Copy_of_CS146_Winter2021_PS3.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Copy of CS146-Winter2021-PS3.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/alyssamw/cs35l/blob/master/Copy_of_CS146_Winter2021_PS3.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZQ57FP3Zdq19"
      },
      "source": [
        "import os\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import torch\n",
        "from torch.utils.data import TensorDataset, DataLoader\n",
        "from PIL import Image"
      ],
      "execution_count": 11,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PyqBmQZE7Uqw",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "fdd48fc9-b0f5-4af2-d36c-8ff8dd9db5dc"
      },
      "source": [
        "# To add your own Drive Run this cell.\n",
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "q4RwWjtN2sf4"
      },
      "source": [
        "######################################################################\n",
        "# OneLayerNetwork\n",
        "######################################################################\n",
        "\n",
        "class OneLayerNetwork(torch.nn.Module):\n",
        "    def __init__(self):\n",
        "        super(OneLayerNetwork, self).__init__()\n",
        "\n",
        "        ### ========== TODO : START ========== ###\n",
        "        ### part d: implement OneLayerNetwork with torch.nn.Linear\n",
        "        self.layer = torch.nn.Linear(784, 3)\n",
        "        ### ========== TODO : END ========== ###\n",
        "\n",
        "    def forward(self, x):\n",
        "        # x.shape = (n_batch, n_features)\n",
        "\n",
        "        ### ========== TODO : START ========== ###\n",
        "        ### part d: implement the foward function\n",
        "        loss = torch.nn.CrossEntropyLoss()\n",
        "        input = x\n",
        "        outputs = self.layer(x)\n",
        "        #outputs = loss(input,target)\n",
        "\n",
        "        ### ========== TODO : END ========== ###\n",
        "        return outputs"
      ],
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JRbW6WPv25tE"
      },
      "source": [
        "######################################################################\n",
        "# TwoLayerNetwork\n",
        "######################################################################\n",
        "\n",
        "class TwoLayerNetwork(torch.nn.Module):\n",
        "    def __init__(self):\n",
        "        super(TwoLayerNetwork, self).__init__()\n",
        "        ### ========== TODO : START ========== ###\n",
        "        ### part g: implement TwoLayerNetwork with torch.nn.Linear\n",
        "        self.layer1 = torch.nn.Linear(784, 400)\n",
        "        self.in2 = torch.nn.Sigmoid()\n",
        "        self.layer2 = torch.nn.Linear(400, 3)\n",
        "        ### ========== TODO : END ========== ###\n",
        "\n",
        "    def forward(self, x):\n",
        "        # x.shape = (n_batch, n_features)\n",
        "\n",
        "        ### ========== TODO : START ========== ###\n",
        "        ### part g: implement the foward function\n",
        "        input = x\n",
        "        out1 = self.layer1(x)\n",
        "        in2 = self.in2(out1)\n",
        "        outputs = self.layer2(in2)\n",
        "        ### ========== TODO : END ========== ###\n",
        "        return outputs"
      ],
      "execution_count": 14,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "I06zHwgTeSs3"
      },
      "source": [
        "# load data from csv\n",
        "# X.shape = (n_examples, n_features), y.shape = (n_examples, )\n",
        "def load_data(filename):\n",
        "    data = np.loadtxt(filename)\n",
        "    y = data[:, 0].astype(int)\n",
        "    X = data[:, 1:].astype(np.float32) / 255\n",
        "    return X, y"
      ],
      "execution_count": 15,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JvP1DJfteUOn"
      },
      "source": [
        "# plot one example\n",
        "# x.shape = (features, )\n",
        "def plot_img(x):\n",
        "    x = x.reshape(28, 28)\n",
        "    img = Image.fromarray(x*255)\n",
        "    plt.figure()\n",
        "    plt.imshow(img)\n",
        "    return"
      ],
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RqaCgw9Z4Eef"
      },
      "source": [
        "def evaluate_loss(model, criterion, dataloader):\n",
        "    model.eval()\n",
        "    total_loss = 0.0\n",
        "    for batch_X, batch_y in dataloader:\n",
        "        outputs = model(batch_X)\n",
        "        loss = criterion(outputs, batch_y)\n",
        "        total_loss += loss.item()\n",
        "        \n",
        "    return total_loss / len(dataloader)"
      ],
      "execution_count": 17,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uEnwn4sC4F3B"
      },
      "source": [
        "def evaluate_acc(model, dataloader):\n",
        "    model.eval()\n",
        "    total_acc = 0.0\n",
        "    for batch_X, batch_y in dataloader:\n",
        "        outputs = model(batch_X)\n",
        "        predictions = torch.argmax(outputs, dim=1)\n",
        "        total_acc += (predictions==batch_y).sum()\n",
        "        \n",
        "    return total_acc / len(dataloader.dataset)"
      ],
      "execution_count": 18,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6OpXc_t7Jc5_"
      },
      "source": [
        "def train(model, criterion, optimizer, train_loader, valid_loader):\n",
        "    train_loss_list = []\n",
        "    valid_loss_list = []\n",
        "    train_acc_list = []\n",
        "    valid_acc_list = []\n",
        "    for epoch in range(1, 31):\n",
        "        model.train()\n",
        "        for batch_X, batch_y in train_loader:\n",
        "            ### ========== TODO : START ========== ###\n",
        "            ### part f: implement the training process\n",
        "            y_pred = model.forward(batch_X)\n",
        "            loss = criterion(y_pred, batch_y)\n",
        "\n",
        "            optimizer.zero_grad()\n",
        "            loss.backward()\n",
        "            optimizer.step()\n",
        "            ### ========== TODO : END ========== ###\n",
        "            \n",
        "        train_loss = evaluate_loss(model, criterion, train_loader)\n",
        "        valid_loss = evaluate_loss(model, criterion, valid_loader)\n",
        "        train_acc = evaluate_acc(model, train_loader)\n",
        "        valid_acc = evaluate_acc(model, valid_loader)\n",
        "        train_loss_list.append(train_loss)\n",
        "        valid_loss_list.append(valid_loss)\n",
        "        train_acc_list.append(train_acc)\n",
        "        valid_acc_list.append(valid_acc)\n",
        "\n",
        "        print(f\"| epoch {epoch:2d} | train loss {train_loss:.6f} | train acc {train_acc:.6f} | valid loss {valid_loss:.6f} | valid acc {valid_acc:.6f} |\")\n",
        "\n",
        "    return train_loss_list, valid_loss_list, train_acc_list, valid_acc_list"
      ],
      "execution_count": 19,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KMmurb4-d39G",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "11e55e1a-f3b4-42c5-f9b2-6846b6b472cd"
      },
      "source": [
        "#@title\n",
        "######################################################################\n",
        "# main\n",
        "######################################################################\n",
        "\n",
        "def main():\n",
        "\n",
        "    # fix random seed\n",
        "    np.random.seed(0)\n",
        "    torch.manual_seed(0)\n",
        "\n",
        "    # load data with correct file path\n",
        "\n",
        "    ### ========== TODO : START ========== ###\n",
        "    data_directory_path =  \"/content/drive/My Drive/cs146\"\n",
        "    ### ========== TODO : END ========== ###\n",
        "\n",
        "    # X.shape = (n_examples, n_features)\n",
        "    # y.shape = (n_examples, )\n",
        "    X_train, y_train = load_data(os.path.join(data_directory_path, \"ps3_train.csv\"))\n",
        "    X_valid, y_valid = load_data(os.path.join(data_directory_path, \"ps3_valid.csv\"))\n",
        "    X_test, y_test = load_data(os.path.join(data_directory_path, \"ps3_test.csv\"))\n",
        "\n",
        "    ### ========== TODO : START ========== ###\n",
        "    ### part a: print out three training images with different labels\n",
        "    \n",
        "    for i in range(0,3):\n",
        "      plot_img(X_train[i])\n",
        "    \n",
        "    ### ========== TODO : END ========== ###\n",
        "\n",
        "    print(\"Data preparation...\")\n",
        "\n",
        "    ### ========== TODO : START ========== ###\n",
        "    ### part b: convert numpy arrays to tensors\n",
        "    X_train = torch.tensor(X_train)\n",
        "    X_valid = torch.tensor(X_valid)\n",
        "    X_test = torch.tensor(X_test)\n",
        "\n",
        "    y_train = torch.tensor(y_train)\n",
        "    y_valid = torch.tensor(y_valid)\n",
        "    y_test = torch.tensor(y_test)\n",
        "    \n",
        "    ### ========== TODO : END ========== ###\n",
        "\n",
        "    ### ========== TODO : START ========== ###\n",
        "    ### part c: prepare dataloaders for training, validation, and testing\n",
        "    ###         we expect to get a batch of pairs (x_n, y_n) from the dataloader\n",
        "    train_loader = DataLoader(TensorDataset(X_train, y_train), 10)\n",
        "    valid_loader = DataLoader(TensorDataset(X_valid, y_valid), 10)\n",
        "    test_loader = DataLoader(TensorDataset(X_test, y_test), 10)\n",
        "    \n",
        "    ### ========== TODO : END ========== ###\n",
        "\n",
        "    ### ========== TODO : START ========== ###\n",
        "    ### part e: prepare OneLayerNetwork, criterion, and optimizer\n",
        "    model_one = OneLayerNetwork()\n",
        "    criterion = torch.nn.CrossEntropyLoss()\n",
        "    optimizer = torch.optim.SGD(model_one.parameters(), lr=0.0005)\n",
        "    \n",
        "    ### ========== TODO : END ========== ###\n",
        "\n",
        "    print(\"Start training OneLayerNetwork...\")\n",
        "    results_one = train(model_one, criterion, optimizer, train_loader, valid_loader)\n",
        "    print(\"Done!\")\n",
        "\n",
        "    ### ========== TODO : START ========== ###\n",
        "    ### part h: prepare TwoLayerNetwork, criterion, and optimizer\n",
        "    model_two = TwoLayerNetwork()\n",
        "    criterion = torch.nn.CrossEntropyLoss()\n",
        "    optimizer = torch.optim.SGD(model_two.parameters(), lr=0.0005)\n",
        "    \n",
        "    ### ========== TODO : END ========== ###\n",
        "\n",
        "    print(\"Start training TwoLayerNetwork...\")\n",
        "    results_two = train(model_two, criterion, optimizer, train_loader, valid_loader)\n",
        "    print(\"Done!\")\n",
        "\n",
        "    one_train_loss, one_valid_loss, one_train_acc, one_valid_acc = results_one\n",
        "    two_train_loss, two_valid_loss, two_train_acc, two_valid_acc = results_two\n",
        "\n",
        "    ### ========== TODO : START ========== ###\n",
        "    ### part i: generate a plot to comare one_train_loss, one_valid_loss, two_train_loss, two_valid_loss\n",
        "    plt.figure()\n",
        "    plt.plot(range(0, len(one_train_loss)), one_train_loss, label=\"One Layer Train\")\n",
        "    plt.plot(range(0, len(one_valid_loss)), one_valid_loss, label=\"One Layer Valid\")\n",
        "    plt.plot(range(0, len(two_train_loss)), two_train_loss, label=\"Two Layer Train\")\n",
        "    plt.plot(range(0, len(two_valid_loss)), two_valid_loss, label=\"Two Layer Valid\")\n",
        "    plt.title(\"Loss against Epochs for One and Two Layer Networks\")\n",
        "    plt.xlabel(\"Num Epochs\")\n",
        "    plt.ylabel(\"Loss\")\n",
        "    plt.legend()\n",
        "    plt.show()\n",
        "    ### ========== TODO : END ========== ###\n",
        "\n",
        "    ### ========== TODO : START ========== ###\n",
        "    ### part j: generate a plot to comare one_train_acc, one_valid_acc, two_train_acc, two_valid_acc\n",
        "    plt.figure()\n",
        "    plt.plot(range(0, len(one_train_acc)), one_train_acc, label=\"One Layer Train\")\n",
        "    plt.plot(range(0, len(one_valid_acc)), one_valid_acc, label=\"One Layer Valid\")\n",
        "    plt.plot(range(0, len(two_train_acc)), two_train_acc, label=\"Two Layer Train\")\n",
        "    plt.plot(range(0, len(two_valid_acc)), two_valid_acc, label=\"Two Layer Valid\")\n",
        "    plt.title(\"Accuracy against Epochs for One and Two Layer Networks\")\n",
        "    plt.xlabel(\"Num Epochs\")\n",
        "    plt.ylabel(\"Accuracy\")\n",
        "    plt.legend()\n",
        "    plt.show()\n",
        "    ### ========== TODO : END ========== ##\n",
        "\n",
        "    ### ========== TODO : START ========== ###\n",
        "    ### part k: calculate the test accuracy\n",
        "    one_test_acc = evaluate_acc(model_one, test_loader)\n",
        "    two_test_acc = evaluate_acc(model_two, test_loader)\n",
        "    print(f\"One Layer Test: {one_test_acc:.6f}, Two Layer Test: {two_test_acc:.6f}\")\n",
        "    ### ========== TODO : END ========== ###\n",
        "\n",
        "    # part K improve:\n",
        "\n",
        "    model_one = OneLayerNetwork()\n",
        "    model_two = TwoLayerNetwork()\n",
        "    optimizer = torch.optim.SGD(model_one.parameters(), lr=0.05)\n",
        "    print(\"Start training OneLayerNetwork with Adam...\")\n",
        "    results_one = train(model_one, criterion, optimizer, train_loader, valid_loader)\n",
        "    print(\"Done!\")\n",
        "\n",
        "    optimizer = torch.optim.Adam(model_two.parameters(), lr=0.05)\n",
        "    print(\"Start training TwoLayerNetwork with Adam...\")\n",
        "    results_two = train(model_two, criterion, optimizer, train_loader, valid_loader)\n",
        "    print(\"Done!\")\n",
        "\n",
        "    one_test_acc = evaluate_acc(model_one, test_loader)\n",
        "    two_test_acc = evaluate_acc(model_two, test_loader)\n",
        "    print(f\"(Imp) One Layer Test: {one_test_acc:.6f}, Two Layer Test: {two_test_acc:.6f}\")\n",
        "\n",
        "    ### ========== TODO : START ========== ###\n",
        "    ### part l: replace the SGD optimizer with the Adam optimizer and do the experiments again\n",
        "    model_one = OneLayerNetwork()\n",
        "    model_two = TwoLayerNetwork()\n",
        "    optimizer = torch.optim.Adam(model_one.parameters(), lr=0.0005)\n",
        "    print(\"Start training OneLayerNetwork with Adam...\")\n",
        "    results_one = train(model_one, criterion, optimizer, train_loader, valid_loader)\n",
        "    print(\"Done!\")\n",
        "\n",
        "    optimizer = torch.optim.Adam(model_two.parameters(), lr=0.0005)\n",
        "    print(\"Start training TwoLayerNetwork with Adam...\")\n",
        "    results_two = train(model_two, criterion, optimizer, train_loader, valid_loader)\n",
        "    print(\"Done!\")\n",
        "\n",
        "    one_train_loss, one_valid_loss, one_train_acc, one_valid_acc = results_one\n",
        "    two_train_loss, two_valid_loss, two_train_acc, two_valid_acc = results_two\n",
        "\n",
        "    ### part i: generate a plot to comare one_train_loss, one_valid_loss, two_train_loss, two_valid_loss\n",
        "    plt.figure()\n",
        "    plt.plot(range(0, len(one_train_loss)), one_train_loss, label=\"One Layer Train\")\n",
        "    plt.plot(range(0, len(one_valid_loss)), one_valid_loss, label=\"One Layer Valid\")\n",
        "    plt.plot(range(0, len(two_train_loss)), two_train_loss, label=\"Two Layer Train\")\n",
        "    plt.plot(range(0, len(two_valid_loss)), two_valid_loss, label=\"Two Layer Valid\")\n",
        "    plt.title(\"Loss against Epochs for One and Two Layer Networks with Adam\")\n",
        "    plt.xlabel(\"Num Epochs\")\n",
        "    plt.ylabel(\"Loss\")\n",
        "    plt.legend()\n",
        "    plt.show()\n",
        "    \n",
        "    ### part j: generate a plot to comare one_train_acc, one_valid_acc, two_train_acc, two_valid_acc\n",
        "    plt.figure()\n",
        "    plt.plot(range(0, len(one_train_acc)), one_train_acc, label=\"One Layer Train\")\n",
        "    plt.plot(range(0, len(one_valid_acc)), one_valid_acc, label=\"One Layer Valid\")\n",
        "    plt.plot(range(0, len(two_train_acc)), two_train_acc, label=\"Two Layer Train\")\n",
        "    plt.plot(range(0, len(two_valid_acc)), two_valid_acc, label=\"Two Layer Valid\")\n",
        "    plt.title(\"Accuracy against Epochs for One and Two Layer Networks with Adam\")\n",
        "    plt.xlabel(\"Num Epochs\")\n",
        "    plt.ylabel(\"Accuracy\")\n",
        "    plt.legend()\n",
        "    plt.show()\n",
        "    ### part k: calculate the test accuracy\n",
        "    one_test_acc = evaluate_acc(model_one, test_loader)\n",
        "    two_test_acc = evaluate_acc(model_two, test_loader)\n",
        "    print(f\"(Adam) One Layer Test: {one_test_acc:.6f}, Two Layer Test: {two_test_acc:.6f}\")\n",
        "\n",
        "    ### ========== TODO : END ========== ###\n",
        "\n",
        "\n",
        "\n",
        "if __name__ == \"__main__\":\n",
        "    main()"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Data preparation...\n",
            "Start training OneLayerNetwork...\n",
            "| epoch  1 | train loss 1.075398 | train acc 0.453333 | valid loss 1.084938 | valid acc 0.453333 |\n",
            "| epoch  2 | train loss 1.021364 | train acc 0.566667 | valid loss 1.031102 | valid acc 0.553333 |\n",
            "| epoch  3 | train loss 0.972648 | train acc 0.630000 | valid loss 0.982742 | valid acc 0.593333 |\n",
            "| epoch  4 | train loss 0.928398 | train acc 0.710000 | valid loss 0.938953 | valid acc 0.640000 |\n",
            "| epoch  5 | train loss 0.887963 | train acc 0.783333 | valid loss 0.899045 | valid acc 0.700000 |\n",
            "| epoch  6 | train loss 0.850839 | train acc 0.826667 | valid loss 0.862485 | valid acc 0.753333 |\n",
            "| epoch  7 | train loss 0.816627 | train acc 0.850000 | valid loss 0.828852 | valid acc 0.793333 |\n",
            "| epoch  8 | train loss 0.785000 | train acc 0.886667 | valid loss 0.797807 | valid acc 0.846667 |\n",
            "| epoch  9 | train loss 0.755688 | train acc 0.900000 | valid loss 0.769067 | valid acc 0.866667 |\n",
            "| epoch 10 | train loss 0.728461 | train acc 0.903333 | valid loss 0.742397 | valid acc 0.873333 |\n",
            "| epoch 11 | train loss 0.703122 | train acc 0.913333 | valid loss 0.717596 | valid acc 0.880000 |\n",
            "| epoch 12 | train loss 0.679499 | train acc 0.920000 | valid loss 0.694488 | valid acc 0.886667 |\n",
            "| epoch 13 | train loss 0.657439 | train acc 0.933333 | valid loss 0.672921 | valid acc 0.886667 |\n",
            "| epoch 14 | train loss 0.636807 | train acc 0.943333 | valid loss 0.652760 | valid acc 0.886667 |\n",
            "| epoch 15 | train loss 0.617482 | train acc 0.943333 | valid loss 0.633883 | valid acc 0.886667 |\n",
            "| epoch 16 | train loss 0.599356 | train acc 0.943333 | valid loss 0.616184 | valid acc 0.886667 |\n",
            "| epoch 17 | train loss 0.582330 | train acc 0.943333 | valid loss 0.599565 | valid acc 0.893333 |\n",
            "| epoch 18 | train loss 0.566316 | train acc 0.943333 | valid loss 0.583938 | valid acc 0.900000 |\n",
            "| epoch 19 | train loss 0.551234 | train acc 0.943333 | valid loss 0.569225 | valid acc 0.906667 |\n",
            "| epoch 20 | train loss 0.537010 | train acc 0.943333 | valid loss 0.555355 | valid acc 0.906667 |\n",
            "| epoch 21 | train loss 0.523580 | train acc 0.943333 | valid loss 0.542262 | valid acc 0.906667 |\n",
            "| epoch 22 | train loss 0.510882 | train acc 0.943333 | valid loss 0.529888 | valid acc 0.906667 |\n",
            "| epoch 23 | train loss 0.498862 | train acc 0.950000 | valid loss 0.518179 | valid acc 0.906667 |\n",
            "| epoch 24 | train loss 0.487470 | train acc 0.950000 | valid loss 0.507086 | valid acc 0.906667 |\n",
            "| epoch 25 | train loss 0.476660 | train acc 0.950000 | valid loss 0.496564 | valid acc 0.906667 |\n",
            "| epoch 26 | train loss 0.466391 | train acc 0.953333 | valid loss 0.486573 | valid acc 0.926667 |\n",
            "| epoch 27 | train loss 0.456625 | train acc 0.953333 | valid loss 0.477076 | valid acc 0.926667 |\n",
            "| epoch 28 | train loss 0.447328 | train acc 0.953333 | valid loss 0.468038 | valid acc 0.926667 |\n",
            "| epoch 29 | train loss 0.438467 | train acc 0.956667 | valid loss 0.459429 | valid acc 0.933333 |\n",
            "| epoch 30 | train loss 0.430013 | train acc 0.956667 | valid loss 0.451220 | valid acc 0.940000 |\n",
            "Done!\n",
            "Start training TwoLayerNetwork...\n",
            "| epoch  1 | train loss 1.098020 | train acc 0.240000 | valid loss 1.098498 | valid acc 0.253333 |\n",
            "| epoch  2 | train loss 1.096157 | train acc 0.283333 | valid loss 1.096622 | valid acc 0.340000 |\n",
            "| epoch  3 | train loss 1.094329 | train acc 0.386667 | valid loss 1.094783 | valid acc 0.380000 |\n",
            "| epoch  4 | train loss 1.092512 | train acc 0.433333 | valid loss 1.092956 | valid acc 0.400000 |\n",
            "| epoch  5 | train loss 1.090700 | train acc 0.470000 | valid loss 1.091135 | valid acc 0.413333 |\n",
            "| epoch  6 | train loss 1.088891 | train acc 0.486667 | valid loss 1.089318 | valid acc 0.420000 |\n",
            "| epoch  7 | train loss 1.087085 | train acc 0.496667 | valid loss 1.087503 | valid acc 0.453333 |\n",
            "| epoch  8 | train loss 1.085281 | train acc 0.526667 | valid loss 1.085691 | valid acc 0.466667 |\n",
            "| epoch  9 | train loss 1.083480 | train acc 0.533333 | valid loss 1.083882 | valid acc 0.486667 |\n",
            "| epoch 10 | train loss 1.081682 | train acc 0.550000 | valid loss 1.082076 | valid acc 0.506667 |\n",
            "| epoch 11 | train loss 1.079886 | train acc 0.560000 | valid loss 1.080273 | valid acc 0.540000 |\n",
            "| epoch 12 | train loss 1.078093 | train acc 0.573333 | valid loss 1.078472 | valid acc 0.553333 |\n",
            "| epoch 13 | train loss 1.076302 | train acc 0.593333 | valid loss 1.076674 | valid acc 0.566667 |\n",
            "| epoch 14 | train loss 1.074514 | train acc 0.633333 | valid loss 1.074878 | valid acc 0.626667 |\n",
            "| epoch 15 | train loss 1.072727 | train acc 0.683333 | valid loss 1.073084 | valid acc 0.660000 |\n",
            "| epoch 16 | train loss 1.070942 | train acc 0.750000 | valid loss 1.071292 | valid acc 0.693333 |\n",
            "| epoch 17 | train loss 1.069159 | train acc 0.776667 | valid loss 1.069502 | valid acc 0.746667 |\n",
            "| epoch 18 | train loss 1.067377 | train acc 0.806667 | valid loss 1.067713 | valid acc 0.773333 |\n",
            "| epoch 19 | train loss 1.065597 | train acc 0.820000 | valid loss 1.065926 | valid acc 0.800000 |\n",
            "| epoch 20 | train loss 1.063817 | train acc 0.826667 | valid loss 1.064139 | valid acc 0.820000 |\n",
            "| epoch 21 | train loss 1.062038 | train acc 0.843333 | valid loss 1.062354 | valid acc 0.833333 |\n",
            "| epoch 22 | train loss 1.060260 | train acc 0.860000 | valid loss 1.060569 | valid acc 0.840000 |\n",
            "| epoch 23 | train loss 1.058483 | train acc 0.870000 | valid loss 1.058785 | valid acc 0.853333 |\n",
            "| epoch 24 | train loss 1.056706 | train acc 0.876667 | valid loss 1.057001 | valid acc 0.860000 |\n",
            "| epoch 25 | train loss 1.054928 | train acc 0.883333 | valid loss 1.055217 | valid acc 0.880000 |\n",
            "| epoch 26 | train loss 1.053151 | train acc 0.886667 | valid loss 1.053433 | valid acc 0.886667 |\n",
            "| epoch 27 | train loss 1.051374 | train acc 0.890000 | valid loss 1.051650 | valid acc 0.893333 |\n",
            "| epoch 28 | train loss 1.049596 | train acc 0.893333 | valid loss 1.049865 | valid acc 0.900000 |\n",
            "| epoch 29 | train loss 1.047818 | train acc 0.893333 | valid loss 1.048081 | valid acc 0.900000 |\n",
            "| epoch 30 | train loss 1.046038 | train acc 0.896667 | valid loss 1.046295 | valid acc 0.893333 |\n",
            "Done!\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYkAAAEWCAYAAACT7WsrAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOydd3hURdfAfye9h1RIAqEmNAWUKiiCWBCw8lpQXyuCn4Id7L6I+trFioqvFaWIBVGKCgIqHSnSSWiBFNJ73d35/rg3sMaUDWSzCczvefbZe+/MvXNuPTNnZs4RpRQajUaj0VSHm6sF0Gg0Gk3TRSsJjUaj0dSIVhIajUajqRGtJDQajUZTI1pJaDQajaZGtJLQaDQaTY1oJXEKIyLvi8hTrpbDGYjIChEZ2wDH6SwiW0SkQETubQjZmjMiclBELnS1HJqTQ0Q+FZHnGuJYTV5JmB+DHBHxdrUszQ2l1F1KqWdP5hgiMkREjtSR51MRKReRQrvf1pMptxGZDCxXSgUqpd5qiAOKSDcRWSAieabyWS4iAxvi2K5CRBbb3duKKvf7/QYuq85nzhWY36JSEWljt+1CETno4P5TROQLpwnoJJq0khCRdsB5gAIub+SyPRqzvFOAl5VSAXa/nq4WyEHaAjtOZMfqnhER6QisArYB7YFo4DvgZxE55yTkdClKqUsr7y3wJX+/33e5Wr6GRkTca0gqApp067wW2U+IJq0kgJuBtcCnwC32CSLSRkS+FZEMEckSkXfs0u4UkV1mLW6niJxtblci0sku37EmWWXtRUQeEZE04BMRCRGRH80ycszl1nb7h4rIJyKSYqbPN7dvF5HL7PJ5ikimiJxV9QQdKKO9iPxmnstSEXnXvjYiIvNEJM2stf4mIt3rOL+HRCRdRFJF5Da7vCPMa1UgIski8rCI+AOLgWi7WmN0Pe4fItLOvO7jzOuUKiIP26V7i8gbZlqKuextl36FGOagfBHZJyLD7Q7fVkRWmTL/LCLh5j4+IvKF+VzkisgGEWlZjWy/AkOBd8xzixeRYBH53Lwfh0TkSRFxM/PfapY3TUSygCnVnPIUYI1S6gmlVLZSqsBsocwEXqpyTW4RkSTz2XjCTi43EXnUPN8sEflKREJruL51PT8rROTZ6q6Tmf5v8zyz7GVwFBFZKSKjzeVB5nmNNNeHicgWu3N60iwr3bzGwSdQ3kgR2Ww+D4dFZIpd2kIRmVgl/18icpW53EVEfhGRbBHZIyLX2uX7VETeE5FFIlKE8VxUx1vAGDEqA9XJFy0i35j344CYJkzzuX0cuM581raKyFAR2Wa37y8issFu/XcRudJc7mrey1wR2SEil9vlq1V2EQkUozX7lhj8412v9aIrpZrsD0gE7gZ6AxVAS3O7O7AVmAb4Az7AuWbaNUAy0BcQoBPQ1kxTQCe7438KPGcuDwEsGC+yN+ALhAGjAT8gEJgHzLfbfyEwFwgBPIHzze2Tgbl2+a4AttVwjnWVsQZ4FfACzgXygS/s0m839/MG3gC21HF+U01ZRwDFQIiZngqcZy6HAGfb7Xekjvt0rJxq0tqZ1322ea/OBDKAC830qRgVgUggAlgNPGum9QPygIswKjQxQBczbQWwD4g379UK4EUzbTzwg3lN3TGen6Aa5FsBjLVb/xz43rym7YC9wB1m2q3mNZwIeAC+1RwvDbitmu1DAaspa+U1+dBc7wmUAV3NvPeZ16S1eV8/AGaf4PNT23XqBhQCg81yXjfP70JH77d5/942lx83y3rJLu1Nu+c0EegABADfAjNrOP4QanjmzLQzzeehB3AUuNJMuxZYZ5e3J5CF8e74A4eB28x7dxaQCXSzO6c8YJB5bJ+anhXzOn1hbrsQOGguuwF/Ak+bZXYA9gOXmOlT+Pu76wuUAuEY7+RRjG9XoJlWYt5fT/PaPW4e9wKgAOhck+yV98jcfz127yc1vOs13u8T/YA7+4fxQawAws313cAD5vI5GB8aj2r2+wm4r4Zj1qUkyqt7OOzy9wJyzOUowIb5ka2SL9q8iUHm+tfAZAfP276MWIyX1s8u/Qv7B63Kvi3Mcwyu4fxK7K8ZkA4MMJeTMD6uQVWOOQTHlEQpkGv3+8xMa2fK1MUu/8vAR+byPmCEXdolHH/pPgCm1VDmCuBJu/W7gSXm8u0YyqaHA9d7BaaSwFAo5ZgfDnPbeGCFuXwrkFTH8SzA8Gq2dzGvQ4zdNWltl74euN5c3gUMs0uLwngX/vG81/b8OHCdngbm2KX5m+dfHyUxDPjLXF6C8RFda66vBK42l5cBd9sdo3NN5+TIM2eX943KZwTj45gDxJnrrwLTzeXrgN+r7PsB8B+7c/rckWcFozKTB3Tn70qif9XnA3gM+MRcnkKVdxf4HbgaGAD8DHwFDMeoVFRe1/MwKh9udvvNBqbUJLu57WNgOzCpSlq173pNv6ZsbroF+FkplWmuz+K4yakNcEgpZalmvzYYH54TIUMpVVq5IiJ+IvKB2UTOB34DWohh82sDZCulcqoeRCmVgmGXHi0iLYBLMey4/6COMqLNMortdjlst6+7iLxomiXygYNmUjjVk1XlmhVj1OrAqI2OAA6ZJoT62s9fVUq1sPvdUiX9sN3yIYxzw/w/VENaXfcyzW7Z/lxmYlQW5ohhwnpZRDwdOIfKGl1VeWJqOI/qyMT4qFelslJh/7zUJH9b4DvTtJCLoTSsQHUms9qen7rKibY/H6VUEUbNuz6sAeLFMOf1wmiJtTFNWv1MeSrLqnpdPao7p9oQkf6m6SRDRPKAuzCfd/PdnQvcJIaJcAzGswDGNe1feU3N63oj0Mru8HXdW8xyMoB3MFpK9rTFMM3al/F4Hee4EkMpDjaXVwDnm7+VZp5o4LBSyma3nyPP5UiMFknVgQX1etebpJIQEV+MpuP5Ytjb04AHgJ4i0hPjgsRK9Z3Lh4Fq7YUYL4if3XqrKumqyvpDGDWe/kqpIIwbCYYZ6zAQaiqB6vgMuAnD/LVGKZVcQ77aykg1y7CXuY3d8g0YpqwLgWCMGmrlvvVCKbVBKXUFhtlnPkaNBv55TU4Ue7ljgRRzOQXj5aourbZ7WSNKqQql1DNKqW7AQGAURv9WXWRi1G6rymN/7+q6Hksx7nlVrsV4DoqrSavKYeDSKkrXp4ZnqLbnpy5Ssbsv5nMW5sB+xzDP508ME9l2pVQ5RivuQWCfXSWvuvtswTCx1IdZwAKgjVIqGOMDaH+un2F8/IcBxUqpNeb2w8DKKtc0QCn1f/anUw85XsGo7fe223YYOFCljECl1Ihajl9VSazkn0oiBUPx2n+vHXkuP8Ro3S0So3/RyFjzu14tTVJJAFdi1Jy6YdROegFdMZpmN2M0zVOBF0XEX4yOykHmvv8DHhaR3mYnTScRqXw4twA3mDXw4Rg3ojYCMUw0uWJ0HP6nMkEplYrRqTtdjM5DTxEZbLfvfOBsjJfn8xMs4xCwEZgiIl6mxr+syr5lGLU/P+C/dZxPtZjHvlFEgpVSFRj9HpW1lqNAmJxAJ2MVnjJrvd0x7MJzze2zgSdFJMKsfT6NYVID+Ai4TYwOUDcRiRGRLg6cz1AROdOsTedjfPhtdeyGUsqK8cI8b3b2tcX42NVn2OIzwEAReV6MgQ2BYnSm3gw84uAx3jdlaGueT4SIXFFD3hqfHwf4GhglIueKiBdGzfhEvgkrgQkc/6itqLIOxn1+QIyBGAEYz+rcGqwBwLEBCPY/wTjfbKVUqYj0w6goHcNUCjbgNY63IgB+xGjx/Nt8Vz1FpK+IdD2B80UplWuWMdlu83qgQIzBL77md+YMEelrph8F2lX52K/GUPL9gPVKqR2YrR6Ot8LWYVRwJ5tyD8H4DsxxQNQJwB7gB1Om2t71ammqSuIWDDteklIqrfKH0cS7EaPmcBlGp3QScATD5ohSah7wPEaNowDjY105MuQ+c7/Kpub8OuR4A6O5lonRkbikSvq/MT5AuzHs+/dXJiilSoBvMIZBfnsSZdyI0QeThdERNRdDMYChfA5h1Ch2mvufKP8GDpomi7vMclFK7cZ4wfebTeiaRjdNlr/Pk8iskr4So/NtGYZp6mdz+3MYivAvjGGjm8xtKKXWYyiUaRg24JX8vTZaE60wPoD5GKaalfz9g1EbEzGGOe4H/sB4jj52cF+UUgkY/Wk9Mcx/qRjN+0uUUqscPMybGLXln0WkAOO+9q8hb13PT22y7gDuwTjHVAxT2InMT1iJ8fH+rYZ1MK7hTHPbAYw+rL+NRKpCDIbys/91xOhTmWpel6epvhb8OUbn9jHlrpQqAC4GrseomadxfJDKifImRmW2sgwrRqu1F8Y5ZmJUWisrWPPM/ywR2WTuU4TxzO8wW2FgmPAOKaXSzTzlGN+tS81jTgduNt/NWlFGJ8Q4jPv6PUa/TbXvek2I2ZGhcQIi8jQQr5S6qQGPORfYrZSqT43RZYgx1+UA4FlbrVGjaShE5GZgnFLqXFfLcirQVFsSzR6z6X8HMOMkj9NXRDqa5pbhGH0QdbWANJrTErNf5W5O8r3THEcrCScgIndidGItVkr9Vlf+OmiFYeMtxJjI839Kqc0neUyN5pRDRC7BGBp/FMOEpmkAtLlJo9FoNDWiWxIajUajqZFm58QuPDxctWvXztViaDQaTbPizz//zFRKRdR3v2anJNq1a8fGjRtdLYZGo9E0K0TkUN25/ok2N2k0Go2mRrSS0Gg0Gk2NaCWh0Wg0mhrRSkKj0Wg0NaKVhEaj0WhqRCsJjUaj0dSIVhIajUajqZFmN09Co9FomiMVVhtZheVkFpaRWVh2bLmozHHnyMO6tqRnm5rinDkHrSQ0Gs0pj1KKonIrmQVlZBWVkWl+oLMKy8kuKsdiqzMmVb2wKcgrriCjsIysQqO8vJKKGvOLg7EkI4N8tJLQaDSnPsXlFvYeLWRPWj670wo4ml9a9071pLTCduwDnVlYRpmlekUQ6OOBl3vDWt5FIMjXk/AAbzq3CmRQgDdh/t6EB3oR5u9NhPkfHuiNv5c74qiWcAFaSWg0Gqdhsdo4mFXE7rQC9qQVsDutgL1HC0jKLqbSAbWvpzvRLXxwa+APpZeHG2EB3nSMCCA80Jswfy/CA7wJCzD+wwO8CfX3wstDd83WhlYSmtOaknIrGw5ms2pfJusPZNfLPqypHYtNcSSnhHKzBu8m0D7cnzOigxl9dms6twqkc8tAYkP9cHNrujXpJoVSjtumGgitJDSnFRarja1H8lidmMmqfZlsOpRLudWGp7vQq00LOkYEuFrEUwYRuLBrSzq3DKRzq0A6RQbg4+nuarGaBxUlkLEH0nfC0R3m/064+FnocW2jiqKVhOaURilFQnohfyRksnpfJuv2Z1Ngtha6RQVxy8C2DOoUTr/2ofh56ddB08jYrJBz0E4RmP/Z+0GZfSju3hDRGToOhaCYRhdRvxWaU5LE9AK+35LCgq0pHMoqBqBtmB+jekYzqFMY53QIIyzA28VSak4blILCdEjfYbQI0it/u8FSYmYSCG0Pkd3gjNHGf8vuENIe3F33qdZKQnPKkJxbwg9bU1iwJYWdqfm4CQzsGM74wR05Ly6cNqF+rhZR0xTJ2AMlOQ17TJsFsvb9vXVQnHU83T8SIrtCn9tMZdANIrqAl3/DytEAaCWhadZkF5WzaFsqC7aksP5gNgC92rTg6VHdGNUjisggHxdLqGmypG6FZVMhcanzyvD0h8gu0HmE0SqobB34hzuvzAZGKwlNs6K0wkpieiE7U/JZvD2V3xMysdgUnSIDeOiieC7vFU3bsKZXG9M0ITIT4NfnYOd88A2BC6dAVM8GLkQgpB20aAtuzXuIrVYSmiaJzaZIyi4+Nr5+79ECdqflczCrGKvNGGAfHezDHee154qeMXSNCmzSE5I0TYDcw7DyJdgyCzx8YPBkGDgBfIJdLVmTRisJTZPgcHYxqxIz2ZSUYyqFQkoqrMfSY0P96NwqkBFnRh0bX98xIkCPr9fUTWEG/PE6bPifsd5/PJz7IAREuFauZoJWEhqXkFVYxup9Wazel8mqxCySso0RSKH+XnRpFcj1/drQpVUgnVsFERcZgL+3flQ19aQ0D1a/A2unQ0Ux9LoRzn8EWrRxtWTNCqe+eSIyHHgTcAf+p5R6sUp6W+BjIALIBm5SSh1xpkwa11BUZmH9wWxWJWSyal8Wu1LzAQj09qB/hzBuG9SOQZ3CiYsM0Gajk6Eoy26Y5Q7IT3G1RK5BKUjZZIxa6n4VDH0CwuNcLVWzxGlKQkTcgXeBi4AjwAYRWaCU2mmX7VXgc6XUZyJyAfAC8G9nyaRpeGw2RU5xOVlFlS6Qy02nasddIacXlLEzJR+LTeHl7kbvtiE8fHE8AzuF0yMmGI8Gdq52WlBeDBm7j8/ErVQMRenH8/iGQkhbkNP0+rY7D857CKJ7uVqSZo0zWxL9gESl1H4AEZkDXAHYK4luwIPm8nJgvhPl0TQAZRYrK/dk8P3WFNYfyCa7qPxYR7I9bgKh/t6Em87Uxp7XgUGdwujTNhRfL+2awWFsVmP2beVY+0qlkL0fMK+7h48xxj7uouNj7iO7Q0Bko/v5OR1RFgvlhw5RtncvpXv3UrY3gfJ9+1Dl5Q7t7+bvj3dcJ7zj44/9PKOjkWpGRSmlGr2l7UwlEQMctls/AvSvkmcrcDWGSeoqIFBEwpRSWfaZRGQcMA4gNjbWaQJrqsdqU6zbn8X3W1JYvD2V/FILof5eDOkcQXSwL2EBXoQFHFcI4QHetPD11J3K9UEpKEj754zcjD1gMd1oixuEdjDG2fe41piMFdndmKXrphWvs1FKYUlLoywh4Z8KocKMFeHmhle7dnjGdQI/X4eOa8nJJn/zJli0+Pg2H0/yooM4GuXD4Qg39oWVsyu4iHuGPs5VcVc54/RqxNW9gQ8D74jIrcBvQDJgrZpJKTUDmAHQp0+ff1ZbNQ2OUoq/juTx/ZYUfvwrhfSCMvy93Lmkeysu7xXNoE7heGoz0YlRmg/pu/6pEOxn/Qa0MloEfceak7C6Gq0FT8c+PJqTw5qfX0UZ7KV0bwKqoOBYnvKwQApah5BxYUdSWnpyINRKQnAx6dZsCiuS6lfgIPAtc6d1BsRmKGIzrXTIzKft5ly6FR//JFZY90Ajd604U0kkA/bDCFqb246hlErBaEkgIgHAaKVUrhNl0tTBwcwivt2czIItyRzMKsbL3Y0hnSO4olcMF3SJbBqmImuFYW6pKHa1JHVjsxmy2iuEPLsGtlegoQC6XWG0Clp2M0xGfqGuk/k0ocJWQXb+UbL3bKNw1w4qEvYh+5PwPnQU3+yiY/lKvIWkCDjUCQ5HuJEUYawX+ZYglBLiE0KoTyihPqF09unIOb6hhHiH4Onu6ZAcXm5ehPqEEuYbduw/2CsYdzd3o/WSkUHZXkNh+Q8a6KzLUSPOVBIbgDgRaY+hHK4HbrDPICLhQLZSygY8hjHSSeMilu48yt1fbsJis3FOxzDuHtKJS85oRbCvYw97g6OU8UG1r20f3QmZe8FWcyjIJombB4THQ5v+pr8es3XQIlb3GzQwFdYKDuQf4HDBYbJLs8kqySK7OIvyI0fwPJiCf1IWISkFRB0tJzoL3BUEABY3OBIOydHuZPcOpqBNKOXto/CKiiLUN4wwnzA6+oYS5nP8Y97CuwUebs77jIoInpGReEZGEnDuIKeVUxtOOzullEVEJgA/YQyB/VgptUNEpgIblVILgCHACyKiMMxN9zhLHk3tLPwrlfvmbKZ7dBAf/LsPrYIb2edRSc5xZXCsk3YXlOUfzxPcxqhlV3bQ+gQ1rownhBjj8sPiwMPL1cKcUiilSClKISEnwfjlGv8H8w4SklNBnwRFbIaiTYaifyb42PUjF0UEUNI2hqwhrfGI64h/l26ExHXnvIBI/D399TBsO0Sp5mXi79Onj9q4caOrxTil+HbTER6et5XebUP4+Na+BPo4seVQUWoO3axiky9IPZ7HJ/jvppdKm7x2n3Dakluae0wJVP4n5iZSVHHcLBTtH01Pt1iGLc+hzfLdiMWKBAfhE98Zn86d8Y6Pwyc+Hq9OcbgHnH7+vUTkT6VUn/ru5+qOa42Lmb0+ice/28Y5HcL43y19Gi7wzt+CqdgphOx9VYKpxEP7848P22zZDQKjtAnmNKXUUsq+vH0k5iT+TSFklGQcyxPsHUxcizgu63AZcSFxxIfE08EtktLP5pA9cyaqooIW//oXYXeOxTMmRrcKThKtJE5jPll1gGd+2MnQzhG8d1PvEwsteSyYys6/T+yqGkwlpJ3RIuh+1XGFENrBpcFUNK7FpmzszdnLutR1bM3YSkJOAkkFSdjMSoSXmxcdW3TknOhziGsRR1yI8YvwjTj24bcVF5P9+UxSP/oIW2EhQSNHEjFxAl5t27ry1E4p9Bt6mvLein28tGQ3l3RvyVtjzsLbox4KwmqBrbNg2zyjpfC3YCoRhomoMphKZDfDn34TDKaiaVyUUhwpOMLatLWsS13H+tT15JQZw35bB7Smc2hnhrcfTqcWnYgLiSM2MLbGTmFbeTm5c78i84MPsGZmEjB0KBH334dP586NeUqnBVpJnGYopXhjaQJvLkvg8p7RvHZtT8fnO9hshg/+5c9DViJEdIXOl9r1H3TXnjU1fyOzJJP1qetZl7aOtSlrSSkyfElF+kZybsy5DIgeQL9W/Wjl38qh4ymrlbwFP5D59ttUpKTg17cvEW+9hd/ZZznzNE5rtJI4jVBK8eKS3Xywcj/X9G7Ni6N74O7IrGiljOhdy6ZC2l+GcrjuS+gyUvcdaP5BqaWUhfsXMnfPXHZl7wIg0CuQfq36cesZt9I/qj/tg9rX2VeglKIiOYUyczJb2d69lGzdSkVyMj7du9Nq6lT8Bw3UfQ5ORiuJ0wSbTTH1x518uvogNw2IZerlZzjmNuPQGkM5JK02omxdNQPO/Jd2A6H5B+nF6czZPYd5e+eRW5ZL55DO3Hf2fQyIGkDX0K641/LMWHJyjk0Yq5zpXJaQgK3o+Oglz+hovOPjiZw8mcCLL9LKoZHQSuI0wGpTPPHdNuZsOMzYc9vzxMiudb9gqX/Br89Cws8Q0BJGvgZn3azH+mv+wY7MHczcNZOfDvyEVVkZ2mYoN3W7iT4t+9T6nFWkpJC/aBF5CxdRtmvXse3uwcF4x8cTfOWVhsO7uDi84+NwDwhojNPRVEEriVOcxPQCnvhuO+sOZDPxgk48eFF87Qoia58R/3fHt+DTwoj/2288ePk1lsiaZoDFZmH54eXM3DmTzemb8ff05/ou13ND1xtoE1hzUB9LTg4FS5aQt3AhJRv/BMC3Z08iHnoQny5d8Y6PxyMyQrcSmhBaSZyilJRbefvXBD78fT9+Xh68NPpMrutbhwfdzAT44HxjefAkOGcC+LZwvrCaZkN+eT7fJXzHrF2zSClKISYghsl9J3NVp6sI8Kq+pm8rKqLg11/J+/FHilatBosFr04dibj/PoJGjsSrjY4U15TRSuIUZNmuozz9/Q6Sc0sYfXZrHhvRhfAA79p3spTDN2MNc9L43wyfQhqNyaH8Q3y560vmJ86nxFJCn5Z9mNxvMkNaD/lHX4Oy2ahITqZ0x04Kfv6Jgl+Xo0pL8YiOIuy2WwkaNQrv+DpatJomg1YSpxDJuSU8s2AHP+88SlxkAHPHDaB/hzDHdl7+PKRuMUYtaQWhwRhdtD5tPTN3zuS3I7/h7ubOiPYjuLHrjXQL6waAJTubosrRRwkJhlvthERUseGh1z0khBZXX0XQqFH49upVbSAdTdNGK4lTgAqrjY//OMAbSxMAeGR4F+44tz1eHg6+kAd+g1VvQu9boeso5wmqaRaUWctYtH8RM3fNJCEngVCfUMb3HM+1HUbjs2k3xTMWkJTwKqV7E7BmZh7bzz0kBO/4eFqMHn3MT5JPt26Ip4u8CGsaBK0kmjkbDmbz5Hfb2XO0gAu7tmTK5d1oHVKPTubibPh2PIR1gkv+6zxBNU2ejOIM5u6Zy7y988guzSY+JJ6pA6YwNDeKkvk/k7PkKqy5uYiPD96dOhEwePAxZeAdF4d7eLg2IZ2CaCXRTLHaFE99v51Z65KIaeHLhzf34aJuLet3EKXgh/ugKAPGzNauM04jlFJklGQcc7O9PWs7y5KWYbVZOT9mMLd4n0+bNQfJf/0dUtPSEF9fAi+4gKBRIwkYNAjx0kOhTxe0kmiGKKWYsmAHs9YlMfbc9jx4cfyJeW/dPBN2LYCLpkJ0r4YXVNMkKCgvYF/uPvbm7D3mWTUxN5G8srxjeSJ8I7g1+BJG7gtC5q6ifP8ysj08CDj3XIIeeojAC4bi5q8rEacjWkk0Qz74bT8z1x5i/OAOPDai64kdJDMRFj8C7QfDORMbVkCNS7HarGw4uoHFBxazJmUNqUXHY3X4e/rTqUUnLmp7keFZtUUnYndlUzLjM0q2fE8F4Ne3L6E330zgJRfjERLiuhPRNAm0kmhmfL8lmRcX7+ayntE8MrzLiR3EUg7f3AHuXnDVB6BHnDR7lFJsz9zOogOL+OngT2SUZODn4ce5Medybedrj3lWjfaPPtZvULxpMxnPv072xo14xsQQOelhgkaMwDMqysVno2lKaCXRjFi7P4tJ8/6iX/tQXr2mh2O+l6pjxX+N4a7XzoSg6IYVUtOo7M/dz6IDi1h0YBGHCw7j6ebJeTHnMaLDCM5vfT4+Hv8MQ1u6ezcZ096gcOVK3MPDafnUk4Rcc43uZ9BUi1OVhIgMB97EiHH9P6XUi1XSY4HPgBZmnkeVUoucKVNzJeFoAeM+30hsmB8f/rtP/eI/2HPgd/jjDTj7Zuh2ecMKqWkU0orSWHxgMYsPLGZX9i7cxI2+rfpy55l3MqztMIK8qo/9XX7wIBlvv0P+woW4BQUR8eCDhN50I25+2uWKpmacpiRExB14F7gIOAJsEJEFSqmddtmeBL5SSr0nIt2ARUA7Z8nUXEnPL+XWTzbg7enOJ7f2JdjvBMedF2fDd+MhrCMMf7Hu/Jomx3cJ3/Hc2ucot5VzZviZPNL3ES5pdwkRfjXH8ahISyNz+nvkfvMN4uVF2PjxhN1+G4HzJLcAACAASURBVO7BOma4pm6c2ZLoByQqpfYDiMgc4ArAXkkooLLaEwykOFGeZklhmYXbPt1ATnE5X40/hzahJ1jrUwp+vB8Kj8Idv+jhrs2Mcms5L6x/ga/3fs2AqAE8NeApYoNqnxlvzc0l84MZ5Hz5JUopQsaMIXz8ODwidGAojeM4U0nEAIft1o8A/avkmQL8LCITAX/gwuoOJCLjgHEAsbGnj8uICquNe77cxO60Av53Sx/OiDmJmt+WL2Hn94ZX15izG0pETSOQVpTGgyseZFvmNsaeOZYJvSbUGpsBoPCPVaQ+9hiWrCyCL7+c8AkT8God00gSa04lXN1xPQb4VCn1moicA8wUkTOUMiOhmyilZgAzAPr06aNcIGejo5TiqfnbWbk3gxeuPpOhnSNP/GCZibBoMrQ7Dwbe23BCapzOhrQNPLzyYUotpUwbMo0L21ZbjzqGrbSU9FdfI+eLL/Dq2JF277+Hb/fujSSt5lTEmUoiGbD3Adza3GbPHcBwAKXUGhHxAcKBdCfK1Sx459dE5mw4zIShnRjT7yRaT/kp8MXVhnfXq97XEeWaCUopPt/5OdP+nEZsUCyfXPIJHVp0qHWf0p07SZ40mfJ9+wi56SYiH34IN59/jm7SaOqDM5XEBiBORNpjKIfrgRuq5EkChgGfikhXwAfIcKJMzYJv/jzCa7/s5eqzYnjo4vgTP1BhBnx+hdFhfcv3ENy64YTUOI3iimL+s/o/LDm4hGGxw3hu0HM1xmoAUFYrWR9/TMZbb+PRogVtPvyQgPPObUSJNacyTlMSSimLiEwAfsIY3vqxUmqHiEwFNiqlFgAPAR+KyAMYndi3KqVOC3NSTaw/kM0j3/zFoE5hvDi6x4k7TCvJgS+ugtzDcNM3ENO7YQXVOIVD+Ye4f/n97M/bz/1n38/tZ9xe6zNQfiSZlEcfoWTjnwRefDGtnpmiZ0lrGhSn9kmYcx4WVdn2tN3yTmCQM2VoTiilmPrjDloF+/DeTb0dd/VdlbIC+PIaSN8NN8yBdvoSNwdWHl7JY78/hrubO+9d+B4DowfWmFcpRf6CBaRNfRaAqBdeIPjKK7QXVk2D4+qOa40dS7ansT05n9eu6UmQzwnOhagogdljIHkTXPsZdKq9o1PjeiqsFby75V0+2v4RXUO78sbQN4gOqHkmvDU3l9Qpz1CwZAm+Z59N9Msv4dVamxI1zkEriSaC1aZ47Ze9dIzw58qzTnCooqUcvroFDv4BV8+Arpc1rJCaBmd/7n4e/f1RdmXvYnTcaB7t92i1rjQqKVq9mpTHHseSlUXEAw8QNvYOxF0PRtA4D60kmgjfb0kmMb2Qd284G/cT8clktcC3d0LCTzDqDehxbcMLqWkwlFLM2j2LaX9Ow8/DjzeHvskFsRfUmN9WWkr666+T8/lMvDp0oN277+J7hh7aqnE+Wkk0ASqsNt5YmkC3qCAuPaNV/Q9gs8EP98LO+UZ0uT63NbyQmgYjvTidp1c9zaqUVZwXcx5TB00l3De8xvylu3eTMmkSZQmJhNxwA5GTHsbN17cRJdaczmgl0QT4auNhkrKL+fjWPvX37KoULJ5szKge8jicc49zhNQ0CEsPLWXKmimUWcp4sv+TXNv52ho7m5XVSvann5L+xpu4twimzYczCDjvvEaWWHO6o5WEiymtsPL2skTOjm1R/1nVSsHSKbDhQ2Mm9fmTnSKj5uQpLC/kxfUv8v2+7+ke1p0XznuB9sHta8xfkZJCyiOPUrxhA4EXXUirqVP10FaNS9BKwsV8uS6JtPxSXr+uZ/2HL/7+Gqx6A/rcYYQg1cMfmySb0zfz2O+PkVqUyrge47ir5114utU8ei3vhx9JmzoVrFainn+e4Kuv0kNbNS5DKwkXUlRmYfryRAZ1CmNgx5pt0tWSsBR+fRZ6XAcjXtUKogmilOL9re/z/l/vE+UfxWfDP6NXZM2xxK15eaQ9M5X8RYvwPessY2hrmzY15tdoGgOtJFzIp6sPklVUzkMXd67fjoUZMP//ILI7XPaWDj/aRPk9+Xemb53OyA4jeWrAU/h71uyevWTbdo5MnIglM5OI++8jbOxYxEO/nhrXo59CF5FXUsEHK/cxrEskZ8fWw9asFHx/D5Tmwc3fg6d24NYUqbBV8MqGV2gX1I5nBz6Lp3vN5iVrXh5H7r0X3IR2s2fje+YZjSipRlM7Wkm4iA9/209+qYUH6+vAb8P/jLkQl74MLbs5RzjNSfPVnq84mH+Qdy54p1YFoZQidcoULBkZtJs9SysITZND2ylcQGZhGR+vOsDIHlF0j65HIKH0XfDzkxB3MfQb5zwBNSdFbmku07dM55yocxjcenCtefPmf0/B4iVETJyI75lnNpKEGo3jaCXhAt5bsY/SCisPXFiPVkRFKXwzFrwD4Yp3dUd1E+a9re9RWFHIpL6TavfgeugQR599Fr++fQkbe0cjSqjROI42NzUyaXmlzFx7iKvPbk2nyJpjBPyDZc/A0e1wwzwIOIkodRqnsj93P3P3zOWa+GuIC4mrMZ+qqCB50mTw8CD65Ze0/yVNk0UriUbm7V8TUEpx37CaPyD/IGEprJ0O/cZD/MXOE05z0ryy8RX8PPy4u9fdtebLmD6d0r/+IuaNaXhGRTWSdBpN/dHmpkYkKauYuRsOc33fWNqE+jm2U1GmOdy1G1z0jHMF1JwUvx/5nT+S/2B8z/GE+oTWmK9440ayPphB8FVXETR8eCNKqNHUH92SaETeWLYXdzdhwgWdHNvhb8Nd54OndurWVKmwVfDKxleIDYzlhi5Vo/Qex5qfT/LkyXi2bk3LJ55oRAk1mhNDK4lGIjG9gPmbk7nj3Pa0DHJwbsOG/8HeJTD8RWip3UI3Zb7a8xUH8g7w1tC3ahzyqpQibcozWI6m0272LNwDap5cp9E0FbS5qZF4/Ze9+Hq6839DHGxFpO82hrt2uhD63+Vc4TQnRV5ZHtO3TGdA1ACGtBlSY778BQvIX7SIiIkT8O3Ro/EE1GhOAqcqCREZLiJ7RCRRRB6tJn2aiGwxf3tFJNeZ8riKvUcLWLQtjdvPbU+ov1fdO1jKjOGuXgFwxXQ93LWJ48iQ1/LDh0mb+ix+ffoQduedjSyhRnPiOM3cJCLuwLvARcARYIOILFBK7azMo5R6wC7/ROAsZ8njSt5bsQ8/L3duH1Sza+i/sfQZOLoNxsyFwJbOFU5zUuzP3c+c3XP4V9y/iA+pft6Lqqgg5eFJ4Oamh7tqmh3ObEn0AxKVUvuVUuXAHOCKWvKPAWY7UR6XcDi7mAVbU7ihXywhjrQi9v0Ka9+FvmOhsx750tR5deOr+Hr4cs9ZNQd7ynzvfUq2biXqmSl4Rkc3onQazcnjTCURAxy2Wz9ibvsHItIWaA/8WkP6OBHZKCIbMzIyGlxQZ/LBb/twF2HseR3qzlxeBD/cB2FxcPFzzhdOc1L8kfwHvyf/zl0976pxyGvxn3+S+f77BF95JUEjRjSyhBrNydNUOq6vB75WSlmrS1RKzVBK9VFK9YmIiGhk0U6c9PxSvtp4hNG9W9Mq2IERTStegNwkuOxNPdy1iWOxWXhlQ+1DXsuTkkh+eBKeMTG0fPLJRpZQo2kYnKkkkgH7iCmtzW3VcT2noKnpoz8OYLHauOt8B1oRKVtgzbtw9i3QbpDzhdOcFPP2zmN/3n4e6vNQtUNeS7Zu5eD1Y1AlJcS8MU0Pd9U0W5ypJDYAcSLSXkS8MBTBgqqZRKQLEAKscaIsjU5ecQVfrD3EqB7RtA2r4wNhtcAP94JfuJ5V3QzIK8vj3S3v0r9Vf4a2GfqP9IKlSzl0y624+fvTdvYsfLvrOS6a5ovTlIRSygJMAH4CdgFfKaV2iMhUEbncLuv1wByllHKWLK7gszUHKSq38n9DOtaded37kLoVRrwMvjrYfVPnlQ2vUFBeUO2Q1+yZX3Bk4r14x8fTbs5svNs7OKJNo2miOHXGtVJqEbCoyranq6xPcaYMrqCozMLHqw5wYddIukYF1Z455yAsfx7ih0O3KxtFPs2J813Cd3y/73vG9RhH59DjYWeVzUb6K6+S/cknBAwbRsyrr+Dmq/uVNM0f7ZbDCcxen0RucUXds6uVgoUPgbjBiFf1pLkmzp7sPTy/7nn6t+rP3T2Pe3m1lZWR8sijFCxZQsiNN9Ly8cf0XAjNKYNWEg1MmcXKh7/vZ0CHUHq3rcN0tO1rSFwKw1+CFm1qz6txKYXlhTy88mGCvIJ4cfCLuLsZSsCSk8OReyZQsmkTkZMnE3rbrbUGGtJomht19kmIyGUi0lSGyjZ5vtuUzNH8Mu4ZWkcrojgbljwKMb2hn3bT0JRRSjFlzRSSCpJ4efDLhPuGA4arjUNjbqB0+3Zi3phG2O23aQWhOeVw5ON/HZAgIi+bI5E0NWCx2nhv5T7OjAnm3E7htWf++SkozYXL3gI3bZpoyszePZufDv7ExLMm0qdVHwBK/vqLg9ddjzUnh9hPPtZxITSnLHUqCaXUTRg+lfYBn4rIGnMGdKDTpWtmLNqexqGsYu4Z2rH2GuX+lbDlCxg4EVqd0XgCaurNtoxtvLLxFc5vfT63n3E7AAW/LufQzbfg5udH29mz8evd28VSajTOwyEzklIqH/gaw/9SFHAVsMl0yqfBMElMX55Ixwh/Lu7WquaMFSXw4/0Q0h7Of6TxBNTUm7yyPB5e+TCRvpE8f+7zuIkb2V9+yZEJE/Du1MkY4tpBD3HVnNrU2XFtzmm4DegEfA70U0qli4gfsBN427kiNg+W70lnd1oBr13TEze3WloRv70C2fvh5u+1640mjE3ZeOKPJ0gvSefz4Z8T5BnI0VdeIfujjwkYOpSY117Fzc/BELQaTTPGkdFNo4FpSqnf7DcqpYpF5A7niNW8UErxzq+JxLTw5fJetXj5TNsOq96EnjdAhyGNJZ7mBPhk+yesPLKSx/o9RvegeFIefpj8RYsJuWEMLZ94Qg9x1Zw2OKIkpgCplSsi4gu0VEodVEotc5ZgzYl1B7LZlJTL1Cu64+legwXPZjU8vPoEaw+vTZyNaRt5e/PbXNLuEq5tdSlJt99ByZ9/EjnpYUJvv12PYNKcVjjSJzEPsNmtW81tGpPpK/YRHuDFtX1qmeuw4SNI3mjEq/YPazzhNPUisySTyb9Npk1gG56KHcehG26k9K+/iHn9NcLuuEMrCM1phyMtCQ8zaBAASqly02GfBth2JI/f9mbwyPAu+HjWYII4uhOWToGOF8CZ1zSqfBrHsdqsPPrbo+SX5/Ne64c4etPtKIuF2E8+xq9PH1eLp9G4BEdaEhn2DvlE5Aog03kiNS+mr0gk0MeDmwbEVp+hOBtmXw/egXDFu9r1RhNm+tbprEtbxwtyNeqep3Dz8aHdrC+1gtCc1jjSkrgL+FJE3gEEI9rczU6VqpmQmF7Ikh1p3DOkE4E+/4wpgLUC5t0CBalw6yII0qErmyrfJXzHjL9m8PDhM2k960u8u3alzfvv4dGMglxpNM6gTiWhlNoHDBCRAHO90OlSNRNeWrIbX093bhvUrvoMPz8JB36DK9+DNn0bVTaN4yw+sJj/rP4PkzZG0feXzQQMGULM66/pIa4aDQ46+BORkUB3wKey404pNdWJcjV5lu9O55edR3n00i6EBXj/M8OmmUaciAH3QK/qw1tqXM+Kwyt4/PfHufZoO/r+kkCLa/5Fq//8B/HQvi81GnDMwd/7GP6bJmKYm64B2jpZriZNaYWVKT/soGOEP7cPqmbGbdI6+PEB6DAULjqtdWmTZk3KGh5a8RA9/OK49sc8vOPjafX001pBaDR2ONJxPVApdTOQo5R6BjgHiHeuWE2bGb/t51BWMVOvOAMvjyqXMO8IzL0JglvDvz4Gd/3BaYpsTt/Mfcvvo21wW6Zu64w1I4OoZ6cintX0LWk0pzGOKIlS879YRKKBCgz/Taclh7OLeXd5IiN7RDGoqqfXihKYcyNUFMOYOeAX6hohNbWyM2sndy+9m0i/SN6NvJfir74h5Kab8O3Z09WiaTRNDkeUxA8i0gJ4BdgEHARmOXJwERkuIntEJFFEHq0hz7UislNEdoiIQ8d1Jc/8sBN3N+HJkV3/nqAULJhoxKq++kOI1F7VmyKJOYmM/2U8QV5BfDj0PUqefx2Pli2JuO8+V4um0TRJarWFmMGGlimlcoFvRORHwEcplVfXgUXEHXgXuAg4AmwQkQVKqZ12eeKAx4BBSqkcEYk8iXNxOr/uPsrSXUd57NIuRAVXcc636k3YNg8ueBK6jHCNgJpaScpPYtwv4/Bw8+DDiz/Ec9ZCchMSaD19Ou4B/q4WT6NpktTaklBK2TA+9JXrZY4oCJN+QKJSar85Y3sOcEWVPHcC7yqlcszjpzsseSNTWmFlyoKddIoM4LaqndV7fzZmVHe7Es572CXyaWontTCVsT+PpcJWwYcXfUjLLCuZ098jcPhwAi8Y6mrxNJomiyPmpmUiMlrq77QmBmPiXSVHzG32xAPxIrJKRNaKSJMN7/XByv0kZRcz9fLuf++szkyAb8YawYOunK5nVDdBMksyufOXOykoL+CDiz6gY4uOpP1nCuLtTcvHH3O1eBpNk8aRoTfjgQcBi4iUYgyDVUqpoAYqPw4YArQGfhORM03z1jFEZBwwDiA2tgb3F04kKauY6SsSGdUjioH2ndUluYbLDXdPuH4WeGmTRVMjuzSbO3++k/TidD646AO6hXUj95tvKF6/nlbPPINnZJO2cGo0LseRGdcnGqY0GbB3i9ra3GbPEWCdUqoCOCAiezGUxoYqMswAZgD06dNHnaA8J8zUH3eYndXdjm8sL4bZYyDnINy8AFo0vvLS1ExxRTFf7PqCT7d/Sqm1lHeHvctZkWdhyczk6Muv4NunNy2u+ZerxdRomjyORKYbXN32qkGIqmEDECci7TGUw/VA1anH84ExwCciEo5hftpfl0yNybJdR1m6K53HR3ShVbCPsdFSBnNvhKQ18K+PoN0g1wqpOUa5tZx5e+cx468ZZJdmM6TNEO49617iQuIAOPrfF1DFxUQ98wzi5lD0Xo3mtMYRc9Mku2UfjA7pP4ELattJKWURkQnAT4A78LFSaoeITAU2KqUWmGkXi8hOjDgVk5RSWSdwHk6hcmZ1nH1ntdUCX98O+36Fy9+BM0a7VkgNABabhR/2/cB7W98jtSiVvq368uZZb9IrstexPIUrV5K/aBHhEybg3bGjC6XVaJoPjpibLrNfF5E2wBuOHFwptQhYVGXb03bLCqO/40FHjtfYvL9yH4ezS5h1Z38j4pzNBvP/D3b/CMNfgrP/7WoRT3uUUvxy6Bfe2fIOB/IO0D2sO1MGTuGcqHP+FiDIVlRE2jNT8erYkbBxd7pQYo2meXEiPiOOAF3rzNXMMTqr93FZz2gGdgw3JsstfBC2fQUXPAUD7nK1iKc1SilWp6zmrc1vsTNrJx2COzBtyDSGxQ6rNnpcxltvU5GSQtsvv8DNS8fM0mgcxZE+ibeBys5iN6AXxszrU5pnftiBp5vwxIiuhoL4+Un48xM490EYrOdCuIq8sjyWJS1jfuJ8NqdvJto/mucGPceoDqNwd6s+MmDJtu1kz5xJi+uvw69370aWWKNp3jjSkthot2wBZiulVjlJnibB0p1HWbY7nSdGdDU6q5e/AGvegX7jYdjTdR9A06CUWEpYeWQli/Yv4vfk37HYLMQGxvJov0e5Jv4avNxrbhkoi4XUp5/GIyyMyAebpFVTo2nSOKIkvgZKlVJWMNxtiIifUqrYuaK5BptN8ezCncRFBnDroHaw6i1Y+SL0ugmGv6gnyzUSFbYK1qasZdGBRfya9CvFlmIifCMY02UMI9uPpFtYt2rNSpVYCwoo+GUped9+S9muXcS8+SbuQQ0xtUejOb1wREksAy4EKiPS+QI/AwOdJZQr+TMph0NZxbx5fS88N30CvzwF3a+Gy98CPWTSqdiUjS3pW1h0YBE/H/yZnLIcAr0CubT9pYxoP4LeLXvXaFICsJWVUbhiJfk//kjhypWo8nI8W7cmctIkAi++qBHPRKM5dXBESfjYhyxVShWKyCkb13HhX6l4e7hxiWUFLHwI4ofD1TOglo+T5sRQSnG44DBrU9eyNnUtG9I2kFuWi4+7D0PaDGFE+xEMihlUpzmpaO068hcupOCXX7AVFuIeHk6L664jeNRIfHr0qLXFodFoascRJVEkImcrpTYBiEhvoMS5YrkGq02xaFsqD8TsxmfhVGh/HlzzmeF2Q9MgZJZksi51HWtT17IudR2pRakAtPRryeDWgxkYPZAhbYbg71mzixOlFKVbt5L340LyFy/GmpWFW0AAgRdfTPCokfj166ejy2k0DYQjb9L9wDwRScHw29QKI5zpKcfGg9mUFOQw1vYSxPSG62eDp4+rxWrW2JSNNSlr+CP5D9amriUxNxGAIK8g+rXqx+1n3M6AqAG0DWpbZ42/LCHBUAwLF1Jx5Aji5UXA0KEEjRpJwODBuHlXE2tco9GcFI5MptsgIl2AzuamPaavpVOOhdtSudprLR7WErj0JfAOcLVIzZaiiiLmJ85n1q5ZJBUk4ePuw1mRZzGqwygGRA+gS0iXWvsXKik/kkz+okXk//gjZXv3grs7/uecQ/g99xB40YW4B+h7pNE4E0fmSdwDfKmU2m6uh4jIGKXUdKdL14gYpqY05vmtgsCuEH22q0VqliQXJjNr1yy+TfiWwopCekT0YOJZExkaOxRvd8dq+pasLPKXLCH/x4WUbN4MgO9ZZ9HyyScJGn4JHuHhdRxBo9E0FI6Ym+5UStkHHsoRkTuBU0pJrD+QTYuifbT33gWDn9dDXeuBUorN6Zv5YtcXLEtahiBc3PZibux2Iz0jHI8bXbJtGxlvvU3R6tVgteIdH0/Egw8SNGIEXq2rhiLRaDSNgSNKwl1ExPSzVBmW9JTza7BwWwpjPH9HuXkgPU7JLpcGp8JawZKDS/hi1xfszNpJkFcQt3W/jeu7XE8r/1b1OlbB0qUkPzwJ98BAwu64g6CRI/HpHO8kyTUajaM4oiSWAHNF5ANzfTyw2HkiNT4Wq41fth3hF88/kLjhEBDhapGaHEopMksySchJICE3gb05e1mTsoaMkgzaB7fnqQFPMarDKPw86z86OnvmFxz973/x6XEmbaZPxyMszAlnoNFoTgRHlMQjGFHhKj3a/YUxwumUYf2BbM4s2UCQVw70utHV4ricooqiY8ogIcf4JeYmklt2PGBguG84Z4SfwXWdr2Ng9EDcpP4TDZXNRvorr5L9yScEDBtGzKuv4Obr25CnotFoThJHRjfZRGQd0BG4FggHvnG2YI3Jj9tSud7zN5RfBBJ3+s7MTcxJ5PE/HmdX9q5j2/w8/OgU0olhscOIC4kjPiSeTi06EeITclJl2crKSHnkUQqWLCHkxhtp+fhjiLuesKjRNDVqVBIiEo8RNW4MkAnMBVBKDW0c0RoHi9XGum17eFY2Ib3uPm0nzv108CeeWvUUfh5+TOg1gfiQeOJC4ogOiD6hVkJtWHJyOHLPBEo2bSJy8mRCb7tVz4rWaJootbUkdgO/A6OUUokAIvJAo0jViKw7kM2Qsl9x97QaTvxOMyw2C29uepNPd3xKz4ievD7kdSL9Ip1WXvnhwxy+cxwVKSnEvDGNoOHDnVaWRqM5eWpTEldjxKVeLiJLgDkYM65PKX7cmsLtHiuxRffGLbKLq8VpVLJLs5m0chLr09ZzXefreKTvI3g6sSVV8tdfHL7r/8BqJfaTj3VsB42mGVCjklBKzQfmi4g/cAWGe45IEXkP+E4p9XMjyeg0LFYbSdv/IE6OwNkPuVqcRmV75nYeWPEA2SXZPDvoWa7sdKVTyyv49VeSH3wIj/Bw2syYgXeH9k4tT6PRNAx1GpuVUkVKqVlmrOvWwGaMEU91IiLDRWSPiCSKyKPVpN8qIhkissX8ja33GZwEa/ZnMbxiGVY3bzhjdGMW7VK+TfiWmxffjBtuzBwx06kKQtlsZH/+OUcmTMQ7Lo52c2ZrBaHRNCPq5SpTKZUDzDB/tWJOunsXuAgjLvYGEVmglNpZJetcpdSE+sjRUPy05SCT3VdDt8vBJ9gVIjQq5dZyXlj/Al/v/Zpzos7h5cEv08KnhVPKUkpRuHw5GW+8SdnevQQMHUrMa6/i5nfKepnXaE5JnOlPuR+QqJTaDyAiczDMVlWVhEuosNqw7PyBICmGs0/9Duu0ojQeXPEg2zK3MfbMsUzoNcEhB3snQtHadWRMm0bJ1q14to0l+rVXCbr0UkQHbdJomh3OVBIxwGG79SNA/2ryjRaRwcBe4AGl1OGqGURkHMaEPmJjYxtEuNX7shhpWUZJYAy+7QY3yDGbGjZlY0/2HtamruXTHZ9Saill2pBpXNj2QqeUV7JtGxnT3qBo9Wo8WrWi1bNTaXHllYjn6TmsWKM5FXB1ZJYfgNlKqTIRGQ98BlxQNZNS6piJq0+fPqohCl61cTOPuu3A1nvSKROWtKZIbwDdw7rz3/P+S4fgDg1eblliIhlvvkXBL7/gHhJC5KOPEDJmjI7voNGcAjhTSSQDbezWW5vbjqGUyrJb/R/wshPlOUa5xUbw3q9xE4VbMzc11RXpbUDUAPpH9XfK3IfyI8lkvvMOeQsW4ObrS/jECYTecouO8aDRnEI4U0lsAOJEpD2GcrgeuME+g4hEKaVSzdXLgV00AqsS07nM9ivZrQYQGtK2MYpsUEotpSw6sIg5u+ccc6ER5BVE/6j+3HHGHfSP6u9QpLcTxZKRQeb7H5Dz1VeICKG33ELYuDvxCDk5Vx0ajabp4TQloZSyiMgE4CfAHfhYKbVDRKYCG5VSC4B7ReRywAJkA7c6Sx57dq9dwlC3DCrOebYximswMoozmLNnDvP2zCOnLIf4kHjuP/v+ekV6fRnKJQAAIABJREFUOxmseXlkffQx2TNnosrLaTF6NOF3/x+erU4pf48ajcYOp/ZJKKUWAYuqbHvabvkx4DFnylCVcouNmIPfUOLmj2/3Kxqz6BNmR9YOvtz5JYsPLsZqszKkzRD+3e3f9GnZp1F8HtmKi8me+QVZH32ELT+foJEjibh3Il5tm18rTKPR1A9Xd1w3Omt2HeAitZbs9lcR49V0x+xbbVaWH17OzJ0z2ZS+CT8PP67rfB03drmRNkFt6j5AA6DKy8n5ah6Z77+PNTOTgCFDiLj/Pny6nF7uSzSa05nTTkkcXT0LXykncnCjTu52mBJLCV/t+YrZu2eTXJhMTEAMk/pM4qq4qwj0CmwUGZTVSt4PP5D59jtUJCfj17cvEW+9hd//t3fn8THd++PHX+9ECEksiV3EruhFkNZWiha9pZRSjeUWVcqX0ttWS3uV0t72Fi1df7pILbe4JbVWUetVJURsoWJLRW1BxCD75/fHjNxEk0giYzKT9/Px8EjmnDOf8/7MiXnP+Zw570/zZvdk/0qpwqNIJYnElFTq/7GC8541qBTwgKPD+ZP4pHhGbRjFvov7aFGpBa8GvUqH6h3sfq3hlrSkJCw//8zFTz8l6dhxPO+/n8pTpuDVto2W8laqiCpSSWLP7p20kaMcb/galQrZm15cQhzD1w8nKi6KGQ/PoEvNLvdkvyY1lRthYVxdtYpr69aTFh9P8dq1qTZrFj5dOmtyUKqIK1JJ4mbYPFJwI6DjEEeHkknszViGrx9O9NVoZnWcRXt/+94Bbowh4eBB4letIn7Nj6RcvIhbqVL4dH6U0t2749W6NVKsSP1pKKWyUWTeCRISE2kS+yNHS7emUZkqjg4n3fnr5xm2bhjnb5zn00c/pVWVVnbbV+KJE8SvWsXV1atJjv4d8fDA6+H2lOneHe+HH9b5pZVSf1JkkkTk1mU0lzguFKLZ585YzjDsp2FcSbzCF49+QfNKze2yn+u//ML56dNJjDwMbm6Uavkg5YcPx6dzZ9xLl7bLPpVSrqHIJInUaxf43a0a9dsVjnkjouOjGbZuGNeTr/Nl5y9pXKGxXfYTt3QpZye9RfHq1ak0cQI+jz2GR0X7TU+qlHItRSZJPNB7LObJMYWiXPXxuOM8v+55UtJS+KbrNzTwLfj7DowxxH78CbGffYZX27ZUm/WR1lRSOUpOTiYmJoaEhARHh6LugqenJ/7+/ngUUPXlIpMkgEKRII5cPsLwdcNxd3Nn7mNzqVO2ToHvwyQlcfYfk7i6fDllevemypTJWq5b3VFMTAw+Pj7UrFlTv9XmpIwxXLp0iZiYGGrVKpgZIB3/rlmEHIw9yNCfhlKiWAlCHguxS4JIvXaN0y+8wNXlyyn/4hiqvDNNE4TKlYSEBPz8/DRBODERwc/Pr0DPBovUmYQjhZ8PZ9TPoyhboixfd/2aat7VCnwfyefOcXr4CBJPnKDKP/9J2V72m7tauSZNEM6voI+hJol74NiVY4zcMJKKpSryVZevqORVqcD3kXDkCKdHvECaxULAnP+HV5s2Bb4PpVTRo8NNdhaXEMeYjWPw8vCyW4KwbN9O9ADrV3tr/HuhJgjltGJiYujZsyf16tWjTp06jB07lqSkpAJpe/DgwXz//fcF0lZe9OrVi8DAQOrWrUuZMmUIDAwkMDCQX375JVfPb+Pg/8+aJOwoJS2FV7a+wvkb5/mw44d2SRBxy0I5PeIFPKpVo+biRXjed1+B70Ope8EYQ+/evXnyySeJiori6NGjWCwW3njjDUeHlifGGNLS0tIfh4aGEhERwVdffUW7du2IiIggIiIi/c0/JSUlx/Zym0zsRYeb7GjG7hnsPLuTqW2n0rRC0wJp0xhDysWLJB6NwrJ1C1fmzcerTWuqzZqFu8+9qRKrXN+UlYeI/CO+QNtsVLU0bz1xf7brN27ciKenJ0OGWMvmuLu78+GHH1KrVi2mTJnCkiVLWLFiBTdu3OD48eP06tWLf/3LOuPxunXreOutt0hMTKROnTrMnTsX71x85dtisdCzZ0+uXLlCcnIy06ZNo2fPnkyaNAlfX1/GjRsHwBtvvEHFihUZO3YsH3zwAUuWLCExMZFevXoxZcoUTp06RdeuXWnZsiV79uxhzZo11MhhvpWQkBCWLVuGxWIhNTWV1atXZxkHgLe3NxaLhc2bNzN58mTKly/PwYMHadGiBQsWLLD7dSRNEnYSGhXKgsMLGNhwIE/Wzd8F5FTLdRKjjpJ4NIrEo0et/6KiSI2LS9+mzFO9qfLWW0jx4gUVulIOcejQIVq0aJFpWenSpQkICODYsWMAREREsHfvXkqUKMF9993HmDFjKFmyJNOmTWPDhg14eXnx/vvvM3PmTCZNmpTVbjLx9PQkNDSU0qVLExsbS6tWrejRowdDhw6ld+/ejBs3jrS0NBYtWsSuXbtYt24dUVFR7Nq1C2MMPXr0YOvWrQQEBBAVFcW3335Lq1a5K60THh7O/v378fX1JSUlJcs4bk8Ae/fu5dChQ1StWpW2bduyfft2HnrooVy+wvmjScIOIi5EMPXXqbSu0pqXg17O9fNSLl0ifu1arv93O4m//UbyH3+kr3MrVYoS9erh07kzJerXp0S9epS4r77OK63sIqdP/I70yCOPUKZMGQAaNWpEdHQ0cXFxREZG0rZtWwCSkpJo3bp1rtozxjBx4kS2bt2Km5sbZ86c4fz589SsWRM/Pz/27t3L+fPnadasGX5+fqxbt45169bRrJl1bhWLxUJUVBQBAQHUqFEj1wkCoHPnzvj6+uYYR+XbpgZ+8MEH8ff3ByAwMJBTp05pknA2566fY9ymcVT2qswHD39AMbecX+JUi4VrGzYQv2o113fsgNRUitesScnAQMo+3deaEOrXx6Nq1UJxM6BS9tKoUaM/XViOj4/n999/p27duoSHh1OiRIn0de7u7qSkpGCMoXPnznz33Xd53ufChQu5ePEie/bswcPDg5o1a6bfYzBs2DBCQkI4d+4cQ4cOBaxv5hMmTGDEiBGZ2jl16hReXl552nfG7XOKI6Os+m9vdn3XEZHHROQ3ETkmIq/nsN1TImJEJMie8dhbQkoC4zaN42bKTWZ3nE2ZEmWy3C4tMZH49euJGTuOqLYPcfb1CSSdPInfc89Ra/ly6qz9kWozZ1D+hRfw6dSJ4v7+miCUy3vkkUe4ceMG8+bNAyA1NZWXX36ZwYMHU6pU9lMNt2rViu3bt6cPSV2/fp2jR4/map9Xr16lYsWKeHh4sGnTJqKjo9PX9erVi7Vr1xIWFkbXrl0B6Nq1K9988w0WiwWAM2fOcOHChXz1N7dxOJrdziRExB34FOgMxABhIrLCGBN523Y+wFhgp71iuReMMUzeMZnIS5HM6jiLuuXqZl6fmsqNnTu5unq1dXKfa9dw9/OjbN++lO72OCUDA/VGJlWkiQihoaGMGjWKqVOnkpaWxuOPP867776b4/MqVKhASEgIwcHBJCYmAjBt2jTq16//p21HjBiRfjG6evXqrFy5kieeeILGjRsTFBREgwzztxcvXpyOHTtStmxZ3N2ts0N26dKFw4cPpw9neXt7s2DBgvT1+TVgwIBs43A0McbYp2GR1sBkY0xX2+MJAMaYf9623UfAeuBV4BVjzO6c2g0KCjK7d+e4iUPMPTiXmXtmMqbZGIY3GZ6+3CQnExcaSuznX5By9ixuXl74dOlC6W7d8GrVUif3UYXG4cOHadiwoaPDKDTS0tJo3rw5//nPf6hXr56jw8mTrI6liOwxxuR5tMae71DVgNMZHscALTNuICLNgerGmNUi8mp2DYnIcGA4QEBAgB1CvTv/PfNfPtzzIV1qdOH5xs8DYNLSiP/xR2Jnf0xSdDQlmzal0muv4d3hYdw8PR0csVIqJ5GRkXTv3p1evXo5XYIoaA77GCsibsBMYPCdtjXGzAHmgPVMwr6R5c3JqycZv2U89/nex9S2UwG4tnkzFz+aReKRI5SoXx//zz7Du2MHHU5Sykk0atSIEydOODqMQsGeSeIMUD3DY3/bslt8gL8Am21vnpWBFSLS405DToXFtaRrvLjxRYq5FWNWx1kQcYjoDz/iZng4HgEBVP3gA0p3e1wvOiulnJY9k0QYUE9EamFNDs8A/W+tNMZcBcrfeiwim8nFNYnC5KsDX3H62mm+qfkmKS9NJnrbNopVrEjlyZMp+1RvLdGtlHJ6dksSxpgUERkN/AS4A98YYw6JyNvAbmPMCnvt+15ITUtl297lTP25DCX3/oOEMmWo+OqrlBvQX685KKVchl2vSRhj1gBrbluW5b3yxpgO9oyloO2KXMf/fX2eyjeKU37USHyHDNHaSUopl6OD5fmQcuUKZuxkfC3g//UcKrz4oiYIpQqAK5YKnzJlChMmTMi0LCIiIsevG0+ePJnp06cDMGnSJDZs2PCnbTZv3kz37t0LNtgsaJLIo1SLhehhw/A6F8+v4zpSJqjlnZ+klLojVy0VHhwczOLFizNts2jRIoKDg3PV3ttvv82jjz5aoDHmhd7JlQdpN29y+oUXSDxyhJm93HjpieF3fpJSzujH1+HcgYJts3Jj+Ot72a521VLh9evXp1y5cuzcuZOWLa0fKpcsWcJPP/3El19+yZw5c0hKSqJu3brMnz//TyVIBg8eTPfu3enTpw9r165l3LhxlCpVyu6F/W7RM4lcSktKImb0GG6G72XVoLpcbF6jwOaIUErlvlT44sWLOXDgAIsXL+b06dPExsamlwoPDw8nKCiImTNn5mqft0qFh4eHs2nTJl5++WWMMQwdOjS9htStUuEDBw7MVCo8IiKCPXv2sHXrVgCioqIYNWoUhw4d+tNcEsHBwSxatAiAX3/9FV9fX+rVq0fv3r0JCwtj3759NGzYkK+//jrbWBMSEnj++edZuXIle/bs4dy5c7l7Ye+SnknkgklO5szf/8717dspOelV5id/yKg6o/TmOOW6cvjE70jOWiq8X79+tGnThhkzZmQaajp48CBvvvkmcXFxWCyW9EKCWTly5Ai1atVKvwN84MCBzJkzJ3cv3F3QJHEHJjWVPyZMxLLhZyq9+Sbf/+U67IUnaj/h6NCUcimuXCq8evXq1KpViy1btrB06VJ27NgBWIeSfvjhB5o2bUpISAibN2/Ocx/sTYebcmCM4dzkKcSvWkWFv/+dcgP6s/L4SlpUaoG/j7+jw1PKpbh6qfDg4GBeeuklateunT5x0LVr16hSpQrJycksXLgwx+c3aNCAU6dOcfz4cYB8JcX80CSRDWMMF957n7j//Ae/ESMoP/x5DsQe4FT8KXrU6eHo8JRyObdKhd+qulq/fn08PT3zVCq8SZMmtG7dmiNHjmS57YgRI/D398ff35/WrVszYMAAdu/eTePGjZk3b16WpcKffvrpTKXC+/fvT+vWrWncuDF9+vTh2rVruepf3759OXToUKZvNU2dOpWWLVvStm3bO5YH9/T0ZM6cOXTr1o3mzZtTsWLFXO33btmtVLi93KtS4Rdnf0zsZ59RbtAgKk2cgIgw7ddp/HDsBzY9vQmf4npfhHItWio8My0VbqVnElm49PU3xH72GWWe6k2lCa8jIiSlJvHjyR/pFNBJE4RSLi4yMpK6devyyCOPOF2CKGh64fo2yWfOcGHGDHy6dKHK22+nV3DdGrOV+KR4HWpSqgjQUuH/o2cSt7m8YCGIWM8gMkxJuOL4CsqXLE+rKll/xU0ppVyRJokMUi3Xifv+e0p37YJHlSrpy68kXGFbzDa61+5OMTc9+VJKFR2aJDK4GhpK2rVr+P7tb5mWrzm5hhSTwhN19N4IpVTRoknCxqSlcXn+fEo2bUrJwMBM61YeX0kD3wbUL1ffQdEppZRjaJKwsWzeQvLvv+P7bOaziONxxzl06ZDeYa2UnV26dInAwEACAwOpXLky1apVS398N+XCQ0JCGD16dAFGmjvvvPNOevzu7u7pv8+ePTtXzx82bBiRkZF2jvLOdIDd5vK331KsShV8unTJtHzl8ZW4izuP137cQZEpVTT4+fkREREBWOdT8Pb25pVXXnFwVHmTmpqafuPdG2+8kV7m3NvbO71vtxhjMMbg5pb1Z/WvvvrKvsHmkiYJIOHIEW7s3EnFV15Giv3vJUlNS2XliZW0rdaW8iXL59CCUq7l/V3vc+Ry1nct51cD3wa89uBrud4+LS2NFi1asGfPHvbt20dgYCDR0dEEBARQp04dDhw4wIULFxg6dCixsbFUqFCBuXPnEhAQkKv2R44cSVhYGDdv3qRPnz5MmTKFjRs3Mnv2bH744QcA1q9fz2effUZoaGi25chr1qxJv379WL9+PePHj+eZZ57Jdp9ZlRR/7733/hQHQIcOHZg+fTpBQUF4e3szduxYVq1aRcmSJVm+fDmVKlXK9Wt5N+w63CQij4nIbyJyTERez2L9CyJyQEQiROS/ItLInvFk5/K8+UjJkpTt2zfT8l3ndnHhxgW9YK2UA7i5uZGQkEB8fDzbtm0jKCiIbdu2ER0dTcWKFSlVqhRjxozh2WefZf/+/QwYMIAXX3wx1+2/88477N69m/3797Nlyxb2799Px44dOXLkCBcvXgRg7ty56Ukop3Lkfn5+hIeH55ggbrm9pHhWcdzu+vXrtGrVin379tG+fXu+/PLLXPfzbtntTEJE3IFPgc5ADBAmIiuMMRkH2f5tjPnCtn0PYCbwmL1iykpKbCzxK1dStm8f3G0liG9ZeXwlPh4+dKze8V6GpJTD5eUTvz21adOG7du3s3XrViZOnMjatWsxxtCuXTsAduzYwbJlywAYNGgQ48ePz3XbS5YsYc6cOaSkpHD27FkiIyNp0qQJgwYNYsGCBQwZMoQdO3Ywb9481q5dm2M58n79+uV6v7eXFM8ujoyKFy+ePlVpixYtWL9+fa73d7fsOdz0IHDMGHMCQEQWAT2B9CRhjInPsL0XcM8LSV1ZtBiTnEy5gYMyLb+RfIMNv2+gW+1ulHAvkc2zlVL21L59+/Szh549e/L+++8jInTr1u2u2j158iTTp08nLCyMcuXKMXjw4PQS4UOGDOGJJ57A09OTvn37UqxYsTuWI8+pTHhO2+YUR0YeHh7p89fcKpF+r9hzuKkacDrD4xjbskxE5P9E5DjwLyDLc0URGS4iu0Vk963TwIKQlpjIle++w/vhhylRu1amdeuj13Mz5aaW4VDKgdq1a8eCBQuoV68ebm5u+Pr6smbNmvSpO9u0aZM+49vChQvTzzDuJD4+Hi8vL8qUKcP58+f58ccf09dVrVqVqlWrMm3atPSpVO+mHHl+4ygsHP4VWGPMp8aYOsBrwJvZbDPHGBNkjAmqUKFCge07fvUaUi9d+tPXXsE61FTdpzqBFQKzeKZS6l6oWbMmxhjat28PwEMPPUTZsmUpV64cAB9//DFz586lSZMmzJ8/n1mzZmXZTkhISHqJcH9/f/z8/GjWrBkNGjSgf//+6cNItwwYMIDq1aunV1LNSznyvGjatGmOcRQGdisVLiKtgcnGmK62xxMAjDH/zGZ7N+CKMaZMVutvKahS4cYYTvbqDamp1FqxPNNUpGctZ+m6tCsjA0cysunIu96XUs5AS4X/z+jRo2nWrBnPPfeco0PJF2cpFR4G1BORWiJSHHgGWJFxAxHJWIO3GxBlx3gyubFzF4lHjuD77N/+NFf16pOrMRi61+5+r8JRShUSLVq0YP/+/QwcONDRoRQKdrtwbYxJEZHRwE+AO/CNMeaQiLwN7DbGrABGi8ijQDJwBXjWXvHc7vK8ebiXK0fp7pkTQUJKAqFRoTSv2JzqPtXvVThKqUJiz549jg6hULHrzXTGmDXAmtuWTcrw+1h77j87SdHRWDZtovzIF3Dz9ExfnpqWyoRtEzh97XSh+QqgUko5ksMvXDvC5fkLoFgxyt5248v03dPZ8PsGXn3gVdr7t3dQdEopVXgUuSSRGh9P3LJllHn8r3hkmEh83qF5LDi8gIENBzKo0aAcWlBKqaKjyCWJuO+XYm7coFyGOSPWnVrH9N3T6VyjM68+8KoDo1NKqcKlSCUJk5LClQULKBUURMn77wdg74W9TNg2gaYVmvLuQ+/iJkXqJVGq0HC1UuFbtmzJVLoDICUlhUqVKvHHH39k+ZzNmzenl99YsWIF7733XpbbeXt7F2ywOShSVWCv/byR5D/+oOIEa63Bk1dPMmbjGKp4V2F2p9l4FvO8QwtKKXtxtVLh7dq1IyYmhujoaGrUqAHAhg0buP/++6lateod2+rRowc9eji+4kORShKXv/0WD39/fDp1IvZmLCM3jMRd3Pn8kc8p51nO0eEpVWice/ddEg8XbKnwEg0bUHnixFxv7+ylwt3c3Hj66adZtGgRr71m/bbkokWLCA4OZteuXYwdO5aEhARKlizJ3Llzue+++zLFFxISwu7du/nkk084efIk/fv3x2Kx0LNnz1y/hgWhyIyt3DxwgJvh4fgOGsjNtERG/zyaSzcv8UmnT6heWu+HUKqwcYVS4cHBwem1pRITE1mzZg1PPfUUDRo0YNu2bezdu5e3336biXdInmPHjmXkyJEcOHCAKlWq5LqPBaHInEnc2LULNx8fvHr15OWt4zl8+TAfdfiIxhUaOzo0pQqdvHzitydnLxUeFBSExWLht99+4/Dhw7Rs2RJfX19Onz7Ns88+S1RUFCJCcnJyjrFu376dpUuXpvfz1pnJvVBkkoTfc89Rpndv3ov8hC0xW3ij5Rt0DNB5IpQqzFyhVPits4nDhw8THBwMwD/+8Q86duxIaGgop06dokOHDneM+fbyQfdKkRluAvg2ZhmLf1vMkL8M4ZkGd55BSinlWK5QKjw4OJgFCxawcePG9OsJV69epVo168wJISEhd2yjbdu2mfp5LxWZJLHmxBo+Cv+Iv9b8K+Oaj3N0OEqpXHCFUuENGzbEy8uLTp06pZ9xjB8/ngkTJtCsWbNcTSA0a9YsPv30Uxo3bsyZM2dytd+CYrdS4faS31LhYefCmB85n+kPT6e4e3E7RKaUc9NS4f+jpcL/p8hck3ig8gM8UPkBR4ehlCrkWrRogZeXFzNmzHB0KIVCkUkSSimVG1oqPLMic01CKXVnzjb8rP6soI+hJgmlFACenp5cunRJE4UTM8Zw6dIlPD0LrsSQDjcppQDw9/cnJiYm/W5j5Zw8PT3x9/cvsPY0SSilAPDw8KBWrVqODkMVMjrcpJRSKluaJJRSSmVLk4RSSqlsOd0d1yJyEYjO59PLA7EFGE5h4Gp9crX+gOv1ydX6A67Xp6z6U8MYUyGvDTldkrgbIrI7P7elF2au1idX6w+4Xp9crT/gen0qyP7ocJNSSqlsaZJQSimVraKWJOY4OgA7cLU+uVp/wPX65Gr9AdfrU4H1p0hdk1BKKZU3Re1MQimlVB5oklBKKZWtIpMkROQxEflNRI6JyOuOjuduicgpETkgIhEikvep+goBEflGRC6IyMEMy3xFZL2IRNl+lnNkjHmRTX8mi8gZ23GKEJHHHRljXolIdRHZJCKRInJIRMbaljvlccqhP057nETEU0R2icg+W5+m2JbXEpGdtve8xSKSryk5i8Q1CRFxB44CnYEYIAwINsZEOjSwuyAip4AgY4zT3gAkIu0BCzDPGPMX27J/AZeNMe/Zknk5Y8xrjowzt7Lpz2TAYoyZ7sjY8ktEqgBVjDHhIuID7AGeBAbjhMcph/48jZMeJxERwMsYYxERD+C/wFjg78AyY8wiEfkC2GeM+Tyv7ReVM4kHgWPGmBPGmCRgEdDTwTEVecaYrcDl2xb3BL61/f4t1v/ATiGb/jg1Y8xZY0y47fdrwGGgGk56nHLoj9MyVhbbQw/bPwN0Ar63Lc/3MSoqSaIacDrD4xic/A8D6x/BOhHZIyLDHR1MAapkjDlr+/0cUMmRwRSQ0SKy3zYc5RTDMlkRkZpAM2AnLnCcbusPOPFxEhF3EYkALgDrgeNAnDEmxbZJvt/zikqScEUPGWOaA38F/s821OFSjHUs1NnHQz8H6gCBwFlghmPDyR8R8QaWAuOMMfEZ1znjccqiP059nIwxqcaYQMAf68hJg4Jqu6gkiTNA9QyP/W3LnJYx5ozt5wUgFOsfhis4bxs3vjV+fMHB8dwVY8x523/gNOBLnPA42ca5lwILjTHLbIud9jhl1R9XOE4Axpg4YBPQGigrIrcmlsv3e15RSRJhQD3b1f7iwDPACgfHlG8i4mW76IaIeAFdgIM5P8tprACetf3+LLDcgbHctVtvpDa9cLLjZLso+jVw2BgzM8MqpzxO2fXHmY+TiFQQkbK230ti/YLOYazJoo9ts3wfoyLx7SYA21faPgLcgW+MMe84OKR8E5HaWM8ewDoF7b+dsT8i8h3QAWtZ4/PAW8APwBIgAGtJ+KeNMU5xMTib/nTAOoRhgFPAiAxj+YWeiDwEbAMOAGm2xROxjuM73XHKoT/BOOlxEpEmWC9Mu2P94L/EGPO27X1iEeAL7AUGGmMS89x+UUkSSiml8q6oDDcppZTKB00SSimlsqVJQimlVLY0SSillMqWJgmllFLZ0iShXI6IGBGZkeHxK7ZCewW9n8EicjFD5dAIEWlUgO1PFpFXCqo9pfJDk4RyRYlAbxEpfw/2tdgYE5jhn9NWFlYqK5oklCtKwTrH70u3rxCREBHpk+Gxxfazg4hsEZHlInJCRN4TkQG2Ov0HRKRObndua2uriKwW6xwmX4iIm21dsK29gyLyfobnPCYi4bY5AX7O0FwjEdlsi+lF27Zetrb32drpl+dXSKlcKnbnTZRySp8C+23zU+RWU6Ah1nLfJ4CvjDEP2iamGQOMy+I5/Wx38d7S2vbzQaAR1ruR12I9s/kFeB9oAVzBWsX3SWA71npB7Y0xJ0XEN0N7DYCOgA/wm4h8DjwG/GGM6QY76QTiAAABj0lEQVQgImXy0Eel8kSThHJJxph4EZkHvAjczOXTwm6VYhCR48A62/IDWN+os7LYGDM64wJreSB2GWNO2B5/BzwEJAObjTEXbcsXAu2BVGCrMeakLfaM5S1W20opJIrIBawluQ8AM2xnIquMMdty2T+l8kyHm5Qr+wh4DvDKsCwF29+9bQgo45SOGevapGV4nEbeP1DdXu8mv/VvMsaUChQzxhwFmmNNFtNEZFI+21bqjjRJKJdl+0S+BGuiuOUU1uEegB5YZ/GyhwdtVYfdgH5Yp5TcBTwsIuXFOqVuMLAF+BVoLyK1wDp/dE4Ni0hV4IYxZgHwAdaEoZRd6HCTcnUzgIzDQV8Cy0VkH9ZrBdfvsv3br0mMsv0MAz4B6mIt2RxqjEmzzQe9CRCsQ0nLAWyzCy6zJZULWMs9Z6cx8IGIpGEdwhp5l31QKltaBVapAiYiHYBXjDHdHR2LUndLh5uUUkplS88klFJKZUvPJJRSSmVLk4RSSqlsaZJQSimVLU0SSimlsqVJQimlVLb+P9x8+D6MrN4DAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "One Layer Test: 0.960000, Two Layer Test: 0.900000\n",
            "Start training OneLayerNetwork with Adam...\n",
            "| epoch  1 | train loss 0.197655 | train acc 0.970000 | valid loss 0.217931 | valid acc 0.953333 |\n",
            "| epoch  2 | train loss 0.131471 | train acc 0.983333 | valid loss 0.160481 | valid acc 0.960000 |\n",
            "| epoch  3 | train loss 0.102530 | train acc 0.986667 | valid loss 0.136721 | valid acc 0.973333 |\n",
            "| epoch  4 | train loss 0.085124 | train acc 0.990000 | valid loss 0.123066 | valid acc 0.973333 |\n",
            "| epoch  5 | train loss 0.073182 | train acc 0.990000 | valid loss 0.114030 | valid acc 0.973333 |\n",
            "| epoch  6 | train loss 0.064365 | train acc 0.990000 | valid loss 0.107544 | valid acc 0.966667 |\n",
            "| epoch  7 | train loss 0.057539 | train acc 0.993333 | valid loss 0.102632 | valid acc 0.966667 |\n",
            "| epoch  8 | train loss 0.052074 | train acc 1.000000 | valid loss 0.098767 | valid acc 0.966667 |\n",
            "| epoch  9 | train loss 0.047588 | train acc 1.000000 | valid loss 0.095640 | valid acc 0.966667 |\n",
            "| epoch 10 | train loss 0.043835 | train acc 1.000000 | valid loss 0.093052 | valid acc 0.966667 |\n",
            "| epoch 11 | train loss 0.040644 | train acc 1.000000 | valid loss 0.090873 | valid acc 0.966667 |\n",
            "| epoch 12 | train loss 0.037896 | train acc 1.000000 | valid loss 0.089012 | valid acc 0.966667 |\n",
            "| epoch 13 | train loss 0.035504 | train acc 1.000000 | valid loss 0.087403 | valid acc 0.966667 |\n",
            "| epoch 14 | train loss 0.033402 | train acc 1.000000 | valid loss 0.085998 | valid acc 0.966667 |\n",
            "| epoch 15 | train loss 0.031539 | train acc 1.000000 | valid loss 0.084760 | valid acc 0.966667 |\n",
            "| epoch 16 | train loss 0.029877 | train acc 1.000000 | valid loss 0.083661 | valid acc 0.966667 |\n",
            "| epoch 17 | train loss 0.028385 | train acc 1.000000 | valid loss 0.082678 | valid acc 0.966667 |\n",
            "| epoch 18 | train loss 0.027037 | train acc 1.000000 | valid loss 0.081795 | valid acc 0.966667 |\n",
            "| epoch 19 | train loss 0.025813 | train acc 1.000000 | valid loss 0.080996 | valid acc 0.966667 |\n",
            "| epoch 20 | train loss 0.024698 | train acc 1.000000 | valid loss 0.080271 | valid acc 0.966667 |\n",
            "| epoch 21 | train loss 0.023676 | train acc 1.000000 | valid loss 0.079610 | valid acc 0.966667 |\n",
            "| epoch 22 | train loss 0.022737 | train acc 1.000000 | valid loss 0.079004 | valid acc 0.966667 |\n",
            "| epoch 23 | train loss 0.021871 | train acc 1.000000 | valid loss 0.078448 | valid acc 0.966667 |\n",
            "| epoch 24 | train loss 0.021070 | train acc 1.000000 | valid loss 0.077935 | valid acc 0.966667 |\n",
            "| epoch 25 | train loss 0.020326 | train acc 1.000000 | valid loss 0.077461 | valid acc 0.966667 |\n",
            "| epoch 26 | train loss 0.019633 | train acc 1.000000 | valid loss 0.077021 | valid acc 0.966667 |\n",
            "| epoch 27 | train loss 0.018987 | train acc 1.000000 | valid loss 0.076613 | valid acc 0.966667 |\n",
            "| epoch 28 | train loss 0.018383 | train acc 1.000000 | valid loss 0.076232 | valid acc 0.966667 |\n",
            "| epoch 29 | train loss 0.017817 | train acc 1.000000 | valid loss 0.075877 | valid acc 0.960000 |\n",
            "| epoch 30 | train loss 0.017285 | train acc 1.000000 | valid loss 0.075545 | valid acc 0.960000 |\n",
            "Done!\n",
            "Start training TwoLayerNetwork with Adam...\n",
            "| epoch  1 | train loss 0.190851 | train acc 0.946667 | valid loss 0.230784 | valid acc 0.906667 |\n",
            "| epoch  2 | train loss 0.017252 | train acc 0.996667 | valid loss 0.046777 | valid acc 0.986667 |\n",
            "| epoch  3 | train loss 0.013669 | train acc 0.996667 | valid loss 0.056629 | valid acc 0.980000 |\n",
            "| epoch  4 | train loss 0.001908 | train acc 1.000000 | valid loss 0.020065 | valid acc 0.993333 |\n",
            "| epoch  5 | train loss 0.000994 | train acc 1.000000 | valid loss 0.013752 | valid acc 0.993333 |\n",
            "| epoch  6 | train loss 0.000749 | train acc 1.000000 | valid loss 0.014171 | valid acc 0.993333 |\n",
            "| epoch  7 | train loss 0.000580 | train acc 1.000000 | valid loss 0.014815 | valid acc 0.993333 |\n",
            "| epoch  8 | train loss 0.000489 | train acc 1.000000 | valid loss 0.015430 | valid acc 0.993333 |\n",
            "| epoch  9 | train loss 0.000414 | train acc 1.000000 | valid loss 0.016043 | valid acc 0.993333 |\n",
            "| epoch 10 | train loss 0.000348 | train acc 1.000000 | valid loss 0.016725 | valid acc 0.993333 |\n",
            "| epoch 11 | train loss 0.000296 | train acc 1.000000 | valid loss 0.017274 | valid acc 0.993333 |\n",
            "| epoch 12 | train loss 0.000263 | train acc 1.000000 | valid loss 0.017559 | valid acc 0.993333 |\n",
            "| epoch 13 | train loss 0.000237 | train acc 1.000000 | valid loss 0.017780 | valid acc 0.993333 |\n",
            "| epoch 14 | train loss 0.000214 | train acc 1.000000 | valid loss 0.017975 | valid acc 0.993333 |\n",
            "| epoch 15 | train loss 0.000195 | train acc 1.000000 | valid loss 0.018150 | valid acc 0.993333 |\n",
            "| epoch 16 | train loss 0.000179 | train acc 1.000000 | valid loss 0.018308 | valid acc 0.993333 |\n",
            "| epoch 17 | train loss 0.000164 | train acc 1.000000 | valid loss 0.018452 | valid acc 0.993333 |\n",
            "| epoch 18 | train loss 0.000149 | train acc 1.000000 | valid loss 0.018589 | valid acc 0.993333 |\n",
            "| epoch 19 | train loss 0.000133 | train acc 1.000000 | valid loss 0.018728 | valid acc 0.993333 |\n",
            "| epoch 20 | train loss 0.000122 | train acc 1.000000 | valid loss 0.018882 | valid acc 0.993333 |\n",
            "| epoch 21 | train loss 0.000113 | train acc 1.000000 | valid loss 0.019041 | valid acc 0.993333 |\n",
            "| epoch 22 | train loss 0.000105 | train acc 1.000000 | valid loss 0.019189 | valid acc 0.993333 |\n",
            "| epoch 23 | train loss 0.000096 | train acc 1.000000 | valid loss 0.019328 | valid acc 0.993333 |\n",
            "| epoch 24 | train loss 0.000089 | train acc 1.000000 | valid loss 0.019453 | valid acc 0.993333 |\n",
            "| epoch 25 | train loss 0.000083 | train acc 1.000000 | valid loss 0.019615 | valid acc 0.993333 |\n",
            "| epoch 26 | train loss 0.000078 | train acc 1.000000 | valid loss 0.019781 | valid acc 0.993333 |\n",
            "| epoch 27 | train loss 0.000074 | train acc 1.000000 | valid loss 0.019935 | valid acc 0.993333 |\n",
            "| epoch 28 | train loss 0.000070 | train acc 1.000000 | valid loss 0.020075 | valid acc 0.993333 |\n",
            "| epoch 29 | train loss 0.000065 | train acc 1.000000 | valid loss 0.020200 | valid acc 0.993333 |\n",
            "| epoch 30 | train loss 0.000061 | train acc 1.000000 | valid loss 0.020294 | valid acc 0.993333 |\n",
            "Done!\n",
            "(Imp) One Layer Test: 0.973333, Two Layer Test: 0.993333\n",
            "Start training OneLayerNetwork with Adam...\n",
            "| epoch  1 | train loss 0.675383 | train acc 0.910000 | valid loss 0.684692 | valid acc 0.880000 |\n",
            "| epoch  2 | train loss 0.463563 | train acc 0.950000 | valid loss 0.478921 | valid acc 0.926667 |\n",
            "| epoch  3 | train loss 0.349827 | train acc 0.970000 | valid loss 0.367965 | valid acc 0.946667 |\n",
            "| epoch  4 | train loss 0.282537 | train acc 0.973333 | valid loss 0.302650 | valid acc 0.946667 |\n",
            "| epoch  5 | train loss 0.238357 | train acc 0.973333 | valid loss 0.260146 | valid acc 0.946667 |\n",
            "| epoch  6 | train loss 0.206962 | train acc 0.973333 | valid loss 0.230248 | valid acc 0.946667 |\n",
            "| epoch  7 | train loss 0.183325 | train acc 0.973333 | valid loss 0.207996 | valid acc 0.953333 |\n",
            "| epoch  8 | train loss 0.164752 | train acc 0.980000 | valid loss 0.190727 | valid acc 0.960000 |\n",
            "| epoch  9 | train loss 0.149677 | train acc 0.980000 | valid loss 0.176895 | valid acc 0.960000 |\n",
            "| epoch 10 | train loss 0.137131 | train acc 0.986667 | valid loss 0.165537 | valid acc 0.960000 |\n",
            "| epoch 11 | train loss 0.126479 | train acc 0.986667 | valid loss 0.156025 | valid acc 0.960000 |\n",
            "| epoch 12 | train loss 0.117292 | train acc 0.986667 | valid loss 0.147930 | valid acc 0.960000 |\n",
            "| epoch 13 | train loss 0.109266 | train acc 0.986667 | valid loss 0.140948 | valid acc 0.960000 |\n",
            "| epoch 14 | train loss 0.102177 | train acc 0.986667 | valid loss 0.134859 | valid acc 0.960000 |\n",
            "| epoch 15 | train loss 0.095862 | train acc 0.986667 | valid loss 0.129496 | valid acc 0.960000 |\n",
            "| epoch 16 | train loss 0.090194 | train acc 0.986667 | valid loss 0.124732 | valid acc 0.960000 |\n",
            "| epoch 17 | train loss 0.085074 | train acc 0.990000 | valid loss 0.120471 | valid acc 0.960000 |\n",
            "| epoch 18 | train loss 0.080423 | train acc 0.990000 | valid loss 0.116634 | valid acc 0.966667 |\n",
            "| epoch 19 | train loss 0.076178 | train acc 0.990000 | valid loss 0.113158 | valid acc 0.966667 |\n",
            "| epoch 20 | train loss 0.072289 | train acc 0.990000 | valid loss 0.109993 | valid acc 0.973333 |\n",
            "| epoch 21 | train loss 0.068711 | train acc 0.993333 | valid loss 0.107098 | valid acc 0.973333 |\n",
            "| epoch 22 | train loss 0.065409 | train acc 0.993333 | valid loss 0.104439 | valid acc 0.980000 |\n",
            "| epoch 23 | train loss 0.062353 | train acc 0.993333 | valid loss 0.101986 | valid acc 0.986667 |\n",
            "| epoch 24 | train loss 0.059516 | train acc 0.993333 | valid loss 0.099716 | valid acc 0.986667 |\n",
            "| epoch 25 | train loss 0.056876 | train acc 0.993333 | valid loss 0.097608 | valid acc 0.986667 |\n",
            "| epoch 26 | train loss 0.054415 | train acc 0.996667 | valid loss 0.095645 | valid acc 0.986667 |\n",
            "| epoch 27 | train loss 0.052115 | train acc 0.996667 | valid loss 0.093811 | valid acc 0.986667 |\n",
            "| epoch 28 | train loss 0.049961 | train acc 0.996667 | valid loss 0.092094 | valid acc 0.986667 |\n",
            "| epoch 29 | train loss 0.047941 | train acc 0.996667 | valid loss 0.090483 | valid acc 0.986667 |\n",
            "| epoch 30 | train loss 0.046044 | train acc 0.996667 | valid loss 0.088967 | valid acc 0.986667 |\n",
            "Done!\n",
            "Start training TwoLayerNetwork with Adam...\n",
            "| epoch  1 | train loss 0.509271 | train acc 0.956667 | valid loss 0.510989 | valid acc 0.953333 |\n",
            "| epoch  2 | train loss 0.275621 | train acc 0.980000 | valid loss 0.290236 | valid acc 0.960000 |\n",
            "| epoch  3 | train loss 0.177928 | train acc 0.983333 | valid loss 0.202546 | valid acc 0.960000 |\n",
            "| epoch  4 | train loss 0.128210 | train acc 0.983333 | valid loss 0.160201 | valid acc 0.960000 |\n",
            "| epoch  5 | train loss 0.098140 | train acc 0.983333 | valid loss 0.135783 | valid acc 0.966667 |\n",
            "| epoch  6 | train loss 0.077636 | train acc 0.990000 | valid loss 0.119698 | valid acc 0.966667 |\n",
            "| epoch  7 | train loss 0.062724 | train acc 0.996667 | valid loss 0.108279 | valid acc 0.966667 |\n",
            "| epoch  8 | train loss 0.051486 | train acc 1.000000 | valid loss 0.099769 | valid acc 0.973333 |\n",
            "| epoch  9 | train loss 0.042822 | train acc 1.000000 | valid loss 0.093204 | valid acc 0.973333 |\n",
            "| epoch 10 | train loss 0.036031 | train acc 1.000000 | valid loss 0.088002 | valid acc 0.973333 |\n",
            "| epoch 11 | train loss 0.030635 | train acc 1.000000 | valid loss 0.083793 | valid acc 0.973333 |\n",
            "| epoch 12 | train loss 0.026297 | train acc 1.000000 | valid loss 0.080328 | valid acc 0.973333 |\n",
            "| epoch 13 | train loss 0.022771 | train acc 1.000000 | valid loss 0.077439 | valid acc 0.973333 |\n",
            "| epoch 14 | train loss 0.019877 | train acc 1.000000 | valid loss 0.075001 | valid acc 0.973333 |\n",
            "| epoch 15 | train loss 0.017478 | train acc 1.000000 | valid loss 0.072925 | valid acc 0.973333 |\n",
            "| epoch 16 | train loss 0.015472 | train acc 1.000000 | valid loss 0.071142 | valid acc 0.973333 |\n",
            "| epoch 17 | train loss 0.013782 | train acc 1.000000 | valid loss 0.069599 | valid acc 0.980000 |\n",
            "| epoch 18 | train loss 0.012346 | train acc 1.000000 | valid loss 0.068256 | valid acc 0.980000 |\n",
            "| epoch 19 | train loss 0.011117 | train acc 1.000000 | valid loss 0.067079 | valid acc 0.980000 |\n",
            "| epoch 20 | train loss 0.010058 | train acc 1.000000 | valid loss 0.066042 | valid acc 0.980000 |\n",
            "| epoch 21 | train loss 0.009140 | train acc 1.000000 | valid loss 0.065124 | valid acc 0.980000 |\n",
            "| epoch 22 | train loss 0.008339 | train acc 1.000000 | valid loss 0.064307 | valid acc 0.980000 |\n",
            "| epoch 23 | train loss 0.007637 | train acc 1.000000 | valid loss 0.063577 | valid acc 0.980000 |\n",
            "| epoch 24 | train loss 0.007018 | train acc 1.000000 | valid loss 0.062922 | valid acc 0.980000 |\n",
            "| epoch 25 | train loss 0.006470 | train acc 1.000000 | valid loss 0.062333 | valid acc 0.980000 |\n",
            "| epoch 26 | train loss 0.005982 | train acc 1.000000 | valid loss 0.061800 | valid acc 0.980000 |\n",
            "| epoch 27 | train loss 0.005546 | train acc 1.000000 | valid loss 0.061318 | valid acc 0.980000 |\n",
            "| epoch 28 | train loss 0.005156 | train acc 1.000000 | valid loss 0.060880 | valid acc 0.980000 |\n",
            "| epoch 29 | train loss 0.004804 | train acc 1.000000 | valid loss 0.060481 | valid acc 0.980000 |\n",
            "| epoch 30 | train loss 0.004486 | train acc 1.000000 | valid loss 0.060116 | valid acc 0.980000 |\n",
            "Done!\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "(Adam) One Layer Test: 0.966667, Two Layer Test: 0.966667\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}