SUBESCO_CNN_4857.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "SUBESCO_CNN.ipynb",
      "provenance": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/Faisal-NSU/CSE465/blob/main/SUBESCO_CNN_4857.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Unzip Dataset (Subesco)"
      ],
      "metadata": {
        "id": "UAetRfWHTQSD"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Download And Extract"
      ],
      "metadata": {
        "id": "BY5QPBv3h7Uq"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#Subesco.zip\n",
        "!gdown --id 1k-afEJdwz5Tf4-bsuKOJzP7xn-KQTlkM"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "PY2QjzdT6Zpj",
        "outputId": "16bd2dea-b177-4690-e5cf-3c5ab148ded9"
      },
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading...\n",
            "From: https://drive.google.com/uc?id=1k-afEJdwz5Tf4-bsuKOJzP7xn-KQTlkM\n",
            "To: /content/SUBESCO.zip\n",
            "100% 1.65G/1.65G [00:13<00:00, 118MB/s]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "id": "kcAXMisz2plC"
      },
      "outputs": [],
      "source": [
        "import zipfile\n",
        "dataset_directory = '/content/SUBESCO.zip'\n",
        "zip_ref = zipfile.ZipFile(dataset_directory, 'r')\n",
        "zip_ref.extractall('/content')\n",
        "zip_ref.close()"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Custom Dataset Class"
      ],
      "metadata": {
        "id": "8m8F1n2QiNXD"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Create custom Dataset"
      ],
      "metadata": {
        "id": "i2kgkj5VWhgx"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import os\n",
        "import torch\n",
        "from torch.utils.data import Dataset\n",
        "import pandas as pd\n",
        "import torchaudio\n",
        "from torch.utils.data import DataLoader\n",
        "import torch.optim.lr_scheduler as lr_scheduler\n",
        "from torch.nn.functional import normalize\n",
        "\n",
        "\n",
        "class CustomDataset(Dataset):\n",
        "    def __init__(self, annotations_file, audio_dir, transformation, target_sample_rate, num_samples, device):\n",
        "        self.annotations = pd.read_csv(annotations_file)\n",
        "        self.audio_dir = audio_dir\n",
        "        self.device = device\n",
        "        self.transformation = transformation.to(self.device)\n",
        "        self.target_sample_rate = target_sample_rate\n",
        "        self.num_samples = num_samples\n",
        "\n",
        "    def __len__(self):\n",
        "        return len(self.annotations)\n",
        "\n",
        "    def __getitem__(self, index):\n",
        "        audio_sample_path = self._get_audio_sample_path(index)\n",
        "        #print(audio_sample_path)\n",
        "        label = self._get_audio_sample_label(index)\n",
        "        #print(label)\n",
        "        #signal, sr = torchaudio.load(audio_sample_path,frame_offset=int(self.num_samples*.3),num_frames = self.target_sample_rate)\n",
        "        signal, sr = torchaudio.load(audio_sample_path)\n",
        "        signal = signal.to(self.device)\n",
        "        signal = self._resample_if_necessary(signal, sr)\n",
        "        signal = self._mix_down_if_necessary(signal)\n",
        "        signal = self._cut_if_necessary(signal)\n",
        "        signal = self._right_pad_if_necessary(signal)\n",
        "        signal = self.transformation(signal)\n",
        "        # produce delta and delta-delta and add to mfcc\n",
        "        delta = torchaudio.functional.compute_deltas(signal)\n",
        "        delta2 = torchaudio.functional.compute_deltas(delta)\n",
        "        signal = torch.cat((signal, delta, delta2), 1) # concatenating deltas\n",
        "        \n",
        "        #signal = signal + delta + delta2\n",
        "        means = signal.mean(dim=1, keepdim=True)\n",
        "        stds = signal.std(dim=1, keepdim=True)\n",
        "        signal = (signal - means) / stds #NORMALIZED\n",
        "\n",
        "        return signal, label\n",
        "\n",
        "    def _cut_if_necessary(self, signal):\n",
        "        if signal.shape[1] > self.num_samples:\n",
        "            # print(signal.shape[1]) # print sample size\n",
        "            signal = signal[:, :self.num_samples]\n",
        "        return signal\n",
        "\n",
        "    def _right_pad_if_necessary(self, signal):\n",
        "        length_signal = signal.shape[1]\n",
        "        if length_signal < self.num_samples:\n",
        "            num_missing_samples = self.num_samples - length_signal\n",
        "            last_dim_padding = (0, num_missing_samples)\n",
        "            signal = torch.nn.functional.pad(signal, last_dim_padding)\n",
        "        return signal\n",
        "\n",
        "    def _resample_if_necessary(self, signal, sr):\n",
        "        if sr != self.target_sample_rate:\n",
        "            resampler = torchaudio.transforms.Resample(sr, self.target_sample_rate)\n",
        "            resampler = resampler.to(device)\n",
        "            signal = resampler(signal)\n",
        "        return signal\n",
        "\n",
        "    def _mix_down_if_necessary(self, signal):\n",
        "        if signal.shape[0] > 1:\n",
        "            signal = torch.mean(signal, dim=0, keepdim=True)\n",
        "        return signal\n",
        "\n",
        "    # for any audio index this function returns the audio path\n",
        "    def _get_audio_sample_path(self, index):\n",
        "        path = os.path.join(self.audio_dir, self.annotations.iloc[index,0])\n",
        "        return path\n",
        "    \n",
        "    # for specified audio index this will return the label from the data csv\n",
        "    def _get_audio_sample_label(self, index):\n",
        "        return self.annotations.iloc[index, 4]"
      ],
      "metadata": {
        "id": "UBSsEU43WdcD"
      },
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "Instantiate the dataset"
      ],
      "metadata": {
        "id": "LY45V_efdoLX"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "TRAIN_CSV = '/content/SUBESCO/train/train.csv'\n",
        "TEST_CSV = '/content/SUBESCO/test/test.csv'\n",
        "VALID_CSV = '/content/SUBESCO/valid/valid.csv'\n",
        "\n",
        "TRAIN_PATH = '/content/SUBESCO/train/'\n",
        "TEST_PATH = '/content/SUBESCO/test/'\n",
        "VALID_PATH = '/content/SUBESCO/valid/'\n",
        "\n",
        "SAMPLE_RATE = 48000\n",
        "NUM_SAMPLES = SAMPLE_RATE*4\n",
        "\n",
        "device = ''\n",
        "if torch.cuda.is_available():\n",
        "    device = \"cuda\"\n",
        "else:\n",
        "    device = \"cpu\"\n",
        "print(f\"Using device {device}\")\n",
        "\n",
        "#mfcc = torchaudio.transforms.MelSpectrogram(sample_rate=SAMPLE_RATE, n_mels = 70,n_fft = 600)\n",
        "mfcc = torchaudio.transforms.MFCC(sample_rate=SAMPLE_RATE, n_mfcc=44, melkwargs={\"n_fft\": 1536, \"hop_length\": 512, \"power\": 2})\n",
        "train_dataset = CustomDataset(TRAIN_CSV,\n",
        "                        TRAIN_PATH,\n",
        "                        mfcc,\n",
        "                        SAMPLE_RATE,\n",
        "                        NUM_SAMPLES,\n",
        "                        device)\n",
        "print(f\"There are {len(train_dataset)} samples in the train.\")\n",
        "\n",
        "val_dataset = CustomDataset(VALID_CSV,\n",
        "                        VALID_PATH,\n",
        "                        mfcc,\n",
        "                        SAMPLE_RATE,\n",
        "                        NUM_SAMPLES,\n",
        "                        device)\n",
        "\n",
        "print(f\"There are {len(val_dataset)} samples in the valid.\")\n",
        "\n",
        "test_dataset = CustomDataset(TEST_CSV,\n",
        "                        TEST_PATH,\n",
        "                        mfcc,\n",
        "                        SAMPLE_RATE,\n",
        "                        NUM_SAMPLES,\n",
        "                        device)\n",
        "print(f\"There are {len(test_dataset)} samples in the test.\")\n",
        "signal, label = val_dataset[0]\n",
        "print(signal.shape)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "irCA4qM4dneT",
        "outputId": "631fd863-9a3c-4dd3-fbb0-ab9a36871888"
      },
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Using device cuda\n",
            "There are 4900 samples in the train.\n",
            "There are 700 samples in the valid.\n",
            "There are 1400 samples in the test.\n",
            "torch.Size([1, 132, 376])\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import math\n",
        "#Calculate outputs:\n",
        "def CalculateConv2dOutput(W, F, P, S):\n",
        "  return math.floor((W - F + 2 * P) / S ) + 1\n",
        "\n",
        "def MaxPool2d(W, F, P, S):\n",
        "  return math.floor((W - F) / S ) + 1\n",
        "\n",
        "\n",
        "#layer 1\n",
        "a = CalculateConv2dOutput(132, 3, 2, 1)\n",
        "b = CalculateConv2dOutput(376, 3, 2, 1)\n",
        "\n",
        "a = MaxPool2d(a, 2, 0, 2)\n",
        "b = MaxPool2d(b, 2, 0, 2)\n",
        "#layer 2\n",
        "a = CalculateConv2dOutput(a, 3, 2, 1)\n",
        "b = CalculateConv2dOutput(b, 3, 2, 1)\n",
        "\n",
        "a = MaxPool2d(a, 2, 0, 2)\n",
        "b = MaxPool2d(b, 2, 0, 2)\n",
        "\n",
        "#layer 3\n",
        "a = CalculateConv2dOutput(a, 3, 2, 1)\n",
        "b = CalculateConv2dOutput(b, 3, 2, 1)\n",
        "\n",
        "a = MaxPool2d(a, 2, 0, 2)\n",
        "b = MaxPool2d(b, 2, 0, 2)\n",
        "# layer 4\n",
        "a = CalculateConv2dOutput(a, 3, 2, 1)\n",
        "b = CalculateConv2dOutput(b, 3, 2, 1)\n",
        "\n",
        "a = MaxPool2d(a, 2, 0, 2)\n",
        "b = MaxPool2d(b, 2, 0, 2)\n",
        "\n",
        "print(a)\n",
        "print(b)\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "lBteY7ubjjQt",
        "outputId": "4ec7eec5-db22-4935-91f2-751af258b88d"
      },
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "10\n",
            "25\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from torch import nn\n",
        "from torchsummary import summary\n",
        "\n",
        "class CNNNetwork(nn.Module):\n",
        "\n",
        "    def __init__(self):\n",
        "        super().__init__()\n",
        "        # 4 conv blocks / flatten / linear / softmax\n",
        "        self.conv1 = nn.Sequential(            # shape = 66,132\n",
        "            nn.Conv2d(\n",
        "                in_channels=1,\n",
        "                out_channels=16,\n",
        "                kernel_size=3,\n",
        "                stride=1,\n",
        "                padding=2\n",
        "            ),\n",
        "            nn.ReLU(),\n",
        "            nn.MaxPool2d(kernel_size=2)      # shape after maxpool = 33, 66\n",
        "        )\n",
        "        self.conv2 = nn.Sequential(\n",
        "            nn.Conv2d(\n",
        "                in_channels=16,\n",
        "                out_channels=32,\n",
        "                kernel_size=3,\n",
        "                stride=1,\n",
        "                padding=2\n",
        "            ),\n",
        "            nn.ReLU(),\n",
        "            nn.MaxPool2d(kernel_size=2)\n",
        "        )\n",
        "        self.conv3 = nn.Sequential(\n",
        "            nn.Conv2d(\n",
        "                in_channels=32,\n",
        "                out_channels=64,\n",
        "                kernel_size=3,\n",
        "                stride=1,\n",
        "                padding=2\n",
        "            ),\n",
        "            nn.ReLU(),\n",
        "            nn.MaxPool2d(kernel_size=2)\n",
        "        )\n",
        "        self.conv4 = nn.Sequential(\n",
        "            nn.Conv2d(\n",
        "                in_channels=64,\n",
        "                out_channels=128,\n",
        "                kernel_size=3,\n",
        "                stride=1,\n",
        "                padding=2\n",
        "            ),\n",
        "            nn.ReLU(),\n",
        "            nn.MaxPool2d(kernel_size=2)\n",
        "        )\n",
        "        self.flatten = nn.Flatten()\n",
        "        self.linear = nn.Linear(128 *10*25, 512)\n",
        "        self.linear2 = nn.Linear(512, 256)\n",
        "        self.linear3 = nn.Linear(256, 128)\n",
        "        self.linear4 = nn.Linear(128, 7) # 7 outputs # added linear layers\n",
        "        self.softmax = nn.Softmax(dim=1)\n",
        "\n",
        "    def forward(self, input_data):\n",
        "        x = self.conv1(input_data)\n",
        "        x = self.conv2(x)\n",
        "        x = self.conv3(x)\n",
        "        x = self.conv4(x)\n",
        "        x = self.flatten(x)\n",
        "        x = self.linear(x)\n",
        "        x = self.linear2(x)\n",
        "        x = self.linear3(x)\n",
        "        logits = self.linear4(x)\n",
        "        predictions = self.softmax(logits)\n",
        "        return predictions\n",
        "\n",
        "\n",
        "if __name__ == \"__main__\":\n",
        "    model = CNNNetwork().to(device)\n",
        "\n",
        "summary(model, (signal.shape))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BDFspOC79_JN",
        "outputId": "1d1943e1-ae78-483b-aa30-25bf2b90ad05"
      },
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "----------------------------------------------------------------\n",
            "        Layer (type)               Output Shape         Param #\n",
            "================================================================\n",
            "            Conv2d-1         [-1, 16, 134, 378]             160\n",
            "              ReLU-2         [-1, 16, 134, 378]               0\n",
            "         MaxPool2d-3          [-1, 16, 67, 189]               0\n",
            "            Conv2d-4          [-1, 32, 69, 191]           4,640\n",
            "              ReLU-5          [-1, 32, 69, 191]               0\n",
            "         MaxPool2d-6           [-1, 32, 34, 95]               0\n",
            "            Conv2d-7           [-1, 64, 36, 97]          18,496\n",
            "              ReLU-8           [-1, 64, 36, 97]               0\n",
            "         MaxPool2d-9           [-1, 64, 18, 48]               0\n",
            "           Conv2d-10          [-1, 128, 20, 50]          73,856\n",
            "             ReLU-11          [-1, 128, 20, 50]               0\n",
            "        MaxPool2d-12          [-1, 128, 10, 25]               0\n",
            "          Flatten-13                [-1, 32000]               0\n",
            "           Linear-14                  [-1, 512]      16,384,512\n",
            "           Linear-15                  [-1, 256]         131,328\n",
            "           Linear-16                  [-1, 128]          32,896\n",
            "           Linear-17                    [-1, 7]             903\n",
            "          Softmax-18                    [-1, 7]               0\n",
            "================================================================\n",
            "Total params: 16,646,791\n",
            "Trainable params: 16,646,791\n",
            "Non-trainable params: 0\n",
            "----------------------------------------------------------------\n",
            "Input size (MB): 0.19\n",
            "Forward/backward pass size (MB): 27.42\n",
            "Params size (MB): 63.50\n",
            "Estimated Total Size (MB): 91.11\n",
            "----------------------------------------------------------------\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Model Loading and Functions"
      ],
      "metadata": {
        "id": "V7ZEOkhKmF3E"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Model Parameter Setting"
      ],
      "metadata": {
        "id": "3TklzFYmttsi"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from torch import nn\n",
        "train_loss = []\n",
        "val_loss = []\n",
        "test_loss = []\n",
        "\n",
        "train_acc = []\n",
        "val_acc = []\n",
        "test_acc = []\n",
        "\n",
        "BATCH_SIZE = 32\n",
        "criterion = nn.CrossEntropyLoss()\n",
        "optimizer = torch.optim.Adam(model.parameters(), lr=10**-4, weight_decay = 10**-6)\n",
        "\n",
        "#optimizer = torch.optim.SGD(model.parameters(), lr=0.1, momentum=0.9)\n",
        "#step_lr_scheduler = lr_scheduler.ReduceLROnPlateau(optimizer, mode='max',factor=0.1,patience=2, verbos=True)\n",
        "#step_lr_scheduler = lr_scheduler.StepLR(optimizer, step_size=3, gamma=0.1)\n",
        "\n",
        "def create_data_loader(train_data, batch_size):\n",
        "    train_dataloader = DataLoader(train_data, batch_size=batch_size)\n",
        "    return train_dataloader\n",
        "\n",
        "train_dataloader = create_data_loader(train_dataset, BATCH_SIZE)\n",
        "test_dataloader = create_data_loader(test_dataset, BATCH_SIZE)\n",
        "val_dataloader = create_data_loader(val_dataset, BATCH_SIZE)"
      ],
      "metadata": {
        "id": "Vn4QrDSngyhX"
      },
      "execution_count": 20,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "def train_loop(dataloader, model, criterion, optimizer):\n",
        "    model.train()\n",
        "    running_loss = 0.0\n",
        "    running_corrects = 0\n",
        "    size = len(dataloader.dataset)  \n",
        "\n",
        "    for batch, (X,y) in enumerate(dataloader):\n",
        "        # Compute prediction and loss\n",
        "        inputs, labels = X.to(device), y.to(device)\n",
        "        \n",
        "        with torch.set_grad_enabled(True):\n",
        "            outputs = model(inputs)\n",
        "            _, preds = torch.max(outputs, 1)\n",
        "            loss = criterion(outputs, labels)\n",
        "\n",
        "            optimizer.zero_grad()\n",
        "            loss.backward()\n",
        "            optimizer.step()\n",
        "\n",
        "        running_loss += loss.item() * inputs.size(0)\n",
        "        running_corrects += torch.sum(preds == labels.data)\n",
        "\n",
        "        if batch % 80 == 0:\n",
        "            loss, current = loss.item(), batch * len(X)\n",
        "            print(f\"loss: {loss:>7f}  [{current:>5d}/{size:>5d}]\")\n",
        "\n",
        "    epoch_loss = running_loss / size\n",
        "    epoch_acc = running_corrects.double() / size\n",
        "\n",
        "    print('Train      Loss: {:.4f} Acc: {:.4f}'.format(\n",
        "                 epoch_loss, epoch_acc))\n",
        "    \n",
        "    return epoch_loss, epoch_acc"
      ],
      "metadata": {
        "id": "Jhk6pTKCSEQY"
      },
      "execution_count": 21,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "def test_loop(dataloader, model, criterion):\n",
        "    #model.eval()\n",
        "    running_loss = 0.0\n",
        "    running_corrects = 0\n",
        "    size = len(dataloader.dataset)  \n",
        "\n",
        "    for batch, (X,y) in enumerate(dataloader):\n",
        "        # Compute prediction and loss\n",
        "        inputs, labels = X.to(device), y.to(device)\n",
        "        \n",
        "        with torch.set_grad_enabled(False):\n",
        "            outputs = model(inputs)\n",
        "            _, preds = torch.max(outputs, 1)\n",
        "            loss = criterion(outputs, labels)\n",
        "\n",
        "        running_loss += loss.item() * inputs.size(0)\n",
        "        running_corrects += torch.sum(preds == labels.data)\n",
        "\n",
        "    epoch_loss = running_loss / size\n",
        "    epoch_acc = running_corrects.double() / size\n",
        "\n",
        "    print('Validation Loss: {:.4f} Acc: {:.4f}'.format(\n",
        "                 epoch_loss, epoch_acc))\n",
        "    \n",
        "    return epoch_loss,epoch_acc\n",
        "    "
      ],
      "metadata": {
        "id": "RngTm7rGSm6v"
      },
      "execution_count": 22,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Model Running"
      ],
      "metadata": {
        "id": "h3_nrlbJUS5i"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import math\n",
        "import time\n",
        "import copy\n",
        "\n",
        "best_model_wts = copy.deepcopy(model.state_dict())\n",
        "best_acc = 0.0\n",
        "\n",
        "EPOCHS = 100\n",
        "start = time.time()\n",
        "for t in range(EPOCHS):\n",
        "    print(f\"\\nEpoch {t+1}\\n-------------------------------\")\n",
        "    ls , ac = train_loop(train_dataloader, model, criterion, optimizer)\n",
        "    train_loss.append(ls)\n",
        "    train_acc.append(ac)\n",
        "\n",
        "    ls,ac = test_loop(val_dataloader, model, criterion)\n",
        "    val_loss.append(ls)\n",
        "    val_acc.append(ac)\n",
        "\n",
        "    # #testing \n",
        "    # ls,ac = test_loop(test_dataloader, model, criterion)\n",
        "    # test_loss.append(ls)\n",
        "    # test_acc.append(ac)\n",
        "    # #testing \n",
        "\n",
        "    if ac > best_acc:\n",
        "        best_acc = ac\n",
        "        best_model_wts = copy.deepcopy(model.state_dict())\n",
        "\n",
        "    #step_lr_scheduler.step()\n",
        "    \n",
        "final = (time.time() - start)/60\n",
        "model.load_state_dict(best_model_wts)\n",
        "print(f\"Done for all {EPOCHS} epochs in {math.ceil(final)} minutes\\n\")\n",
        "print(f\"Best Validation accuracy: {best_acc} \\n\")"
      ],
      "metadata": {
        "id": "rBPMtivT4pmH",
        "outputId": "57911458-cd73-410f-c7df-0860acc6078a",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            "Epoch 1\n",
            "-------------------------------\n",
            "loss: 1.946487  [    0/ 4900]\n",
            "loss: 1.875505  [ 2560/ 4900]\n",
            "Train      Loss: 1.8694 Acc: 0.2720\n",
            "Validation Loss: 1.9058 Acc: 0.2443\n",
            "\n",
            "Epoch 2\n",
            "-------------------------------\n",
            "loss: 1.814022  [    0/ 4900]\n",
            "loss: 2.075772  [ 2560/ 4900]\n",
            "Train      Loss: 1.8524 Acc: 0.3039\n",
            "Validation Loss: 1.8537 Acc: 0.2943\n",
            "\n",
            "Epoch 3\n",
            "-------------------------------\n",
            "loss: 1.704497  [    0/ 4900]\n",
            "loss: 2.017435  [ 2560/ 4900]\n",
            "Train      Loss: 1.8140 Acc: 0.3431\n",
            "Validation Loss: 1.8419 Acc: 0.3057\n",
            "\n",
            "Epoch 4\n",
            "-------------------------------\n",
            "loss: 1.825765  [    0/ 4900]\n",
            "loss: 1.907987  [ 2560/ 4900]\n",
            "Train      Loss: 1.8153 Acc: 0.3398\n",
            "Validation Loss: 1.9128 Acc: 0.2357\n",
            "\n",
            "Epoch 5\n",
            "-------------------------------\n",
            "loss: 2.036413  [    0/ 4900]\n",
            "loss: 1.866221  [ 2560/ 4900]\n",
            "Train      Loss: 1.8275 Acc: 0.3306\n",
            "Validation Loss: 1.8646 Acc: 0.2986\n",
            "\n",
            "Epoch 6\n",
            "-------------------------------\n",
            "loss: 1.861326  [    0/ 4900]\n",
            "loss: 1.902948  [ 2560/ 4900]\n",
            "Train      Loss: 1.8105 Acc: 0.3480\n",
            "Validation Loss: 1.9740 Acc: 0.1829\n",
            "\n",
            "Epoch 7\n",
            "-------------------------------\n",
            "loss: 2.160332  [    0/ 4900]\n",
            "loss: 1.923504  [ 2560/ 4900]\n",
            "Train      Loss: 1.8686 Acc: 0.2892\n",
            "Validation Loss: 1.9147 Acc: 0.2314\n",
            "\n",
            "Epoch 8\n",
            "-------------------------------\n",
            "loss: 1.990441  [    0/ 4900]\n",
            "loss: 1.959932  [ 2560/ 4900]\n",
            "Train      Loss: 1.7748 Acc: 0.3818\n",
            "Validation Loss: 1.9365 Acc: 0.1986\n",
            "\n",
            "Epoch 9\n",
            "-------------------------------\n",
            "loss: 2.157390  [    0/ 4900]\n",
            "loss: 1.864001  [ 2560/ 4900]\n",
            "Train      Loss: 1.7573 Acc: 0.3969\n",
            "Validation Loss: 1.9551 Acc: 0.2000\n",
            "\n",
            "Epoch 10\n",
            "-------------------------------\n",
            "loss: 2.148390  [    0/ 4900]\n",
            "loss: 1.928742  [ 2560/ 4900]\n",
            "Train      Loss: 1.7255 Acc: 0.4341\n",
            "Validation Loss: 1.9595 Acc: 0.1886\n",
            "\n",
            "Epoch 11\n",
            "-------------------------------\n",
            "loss: 2.135762  [    0/ 4900]\n",
            "loss: 1.997136  [ 2560/ 4900]\n",
            "Train      Loss: 1.7455 Acc: 0.4176\n",
            "Validation Loss: 1.8953 Acc: 0.2514\n",
            "\n",
            "Epoch 12\n",
            "-------------------------------\n",
            "loss: 2.028422  [    0/ 4900]\n",
            "loss: 1.970940  [ 2560/ 4900]\n",
            "Train      Loss: 1.6817 Acc: 0.4829\n",
            "Validation Loss: 1.9203 Acc: 0.2343\n",
            "\n",
            "Epoch 13\n",
            "-------------------------------\n",
            "loss: 2.065348  [    0/ 4900]\n",
            "loss: 1.975029  [ 2560/ 4900]\n",
            "Train      Loss: 1.6784 Acc: 0.4884\n",
            "Validation Loss: 1.9102 Acc: 0.2371\n",
            "\n",
            "Epoch 14\n",
            "-------------------------------\n",
            "loss: 2.031106  [    0/ 4900]\n",
            "loss: 1.978818  [ 2560/ 4900]\n",
            "Train      Loss: 1.6628 Acc: 0.5053\n",
            "Validation Loss: 1.9095 Acc: 0.2400\n",
            "\n",
            "Epoch 15\n",
            "-------------------------------\n",
            "loss: 2.002708  [    0/ 4900]\n",
            "loss: 1.891586  [ 2560/ 4900]\n",
            "Train      Loss: 1.6895 Acc: 0.4757\n",
            "Validation Loss: 1.8546 Acc: 0.2986\n",
            "\n",
            "Epoch 16\n",
            "-------------------------------\n",
            "loss: 1.884585  [    0/ 4900]\n",
            "loss: 1.926283  [ 2560/ 4900]\n",
            "Train      Loss: 1.6524 Acc: 0.5110\n",
            "Validation Loss: 1.8674 Acc: 0.2857\n",
            "\n",
            "Epoch 17\n",
            "-------------------------------\n",
            "loss: 1.883614  [    0/ 4900]\n",
            "loss: 1.900129  [ 2560/ 4900]\n",
            "Train      Loss: 1.6271 Acc: 0.5414\n",
            "Validation Loss: 1.8422 Acc: 0.3071\n",
            "\n",
            "Epoch 18\n",
            "-------------------------------\n",
            "loss: 1.830470  [    0/ 4900]\n",
            "loss: 1.826782  [ 2560/ 4900]\n",
            "Train      Loss: 1.5959 Acc: 0.5759\n",
            "Validation Loss: 1.8305 Acc: 0.3200\n",
            "\n",
            "Epoch 19\n",
            "-------------------------------\n",
            "loss: 1.856239  [    0/ 4900]\n",
            "loss: 1.810299  [ 2560/ 4900]\n",
            "Train      Loss: 1.5957 Acc: 0.5665\n",
            "Validation Loss: 1.8513 Acc: 0.3043\n",
            "\n",
            "Epoch 20\n",
            "-------------------------------\n",
            "loss: 1.832421  [    0/ 4900]\n",
            "loss: 1.835834  [ 2560/ 4900]\n",
            "Train      Loss: 1.5705 Acc: 0.5951\n",
            "Validation Loss: 1.8254 Acc: 0.3314\n",
            "\n",
            "Epoch 21\n",
            "-------------------------------\n",
            "loss: 1.763113  [    0/ 4900]\n",
            "loss: 1.821158  [ 2560/ 4900]\n",
            "Train      Loss: 1.5578 Acc: 0.6106\n",
            "Validation Loss: 1.7986 Acc: 0.3586\n",
            "\n",
            "Epoch 22\n",
            "-------------------------------\n",
            "loss: 1.706766  [    0/ 4900]\n",
            "loss: 1.763495  [ 2560/ 4900]\n",
            "Train      Loss: 1.5467 Acc: 0.6210\n",
            "Validation Loss: 1.7704 Acc: 0.3886\n",
            "\n",
            "Epoch 23\n",
            "-------------------------------\n",
            "loss: 1.751251  [    0/ 4900]\n",
            "loss: 1.741329  [ 2560/ 4900]\n",
            "Train      Loss: 1.5395 Acc: 0.6318\n",
            "Validation Loss: 1.7787 Acc: 0.3743\n",
            "\n",
            "Epoch 24\n",
            "-------------------------------\n",
            "loss: 1.759184  [    0/ 4900]\n",
            "loss: 1.720528  [ 2560/ 4900]\n",
            "Train      Loss: 1.5155 Acc: 0.6541\n",
            "Validation Loss: 1.7536 Acc: 0.4043\n",
            "\n",
            "Epoch 25\n",
            "-------------------------------\n",
            "loss: 1.711501  [    0/ 4900]\n",
            "loss: 1.688468  [ 2560/ 4900]\n",
            "Train      Loss: 1.4985 Acc: 0.6714\n",
            "Validation Loss: 1.7729 Acc: 0.3843\n",
            "\n",
            "Epoch 26\n",
            "-------------------------------\n",
            "loss: 1.700655  [    0/ 4900]\n",
            "loss: 1.678243  [ 2560/ 4900]\n",
            "Train      Loss: 1.4764 Acc: 0.6943\n",
            "Validation Loss: 1.7538 Acc: 0.4057\n",
            "\n",
            "Epoch 27\n",
            "-------------------------------\n",
            "loss: 1.635239  [    0/ 4900]\n",
            "loss: 1.670427  [ 2560/ 4900]\n",
            "Train      Loss: 1.4628 Acc: 0.7078\n",
            "Validation Loss: 1.7305 Acc: 0.4286\n",
            "\n",
            "Epoch 28\n",
            "-------------------------------\n",
            "loss: 1.601314  [    0/ 4900]\n",
            "loss: 1.680213  [ 2560/ 4900]\n",
            "Train      Loss: 1.4422 Acc: 0.7296\n",
            "Validation Loss: 1.7031 Acc: 0.4529\n",
            "\n",
            "Epoch 29\n",
            "-------------------------------\n",
            "loss: 1.548665  [    0/ 4900]\n",
            "loss: 1.708929  [ 2560/ 4900]\n",
            "Train      Loss: 1.4286 Acc: 0.7412\n",
            "Validation Loss: 1.6954 Acc: 0.4629\n",
            "\n",
            "Epoch 30\n",
            "-------------------------------\n",
            "loss: 1.544101  [    0/ 4900]\n",
            "loss: 1.638580  [ 2560/ 4900]\n",
            "Train      Loss: 1.4182 Acc: 0.7531\n",
            "Validation Loss: 1.6798 Acc: 0.4814\n",
            "\n",
            "Epoch 31\n",
            "-------------------------------\n",
            "loss: 1.542454  [    0/ 4900]\n",
            "loss: 1.641332  [ 2560/ 4900]\n",
            "Train      Loss: 1.4040 Acc: 0.7655\n",
            "Validation Loss: 1.6974 Acc: 0.4571\n",
            "\n",
            "Epoch 32\n",
            "-------------------------------\n",
            "loss: 1.541229  [    0/ 4900]\n",
            "loss: 1.621660  [ 2560/ 4900]\n",
            "Train      Loss: 1.4026 Acc: 0.7655\n",
            "Validation Loss: 1.6876 Acc: 0.4643\n",
            "\n",
            "Epoch 33\n",
            "-------------------------------\n",
            "loss: 1.538710  [    0/ 4900]\n",
            "loss: 1.630558  [ 2560/ 4900]\n",
            "Train      Loss: 1.3946 Acc: 0.7743\n",
            "Validation Loss: 1.6729 Acc: 0.4900\n",
            "\n",
            "Epoch 34\n",
            "-------------------------------\n",
            "loss: 1.523898  [    0/ 4900]\n",
            "loss: 1.667829  [ 2560/ 4900]\n",
            "Train      Loss: 1.3920 Acc: 0.7794\n",
            "Validation Loss: 1.6993 Acc: 0.4586\n",
            "\n",
            "Epoch 35\n",
            "-------------------------------\n",
            "loss: 1.507452  [    0/ 4900]\n",
            "loss: 1.653056  [ 2560/ 4900]\n",
            "Train      Loss: 1.3695 Acc: 0.8002\n",
            "Validation Loss: 1.7198 Acc: 0.4371\n",
            "\n",
            "Epoch 36\n",
            "-------------------------------\n",
            "loss: 1.539641  [    0/ 4900]\n",
            "loss: 1.708063  [ 2560/ 4900]\n",
            "Train      Loss: 1.3607 Acc: 0.8088\n",
            "Validation Loss: 1.7083 Acc: 0.4471\n",
            "\n",
            "Epoch 37\n",
            "-------------------------------\n",
            "loss: 1.529095  [    0/ 4900]\n",
            "loss: 1.651079  [ 2560/ 4900]\n",
            "Train      Loss: 1.3523 Acc: 0.8188\n",
            "Validation Loss: 1.6988 Acc: 0.4629\n",
            "\n",
            "Epoch 38\n",
            "-------------------------------\n",
            "loss: 1.515919  [    0/ 4900]\n",
            "loss: 1.680529  [ 2560/ 4900]\n",
            "Train      Loss: 1.3409 Acc: 0.8298\n",
            "Validation Loss: 1.6611 Acc: 0.5057\n",
            "\n",
            "Epoch 39\n",
            "-------------------------------\n",
            "loss: 1.533838  [    0/ 4900]\n",
            "loss: 1.634629  [ 2560/ 4900]\n",
            "Train      Loss: 1.3322 Acc: 0.8380\n",
            "Validation Loss: 1.6781 Acc: 0.4786\n",
            "\n",
            "Epoch 40\n",
            "-------------------------------\n",
            "loss: 1.509331  [    0/ 4900]\n",
            "loss: 1.613034  [ 2560/ 4900]\n",
            "Train      Loss: 1.3213 Acc: 0.8486\n",
            "Validation Loss: 1.6762 Acc: 0.4814\n",
            "\n",
            "Epoch 41\n",
            "-------------------------------\n",
            "loss: 1.509740  [    0/ 4900]\n",
            "loss: 1.601875  [ 2560/ 4900]\n",
            "Train      Loss: 1.3189 Acc: 0.8518\n",
            "Validation Loss: 1.6803 Acc: 0.4743\n",
            "\n",
            "Epoch 42\n",
            "-------------------------------\n",
            "loss: 1.516034  [    0/ 4900]\n",
            "loss: 1.655250  [ 2560/ 4900]\n",
            "Train      Loss: 1.3094 Acc: 0.8590\n",
            "Validation Loss: 1.6904 Acc: 0.4714\n",
            "\n",
            "Epoch 43\n",
            "-------------------------------\n",
            "loss: 1.533594  [    0/ 4900]\n",
            "loss: 1.574589  [ 2560/ 4900]\n",
            "Train      Loss: 1.3093 Acc: 0.8594\n",
            "Validation Loss: 1.6614 Acc: 0.4971\n",
            "\n",
            "Epoch 44\n",
            "-------------------------------\n",
            "loss: 1.476049  [    0/ 4900]\n",
            "loss: 1.500034  [ 2560/ 4900]\n",
            "Train      Loss: 1.3022 Acc: 0.8676\n",
            "Validation Loss: 1.6856 Acc: 0.4786\n",
            "\n",
            "Epoch 45\n",
            "-------------------------------\n",
            "loss: 1.495363  [    0/ 4900]\n",
            "loss: 1.480211  [ 2560/ 4900]\n",
            "Train      Loss: 1.3026 Acc: 0.8647\n",
            "Validation Loss: 1.6727 Acc: 0.4757\n",
            "\n",
            "Epoch 46\n",
            "-------------------------------\n",
            "loss: 1.466743  [    0/ 4900]\n",
            "loss: 1.465321  [ 2560/ 4900]\n",
            "Train      Loss: 1.2954 Acc: 0.8720\n",
            "Validation Loss: 1.6658 Acc: 0.4871\n",
            "\n",
            "Epoch 47\n",
            "-------------------------------\n",
            "loss: 1.459339  [    0/ 4900]\n",
            "loss: 1.494412  [ 2560/ 4900]\n",
            "Train      Loss: 1.2916 Acc: 0.8755\n",
            "Validation Loss: 1.6877 Acc: 0.4729\n",
            "\n",
            "Epoch 48\n",
            "-------------------------------\n",
            "loss: 1.419662  [    0/ 4900]\n",
            "loss: 1.439152  [ 2560/ 4900]\n",
            "Train      Loss: 1.2883 Acc: 0.8798\n",
            "Validation Loss: 1.6849 Acc: 0.4686\n",
            "\n",
            "Epoch 49\n",
            "-------------------------------\n",
            "loss: 1.364430  [    0/ 4900]\n",
            "loss: 1.414858  [ 2560/ 4900]\n",
            "Train      Loss: 1.2830 Acc: 0.8837\n",
            "Validation Loss: 1.6672 Acc: 0.4914\n",
            "\n",
            "Epoch 50\n",
            "-------------------------------\n",
            "loss: 1.342522  [    0/ 4900]\n",
            "loss: 1.422577  [ 2560/ 4900]\n",
            "Train      Loss: 1.2783 Acc: 0.8884\n",
            "Validation Loss: 1.6848 Acc: 0.4743\n",
            "\n",
            "Epoch 51\n",
            "-------------------------------\n",
            "loss: 1.367048  [    0/ 4900]\n",
            "loss: 1.402821  [ 2560/ 4900]\n",
            "Train      Loss: 1.2755 Acc: 0.8920\n",
            "Validation Loss: 1.6521 Acc: 0.5057\n",
            "\n",
            "Epoch 52\n",
            "-------------------------------\n",
            "loss: 1.321669  [    0/ 4900]\n",
            "loss: 1.384501  [ 2560/ 4900]\n",
            "Train      Loss: 1.2692 Acc: 0.8980\n",
            "Validation Loss: 1.6781 Acc: 0.4786\n",
            "\n",
            "Epoch 53\n",
            "-------------------------------\n",
            "loss: 1.347807  [    0/ 4900]\n",
            "loss: 1.383743  [ 2560/ 4900]\n",
            "Train      Loss: 1.2670 Acc: 0.8998\n",
            "Validation Loss: 1.6523 Acc: 0.5086\n",
            "\n",
            "Epoch 54\n",
            "-------------------------------\n",
            "loss: 1.322460  [    0/ 4900]\n",
            "loss: 1.404265  [ 2560/ 4900]\n",
            "Train      Loss: 1.2702 Acc: 0.8973\n",
            "Validation Loss: 1.6626 Acc: 0.4971\n",
            "\n",
            "Epoch 55\n",
            "-------------------------------\n",
            "loss: 1.323653  [    0/ 4900]\n",
            "loss: 1.397745  [ 2560/ 4900]\n",
            "Train      Loss: 1.2666 Acc: 0.9004\n",
            "Validation Loss: 1.6668 Acc: 0.4943\n",
            "\n",
            "Epoch 56\n",
            "-------------------------------\n",
            "loss: 1.394736  [    0/ 4900]\n",
            "loss: 1.392776  [ 2560/ 4900]\n",
            "Train      Loss: 1.2613 Acc: 0.9055\n",
            "Validation Loss: 1.6812 Acc: 0.4757\n",
            "\n",
            "Epoch 57\n",
            "-------------------------------\n",
            "loss: 1.320828  [    0/ 4900]\n",
            "loss: 1.387863  [ 2560/ 4900]\n",
            "Train      Loss: 1.2592 Acc: 0.9076\n",
            "Validation Loss: 1.6866 Acc: 0.4729\n",
            "\n",
            "Epoch 58\n",
            "-------------------------------\n",
            "loss: 1.321701  [    0/ 4900]\n",
            "loss: 1.358080  [ 2560/ 4900]\n",
            "Train      Loss: 1.2549 Acc: 0.9127\n",
            "Validation Loss: 1.6851 Acc: 0.4771\n",
            "\n",
            "Epoch 59\n",
            "-------------------------------\n",
            "loss: 1.321563  [    0/ 4900]\n",
            "loss: 1.384514  [ 2560/ 4900]\n",
            "Train      Loss: 1.2525 Acc: 0.9145\n",
            "Validation Loss: 1.6774 Acc: 0.4843\n",
            "\n",
            "Epoch 60\n",
            "-------------------------------\n",
            "loss: 1.321841  [    0/ 4900]\n",
            "loss: 1.358305  [ 2560/ 4900]\n",
            "Train      Loss: 1.2473 Acc: 0.9186\n",
            "Validation Loss: 1.6789 Acc: 0.4771\n",
            "\n",
            "Epoch 61\n",
            "-------------------------------\n",
            "loss: 1.321834  [    0/ 4900]\n",
            "loss: 1.352028  [ 2560/ 4900]\n",
            "Train      Loss: 1.2440 Acc: 0.9229\n",
            "Validation Loss: 1.6926 Acc: 0.4671\n",
            "\n",
            "Epoch 62\n",
            "-------------------------------\n",
            "loss: 1.369562  [    0/ 4900]\n",
            "loss: 1.354099  [ 2560/ 4900]\n",
            "Train      Loss: 1.2441 Acc: 0.9231\n",
            "Validation Loss: 1.6941 Acc: 0.4600\n",
            "\n",
            "Epoch 63\n",
            "-------------------------------\n",
            "loss: 1.331590  [    0/ 4900]\n",
            "loss: 1.327701  [ 2560/ 4900]\n",
            "Train      Loss: 1.2416 Acc: 0.9251\n",
            "Validation Loss: 1.6847 Acc: 0.4743\n",
            "\n",
            "Epoch 64\n",
            "-------------------------------\n",
            "loss: 1.321548  [    0/ 4900]\n",
            "loss: 1.356508  [ 2560/ 4900]\n",
            "Train      Loss: 1.2427 Acc: 0.9239\n",
            "Validation Loss: 1.6801 Acc: 0.4771\n",
            "\n",
            "Epoch 65\n",
            "-------------------------------\n",
            "loss: 1.348844  [    0/ 4900]\n",
            "loss: 1.318890  [ 2560/ 4900]\n",
            "Train      Loss: 1.2400 Acc: 0.9269\n",
            "Validation Loss: 1.6856 Acc: 0.4700\n",
            "\n",
            "Epoch 66\n",
            "-------------------------------\n",
            "loss: 1.321670  [    0/ 4900]\n",
            "loss: 1.324194  [ 2560/ 4900]\n",
            "Train      Loss: 1.2372 Acc: 0.9288\n",
            "Validation Loss: 1.6656 Acc: 0.4943\n",
            "\n",
            "Epoch 67\n",
            "-------------------------------\n",
            "loss: 1.321536  [    0/ 4900]\n",
            "loss: 1.333703  [ 2560/ 4900]\n",
            "Train      Loss: 1.2316 Acc: 0.9345\n",
            "Validation Loss: 1.6588 Acc: 0.5043\n",
            "\n",
            "Epoch 68\n",
            "-------------------------------\n",
            "loss: 1.321290  [    0/ 4900]\n",
            "loss: 1.293848  [ 2560/ 4900]\n",
            "Train      Loss: 1.2303 Acc: 0.9361\n",
            "Validation Loss: 1.6662 Acc: 0.4929\n",
            "\n",
            "Epoch 69\n",
            "-------------------------------\n",
            "loss: 1.327568  [    0/ 4900]\n",
            "loss: 1.294587  [ 2560/ 4900]\n",
            "Train      Loss: 1.2301 Acc: 0.9359\n",
            "Validation Loss: 1.6792 Acc: 0.4871\n",
            "\n",
            "Epoch 70\n",
            "-------------------------------\n",
            "loss: 1.351238  [    0/ 4900]\n",
            "loss: 1.330625  [ 2560/ 4900]\n",
            "Train      Loss: 1.2319 Acc: 0.9343\n",
            "Validation Loss: 1.6693 Acc: 0.4871\n",
            "\n",
            "Epoch 71\n",
            "-------------------------------\n",
            "loss: 1.328077  [    0/ 4900]\n",
            "loss: 1.307462  [ 2560/ 4900]\n",
            "Train      Loss: 1.2299 Acc: 0.9361\n",
            "Validation Loss: 1.6451 Acc: 0.5114\n",
            "\n",
            "Epoch 72\n",
            "-------------------------------\n",
            "loss: 1.321051  [    0/ 4900]\n",
            "loss: 1.310021  [ 2560/ 4900]\n",
            "Train      Loss: 1.2296 Acc: 0.9361\n",
            "Validation Loss: 1.6619 Acc: 0.5000\n",
            "\n",
            "Epoch 73\n",
            "-------------------------------\n",
            "loss: 1.353174  [    0/ 4900]\n",
            "loss: 1.317963  [ 2560/ 4900]\n",
            "Train      Loss: 1.2273 Acc: 0.9392\n",
            "Validation Loss: 1.6453 Acc: 0.5186\n",
            "\n",
            "Epoch 74\n",
            "-------------------------------\n",
            "loss: 1.327668  [    0/ 4900]\n",
            "loss: 1.290779  [ 2560/ 4900]\n",
            "Train      Loss: 1.2221 Acc: 0.9437\n",
            "Validation Loss: 1.6555 Acc: 0.5071\n",
            "\n",
            "Epoch 75\n",
            "-------------------------------\n",
            "loss: 1.321528  [    0/ 4900]\n",
            "loss: 1.297932  [ 2560/ 4900]\n",
            "Train      Loss: 1.2221 Acc: 0.9437\n",
            "Validation Loss: 1.6520 Acc: 0.5114\n",
            "\n",
            "Epoch 76\n",
            "-------------------------------\n",
            "loss: 1.325331  [    0/ 4900]\n",
            "loss: 1.289958  [ 2560/ 4900]\n",
            "Train      Loss: 1.2187 Acc: 0.9467\n",
            "Validation Loss: 1.6620 Acc: 0.4871\n",
            "\n",
            "Epoch 77\n",
            "-------------------------------\n",
            "loss: 1.318120  [    0/ 4900]\n",
            "loss: 1.290438  [ 2560/ 4900]\n",
            "Train      Loss: 1.2162 Acc: 0.9494\n",
            "Validation Loss: 1.6525 Acc: 0.5029\n",
            "\n",
            "Epoch 78\n",
            "-------------------------------\n",
            "loss: 1.259726  [    0/ 4900]\n",
            "loss: 1.289245  [ 2560/ 4900]\n",
            "Train      Loss: 1.2178 Acc: 0.9480\n",
            "Validation Loss: 1.6666 Acc: 0.4914\n",
            "\n",
            "Epoch 79\n",
            "-------------------------------\n",
            "loss: 1.228114  [    0/ 4900]\n",
            "loss: 1.295597  [ 2560/ 4900]\n",
            "Train      Loss: 1.2175 Acc: 0.9488\n",
            "Validation Loss: 1.6502 Acc: 0.5086\n",
            "\n",
            "Epoch 80\n",
            "-------------------------------\n",
            "loss: 1.230028  [    0/ 4900]\n",
            "loss: 1.290080  [ 2560/ 4900]\n",
            "Train      Loss: 1.2160 Acc: 0.9496\n",
            "Validation Loss: 1.6344 Acc: 0.5271\n",
            "\n",
            "Epoch 81\n",
            "-------------------------------\n",
            "loss: 1.238492  [    0/ 4900]\n",
            "loss: 1.288177  [ 2560/ 4900]\n",
            "Train      Loss: 1.2171 Acc: 0.9490\n",
            "Validation Loss: 1.6740 Acc: 0.4871\n",
            "\n",
            "Epoch 82\n",
            "-------------------------------\n",
            "loss: 1.277135  [    0/ 4900]\n",
            "loss: 1.389626  [ 2560/ 4900]\n",
            "Train      Loss: 1.2292 Acc: 0.9367\n",
            "Validation Loss: 1.6338 Acc: 0.5271\n",
            "\n",
            "Epoch 83\n",
            "-------------------------------\n",
            "loss: 1.252756  [    0/ 4900]\n",
            "loss: 1.311747  [ 2560/ 4900]\n",
            "Train      Loss: 1.2192 Acc: 0.9465\n",
            "Validation Loss: 1.6688 Acc: 0.4929\n",
            "\n",
            "Epoch 84\n",
            "-------------------------------\n",
            "loss: 1.228292  [    0/ 4900]\n",
            "loss: 1.322604  [ 2560/ 4900]\n",
            "Train      Loss: 1.2168 Acc: 0.9494\n",
            "Validation Loss: 1.6424 Acc: 0.5271\n",
            "\n",
            "Epoch 85\n",
            "-------------------------------\n",
            "loss: 1.229206  [    0/ 4900]\n",
            "loss: 1.319373  [ 2560/ 4900]\n",
            "Train      Loss: 1.2126 Acc: 0.9529\n",
            "Validation Loss: 1.6651 Acc: 0.5000\n",
            "\n",
            "Epoch 86\n",
            "-------------------------------\n",
            "loss: 1.229280  [    0/ 4900]\n",
            "loss: 1.290207  [ 2560/ 4900]\n",
            "Train      Loss: 1.2180 Acc: 0.9484\n",
            "Validation Loss: 1.6559 Acc: 0.5086\n",
            "\n",
            "Epoch 87\n",
            "-------------------------------\n",
            "loss: 1.257793  [    0/ 4900]\n",
            "loss: 1.289782  [ 2560/ 4900]\n",
            "Train      Loss: 1.2122 Acc: 0.9535\n",
            "Validation Loss: 1.6301 Acc: 0.5286\n",
            "\n",
            "Epoch 88\n",
            "-------------------------------\n",
            "loss: 1.227967  [    0/ 4900]\n",
            "loss: 1.290103  [ 2560/ 4900]\n",
            "Train      Loss: 1.2094 Acc: 0.9563\n",
            "Validation Loss: 1.6428 Acc: 0.5171\n",
            "\n",
            "Epoch 89\n",
            "-------------------------------\n",
            "loss: 1.227997  [    0/ 4900]\n",
            "loss: 1.318781  [ 2560/ 4900]\n",
            "Train      Loss: 1.2098 Acc: 0.9559\n",
            "Validation Loss: 1.6680 Acc: 0.4843\n",
            "\n",
            "Epoch 90\n",
            "-------------------------------\n",
            "loss: 1.228004  [    0/ 4900]\n",
            "loss: 1.290243  [ 2560/ 4900]\n",
            "Train      Loss: 1.2119 Acc: 0.9537\n",
            "Validation Loss: 1.6422 Acc: 0.5100\n",
            "\n",
            "Epoch 91\n",
            "-------------------------------\n",
            "loss: 1.227918  [    0/ 4900]\n",
            "loss: 1.287897  [ 2560/ 4900]\n",
            "Train      Loss: 1.2114 Acc: 0.9537\n",
            "Validation Loss: 1.6355 Acc: 0.5243\n",
            "\n",
            "Epoch 92\n",
            "-------------------------------\n",
            "loss: 1.228023  [    0/ 4900]\n",
            "loss: 1.325051  [ 2560/ 4900]\n",
            "Train      Loss: 1.2093 Acc: 0.9565\n",
            "Validation Loss: 1.6603 Acc: 0.4986\n",
            "\n",
            "Epoch 93\n",
            "-------------------------------\n",
            "loss: 1.227919  [    0/ 4900]\n",
            "loss: 1.290165  [ 2560/ 4900]\n",
            "Train      Loss: 1.2075 Acc: 0.9580\n",
            "Validation Loss: 1.6571 Acc: 0.5014\n",
            "\n",
            "Epoch 94\n",
            "-------------------------------\n",
            "loss: 1.256500  [    0/ 4900]\n",
            "loss: 1.293139  [ 2560/ 4900]\n",
            "Train      Loss: 1.2089 Acc: 0.9563\n",
            "Validation Loss: 1.6648 Acc: 0.4957\n",
            "\n",
            "Epoch 95\n",
            "-------------------------------\n",
            "loss: 1.227898  [    0/ 4900]\n",
            "loss: 1.279862  [ 2560/ 4900]\n",
            "Train      Loss: 1.2102 Acc: 0.9557\n",
            "Validation Loss: 1.6602 Acc: 0.5014\n",
            "\n",
            "Epoch 96\n",
            "-------------------------------\n",
            "loss: 1.227929  [    0/ 4900]\n",
            "loss: 1.257954  [ 2560/ 4900]\n",
            "Train      Loss: 1.2095 Acc: 0.9565\n",
            "Validation Loss: 1.6365 Acc: 0.5271\n",
            "\n",
            "Epoch 97\n",
            "-------------------------------\n",
            "loss: 1.227614  [    0/ 4900]\n",
            "loss: 1.258679  [ 2560/ 4900]\n",
            "Train      Loss: 1.2085 Acc: 0.9571\n",
            "Validation Loss: 1.6760 Acc: 0.4829\n",
            "\n",
            "Epoch 98\n",
            "-------------------------------\n",
            "loss: 1.310173  [    0/ 4900]\n",
            "loss: 1.262823  [ 2560/ 4900]\n",
            "Train      Loss: 1.2126 Acc: 0.9522\n",
            "Validation Loss: 1.6499 Acc: 0.5129\n",
            "\n",
            "Epoch 99\n",
            "-------------------------------\n",
            "loss: 1.227922  [    0/ 4900]\n",
            "loss: 1.261586  [ 2560/ 4900]\n",
            "Train      Loss: 1.2085 Acc: 0.9576\n",
            "Validation Loss: 1.6646 Acc: 0.4957\n",
            "\n",
            "Epoch 100\n",
            "-------------------------------\n",
            "loss: 1.228211  [    0/ 4900]\n",
            "loss: 1.257712  [ 2560/ 4900]\n",
            "Train      Loss: 1.2036 Acc: 0.9620\n",
            "Validation Loss: 1.6425 Acc: 0.5229\n",
            "Done for all 100 epochs in 107 minutes\n",
            "\n",
            "Best Validation accuracy: 0.5285714285714286 \n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Draw Graphs - Training and Validation"
      ],
      "metadata": {
        "id": "ZiTdXyXnUdGy"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from matplotlib import pyplot as plt\n",
        "plt.figure(figsize=[6.4, 4.8],facecolor='white', edgecolor='black',dpi=100)\n",
        "\n",
        "plt.title('Loss Curve')\n",
        "plt.plot(train_loss,label='Training Loss')\n",
        "plt.plot(val_loss,label='Validation Loss')\n",
        "#plt.plot(test_loss,label='Testing Loss')  #comment this for inference\n",
        "plt.xlabel('Epoch')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "plt.grid()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 465
        },
        "id": "v4wfvYkfTZ9E",
        "outputId": "8191ff3c-2217-4f4c-ff7a-8109735e876b"
      },
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.figure(figsize=[6.4, 4.8],facecolor='white', edgecolor='black',dpi=100)\n",
        "plt.title('Accuracy Curve')\n",
        "plt.plot(train_acc,label='Training Accuracy')\n",
        "plt.plot(val_acc,label='Validation Accuracy')\n",
        "#plt.plot(test_acc,label='Testing Accuracy')  #comment this for inference\n",
        "plt.xlabel('Epoch')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.legend()\n",
        "plt.grid()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 465
        },
        "id": "jLHvmjjeT03M",
        "outputId": "81c23825-b526-45a5-d4a5-8a3500d0ddca"
      },
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjcAAAHACAYAAABeV0mSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAAPYQAAD2EBqD+naQAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzdeVxU5f7A8c8M+7AjCIIoAu4bikvumhpmWa6p5Z6Yde2mXH+iuedNbzf16nXJbrmkaVqmZddSkTLNSC2X3BdEUWRXdhiGmfP7g5zrBCgoMArf9+s1r5ozzznnex5Gzpfnec7zqBRFURBCCCGEqCLU5g5ACCGEEKI8SXIjhBBCiCpFkhshhBBCVCmS3AghhBCiSpHkRgghhBBViiQ3QgghhKhSJLkRQgghRJUiyY0QQgghqhRJboQQQghRpUhyI4QQQogqRZIbIaqw1atXo1KpaN++vblDeSIlJiYydepUGjVqhEajwd7enuDgYP7+97+TlpZm7vCEECVQydpSQlRdnTp14tatW1y7do3Lly8TGBho7pCeGMeOHaNv375kZWUxYsQIgoODAfj111/ZunUrHTt2ZN++fWaOUghRHEluhKiiYmJi8Pf3Z8eOHbz22mv85S9/Ye7cueYOq1jZ2dnY29ubOwyjtLQ0mjVrRkFBAQcOHKBRo0YmnycmJvLRRx8xa9asRz7X43btQlQF0i0lRBW1efNmXF1dee655xg8eDCbN28utlxaWhpTpkzBz88PGxsbateuzahRo0hJSTGWycvLY968eTRo0ABbW1tq1arFwIEDiY6OBuDAgQOoVCoOHDhgcuxr166hUqnYsGGDcduYMWNwcHAgOjqavn374ujoyCuvvALAoUOHGDJkCHXq1MHGxgZfX1+mTJlCbm5ukbgvXLjASy+9hIeHB3Z2djRs2JCZM2cC8MMPP6BSqdi5c2eR/bZs2YJKpSIqKqrEuvvwww+Ji4tj6dKlRRIbAE9PT5PERqVSMW/evCLl/Pz8GDNmjPH9hg0bUKlU/Pjjj7zxxhvUrFmT2rVrs337duP24mJRqVScOXPG5NoHDx6Mm5sbtra2tGnThl27dpV4PUJUN5bmDkAIUTE2b97MwIEDsba2Zvjw4XzwwQccO3aMtm3bGstkZWXRpUsXzp8/z7hx42jdujUpKSns2rWLmzdv4u7ujl6v5/nnnycyMpJhw4bx1ltvkZmZSUREBGfOnCEgIKDMsRUUFBASEkLnzp1ZvHgxGo0GgC+++IKcnBxef/11atSowdGjR1mxYgU3b97kiy++MO7/+++/06VLF6ysrJgwYQJ+fn5ER0fzzTff8O6779K9e3d8fX3ZvHkzAwYMKFIvAQEBdOjQocT4du3ahZ2dHYMHDy7ztZXGG2+8gYeHB3PmzCE7O5vnnnsOBwcHPv/8c7p162ZSdtu2bTRt2pRmzZoBcPbsWTp16oSPjw/Tp0/H3t6ezz//nP79+/Pll18WuV4hqiVFCFHl/PrrrwqgREREKIqiKAaDQaldu7by1ltvmZSbM2eOAig7duwocgyDwaAoiqKsW7dOAZSlS5eWWOaHH35QAOWHH34w+TwmJkYBlPXr1xu3jR49WgGU6dOnFzleTk5OkW2LFi1SVCqVcv36deO2rl27Ko6Ojibb7o1HURRlxowZio2NjZKWlmbclpSUpFhaWipz584tcp57ubq6Ki1btrxvmXsBxR6zbt26yujRo43v169frwBK586dlYKCApOyw4cPV2rWrGmyPT4+XlGr1co777xj3NazZ0+lefPmSl5ennGbwWBQOnbsqNSvX7/UMQtRlUm3lBBV0ObNm/H09KRHjx5AYbfJ0KFD2bp1K3q93ljuyy+/pGXLlsX+ta9SqYxl3N3defPNN0ss8zBef/31Itvs7OyM/5+dnU1KSgodO3ZEURROnDgBQHJyMgcPHmTcuHHUqVOnxHhGjRqFVqtl+/btxm3btm2joKCAESNG3De2jIwMHB0dH+q6SiM0NBQLCwuTbUOHDiUpKcmka2/79u0YDAaGDh0KwO3bt/n+++956aWXyMzMJCUlhZSUFFJTUwkJCeHy5cvExcVVWNxCPCkkuRGiitHr9WzdupUePXoQExPDlStXuHLlCu3btycxMZHIyEhj2ejoaGN3R0mio6Np2LAhlpbl14ttaWlJ7dq1i2yPjY1lzJgxuLm54eDggIeHh7GbJj09HYCrV68CPDDuRo0a0bZtW5OxRps3b+app5564FNjTk5OZGZmlumayqJevXpFtvXp0wdnZ2e2bdtm3LZt2zaCgoJo0KABAFeuXEFRFGbPno2Hh4fJ6+5g8aSkpAqLW4gnhYy5EaKK+f7774mPj2fr1q1s3bq1yOebN2/mmWeeKddzltSCc28r0b1sbGxQq9VFyvbu3Zvbt28THh5Oo0aNsLe3Jy4ujjFjxmAwGMoc16hRo3jrrbe4efMmWq2WX375hZUrVz5wv0aNGnHy5Eny8/OxtrYu83nvKun6722husvGxob+/fuzc+dOVq9eTWJiIocPH2bhwoXGMnfrYOrUqYSEhBR7bHncXwhJboSocjZv3kzNmjVZtWpVkc927NjBzp07WbNmDXZ2dgQEBJg8hVOcgIAAjhw5gk6nw8rKqtgyrq6uAEUmtrt+/Xqp4z59+jSXLl3ik08+YdSoUcbtERERJuX8/f0BHhg3wLBhwwgLC+Ozzz4jNzcXKysrYxfP/fTr14+oqCi+/PJLhg8f/sDyrq6uRa49Pz+f+Pj4B+57r6FDh/LJJ58QGRnJ+fPnURTFJN67125lZUWvXr3KdGwhqhPplhKiCsnNzWXHjh08//zzDB48uMhr0qRJZGZmGh8bHjRoEKdOnSr2kWnljymwBg0aREpKSrEtHnfL1K1bFwsLCw4ePGjy+erVq0sd+90xKMo9U28pisLy5ctNynl4eNC1a1fWrVtHbGxssfHc5e7uzrPPPsunn37K5s2b6dOnD+7u7g+MZeLEidSqVYu//e1vXLp0qcjnSUlJ/P3vfze+DwgIKHLt//nPf0psuSlJr169cHNzY9u2bWzbto127dqZdGHVrFmT7t278+GHHxabOCUnJ5fpfEJUVdJyI0QVsmvXLjIzM3nhhReK/fypp57Cw8ODzZs3M3ToUP7v//6P7du3M2TIEMaNG0dwcDC3b99m165drFmzhpYtWzJq1Cg2btxIWFgYR48epUuXLmRnZ7N//37eeOMNXnzxRZydnRkyZAgrVqxApVIREBDAf//73zKN/2jUqBEBAQFMnTqVuLg4nJyc+PLLL7lz506Rsv/+97/p3LkzrVu3ZsKECdSrV49r166xe/duTp48aVJ21KhRxke6FyxYUKpYXF1d2blzJ3379iUoKMhkhuLjx4/z2WefmTxKPn78eCZOnMigQYPo3bs3p06dYu/evaVKpO5lZWXFwIED2bp1K9nZ2SxevLhImVWrVtG5c2eaN29OaGgo/v7+JCYmEhUVxc2bNzl16lSZzilElWS+B7WEEOWtX79+iq2trZKdnV1imTFjxihWVlZKSkqKoiiKkpqaqkyaNEnx8fFRrK2tldq1ayujR482fq4ohY9oz5w5U6lXr55iZWWleHl5KYMHD1aio6ONZZKTk5VBgwYpGo1GcXV1VV577TXlzJkzxT4Kbm9vX2xs586dU3r16qU4ODgo7u7uSmhoqHLq1Kkix1AURTlz5owyYMAAxcXFRbG1tVUaNmyozJ49u8gxtVqt4urqqjg7Oyu5ubmlqUajW7duKVOmTFEaNGig2NraKhqNRgkODlbeffddJT093VhOr9cr4eHhiru7u6LRaJSQkBDlypUrJT4KfuzYsRLPGRERoQCKSqVSbty4UWyZ6OhoZdSoUYqXl5diZWWl+Pj4KM8//7yyffv2Ml2fEFWVLL8ghKjSCgoK8Pb2pl+/fqxdu9bc4QghKoGMuRFCVGlfffUVycnJJoOUhRBVm7TcCCGqpCNHjvD777+zYMEC3N3dOX78uLlDEkJUEmm5EUJUSR988AGvv/46NWvWZOPGjeYORwhRicya3Bw8eJB+/frh7e2NSqXiq6++euA+Bw4coHXr1tjY2BAYGGiy2rAQQty1YcMGCgoK+PXXXx84m7EQomoxa3KTnZ1Ny5Yti51srDgxMTE899xz9OjRg5MnTzJ58mTGjx/P3r17KzhSIYQQQjwpHpsxNyqVip07d9K/f/8Sy4SHh7N7926TmUmHDRtGWloae/bsqYwwhRBCCPGYe6Im8YuKiioy5XhISAiTJ08u9TEMBgO3bt3C0dHxkVY0FkIIIUTlURSFzMxMvL29i6xN92dPVHKTkJCAp6enyTZPT08yMjLIzc0tdjE6rVaLVqs1vo+Li6NJkyYVHqsQQgghyt+NGzeoXbv2fcs8UcnNw1i0aBHz588vsv3jjz9Go9GYISIhhBBClFVOTg7jx4/H0dHxgWWfqOTGy8uLxMREk22JiYk4OTkV22oDMGPGDMLCwozvMzIy8PX1pX///jg5OZVrfDqdjoiICHr37l3i6smifEhdVx6p68ojdV15pK4rT3nVdUZGBuPHjy/VkJInKrnp0KED3377rcm2iIgIkwXs/szGxgYbG5si262srCrsC12RxxampK4rj9R15ZG6rjxS15XnUeu6LPua9VHwrKwsTp48aVzFNyYmhpMnTxIbGwsUtrrcO2X6xIkTuXr1KtOmTePChQusXr2azz//nClTppglfiGEEEI8fsya3Pz666+0atWKVq1aARAWFkarVq2YM2cOAPHx8cZEB6BevXrs3r2biIgIWrZsyZIlS/j4448JCQkxS/xCCCGEePyYtVuqe/fu3G+aneJmH+7evTsnTpyowKgK6fV6dDpdmfbR6XRYWlqSl5eHXq+voMgESF0/CisrKywsLMwdhhBCVJgnasxNZVAUhYSEBNLS0h5qXy8vL27cuCFz6FQwqetH4+LigpeXl9SdEKJKkuTmT+4mNjVr1kSj0ZTpl7/BYCArKwsHB4cHTjAkHo3U9cNRFIWcnBySkpIAqFWrlpkjEkKI8ifJzT30er0xsalRo0aZ9zcYDOTn52Nrays33Aomdf3w7k6bkJSURM2aNaWLSghR5chd4R53x9jI5H6iqrv7HS/ruDIhhHgSSHJTDBmHIKo6+Y4LIaoySW6EEEIIUaVIciNK5Ofnx7Jly0pd/sCBA6hUqod60kwIIYQoL5LcVAEqleq+r3nz5j3UcY8dO8aECRNKXb5jx47Ex8fj7Oz8UOd7GE2aNMHGxoaEhIRKO6cQQojHmyQ3VUB8fLzxtWzZMpycnEy2TZ061VhWURQKCgpKdVwPD48yDa62trau1LlToqKiyM3NZfDgwXzyySeVcs77kcG5QojqTlEUjsbc5kTsHbPGIclNFeDl5WV8OTs7o1KpjO8vXLiAo6Mj3333HcHBwdjY2PDTTz8RHR3Niy++iKenJw4ODrRt25b9+/ebHPfP3VIqlYqPP/6YAQMGoNFoqF+/Prt27TJ+/uduqQ0bNuDi4sLevXtp3LgxDg4O9OnTh/j4eOM+BQUF/PWvf8XFxYUaNWoQHh7O6NGj6d+//wOv+9NPP2X48OGMHDmSdevWFfn85s2bDB8+HDc3N+zt7WnTpg1Hjhwxfv7NN9/Qtm1bbG1tcXd3Z8CAASbX+tVXX5kcz8XFxThr9rVr11CpVGzbto1u3bpha2vL5s2bSU1NZfjw4fj4+KDRaGjevDmfffaZyXEMBgP//Oc/CQwMxMbGhjp16vDuu+8C8PTTTzNp0iST8snJyVhbWxMZGfnAOhFCCHPI1hbw6S/XeXb5IV76MIr39140azwyz80DKIpCrq500/sbDAZy8/VY5heUy9wrdlYW5dYKMn36dBYvXoy/vz+urq7cuHGDvn378u6772JjY8PGjRvp168fFy9epE6dOiUeZ/78+fzzn//k/fffZ8WKFbzyyitcv34dNze3Ysvn5OSwePFiNm3ahFqtZsSIEUydOpXNmzcD8N5777F582bWr19P48aNWb58OV999RU9evS47/VkZmby9ddfExUVRZMmTUhPT+fQoUN06dIFKFyUtVu3bvj4+LBr1y68vLw4fvw4BoMBgN27dzNgwABmzpzJxo0byc/PL7LifGnrdcmSJbRq1QpbW1vy8vIIDg4mPDwcJycndu/ezciRIwkICKBdu3ZA4YKwH330Ef/617/o3Lkz8fHxXLhwAYDx48czadIklixZYlzN/tNPP8XHx4enn366zPEJIao+RVFIzc4n9nYON/542Vlb4u9hT4C7Az6udlioS3cvURQFbYGBjDwdWXkFuGqscbW3LrF8TEo2n/x8jS9/u0mmtrBXwM7Kgro1NBToDVhamKcNRZKbB8jV6WkyZ69Zzn3unRA01uXzI3rnnXfo3bu38b2bmxstW7Y0vl+wYAE7d+5k165dRVoO7jVmzBiGDx8OwMKFC/n3v//N0aNH6dOnT7HldToda9asISAgAIBJkybxzjvvGD9fsWIFM2bMMLaarFy5slRJxtatW/H396dp06ao1WqGDRvG2rVrjcnNli1bSE5O5tixY8bEKzAw0Lj/u+++y7Bhw5g/f75x2731UVqTJ09m4MCBJtvu7QZ888032bt3L59//jnt2rUjMzOT5cuXs3LlSkaPHg1AQEAAnTt3BmDgwIFMmjSJr7/+mpdeegkobAEbM2aMPL4txBPOYFA4n5DB4SspHL6SSmJGHnbWFthaWmBnXfgK8HAgyNeZFrVdcHewKfY4eoPC7zfTOHwlhZ+upPD7zXRy8kv+I9zaUo2vqx0uGmscbCxxsLXEydaSAr3CnZx87uTouJOdT1qujsw8HTr9/9Z8tFSreCHIm9Au/jSu5WTcfiUpk5XfX2HXqVsY/iju727PiKfqMii4Ns52VuVTaQ9Jkptqok2bNibvs7KymDdvHrt37yY+Pp6CggJyc3NNVmEvTosWLYz/b29vj5OTk3Eq/+JoNBpjYgOF0/3fLZ+enk5iYqKxRQPAwsKC4OBgYwtLSTZs2GC8+QOMGDGCbt26sWLFChwdHTl58iStWrUqsUXp5MmThIaG3vccpfHnetXr9SxcuJDPP/+cuLg48vPz0Wq1xrFL58+fR6vV0rNnz2KPZ2tra+xme+mllzh+/Dhnzpwx6f4TQjw+btzO4UpSFm38XHG0LXpDVxSFqKupfHb0Bj9fSSE1O7/Ux/ZxsSOgpgMW9/xdo9MrnLqZRmae6dhJlQpqOdni66ahtquGbG0BMSnZxKRmk19gIDo5G8gu9blVKtBYWZCdr2fH8Th2HI+jawMPhgTXZu/ZBHafjufuutc9G9VkTCc/OgW4oy5lC1FFk+TmAeysLDj3TkipyhoMBjIzMnF0ciy3bqnyYm9vb/J+6tSpREREsHjxYgIDA7Gzs2Pw4MHk59//H56Vlek/XpVKdd9EpLjy91sJvjTOnTvHL7/8wtGjR02eBNPr9WzdupXQ0FDjEgMledDnxcVZ3IDhP9fr+++/z/Lly1m2bBnNmzfH3t6eyZMnG+v1QeeFwq6poKAgbt68yfr163n66aepW7fuA/cTQlSOK0mZ7DmTwHdnEjh7KwMAe2sLBrauzcgOdWng6YjBoBB5IYlVP1zh5I3/TY+hsbbgKf8adAyoQX1PR/ILDOTq9OTl68nI03EuPoPfb6YTnZxFXFoucWm5xcbgZGtJxwB3OtV356l6btSpocHGsug9Q29QuJWWy43bOWTkFZCZpyNLW0BmXgEWahWuGmvc7K1w0VjjqrHGyc4SBxtL7K0tUatVnLyRxkcHr/LdmXgOXkrm4KVk47H7NPVi0tOBNPOpvCdkS0uSmwdQqVSl7hoyGAwUWFugsbZ87Nc7Onz4MGPGjDF2B2VlZXHt2rVKjcHZ2RlPT0+OHTtG165dgcIE5fjx4wQFBZW439q1a+natSuLFi0yWThz/fr1rF27ltDQUFq0aMHHH3/M7du3i229adGiBZGRkYwdO7bYc3h4eJgMfL58+TI5OTkPvKbDhw/z4osvMmLECKDwO3Hp0iWaNGkCQP369bGzsyMyMpLx48cXe4zmzZvTpk0bPvroI7Zs2cLKlSsfeF4hqrLULC1HY25zJOY2R2Nuk683UMdNQx03Db5uGmo525KTr+dOdv4f3Sz53MnWcTsnn7Q/ul3Sc3XYWKpxtLHE0dYKexsL7LRquuQV4GZ1/y6UjDwdv0SncvhKCocup3A15X8tIGoVeDjakJihZdMv19n0y3Xa13PjTk4+lxKzgMJuoZfa1OaFlj4E+bpgbfng+0Nmno7TcenE3TFNblQqFYE1HWju41yqcTQWahW+f9TTwwjydWHVK62JTc1h3eEYIs4lEuTrwqSnA026qR43ktxUU/Xr12fHjh3069cPlUrF7NmzH9gVVBHefPNNFi1aRGBgII0aNWLFihXcuXOnxPElOp2OTZs2MW/ePJo0aYKTk5MxuRk/fjxLly7l7NmzDB8+nIULF9K/f38WLVpErVq1OHHiBN7e3nTo0IG5c+fSs2dPAgICGDZsGAUFBXz77beEh4cDhU8trVy5kg4dOqDX6wkPDy/SClWc+vXrs337dn7++WdcXV1ZunQpiYmJxuTG1taW8PBwpk2bhrW1NZ06dSI5OZmzZ8/y6quvGo9zd2Cxvb29yVNcQlQHeTo9R2Ju8+PFZA5dTuZyUlaRMleK2fYg+QWGwu6c9Lw/tqh5c+spNoxrh1UxA1/3nU3ggx+jOXUjzTiuBMDKQkWnQHeebeZFr8aeuNlb83N0KhujrhFxLpEjMbcBcLCxZMRTdRnX2Y+ajrZlitXR1oqOAe5lvsaKUqeGhnkvNGXeC03NHUqpSHJTTS1dupRx48bRsWNH3N3dCQ8PJyMjo9LjCA8PJyEhgVGjRmFhYcGECRMICQkpcaXqXbt2kZqaWuwNv3HjxjRu3Ji1a9eydOlS9u3bx9/+9jf69u1LQUEBTZo0YdWqVQB0796dL774ggULFvCPf/wDJycnY+sRwJIlSxg7dixdunTB29ub5cuX89tvvz3wembNmsXVq1cJCQlBo9EwYcIE+vfvT3p6urHM7NmzsbS0ZM6cOdy6dYtatWoxceJEk+MMHz6cyZMnM3z4cGxty/ZLUYjHnaIo/Hr9Dkdjbpt0/+oNcPLGHaKuppKnM/1jq4GnA+3r1aC9vxsudtbcuJND7O3CV0J6HvY2lrhqrAqf7tFY42pv9UeXizUuGiuc7ayMyU2WtoCbt7OY/dUZDkenMmPHad4f3MLkj6pPfr7GvG/OGseV+HvY0ynAnU6B7nQMrIHTn8bXdAos/OxWWi47T8RhY6lmSBtfsw+sra5UyqMOgHjCZGRk4OzsTHp6Ok5Opk1qeXl5xMTEUK9evYe6oRgMBjIyMkxaE0TZGAwGGjduzEsvvcSCBQvuW64q1/W1a9cICAjg2LFjtG7dutyPX5bvuk6n49tvv6Vv376lar0SD6+q13W2toCvTsaxKeo6FxIy71vW08mG7g1q0q2hB0/518DtPo8jPwydTsf7m7/j44sWGBR4q2d9pvRugKIoLI24xIrvrwDwcvs6TOoRiLfLg8fLieKV1/f6fvfvP5OWG2FW169fZ9++fXTr1g2tVsvKlSuJiYnh5ZdfNndoZqHT6UhNTWXWrFk89dRTFZLYCFGZ8nR6jl27TcS5RHYejzPOhWJrpaZnY0+cbE1vQ3Xc7One0INGXo4VPv1BU1eF+f2aMHvXOZZHXsbL2Zbfb6bz2dHCp0bDejfgzacDZRqGJ5AkN8Ks1Go1GzZsYOrUqSiKQrNmzdi/fz+NGzc2d2hmcfjwYXr06EGDBg3Yvn27ucMR1cjt7HyuJmeRqS3ASq3G0kKFlYUKawsLarvaFTuRm6IoJGTkEZ2UjU5v2o10404OBy4mExWdajIRql8NDSOeqsuQYF+cNeZvnRrWtjYJmVpW/RDNjB2ngcJBwgv6N+OV9vKU4pNKkhthVr6+vhw+fNjcYTw2unfv/siPygtRGr9cTeXL325yJTmLmJRs0nLuvzaaq8YKfw8H/N3tsbRQcykxk0uJmUXmWymOp5MN3Rp48FwLb7oEPj5zodw19ZmG3ErLY+eJOKwt1Px7eBB9mtUyd1jiEUhyI4QQ1ciZuHTe33uRH++Zr+QuHxc7XDRWFOgVdAYDeoNCTr6e5Ewtd3J0/Hb9Dr9dN10Q0UKtom4NDfZ/mjLDyc6SzoEeldbF9ChUKhXvDWpBGz9XWtZ2eSznbRFlI8mNEEJUcQaDwuWkLFb+cIVvTt0CCqfVH9KmNp0C3fF3d6Ceuz121sU/pZiTXzjb7dXkwpdOb6C+pwMNvRyp525f7ORxTxprS7V0Q1UhktwIIUQVcnfdoWPXbnMxIYvLSZlcTswyGffyQktvwno3wM/d/j5H+h+NtSVNvZ1p6i0tGuLJIMmNEEI85hRFIfZ2DuduZXD2Vgap2fl4OdlSy8UWb2c7PBxtOHsrnQN/THp3p5jxM9aWarrWd2dK7waSpIgqT5IbIYR4DMSm5vDZ0WscuqBmR8pxDIBObyBPZyA6Kcv4CHVpONpa0jGgBk29nWng6UADT0fquGmwLGYWXiGqIkluhBDCTPILDOw/n8hnR2M5dDnlj61quJNSpKy1hZqGXo409XaippMtSRl53ErP41ZaLokZedRx09C9oQfdG9akla+LJDKiWpPkRhh1796doKAgli1bBoCfnx+TJ09m8uTJJe6jUqnYuXMn/fv3f6Rzl9dxhHgSXErM5Itfb7DzRBwpWfnG7Z0Da+ClTyI4qAW21pZYqtVYWajxc9cQ4OFQ7PpHQoiiJLmpAvr164dOp2PPnj1FPjt06BBdu3bl1KlTtGjRokzHPXbsGPb2pRtwWFrz5s3jq6++4uTJkybb4+PjcXV1LddzlSQ3NxcfHx/UajVxcXHY2NhUynlF9ZaapeXbMwls//UGp27+b60xD0cbhrbxZWhbX7wcrQqnqW/tUyWXXxCiskhyUwW8+uqrDBo0iJs3b1K7dm2Tz9avX0+bNm3KnNgAeHh4lFeID+Tl5VVp5/ryyy9p2rQpiqLw1VdfMfsuWK4AACAASURBVHTo0Eo7958pioJer8fSUv4pViW3s/PZdzaBCwmZXE7K5GJCFilZWuPnlmoVTzeqyZA2vnRv6GFskdHp7j+RnhCidKSNswp4/vnn8fDwYMOGDSbbs7Ky+OKLL3j11VdJTU1l+PDh+Pj4oNFoaN68OZ999tl9j+vn52fsogK4fPkyXbt2xdbWliZNmhAREVFkn/DwcBo0aIBGo8Hf35/Zs2cbf2Fv2LCB+fPnc+rUKVQqFSqVyhizSqXiq6++Mh7n9OnTPP3009jZ2VGjRg0mTJhAVlaW8fOxY8fyyiuvsGTJEmrVqkWNGjX4y1/+Uqqbw9q1axkxYgQjRoxg7dq1RT4/e/Yszz//PE5OTjg6OtKlSxeio6ONn69bt46mTZtiY2NDrVq1mDRpElC42KVKpTJplUpLS0OlUnHgwAEADhw4gEql4rvvviM4OBgbGxt++uknoqOjefHFF/H09MTBwYG2bduyf/9+k7i0Wi3h4eH4+vpiY2NDYGAga9euRVEUAgMDWbx4sUn5kydPolKpuHLlygPrRJSPW2m5zP/mLJ3+8T3Td5xmw8/XOHwl1ZjYNK7lxKznGvPL2z35z6g29G7iKV1NQlQA+XPxQRQFdDmlK2swFJbNt4DyWKnaSgOlmNXT0tKSUaNGsWHDBmbOnGmcCfSLL75Ar9czfPhwsrKyCA4OJjw8HCcnJ3bv3s3IkSMJCAigXbt2DzyHwWBg4MCBeHp6cuTIEdLT04sdi+Po6MiGDRvw9vbm9OnThIaG4ujoyLRp0xg6dChnzpxhz549xhu3s3PRR1Kzs7MJCQmhQ4cOHDt2jKSkJMaPH8+kSZNMErhDhw7h6+vLDz/8wJUrVxg6dChBQUGEhoaWeB3R0dFERUWxY8cOFEVhypQpXL9+nbp1CyfviouLo2vXrnTv3p3vv/8eJycnDh8+TEFB4ZMqH3zwAWFhYfzjH//g2WefJT09/aGWj5g+fTqLFy/G398fV1dXbty4Qd++fXn33XexsbFh48aN9OvXj4sXL1KnTh0ARo0aRVRUFP/+979p2bIlMTExpKSkoFKpGDduHOvXr2fq1KnGc6xfv56uXbsSGBhY5vhE6ekNChcTMtnwcww7T8Sh0xcun9GklhMdA2rQwMuRhp6OBNZ0wN5GfuUKURnM/i9t1apVvP/++yQkJNCyZUtWrFhR4s1Wp9OxaNEiPvnkE+Li4mjYsCHvvfceffr0qbgAdTmw0LtURdWAS3me++1bYF26MS/jxo3j/fff58cff6R79+5A4c1t0KBBODs74+zsbHLje/PNN9m7dy+ff/55qZKb/fv3c+HCBfbu3Yu3d2F9LFy4kGeffdak3KxZs4z/7+fnx9SpU9m6dSvTpk3Dzs4OBwcHLC0t79sNtWXLFvLy8ti4caNxzM/KlSvp168f7733Hp6engC4uLiwYsUKrKysaNSoEc899xyRkZH3TW7WrVvHs88+axzfExISwvr165k3bx5Q+H10dnZm69atxjEPDRo0MO7/97//nb/97W+89dZbxm1t27Z9YP392TvvvEPv3r2N793c3GjZsqXx/YIFC9i5cye7du1i0qRJXLp0ic8//5yIiAh69eoFgL+/v7H8mDFjmDNnDkePHqVdu3bodDq2bNlSpDVHPLqkjDz2n0/i7K10zsVncCE+02SCvKf83XijeyBd6rs/1ksOCFGVmbU9dNu2bYSFhTF37lyOHz9Oy5YtCQkJISkpqdjys2bN4sMPP2TFihWcO3eOiRMnMmDAAE6cOFHJkT9+GjVqRMeOHVm3bh0AV65c4dChQ7z66qsA6PV6FixYQPPmzXFzc8PBwYG9e/cSGxtbquOfP38eX19fY2ID0KFDhyLltm3bRqdOnfDy8sLBwYFZs2aV+hz3nqtly5Ymg5k7deqEwWDg4sWLJtdsYfG/ad9r1apV4ncHCuvgk08+YcSIEcZtI0aMYMOGDRgMhSsanzx5ki5duhQ7mDMpKYlbt27Rs2fPMl1Pcdq0aWPyPisri6lTp9K4cWNcXFxwcHDg/Pnzxro7efIkFhYWdOvWrdjjeXt789xzzxl//t988w1arZYhQ4Y8cqyi0PXUbN7eeZrO7/3A2ztPs/lILCdi08jV6bGzsqB3E0++fL0jWyd0oGsDD0lshDAjs7bcLF26lNDQUMaOHQvAmjVr2L17N+vWrWP69OlFym/atImZM2fSt29fAF5//XX279/PkiVL+PTTTysmSCtNYQtKKRgMBjIyM3FydERdXt1SZfDqq6/y5ptvsmrVKtavX09AQIDxZvj++++zfPlyli1bRvPmzbG3t2fy5Mnk5+c/4KilFxUVxSuvvML8+fMJCQkxtoAsWbKk3M5xrz8nICqVypikFGfv3r3ExcUVGUCs1+uJjIykd+/e2NnZlbj//T4DjD/ze1f1LmkM0J+fQps6dSoREREsXryYwMBA7OzsGDx4sPHn86BzA4wfP56RI0fyr3/9i/Xr1zN06FA0mrJ9h4QpRVE4F5/Bhz9e5b+/38Lwx4+2pa8LT/m70dTbmSa1nKjnbo/FY7bStRDVmdmSm/z8fH777TdmzJhh3KZWq+nVqxdRUVHF7qPVarG1tTXZZmdnx08//VTiebRaLVrt/55SyMjIAApvOn++8eh0OhRFwWAwmN4kLR98Y4E/bmpWehQrDYby+KtNUQpfpTR48GDeeustPv30UzZu3MjEiRNRFAVFUfjpp5944YUXePnll4HCROzSpUs0btzY5FrvXv+f3zds2JAbN24QFxdHrVq1APj555+NxzIYDBw+fJi6deua/EyvXbtmLAOFCYlery82Cbl7nIYNG7JhwwYyMzONScChQ4dQq9XUr1+/xHjvJhUlJTgff/wxQ4cO5e233zbZvnDhQj7++GN69uxJ8+bN2bhxI1qttkjyZG9vj5+fH/v37y+2BaVGjRpA4bidu11Mx48fN7m2u7H9+Tt2+PBhRo8ezYsvvggUtuRcu3bNeH1NmzbFYDDwww8/GLul/qxPnz7Y29uzevVq9uzZw4EDB0qsC4PBgKIo6HQ6k9av4tz9d/KkP8mTm6/nbHwGlxKzuJxU+LqWmkNtFzta1XGhta8Lreo4Y2dlwS8xtzl4OYWDl1K4mZZnPEbX+jWY2NWftn6m0xYY9AUY9H8+Y9lVlbp+EkhdV57yquuy7G+25CYlJQW9Xm8cP3GXp6cnFy5cKHafkJAQli5dSteuXQkICCAyMpIdO3ag15f8W2XRokXMnz+/yPZ9+/YV+av27liQrKysR2rRyMzMfOh9H9WAAQN4++23yczMZODAgcZkrm7dunz99ddERETg4uLC6tWrSUhIoH79+sYyBQUF5OfnG98bDAby8vLIyMigXbt2BAYGMnLkSObPn09mZiYzZ84ECueNycjIwNvbm9jYWNavX0/r1q3Zt28fO3fuRFEU4zFr1qxJTEwMhw8fxtvbGwcHB+M8M3eP069fP+bNm8eIESMIDw8nNTWVv/71rwwdOhQ7OzsyMjKMX/J76zo/P5+CggLjue6VkpLCf//7X7Zs2WIcoHvXoEGDGDlyJNevX2fUqFGsWLGCIUOGMGXKFJycnDh27BjBwcHUr1+fadOmERYWhpOTE7169SIrK4sjR44wYcIEoHD8zcKFC/Hw8CAlJYW5c+cCkJOTQ0ZGBjk5Oca4723d8/PzY/v27fTo0QMoTLgMBoPx5+Hm5sbw4cMZN24c7733Hs2aNePGjRskJyczYMAA43GGDRvG22+/TUBAAE2bNi22Lu7WVW5uLgcPHjQOln6Q4p6OexLoDfBzkoo9N9RkFRT9oyMxQ8tvsWnG92oUDPyvnIVKobmbQm8fA7XtE0k+l8i35yo25ie1rp9EUteV51Hr+u7vz9Iw+4Disli+fDmhoaE0atQIlUpFQEAAY8eONY4zKM6MGTMICwszvs/IyMDX15dnnnkGJycnk7J5eXncuHEDBweHIi1EpaEoCpmZmTg6Opqtv/21115j06ZNPPvsszRs2NC4ff78+dy8eZPBgwej0WgIDQ2lf//+pKenG+vB0tISa2tr43u1Wo2tra3x/c6dOwkNDaVXr17Gx8T79u2LnZ0dTk5ODBs2jBMnThAeHo5Wq6Vv377Mnj2b+fPnG48xYsQI9uzZwwsvvEBaWhpr165lzJgxAMbjODk5sWfPHqZMmULPnj3RaDQMHDiQJUuW4ODgAPyvS+reura2tsbS0rLIzxUKW23s7e3p169fkRaZfv36YWdnx65du3jzzTeJjIxk2rRpPP/881hYWBAUFESvXr1wcnLitddeAwq/i7Nnz8bd3Z1BgwYZz7l+/XpCQ0Pp0aMHDRs25B//+Ad9+vRBo9Hg5ORkTKgdHR1N4ly+fDnjx48nJCQEd3d3pk2bRm5ursnP46OPPmLmzJn83//9H6mpqdSpU4fp06ebHOf1119n6dKljBs3rth6uCsvLw87Ozvjo/33o9PpiIiIoHfv3k/UxHKKorDvXBL/irjMtdTCX4oeDtY08Xaifk0HGtR0oG4NDddSszkem86J2DQuJ2dhUFTUdrWjW313ujZwp72fa6U95fSk1vWTSOq68pRXXZf0x1pxVIpShn6PcpSfn49Go2H79u0mU+6PHj2atLQ0vv766xL3zcvLIzU1FW9vb6ZPn85///tfzp49W6rzZmRk4OzsbHJTv/e4MTEx1KtX76GSG4PBQEZGBk5OTuUz5kaUSOq6eIcOHaJnz57cuHGjSKvovcryXdfpdIWz5vbt+0TcBLK1Bew/n8imqOv8ev0OAO4O1rzVqwHD2vred16Z9FwdmXk6fFzszPIHypNW108yqevKU151fb/795+ZreXG2tqa4OBgIiMjjcmNwWAgMjLSOClaSWxtbfHx8UGn0/Hll1/y0ksvVUbIQjy2tFotycnJzJs3jyFDhtw3samKcvP1/HAxif/+fovI80loCwrHGtlaqQnt4s9r3QJwKEXri7OdFc52cqMT4kln1m6psLAwRo8eTZs2bWjXrh3Lli0jOzvb+PTUqFGj8PHxYdGiRQAcOXKEuLg4goKCiIuLY968eRgMBqZNm2bOyxDC7D777DNeffVVgoKC2Lhxo7nDqTQ37+SwKeo6nx2NJSPvf2OH6rnb83yLWrzSvi5ezmVvhRVCPNnMmtwMHTqU5ORk5syZQ0JCAkFBQezZs8f4V2dsbKxJl0NeXh6zZs3i6tWrODg40LdvXzZt2oSLS7lOnSfEE2fMmDHGsUtVnaIo/Hr9DusPx7DnTILx8WwfFzueb1mLfi28aertJPPMCFGNmX1A8aRJk0rshrq7Hs9d3bp149y5Cn5MQQjxWLpxO4evT8ax80Qc0cnZxu2dAmswrlM9ejSsiVrmmhFC8BgkN0IIUZIsbQHfnLrFjuM3OXbtjnG7jaWaAa18GNPJj0Ze9x9YKISofiS5Kcb9ZrkVoip4nL/jiqJwOi6dz47G8vXJW+TkF85jpVJBx4Aa9A/yoU8zLxxtZeCvEKJ4ktzcw9raGrVaza1bt/Dw8MDa2rpM/fZ3J13Ly8uTx5MrmNT1w1EUhfz8fJKTk1Gr1VhbW5s7JKMbt3PYezaBnSfiOHvrf/NZ+HvYM7SNLy8G+cjgYCFEqUhycw+1Wk29evWIj4/n1q3SrSd1L0VRyM3Nxc7OPHNkVCdS149Go9FQp04dsyeGV5Oz+PZ0PN+dSTBJaKwt1fRt5sXwdnVoV89NfsZCiDKR5OZPrK2tqVOnDgUFBfdd1qE4Op2OgwcP0rVrV5kUqoJJXT88CwsLLC0tzZow5On0LI24xEeHrhqXT1OroH29GvRp5sULLb1xtX98WpWEEE8WSW6KoVKpsLKyKvNN08LCgoKCAmxtbeWGW8Gkrh9viqJwKTELR1tLvF1MF549E5dO2OcnuZSYBUCX+u4836IWvRp7UsPBxhzhCiGqGEluhBDlKiNPR/j23/nuTAIA/u72dAp0p1OgOxcTMlnx/WUKDAruDjYsGtic3k2q12zKQoiKJ8mNEKLcnIlL5y9bjnM9NQdLtQoFuJqSzdWUbDb9ct1Y7rnmtVjQvxlu0vUkhKgAktwIIR6ZoihsORrL/G/OkV9gwMfFjlWvtMbfw54jV29z+EoKhy4no9Mr/O2ZBrzQ0lsGCQshKowkN0KIMjkTl866wzHcyc4nV6cnV2cgM0/H1T9mDe7VuCaLh7TERVPYKtO7iad0PQkhKpUkN0KIUtEWGFj+/UU++DEa/d0Fne5hoVYR3qchoV38pVVGCGFWktwIIR4oNgsGfBDF5aTC1pm+zb3o3rAmdlYWhS9rC/zc7fH505NRQghhDpLcCCGKVaA38Ov1O3xzMo7PTltgIJsa9tYs6N+Mvs1rmTs8IYQokSQ3QgijAr2BvWcT2X8+ke8vJJGeq/vjExXPN/finf7N5QknIcRjT5IbIQRQOGtw6MZfOXQ5xbjNRWNF9/rueOXf5G8vtZAJE4UQTwRJboQQ5ObrGb/xGIevpKKxtuCV9nXo3cSL1nVcUAx6vv32hrlDFEKIUpPkRohqLltbwKufHOOXq7ext7Zg/dh2tKvnZvxcZyjbGmtCCGFuktwIUY1laQsYt/4YR6/dxsHGkk/GtSW4rtuDdxRCiMeYJDdCVFNJmXlM3PQbx2PTcLSxZOOr7WhVx9XcYQkhxCOT5EaIaigqOpW/bj1BcqYWJ1tLNr3anpa+LuYOSwghyoUkN0JUIwaDwgc/RrNk30UMCjTwdGD1K60JrOlo7tCEEKLcSHIjRDVxJzufydtO8uOlZAAGtvbh7/2bobGWXwNCiKpFfqsJUQ2kZmkZ/tEvXErMwsZSzYIXmzGkTW1ZA0oIUSVJciNEFXcnO59XPj7CpcQsPJ1sWD+mHU28ncwdlhBCVBhJboSowtJzdIxcd4QLCZl4ONqwJfQpAjwczB2WEEJUKLW5AxBCVIyMPB2j1h3hTFwGNeyt2TK+vSQ2QohqQZIbIaqgjDwdY9cf49TNdFw1VmwObU99T3kiSghRPUi3lBBVTEqWltHrjnL2VgbOdlZ8Or49jbxkjI0QovqQ5EaIKiQuLZeRHx/hako2Neyt+WRcO5p6O5s7LCGEqFSS3AhRRVxJymLk2iPEp+fh42LHplfb4S9jbIQQ1ZAkN0JUAb/fTGPM+mPczs4nwMOeT8e3p5aznbnDEkIIszD7gOJVq1bh5+eHra0t7du35+jRo/ctv2zZMho2bIidnR2+vr5MmTKFvLy8SopWiMfPzhM3GbImitvZ+bSo7cwXEztKYiOEqNbM2nKzbds2wsLCWLNmDe3bt2fZsmWEhIRw8eJFatasWaT8li1bmD59OuvWraNjx45cunSJMWPGoFKpWLp0qRmuQAjz0ekNLPz2POsPXwOge0MPVgxvhaOtlXkDE0IIMzNry83SpUsJDQ1l7NixNGnShDVr1qDRaFi3bl2x5X/++Wc6derEyy+/jJ+fH8888wzDhw9/YGuPEFVNSpaWER8fMSY2bz4dyNrRbSWxEUIIzNhyk5+fz2+//caMGTOM29RqNb169SIqKqrYfTp27Minn37K0aNHadeuHVevXuXbb79l5MiRJZ5Hq9Wi1WqN7zMyMgDQ6XTodLpyuhqMx7z3v6LiVOe6jkvLZdhHR0nI0GJvbcE/BzXjmSaeGPQFGPTlf77qXNeVTeq68khdV57yquuy7G+25CYlJQW9Xo+np6fJdk9PTy5cuFDsPi+//DIpKSl07twZRVEoKChg4sSJvP322yWeZ9GiRcyfP7/I9n379qHRaB7tIkoQERFRIccVRVXHut54WU1ChhoPW4XxDbUUXPuNb69V/HmrY12bi9R15ZG6rjyPWtc5OTmlLvtEPS114MABFi5cyOrVq2nfvj1XrlzhrbfeYsGCBcyePbvYfWbMmEFYWJjxfUZGBr6+vjzzzDM4OZXvxGY6nY6IiAh69+6NlZV0D1Sk6lrXlxOzOP7LzwD8Z0wHmvlU/OR81bWuzUHquvJIXVee8qrruz0vpWG25Mbd3R0LCwsSExNNticmJuLl5VXsPrNnz2bkyJGMHz8egObNm5Odnc2ECROYOXMmanXRIUQ2NjbY2NgU2W5lZVVhX+iKPLYwVd3q+t8/XEVRoE9TL1r51ajUc1e3ujYnqevKI3VdeR61rsuyr9kGFFtbWxMcHExkZKRxm8FgIDIykg4dOhS7T05OTpEExsLCAgBFUSouWCEeA6dvprPnbAIqFYQ908Dc4QghxGPLrN1SYWFhjB49mjZt2tCuXTuWLVtGdnY2Y8eOBWDUqFH4+PiwaNEiAPr168fSpUtp1aqVsVtq9uzZ9OvXz5jkCFFVLY24CMCLLb1pIItgCiFEicya3AwdOpTk5GTmzJlDQkICQUFB7NmzxzjIODY21qSlZtasWahUKmbNmkVcXBweHh7069ePd99911yXIESl+O36bX64mIyFWsXkXtJqI4QQ92P2AcWTJk1i0qRJxX524MABk/eWlpbMnTuXuXPnVkJkQjw+Fu+9BMCQ4Nr4udubORohhHi8mX35BSHE/f18JYWoq6lYW6h5s2d9c4cjhBCPPbO33AghiqfTG/jm1C2WRhS22rzcvg4+LrJmlBBCPIgkN0I8ZvJ0er749QYfHrzKzTu5ANR0tOGNHgFmjkwIIZ4MktwI8RiJPJ9I+JenSckqXDKkhr014zrXY2SHujjJulFCCFEqktwI8RhQFIXVB6JZvO8iigI+Lna81s2fl9r4Ymsl0xwIIURZSHIjhJnl5Bfwf9t/Z/fv8QCMfKous59vgrWljPcXQoiHIcmNEJUoIT2PnPwC4/ssbQHTvzzNufgMLNUq3nmxGS+3r2PGCIUQ4sknyY0QlUBRFP6++zxrf4op9vMa9tZ8MCKYdvXcKjkyIYSoeiS5EaISrP0pxpjYONma/rNr5uPM+0NaymPeQghRTiS5EaKC7TmTwLvfngdg1nONGd/F38wRCSFE1SYjFoWoQCdi7zB52wkUpXCg8Kud65k7JCGEqPIkuRGigsSm5jD+k1/J0xno0dCDuf2aoFKpzB2WEEJUeZLcCFEB7mTnM3bDUVKz82nq7cTKl1tjaSH/3IQQojLIb1shyllOfgHjPjlGdHI2tZxtWTemLfY2MrxNCCEqiyQ3QpQjnd7AG5uPcyI2DWc7KzaOa4enk625wxJCiGpFkhshyonBoBC+/XcOXEzG1krNujFtqe/paO6whBCi2pHkRohysui78+w4EYeFWsXqV1oTXNfV3CEJIUS1JMmNEOXgwx+j+ehQ4SR9/xzUgqcbeZo5IiGEqL4kuRHiEW07Fsui7y4A8HbfRgwKrm3miIQQonqT5EaIR7DnTDwzdpwGYGK3ACZ0DTBzREIIISS5EeIh/XQ5hb9+dhKDAsPb+RLep6G5QxJCCIEkN0I8lBOxd5iw6Vfy9Qb6Nvfi7/2by+zDQgjxmJDkRogySkjPY+yGY+Tk6+lS351/DQ3CQi2JjRBCPC4kuRGijD44cIW0HB1NvZ1YMyIYG0sLc4ckhBDiHpLcCFEGiRl5fHbsBgAz+zaWZRWEEOIxJMmNEGWw5sdo8gsMtPVzpUNADXOHI4QQohiS3AhRSkmZeWw5EgvAX3vWlwHEQgjxmJLkRohS+ujgVbQFBlrVcaFzoLu5wxFCCFECSW6EKIWULC2f/iKtNkII8SSQ5EaIUvjo0FVydXpa1HamewMPc4cjhBDiPiS5EeIBbmfnsynqOgBvSauNEEI89h6L5GbVqlX4+flha2tL+/btOXr0aIllu3fvjkqlKvJ67rnnKjFiUZ385+BVcvL1NPV24ulGNc0djhBCiAcwe3Kzbds2wsLCmDt3LsePH6dly5aEhISQlJRUbPkdO3YQHx9vfJ05cwYLCwuGDBlSyZGL6uDnKyn852A0IGNthBDiSWH25Gbp0qWEhoYyduxYmjRpwpo1a9BoNKxbt67Y8m5ubnh5eRlfERERaDQaSW5EubtxO4e/bDmOQYGBrX14pomnuUMSQghRCmZNbvLz8/ntt9/o1auXcZtaraZXr15ERUWV6hhr165l2LBh2NvbV1SYohrKzdfz2qbfuJOjo7mPMwsHyMKYQgjxpDDr3PEpKSno9Xo8PU3/Ivb09OTChQsP3P/o0aOcOXOGtWvXllhGq9Wi1WqN7zMyMgDQ6XTodLqHjLx4d49X3scVRVVkXSuKwrTtpzkXn4GbvRWrhrfEAgM6naHcz/UkkO915ZG6rjxS15WnvOq6LPs/0QvjrF27lubNm9OuXbsSyyxatIj58+cX2b5v3z40Gk2FxBUREVEhxxVFVURd/3BLxTfXLVCrFF6pm8uJw99zotzP8uSR73XlkbquPFLXledR6zonJ6fUZc2a3Li7u2NhYUFiYqLJ9sTERLy8vO67b3Z2Nlu3buWdd965b7kZM2YQFhZmfJ+RkYGvry/PPPMMTk5ODx98MXQ6HREREfTu3RsrK6tyPbYwVVF1/fvNdHb9cgQoXBhz1FN1yu3YTyr5XlceqevKI3Vdecqrru/2vJSGWZMba2trgoODiYyMpH///gAYDAYiIyOZNGnSfff94osv0Gq1jBgx4r7lbGxssLGxKbLdysqqwr7QFXlsYaq863r1jzEYFHiuRS3GdfaXcTb3kO915ZG6rjxS15XnUeu6LPuavVsqLCyM0aNH06ZNG9q1a8eyZcvIzs5m7NixAIwaNQofHx8WLVpkst/atWvp378/NWrIysyifFxIyCDyQhIqFfytdwNJbIQQ4gll9uRm6NChJCcnM2fOHBISEggKCmLPnj3GQcaxsbGo1aYPdV28eJGffvqJffv2mSNkUUWtOVA4n03fZrXw93AwczRCCCEeltmTG4BJkyaV2A114MCBItsaNmyIoigVHJWoTm7czuGb3+MBeL17gJmjEUII8SjMPomfEI+DbS8JpwAAIABJREFU/xy8it6g0KW+O818nM0djhBCiEcgyY2o9pIztXz+6w0A3ugeaOZohBBCPCpJbkS1t+5wDNoCA0G+Ljzl72bucIQQQjwiSW5EtZaRp+PTqOsAvNE9QJ6QEkKIKkCSG1GtffrLdTK1BdSv6UCvxrIwphBCVAWS3Ihq6052Ph8figFgYrcA1GpptRFCiKpAkhtRbf3juwvczs6ngacDLwR5mzscIYQQ5USSG1EtHY25zbY/npBaOKA5VhbyT0EIIaoK+Y0uqp38AgMzd54GYFhbX9r4yRNSQghRlUhyI6qdjw5d5XJSFjXsrZn+bCNzhyOEEKKcSXIjqpXY1Bz+HXkZgFnPN8ZFY23miIQQQpQ3SW5EtaEoCrO/PoO2wECnwBr0D/Ixd0hCCCEqgCQ3otrYezaBHy8lY22hZsGLzWTCPiGEqKIkuRHVgsGgsGx/YXfUhK7++Hs4mDkiIYQQFUWSG1EtRF5I4kJCJg42loR28Td3OEIIISqQJDeiylMUhZXfF7bajOxQF2eNlZkjEkIIUZEkuRFV3qHLKZy6mY6tlZpXO9czdzhCCCEqmCQ3ospb+f0VAF5uVxd3BxszRyOEEKKilTm58fPz45133iE2NrYi4hGiXB25msrRa7extlAzoauMtRFCiOqgzMnN5MmT2bFjB/7+/vTu3ZutW7ei1WorIjYhHtnKHwpbbYa0qY2Xs62ZoxFC/D979x0eVZU+cPw7LQkJBBJCAoQWOqF3UEFAyoqCoChWiq66CGtBdxU76ooFlbWi/gRRRFhREQslBpEOQgi9t0AgFUhCAsmU+/vjZDIJk4SZMJlJeT/Pk2du7tx758xldN6c8573COENZQpu4uPj2bJlC+3ateOf//wnDRo0YMqUKcTFxZVHG4Uok/iT51l7KA2DXsc/rm/h6+YIIYTwkjLn3HTr1o3333+f06dP89JLL/F///d/9OzZky5dujBnzhw0TfNkO4Vwi6ZpBcssjO4aSePQQB+3SAghhLcYy3qi2Wzmxx9/ZO7cucTExNCnTx8eeOABTp06xbPPPsvvv//OggULPNlWIVyiaRpvLN/Pqv0p6HUwaYD02gghRHXidnATFxfH3Llz+fbbb9Hr9YwbN4733nuPtm0dqyuPHj2anj17erShQrhC0zTeXH6AT/88CsD0ke1pIdWIhRCiWnE7uOnZsydDhgzhk08+YdSoUZhMzgXRoqKiuPPOOz3SQCFcZQ9sZv95BIBXbmnPfX2b+bZRQgghvM7t4Obo0aM0bdq01GOCgoKYO3dumRslhLs0TeOtFY7AZvrI9oyTwEYIIaoltxOKU1JS2Lx5s9P+zZs3s3XrVo80Sgh32HtsPlmtApuXR0Qz/ppmvm2UEEIIn3E7uJk8eTInT5502p+YmMjkyZM90ighXKVpGtN/3lvQY/PSiGgmXCtLLAghRHXm9rDU3r176datm9P+rl27snfvXo80SghX2DR48ed9LPzrFACvjurAfX1KHzIVQghR9bndc+Pv709ycrLT/jNnzmA0lnlmuRBusdo0vj2iZ+Ffp9Dp4K0xnSSwEUIIAZQhuBk6dCjTpk0jIyOjYN/58+d59tlnGTJkiEcbJ0RxbDaNpxbvYkuqHoNex6yxXbijR2NfN0sIIUQF4XZwM3PmTE6ePEnTpk0ZOHAgAwcOJCoqiqSkJN555x23G/DRRx/RrFkzAgIC6N27N1u2bCn1+PPnzzN58mQaNGiAv78/rVu35rfffnP7dUXl9eWG4/yyKwmDTuO/d3Tili6Rvm6SEEKICsTtcaTIyEh27tzJN998w44dO6hRowYTJ07krrvuKrbmTWkWLVrE1KlTmT17Nr1792bWrFkMGzaMAwcOEB4e7nR8Xl4eQ4YMITw8nMWLFxMZGcmJEyeoU6eOu29DVFKHkrN4Y/l+AEY3szGsfYSPWySEEKKiKVOSTFBQEA899NBVv/i7777Lgw8+yMSJEwGYPXs2v/76K3PmzOGZZ55xOn7OnDmcPXuWDRs2FARSzZo1u+p2iMohz2Lj8UXx5Fls9G9Vl+vqOud+CSGEEGXOAN67dy8JCQnk5eUV2T9y5EiXzs/Ly2Pbtm1MmzatYJ9er2fw4MFs3Lix2HOWLl1K3759mTx5Mj/99BP16tXj7rvv5umnn8ZgMJT1rYhK4r+xB9lzOpM6gSZmjO7A1rUS3AghhHBWpgrFo0ePZteuXeh0uoLVv3U6HQBWq9Wl66SlpWG1WomIKDqsEBERwf79+0t87VWrVnHPPffw22+/cfjwYR555BHMZjMvvfRSsefk5uaSm5tb8HtmZiagFv40m80utdVV9ut5+roC4hLOFxTpe3VkNCEBKl1M7nX5k8+198i99h65197jqXvtzvluBzePPfYYUVFRxMbGEhUVxZYtW0hPT+fJJ59k5syZ7l7OLTabjfDwcD777DMMBgPdu3cnMTGRt99+u8TgZsaMGUyfPt1p/8qVKwkMDCyXdsbExJTLdaurXCu8ucOATdPRs54N24ltxJxQz8m99h65194j99p75F57z9Xe65ycHJePdTu42bhxI6tWrSIsLAy9Xo9er+e6665jxowZPProo2zfvt2l64SFhWEwGJxq5iQnJ1O/fv1iz2nQoAEmk6nIEFS7du1ISkoiLy8PPz8/p3OmTZvG1KlTC37PzMykcePGDB06lODgYJfa6iqz2UxMTAxDhgxxO7lalGzGsgOk556gYe0APn2oL7UCTHKvvUjutffIvfYeudfe46l7bR95cYXbwY3VaqVWrVqAClBOnz5NmzZtaNq0KQcOHHD5On5+fnTv3p3Y2FhGjRoFqJ6Z2NhYpkyZUuw51157LQsWLMBms6HXq2GJgwcP0qBBg2IDG1BFB/39/Z32m0ymcvtAl+e1q5tLZiuL4xIBVYE4tFbR3ja5194j99p75F57j9xr77nae+3OuW7XuenQoQM7duwAoHfv3rz11lusX7+eV155hebNm7t1ralTp/L5558zb9489u3bx6RJk8jOzi6YPTVu3LgiCceTJk3i7NmzPPbYYxw8eJBff/2V119/Xda0qsJi9iaTeclCZJ0aDGzjXB5ACCGEuJzbPTfPP/882dnZALzyyivcfPPN9OvXj7p167Jo0SK3rjV27FhSU1N58cUXSUpKokuXLixfvrwgyTghIaGghwagcePGrFixgieeeIJOnToRGRnJY489xtNPP+3u2xCVxHfb1LpRt3WLRK/X+bg1QgghKgO3g5thw4YVbLds2ZL9+/dz9uxZQkJCCmZMuWPKlCklDkOtXr3aaV/fvn3ZtGmT268jKp8zGRdZeygVgNu6N/Jxa4QQQlQWbg1Lmc1mjEYju3fvLrI/NDS0TIGNEKX5IS4RTYNeUaE0rRvk6+YIIYSoJNwKbkwmE02aNHG5lo0QZaVpGovzh6TGSK+NEEIIN7idUPzcc8/x7LPPcvbs2fJojxAAxCWc41haNoF+Bm7q2MDXzRFCCFGJuJ1z8+GHH3L48GEaNmxI06ZNCQoqOlwQFxfnscaJqu+dlQf4IS6RN2/rxHWtwgr2f7dV9drc2KEBQf5lXiVECCFENeT2t4a9Jo0QV+uS2coX646Rk2dlwtwtvHFbJ8Z0b8TFPCu/7DwDwO09ZEhKCCGEe9wObkpa5kAId204kkZOnhWdDiw2jae+28Gpczk0rRvIhVwLjUNr0KtZqK+bKYQQopKR/n7hMyv3qKU37undhFoBJj5ZfYRZvx8i0E8trzGmW2OpbSOEEMJtbgc3er2+1GnfMpNKuMJq0/h9nwpu/ta+Ade1CqNxSCAv/LSbnDz1Gbq1W6QvmyiEEKKScju4+fHHH4v8bjab2b59O/PmzSt29W0hirM94RxpF/KoFWCkd3M19HR37yY0qBPA4wvjGdCmHo1Dy2fVdiGEEFWb28HNLbfc4rRvzJgxtG/fnkWLFvHAAw94pGGiaovZq3ptbmgbjsngqEgwsE04254fjNHgdpUCIYQQAihDnZuS9OnTh9jYWE9dTlRhmqaxYk8SAEPb13d6XgIbIYQQV8Mj3yIXL17k/fffJzJSciTElR1OucDx9Bz8jHr6t67n6+YIIYSoYtwelrp8gUxN08jKyiIwMJD58+d7tHGialqZPyR1XcswakqBPiGEEB7m9jfLe++9VyS40ev11KtXj969exMSEuLRxomqaWX+kNSQ6Agft0QIIURV5HZwM2HChHJohqguzmRcZMepDHQ6uKFduK+bI4QQogpyO+dm7ty5fPfdd077v/vuO+bNm+eRRomq6/f8IaluTUIIrxXg49YIIYSoitwObmbMmEFYWJjT/vDwcF5//XWPNEpUXfZ8m6EyJCWEEKKcuB3cJCQkEBUV5bS/adOmJCQkeKRRomrKuGhm45F0oPgp4EIIIYQnuB3chIeHs3PnTqf9O3bsoG7duh5plKiaYvclY7FptAyvSVRYkK+bI4QQoopyO7i56667ePTRR/njjz+wWq1YrVZWrVrFY489xp133lkebRRVxE/xpwG4qWMDH7dECCFEVeb2bKlXX32V48ePc8MNN2A0qtNtNhvjxo2TnBtRorQLuaw7nAbAqK5S7FEIIUT5cTu48fPzY9GiRbz22mvEx8dTo0YNOnbsSNOmTcujfaKK+HXnGaw2jc6NasuQlBBCiHJV5vKwrVq1olWrVp5si6jClsQnAnBLF+m1EUIIUb7czrm57bbbePPNN532v/XWW9x+++0eaZSoWk6kZ7M94Tx6HdzcWfJthBBepmlwbA1cyij9uJN/QXZa6cecjoeMU+69/sGV8E5b2DrXvfO8yWaFI3+AJdfXLfEIt4ObNWvWMHz4cKf9N954I2vWrPFIo0TVYk8kvrZlmBTuE0J434b3Yd4IWPZ0ycccXw9fDIbvJpR8TMp++HwQzL0RrBbXXtuSC789CVln4JfHYcdCt5ru5OJ511/bHX+8Dl+PgnXvef7aPuB2cHPhwgX8/Pyc9ptMJjIzMz3SKFF1aJomQ1JCCN/JSoI/31LbB5apHori7P9VPR5fC5mniz9m7xLQrHA+AQ6tdO31t85Vx+sM6vclj8C+n11vf2GHfoeZreDHh8p2fknMF2HrF/mv4eL7KkluFiy4E/YsAZvt6ttWRm4HNx07dmTRokVO+xcuXEh0dLRHGiWqjj2nMzmamo2/Uc+w9lKVWAjhZb9Ph7wLavvSeTgTX/xxR/9wbB/4rfhj9v/i2N725ZVf+1ImrMkPrG6aCV3uUcHR4vvhyKorn19Yxin44UGw5sHuH+DsUffOL83u7+HiObV9Oh5yL1zdtQ4ug1WvQqFFtr3N7YTiF154gVtvvZUjR44waNAgAGJjY1mwYAGLFy/2eANF5bZku+q1GdwugloBJh+3RgjhE5cyIfYVaNIHOo7x3uue2gY7FqjtsNaQdlDllUR2L3pcVhKk7HX8vu8X6Pn3osecOw5Ju0CnB80Gh2NUwFG7Ucmvv+EDyEmHuq2g6zj1k5sF+5bCwnvgvh/VPbkSq1kFRBfP5u/Q4K8vYNh/rnzulWgabPms0O9WOLkZWt5QtuvZg75u430a3LjdczNixAiWLFnC4cOHeeSRR3jyySdJTExk1apVtGzZsjzaKCopq01j6Q7VvXtLl4Y+bo0Qwmf+eB3++hy+fwDivvbOa9pssOzfarvzXdArfyjn6GrnY+37auVPeDi+VuW2FLY/vzenyTXQrJ8KcLbPL/n1s5Jh44dq+4YXwWBUP7f9H7S4Acw5sGCs8+sUJ/YVFXD4B8Pf8if0bJ8PeTlXPvdKTm2FMzvA4A+thqp9JzaU7VpndsDp7aA3QZe7r75tV8Ht4AbgpptuYv369WRnZ3P06FHuuOMOnnrqKTp37uzp9olKbNPRdFKycqldw8SANuG+bo4QwhdS9hftGfj5UdjzY/m/7s5FkLgV/GrC4JehhRppIGET5GUXPfZI/pBU5zuhXluwWZxzT+w5Oe1uVr0SoAK1knJ4/nxTBTCRPaDdCMd+oz+MnQ+hLdQw2YFlpb+PA8tVQjTALR9CrwehThN17u7vSz/XFX99rh47jnG0s6TgRtPU/TNfKv75bfPUY7sREOS8wLY3lSm4ATVravz48TRs2JB33nmHQYMGsWnTJk+2TVRyP+UnEg/vWB8/Y5k/akKIiuxCKvzwMBz+3fk5TYPlz6ihjjbDVVCg2eD7B+FQjOM4mw2O/glLHy29N8RVuVnw+8tqu/9TUKs+hDaH2k3AZi765a1pjp6b5gOh7U1qu3B+TXYaJOSf02a4+vKuEQKZp+BwrPPrpx12DM8Mme48POMXCB3zS6fsW1ry+zh/En58WG33/gdE3wJ6g2PIbMtnqv1ldSHVEWj2/Ds0vVZtJ24tPoDZ8S3MGQb/u8/5dfOyYdd3arv7+LK3yUPc+sZJSkrijTfeoFWrVtx+++0EBweTm5vLkiVLeOONN+jZs2d5tVNUMmarjeW7kwAY2VlmSYkKyGqGLZ9D+hFft6RyW/MW7FyoZshc/kV/YJlK1DX4qfyQm9+D9reqAGPRvSq3Zd0s+LA7fDUS4ubBz4+VPFvJVWvfgQtJEBIFfR5R+3Q6aDFAbR8plDycul8da6yh8l/a3qz2H/pdzSICOLhcBWX1O0FIUzAFqKEucE4s1jSIna4CulbDoNl1xbcx+hb1eDhW5SQV56dHVA9Nw24w5FXH/q73gTEAknbCqb9cuSPFi5unEpQju0NkNxUA1oxQ+xK3Oh9v75k5tNK512jPj5Cbqe55s/5lb5OHuBzcjBgxgjZt2rBz505mzZrF6dOn+eCDDzzSiI8++ohmzZoREBBA79692bJlS4nHfvnll+h0uiI/AQFSO8Xbci1W8iwlT/P769hZMi9ZqBvkR6+oUC+2TJQoL1slRFYHmgaJ20qvB7J1Dvz2FHwxFFIPeq9tVcnFc7D9G7VtM6sk2YT8HnxLLqx4Vm33nay+OPUGuPUzaP03sFyCRffA7y+pmT9+tSA4Ug0JFR7GclfqAdiQn+sy7HU1DGTXfKB6LDQzSn9stdpoeo06tmFX1Q5ztupNAhWEgSPwAcfQ1MHlkHlGbeflqJ6WfUsBHQx+qeR2hrdTicbW3OKnX5/ZqQoP6o0wZg4YC5VgCQyFDrep7S2fl/wapbFaHEUF7flIOp26D+A8NHX2GJwsNDqz/BnHDCtwBD7dx4Pe9z31Lrdg2bJlPPDAA0yfPp2bbroJg8HgkQYsWrSIqVOn8tJLLxEXF0fnzp0ZNmwYKSkpJZ4THBzMmTNnCn5OnDjhkbYI12RdMnPjf9cycOZqcvKK//KI2ZcMwKC24Rj0vsuYF4UsexpmX+edfAdf2/K5KrYW+3LJx+z4Vj3mpKniZecTvNK0CiNhk+oluZqeq23zVBAQHg0th4DlInxzu5pOvOkTOHcMataHfk86zjGY4PYvIep69XtkDxj5ITx1AIa/rfZtneucF2NntZQ8FGOzwc+Pq0Cr1VBoc2PR55sPAHRqZlSW6lnWHcsPYFrkBz46XaGhqZ/VtGj7tO12hYKb8LbQpK/qoYmfD+dOqCGbnYtUTZvhb0NE+5LvnU7n6L3Zu8T5eXsuTLuREBrl/Lx9aGrvErhQ8vdliQ4uU8NqgXUhepRjv31o6sT6osfvyp8N3fQ6NfMsOxVi8oO35L1waosKxLrc435byoHLwc26devIysqie/fu9O7dmw8//JC0tCuUqXbBu+++y4MPPsjEiROJjo5m9uzZBAYGMmfOnBLP0el01K9fv+AnIkLqp3jT2ysOcDQ1m8TzF1m5J9npeU3T+D0/uBkcLf82FYLN5qjdsWm2b9viDdu/Uo/b5hU/oyT1oJrVoTOov54zE2HeyIIvvCpv/6/q/W77UlXuLUtgZzU7elj6ToY7vlIziXIzYf6tsCY/UBn8MvjXKnquqYaaBj11PzwYC93uA78g1aMTEqWGYuIXOL9mdjp80hc+7lP8Egjx81VujCkQhs90znUJDIUGauKL7vgadDYLuoSN6jl7rw44gpsDy9SUb2suhDRTQVxh9t6bLZ/DZwPUMFFgGIz7SSX+Xok9uDkUU7S2zMVzsDM/f6Wk60R2U4GhNU8NL4EK+pL3qmniR1eXHARaLY7/D3Qbp4bZ7OzBzckt6t/Yft2d+fXtut4LI/6rtuPmqR4e++u3GQ41K8bkEZfr3PTp04c+ffowa9YsFi1axJw5c5g6dSo2m42YmBgaN25MrVq1rnyhQvLy8ti2bRvTpk0r2KfX6xk8eDAbN24s8bwLFy7QtGlTbDYb3bp14/XXX6d9++Ij5NzcXHJzHWtl2Ksom81mzGazW+29Evv1PH3dimR7wnm+3uToKft+20lu6lD0w3wwOYuTZy/iZ9TTp1ntcrkf1eFee1TKPkw56Wr75CbMp3epWSEuqHT3Ov0QJvvwW24mlp2L0ez5Efn08QsxALYWg7De+C7Gr29Gd+4Y2lejsNy3VCWL+oA37rVu50IMvzyGTrOiGQPQZSaizRuJ5b6fVeKtq9fZ8z3GzES0oHAsbUeBzgR3fINh/ij0STsAsDXshjX6Vijp/dQIc3pO3/NhDCufQdv0MZYu41RdGQBNw/Dz4+jT1BCi9tUtqs1B9dTz2akYV76ADrD2fxpbzYbFvq4+6noMZ+LhyCpCc1ujM2ejBdXDEtrKcXzDXhgD6qDLSUeLfVVds/WN2CyX9VS3vgljQG10F9Qfc7YGXbCOmZc/rOXCv2HdthhDotCdO4blwHK0dirY0W/7GoPlIlp4NJYGPUq8lq77/RgTt6Jt+RwtZT+6Y2vQZTt6cWxN+mIb+AJao14F91C3/2cMf76OLv0wmk6v7nHh64e0wFgjBN3Fc1hObkWL7AFn4jGlH0Iz1sDSchj418LQ5V708fPRlj4K2SnoAEvne9GKaaunPtfunK/TtLKnWh84cIAvvviCr7/+mvPnzzNkyBCWLi0l8/syp0+fJjIykg0bNtC3b9+C/f/+97/5888/2bx5s9M5Gzdu5NChQ3Tq1ImMjAxmzpzJmjVr2LNnD40aORdTevnll5k+fbrT/gULFhAYGOhyWwVYbPD2TgNJF3W0qW3jQIYeHRrTu1upXWg4eOUpHb+eNBBdx8bD7XxXfls4RKWupNMpxyyUI/WGsbuR57qPG53dQFBuMgfq3+L4MvKR1meW0C7pB2w6A3rNSnpQK9a1fsFxgKYxeO9TBOWlsrXZIySG9CEwN4V+B18jwHKe8zWacLTeMFJrRXPJr67v3kg5aJ6ynI6JqkckIbQf+xuM5tpDMwjKSyUzoBHrWj2L2VjzyhfSNPoffJmQnGPsq38rBxs4hjX8LFlcc+gNgvJSWN/yGc4HtXCrjQbrJYbueRw/aw6bmj9Bcu2uAESe3UCPE7OxYSDXFEwN8znO12jC+pbTsBiD6Hb8Exqf28j5Gk1Z0+ZlNF3xqRNhWXu49vCbXDLWIaFuP1on/8zJkL7ENZtU5Lhuxz+l8TnH0MzaVs9xtmYbp+u1OfMjbZN+JCG0Hzsaj8emd16eqDTRiYtolfIriXV6sTVqCmg2btj7b2rmpRDfeAInwgaVeK7elsfQPU/gb8kq2GfR+XE+KIqQ7KMYNBUMnAnuSmJIb1qkriAk5xgAuYaa7Im8m5N1nROeex2dRYOMOPY0HMvhiJvocOobWqSu4FSdPmyLUgnaJks2g/Y9TYBFdRjk+IUREz2zXP/7z8nJ4e677yYjI4Pg4OBSj72q4MbOarXy888/M2fOnHIPbi5nNptp164dd911F6+++qrT88X13DRu3Ji0tLQr3hx3mc1mYmJiGDJkCCZT1avG+9Hqo8yKPUxokInlj17Lw/O3s/1kBtP+1pr7r21WcNyYTzez41QGr46M5s6epVTvvApV/V57mmHxBPQHfsHWpC/6hI1oAXWwPLZbzbi4give66wkjB92QWezYL1xJrZuEzz/BlylaRg/vQZd+iGsg15E/8d/0GlWzA+uVQmcgO7UFozzhqP5BWF5fJ8awgBI3Y/x65HoCqrAghbaAlvU9dh6PazqkpSz8vxc69e8iWGtGiqy9p6E7Ybp6ovo3HGMX92E7kKy6mm5+3vnYaTL6E5uwvjVzWgGfyz/3OFc00SzqeJ0gWWbTKBfNR3Dxg+wNb0W670/QeYZjJ9fh+5SBtb+z2BrPxrjVyPQZadga9QLW+/JGL8fj6bTY52wAq1h15IvbrmE8Z2W6CyXuGQMJsCSieXmD5x693T7f8H4/QT1dgLDsDy2RyVEX07TVOXgwLIFwrrT2zHOHYJmCsLyxD50CRsxLhyL5h+M5dGdqk5PaefvXaJ6UBp2Q4vqjxbZUyVGZyZiWPs2uh0L0GmOPzI1UxC23pOw9Zlc4r+zfvMnGH5/AVvLIVhv/xrj+53QZadguWMBmr3QH/m9d0vUVHXr9c9iu25qsdfz1Oc6MzOTsLAwl4Ibt5dfKI7BYGDUqFGMGjXqygcXEhYWhsFgIDm5aN5GcnIy9eu71j1qMpno2rUrhw8fLvZ5f39//P39nfabTKZy+1Isz2v7ypHUC3y8Wq1l8tKI9oTXDuLW7o3ZfjKDJTuSeHhAKwBSsi6x41QGAEM7NCj3+1AV77XH2WyQoP4C1Q9+GRY/gC7zFKZDy6DTHS5fpsR7vXOBmuECGFa/jqHTmPId1rFa1PTdiPbOORVJuyH9EBj8MfR6EE7Hwf5fMO38Bm7Mr+y6R01h1bW7BVNgbce5DTvCg6tUnZWjq+F0HLqzRzCcPYLh4DKYshX8XejV8ACPf65PbIT8wIZBL2Do9yQG+70Lb6VyROYOR386Dv23Y1Qy7OVLFBS2ReVr6DqPxVSnQfHH+F1Fvl2ff8Cmj9GfWI8+dY+aWn0pAxp2w3D9vzAYjCpn58vh6E9tQX9KzbDV9XoIY9NepV/bZFIzgo6sKuh1MLYerPblit5mAAAgAElEQVQX1maoCv4tl9C1HY7Jv5Q/BPxcH85z0qQn1G6CLiMB04k1BXV+dF3uwRTkwn9HnW9XP5er2wxGfQTXPQ6rXlNTzrvei67fkxhq1qPUKUHN+wGgP7kZ/Ym1kJ0CgXUxthmqEsILXnusmnl26i8MPSdiuMJn9mo/1+6c69P+Yz8/P7p3705srKM2gs1mIzY2tkhPTmmsViu7du2iQYMS/gMTV81m05j2wy7yrDaub12PkZ3VUgojOjXAZNCx70wm+5PU/yRW7VPjvZ0b1SYiWKbou+1CKix7RhUB85SUvSpB0RSkvrC6jVP77VM3r4bVDNvyp5OaAtVfsPYVmMvL7y/B7GthzUzn5+y1N1oNgYBg6D5B/b5joapZYsmDPT+ofcUFdqFRcMMLKsn138dg7DeqGmzWGVj3Xrm8Ha+wLwPQ9T5V1O7yoDC8Hdz3gyrvn7hNzTRbdF/xU+TPHnVU67XXkPG02o2g/Wi1veheNVvJGACjP1VLGADU7wD3/qA+16DyXAY979r1CyUPa2GtIbiY5WH8gqDDGNW71bkclxLQ6SB6pNre+BEcXKG2L1/bqqzCWsEd82DaSbjxDahZ78rnRHRUU/NzM1VgCWrqueGy4EKng1s/hUfjKkwisZ3PJ6NPnTqVzz//nHnz5rFv3z4mTZpEdnY2EydOBGDcuHFFEo5feeUVVq5cydGjR4mLi+Pee+/lxIkT/P3vHvogiCISz1/kqcU72HLsLDVMBl4b1QFd/v8Y6wT6Mait+kD/GKeqERfMkmons6TKJPZl2PwJLH/ac9c8vk49Numj/ufU9V71P+wT6yDt0NVde/+v6os/qJ6a3gtqBk3qgSufa7Op9XrOuVHK4UIq/PV/anvNW0W/fDXNEdzYa4C0GAS1G6vZN3uXqiq6F8+p6clRVyg0VqOOmvo77HX1+4YPVK2PyqZwMHLNP0s+rmFXmLQ+/4tcp2q1fNwbljyi3rv9Z9nTgKbWR8of6isXffMDp4yT6nHwy1CvddFjGvWAexerWVZj5l5xOK1AC0dwY4saUPJxN78HT+yFpq79sV1m9llTCRsBTQVfYR5eq9GdRSwNRmjSW20n71aPncZ6tj3lzOfBzdixY5k5cyYvvvgiXbp0IT4+nuXLlxdM705ISODMmTMFx587d44HH3yQdu3aMXz4cDIzM9mwYQPR0dElvYQog+TMS7z4024Gvr2aH/IDl2eHt6VxaNEk7NFdVU7NkvhELuRaWHtIlQeQKeBlkHPWUUvi6Go17dUTjq9Vj1Gqq5nakY4F8uKusvfGXkCs23hoPQza3KSGqJY/U3pZeHtZ/oV3qV6CzDMlH1vY5k9U8TdQU2B/eVwFSaCGoM6fUH/Jtx6m9ukNjp6quHmO6awdxxSfP1GctjermizWXIh54crHlyQrSS2m6G2bPwU0VYemnnNCbBF1msDoT2DSBvVvqdkg/htY+bzjx15wru/k8m13ZHdVRwbUQpW9Hi7+uKbXwN2LHF/Grghvjxak/jDTSgtujH4Q7IVRgcgeUKtQ75G9qJ4v2aeEgyrAWNowZQXk8+AGYMqUKZw4cYLc3Fw2b95M796OD+nq1av58ssvC35/7733Co5NSkri119/pWvXUpLHhFs0TeOdlQfo/9YffLXxBHlWG32b1+W7f/Tlvr7NnI4f2LYetWuYSM7MZeaKA+RabETWqUHb+u6VBRCoLxH7F7fNUvqaM66y2Rw9N836Ofbb63PEL1CVZMsiea/q/dEZoIfqaWXYa6rU/pFVqnJrSf74D2z5VG3npMHi+0uvJgyqRP2W/F6bYTPUMNiJ9aq2CcDu/OGmNn9TQwp2Xe7J76la7+jBcOevUJ0O/vaGep/7fnZUrXWV1aKGtGZ1gv92gvXvl7zYoqddPO9YhbuvG0NIEdFw1wJ4IEYN7XW6s+jP4OmOhSjL04j3VW/TmDmerXqr12Md9Sl7Go5FaznYc9e9ivYUDE3VbuIIzn2pcHDTaax7PT8VgEcSikXVcTw9hw9WqXyPHk1DmDq0Nde0KHl1V3+jgZs7NeCbzQl8ueE4AEOiIwqGrqqtnLOqrH9IU7hn8ZX/x2CzqcJbAOHtIWWPGmKxBw1llbJHDcn41SwoXgaonptaDdSQ0s7/qfVkjq5WP+cTVAn1gc+pBf5KYh8eajtc5UiA+guv72T1Zb7iWfUFaLwsoX/9+44Cb/2ehM2fqcJrf/yn9HL1W+dAbgaEtVGLCGrW/N6EF9QaPvbgxj4kZVc7Uj1/cJmqXFuvHdTveKU7V1RENPR8QA25LX8GHl7ryP0oTco+NaxzOs6xL+YFFbje8rHzMIunxX3lqCBcuEidqxr3Uj++Uq81DH2tXC6tNevH4YgsWvu4dEGBvpPV56XPJNd7FctTw64QUEctQtqxmITlCq6C/KuKiuJAkqqXEN0gmO/+0bfUwMbu1m5Fp3sPkSEpNTsh/ZDK8bCvOFyaI7GqVL1/bRiTH+QcX3f1FXOP5Q9JNelbNBnQYFS5NwBLp8CC22HTRyoYystSCaizr1OzbIpzKdMxxNPzsgqq/Z5UOS1nj6q1hrbPd1ST3TrXMbRzw0tww4sw8n31+7p3i64UXZj5Emz6WG1f97j6S7f3JBWkXDoP82+DrNPq/hX3l3jhVYo73V62v0IHTFOzwFL2OpKoS2LJUwnPn/ZXgY1/bbjlIxj5gUraPfWXur/rZl25x+pKbDZ1ry9fxdlqzh+SQiX+Vvc/OCq6Ok1g/FLnJSN8xegHE36Bib9B3fIvg+BpEtyIIg6nqOCmbf1aLve+dGtSh2Z11V/4tQKMslBm0u6iX372L+XS2HNXut6rkjQjewAa7P3p6tpSMCRVzMrE3cY5ZprUaaKGqsbMhTu+VuP/Z4/A3BvRxzyHwXbZ0NWOhZB3QfWiXJ6Y619LrQANqnT9T5Phvfbwflf45Qm1/7onoF9+TYwOtzoCpB8eKqGs/jdwIRmCG6kZLKACtBH/VUNOyfkVidvd7NxTBCrfpG4r1YNV1sTIwFDVmwWql2nXYpXgXFj6EYh5Ed5tB6teVXlBrf8Gkzepf9tu4+CRjSoAs+aqmV8L73asPu2qc8fV0gnfTYC3W6h7+0E3tZK13d6f1NpBQfUq5V/eogKo31FNRKiEZFhKFHEoRa1v0jLC9XoeOp2OMd0bMXPlQQa3i8BkqMYxsz1RVrOpdXYSNqoEzNQDJSdznj3mSNLs+YB67HAbJG5VQ1O9S0ikvBKbVeXEQNF8G7s6TWDKX2qoJqRZ0eei+sPK52D7fAxbPmWI4RsM5l/VMFPz6x2L+vV6sPgegY5j1DUPLlc9V4nbVO8CqCmuN1w2/DTsP6o340y8yr+5e5GjVo7VAhvye3eu+WfR1ZEju6vky8356+S0v7X4e2Ewwt9jVBBR3LRfV3WfqHqfUvbA9/n/VhEd1P1N2aNWcbarWR+GTHfOV6jdSA1Vxn8Dvz4Jh1aonqe7vgXDFaqma5paddrea1ZAp9bH+uY26HKvup8bP1JP9fx70bWDhKgGJLgRRRxMVsFNq3D3EoIfvr4FkSE1GNimYtU68Lp9S9XsJGMAjJ6t8k72/6JWSB4xq/hztn5BwdRae/dv+1Hq3JOb4fxJqNPY/bYk71aFz/xqFc23Kax2ZPH7a9RRwyjRo9F+fhT/zES1+nDh1Yuv1AvSqIf6GfS8asfxdWql5w5jnAMio7+aSv7p9eo9v9NW1TnpPkH15Jw7rirA2mc+FTboeRVU6A0q8CpJjZCrLy5oMKp6MBs+gGN/QtIudZ/t02XRqV6Z7hNUUujldUHsdDrVkxPaHBaMVcnOX94Md14etFxmwweOVacb91KrXDcfoHr7Vr+pegnj56tFUi+eBYM/9Hjg6t6zEJWQBDeigNWmcSRVBTet3ei5ATAZ9AXTwqst80WV4ApwzaMqmbjPIyq42fEtDHoBguo6n5NfkbTI6r/BDdVshRPrYM+PcO2j7rfHPiTVtK9rya/FaTUYyyNb2fT9h1xTPw/D8bWqh0Wzqi/wABeXMAmo7VhpuSShUWqGzrKnVbCw41v1Y18jqPek4hOc/WvBP9Z7djZNaWrVdwy7ZaepIOfEBjX80+Vu1SPmqqbXqLyGr2+FpJ0Yv7qZGg1KmGKdsBl+f1lt3zQTetxf9Pm/va5m3Cx5RA0pgipU6ErRNiGqmGo8fiAud/JsDnkWG/5GPY1CZFFRt234UM00Co5USa+gvrwadFFTvLfNcT5n9/eqqFydJo7aM3YdbnUcUxb2ZOLi8m3cYTBxtmYbbP2fhgdWwDMn1DIFg1++uusWp9l18I918PdVqpquKVAFUn61oFcphTq9FdhcLihMDSHe9A4MeMa9wMauQWe4fwXUbozu7BH6HXwV3ZFVRY/JTofFE9W96DBGDY8Vp0kfVYjvuifUUNn1/3a/PUJUARLciAL2fJsW9Wpi0MvMCrdkJKrZPgBDXnHUWdHpHMXOtnxetKZMdrojL6LHA87TP6NvUb0WZ+JVoqo7bFbVmwDF59tcDf9aKtelpCGXq6XTQaPucMuH8OQBGP0ZjP+pfNer8rWwlnD/CrSwNtQwn8O48A6ViH3xvJoN9eNDKqembks1vFlasr+phgo8J/xStmBLiCpAghtR4GCyminVys0hqWovN0vN8jHnqCnXl9dZiR6lZh9dSHbUYtmzBD7qpaYV+9VSvRSXCwpz5JDYz3PVsT9VTRj/YKjfyf33VFEEBKvF+SpZddQyqR2J5f4YjtQbhoZODVd+3FcFOYd/V3lct89zfYkBIaoxCW5EgcMp9mRiCW5clp0O80ao3Bi/WjB8ZjHJsn7QO7+c+ob34X/j4bvxqjJveDRM+Nk5F8fOHijtcSO4yc2Cn/OHxTqOKXu+jfA+UyC7G92DddzPENpC1e7ZsUA9d+NbarFIIcQVSXAjChxKsffcyF+GLsk4BXP/Bqe3q5k8E34u+cun+wSVP5KyV8040hmg/7/godWqEmhJ2t4MepM679RW19q14jm1xlLtJqpMvqh0tMZ9VO5R3ynqs9J9YvEzxYQQxZLgRgBgs2nSc+OOtMMw52+QdlAlEE9cXnqQUiPEMbslooNKyB30fPEF54qcV8dRgG31G1du18GVjsUwR33s+mwmUfH4BapZWc+evnKejRCiCOmvFgAknr/IJbMNP4OeJqEyU6pUWcmqxyY7VSV43rfEtTo0Q15RRebqdyxaiO5K+j+lapscjlG9N416FH9czlm1lAJAn8mOVcBF5SYF+IRwm/TcCMCRTNy8XhDG6lxh2BWbP1GBTb22qsfG1QJ7eoOaBeROYAOqsF/nO9X26hklH/frkyppOawN3PCCe68hhBBViHyLCcAxDVzyba4gL1uV3wdVlM9bBdL6P6VyLw7/Die3OD+/a7FKOtYZVGVkUw3vtEsIISogCW4EAIeSJd/GJfEL1CrUIVHeXb03tDl0vkttX557s/cnWDJJbff/F0R28167hBCiApLgRgCO1cAluCmFzabWiALoM8m56F556/8U6I1wJNbRexP3lVoZ2poH7UaqY4QQopqT4EZgs2mFhqUkuCnRoRVqzR7/2tDlHu+/fmhUod6bGbD+v7D0n2oF8q73qYUny6tqsBBCVCIS3AhOZ1wkJ8+KyaCjad0gXzen4ipYKmEC+PsoCCzovVkFMS+qfdc+BiM/8H5PkhBCVFAS3IiCXpuosCBMMlOqeGd2wPG1KmG310O+a0dIM7XytN3gl9UUc6mBIoQQBaTOTTWSfiGXWz/ZQLcmIbx7R2d0+V+IhwuSiWWmVIk2fqwe24+G2o1825ZBL4LVDC1ugE63+7YtQghRAUlwU40s253EifQcTqTncF3LMG7rrr6k7csutJRk4uJlnoHd36vtvo/4ti2gpp+Pnu3rVgghRIUlYxDVyB/7Uwq2X/llLylZlwA4mCzJxKX663OwmdWK39VhdWohhKjkJLipJi6Zraw/kgZAw9oBZFw08/LSPWha4TWlZFjKSW4W/PV/artPBei1EUIIcUUS3FQTm4+d5ZLZRv3gAD4b1wOjXsdvu5L4csNxLuRaMOh1RIXJTCkn276ESxlqDam2N/m6NUIIIVwgwU01YR+SGti2Hh0ia/OP61sA8Nqv+wBoVjcQP6N8HIqw5Dqmf1/7uEy1FkKISkK+zaoBTdP444AKbga0CQdgyqCWtKgXhNWmATIkVawdCyHrDNRqCJ3G+ro1QgghXCTBTTVwLC2bE+k5mAw6rm0ZBkCAycBbYzoXlEeRZOLL2KyqAjDANVPcX8lbCCGEz0hwUw38cSAVgN5Rdanp75j9371pCI/d0IpAPwNDoiN81byKad9StdRCjRDoNt7XrRFCCOEGqXNTDdjzbQa0qef03OODW/PooFbo9VLhtoCmwdp31Xavh3231IIQQogykZ6bKi4718LmY+kADGwbXuwxEthc5kgsJO0EUxD0ftjXrRFCCOGmChHcfPTRRzRr1oyAgAB69+7Nli1bXDpv4cKF6HQ6Ro0aVc4trLzWH07DbNVoEhpIc5nq7Zp1s9Rj9wkQGOrTpgghhHCfz4ObRYsWMXXqVF566SXi4uLo3Lkzw4YNIyUlpdTzjh8/zlNPPUW/fv281NLKyT5LalDb8IK1pKotqwX2LFGF+UpycotaIFNvgr6Tvdc2IYQQHuPz4Obdd9/lwQcfZOLEiURHRzN79mwCAwOZM2dOiedYrVbuuecepk+fTvPmzb3Y2spF0zT+2K+SiYvLt6l2Nn4A342HH0oZalr9hnrsfCfUjvROu4QQQniUTxOK8/Ly2LZtG9OmTSvYp9frGTx4MBs3bizxvFdeeYXw8HAeeOAB1q5dW+pr5ObmkpubW/B7ZmYmAGazGbPZfJXvoCj79Tx93bLan5RFUuYlAkx6ujcOrjDt8gS377WmYdw+Hx3AgV+xHF2L1rhPkUN0p/7CeCQWTW/E0vcxqEL362pUtM91VSb32nvkXnuPp+61O+f7NLhJS0vDarUSEVF0GnJERAT79+8v9px169bxxRdfEB8f79JrzJgxg+nTpzvtX7lyJYGBge432gUxMTHlcl13xSTqAAPNgyysilnh6+aUC1fvdZ2co1yffrjg94zvn2Bdq+eh0FBd38NvEQ4khFxL/Ma9wF4Pt7Zyqyif6+pA7rX3yL32nqu91zk5OS4fW6mmgmdlZXHffffx+eefExYW5tI506ZNY+rUqQW/Z2Zm0rhxY4YOHUpwcLBH22c2m4mJiWHIkCGYTCaPXttdl8xW/vvxRiCHO/q1Z3ivxj5tj6e5e6/1K58DwNasH7pTW6mbfYibWurQ2gwHQHdyM8btu9H0Rhre+R4N6zQp1/ZXJhXpc13Vyb32HrnX3uOpe20feXGFT4ObsLAwDAYDycnJRfYnJydTv359p+OPHDnC8ePHGTFiRME+m80GgNFo5MCBA7Ro0aLIOf7+/vj7+ztdy2QyldsHujyv7ar/LDvI0bQcwmr6c0uXRj5vT3lx6V5bLbD3BwD01/wTTm6Gte9gXP0faHcTGIyw7m0AdF3uwVSvRWlXq7Yqwue6upB77T1yr73nau+1O+f6NKHYz8+P7t27ExsbW7DPZrMRGxtL3759nY5v27Ytu3btIj4+vuBn5MiRDBw4kPj4eBo3rlq9E2X1x/4UvtxwHICZt3ciJKiaLx1wdDVkp0JgXWgxCK59TFUeTjsAOxbAiY3qGL0R+j3p69YKIYS4Sj4flpo6dSrjx4+nR48e9OrVi1mzZpGdnc3EiRMBGDduHJGRkcyYMYOAgAA6dOhQ5Pw6deoAOO2vrlKzcvnX4h0ATLimWcFCmVXe6e3w59sw4Blo0KnoczsXqccOt4HBBIba0P9fsOJZ+ON1CIlSz3e9F0KaerfdQgghPM7nwc3YsWNJTU3lxRdfJCkpiS5durB8+fKCJOOEhAT0ep/PWK8UNE3j34t3kHYhjzYRtXjmxra+bpL3xL4CR1apysIPr3EU38u9APt/UduFV/bu+XfYNBsyEtTK33qT9NoIIUQV4fPgBmDKlClMmTKl2OdWr15d6rlffvml5xtUSX218QR/HEjFz6jn/bu6EmAy+LpJ3pGVpIaVADJOwpJJcOe3oNfD/l/BnAOhzSGyu+Mcoz8Meg5+zK950/VekCRiIYSoEqRLpIo4eTaH//y2D4Bnb2xLm/q1fNwiL9r9PWg2Nbxk8IeDy1XBPoBd/1OPncYWmfYNQMfboUlflYsjvTZCCFFlVIieG3H11h9OI89io3PjOoy/ppmvm+Nd9pyavpNBb4BfnoDfp6tg58gq9VzH253P0xtg/C9gs4ApwHvtFUIIUa4kuKkiTmdcAqB9w+DqtYZU2kE4s0PNdGo/WvXCHF8PuxfD/8YBGjTqCXVLmN5tMKofIYQQVYYMS1URp89fBCCyTg0ft8S79LsXq42WgyEoTA09jZgFdVsCmnqucCKxEEKIKk+CmyrCHtw0rFONhlc0myO46XSHY79/LbjjKzDWAFOQ6tERQghRbUh/fBVhD24a1K4+PTeh2YfQZSSAXy1ofWPRJyPaw6T1Kp8myLWlOoQQQlQNEtxUAZqmFeTcVKdhqUZnN6iN6JHgV8wiqCXl2QghhKjSZFiqCkjPziPPYkOng4jgajIsZckl8vwWtV14SEoIIUS1J8FNFWAfkgqv5Y+fsXr8k+qOxOJnzUarWR+a9fN1c4QQQlQg1eObsIqrjvk2+t3fAWDrcJuqVyOEEELkk+CmCjh9vprl21w8h+7QCgBsHYopzieEEKJak+CmCqh208B3/g+dNY+MgMYQ3t7XrRFCCFHBSHBTBZzOsAc31aDnRtNg2zwAToQNcF4vSgghRLUnwU0VkJg/LFUtcm5ObYWUPWjGAE6GXOPr1gghhKiAJLipBDJyzMz6/SDH07KLff5MdVp6Ie5LALR2t2AxBvm2LUIIISokCW4qgWd+2Mms3w/xTsxBp+dyLVZSsnKBapBzcykTdv8AgK3rOB83RgghREUlwY2XrdyTxLdbEjiTnydzJav2J7NsdxIAexIznJ5PzlCBjb9RT2iQn+caWhHt+g7MOVCvLVqjXr5ujRBCiApKll/wouNp2Tw8fxta/mLVHSKDuaFtBEOiI+gQWdvp+Jw8Cy8s2VPw+7H0bHLyLAT6Of7ZEs87kol1VT25dtuX6rHbeEkkFkIIUSLpufGiPw+momkQ6GdAp4PdiZn8N/YQN3+wjqcX7yTPYity/H9jD5F4/iKRdWoQGuSHpsGBpKwix5zJqCbTwE9vh6SdYPCHznf6ujVCCCEqMAluvGjtoVQA/jmoFX89N5i3xnRiaHQEeh0s2nqS8XO2kJFjBmB/UiZfrD0GwPSR7WnfMBiAfWeKBjcFNW6q+kwpe69N9EgIDPVpU4QQQlRsEtx4SZ7FxsYj6QD0axVGWE1/7ujRmM/G9eCL8T0J8jOw8Wg6oz9Zz7G0bJ77cTcWm8aw9hEMjo4guoE9uMkscl37NPAqXeMm9wLsWqy2u0/waVOEEEJUfBLceElcwjmy86yE1fQrCFTsBrYNZ/Gka2hYO4CjqdkMm7WGbSfOEeRn4OWRqgJvuxKCm2pRnXj395B3Aeq2hKbX+ro1QgghKjgJbrzEPiR1Xcsw9HrnZNh2DYJZMvlaOjWqXZB7M3Vom4LCfPbgZn9SFjabVnDe6fPVoDqxJBILIYRwg8yW8pI1B9MA6N+6XonHhAcHsOihvry5fD8Wm43xfZsWPNe8XhB+Bj0Xci2cOneRJnUD0TSt6gc3Z3bC6TjQm6DL3b5ujRBCiEpAghsvSL+Qy+7TqkbNdS3DSj22RqGhqMJMBj2tImqy53Qme89k0qRuIJmXLGTnWYEqnFAcp9aRot3NEFT6vRNCCCFAhqW8Yv2RdDQN2tavRXhw2XNjLs+7sffahAb5UcPPcPUNrWjysmHn/9S2JBILIYRwkQQ3XrDmoMq3ub6UISlXtK1fC3AObhrUrsTJxDsWwY+TIDvd+bk9P0JuJoREQbP+3m+bEEKISkmGpcqZpmkFycT9Wl1dcFMwHTwpP7jJqOTTwPcuhR8fUtu5mTB2ftGE4W35Q1LdxoFe4nAhhBCukW+McnYw+QLJmbkEmPT0aBZyVdeyD0udPHuRrEvmgp6bSrka+Jkd8OPDjt/3/wI7Fzl+T94Dp7aA3ghd7vF++4QQQlRaEtyUM3uvTe+ougSYri4vJiTIj/r5OTv7k7Iqb42brCT49i61CGaLQTBgmtr/278h45TatvfatLkRakX4pp1CCCEqJQluytmaQ2oKeL9Wnpnp066BI+/GkXNTiXpuzBdVYJOZCGGtYcxc6PcURPaA3Az4aTLk5cDOhep4SSQWQgjhpgoR3Hz00Uc0a9aMgIAAevfuzZYtW0o89ocffqBHjx7UqVOHoKAgunTpwtdff+3F1rruktnK5qMqUba0+jbucMyYyuJ0ZVt6QdNgySOqbk2NELh7EdSoAwYjjJ4NxhpwdDV8eydcyoDaTaD5IF+3WgghRCXj8+Bm0aJFTJ06lZdeeom4uDg6d+7MsGHDSElJKfb40NBQnnvuOTZu3MjOnTuZOHEiEydOZMWKFV5u+ZX9dfwsuRYb9YMDaBVeEyy5sOI52Pxpma9pD272nM4gKVMFN5Um52bHQtjzgyrIN3Y+hDZ3PBfWCoa8oraP/akeJZFYCCFEGfj8m+Pdd9/lwQcfZOLEiURHRzN79mwCAwOZM2dOsccPGDCA0aNH065dO1q0aMFjjz1Gp06dWLdunZdbfmX2KeD9WoWhs1nh+7/Dxg9h+TQV6JSBPbjZlZiB1aZh1OuoV8vfY20uN1YL/Pmm2h44DZpd53xMz79D1PVqW2eArpJILIQQwn0+DW7y8vLYtm0bgwcPLtin1+sZPHgwGzduvOL5mqYRGxvLgQMH6N+/YtVB0TSN5XuSABjQOgyW/vKbgfUAABxISURBVBP2Lc1/0gpnj5bpulFhQQSY9Gj5y0tFBAdgKGatqgpn50I4dwwCw6D3P4o/Rq+HUR9Dg85wzRQIbujdNgohhKgSfFrnJi0tDavVSkRE0dkwERER7N+/v8TzMjIyiIyMJDc3F4PBwMcff8yQIUOKPTY3N5fcXEcvSWamqhFjNpsxm80eeBcO9uuZzWZ2JWZw8uxFaph0DD7xDuxYgKYzQI066HLSsSTvQwtpWex19H99BpcysPX7V7HPtw6vyc5E9T4a1Pb3+Psok6wzGFb/B1vXcWiNehV9zmrG+Odb6ABr3ynYdH5QUpsDI+D+WLVdyvsqfK9F+ZJ77T1yr71H7rX3eOpeu3N+pSziV6tWLeLj47lw4QKxsbFMnTqV5s2bM2DAAKdjZ8yYwfTp0532r1y5ksDAwHJpX0xMDD8d1wN6pgcsxD/uJzR0xDX5O+GZu2mcs56DG37l0FHnqeFG60WG73wOHRoxZyPJNdVxOibQrK4NoGWf5bfffiuX9+GONmd+oG3SEqx7lrKm9QtcCIgseK5J+p90PX+CS8Zgfk+NxOrB9sbExHjsWqJ0cq+9R+6198i99p6rvdc5OTkuH+vT4CYsLAyDwUBycnKR/cnJydSvX7/E8/R6PS1bql6PLl26sG/fPmbMmFFscDNt2jSmTp1a8HtmZiaNGzdm6NChBAcHe+aN5DObzcTExDB48GDefH8TdxiWcYf5JwBsf3uLTt0nol/3Lvy5nrZhBloNH+50DV3iVnQ71ZjTDb3aqyGay6RtSmDTr6pnq2d0C4YPaeXR91EWhv8tgCQwWXMYlPQZlgkrIDBU9drMfgEA0/VPMqzPaI+8nv1eDxkyBJPJ5JFriuLJvfYeudfeI/faezx1r+0jL67waXDj5+dH9+7diY2NZdSoUQDYbDZiY2OZMmWKy9ex2WxFhp4K8/f3x9/fOeHWZDKV2wd6T3IOpzMucZd//qyf/v/C0OchDAARbQHQnz2MvrjXTz/oaOOls1DMMR0iHb05kaFBFeM/zJS96tEUhO7cMUw/3A/3/Qh7voPzJyCoHobeD2HwcFvL899RFCX32nvkXnuP3Gvvudp77c65Ph+Wmjp1KuPHj6dHjx706tWLWbNmkZ2dzcSJEwEYN24ckZGRzJgxA1DDTD169KBFixbk5uby22+/8fXXX/PJJ5/48m0U8duuJExYaK8/DhrQ6U7Hk3Xze1nSDqu6L7rLkoFT9jm2LxTt0bJr28DR4xRZEaoTXzwPGQlq+97v4Zvb4cQ6+OUJOL5G7b/2cfArn2FAIYQQojCfBzdjx44lNTWVF198kaSkJLp06cLy5csLkowTEhLQF6p1kp2dzSOPPMKpU6eoUaMGbdu2Zf78+YwdO9ZXb6EImwbL9ibTVpeAn5YHAXWgbgvHAaHNAZ2qxnshxXlpgdRCwU128bV+atcw0bZ+LQ4mZ9E6opbn34S7kveox9qNoWlfGDMHvh0L8fPV/qBw6HG/79onhBCiWvF5cAMwZcqUEoehVq9eXeT31157jddee80LrSqbY1mQnJnLSP/8qd6R3Yv2zpgCIKQpnDsO6Yecg5siPTfFBzcA8+7vRdqFXBqFVIDekKRd6jGig3psPRSG/gdW5K8Zdd0T0msjhBDCaypEcFOVbE9XvUxD65yCDKBRD+eD6rZSwU3awaLF7LLTiw5FlTAsBaq+TURwBRiSAkjOD27qd3Ds6zMJ8i5A+mHoMdE37RJCCFEtSXDjQVabxo501UvT3nZI7YwsJrgJaw2HY1TeTWGFh6Sg1J6bCiVpt3qMKBTc6HRw/b990x4hhBDVms+XX6hKtp44R6ZZR6OASwRmHVM7i+u5Ccsv3pd+qOh++5BUQG31WErPTYVhtTjaXb+jb9sihBBCIMGNR/22Wy23MK6JWgmc0Oaq1svlCmZMHSy63z6dOip/KYnK0HOTfhisuWAKgpAoX7dGCCGEkODGUyxWG8v3qJ6WwbXyp0UXNyQFalgK4HwCmC859tt7QOyLR+ZmgvliObTWg5LtQ1LtZQVvIYQQFYJ8G3nIpqNnOZttJsio0fRifg9McUNSADXDwT8YNJtjAU1Nc/TcNOkDhvzCgxW99yapmGRiIYQQwockuPGQJqGBPNSvGf0ibOjPbFc7S+q50ekgLH9oyp53k5UElzJAZ1A9OzXzp4hX9OAmuZhkYiGEEMKHJLjxkCZ1A/nX0NaMiUhCd/EsGPxK780oyLvJD27svTZ1W4DRX/XuQMVPKrbPlJJkYiGEEBWEBDceFpJ9RG006KyClJLYZ0wVBDf5+Tbh7dRjQc9NBQ5ustPgQhKgg/BoX7dGCCGEACS48biQnPzgpqQhKTt7UnH65cFNfpBQ0HNTgYel7Pk2oVHgX9O3bRFCCCHySXDjYQU9NyUlE9sVHpYqnExc0HNTCYalJN9GCCFEBSTBjSdZcgm+aJ8G3r30Y0Obg06vpntnJUHqfrX/8p6b7NTyaasnSL6NEEKICkiCGw/SJe/CoFnQAutCSLPSDzYFQJ0mavvw72DOUdO/7YXwKkPOjfTcCCGEqIAkuPEgXeI2ALSG3YquBF4S+9DUvqXqMaw1GPKX+/JWcHNsDSy6D86dcO88Sx6kHlDbUuNGCCFEBSLBjQfpTucHN1dKJrazJxUf+UM92vNtoGhCsaZ5qIXFWPuOCq5+muze66QdAJtZrYNVu3H5tU8IIYRwkwQ3HqRLjANAa3iFfBs7+3Rwm1k9Fg5ugvKDG8sllZdTXuyztI6vhfhvXD/PPlMqooNrvVRCCCGEl0hw4ynZaejOHwdAa9jVtXPsPTd2hWvF+AWqJRqg/KaDZ6cXHfZa+byqXeOKJMm3EUIIUTFJcOMp+fk2Wf4N1FCNK+w5N3aFe24Aguqpx/LKu0nN77UJbqRmPF08Byuede3cZFlTSgghRMUkwY2nNL0Gy12L2dvwDtfPsS+gCWAKcs5dKe/1pexDUvU7wIj/AjrYuQiOrCr++IvnYO9S+GUqnPxL7ZOeGyGEEBWM0dcNqDL8a6E1H0DS/hzXz7EvoJm4DcLbgv6yWLO8qxQXXvIhsjv0egi2fKqCl0c2Ajo4uRmOrlY/Z+LVSuZ2wZEQ0b582iaEEEKUkQQ3vlbXHty0c36uvKeDX77kw6DnYd/PcO4YfNofzieohObCwtpA8wHqJ6pf6etnCSGEED4gwY2vdboDTv0FncY6P1daz01WshomCm9bttctbsmHgGAY/jYsugfSDua3oX5+MHO9egxuWLbXE0IIIbxEghtfa3kDPBpX/HOl9dx8cxuk7IdJ66FeG/dfNysJLp0HnaFoYnO7m2H0Z+q5qOvVtWWqtxBCiEpEgpuKrKTFM7OSHHVmdn8PA12c4VSYvdembgu1FERhnYvpRRJCCCEqCZktVZGVtHjmqb8c23t/+v/27j0oqvrvA/h7byw3uccCCYLpiHdNlFD7dZES9ak0rXTI0BodFAxzKjVTcRqDLmNTjYOjpfU8mZaNt8xLimLpICiKl8RLo6k/dUEjWhAF3P08fxCrK2ogyx44vl8zO7t7zncPn32X8Jlzvuece9v2zZOJiYiIVITNTUt286ngtpvOUjqXf+P1pWO1h6ca69bJxERERCrB5qYlq7uIn1iBq6U3lv93X+2z7p8zlepuvNkYdYelHrjHCclEREQtFJublkxnADwDa1/Xzbux1gAXDtS+fiS59rmxh6Zstht39OaeGyIiUhk2Ny3drWdMmQ8D168C7n7AgKmAVg8UHwEu/97wbf59Fqi5AujcgID2zq+ZiIhIQWxuWjr7/aX+udZN3WTitn0BzwAg6j+174sasfembr5NUCdAxxPmiIhIXdjctHS37rmpa27C+9U+d3mu9vloI+bd3HrxPiIiIhVhc9PS3XqV4rozpdrG1D5H/w+g0dbe9+mvPxq2TZ4GTkREKtYimpuFCxciMjIS7u7uiI2NRX5+/h3HLlmyBI8++ij8/f3h7++P+Pj4u45v9W4+HbyiBCg7A0ADPPhPc+MVBEQOrH3d0L03PA2ciIhUTPHm5rvvvsO0adMwd+5c7N+/Hz179sTgwYNRUnL7O2Hn5ORgzJgx2LFjB3JzcxEeHo6nn34a58+fd3HlLnLzYam6Q1LBnWvvA1XHfmiqAfNurDU37hvFPTdERKRCijc3CxYswIQJEzB+/Hh06dIFixYtgqenJ5YuXXrb8cuXL8fkyZPRq1cvREdH44svvoDNZkN2draLK3eRmw9L2Q9J9XUcE/0MAA1wfh9Qdu7GchHHi/8BQOkpwFoNuHkDvuHNVjYREZFSFD1Vprq6GgUFBZg5c6Z9mVarRXx8PHJzcxu0jcrKStTU1CAgIOC266uqqlBVVWV/b7FYAAA1NTWoqalpQvX11W3Pqdt1D4ABgFQUQ87lQQvgeujDkJt/hnsAdOGx0J7bA9u6VMDgAU3ZGeCvM4BGA+sL/wuJrD2rSnPxMPQAbEGdYLVaAavVebW6ULNkTbfFrF2HWbsOs3YdZ2XdmM8r2txcvnwZVqsVJpPJYbnJZMKxYw27pcD06dMRFhaG+Pj4267PyMjAvHnz6i3/+eef4enp2fiiG2Dr1q1O25ZbjQVDAGiulsJ6rgBaADtPVaLiwkaHce1tHdAde6A9nVN/IyvGILfDDPzl9RA6XfwR0QDOVXmjcOPG+mNbGWdmTXfHrF2HWbsOs3adpmZdWVnZ4LGt+iInmZmZWLlyJXJycuDu7n7bMTNnzsS0adPs7y0Wi32ejo+Pz20/c69qamqwdetWPPXUUzAYDM7ZqNggv6VBI1bopRri7ov/jHit9gwphx/+BKw7//k+fu0gfu0gvuHQ/TwT+j9+waNnP8X1l9dBt8sKmIG2veMRFjvUOTUqoFmyptti1q7DrF2HWbuOs7KuO/LSEIo2N0FBQdDpdCguLnZYXlxcjJCQkLt+9uOPP0ZmZia2bduGHj163HGc0WiE0Wist9xgMDTb/9BO37Z3MFB+EQCgadsXBrf63wcGAzAko/7yMSuA/xsBzX/zYVjxgr0p0oV2g04F/6Cb878jOWLWrsOsXYdZu05Ts27MZxWdUOzm5oY+ffo4TAaumxwcFxd3x899+OGHeO+997B582bExMS4olRl1U0qBupPJv43Rm8g8XvA1B24UgJUmGuX8zRwIiJSKcXPlpo2bRqWLFmCr7/+GkVFRZg0aRKuXLmC8ePHAwBeeeUVhwnHH3zwAWbPno2lS5ciMjISZrMZZrMZFRUVSn2F5ud905ykxjY3AODhD4xdAwR2uPHe23T3zxAREbVSis+5eemll3Dp0iXMmTMHZrMZvXr1wubNm+2TjM+ePQut9kYPlpWVherqaowaNcphO3PnzkV6erorS3cdr7o9N5obVyZuLO8HgFfWAetSgPZPABqN08ojIiJqSRRvbgAgNTUVqampt12Xk5Pj8P6PP/5o/oJamrrDUg9EA+6+974d37a1DQ4REZGKKX5YihrA1LX2ucMgZesgIiJqBVrEnhv6F91G1s6XeSBa6UqIiIhaPDY3rYFGA4T1UroKIiKiVoGHpYiIiEhV2NwQERGRqrC5ISIiIlVhc0NERESqwuaGiIiIVIXNDREREakKmxsiIiJSFTY3REREpCpsboiIiEhV2NwQERGRqrC5ISIiIlVhc0NERESqwuaGiIiIVOW+uyu4iAAALBaL07ddU1ODyspKWCwWGAwGp2+fbmDWrsOsXYdZuw6zdh1nZV33d7vu7/jd3HfNTXl5OQAgPDxc4UqIiIioscrLy+Hr63vXMRppSAukIjabDRcuXECbNm2g0Wicum2LxYLw8HCcO3cOPj4+Tt02OWLWrsOsXYdZuw6zdh1nZS0iKC8vR1hYGLTau8+que/23Gi1WrRt27ZZf4aPjw//sbgIs3YdZu06zNp1mLXrOCPrf9tjU4cTiomIiEhV2NwQERGRqujS09PTlS5CTXQ6HR5//HHo9ffdET+XY9auw6xdh1m7DrN2HVdnfd9NKCYiIiJ142EpIiIiUhU2N0RERKQqbG6IiIhIVdjcOMnChQsRGRkJd3d3xMbGIj8/X+mSWr2MjAz07dsXbdq0QXBwMIYPH47jx487jLl27RpSUlIQGBgIb29vjBw5EsXFxQpVrB6ZmZnQaDSYOnWqfRmzdp7z58/j5ZdfRmBgIDw8PNC9e3fs27fPvl5EMGfOHISGhsLDwwPx8fE4efKkghW3TlarFbNnz0ZUVBQ8PDzw0EMP4b333nO4fD+zvne//PILnnnmGYSFhUGj0WDt2rUO6xuSbWlpKRITE+Hj4wM/Pz+89tprqKioaHpxQk22cuVKcXNzk6VLl8pvv/0mEyZMED8/PykuLla6tFZt8ODBsmzZMjly5IgUFhbK0KFDJSIiQioqKuxjkpOTJTw8XLKzs2Xfvn3yyCOPSP/+/RWsuvXLz8+XyMhI6dGjh6SlpdmXM2vnKC0tlXbt2sm4ceMkLy9PTp06JVu2bJHff//dPiYzM1N8fX1l7dq1cvDgQXn22WclKipKrl69qmDlrc/8+fMlMDBQNmzYIKdPn5ZVq1aJt7e3fPrpp/YxzPrebdy4UWbNmiWrV68WALJmzRqH9Q3JNiEhQXr27Cl79uyRX3/9VTp06CBjxoxpcm1sbpygX79+kpKSYn9vtVolLCxMMjIyFKxKfUpKSgSA7Ny5U0REysrKxGAwyKpVq+xjioqKBIDk5uYqVWarVl5eLh07dpStW7fKY489Zm9umLXzTJ8+XQYOHHjH9TabTUJCQuSjjz6yLysrKxOj0SgrVqxwRYmqMWzYMHn11Vcdlj3//POSmJgoIszamW5tbhqS7dGjRwWA7N271z5m06ZNotFo5Pz5802qh4elmqi6uhoFBQWIj4+3L9NqtYiPj0dubq6ClanP33//DQAICAgAABQUFKCmpsYh++joaERERDD7e5SSkoJhw4Y5ZAowa2dav349YmJi8MILLyA4OBi9e/fGkiVL7OtPnz4Ns9nskLWvry9iY2OZdSP1798f2dnZOHHiBADg4MGD2LVrF4YMGQKAWTenhmSbm5sLPz8/xMTE2MfEx8dDq9UiLy+vST+fVy5qosuXL8NqtcJkMjksN5lMOHbsmEJVqY/NZsPUqVMxYMAAdOvWDQBgNpvh5uYGPz8/h7Emkwlms1mJMlu1lStXYv/+/di7d2+9dczaeU6dOoWsrCxMmzYN77zzDvbu3YvXX38dbm5uSEpKsud5u98pzLpxZsyYAYvFgujoaOh0OlitVsyfPx+JiYkAwKybUUOyNZvNCA4Odliv1+sREBDQ5PzZ3FCrkJKSgiNHjmDXrl1Kl6JK586dQ1paGrZu3Qp3d3ely1E1m82GmJgYvP/++wCA3r1748iRI1i0aBGSkpIUrk5dvv/+eyxfvhzffvstunbtisLCQkydOhVhYWHMWuV4WKqJgoKCoNPp6p01UlxcjJCQEIWqUpfU1FRs2LABO3bscLije0hICKqrq1FWVuYwntk3XkFBAUpKSvDwww9Dr9dDr9dj586d+Oyzz6DX62EymZi1k4SGhqJLly4Oyzp37oyzZ88CgD1P/k5purfeegszZszA6NGj0b17d4wdOxZvvPEGMjIyADDr5tSQbENCQlBSUuKw/vr16ygtLW1y/mxumsjNzQ19+vRBdna2fZnNZkN2djbi4uIUrKz1ExGkpqZizZo12L59O6KiohzW9+nTBwaDwSH748eP4+zZs8y+kQYNGoTDhw+jsLDQ/oiJiUFiYqL9NbN2jgEDBtS7pMGJEyfQrl07AEBUVBRCQkIcsrZYLMjLy2PWjVRZWQmt1vHPnE6ng81mA8Csm1NDso2Li0NZWRkKCgrsY7Zv3w6bzYbY2NimFdCk6cgkIrWnghuNRvnqq6/k6NGjMnHiRPHz8xOz2ax0aa3apEmTxNfXV3JycuTixYv2R2VlpX1McnKyREREyPbt22Xfvn0SFxcncXFxClatHjefLSXCrJ0lPz9f9Hq9zJ8/X06ePCnLly8XT09P+eabb+xjMjMzxc/PT9atWyeHDh2S5557jqcn34OkpCR58MEH7aeCr169WoKCguTtt9+2j2HW9668vFwOHDggBw4cEACyYMECOXDggJw5c0ZEGpZtQkKC9O7dW/Ly8mTXrl3SsWNHngreknz++ecSEREhbm5u0q9fP9mzZ4/SJbV6AG77WLZsmX3M1atXZfLkyeLv7y+enp4yYsQIuXjxonJFq8itzQ2zdp4ff/xRunXrJkajUaKjo2Xx4sUO6202m8yePVtMJpMYjUYZNGiQHD9+XKFqWy+LxSJpaWkSEREh7u7u0r59e5k1a5ZUVVXZxzDre7djx47b/o5OSkoSkYZl++eff8qYMWPE29tbfHx8ZPz48VJeXt7k2nhXcCIiIlIVzrkhIiIiVWFzQ0RERKrC5oaIiIhUhc0NERERqQqbGyIiIlIVNjdERESkKmxuiIiISFXY3BAREZGqsLkhovueRqPB2rVrlS6DiJyEzQ0RKWrcuHHQaDT1HgkJCUqXRkStlF7pAoiIEhISsGzZModlRqNRoWqIqLXjnhsiUpzRaERISIjDw9/fH0DtIaOsrCwMGTIEHh4eaN++PX744QeHzx8+fBhPPvkkPDw8EBgYiIkTJ6KiosJhzNKlS9G1a1cYjUaEhoYiNTXVYf3ly5cxYsQIeHp6omPHjli/fn3zfmkiajZsboioxZs9ezZGjhyJgwcPIjExEaNHj0ZRUREA4MqVKxg8eDD8/f2xd+9erFq1Ctu2bXNoXrKyspCSkoKJEyfi8OHDWL9+PTp06ODwM+bNm4cXX3wRhw4dwtChQ5GYmIjS0lKXfk8icpIm31eciKgJkpKSRKfTiZeXl8Nj/vz5IiICQJKTkx0+ExsbK5MmTRIRkcWLF4u/v79UVFTY1//000+i1WrFbDaLiEhYWJjMmjXrjjUAkHfffdf+vqKiQgDIpk2bnPY9ich1OOeGiBT3xBNPICsry2FZQECA/XVcXJzDuri4OBQWFgIAioqK0LNnT3h5ednXDxgwADabDcePH4dGo8GFCxcwaNCgu9bQo0cP+2svLy/4+PigpKTknr8TESmHzQ0RKc7Ly6veYSJn8fDwaNA4g8Hg8F6j0cBmszVHSUTUzDjnhohavD179tR737lzZwBA586dcfDgQVy5csW+fvfu3dBqtejUqRPatGmDyMhIZGdnu7RmIlIO99wQkeKqqqpgNpsdlun1egQFBQEAVq1ahZiYGAwcOBDLly9Hfn4+vvzySwBAYmIi5s6di6SkJKSnp+PSpUuYMmUKxo4dC5PJBABIT09HcnIygoODMWTIEJSXl2P37t2YMmWKa78oEbkEmxsiUtzmzZsRGhrqsKxTp044duwYgNozmVauXInJkycjNDQUK1asQJcuXQAAnp6e2LJlC9LS0tC3b194enpi5MiRWLBggX1bSUlJuHbtGj755BO8+eabCAoKwqhRo1z3BYnIpTQiIkoXQUR0JxqNBmvWrMHw4cOVLoWIWgnOuSEiIiJVYXNDREREqsI5N0TUovHIORE1FvfcEBERkaqwuSEiIiJVYXNDREREqsLmhoiIiFSFzQ0RERGpCpsbIiIiUhU2N0RERKQqbG6IiIhIVdjcEBERkar8P4CQgvNktRg2AAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Testing The Model"
      ],
      "metadata": {
        "id": "UsUgOQmt1jKk"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "test_loop(test_dataloader, model, criterion)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1c8hjIdfZmLT",
        "outputId": "2da5bf0d-4d06-4cdc-fe7f-a39779ffcb03"
      },
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Validation Loss: 1.6762 Acc: 0.4857\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(1.6761793865476335, tensor(0.4857, device='cuda:0', dtype=torch.float64))"
            ]
          },
          "metadata": {},
          "execution_count": 26
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "test_loop(val_dataloader, model, criterion)"
      ],
      "metadata": {
        "id": "nvUSGBKz5BKR",
        "outputId": "4d71d29f-429d-455b-da80-ef67cc05de92",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Validation Loss: 1.6301 Acc: 0.5286\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(1.6300642776489258, tensor(0.5286, device='cuda:0', dtype=torch.float64))"
            ]
          },
          "metadata": {},
          "execution_count": 27
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "torch.save(model, '/content/Subesco_resnet50')"
      ],
      "metadata": {
        "id": "jMtikt3d1Vww"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}