx7z79g8o0hADRRzmjGdLems22pwwDx9giIBBf5s4O6TlaDZWLqC0S1RMseyANEuJkxPy4ghpWfTa5+k6m4xPQpbpD1G1RKuaZXiI5VhESYmqxvzuf/u79Pp9+htnUXh8/jOf4c69+2ysjrh/933SIuTxk3tkeYjfchFSE6cxy+WCXrdD99TQXNc1UhpNPovn4kgDEZXIhcapLGaHIa1Wj8tXr3D+0mVa3T4n4zEISVFVxGna0AuyAq3UKWZHn2qxa6QhcWybVr/DMktQWtHtdxj5fcokRdclRRyy++ABD25/gGFabJ6/ij/YpLO2g98b0B8OCdod9u99QHLwBNuCMs/ptNunqB3VBGgZksVyyY2XP/vT7zndIODg8JC1VpvxZMpgMMCyHIYrLmlZ0XZ8hLQoKk2aNHHbZZFhyIZhW1cVRSFJsoxCCzItCNptfEdioYnnc1RZErTblGVBp92mzHJMy8E2LVzHwTJt0qxmejyllBLHcUnSmCTL8YOA4+kMy7HJy4IfpVk7toPruMRxwsnxEQiBaVoIFKOdDn/7v7rG9icEUTVDGlYjr6szlosnQIVlCZI0x7YkUgqqsgBZU4U5SmsM0wRhMjnM+PDtJVXhEJ9UPHz3GF1qHNtmMY1BScbRFGlCK/CZjI9ZLpaYls3RwTFSKFStmM/nVGUJWhBNQoqoIdMrXWMIC60UNboZcCnNnbsf4jouk1KTLZZUSYg0XK6/cJk7B0+RShMu58TJiCvXXufb3/0+aRrz27/zizx58IR0WaBKi26vw/blK3RWVsiKMXG8pFYeWZbj2R3abpc6q3l6+AGH83t4vkWaANohTBbY9oSNlYt8dHiPsztnOfkwJa0TVlcvMp+NMWzoBT32PvwQQykMI+d//B/+G65fe5nvvrtGpROWiyOG/S6VodjeGXG4/5Q0npPEc2zXZrDSwnN9XNcmSSJUVVMmGdFyiaw16eGUep4w2RuTpBlCSi5evMjx3lPu336PvCiZThbsnD1LnORUpaIoYmzztCOpGw6zVhqzNpGGxDAEuA5pllPRMHs1zTqmrpv1jGlY9Fod5nt7vH10gBKCrXMXWT13idT1SOKYweoGR9Mj3EDQH61hAUkSsQyXbGxscPHCJfK0+LH19xNPzj/9yz/76nixZDxfUChIswJBc6+zHZfFqXUsz3Nsx8FxXXrdDlmaNhzZsqJEEpc12rAJfIet0YDAsk51tM1e6EeOk6IoUKq54xqmJMty4rSh9gWBT5iE+EFjI1ouF7T9xkKmEQgp8XyfOIwa/EQcNwR4mtWKQKFEzWd+5Swv/XKHaT4nTCDJFWVVEXg9ai2IkhDbFmgytK5AKAxT47gmSld4rsv+/owk9PjwrYj33zyg569x/50n1JFCl82oPosLBI2IezDqE8cReV5iiAb5Ei+W6LIZXMRhhKwl6TIjWTQJZI0QocC0rGZ1oMC23WY1VFeYVmOhW93YQGpBOJ5jmPDcZ1/mZDxne/s8TtDH8VvcuPIy9+7eo8wNOm2Th3fuYVkSNwg4c/7SqZStpNVeY239Cr5vURYRnmcSZYcs0jlSWswWuywWh1hGn363h2lajaLLzniy/4CiULQ7BoYhWRmuce/OfQbdDseP9zi4FXF0cEiSpQSWgWGUfO/7f0aZp7z08svMZlMcy6SIltRFQp4vicMxeRxS5ilVnpPHEePdfR69e5vZ0yPmjw/R84Q6LsirkgqNFoLxdMI0XLC2sYljOcxmY65cv4LjmmRJRlk0O3HTNImz7OPgLSmayA7Htsik4vHkAC0VgWHQM9t4lotlmU0rbEiCU4aQFDDsdcmjmPnhHvPDpxhVSX9lDbs7xGz3mSUlthdgyKYDOtjfQ6maOI64/tJnfvqT897TXXIN02VIGqfYQvD6Sy+wnDfgYWFIalU3f5C0MU2DdrvN/sEBWa2IioosTrFMg8tb67imQRyFlEXB+voG4XKJaZlEYYSUIKWJVo1C5ujwmCBog6hwThVHnaCFQNPxfQzTIC8LDNfFQKCqmrXVNcJFSBLHpxHtCq3rZp1SQ3vDpxqEPDhRCCvANG1UlVJWYJgDZHmEaaomokCK08ClgKouG7tZVnB0siBcwuxAc/vNI4zc46Nbu6TjHLTAclyyOENoqI2S4dqQOE4pihIhGsT/bDLFsWzKOCMLQ6hBBAGWdDDNglrVGKaBxKGsKoQhMYRFUWg816Eoc2zbZpFETGdTVoMeWlVE4YJHD++wvnmZzbPXKMucv3nrB1w8d40v/Owv8Rdf/+f8+m++wbeTAlO6tAYWYRRh1yUQYEifPM3Zi5fYdoFthziuQhgFtrJoe1uUxQGWLbAdh+ViSW+4wZOjHyIsn52dFzk4/itmiw+IvQ3QCl2ndLsBbr/Lye4hKxvrPH7nHne/80OCfocwf8I3Dv+AqtAYwiBazsjzlDiLiOO42SmaglTVSMuASiGVJHA80jphGsfUlUbXNYYUCMvENG3slsuT6SHZPEZIGM+njIYD5nfuIWncLFpI8ixDqCZt/GQyxbVdFmkMqwFaaFq2Q1s6GFpTqZK6rglaAaquyYsG/D0YrhLFCVVZYtsutmky+egB1WLOJE5wuwPWts4SVTn90Qqd/ibRo/ukFbhe8GPr7ycW53IZYjkuSZJx8+oVFpMTDENi2Q6qbh76PM9xLBOtm4jy4+mcWCniokIJSbvl03EsinBJWjdEtaqqWV1dJc1SqrJEGpIszXAchzSP6bS7DYq/rJo7npDEcUwQBIRJAlJgmw5CStZ21pkcHxHNlzz66CGGlGjVnEiiSfxF1zWamjMvDjn3ep+oWOIbLQwEtdJoXTBd3EbKGssS6FphWSatlo8wFGFcNnY1zyMKYX5ice+dKbODnMB2iI6WiEqCkE1moyMxHJNWt02eFaRhimmZlFXz51rSRCrI0hSVV3S6A9xeh/liibQsRC3xbA+rZRDFMUo3e2PL8QnDCe12i8Fwg1prFod7PPnwDgjJfDrnzNlLWK3L7Oxc5979N7l+/Sb/9k/+Fb/8K3+XPFry4a07VEUNGFy68jxhGVLnFhfPvczR4hEnx/eRlmJz6wyKxr7n+SWW4bLcm+G5HeJ43MDU6gADCwObZXSMb3fw/BHTxUOiMOL6hZ+hrA6wOg9whwovbKN14xzJjuccPTwke7qkVjWuZTPo9lmGc7QALIO6Kum0PCxpocqESmj8ls3amRb7D46xfJee6xBGMVpobMtBoRGOhTI0qlDY/QBR15ykEcdP59ByaAcdhIbp0QTfDxhtbZFFCxSCQtTksiJdzuhZPt3KwckN4nqBGwQfkx0916PMCwzDYnx0ghYax7EJAg/fdZnP5kxnDyhUDbtPiA+e0VvdwjLPELS7rFx8kXbLIQ9n/3HFeTKe0e336Hc7XL98kTt5xmw6A92EjmqtQDX2p0ppDsdTslqTlWVjeFYV/ZaHa5lUZXXaBigsy+L+/XsUZdmkkJ2KFLQC1/HJ8+JUftfsU/OqcZLUqkZpEFKwMhxycHiMYxosprPGslU1A4OGbd181rp5q7bPWgxvGqRE2E4Hx22jVYnlWOgio6wrpFJ4roFhm3Q7bQRwdHRIWmhAkqeS8bFi/yPNw/fGWHgk4xgqjdSgVWOshiYbNE1SNPpjRYk0m72YqiqKosb2Pfxul3bQRmtN4DmgFGWi6Xd6lHVBbhUUVUXgB2xtbXPn7oLV4RrSadNqeRzf/hCBwOoGlCLnpec/z7ffvc362nl+7w/+Mf/pb/0OqhJ876//Am0kPPzobvNbGyYYBmUc4zg+qo44Gd/GDzr4nodr2ajaY5k8aiaoszFJukTQAwzms2PWRtd5cvyYsjKwLYej8X12Nl9jMjnBUBt88uVf5d0P/zW5kLz+xSHf/P1j6rzEcm02rp0jK3LCgylCCWppEqUpyziipkY6TRaJihSt3MbVkAmNZzp0Maj9FvNUoz0LU5fUhgTDInDc5lqlFUhBb60BAwihMRyPVJdUNqR5gdVvIW2bmc44np+wuTLC3wioLYmRFLQrB0/b7D07wZQa3/dotdp4rodjNgaOqizI85QszxkMB6RJ0rTIp0GgvU4LKQVFlpAvpxw/KjAsm+7KCqLqYfs/Xlv7E4vTdm0c26Lrd3jvhz/EPTWexklDwTONJmx2mRUskpQsy/Adm5VeG1NrylxhIDAt6zTe2wAaskJalAgh6XZ7LOYLqpqmWKVkuVjgOD9KJqsxbYtKKaIsQ5oO22e2UWXF+GCfKg6hrtGiSSf7UXFWlQItKOuKlR2fV39jk81PuAjLxrZbFFWBKaEoFVlRIzQ4rkCjsF2PWmnm0yV5rpHSQgqX6Vhysmfw4ZvPqEITUVeorEIoqKqmzbeEiWFYzWK7Vri2TZ014cG9TpdFvCRdRMiixrYdOsMRve6AIoupigpEhpAGUZRQqBwQSGE0ezgpMYTBYDgiyyuOHzwgWcywuy7muoeUDt/5wdd5crIgTb5EfzDgrb95k0+//ga3fvAXsNPnzvu3Wd9cZ7C+ye7eY6bzA154/gxRfR/FIbNJxc72Z3Fdn939B9jBtJnKJx6tYJs8q8nqJRKD2fwpvf6lxuObawpqFvMl2+uf5+RkjOHW5Pox7WGLwLWoikcI7ZDlKVmS4fdb6LwmmcQopSmLgnarRavjMwtDkrKkojGlty0Ht65xQ1jGBbblo5I5UZygTUmdZ9SOxkCQhzHDQY+kKuj5AbPjIyo0/bZPWJUkIkfbBmWeYHoeqSqx+l0O85j1yiWfLlmxAlquC6VClSVJVZEkBWGY4rsey3mI6zk4tkmaJghpNIFGhtnMUEyTbrcHUqNPd5q6rjBURZ0rTnafIo0D/MCHz3/upy9OKRu/4XQ2ZSUIGntTqUHrJl+irjleRizjBClgY22EIxS2FBRF438zpETVjZY2jhMaj0jzg2shSLOcLCtoDmHdUAR+JLlybCqVk5UZSjbtpqSkSDOODw/ptRtchGlaVFXVnJRKUOmKPC+p6hq3b3DjF0ac+1SPXBpo5VKWTYxcWZZkWQHSRuLQDgYofUxV141euDJQysGyPNLMZD5W7N6bkU4E5FBXFaIW1HUFUja+0bpCqBJDmciyJl8u6Tg+faeNqAWH0yVVmrM6XOOlV99g9fwVlBbEiyPe++H3yU9O2F7foioVyTxFSFhd2SDLC8aTE9I4IYzmXD57kXt/8x2ckY/XbXHhxZscPH3C7Q+/x+aVV3j06B2u33iN733zO1w+dxOykLX1ISeHA7IyY7TVZ3y8j64Uwo7Yn76HFBLf6+J566TZjKyc0/e6lGVDIMirJaq2KfMawyygzlkudxkMLpNnCaAYz+/xynO/RprG5PWSZXoCqc2ws8n8YIzZ6eA5PmVSIFD4wzZtp8P06AiEptVqMeh22drY5Mmzp0gpaJsGnmniuh3yLCPLSiZxxLLM0Z4FQmBgoEpFloR4QuLUYEoLr9TcWN9GmAar587wYVEzjRIM28FdaQZqlaHZPz7AqE3G4yU6LVG6pAhqNoYrBJ02k5MxnZaPbVqUacH+bEGlKnzPpd9rM1pbwTQ0aRJRlTbXr98gimKEbHA0qKZzzNOIWku0kFiORtTOj6+/n1SclmUipcT3W2hhNFyfOCYrKg7GE54dH5PkOWc21zmz2mO1bdP2LLI0IcsyDFM2bpCipCxKslPdreM2CcECyWK+IMsaQXKn3UIITa/Xa+IUDJNCK8q6ot3pYJkWqio52j9oMJi1Qimae8BpOpnlWFy/cR2/5WO5Ft2zFtf/1gYEbWBIkijCKCPNIC81luMz6L3I2uhz+ME6Ve0wHmekmY3rrmLZfVQdkCcdxruCo8cJEo+qrKmUxnZ8hCGRCEwlGHptPnH9ZZ7bvERfO2z6PVrKwhM24fEcGVV0pcv2uat88ue/zNXnXsdr9Wh1ethuowS6euMGg9GoMZ8LydkLFynTjEW4QBgaR2gOnz7l2idewRl20a7Br37lH7LW3iCLcubLBX/6J3/E1toW2xtbpGWK7FkIx+DM+bP0VlqYpqYuFN1eh6JYYNY92t5ZLl2+ie1KkmxCWWZkqUYajZikKiNUneC5AY5p0esOKcoJVRXTDlaxnQDbMVgsD1gdnOPp/gNq1aLMVwlaFykWWYMNKSssz8HxPeyWw7mLO7QDj5bnsjYYsJxMEHlBSxi0hKTj+adDwZqiqhjHEWFdNlkmZUUZZ+iioi0t1oMOO/0RXcNhq9PF15qN/gBR1hzvHaKSihtrG7y+fZZXz55HxgmGKWh1W3TbbVRdI1yThayIHHgSztjPEkTHZ7g5QjgmaZ3Rbrlsrq/SH/Qpq5rxeMIyDEnjhDgMWcxnRMs5aRSh6xrTFCRJxHgybdY3RY6oa0z5/ykO+v8/OVeGA/KyMT8vwogyS5t7H4KiKhkNe/imZBiYqNohzxrjtOf7iKwhHlSqRp0OZ35kq2loBQJVK6qyEajXdXVazJpSK0olUFmB3wrIs4w8acbplmWglfp4XaO1oiwbol2tFFIbzGYLbMfg0ouXufkrI2TbZhn5lDlUtY2kjSlskBG2NaDffQHEgmX4iDQTJKlJpzWgrA2U9ogjwXxiUhcmvX6byWzB6uqIaLng+Wuv8tLLrzM/3OPd73yDush47fkrvPvu+2ysrWFLi8V8zjxeYABeDf1eB8fzmU3n+EVFVSUcHu3R7fdZzTfRpkmFxvbcJitzMODi+Qs82HvA1tkthr0eYS25cv0atW+RJRn9tW0u33iF/CnYxgp3Pvj3uLbH8zdfYmVzne1LNd/8D/8Gr+XSXl8jikLitGTljE+SzTDoEPT69IZDsnxOWSTUZc1ssaDT62E5LnWdkiYx7U4frX3CeIFGcnR8l0sXPksYrWAaMJ7d4fXn/z4P97+DVh6t9gatYIfVM1vkFpjSxBKCIssRpsV0PsZ3LdqOg6wKjLpmdnzcuJcMi1oJ4jxreMWORy2jZj9uWfiuhy0ktmXhuTYoRbfTvNyP51OWkxBxeMC8TNGOCYXGnltEy5jtlR59KjqmzVPDYaJLbN9BCc3O+TOsr6zy5g/exPdceuvrTIocN7DZWT1Px/aoNTzd2+XZ7h69bo+gO6Tf7nByfMw3vvUtNtbWCTwXz3PprwyJ4gVeECBNA8toDr7qlKjw0xdnv8/9x0/QdcO+SfO8mbZ12nzmxVfJozmLyZgsibAsGymMj7H0ruucYkIMhGmCFhRl0cj9KoVGYAgDEJSVAioWy5hSQVkXLOKIzY0N0iTCkhbHJ8d0Ox3KUyqf4zTtgKqbySxCUCsoasW9e/e5fHmDzkbG1S8MmUaSMAwxTZ/Au4BtuJRqiWEOGHQvk2Zzan3EYpmS5yaD/gWUyqhKKEufJK+oak24XCAKRaAtLm7s0B/BztlrXL/6Gv1P/zwv3Hydf/cH/zN5khJGC5QhuHj1ElJL3nvnXTp+i/fGMwqtcDo+0hREyZzHjz4ENNeee5m8yPnoo3tEcQooiiTFtV1awxXEwUOkZWAPh8TH+6yduUB7fY1kEfPum99i88oL9C/fIBERt/7qGxjYbG6dwe24WIFPWafUdclopUscxzgth6RMmS9TRgMPaXoskwVxfECUHKOUpOtdRlSSUd9nP71LVZbMJsdcufIaT57lmNKgrjOePL7FxsaLhKkgife5+9G3OX/2eeLskI6/jqwdti6f596jR2xtniGNU6rqmDzPmKsSxxa0fYc8iei2AmqlSPICjWSZZFRVyag7IC9Krly4iKoq8jTFoGn/tIAoigl6Xe7vP2Nle5txljBZLNBa4Qy7SNNEqYL9NERpRWxqLFWyktdc2VxnXITs9vtUhsYNHCqhOX/pApPpCZVrMl5MGNku5wYbDDurdNfPcOb5nOO9Xe7eu81RHfH09jGB4XF4fMQ8raiqkq7nceP557GoGa6s4vsNBSSKE6bzHz+t/YltbR5FzKczZvMFy3hJr9vmS7/w8/RsA5XMiRczpGhE6kXe+DvRCiFACnnalolm7C4EZVkjhAnCoKoUSjXj76pWGKZNkhdUWjdvasOkSFJQmrqq8X2f5ibayPK0BnUaWlTXNZZpYpsGrg2bI59e18EZ1hzGzxjP5kgpcKwVLKtDLTJst8/q6BW0UZKVy8bCltdUtUFVm2RZjhA2lmtimBrb0ljSIF8W9L0Wr998hXwZYSZjoif3+OjuQ577/M/w6V/8Eif7U6bPpmTTlKsXr2FiYgsb03AwDJuiVsRlQVwmZFlDUbhy9TmWy4jZbEKcLCmLDHTNjRsvUGmIirRJyKolwWiNZ4/vMl9McVyPuky5/+HbfHDr+7z0/A1kUvKZL34F2/LQGNy/f5fl4oTBcKXJI7F9pLDYOXee565/EmqN77Q5d+4ySlWkWUyNQNUdPLOLYM4iuk0Uj5FCM5sckmb7nNl8njObL6B0ynTygK31EZbZwbI7PNu/xVrvLFfOf5KySsjjlIsvvcrWziU2NrbwXI+GhE01AAAgAElEQVSXPvUGN1/7BEbXp+65THVJd30d2/eZxQmHsxlH0xnjxZLz126wfvYileEwXSSMJwuiZdTc+0/RM4Hv0fU9VjttfAGr7Ta+5WB7LihQZUldlqgkRVuCyISniwX7uuBZnrK7d8xlr805y0dEKUmeUBuaze0zRLMQVAMtT/OSjY0VqiLm6pVr/MZ/8p/x5V//ewxGm8SqZnc6wQoCkJLpbMnRdMatd97j/PlzvP3WW9y7c4fp+BhDwGL244vzJyqE/uk//cdfXSYp7aDNlXPbtE2Nb0u0LqnriiiMaLVaSNlAjrKs0bgqJagqjWla2LaD7wUURUlZK5xTCFcjODZIshRhGNRo4jTh0qVLLJdLhFLUeQnyFDWiNUop8lP0ZZEXTbu9jOj1+3iugyE066tDfMeiv97muS9dYKpmpFmbXuccrjuiyCrCWcmgd4Wq+n85e69oy+67zvOz48k53HNzrFt1KwepJJVkyZaEJVs2Tm1sMAZDe+gmNNCwwA+zukc9PTMdvJpuYJgFLJqmTRtwO2CBs61oSaWSVDnXDXXjyfnsc3beex524XkBzzIP9/ncl//e+//7fb+fz4iutookSfR7TTzXI5NcoD+oYtoB8X0iv0jtSp9zX7jEzptbpFwZxXKIujbaQKNZq3Dy1CFanT2SEZlIGJx7YfZyuUosEqZerbK9uYlp6Hi+R3KixMEHH7w3le0Si6cIhVWazT1M08CxrXsDNYnFlaMcWD7M9t4mvW6dn/z4z1KutIiFXRQljKx6vP761xBFnWa1yu6dG1QuX+cnf+HX2dzZpttpY5kdLC9Ai4oSJFJRBEFBiSiASKuxRSxaZHJumd6ohmF08DwHy3Tp9vewvA4IEVxfYX7+EKNRO9DxTZ5iYmwfMVWk19mjq7VByCELMqcOPcEbbz3P/MxBtqtX2LvdYXH+KMdOP4rp+jiOjm1bHH3wYZKZItFEgmavi6UEgLduT0MKhfDwsGyL4XBEtVqjUW/QbDbuTbM9TNulr40IRxMIooxjmRSzGXzTIhULk1JU5vIF5nNFIq5PPhLHGpnkJsbQTZtQJMTQNhkILo1mEyWaoLKxTUpSiYgSrUGfYW+EqZuEQyGikSiDrkYiEmVqapZ0Oku1vIulGyhIJEJhSmNFOqMeA99Ex2b20BLXL1yjPeiQjcdIp5M0ao1AjymKPPbku3/0hFBrMGJ+ehpVFMjGwriSQ72yhxIKdjypZIJQKIRpmviiQCaTCe5/oo/nO3hImJaFbgTpGTUUwnEdPN+7dzjlAM4ly+zbt8TO1jatap1hr4ciybiCgGu7IPj3uEAE01/Px/MFdveqhMIqw+GIXt9hpA2oNVp4ApxeTJOcSbO21gHShMcWEEURw+8xNj6LL8JgtI4akjCtEZ1OD20wQhbymLbO/v2HmZEOc/WbW7z4x28iGRIJM0SrVmM8l6Kyt0ul3iSZiOPicuvmZUQc4pEQnXqFwaDP4uIiL3/vBaYnJijmc1iOQ2fggCzguBbeqMtg0CGZzLGzs0E4FCKZSqANGiih4MFm2S7btRrhaJR8PsOLr34Nz0/iaUPmp2d57jtfpDfYQVZK9DUddurkxBDXz3+H/OKDXL3wJvOLaSp7dUbDPtOLJUzTxvMdHHdIt2chSzEsx8ZybeKxLPqozkCr4voCgizheHEgjef1qNX2sCwH1/Vpd/bod10cx+DIsVNcvHSF++57N9XGGt22QaVcRwnp2NYW1W2F5lYD25Y588hj2J6L6Dn0hwb7jx6nVighJTNEVJXLL72CtrVLyA3heS6haBzDshlqnSD1RZBvHhkGA03Dcz36usEHP/xhzFGf8sYqiWgU3bERpRCra1tIgsTP/+wnefHll6hbdcJDJ1AARkOEBHDaGq7l4KgKtqJS6/SJD02SIYUBLslMmlQ2jTEcYfR07myVKU0tMuY5TE5OsP7yy1TqFfLFHLlkClUWaVhDNMVlbXOdMx9/D9dePcf+wwfpaD1Cksr3z57jxIlj/7g353PPfeXZeFghJPhIfrBIz2ZzTIyX6Pd6+H5QDfO8AMwlisGu8e/WLL5/b5rqeEGFxgrWGwgBmcC/RzFTZRlZEHBMi163G9TM/KCnKStysLsEJEmh39PwgUg0hqyGSCUTVOtNioUxdG2AJEjksxkmpgvse/Q4idRJqmWDUChLf9ij1R7iOQLD0R1Mu08uO4Vld6lXqzg2OK6FLCqwnuDCX17iwgtX6TSGhKMJFFEgmwiTjIYCNq4PYxMTJNIZdja3adZrSMDO5g6RmBrE6VbXUCWR/cv7qNaqGL5DrJQnWcpj6EOMkUa706Df62IaOp1uBUkRiMYSxJJF0ukUjVaVSFTFl1zKW+t8+OM/x9k3XiSWznDlypv4okYqlkbrDjggZoiYFlt3b6AlQhxemWft7mXK1T1CMQtXgHSqiGFoRGJxHNtEEiRK04tIoSgJNUVY8egNNpFElWg0gSQFD2ARiaE2Ynx8Gtt2aHdaRCMlQmGRTrdCt91hanqFod6jkE3St9cI5yusX9nk9tkteu0u7XaZ1Ts32b+8wtTMIkv7lrm7s8PM3AJjYxOEwhHe89GPkogk2LmzRiQUIZsr4OHT6XZwfR9ZkohEwuRzOURRwrAMlg8scfj4SaZmJ2k1m1y+fpOIqpJKxonGYvQ0jXqzjS9JbOxViMZTwQEfaEwuzVHdrZCSI6SjEazhEFsUcCIqpmkh2z56p0N31McLSTxw/2lOHTnF7Y01NHPI0WMnWVpY5MW3XqVhDbi0ehPHMBFkiY5j4Mk+UlKlVanSlUwOHD/MN5/7Bvvm5tkrl/nEJz/1o1fGvvKlzz8bVkV6nRaCKKJICs1mEzUUwrTMgNEJxGIJdMMI9O+2HXQr7w1qJEnGNC1M07pnoA4OGpLI/adPM9IGeLbDoN/Hdpwf/LZt20iSFHRGLQvb9mi1Ogx1neFoRKPVIpGMsVuuYugGmjkiMVFg8egBGu0Ockxg7v4VwvFlbEensrfNaKRxYOkUK8uHcXwNRUkx0gRqlV2MkY1puMRiYzx+4H18/T/9DaO9Nj/zqV/m6NETXDz/Nro2IBmPs39hmsFAI5Ufw0SkUt5jcrzIyvIS05OTPP/SK+iWQ7/Tx3McJsbyNGo1DN3AFSE3P4+jSGxuXg96hpKEKIKPgSi5LO47QD4zxoOPPEWzuYltjpibm2Mw7FGub3Li5KPcWr1Aq1/Ht0Z4io3iyfTKLY6Givi+hzBw6UQcopkEXW2PRr3KzMI0SlxEViRs1yeTGSedTpPLjpHOzWA6LoLnEYnqdLrbiEIE1xWCnLJrIYoSjmvS0+p4ro/tGkxPHGNxcoW2to5lWMyWDtMddFjbfpHZfSpnX30Vp5VGb0Vw3aCHmUzE8V2b3co2kgzdXoex0hiTk9P4gKYFKr7VKxfQRyauJ5CKRMiEFWKKhO8GBepkIoauDUgl4+BbdGplRv0OuB7NTo+t3WoQdUzGCCkqvW4XWZKZX1rCVl00ycF0LXTRRYpH8GQBRzcIIxJVVXQFiKjYIoREmYwaRRRlOiONcxfPs17bYXVvi+89/x2yyQRzi/MkJ/Oo2SSVZp2oIyAYNsl8hvzUGGNzU9Q7DZ74yI/z2mtvIpoOxWySj/zEJ370z1rTHOG7MhMTk4wGGoORRjQeo9vtoqoqzXoLURCD2ovr4Tkunu8FiBDXQxAF7HtaPggGNx4+yViKmelpqjvbeJYFuMEbS5bxxCApIocURFnE6ltYtke/30c3TGRVIaJGaPc1Or02J04eRDddKuYAuZDkyvY6MR9c3WH3zi7dsMZ4YYHFxQwbW+tsrO1SrVRRowNmJ/djOQNcZ4ywrFKvNxj1TD77Lz/L5rl1Du+b4Ut/9Tmi4RTecMj81ATNRg01FGXf4YO0Kh3+7e/8Ab/xi/+UtbU1er0e+WyO+bkpUukU73rnE3zpf34FUfRYv7OObdrEx/LEEzGiyQQDLYekKgy0AYLgY9l9UqkkmXSa27duUWpX0PU2c4vzzO3bT7NbJZFOo0ZkSlOz7G5fJxpXCAsp5iMFTkzniBkSouOBIuDcqpB8VOB6t4Ysirz52kV+6lc+yPWrl4nFSgyNIc1mg3xqnrn5MXojDW20ytrmdbo9g9JYFt9xiUTiWKaGIPrMTC2zs3ce3dQIqXlE2UIIRej2hqjRCBvl2+iGjiiriNYE48n9vP7qJnrfJByOk87mqVQrAWtKkHAsncWlfbz9xotIwObdu/iug+g4pKfy6DfLTE/OUsgkqG2tYikStu0w1AcocoZ0Okk6HSeTTZGMRgEfJRLGd2z2zc/iWh6SEgZhyNhYEVGUuHrtKpmJIqJrUZqfxhI89JGOhU+702c+N4HqO+iWycAcES8UqaxvoXYFovE46VwWJxll9/YWpelJ3FSUl1Yv0mt20OwR0VyKxGSefqtLKTdGZKpAOBtnc2uDMw8/gi2IlI4sIvkiL597+x88fz90WhtRw6TTabrdLrbn0u32QBDJZLPYtoXjOJimiWmY6LqO47qIYhBfC0BeATXP9T0cz8P1fYpjY0iCwPb6XeqVKsOhxsLCAtls9ges2VAkEigHRyMM26Y/0II1Dj79vsbRwwd48NRRPvTedxOPhOmZA7ThgHQ8jdXqYw9MJMPn6vnbWK6NDxi6ycr+EwGo2ujgCRJqJEUhv8hwYFGvdkglxpB8j+2NDVRFJlssMDM7ybWrl3A9j/XtHW5sbDOQMzz2U7/CP/9f/w/+7//yO2ysrXL86CGuXr1MpVZjrFAiEo7z/PMvMjk+wcmVY+jDEaFQGN+0Ke9ssle++4PJdiQSYqA18YUhkZjItWvnePqZ95MpCrT7dexA8YlpB8OwcCSEGo7h+QrZwiSSkqBjWuRjCQRFQAlJKKJAQnN44b/8OVani49HXAihtdokM5McO/QYHiPC4SQe4HkyruOTihwkGz9JaWwf2fQUvifS77exXRPd0IkoU0RjMpIwwndcapU9NG2IaTr0tB67lVUMu4UomWxvrnHj0iYf/eCnOX7/Q7i2yXDQJxFPousWiijiOy5vv3GWu6t32NxcxzAGDIctGu09huKIZDHJ9s4WGzu7hNJppuenue+BIzz5xBlWVpaQZQF8l5AcqEIkUaS8t0c8kSCby9HsdTl/+SqqrKL3epy+7yRjxRJb61uELMgMPca8EHlXZjGZw7Ed2l5gFchEo4TjCXxZIpxLIRdS2GGZtfU1XFze/4mPMrs0j4+Ll1AptyvkJoocPHEUJRaiow/oxTyIyLzy7RfIJHOs7+7wlS99kd3GLstnjnL4vX9/dA/+fz5rP//n/+1Zz/UJR8NYlnmPXubSajXxfYL1huf/AEeZTCRxveBtaVnBbtDxvHtdywj79y/hOw6NWvVeBjawADebzR+Ew03bRtcNBoMh/Z6G44v0BkE7Rjd0fuFTn6C5t83m3W08y8VTFCgkEVJRRODUoaN4HY1YTKYnwNzJRe7evclo2GNh6hDpRJGOViaRjOG6Ct1Og52tVbrtNp4HgufSvlHl2PwRbNPkxo2bbO9UcV2R0w8/wvFTp9m8u8E7nn6K//3f/G9875vPc+zkUYa9HidO3kd5d5fzl65i2C63b6/SbLfYrFYRQiqF8THK/TaRsQy2F+AyGuUyjjEkkgDb7eN6FiPDYmHxCL5VoVYrMz51kJ3yNoaugW8QUkK0Wg1Go+BQRMMhVFFhvx9HNp3gIef6ODhEDZ9hZUByNs+g73Dtjbu8/6c+Bbjs7K0SCccoFicIR8cYDh267QFqyKPV3cFxHGxTRxCce0kvl253D21QxXMAN4QghNi3+CCN5h75QoROs0Y0FCYc9jE1nWtnN9nZ2ubAoWNcuHAOXBvBB8swEUWRSrlMt9MlFArhOhaOY2HoIyRFIZJIMF6aJBlJsH13K0DDxENsbW1Tq3VotdokYjGmJ0uEJAHX9XBth263g+fDpRs36fR6hNQwW7s7LMzPEvJt4rJHrVrDlSWUQor2qE8qHCEhhTixtI+oD2ooxuTYGLFYGN0yEEIKY7OThKMhoskE2VyeTrVBdbeCpg8RVJF9h1dYu36VgT5EjYSJRCPMT03THWq0Gk3ksIrjOYRlmUwmS28woLK1xy9++B9x5/ziF/7Hs4ahI8sSvW4vaJV4fkC7k4NMo24GAiJFVXE8n8FAw/N8EILprCAI7FtaIpmIUd7ZYdgf4LkumUwa8O+tXvzgM8f38D0f3bCwPZG9SpMzDz9IdbfM7PgE9993lINLU4RiEYbtBu988DTrd9aIiCp6e0DU8PHbfWRRIZ1M0R40KKzkicbyGIZAMjvGxPgsatSmWJxhurRCrbrB3t5tTMPGsQV8QWD7/AaqqVAr13A8j6GmYbg2qUSU9euX8UyLP/ujPyKVSpGMx/ixp58ml4zT63QplaY4cuwoRw4fIRqNcvXWDfZqVVx8REQU20aRRPykiqzKxJNpkD063S0iEQndMihNHSAsJ/ju579Fv6Mxe+gYq3cuo6gJ+o0dVtdukp8oMRpqjOWzdPUBkaaN3DOQRRdJiqBbOoIXiEkV22e41yY2l8P1BR546HEGfpeZmTHSiRjlSpVScT+Z3BSxaBpBGoHUR5GloNKn+KRTGazRCEEWkaUM8VgG3dLwvDj7Fx+k3lql297EFQwkMY7jbvPSV98iFRrDsHXMocahY0e4ef0GyViCfDSObpvIokI8lkCWFfBFXNvFsHQEwcO0AjcnpsDRU49y6P5HaXZ1yntlwopCOhlHVRUEUeLGzVWS8QiubyMoKhev3UY3DKbGp0mnEgEpQwmkzIeWZlmcHCcuCww6GhuVFh3LoG0NaVsjstEYi8Uim5t3kRSZuq4RTyUZ9PuBYj4WRlQkdtY2yBcKzB7YRyIep7pTZm5pgcWVA9y4dg3PcomIMuevXWVmfIJoPkmqmCedzHDj6nX+yUc/xl6jzqee/MiPfud0XZtQSGU0HJFMJoOnm+sSiUQQfALnphLCsix828V2XFwf9OEQJIlsJEwxm2PQb6P1ukEJWRDJ5/M4jnNv0htwREVRoa8NGZkmluXT6fSZnBinWS2Tz4TRBh1OHXkXF67c4K3vv8GhYwf5+o0LZLNptEaLUijOcORih31KsSiS4LGQGifZl1l3t8gkp4mGItQaO9y5e5Px0hJJdY7RaIRhBUmURCKBNtLQNZuuqqM7FoN+n1wqSUzw8PoDWp0h8YTIyr4FwhEJxfJ4/Ztfw3NdfHyiqoxu2giShOW5REJhisUSvWYL2bQZz+YRE1F2bAspFkYSZLAtZFmg3eiTTGVp7zW50HydT/6LT/Pit75DKhnDw+Hk6TNctHs0GlVwPSLxMDY+6USOpw6f5IXnnsfqdolHdOyhTr1ap5AL0DJxQUXfFYiPp7n+/JssPb3ArfUXiUXGSCRT6MMuhdwsrjmg1lxlaK8i+HFsR0QWPdrtJr7nE1ZDRGJRLGMPSTbwHR1z5NEf1LFdG8u2EKUyhVSGEyee4tqFq0iST62xSyKTYHJ6AsGFazduEEtnSSZTmKaObwQZ04B5nMZxLETJQ/Ad+r0q51/exnEhW5xEjWbotvfwXQtJlvBxmZgsMjIs2p0+d7fLWJbLeGkcwfPp97o8cP99bGxsIUST/OU3v89DJw4xkc8xVVIoREK0LZON/pC2J9JtbrF2d5t909NYuklUkjA8i86gy/zsPMXJcXY27mJbFrlSkYmpSS5evoisypx++FG+981vIZgeB+5bIZRMIrx1gflD+7ny5tvc/4EneeG/fYn3feLj/P7/+VkKxfw/eP5+eELINCmVSqTTaUKhEIPBANu2CYfC9xI/Qfi8PxjiIaAZBqbvEc9kmJmZoVQs0G+3GA0GKNL/J4EZDAbouo4o/h1IJBjRmyODbqtHNBLm+KElJnMx7j+ywv0nj/PUM+/mz77wZXZur/LwqUMkZAH6OkNzyOx0CUWwEUSfnb0GqDJvXbmF6CgsRBcQtSDkNeqZWKaNYQ4Z9Lvooz6KonD8+KPkC3NofQOn7/IHf/RfOXX6ISZLk2QTgSzXVlQ+/s9/gfsefxelmUkee+JxtnaqqLEclb0mjVqTXn9ApdYmHIlQqVWZnSkxVcxRTEaZyqcopePIePTanaDD6HtE4yEiMQkl5BNLSBhWj/0rCySzNl9/5ats13cRZREPATkUodluU5ycQlEUNK2PZVgM7zZ566vfwGzX6NbaxDNJEuN5apbFjmXz5tY2Fyo1jj3zFAcfeJxrF9/GdWpYdo9wVEQOi9Tae9TbbaKRBGPZCYa9PqJgIfgKITEBGCgKRKNJ+j2DfkvEM5OYuoWud+6lwwQyWYVM0uOFL1/iwMKp4J7sC/iIrK/dQQmH0W2LZDFLPp9DFH3anSaKIgXJMlFEllQUORSwg2IRdNeg3arh6hqtrVuoZlBZ29wuo6gqrgutjsb6ToUba1uE40kWFibJpsPMzJbIFVK02nXCYYnlffMsHznIdqNJc6Rz4cYdSvkcx6cmOZbJInZ6eJ5Hy7Qo9/uMbA/DC7wtTz3+bg4dOEwyEkdxBQ6dPE5+fIydzbv063WyqTR/9bk/Z3PjLrFUmiEedzfW0bUe1Z1NCgemqO1sse/BIzz31a+QSCYgGv0Hz98P/ax95aVvP1utVtH1UZACUmRkRcLQdRzPJRYLiHs+IpphgiCSLxRJpZJIvk+zXP5B2sUwTSAoJLuuExS17/F9nHvTWFGUUUMRGpVdHnv0AS5fu8H03CxvXrzJ66u3MCMCUVklqYQoFNIUElla5QoiAul0Esfok4upeKJEJl/g1IljDMwhjurjyg75yTlkVaXWWEPTe1iWhakPWF+/gjEaIBI0HL7/Z3+N2B3h4fDM+58hnEkzsbjAxSsX0HodZhbmGRk2jWqNVnkPCxPL9XF9ATWkcubx08RjUTY3thj0BlhDnXQ0jKKI7FTrOJ5PLBen79vowzbaYBfbHeF4LuFwhF6vx2DYYOXISfptjWxpFk3XGB/fT71yiUQqjTYcEAtH8VwB/cou1tBG0G1CkSivvH2RvjHi0MElIgIMRgaCJ3Ptwts88dMfQwzDdz//N3zoUz8LkopuG5gjgXx2imRSRevViYgKhUwWTTdBglQWTN1D0zxOn3qC6zcuYOgmipRjYfEIi4sZtipXUEWd9fMtahs+t25eJR6P0+0OUBSJdqvH8oFlHC+oBXY7bRzPJRKNIUoyruMFU2vLwPc9HNvB8Sxc14aRh0JgRpcFAQmPidIEg26fZrNJpVqnp43wZYnJyRKTE3mWDywyMTVGvphn3/Iyi0sLdNsNTp1YYXZuim6vTyyTYbfRIayqLJRyjMXixGMZyq0q3f4AOawQjkZI2wqDnSblO6scLGT5qSefYCYR59LF82w3m6TzeTL5FKlChvF9s5S3dyikc0xMjpFfHuetN9/GL/cpjE/Qb/dI5tOUFiYZ1ev8sw/9/XfOH/pZu76+RTqdwnU9hsMh4WigO/BDPp4Dvf6QvqZjOi7JZIp4IkGxVCKTSnHh3DlCsoRtOSiKgiLLePcqXp4HiiIhCjKW5TMyRkiKSrlc5sihA+ybL3JgcYlqvc2t1Tvsny1wPL+Ps5eu4KQjVA2duOWB44EYZWO7jSB3SSZipBJxTMujPWjy7ZdeI18aY9Pc4fRPv5OoKuH5EtHQOFOTB8gks6zeOYvnWvi+h2mMiCohtqp1Pv1LH6NtmkQiCYRuhO3tHcKqyuKR47z16mtUazVkWSY/lscXglB8SBGR/BEvfedVUok4iiCST8ZIqhKiGsZRwtz/yD7OvfEqo3oDsRTFEDQ8R0cOyYiSSjQaRUDHMYwACKaKjE+Ms7m3iijpGJbNzPwc185fxBY8xEafsCRgWA6mY2P2+sREhb1ai9poQDQcZiyXA1GgL1vY7oBMMctEfIxbX9vk4Y+9n0v9b1HMx5jIjaOGYkihJA2zSqfrBzOGrsvO3QrpdB7b73L23PPEoxFC6RgiBRrtDeoblzmw/Dh3bnyT3TWJVCrGrVurnH7ozD20J4TUKJcvXyaXzeHhk0jFGfVHmJaO4AWg778DOweJsGA1l8ymcWpD9JpGOhohFY8gyRKtbpuRPqSQzzM5PUkyX+Tc+Qusr61z8vAzGFof2zA4fPgwo5HO0tISmXQcwxgRiURRQyF2dstomsG24zIYjCgVUxwOS9SbKYa+iBQOsTC3hFVu8rEPPUMyIZHLxnEcj9nJaR58x2P8xcvP8/at62idLkJYJSRLZJIpttbWWD78PrZfX+WhM4/iazpDbcRooFFp1XnHI4/QTP0j35x/+P/87rMBDlMmkUgwHI3wfY9oPIE2MugPg1xsKpVGlERcBGq1GjubW8HbER+foDstSwogBEF1z0c3ghWJadrYpkmv1+XB+0/Q73bYt2+J18+doznUuXzrOpPzE0Q9gX/6T36S8to6qWgCBZep2TESqTT1dg8pnAi8Iz0NNRbjPT/xSUxfZm9nl1P3HcfPuHghCcf2MR2dcDiC68BQazMc1ZEkAdexsJo9juXn+OZz32Zqdg7Xdjn32hv0Ox2qe2XarQE4HktzU0xMjnPx+nWa9SqLc5Pcf+Iw+xZnyKVUGvUyruvi2h6pVJxqs8XIceh022RSSWRFwhSGDL0BsgIIPq4jMBoNEQQHEJmaWaZWrXLw5KOocgLTGrC3dYvBsI9pWZTCGUarFfYvLtBotRjoOhMTM3QaLYqxNIblUJifZWtvB9vziMTjPPT0o0gxFdV3uP3tl0hMTzI5dRpFDuN7EpVWGRQbbVjGGtm4voQtOKTiOSRRBkEkrAi0ql1GI5P9y2eod+7gOh0iYZ0bF9ZwNBXHlrEsk2KpwHCk02l1CUdClPf2KJXGAxuc7+ELPr1+D9e1AkCXAGJQxweC2KYgQCqZIiqHcPSgitjVhtTbPVBU+kOD7shir1xGlFbg5fUAACAASURBVEQK2Rxrd+5w8OB+kvEEvX6X8fESd+9ucOvWTR544AG2Njf5sSefYHoqz0SpQCGbZmyyxNtXb3N1s8KJpXniiozpWLznnY/xyz//c4TVMLu7NXK5PEJI5asvfJf/+pUvUtc15ESEwyePk85kaNabWLbJ5NQE3/riX3PosTNsX7rJ1MoiumWghsNM5ovslSskUyl+9ol/xEBoZnaK4UgL1hx+QCowLYee1qTT7ZHN5RGFwCtpGgbYNpKs4NoOvguuJxAORXEdG90I9qEAgiAiIJKIxmm2m0gCHFleQh8MmJ6eY7tS59y1O5i4zD16FPnkQV546VVufPkvKYQz/F//46/4t//i13n5hbf4ld/+NW6ubWAKUVq7W4QiMeKlCfafvJ9XXnmZXDTE9Vcv4e94LLzHwjIVZhYOk80VsU2H/FgObRiQ1xECqsGtq6u4lsjZF17jwLFlREfnox/6IN976WU8L8x2uUxxrIAcDtOs13jvU48RDYW4dfMWjmUjqQKFsRKj/ojBQOf1q6scOXIIz7bZK1fQtAGL+2cZy+bp9Qb4HvgIqLKK64sIYgA6rtdrWJ6JJNo8896P8r0XvowScXBGGv3tFtrWOrlQlHq1yv7lefYqDba2yqQnxug0mkiCSK/TQYipDH2LqGkhyRaep+MLPu3NDm//6Tf44L+5j1LhCILqIcpNWgMXQY6QzDg4I59GtUtT75JOF1hc2I/Wb9ENtRnqAyYnsnTNFo7WIZ/J4WtjiNIIZIvi+Dj1epNwOEwun0MUBTLpFOFQiM3NLYpjBWzfpTiWx9BGWJaN53tBk0kKGFWu42C7DoIvIYUkhqZBZ2ATjSXI57NB+qpRQ5BkJCVQgJTyOcaOHkQUI0zNzpJIxAlFw7R7HU6cOMa518+yvHKQSqNBKJYiX5IYPzmGIitMLU3ieDJ/+qef58zpE/zkO86AKLBX3qZa6bJXbrJX73J+a422YrA71LG6d5HDIt1ak/nlJcob28wtL9GuNfnkb/4SnUqD4vwUhXSWvdt3OfzgKWYmp/jP//4/4v6Qsc8PfXP+8R/9/rMQZGVdx2dkWEEm9p6vUBQDUrZtWiAE9DnLtpElmXA4zGg4ZDgcYugGju0gyTKiECjhPdej0+2RyqQoZbNMTpQQFJXtTou+7CLn4pRW5kknYty4fIWn3vteXjn3Orok8Puf/c/EIhF+/Tc+gxrOUR35/OJv/2ue+OD7ef6732V86QhiuER9bwPBNDF7GtlUGjkhkJycZHriGMXMGNFwnHavTru9RavVxDZt9EqfSTfOdC5OKhnnAx95H1PjY2xvb3P91hrdVp92q8bjH/oZ7pab1Nau47ugqCrRRIwjJ44zNbePq7dX2S03qbU6hNUwluPxyZ//NOVqlQfvP8XcVImOPaRhDpBViXQqh6JEMMwRkiQjSjLLB45Tq1e4futFGrUKIkP2yjexRj43vn+JuB8mm8zR6/fY3t5hfKxAJBZBCUnsdlpYro3pORTGCoBEfCzJmScfRlGTFCjxnS98nZ2du/iuzZEn3ollC4wsE8+z8OnR75VxTAtJchDxGGom4Wg4yFXbDpliBlvZBUHH00W+8Vdr9No9RCHESDdJp9NcvHiBhYUFBgON0WhEKpVmYnKCRqNBOBRiZmaGoR7wpzzXxXXs4I3qB1o93/OwTAPfMAgZFgtjJR647zjLs1PMT00y6DQxhkMcy8VzfI4fP8HU7DTZVIryTplXX3udhaUDRKIJstks5b0Kx06eotNsMj87zfjMLK12m3yhSDKRQhEUNtbWWdm/xAc/+AES2QRTcwt0BgP+4q++TLM34BOf/jRnnnyatY1tuoMGsVScfG6MXq9Du9kmHIoiuz7veNc7ubO2Gpje9RF7Wzu0y3UuXr3MS6+/ysGVQzzwzjN84MTjP/qe8/Of/9yzoiSjqCEGQw3T9pBl6QcClr/r0iUSqSDcroZAgFA4zHA4ZDQcIoqBizK48Lvouo53700xXixRL5cZy+fIFsb4zoVz/MSv/S+8dP51OvUqM1PTtO6WyUbTXL9zm5nleSRFZilT5I9+53f56l98CVST6NxhTpx8gNzEBO969F34gsCw3aW7t0qvWcUXfURLRhDDHH34YdR4nu6oh+708HwD02xhOzrYDnLLJdRxkW0bMZJkdt8JkuP7+M5LZ1nZfwBDN5meKLK3dh1F8Hnv+34cjDYjbcT2ToOdnTZXr95CFiSiskKpUMAxLXLZAsWpKf72ua9g9ZqUCklaTo+K0b6HdYzhOB6GZZDJFcEH3TAZ6hof/div0hw0qdZ3UQRo3e2hl4d89j/8B/7kzz6HGk2QSCaplSu845GHMSyDiKLSMUaIiQjzi0s0m3USoSiP//hThLo+n/2l30LyXWJqmE6lihAVWDl5AjwLQXAwjDaCOMQ2dSQf2vUOxUIJUZZo1Fvcd/9jINfp9raQnTiXXhjRbjRIxovEEiES8QSu6xEKqaiqSjgcQRIFRPzgYe26VCu79Hs9LMskGolim8Y9fpQQ9HVdF1Eg8F+6PsV4Dmdk0mk1qJXLtPa2kVwXVY1gOj4IEooqkUhE6Pf6GJZNNpfnwqWbfOmv/4anf/wDTMyvEErmiUTjrN6+TTGbResNqO7uUavWSSSSdFvtgJkruMwvLgJgmQZHjx4jnytw984dJNfn8Yd+jJCaYrO8SadfR1ZVECRMx6Ez6KBEVG6+9ia255HKZ+n3+9w4f4EHH3iIxeVl3vPUu/nDf/e7fObTv/ajH87//t//5Nler3dviOMTiYZpd/uUxopogwECEr4vMBhogYVMGwS2akn6wZ/n+vdI7kHAwDBNRFEOLGLdLrF0go3yHq9dPc8jP/shcgdmaPZ7LC4t0+0P6Az1IB00GiIOdRKWQiKc4vqV6+gjjbttm6d/+hfIZ8PUVm9Svn2es9//HhvXrxJWBCTHJhcLs1epIWkC+qiLoTqk83lkuY/vD3G9IY6tIeJh7w2wGybFwhi5iVm+9ZWvMjs1SaNW41tf+1ti8Ri3V28TjsVwXYe//cpfEo9k0TQDz3WQRJ/JyQl6nTa7O5uYhkEoEThG+toA0dB5/P7TvH35Om9trBHKhZEVlUw6i+fZJFMZPB9sxyAUjtPt9lFDsG/fwzxy+sNUrl3D3hmwu76DYerc3tigYWiMLc6imRY3V9fYbrZotmrMLy5S3dylW2khGTb5RJaXvvx1XvnS18ELBnSNoUGr2SJbyDF/3xE8WQn+X7uNYdRp1eqs3biL1jYwRja+LCKLUertMnKsjavbXH65g+eKtFseuqHRqLeJRGJUqxXSqRTNZpNCMRBfWYbO9tYmU7MzYDv0Om1URcW2THLZTFCM8H1c10OW5SAEIYoogkhzu8JErsDK8iJRWSIZC2NZNkPdoDvQSBeyPHjfSRZmp2n2uly+fg1Jkul0BszMzPLnn/8c3/z619nd2OYdDz2IospE4lHSuQzj09PMzs+TK2bJFXIsLi0yNjFBMp4grMhYIw3wScQTOLZNKqEy6jcpZXN87CM/T7EwR6mQZ3dvh0QujS+KiKqELfgkwhFWV9d45sMfQMkl2Vnd5OxrbzAxP8VjP/YYTxx48B93OB3XB1Ekl89w9PBhep0WkigF7pF7FTEAx3HwvXuQLVUNLNKKiut4mEMd13FpNRsIknRPZBpiamqCcqfOz/3r36Du9kiW0rz92uvsn9vHxsYm08uLqHGVTq+F3xuSU6IQDvHGlbeo6j32zS3zq8/+R4ajNvliGkcb8sb5y3z4kz/H1Re/SzYhgauTiMdwPZF6q01KTbE0WSAc7oG8TWtwlt5gCykiYQ5shIqA3tS5eWsdU9c5ePQg33/7ItF0kWQ8jOo73H/fKVzXD1ylrsDQk1h5x7v44M98kmq3w6FjJ+kPuqTTaUQ1TCqd4TOf+S1eevkltnbL3NneoWfpZItx5JSCEokQu2dvfvLx9+Dj0mzX8LzgDjZVmiSVHkcmROetSyymCsyOF/AMk05vwMAe0R9qTC8scvv2bcR0jF6nw9DWSSXiFNNZsokE5mBAZbuMNjLp9QdYrkNYldGHOpu3VsHyefjJ9+I5Dv1eja5WxdR6CJaC6Mv4vkc8ESWfX0RNWoSxSDceoLM3JJLI4joGsiwxOTkRSIAjEVRVpd/pooZUFCVMu11DFAViiTiiIBFRQ6iyhADYjokgeNh28DBXRDkIwQsCIVEG3aO8WcYY9Ckk44iAKwgsrRzk45/8aaYnxqmUK7x95RovvPoGg/6QaDzOzk6ZiYnpYIUTE6m06gxqfRYmpnnjlVcpFMcYn5rBlVRajRa2OcKzdfRhn9FwgGmYrK9t87d/+y26nT5zcwuIskqn28MamYxaAx45fYYzpx7hlee+xsLhZQzJJRYKofoirWaLaDzKsNtnbn6ebqXDL/zmr/LVL3+B5rDPzz/x0R/9cP7BH/7es7lchnNnz5NKxWk2m3iedy/CZwWiItPCc72AGes4uE7AY+11e6ytrmEbJpGQiufYjBXzJJMpIskEe90azZjLh3/pU1y49BZXLlwgO15k/8ohdtc3kSWZlROHuX39Jo21DaYK49ytVygbTZZPHEPow2//q3+Pko4Tzha59NZFfuOTP8lYtojsukSx6DZrIAqkEjFcT6A7GOLaNkJHx77TRNZ6TOQEpnMpcA0UT+TyN27i9Dz26g3a3TZLBw4wu2+Ff/mbv8Wf/PEfcnB5H5euXuGRRx/li1/6CvF0lr4rUGn2aNXrvPHCS9xZvUMqHiMZVomFw7g+nDxxlEajxeUbqziWy+LUOPl0jLbRw5BdJFlAVmBt8zb1eg1B8BEFGR+fbr+DKIK2VmX36k32GjV826S6vc1kPs9kOoVr+2yWdyhOTBCLRvAEER8RyRdJKGGMgU4klqQ7GCFLMrl0ksWFOZYX5pjIZxhPx4lKIl17RCgNjriLbpXpt9sszKyQyiVotbokU3GarS1k1+ev/933GZbhlee/jxiViSSiNBpVQqpKtxNIlgFy+TytVgNVVYN5RTSOYepEYwm2NjfxHYdIJAICGIaOod9DqN6b3Hqei2/bxH2JfCjKVDGPpMpojkcyN0Yyk0cNh9h/cIWvf+s7XLtxGxSF8alpRFUhkcqRzxfQ+13iyRixdJJBs023XuOJJ57gc3/6Z7z8/As4hsXmXo3S+AQgkEhG0Yc9RsM+mXSKUyeOc/DwQURZ5vqN63Q6Da5dfpu9u7cRbB1PG7J/dj9xV2Xz8jWqWp8n3/040XSCy1ev0jd1HMNidnqeV197ldrdLZ7+wPt4euXvt4z9UHnuo+98wH/qySc4d/Z1DMO5dzA9HMcKnJj3QF2+D6PRCO7pDTwPLMthqI3wXZdCLhPcK2IxDM+jqfeIjKeYXJknUkxhmzZXX34TRze5/13vQAjJXLt6hWg4RC5X4NjMErfPXaAbhVg2y4WzbxCP5vhnH/k009kc9/34h7l99g3uXnyRbrPLnRurpAspRFtHcnyKxTyNTp9ytcPhQweZmhjjzt1Nrt64RTQU5tCheSrtCnfv7hGRYwy1ISfvO0kuV0D0BL73/AscPnyAkCzz0suvcOr0fXzzO8/z1PveRzqb57nn/ietZo+JYhYPyCTTfPCZ9/Dm668w0DRsx0ENhdmtdxhoOsvz0xxdmsK3h9xp77GXMJGzUSQ5kCCF75nUfA8My0SWVJZzc5z9wnlq9QbxxVn8YY+UrFLZrnHmxHGa/RZ9w2J3r0IkHcfoa0yOT+DZLrIsUd6rIsgiy8uLPPzQfTTqDd548030kcnBpQM8/vjj3Ll7Bzed4WOf+WU64h3q2lXWr29w6a2LHFo5SCSq0Ou2GGxJXP7qTd73zNPc2d3hrQuXmUxncUQLdSyGExZJZnNBTtbxyGQyDLUu6WwGAQXB8bh55ybjk7OUd7eCgIYsoQ11opEwtm0hiQrmPU2fIkvY2pDoUGRfdpa126tsV2ukMtkgkysEE892u45jeyRTKcKJCI4vUqtXmChMUMpksUZDLN/hyJGD3Ll+i1KpSExVSSQieDhMTk/SabZZOXSQZquBbZvMTE/R7XWJRKIcPnIMX5TodAd4ns/0zDiDbgdJVpFlhWw8wZ07d9AMKI1P8ZcvfZWrrW2a/S6pdAZPcJmYHMdqGbx19TJRRWFxZYk3fu+5H12e+we/95+e7fZ695yGHh7BpNb173F9EHFtB8e0glXLPRPZYDBCH47Ip5JMjRUQVJmBY1DxBux/6n4mji2iSQ5IcPjYEeYW5ug7OqfedYaN67e5efEaiXgSbXWXmeI0q2cvsXj6OF1nhGB6GK0+yyeP8Ma3v837n/gx9tZvYVQ3MAddwrJPRLE5sH+ZerXG+naZZCrF8spBjh4/wXA05Oy5N7l4+TqyEiIaDWEaHvvnF5mbmiCfyRGOJ8H12K2X6bZ6ZNIJEAWe+8Y3WVyYZ3N9i0w+w2//q88wtzLP448/xuvnz+NIKmcefIheq8WbZ89SKBRIZ1OMRsN7khsFyzTJpNMMDZNiaZbp6Tk0c8QAHVGRCKkhLNvCdT0ikSiu4+M4PsmWS3Pj/6XsTZ8su+/zvs9Z7zl33+/te3vfZunZZ7AMMAOAAEmAImWJZBRaSlKyYqUqkR1HkZw4VipVLJdTcVKOqxxZiSNLsqJYSizKlCKKIAWRBIghgMHMAJh9unt6X2/ffTvnnv3kRbP0KmSKf8F59dT3fL+/53k+h0xlc8wkE1yanWO6OgaSRGFiFpBpNhpUSiWqhSIiIvlsijAIUFWNdDqFrsd4urHJ/eUVduoNkrkc1XSefCpLZXIcUZLYeLrK1toGi2fPki7EsEYtjp42WL+/TSafQBOi3Pg/7/AzX/w5bn30MbutJnIIufCYxHX/g/tMnZzGCt1jo0kIsnocI6w3mqiyihbVUCMKcS1CMpFAjURoN5okkwnMkYGqKgTh8YHRdV0IjvdPt2+z8XiT/mCIHo1hmCa6pqJpKq7rMDUxwSsvv4gshTQazeODUDKNQsj5UwukkjGimsZoNCKiqSzOzRFPqExUilRLBVKJKLIoIIQ+qUScmBYjDEOy2QylYp5eu817N97BdmxOnjrF0/VNIrEoY5UxNO244rNYzKKrAndufUhSiPLc5ecYCi6d0YBPXf8UH/7xt5k6tcji7DxTJ+fIlwv8e8++/pO/c6ZSKXzHxbM9ZFk6Br2Gwg/dGwKEwXF9hSgRItFot+l2BqTiCUqZNLoWYb/ZYiB4vPDF12gNWyi6wsHTLSKhgKKIrD5c5vSF83zmtc/y6ON7tPcOmDt7gubTPb706c/xJ9/8BuFEmsulBFk7zQc/uMHiwinSsSRmMsWg1+TDd7/P1FgGYzBkf2+PXLFEoKaIpPKcPl+mkE2ysrKCqmhYtsfO9j5RReL0dJHL509THa9QbzVJ6FFCJP7NX7xFeWaGnfo2pUiMYaPF/mGdUqHEUa1OsVziv/5vf51/8Vv/nHdu3UFOR7D9gJdPnmf30X3OzE3zYa/H/adbKEJIIhYlHtOo5rJkMlkWFhbRozrDfp/Dx1tI3pB4UWYUCY+F6R17lm3RRhDAGY0wOgIvPXMRwwtI5gqYhkE8kUIQ9xiNTCxrRCIWPUbutTqMl4sIIoixgG63T73ROLZXphOoaoRCPkUxnSFwPNzgGCYlyRLTk1W2Hi3z6OtZnv3KPNagiTEa4LguB+sDHr37CfnCOG9++9v0hgPSpQLj5TL1nRqHBw2y+SK9J4eMYiKF+TKhGDLotdBjCUqlAiPjmKMaieisb60zPT2L5znEk/EfusaOKd6e6yOqEIkoOPZxYEKUIaIpZBLJ41xxJoWiqJw4eYpWu8HizDSLJ2bJpuIsL6+yOH+CkTFiOOziCyHrW2s898KLWCObnbVNbt+9RzQa5cbtB+A4ZOJxpuenkGMREqkkgecwWSkdR9DsEYlEglML8wgRjRvvvoOsJ/EFhVCMUyrl6AxGuEGIFlPJlOIkkjne+fATev6Aqxef45Pvv8+LX36DdCHH/Q/vUD41R+B4P1J/P1acYRgwHA6Op6UXokeiOI6D5XoIggShgIBEEIpIkooxtMil0xTyWVqDHjvDNtnpKmcvn8RNCpTHJ5mYmMAYGNQP6xxubOG5HvWtGo3Dfa699gr5qXGidkhKlFk4OU/sYRZBiPLht9/nxOkTzBTHkQhRRYH/7td+neHRPtlkklQ6TzRyfICaOnOR1Ng87737PomozkAW0KMaUT1BzA959rlLhGHAzHiJIfB7f/5Nlh+t89pzl3jjpWe5fGaBBzs1RBECs82piTyCbWH4IjNzU+zs7PD3f/Xv4ypw6twCpVIZt92lGtHoRFXu7OywbrQoF8oIpocoqXi+gBbRqR01WHm6fFyt0uujhh65lI4qhXiagpJQkCURx/GAkDCEhBtle6PLht1kbLKIFYBrmuSiGr/8pc8zMIbUehXur27Q6/ZJJ2Lcu/cAWRL/+mjnI5BOxojGYhAIhJ7EYDCi3+0xPjbG1vIqpYkqeiRKKp1GHNr4aw7j8RLFeBT3rod30Cfr6WC7FLI5hoZBPB7lsHZAppRipIRYikD18nn2t/Zo7XTIVzM4gU1nNCSRzSGJCsNhD0VVmZ2dYzAYoAjiMaGu3iEW03E9B13TkSUZXwpxHDCdESfnpjGEFhFUAgIi0RhLZ86zvbVDv9vl3Rs3OKztEo1FiWtx1FDGQkSSdL751vdIpTL80de+ztn5WUrZCql0ge/94DssnJg+xo30Riw/2eDR/haxsQJu6FDI58hHVK6fO4sc7FMq5IkLkIhEOBgZvPnWN3i0ukyxWMBzbVzHZaYyzhdff530RJnJ1hRv/fENwsBn7tQiR2YPvy8RK2W4+YN3OXty6Ufq78funK++9Ezoed4xMDc8hto6joPr+4CAY7uYlovt+gQBqKJALKozEDyEuIqcjCDlE5w5f4Yw9Lj9/ofE43HGKhW27q9giwFKVGXp4gXiiTj5fI4ntz5h1DFxPJ/np6Z5ur7FWPUEax99wn/xD/4hH9/6kJWdxzimRTWZpDo2xsb6FmIQUMxnkBUZQdX54IOP8L2QZDrJ9OQE6WSSwcBgd28fwxjiCQKxVILuyMCzXURkkqrMF16+gK7r/O6ffJPAD1gcy4Pt4oXQ7nZYnJymOxhw8ewSpnMMZJo4cxJDkjlstvjWje9j4tHYO+DU2BSKYTNerpLNp2m0OuwcHrK6vYUkSxSSccqJBFFJRIoLuNMaRizA8TwczzsuTBZUpL0QY7fHyy+8gBJRqdfbjMwRtu1Qa9Rpd/qMbAuCY7vb+Pg4ljGg2x3SGYzAD0hnk2gijFfLLC6cRIlEqO3t4o5MwsDDdWzUaIxMpUKgSDx+9IRUVuGLv/oqrZ0a1oM+RtOkbws8/6lXyRZK/Mvf+V3urTzk6rVnmMyXuL28jBtROdzfI10pcrBzwMULF+n36+iST9Pvo+eTRBMJRI6b3COKSuD7mEODiCQxHHRxnBGu55FJpXFdD8u0GfUNvnTtDZSux+H+AYl4kl6vx8iyMUc2g8GAYqlAMhHn6cY69cP2MWpRjlCqVjBNg7h2/C3Rd5g6eZbCwmX++Bv/ik99+lXee+dtxqIpSvEk+40GVkzh0OqiRBTi8Tit1U1OVqv81PUXiesqxbEKjzZ3eW/5CQNFQApCMmN5GoMuoeWSUqMkS3mePl7hyovPcurKWfafbrJ+tEcyHuOoXce1Xdz6gJu/95c/+c75+7/321+VJRnX9RFFEcMwcb3g+DfWh8HAZGiOsG2XVCqJGJFwoyKXXrjC5evPopXSdPstdjbWONjeQVJkypUyQ8vEcgKqY1VmFmZptOoksxnuvXsLNaqxt7vNudPn6Ow3+MKrf4NkPMn2yiNSySyvfOaLTBSqMDIRAkglsxSKeXr9HtlSBdPyMCyPscoET58+pT8c4jkugePihwFBcAwwSiaSJHUd0QsQHI+F8RK5uMrB/iGEAuOlMomIBl7AyLIZGCOazQ59z2dufpanW+vc+MFtNlstbi+v8J07t9nutXjt059mbe0pju+SzWaJaRpzM1PUj474i3duYKsK8UIGMa4hSRIhAtXqFNlUlrYxINAlQjH866ZCz/K5tvQMk/kKg8Fxn+vdu/dxPY9sLs9oZGGNTHLpNKlYDFmSMUcWpXyO8fEqre6QSDTOG298hlNzM1w4f5EHD+9xeLjP4uIcr3/2NZLJGE9XVxn0urQbTfqdAZKqsHh9Fn0+zeqtTcJGyIUTJ4koEt/6ztt8cOc2tmvTb/e4fuUZfvZTr/Ho4Qqry6uMTU2QzKZxPY+D2hG6FGF01CMRyKgRlTACQugSix2HpS3bRtO04/ddx0ZVZILARxCO/94C30MWJby+g9u38WyXxlGD4bB/XEF5VGN6egrfD7j/cJlGu09c1YjKIplkjG6vgxbRUSWZTDbF2FgFURQYDtvMnZ5h/eiAWr1OStV47tIFDusNGs0GUlLj7JXzxKMxHMulWWvS6A9xBQkLSGfyDIcjhqbJ5avP0HFMClMVvL6B4TtEVIU+NvnpMe5+/DFdc0iv2UJ3OKac1Qf8xpd+mcWF0z/5zunYHkEQIooClmXj+QEg0u0OiMUT+CHkMllMx8JSfLJTJabOL1CcnmDoOnTrAxZOnUAg5GB7j9rWAUfLO2QqBWJRlbmlBToHR8QkneW7D5HFENN1mYhX+MIzL5O9HgUpxtaDB0iBy1t/9W0anS6Lp85y6vwFPrn5A7r9HrIk43gCa5s7GP0hakSn3x8gRjUCKcQTQ3q9LqVinmIuR+j5dLttpCDG1FgeL0jjuRam57LZaNEVZaJajCAAVxTZHfTQBIVAj3KvXufgicn1+UWikQgXFxZI5BLcun8Xw7D5t//r7yAldZK5NK1Bj8XqPJIfIkdjnL94ntmTc+SKGeqHR1SyBez+kCePa8BYlgAAIABJREFUn6BPTnL17PN8dHCP/VHvuAVfVTBqPSiFPLm/Qm8wYHxyklQixeFeDdwQ2xxSyaaIx+JoEQU/hHqrh0yI4DtEBI9sNoPZPuLKtedIp/NE5BfRVZVCIU/9YI/9zTUyqQS6lkeVJeq9PuGYTvHyNNnyGFFxk8AeMOz1SCki107Nsbp9wPyZ09zRVWxzwD/9H/57/ulv/XP+k7/3awgjm9bGFsVkGl+LomgRIok41xZOsby6xpP1A1KVJLY4RFZFREXGNE0iiooW02k3GggEDI0BmqoiyyKSovJ4aw19oDJTHEOTBTRVxR2Z5JJxNE3hxu17SGoUPxSYHy+TUyVUVWW302cQSkSiEb74lZ9laek0/sjkW9/6SyLJFB+/e5MLhUmuXbzMjZvvMTYzy367STZUqIgxDup1JjQdK5tnYNgcErKxuY7jP0FWJC49f5HQsjha38LfFkllM3iBTX3YJhGLk1R01gZD+qbJ+FiVqKBi1Y6YTYwhjn70n+uPnZy/89v/8qujkYXt2FgjF3PkoukxBoYFokwgCEgRGSsSMnPpJEPXYGZpgbXNNeQwRI9EkBSZkWUiCxLxVBLHdSiWi6ixKOuP1xjZIxZPL6BENELPQ/MlfuPv/Df4R3vcvf+Y/+Vf/CZm45DPPX8Z07H45P59/MBm5ePbCF6IYZiIikx1YpzAC+j2ukiCwMPVJwzVgNL8NIHrEXFcWrU6h40WalQnmUjQ7naxhJC1+iH3Wvs87beIzo0xPjvP9t4eR/s17m+sUamWmMhkkZJxzJSMUEqx06wxe2KG2sYmyWic+fES5yYmSEYUfDekWKkQTSYoawm2tvfxJZG/+yv/KV/8wudZmJ7m0qlT6KJIIZMhk0zSrjdo7tcZtvtYrosbegg+REcKVtNEi2gYwxGh4xK4Dgldp5DLsrm5xqWL5+m0WkQiOvXaEYPhCD+Esco4BD75bBJJDGg3WtQbDXqtJtgWvXqNBx/dwjBMcsUSCydPoEghyfE0l/79FyienaT/tIG30aexUycejyOJAngelXyalCJweW6KsqayMF7FNrsojsmJQoELkxUWymVee/YZ7G6PTr3O6y9eY9AcsrFfQxJUrIFDLh0lGouix3Usx0SWZSQJPMch9IMf8lIVBEnAd0IwA3KxJAlVJh6RmawU0HSN/UafZtfE8yxyus61s6fQZJFUMsFWc4AZwq/8yi9x+bkX2d3e4b13vsvRbp1Pbt2mFI2SjcXY3d5hMBrhmiZLk1PofoDYM5jLFUnpCQzXY7fdwJB8xLhCfqKME3jsdetYoYdg2DRqdVwhRBAFIpqOIqk8uveAyngVwzSZLJVRiPDk9id86ZXXEWw4c/HCT25C+Ne/879/1RyZtLt9XE9gYJpEIjpj5Qora6uIUZXMfJX0TBFXCjBtk8ZRHVEEOQRcn5Fp4hHSbncwzSHTp+eJqCq7T7eRQpg+Pc+Tj+/j1Ib4Q5f/8j/628RVhZX3f8AfvflXpMfLWMaQajrFzFgW27W5d+8+PcPAs2zCQEBVZWRRxPVsotEolflpVtv7pKo5FmaniQegB2B7PnI0Tm9kMPAdCmMVvCBgq7aLXEmTqhSIxWOs7myheMeMlpiuc2lmgbWdHda8PrFqFikio+gR0qcmkW2f7a0dRo5AOqGxOFWlFI0ieyHtVpexQpk/+8736A76hK5HQo/jGCOePHjABz94j09uf4RvW3S7HQzDod8aEpouEWScvk3c1+gethh0BxhDg2Q8hu+66GoEY2gQ+gGVyhi5XJrLl84zGtkcNrvoyTwRXUfXj+lv7e4A1/XZ2T9kY2WVqBwgeBaBKGEGIvFMFi0eJ13OMX1ticqVM7i2x90/+R7Nhwdkszlsz2P/qMXGXo1at0d7MKLZNRCUCEeNDt9//yZBEJJOpFCUCHokApbByWqFM5NTPLj3MYEXMKh3GQ5Mor6KXbfRHQ9RhEw2fdzmLvr4rgN+iCgIeK6FJIIsqMSIMBaJcnZhmtmJMcIgxBFU7jxYJZdKMVktUy0UmZ+ewLVHmKHMzUcrFMp5ioUcy8tP+fq//Xdsrm7jeiH9QY9SIYsoSaxu7XF6aQlJkTk4OMAKA7qmQc92qI1M+qLPSPZIjmUQwhBzYNLcqeEHPoKuEloe2UoZK/TRdJ3T585gDHuEkkgpUyCuaUhahI279xkvltlcfsp4tsiVq1d/cnHeePfdryqqRjyRQ9VitFu940hY6GP4BmMLVcYuzLH07AU8zyEE8mNlZERGnQGSL+D2DA4PDokXs+RKBTzDYfXhCr1hHz2VZG5ympgaIyGoeLstXr50mdHRAbIfMDZ/jiDwGXYHDLoDxgoFJgtpVF1jo9HEd30SWgTftRn2u1SnZ5laWGS9vktqsghCiDSySPkB/VaXQI1giAEb9X30So4Lzz2HbLtIrofRGyDIMsNBD0GWmV1YwDINxqNpLHPE7nCAERVRUirpXJpBq8uZpXM0dmsgivRaHYamTbvTp1zIMTs+STqeoXZ4RGPYJZnPsbmxxdrjJ9T397j/0S1M06TdM2i3+xgji2hMRxR8RC9EdmWMuoEaiIS2hyyKx6Xdjo3vuscdv2FAoVRA1VREUcYLQ2ZnphkOB5w+f5GV5cfkczma9QY+Al4YkIzH0EKX+fEyHdtlb+QRyjILszP4rk9hrsyF119Bj5U5vLlC7cNlxjMFCHw2tnY4bPUYuD4iCvFkmlqnw+SJRdZ29+iYNmYgMQpE9ttDvv/ebVZ39jnz7POIqkYsk8foDBFcj7Qe5dzcFAlFJqfk6e/1Gc8WEZ0Q37OIxxX80RDVA9Hy0QUZ0QwZT2Q5PzNFpZhDkiRqjS47tQ6D0YhoNEIYQLkwxvzMDMl0iiCis7zylBevXuUP/83/zc7mFnIYksvmWVlfoVoZpzcYsVurky0WKZXKIAg07CFGUsZLKNgykNQ46DWozIwTuD79Zg8RCde2UUQJRVGpzs+SyWXo9/vIooTVGxBNJEilUuD59PfrXJycpuwoJHoBsuly9eUXmD+59JOL8+69+189cWqJcqWKbdtsb+/gCB5SXCYxnqKyMI4X+gy7PRqHR9hDk3gsRjKZpNHuUJqZRI/G2FxewzUt9HjsmH0SAhIkkgmCocdkcZxTM3OU4mnaR3Xu3rqF4cL0whnyiQQCAo+Xn7K1e8TVK+fJxDQM06E9OL7A9boddDVKrjKNkshgGgOatRrV3Bih4dLYrdNr96n1evRFHymnkxkv0h4OONzbI+L6ZCQVMQwQshqhKvHcpStE+hYMbZZ3d0hMjqEVE+jJKLlSAUWQSQ1Crp25xP21VWRdwRoMMYYWohyj1h/xpa/8LTzPYTjs0rcsRqGLg8fy6jIJPUpMj9ExbVRJRZEE9IhCGITH+6ao4DsesiAd2yVtlyAUEIWAKxdPkU7opNJJEskU9XaH7f0DhoZJsVym2+vQ7/eIqhEaR0fIisza/h6CLFHJpLhyeoGDVoenzQHVhXkK6eMkRhiGjFdLTFQr1NY3+fYffYPOboPxUhHPc1g8eYKBaXHY7DAwRliuTT6XQgpdpqeKTE5WaDb7BKpONFPmmWefozw1xyuv/zSdvsHQ9HjzT/+CWCyO7XsktQgnFmeJ6Vk69S66p3L09BBsAUkQMBp94nYUxVQQDRHZEliamOT03BSyKIIgc2d5m54NuVwWWZZIZvNIocjpxVks06Db76PICr12F02LEFgjIqJIt9eh0WoiILK9e8DC4hxnz51kZ2udvaMDWlaPZClNcXyMzESRycU5Bn3jOO5IQDyXpro4iy8GVCaqx2BoOWR7Y4uDrV1GpoUuKFQnxnFNl6inkBiJzIsxEpE0f3X7A04+c5a1wy1ef/WnfvKD0HvvvU88HueNN97gz//s68gJiUg2TrScRIjLzFw5y9HhIfjB8eKuqexubZLNFrCNEd1GG0SBTKmI5dlouo6mqoSWQy6fRU/ESQQqr119kZlyhRtvfZc7H9ym1TZ5cvCEmi3z6muvYrkwGNpoEZVuGCWbi3NivE+r0SGbTROPRVHiGVpDh6KeYOPxOoN2A3e3w+zEJLmxcfYUHcQQU/FwhID2XgM34dAZ9JGiMXKIFKNRjowRh9aI23/6FicK49xZf0rdN6loRXJjZXqdNsN2D7vdY/LCGcaq49R7HfxMhMlSCo6GtHpdVut9qp/c5ac//7OIksiffvdN5EyUSDLKMBzR8d3jEi1VRZUVRM9B8I9zroY5IhQc3DCEMECL6lTHxugPBqRSGpXxEkEQIsgqgiSjJ3SUPZF0KoMgHE/Y+YUqghuwLVho8SiS6hFVdSYKeRxB4mBko2o6z545h2fZPH6ySmVukZ29Lo9+8//AtiyePlonFCUGj5eZKhWYyeS4sHSCublpPrj5EY5jISoRNnf28IIy6VgM3/PpdXpMzS/y8huv8/jxYx7dvcvDD96ldlQnGovTHgxRNI0n6we0Bx6XLz9PRNM5aNZxRZH+fg9zfUgoyuiJCKokoWsqi1PjnBgvEY8dg4dXt48Y+QpB6HD+wiU8x6YyNcv/9fu/z9ZOlcC1eP/2h8TTWRJalNCz8RyLoe+g6yqvvPgMASELC1VeuP4Ms7MTXD4/w8A02akdYXsBtVqDeDxDb6uG7gmICY1iPkMmnyFTzDOZSTM+Psmj+w/Z2N3BGg7Jp9Kk0ylMb8TB3i4ZPYNd64ELy0ddzGjA1PVL3HPahLbzI/X3Yyfnf/Vrv/7Vg4MakYiKbRtMP7+Emo8xcgbIMR0pGmdkWaTSKaSIghIIuB0TXdfREjqKLHGws4ueipOpligU8xzs7WP1h+SLRYaDIbP5Ma6dv8L2yip3P7zF6uoGM/MLNJpN9HSKN778c5imje/4nFhaIpUv0u31kQIf27KP4aOyTGFqkp2DfVqHh4zaTaaLZfp9k1K5ghRRWdnbJTlRIlstELgBnXoLQYJoLkpiqkjPNIkcM6/x2kOykSQrm1sc2H3yp6dITuQYjobYtkOn3UIQBTLFAjdu3mJ1Z51zn3qBmbMnUS2P/e1dDN/no/VHGOYIOxR4+doL3HnwMVoqRiGfJSZFaLe6RASRQjLFyDSJx+MEokDPsui7Dleuv0B1cpyZ2UmWlhaRZREB2N2v0RkYOJ5/3FDebP71G+fO+lPmJ6sYvQ6KKJCJa0wV0kwXM2iBRylb4P1P7iGnEgjRCIbgs91oU5k/x0uf/TKfPHjCH33ta8RSRSQ1wtbRASNZwlVVao0W42MVUnEdZ2gQ1XQCBI6abWq1OlefuUI0HqfV7XL69Gnefftt3n37uzx58pidnW2mKmPs7O3g+B7PPfc8/WGfhRNLtFotHq+vctCuE6gCyWiESr6EomhkExmK6QzFbIbxXJqZahnP99nbr/POh/fxQtjY3MAOfS5euMTk/CJvfuvbBKFHbdBjq9mk79iEtoPnHIOyXvn0i4yN57lw4STzi5NcvLBENhPHMoekUnF0TUYTZCrZHKqsEPZHbDxeo93rEcukmJydYnpqAtd2qB/WyKYyyLJKa/+IuKwylc1RjcYo6VGmM1muXb5Av3ZIOpdie9jl0c5TXvqZz7NT2yNfyfPl59/4ySen5/uI0ZB3H9zi7PkzzFw8Q9Pq8sE7tWPw5w/T7N1aE2M0whtZJJUYcSFCezSg2enhE+D4DkG3w75p09xrcPbCGdr1FnZ9wKt/4xoPb97mk5u3cL2A00tLZDM5Stcz7O1tk1M8ylFYGdZ59LCG5wXU9/bAMshn0wShiBiJHZvF200aq08pF4uoyRyzpQk+fvKA9EQObX4MOyZARCCURQbmgBEWF69cRY/HGLkunqChtQ2SaoQwDGgEBhd/6hrNYY+jg0MkTSFbKOB6LkJUZ89o8+BoGzkRRdQUXCHkUe+QicoYccvF0RT+9R//PqWxcd64fp3JSpVRJCRqh/Q6bZKihiap7B4eYnkuSjqBhY+Y0Ll2/XmSmSyS64FpcLC3j9ntUTuoEQQhfhiCHWIrfUb9Hr4oAR7JaIxmo8X9ByvEUilS8Si5mELo2HQ6bTr1PqFh0HcspESMB4MelqiQKE+TroyRKuVRC0nk8QxVqUgkrRMrlnAtCwyXMJrmBz94l8tnFokldNrdLlMzYwgIjE1VyVdKePhsrj7GMxzmJsY5arVoewHv3/mEailHt9dn0O/jE3Jn9QH9ThdnZOD5Do7rkC+XmatOsXdwRESSSOgquhzgmwN812HQN1g/aFPrGCQQMUOf3dohjX6XsjUimoqz0tgnmUohJ1I0Oy0KkSimNeSlT13l9c+/hm0NsUwDRVHoD/ukxAR6RMMyR2gR/bhoe6zE/t4BtaMmo4FJuVJgaNnc/uQum7UDEpk0PdchORoSyiHnX75CQtW59+5NAk2nnMuycHqRB7dvs732lPm5WapRmVDWOLx1l89ev4av/H/6D/7/J+cf/tnXvzrzzCKnXrnEk51NfCHEk0IURaFdbxIQUC7lGbS67KxvYQwMRMsn8AOarRaGa1GcKBPXdAadHoPRkHgsCq7IdH6WxfIcrZ1D7rz3PoeNFoIWByQ0RSOejPDitYsU01H2n9znrW+/hW8b7Kwts7e1yXBoUJ2YYGq8hOsEWKaNPbCIRxOIiRTXP/dZntY2ODQbiAmNsRNTKHGFeq9NRJZYnJrmyvWrxEpZDrf3Odw/JKvpnMmU6A0NNs0285fPMQxs+sMuSiRCZXqSS6fPMugMqBSK7D5cJVrNceH5S6x8dI9Ak+l0e0yNTbFzsEc2FqecSiIELmvbT5mZmKS2e0jMFtlc3fxhdxA4novr+zS7XeZPLiJFVMSoyn67wW7rkMfra1hhQGc4oG0YHLZbeKGMYdhY1oh8rkDXcVjb26Hvemx3etQ6bUjGaXQ7jAg57A7pWA6uHxKJqASOi2dZ+MMRguPS6BxiCQamb3DUrWOGDq5jMTM7jSBC56jB2ROL5BNRKqUs45NjqBGZfCnL6aUTTE1P4nkOzUaDaqVCKh4jl04jhj69VoMLZ5eo1WpMjFeQZZF4Isr2zg4PV5bRE1FkVSalx3EtB9+2GA0NMpk0p5ZOUyzliSsCpUwSWRK5u7xJ3RTYPqzRtvqce+kKYVRhaBpEFY2nu2t4yjGKvlgs0u10mEilicZkZmbGaTUaeI6DJMnIsooiSWhahGw+y9CweOfWffYsm8kTS3z35i16vku71T4OLWRTJDPpYxNB6JPOZahOTNDtd+gZfZRcnFs33kfJpumJIVuDFp+srzMURfa295i/cIphv8+nzj5DVNBY/mSZn3rpMz/55CyenaKyNMEodBibLHPh8nlu3HiXQqXM0BhyuLvPzPwUVuBw8vRJOrUmGUFn4Fs4ckgikyYSjxLYLqqsMGi38WQL2RSRoz6hYLFbO6LfGzIzN4cfiDx+9JCF61d5cus2P/9Lf5P/+R/9E4KByc7eHmXbZL5a5UQhh6hE6He7qIrE0eERrVaT+bkFZpbOctRvY6gjBsqQCy9fYnNjn9XlZRbKBV6eWWTr6IBGt8u5uUm6Zo/WUYOIHxLtDBlJCTYGXeILEwRxic5uh/xYEaM94I2rn+Z7771LSkvQbXWxIyJpUURPRdGjOtbAJBZNsWf3afg2+VyC169/nt/6n/4ZfhjQOToiI+nHpdKyyF6zRjKeolosoHkOoSpz+fwFbt+/x0cPHhBqMtGYTm3YxdJkXN9GTuhMTIxT1BJ4pgmSRGp2lrUndzlUbeKTWUY9g0S0Sn5xhicf3MYVdVJTBULHx1BUXE0Gy6azf8SYnsQUoNau8f3736c0VkaKSdSbh9golGJJni6vEU/EqFSLvPfdt/HsEdNjZZqNI06fWmLQN1FkFUWS6baHDHoWruNSb7Z48OAx3X4fy3IRAoGjow7gEwYB+UwGJIUghEFvgBuapGIxCpk0oe+QTev0Wkd06k3my2n0xHEe9aA9wAlUYnGdVq2BYfZo1pp0jxp89nNv4Ig+8VSKVqtF0DxispBnqlqi0T8iCHwCz8HzXJCPmxG1iMpoZHPzzj2+//EjVptNTl+9yO997U/oRQLmnjmN1igTuj6j0EcKXGRfIrAdcFxG5pD+oE9ta4e1x49JFNNkkwlkQeb2h3eJFVK4YsjZsy9x4+P7XL3+PPOvPMs//o1/TKI08SP192PFOX5iGimhkUmXSUfjHNUOkEWJerNBYbKMKAv4tkfvoI6SCZiemOZo94CW0UNWJApjBURVpd5qkYvHOJ/I090+YDyTwtpZY73RZqwyRTqZxrMshv0RuhQi2kOmCnl+9zd/G1HUeNDcpaVJNNt1ataIy+dOoA08xjIF1rb2yGfSzJYKpONRnKMtZNPg5l/8OR3bQDYcFrUIE3MLKKLC9n4DHzBNhw/fex81G0EQYUqPcSpT5n67ztNujQv6FHZEYO7EHLZn4+21SewNePzhbWZm5tk93COaTrLx/n3OPfs8C5ev8OTeAzLIvHTtGl87OGRzb5NW/CUiswViepSO6pHMpIindWKSj1BJ4zgeHdtECwUqyTyZZJrPvPwyuVKJd957j08/d5UPzZs8WV3h1JmTxFSNS4unyasabujjWQG2JGO4NpXZKbInKgzW9ykm0nieh5SKEiZVvvDzX+bb3/wWgiRxsL/PhdNn8dsjlHSSsfkJ0jJ0zR7m0MAdGOR8iUgIK4+XUUpZkuU8f/XR+zhmn5lcnhMLCxiuy/fe+4iYIhOIEXzPpZhNUi7k2NvexrYd5uZmyZZKEAq8/fbbKIpARIn+ELugEU/EGA5tmgc1dF2nWsrhWCNa7Q7D4QgFkUI8ilxOEoY+7aFFPFekkh/D9jzSukIVnenxmeN31aM6p8eqiLpO96COJfqUszkKYyWeeek59rbXOH/uNK16g72tLdbuP6ZUrLCxW+eDx6vsDgdcevU6xeIY3//e93nu1RdIFZKsrT2hNFFFEyTW7jzCsx2mZibxZQn7sEZq5GJYIaur21x55UV8AtqDDloyThD4dAZ91g730LNxUhMlvvbBX6LO5wikH+0Q+rHixDK5//ZdErk05YkyzcM66082GZuuMPIDzi6dx+gNkDyBrmcjdFsMjAGFQgE7sLEMg4ymkwvjOKuHx00HxQKe69M2TU7MT9ORZb788/8Bf/C7/wpfELn2wrNMFrO8+4MtPvr4LtlSBamQJpdZYHFimo29TVZ7fTIdn0G9S6vd5EI2ieFZOM0eoe/iux7peJyIB8Jug0Qxh43KsjNklE0gRuMo/R6H21vMpk8gDEeMxUosHx7Q0HwWXrqIG5eIZxLElAjS9hEvXLhG0BsR1Ebsq7tIkkBcjqBMLHD4ZJvNzW0+/TOvs3z/Y07Mz2Hj0Tk84P2bN3jujZdYW1ljd30H+bBBLJuiL1qImowUi2D0TBbPnkMwXAZDkzMn59FVnZlClaXFBbxmi3MT4xRLecar40RkBc/26Yc+f/gHfwj5JEbg8uKrn6LdPSKZiRGJ69R365iBzWRljgcP7tE1e6iaRmlinM1bD9FUjW5Koe/08NsjJNtFdqAcTdENh5ieT2ayTCSu0Rt0qYgy1WqO55+7yjs3H/L+J08IFIkriyep7+5Trx2ST56gXMiy+vAe84uLrKxvce3Vl2jW6yydniKqqPieTKPTYWwsj6bF2N7Zp5DPoSgSpXKRWqNDs3dEPJ3i1MkTJAWLZCp+HGrWMxQnJ4mmiyyJEXTOENM0dE1Fl0MCx+GlS5dw/JCwayKqIp+6/iLWcMjRQYvdnQZLJ+H7b32HqbEK/W6fRq3PVqPLUbtNsVogNA16B3VevPw8wkiguXqA1PVx3BaCqpLQojQdGz0dJ1rMsN6okdPSDLsmz7z0MkJap9FvsXjhJMr+ER++/Q6Ti/MMLYOzF87xwa0PKVfLiOkoH9+49SPl92N3zm89+MFXa0eHOJbFw1sfkS+PgQe9VoNLFy6ycOIET3c3UWIqjuJRyGSO8W+JGPFshlajhTT0+Mzzn+HC0vN4QYRmu4NpOWiKgqYKTOVLXHrtp1n5+ENUQSYSkXjw6CGtnsmn3/hpPvnoY/7mL36Fldo6MwszjC1O07u3iSzovPKVL/O99z7AUiXyS3M8OtxFL+RoNts4A4teu4OqKJiGTadvUDOHBHENUQLkEMO3cCyblCmwt1XjyPVpdTr4AuzeW6HX7hG4AV986QvIA4dG44C333sPvZQjOZmnvr2PFuiEckCylKRvdugOB9QPDkjGUoRBSGNvn1Q+z+bDFTxJOo4gVUvHlR71OslijiCi4Eghh502O7u71A7q2IZLKZthslrGs0ZcvnCBaCyGKuv8wdf/HaYuk52b4vHOOr14wFf+9i9Sax6xs7fL/PwJDGPAyBwyNlllrDqOnk3geR6YLmHfwjloI0oiqXgU+gOKio7mQrPZwZYlPEVBjcdIFtJowxEXE0mWcikKiThv33nIw61dtjcOSMbi/OIv/BJub49rz1xAFgRiepR8IYcaiTIwXTwgl82QzWbYO2hy/8EyU5NVSoUsuq4dUwRCn26nS6Pd57DZJRFLcv7UadKxGDh9dDnADQKODJ/M1CkELUmhPMbi6XOUZ09QmllASxe4t7LN2PgMoRdw8vQSw1aHYirJzVs3uXP7Nvtb+wSGgezbyHKU3c6QB2tb7NaPyJTzeDGZ4lgJWw4JJYFQExmEJolimlqnRW9okEgkufTi8+SnxxEiKnbg0XdGBKpEeqLM1MIM+/t7zM7M4vsuA3tEJp0gnUuRrpa48977HGxvkC4W2PjoEf/gl3/1Jzch/N1/+Pe+qiWinDx3mkg6xl79iMpUlVQ+w+TkFKEbcPvObQzXgMDHbA/JRtOMLAtN1yjqWWYKc7hdF1mQ8aIpgliGg2aDWCKGrum88unP48czLG9usb+1zv2H9xAliaZlc/7kGV545TVufvdtwmafv/Mf/2ecrE7zwZ99B1mJYMkCK+tHUMP/AAAf20lEQVRPKJ2YobA4QTabI4PK0sIiQ1WgFZdoxyUGCZmhLqJkU9SPDjAtg5Hg4Qo+ru0yXp4gHM8yf+UMnX6XAJ9UJkUkHsUTQ0atLsPAZKe+T8/zcMUQwxwS8SQ2bz8iEBwWryzx9ptvcuLkaYZHPXKOguCKpCWNo3aLZ648w2AwRI3r9I0hhXyRfKmI5Tr4oU80FsXqD/Bdi3g0iiRrFNJxUjEd1wvw/JCVpxvc21xnubWHn9T45OEDkCVq9SM+87OfY/XxYzZXN5ian+bxkyfEE0lC30eMquytrFGcqOAOR5gbNTRdJ15I0RsN8UQBQdfQx4uYMYm+aVAullFCiXh/xAvlHAv5BJIo0ugOuLuyy0AQUZJRJNOn0R7imQa5hIYig6pGIAw4qDVodgek00mk0Gc0HPFn33iL55+/QkQXee7a86yuLKNIIlFNZWqyyvzMBNV8lpweZbxYxBsNkcIRMU3FDwWqM/PUOkfc+ORdDob7vPf4fepuh7/8/lt8692/ZOtwh/XDbVZ2Nuj2uoxGJo49IgRGhoU1GjE0hghanHsrm9QMA0sIkeJRzly+gCeETF4+hRsJIKkgaiIhAb7gcfL58xjOCE2NEEsm2N/fIxBFpiZn8UY+Ow9WMUYGkXScVDKBYY1Yf7rC3Mw8ru8wMi2G/QExRWf7w0dMLJ1i8dQpfu7a535ycf7RN7/2Vc9xePHVlznoNInJEfaeboAksrmxw8AwmJqY4uBgH98PkGSVfLGEqqrUD49IeCpTmTK5dJJ4XMMxTfJRhUQEnm5v83hjh3t728yeWOR7b76J6w7RVHAJ+R//2f+G5zjERBgNBxQSWbaebnK0XuPhJw9JJJN88uQBf+sf/efMXj1Lp9NgNlnm4zt3cIoJ7u5uUDm7gKUJ2LqMnklj2TaW5RDRoth9g1w2R0xLkM/kMW2TtXsPSI7neeb6VWzXIZBCyrPj7Nf36bhDTs+f4uatj3DEAM+3uHrtGruru5y5fgUzElAdH+fwwTpe3eLU1DzJRIKP799DSkep1Rv4lkPguuytrRNVdCLZBG7osrAwj9np0jF6yKqMZ1q4psmphRnSuTx/8uabnDh7iWK6gtnu8wu/8B/y/3zjG2QrRbZWnzJVrrC8scrpKxe4/+FHpPIZls6c5dzzV1m6cJ7GoIk1MilkCqzdfUxWizOKCmjVLEMtQK/kEWIRjppHtFrHrQUXli7wxpVrPDM1RSWh4IxMLCfk8foBhh5jc3eLWDVPe/OQbqfP+QvP0G81yKdi2I6FZQ5ZXJin3urwzo0P8IB0QuPme3f42c+/huDbDJoNNp48wTZM5ubmSCRi/L/t3deTpXd54PHvyTnn0H06h+k4Mz05Bw3SKI5GQkJggZYCDNhlFuyttb1ay9h41yy7xkWtt4Bii7CAAUkgWUhCSBrNjGZ6Ys90zn26T5+ccw57oVvKVfKNuZjPX/DefC/e3/v+nseo19HX04PDaiWdSFDKZTHqlFiMGnLlCm/evM2F6RsEs1ES+RiFZp6VzSUErTr2ThuoBBQFFRKZBBKrkqntFSpqGSsBP6lSkZKwhcFuRapUUG/UObp3N08/9DBSQQutXkVcVCcYCLGyvIjeasDtdFGv1xDVGgSWN7E67NQEDZpiIQqFCqFQgslkRiKRUMzlsNqsxBNxJCIJ3ulF7O428pE0Q0M7iKYSpDIJBgYGOX7mDN7ANjqhiPNHHvzwcX73zZ+8UBVDtVxl6e48YqEUoaBFNBxgz9FDyFtCbr9/HbPGQDWWwakzoq63qEVSUKzSypRIBwNkIgGSoRARv49CNo1Uo6J9oJ90Lks6FeHu1DWi8SCtRgmr2Un/YD9urYahEyf4wfe/R0kASpWOVr3FrRs30ekNmAwGzpx/iGI+CyoJQoWIC5NXMXZ3EEnHKMlbiBRSPO0eTGo1kVAAo8OMqNVCKpPRM9BPwutHppTSELWoCGvITVrS6RR+fwBPRyczd25jMBoQq5RQb/Ds/Y/xy7feIlfMUE7nMLvslKslkIgRNetYFEr8V2c5MLYXT6eHde86kVoBjUkPIlBbNFi7HeidJuoCEImF5NMJlu/OUWs0UKlVtHV3oTbo2Vzb4L2rk0QaRXzFJJdmb3BzfoomNZbmZ/nok08RCkcJxUJU1WISvgDtOzqxdbeh0ipZWl5HrVWyvrJIoVxErpaztrhKZjOIQiFn4tx9SNVKajRR67WohHLioRgmqZT7enrYZ7djEkuopqJIJUKaTQG5UpmZZIqCVo4vGqZWq9Hn7qSQLXDj1jQiYZOhHg86nQ693gCNJmODvYjqFXra3RQzGTxOC61qgXwmSS2XRVRrIm0KKOUKpKJRItsBkvE4gVCQO3Pz6F1u+kdHyGXT1EVSrq1vEm/VsHY6yWZzpIIRHFYbxVIJmU5FpVImGgjSOdjH0voKQqmYRCqJSCGhImlREUGxWUNs0lA2SNnY3qbebKDXqmjUKuzpHWAzuI0vHKEWz6N32oEWhUqRUrWEWC4jsR4i4Y8glUswGAwk/SEcbjtyqw5buxORWEQ+FCeajFIVtXC43fhjAYQSEXqdgR0jo7z+i5cwGwwoyg3OnXzkw8f5nQs/f6FWKiIVS4n6YxhsZuKxKCKxmEggxOaGl1I8A9UWapkcm1hOn70NrVBCn8ONU2/AbTVhMpgolmrk83k2/D7Wt7ZYW1slFIzQEAtIKODM55/Fv7JKtVjn3LnHkQkqbK6u8LO33sS5Y5xUpkzvjnEuX7/K57/0x1y68DbrgS2aJjkV4QfzjGZuTqFSqNC7bNRjKeq1Gk1aBPx+JBIpA8NDKLRq8rUCUpWcRqOBXKXE2dlGsVpCrzcQXt5EWINEMkGlVUUql2PVm5C1hHTLDfi2/OTyOZx9HvpcPaTTGcqxDGMWF9HZVQbdHRhNVsr5DLlyjYpSRPdIH7Vmg47uLuLJBNVGnf7BAcw2K00BJCNRTGYznsF+qpUa8XgMhVxBs9Eilc2TTCW578EzLGwsoTIaiCbjpHNJlIPtyNUqpBoJTnc7MzfvUms26O3pxm2wcf3qFXLJJI+dP8/7ly9RKVfo7Onh3KeeQaSQEMulEEsk6BV6ZA0R7WIlR9o8WIUt7FoVCkGDVqVCIBgllinw87ev0rCZOfHUY0S82+ztHSG6FaKQzbNzrJfBrnY2onGkMjX7jh3mjX95jR5PB2atAnGjjlZnR+vZwQ//+UWmV7ZoNkVoNEbS+Tz5co16Q4inw4Oz3UW11WJ1c5tGo4bVoKFSqTG9vMV6PIZQK0Or1dLX34uzq51YOIJAKEIiUxAMBhFLRDSkQk6cPU0sGkCnV9EzMMj20hpNGrTtHuSJp5+kLBEwszDN2U9/gtm5BRbXvfjLOSq1EkcOHKRYrqKyGwn6/GS8Uey9nVSEddy9HoYnxtDqtRRzWTaXV/HOLRPzBXD0eJBIJZgsZuLpNDt2DFHKFJi8dAFnWwfLk3dAKMKs0SOr1Tl84iQTnaMfPs7vvfKDF6rZCjTFnH3gLN3DvWQLedx2F+GtMCaljpZUikKpRK1VodcZuXp7Bq93C4tKjVampIqUpsKIUKEhFNymUSmSTaYAAZlSlZJcQM9wL5pSnT6rnTajFQktgpEwkUACjdGBsAYHDhxiz9G9ZMo1rt+cZGV1gbSoRM0gY2B0mLnZu/g3vOhsBprUMbrsH7z7iAREoiHsHg9KrQqlTkW91aQlBJlZhdFjxuvzEtgOsuH1ojBosbpt5Ao53G3t7Nyzh7WVdRpCaJeoOL7vIA/tP8oOrY7l2UU25lYZ27eLyKYXq1BGW6eHp595kvXVDS5cuUEinyRXzGK2WlDLlKzML2N3ulCJ5bz/2/dQypQIxWJkKhV6k5FipYzWoEeuVGA3mWnlSkhqQtYWlkmE41CoYjNYKdZLRLMRKpUyEqGYjdkFHnn6oxiMRlav3UHTZkOSLiBpgEypxNblQi4Uo1aoKOcrXH7zPeLrcSxyN3/07J/RqXLSpRShlUCtJkBvtKPTmQglkkyubvDGzAwSl52mTIilx83U5E3S+SwViZBaNsOp8VF+9d47bBZKBAIhbGoFaqWSV9++RKEhRFBvMHH8BKauMc498xmMai13N7b5D3/xdbJCPUqNjny1SiyRIZZIoter2D0+yEBfJ6FQlExNTu/uIxSLBbLJOIODg6QyWcLBEKlIFLFSRr1SJba6jaW7E6VaTrlYwKQ3U69UyZdLPPDIowTDYaR6Fd7VVSrNJga1jvXgOhi1HD58HGN3O3mRgGw8xa6BEVbmlzE5HFTELSqFInKFnM6OTpLJJIVqGa1Oj1quApGIEk1MZhNqg471+WWSsQSSkgi5Qs2ZM6dZnF/A7fEgawpI+oJkBHUSrRLnJ858+Dj/8m//+oUHHn+Yw/efJJqIMjN1l+7+fjRiBUqZnAp1BvoHUNRF9JrceOeXsakUPHX6JFKhmHi2gtJgo1quMX3jOtlYhHw+h0gsQaWx8OgTz1D0R2gkc5SjOQr5KmpXJ3JrO1cuXGMrHCRdLqFoVkiFN1ibusGdm5PkikkcNgvVShWZTEwwGsTWZsHe2Ya0IcTldH2wiiAeQSQUoNOZUQlkZOIJork0crWCZrNB2BcjG81TzVcY7Bng8cefxr/mpVKvIdUqaLc62ZhbwuZpg0yZH/7f7/O5jz7D0swt/KFtEoIarvEerr71DseGRmlWaxw9cZJUuki5VeXVd96mbbSXvoEBwn4/KpeZvqEBUuEokWgEpV6LWCIh7g8TC4SJRWPkczmq9RrJbJpgNEqrXkfUEiCoNxDVm0TCQezDnQSKcTbnV7F62vH0ecgVC2QySURyKZlynmgywdjEKCJ/ktT8JrVkAYXJRHQzyMbyGuUaGIVmui09aGo18r4VxK0GuVwZ99BObIO7eO3yNb796itg0qOxmalIWoQjAQr5DKVcjo31Vdr6OujUaSnmC6R00g8O6WJ+IpUcEmB0bIT/9Z3vo9Zp0EkkCKplnCPjOLr6EIvF/OrXb9E5NE65KaRYb1CsN4lly/hCadY2w8ysbrMZTqLQmRAIBUS2N3DpDIhyJbzeNVKVPD1DAxzfe4iQ109V3OD4I2d450cvolDryWdyiIRC7C4XK5tr9HV14Y+G2VpYQSAR42p3oXZYqFbLWPrczC3OIUyXcQ72I1LKMKm1mBw2UtEwtXKFob3jBLeD5NN5YqEobWYnMpkKbzCA3WElnoxy8MBB0tkMAkELvcuMu8OFQNpEkM2RWN3mzvQMO/dNIG4JUerVPLr79IeP87W1Ky+EkyFe/elLlIslsokUdpeDSDzE7bvTaFVqchshnDIjhUQKLU2O7xyl2RIRSNeoCZVEQyGWZmdJxiKEE0l2H7mPilDF0Pg+fJte4pEQ2UoJmdPEzOIc7o4OJt+/wvjwCF/6yleIb2+TEZa4uHCHycUZFBYdSzML7HvsIyz6VvHsHKQpFNDe1k08kEAgElHI55iZnUZvNtLT308oHgGxgLKgilQtp5ArYLfZodakki3hsNtJZRLcnZnC7XR/MM9G3KLeajG2b4Lg1haFagmxQsn1lQVefOcCcQnUTSoW3rvLs2efoBiNY7KbUegs5PJ52rt7eO/2JHKdHIVOg0KrpSUWsbq8TKtUY+32HK1WE3NvG0aTCUeHB3dHB3Gvj2OnjlNr1NlYWmJw9wjlRplgMMy+jxwn2szij0VQCSW0DfWjUSm4euEiSpUGkVQG1QZmhZZMOEYin0drt7EVCLA+78W/sk2+VKJzdAeiXAmnwczxfftxmTWoFAKuz87w0sXLLKciXN2Y4VdvvEZRUOPgw6cRVhtE/EEqjSa5Yp6+/WNUSkUUKhUDZhOJVJbBiQlqUgFDozsIJuMITVreu36FXk871UIFi9tNwutFmvKj19vZfeoUgyND/MvPf4Y3GMVscyLUGtG42tA522gbGMDd52EttUVBWmVxexWFTY/EoaMkqGI0GWnTm2lUqoxY3SxML5BuVJBIJLT3eNi6u4Sj3YXBbSVVyGIxmWjv7SSwHUCuUmK2mblz8SqVZpP+HYOEtgMc2XuIeDZHOpkivL3N6vQi+/fto8PVzs2L7yNpCtEajKi0GiwuO7Fk/IM1FQYN+XiadDaDyaonXyxzYOIg115/g36LidJmkBu351G5rUi1ajR2M4Vimlg5xyePPv5vOBB65ccvrC2sYrFZ6e3tY/b21Af7ExsCPE4PRr2Rgi+CMl/m1MQYdoOOFX+UqtxESSBDIJbR1TeKxW7HG/Sz8+Sj3PfUH7Lz8BlkEhAp66itSuKNLBOPn8TUZmFry0s8GwetlBev/IaL1y6z+9FT5CUNPvpHz6HvduI6Pk65WqR9oAuTzYpap6WSK6FNV5FkC+gMOgZ3j5ArZCnm8ygkchQyBe09XcSWt5j99ftUi2WaAgFut4uB8QGEciG0WvgDIRZm5+nr6iEYi7Dl8zGwZ5z07DpPHL0PlUjMVijIn3z1eQS1Kp/Ze4KNpQ06O7vQGQ2MjO9ifOduPv2Fz5NrFhk/tp94Pkm6kKWY/mAFvaOzjROP3s9g7wCBlS0snjaalSq1eAaVWkMylcY/v44ACZV0gbAvir2jDc/efuw6A72jI1y7Msnhj5wklknSMTyAUi1FiYh0MY9CrWTffYfYWF9ne32TQCjM+S9+kngshbwq5PzxEzx9/BQP3/8g7T29yNQKwpk4SXEdUZeBWCvN2uIiKq0ax0Abc4t30MgVaIVyBifGkWnU3Lh0FafRiafVxCxQ8MrNW9SUEoqlIm6ngzXvBvlcjv59E/TsG2Ln2ADZSIrVlgCTScfG1QvYpQokIhU9/QN86lMf59SJw9hMapbn75KKhnj3ndfJlFKIxSISkQQOTxt6o558KEGt1IBqE4dMQa/JhlosY3JmFpXVwNbsPJSh1KjTO9TP1K3b2FRGBDV448WX2bPvIL0DvWyEtjh1/BTFZpP16SVSwQStchOFxURHXzeLAS8NiYDEohddVYRKJCMSSbC+usnCnTnmL04SXFxjYW4BuUyGVCXjiXPnuPby2xSCYdokUvKpFLFMiV+9+Dq7Tx8jS43o3BrZSIKtTIg9e/fy8MiJDx/nK5fffGHt4hT7Th5ELpZSaYnwrq6zZ/9+Vt+f5IH2PvY7XXjcDmL1Fi2ViVZTgKhZJ53KY3B1kMjnef21VxGIJMSicRr5BLcuvcb1a++SykaIVdNsLi9hsZmRtBuQugwoDGq2E0E6d/ejHHDisFnYXF9HpFYwPDLM9vo6rk43oXCQQrZIeTXIuNqKVSzFrdNiEyuYW10Fgwafb5uQL4TZaqNSqrC0tcrQwQmyuTxSlZqxnbvIZdK0aGL3uOno6kQskJBO5Tl25AgLk7doLob5g1NnsUgVHDt4iPcuX2UqtEJHU0wtniWZSDF+8DDlchW5XIo/FOIXb/+aviM7OXr/KX7769cZHx5l99AoTWGLQDSMb2sLXzCApClErlZQFTSoVKsU0nmq4QwOgYY2uZ6Y14fTbCQWjlCUwczyAk2ZiE9+/BOEklHy6QzxcAShWEwkFGJz08uOvTtRqbQszSygV6p45InHefkXL1NdC/P9b3wTJQIS8QyRVBqFSk0wFeLP/9vzbMUCuMa70NtMNBpCEv4o7q4ORgaGaGzGCWxuE0rE6Orvo1AsIc8U2OXu5E4mRi5f4T998cuMtnfzrf/+D1jaXcTKORZvz1CIZqio1ZQksLy2zJu3b7I0v83mZojRHUMYtB98QqlkEhiKCcb0Urr0UgbcdjwWPR6LmUatQtDnJby+hqIhIOWL4F/ZoN4QQVnI1RtTQAutyYDD086j5x5D1gKpUom0LiTlj2KwWJEIpER92ywENjh55DjhrSChlQ0MZjMtgYBULEYxmcS3ukqnwoCmDM1yg2g4glmrZWx4GEdXOy2JAK3DhLWvE0GtjqfdRW+3BxVCxCIxpWaZ95dWWArGqNarHDz/AEurXuRCEZ2dXTgHunEP9SIUCnj03xLnn3zjqy+Y3XZ27dnD3bk5FIUqHqScNrdxwtOFWaEhWmig6hih0pTwxquvEo5FmV1exO3pwGi1Etr2smu0n/7+DnQ6KWaTCrVChM+/SkXVon1HD0KbjkAxSyGbpxBMcPbZJ6lJxUTTaXaNj1EuV9FodLz79jskkkkW1xeQiMRMvn6JXe5eotk4P3v5ZRZX1lDrNKjUOq4tzbN0ewGdWk+xWabZbFCpFvDPryFMV5ErlGyvb5HN57gzN006maJebbLwzg0ESjkWm4VbV6/jkBv58nNfYGJkDzKdnmeefQ7sOo5+5CiudIlKJsvrb77LG5cn+fo3/xGZUsGVWze4du0OFosBpVNPqwFTM3NEMimq1FGrVaxNzhK4vUqtXmfnySPYnXaKuRxmtOgqUirlCndu3aZnsINjh/ew5N/C4rAikIgILvrx+TaxmS1c+N6LlBUCxCIJeouJY6dPI46X2VjeoCCs0dbRRTwUJ/b+Ar986WVykRjerU1+c/0KtT4dN31T+PIR3INdRLMxqoUKggrk0hk6XE4SWwHm3plkpL2X82cfQtRscn3yOvpqkYdH9/GjC+9SLME/f+fHjA+P8pU//Y/UGjXMXW2E17YY2jWGRq+lb+84N+amadLCpNdjdbvZOb6f//q1F7gxe4vhng60giLJjXmarRIGqZQup5Uht4s+q5UBm4Wdbie72j0Y5UpUMgXVeoOh4V2ce/o5wqEQpWKJNlcbl99/l5y6QcMo587VG3z+S18iJSyid2gp50vcmZqlHs6izgvJ5tMfXI0sNqlF82SKWUrlIsqqFEuHi1szU2QrJaLhEFqljFo8jtaoxTXWR65WAmDnrp3E51fAFybi9aHU2fFFQmisGhrCJufPfQxaLeKFBDKpFF8izM7hIWLhGIurq3zuvqc/fJyJYvSFA6MTZDeDHDTaGdE7MLakUJdxY87HerzG4JmHUFuc5JNJCvUSMruRseOHKDdqNGolbt+6SjgaZHnTS6ZWYXp5nlvLd2hYFahsFs4cOoFKr0esVSM3GTn28P3cvTzJjds3OHH8KMVKBY1Bz+T1SRwmE3dev8jHP/VJXvvlr3nyqWcoV0rkwmFy20HC4RgbyRDJShFTRxvjRyewdTjo6OtBY9IR2Qogl0jp3D+C3mHE5XHgcNsIzi8z0j5EsSUgJ6igkMpIXF9huHeYv/7Cn+Ix2yhXq9z/9Dm69o0QiUbZObwTbaGCJJPksWee5Y33LvPpz34OtdHKn/3Ff8bT50FslvP2y6+CUsrI2AjxUITt5Q0qiQKZzRDCpoB8KUsmGqND4URZVrF4a4ZgIoRcoyRbLlKSCxnp7KUmF2Lv7cLkceFw2ijWy6x7V/k/L/2YoY5OrC4n2VwauVTO1PUb3Lh4kUFPF1d+9jpDu/byk29+F61Sx/976UX+57f/CYlawkBbOwsbG8SiSbaX16lm8pg8dkQyCX1DgwQTUdQNMd32DpKBCK+9/iaVag2HXM6AwcVMOMzZ02d54NQZnv+b5/nuT3/AJ778OS5NTxLb9jMwsYtSIkM6myY5vYazo43tzS3UejV79x5AVGrSEDdItYpsRLf52jf+nlS5QHdHNxqljmA8zY3pJXzBBKFomi1fmK1gBF8my7tTU7jG+2m4VVy/dYNKJk+1UWVqeY69DxxHplWxvulloGeQV197FbFKgsPpQNCEbCHL6LEJ5DY924vr7Dx5gLubc6CR0D7Yg9lh4e61axiHuxDIpSTicRpyCdvxCPF0FrlRj1SjJJVNcmZgjI/uPsIDuw8xceggs8kIplE360tLxJNpyvUq8XCc5bUl9p8+jNliZXFrDavDwcyVm3T19PDUgd/9h9C/OvF9+IljrcjGFmq7lVKhxJ8/+8f89ucvUihW+dhnv8jgjmG+/a1voVfKuTw9yZ4zB0jH4mwur1FK59AajMzPL2B3OBnfs5e1+QWqpRIaiw6hVcPI3gOEVzdpdehQiaW89g//xGe//l/wRYM4jSa83i2iwTClVJ7F6VkUAgnnP/sHDA0M0NXZxd89/1VOOHspLM5x+sAEhUqOLlc7AomYlrGNSlsnr17+DZpuMxvxCOFIlO2pRVxmBx07+kil4+wYGSddKSMQNfH7tzFI5Rzt2oUyI8RmNeMwGfnN1Yv86sq7KOxaOvp7yHvjTE1P8WBbN4M9nSz7IvzjT3/G83/5PP/jb77GI488gt6m58D5h1iMb5DMxtFq9KxML7G14MWq1nHnyjV6RgcZ3DvG1DtX2TG6m6MPnubqzUu8/dIrHHIPEgtGOXn/fdy4fpOlapCu8SGuTt0CuZBqKEPfxAjP/eFz3JmappGv0Nbh4d0332JjeQ2xXsWRvRPEohlMUi1f+fhnOHTkCIeOHObwjhE6rHokIgF+rYiNYoZMKEE2mwZpi9GhIS688Q65QpUus5uIL8iZPQexyVTMzt7Bt7XG2aeeJJPJ8e6V64QKGY48dj+BTJRtrxerQU/P4ADXLk4iEokQq5Qk13381Tf+jv/9V3+PtNFEI1Rz7MgD3Lg+yXo0wFPPfIxrV6/gjWzRbEIun6UQS6LQ6JHKZNjsVlqVEjq5nHgqiaLdTNtAF8VaiRHPOO1SK3cnL+GvxHDs6iWeStPb1c3q0ioPnX2QV954BYVBRXdvH+JyC6VSzk9++BO6RvpJV9J0d3eza3icH33/p0grLRw6He37d3Pl3bfo3DGIP+AnFYlhttlRNkU0aeAe7UFSbnL9rQuIZVK6xgYoFMr07x3A3dZBIplhc2ODpDdOb1cnl+Zvo21C/+5xbt26xeadGQYOHeDiX/3od964/lfjvOeee/79CP+9H+Cee+753e7Fec89v6fuxXnPPb+n7sV5zz2/p+7Fec89v6fuxXnPPb+n/j9oXZGO7PWnDwAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "NtaVNwxWP4-A",
        "outputId": "5883894d-287e-4a7a-ba55-edfa96e49c29"
      },
      "source": [
        "train_data.class_names"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['chicken_curry',\n",
              " 'chicken_wings',\n",
              " 'donuts',\n",
              " 'fried_rice',\n",
              " 'grilled_salmon',\n",
              " 'hamburger',\n",
              " 'ice_cream',\n",
              " 'not_food',\n",
              " 'pizza',\n",
              " 'ramen',\n",
              " 'steak',\n",
              " 'sushi']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 18
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BUW1AprHQgHt"
      },
      "source": [
        "## Model 4 (experiment)\n",
        "\n",
        "Going to use same model as #3 but try mixed precision training and see what happens.\n",
        "\n",
        "See: https://www.tensorflow.org/guide/mixed_precision"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "pa2mlpw3Qruv",
        "outputId": "aa4fefe3-dbd9-4a27-f8af-24a4b487e23f"
      },
      "source": [
        "!nvidia-smi -L"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "GPU 0: Tesla T4 (UUID: GPU-6c111d17-da04-3fff-27c9-ca59a1321b70)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "yoNj3STVQhfv",
        "outputId": "3771ae3c-4acd-4ecd-e6b3-dde1e3f842bf"
      },
      "source": [
        "import tensorflow as tf\n",
        "from tensorflow.keras import mixed_precision\n",
        "\n",
        "print(tf.__version__)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2.4.1\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "DpzZGuAiQ42P",
        "outputId": "38662ab0-ad73-4327-a3e0-3d57cd7357b4"
      },
      "source": [
        "# Set mixed precision to float16 & float32\n",
        "mixed_precision.set_global_policy(\"mixed_float16\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Mixed precision compatibility check (mixed_float16): OK\n",
            "Your GPU will likely run quickly with dtype policy mixed_float16 as it has compute capability of at least 7.0. Your GPU: Tesla T4, compute capability 7.5\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Kb9ZuxlqR4cr",
        "outputId": "a6bdb7bf-80af-4300-ed5c-564e39054906"
      },
      "source": [
        "# Get data\n",
        "import zipfile\n",
        "\n",
        "# Download data\n",
        "!wget https://storage.googleapis.com/ml_deploymet_lec_bucket_0/11_food_classes_plus_not_food_class_all_data.zip # note: this may be broken if using well beyond Feb 10 2021\n",
        "unzip_data(\"11_food_classes_plus_not_food_class_all_data.zip\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2021-02-03 23:44:36--  https://storage.googleapis.com/ml_deploymet_lec_bucket_0/11_food_classes_plus_not_food_class_all_data.zip\n",
            "Resolving storage.googleapis.com (storage.googleapis.com)... 74.125.195.128, 74.125.142.128, 74.125.20.128, ...\n",
            "Connecting to storage.googleapis.com (storage.googleapis.com)|74.125.195.128|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 1053780900 (1005M) [application/zip]\n",
            "Saving to: ‘11_food_classes_plus_not_food_class_all_data.zip.1’\n",
            "\n",
            "11_food_classes_plu 100%[===================>]   1005M   202MB/s    in 4.9s    \n",
            "\n",
            "2021-02-03 23:44:41 (204 MB/s) - ‘11_food_classes_plus_not_food_class_all_data.zip.1’ saved [1053780900/1053780900]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jILF46l8R4c0",
        "outputId": "67ee54bf-f147-435a-bd4d-4f9526223ad7"
      },
      "source": [
        "# Create data loaders\n",
        "train_data, test_data = create_data_loaders(train_dir=\"11_food_classes_plus_not_food_class_all_data/train/\",\n",
        "                                            test_dir=\"11_food_classes_plus_not_food_class_all_data/test/\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 14343 files belonging to 12 classes.\n",
            "Found 4273 files belonging to 12 classes.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "NmHVq--pUcAR",
        "outputId": "8f87c64b-a67a-4d29-8bf3-3d20b9af3330"
      },
      "source": [
        "train_data.take(1)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<TakeDataset shapes: ((None, 224, 224, 3), (None, 12)), types: (tf.float32, tf.float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 24
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "I3dklhdTR4c1",
        "outputId": "16074a22-0045-4d3e-906e-a121af4cd079"
      },
      "source": [
        "# Create model (with mixed precision training)\n",
        "def create_model(input_shape=INPUT_SHAPE, base_model=BASE_MODEL, num_classes=10):\n",
        "  # Fine-tune?\n",
        "  base_model.trainable = False\n",
        "\n",
        "  # Create input layer\n",
        "  inputs = layers.Input(shape=input_shape, name=\"input_layer\")\n",
        "\n",
        "  # Add in data augmentation Sequential model as a layer\n",
        "  x = data_augmentation(inputs)\n",
        "\n",
        "  # Give base_model inputs (after augmentation) and don't train it\n",
        "  x = base_model(x, training=False)\n",
        "\n",
        "  # Pool output features of base model\n",
        "  x = layers.GlobalAveragePooling2D(name=\"global_average_pooling_layer\")(x)\n",
        "\n",
        "  # Put a dense layer on as the output\n",
        "  x = layers.Dense(num_classes, name=\"dense_layer\")(x)\n",
        "\n",
        "  print(x.dtype)\n",
        "\n",
        "  ## NOTE: Mixed precision requires a separate output activation\n",
        "  outputs = layers.Activation(activation=\"softmax\", dtype=tf.float32, name=\"predictions\")(x)\n",
        "\n",
        "  print(outputs.dtype)\n",
        "  \n",
        "  # Make a model with inputs and outputs\n",
        "  model = keras.Model(inputs, outputs)\n",
        "\n",
        "  # Compile the model\n",
        "  model.compile(loss=\"categorical_crossentropy\",\n",
        "                optimizer=tf.keras.optimizers.Adam(),\n",
        "                metrics=[\"accuracy\"])\n",
        "  \n",
        "  return model\n",
        "\n",
        "model_4 = create_model(num_classes=len(train_data.class_names))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<dtype: 'float16'>\n",
            "<dtype: 'float32'>\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4bbPH5TDUqr6",
        "outputId": "502a0efd-3539-4151-e2ca-b4fd61cf4fd9"
      },
      "source": [
        "model_4.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"model_1\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_layer (InputLayer)     [(None, 224, 224, 3)]     0         \n",
            "_________________________________________________________________\n",
            "data_augmentation (Sequentia (None, None, None, 3)     0         \n",
            "_________________________________________________________________\n",
            "efficientnetb0 (Functional)  (None, None, None, 1280)  4049571   \n",
            "_________________________________________________________________\n",
            "global_average_pooling_layer (None, 1280)              0         \n",
            "_________________________________________________________________\n",
            "dense_layer (Dense)          (None, 12)                15372     \n",
            "_________________________________________________________________\n",
            "predictions (Activation)     (None, 12)                0         \n",
            "=================================================================\n",
            "Total params: 4,064,943\n",
            "Trainable params: 15,372\n",
            "Non-trainable params: 4,049,571\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "NkiOhbofUsYA",
        "outputId": "f6704048-03a7-4589-ba05-3e22605f00d7"
      },
      "source": [
        "print(model_4.layers[3].dtype)\n",
        "print(model_4.layers[3].dtype_policy)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "float32\n",
            "<Policy \"mixed_float16\">\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8x_1PEhgSzNL",
        "outputId": "eb000509-8041-46b1-def1-d8a8db4a928c"
      },
      "source": [
        "# Fit model\n",
        "history_4 = model_4.fit(train_data,\n",
        "                        epochs=5,\n",
        "                        validation_data=test_data,\n",
        "                        validation_steps=int(0.25*len(test_data)))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/5\n",
            "449/449 [==============================] - 70s 131ms/step - loss: 1.2719 - accuracy: 0.6387 - val_loss: 0.5272 - val_accuracy: 0.8665\n",
            "Epoch 2/5\n",
            "449/449 [==============================] - 57s 126ms/step - loss: 0.6092 - accuracy: 0.8274 - val_loss: 0.4533 - val_accuracy: 0.8788\n",
            "Epoch 3/5\n",
            "449/449 [==============================] - 57s 126ms/step - loss: 0.5427 - accuracy: 0.8433 - val_loss: 0.4074 - val_accuracy: 0.8883\n",
            "Epoch 4/5\n",
            "449/449 [==============================] - 57s 127ms/step - loss: 0.4959 - accuracy: 0.8554 - val_loss: 0.4069 - val_accuracy: 0.8949\n",
            "Epoch 5/5\n",
            "449/449 [==============================] - 57s 126ms/step - loss: 0.4815 - accuracy: 0.8568 - val_loss: 0.4127 - val_accuracy: 0.8807\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_51dKGFZVeot"
      },
      "source": [
        "## Model 3 (not mixed precision)\n",
        "# Epoch 1/5\n",
        "# 449/449 [==============================] - 117s 235ms/step - loss: 1.2646 - accuracy: 0.6447 - val_loss: 0.5087 - val_accuracy: 0.8778\n",
        "# Epoch 2/5\n",
        "# 449/449 [==============================] - 88s 195ms/step - loss: 0.6192 - accuracy: 0.8225 - val_loss: 0.4408 - val_accuracy: 0.8873\n",
        "# Epoch 3/5\n",
        "# 449/449 [==============================] - 84s 186ms/step - loss: 0.5418 - accuracy: 0.8443 - val_loss: 0.4145 - val_accuracy: 0.8864\n",
        "# Epoch 4/5\n",
        "# 449/449 [==============================] - 78s 173ms/step - loss: 0.5053 - accuracy: 0.8512 - val_loss: 0.4024 - val_accuracy: 0.8864\n",
        "# Epoch 5/5\n",
        "# 449/449 [==============================] - 77s 170ms/step - loss: 0.4768 - accuracy: 0.8578 - val_loss: 0.3706 - val_accuracy: 0.8873"
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}