Day 13/Fashion-MNIST.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "PyTorch Tutorial Basic v1.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "DPue4B2vPu2s",
        "colab_type": "code",
        "outputId": "479fd42c-3038-4eb4-af06-3e53abc1ea45",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "source": [
        "# import standard PyTorch modules\n",
        "import torch\n",
        "import torch.nn as nn\n",
        "import torch.nn.functional as F\n",
        "import torch.optim as optim\n",
        "from torch.utils.tensorboard import SummaryWriter # TensorBoard support\n",
        "\n",
        "# import torchvision module to handle image manipulation\n",
        "import torchvision\n",
        "import torchvision.transforms as transforms\n",
        "\n",
        "# calculate train time, writing train data to files etc.\n",
        "import time\n",
        "import pandas as pd\n",
        "import json\n",
        "from IPython.display import clear_output\n",
        "\n",
        "torch.set_printoptions(linewidth=120)\n",
        "torch.set_grad_enabled(True)     # On by default, leave it here for clarity"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<torch.autograd.grad_mode.set_grad_enabled at 0x7f41eb45dc50>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 1
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3iDlvtrBQZO3",
        "colab_type": "code",
        "outputId": "16c19bd5-6647-437d-b9f0-cdbaffd167a0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 54
        }
      },
      "source": [
        "# check PyTorch versions\n",
        "print(torch.__version__)\n",
        "print(torchvision.__version__)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "1.3.0+cu100\n",
            "0.4.1+cu100\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vvY6qsJ0Qy5n",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Use standard FashionMNIST dataset\n",
        "train_set = torchvision.datasets.FashionMNIST(\n",
        "    root = './data/FashionMNIST',\n",
        "    train = True,\n",
        "    download = True,\n",
        "    transform = transforms.Compose([\n",
        "        transforms.ToTensor()                                 \n",
        "    ])\n",
        ")"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rZPckqWTRXUm",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Build the neural network, expand on top of nn.Module\n",
        "class Network(nn.Module):\n",
        "  def __init__(self):\n",
        "    super().__init__()\n",
        "\n",
        "    # define layers\n",
        "    self.conv1 = nn.Conv2d(in_channels=1, out_channels=6, kernel_size=5)\n",
        "    self.conv2 = nn.Conv2d(in_channels=6, out_channels=12, kernel_size=5)\n",
        "\n",
        "    self.fc1 = nn.Linear(in_features=12*4*4, out_features=120)\n",
        "    self.fc2 = nn.Linear(in_features=120, out_features=60)\n",
        "    self.out = nn.Linear(in_features=60, out_features=10)\n",
        "\n",
        "  # define forward function\n",
        "  def forward(self, t):\n",
        "    # conv 1\n",
        "    t = self.conv1(t)\n",
        "    t = F.relu(t)\n",
        "    t = F.max_pool2d(t, kernel_size=2, stride=2)\n",
        "\n",
        "    # conv 2\n",
        "    t = self.conv2(t)\n",
        "    t = F.relu(t)\n",
        "    t = F.max_pool2d(t, kernel_size=2, stride=2)\n",
        "\n",
        "    # fc1\n",
        "    t = t.reshape(-1, 12*4*4)\n",
        "    t = self.fc1(t)\n",
        "    t = F.relu(t)\n",
        "\n",
        "    # fc2\n",
        "    t = self.fc2(t)\n",
        "    t = F.relu(t)\n",
        "\n",
        "    # output\n",
        "    t = self.out(t)\n",
        "    # don't need softmax here since we'll use cross-entropy as activation.\n",
        "\n",
        "    return t"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cOe6m7forHCM",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def get_num_correct(preds, labels):\n",
        "  return preds.argmax(dim=1).eq(labels).sum().item()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mBOxsN4A7N8K",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# import modules to build RunBuilder and RunManager helper classes\n",
        "from collections  import OrderedDict\n",
        "from collections import namedtuple\n",
        "from itertools import product\n",
        "\n",
        "# Read in the hyper-parameters and return a Run namedtuple containing all the \n",
        "# combinations of hyper-parameters\n",
        "class RunBuilder():\n",
        "  @staticmethod\n",
        "  def get_runs(params):\n",
        "\n",
        "    Run = namedtuple('Run', params.keys())\n",
        "\n",
        "    runs = []\n",
        "    for v in product(*params.values()):\n",
        "      runs.append(Run(*v))\n",
        "    \n",
        "    return runs"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HSddFGL7zisa",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Helper class, help track loss, accuracy, epoch time, run time, \n",
        "# hyper-parameters etc. Also record to TensorBoard and write into csv, json\n",
        "class RunManager():\n",
        "  def __init__(self):\n",
        "\n",
        "    # tracking every epoch count, loss, accuracy, time\n",
        "    self.epoch_count = 0\n",
        "    self.epoch_loss = 0\n",
        "    self.epoch_num_correct = 0\n",
        "    self.epoch_start_time = None\n",
        "\n",
        "    # tracking every run count, run data, hyper-params used, time\n",
        "    self.run_params = None\n",
        "    self.run_count = 0\n",
        "    self.run_data = []\n",
        "    self.run_start_time = None\n",
        "\n",
        "    # record model, loader and TensorBoard \n",
        "    self.network = None\n",
        "    self.loader = None\n",
        "    self.tb = None\n",
        "\n",
        "  # record the count, hyper-param, model, loader of each run\n",
        "  # record sample images and network graph to TensorBoard  \n",
        "  def begin_run(self, run, network, loader):\n",
        "\n",
        "    self.run_start_time = time.time()\n",
        "\n",
        "    self.run_params = run\n",
        "    self.run_count += 1\n",
        "\n",
        "    self.network = network\n",
        "    self.loader = loader\n",
        "    self.tb = SummaryWriter(comment=f'-{run}')\n",
        "\n",
        "    images, labels = next(iter(self.loader))\n",
        "    grid = torchvision.utils.make_grid(images)\n",
        "\n",
        "    self.tb.add_image('images', grid)\n",
        "    self.tb.add_graph(self.network, images)\n",
        "\n",
        "  # when run ends, close TensorBoard, zero epoch count\n",
        "  def end_run(self):\n",
        "    self.tb.close()\n",
        "    self.epoch_count = 0\n",
        "\n",
        "  # zero epoch count, loss, accuracy, \n",
        "  def begin_epoch(self):\n",
        "    self.epoch_start_time = time.time()\n",
        "\n",
        "    self.epoch_count += 1\n",
        "    self.epoch_loss = 0\n",
        "    self.epoch_num_correct = 0\n",
        "\n",
        "  # \n",
        "  def end_epoch(self):\n",
        "    # calculate epoch duration and run duration(accumulate)\n",
        "    epoch_duration = time.time() - self.epoch_start_time\n",
        "    run_duration = time.time() - self.run_start_time\n",
        "\n",
        "    # record epoch loss and accuracy\n",
        "    loss = self.epoch_loss / len(self.loader.dataset)\n",
        "    accuracy = self.epoch_num_correct / len(self.loader.dataset)\n",
        "\n",
        "    # Record epoch loss and accuracy to TensorBoard \n",
        "    self.tb.add_scalar('Loss', loss, self.epoch_count)\n",
        "    self.tb.add_scalar('Accuracy', accuracy, self.epoch_count)\n",
        "\n",
        "    # Record params to TensorBoard\n",
        "    for name, param in self.network.named_parameters():\n",
        "      self.tb.add_histogram(name, param, self.epoch_count)\n",
        "      self.tb.add_histogram(f'{name}.grad', param.grad, self.epoch_count)\n",
        "    \n",
        "    # Write into 'results' (OrderedDict) for all run related data\n",
        "    results = OrderedDict()\n",
        "    results[\"run\"] = self.run_count\n",
        "    results[\"epoch\"] = self.epoch_count\n",
        "    results[\"loss\"] = loss\n",
        "    results[\"accuracy\"] = accuracy\n",
        "    results[\"epoch duration\"] = epoch_duration\n",
        "    results[\"run duration\"] = run_duration\n",
        "\n",
        "    # Record hyper-params into 'results'\n",
        "    for k,v in self.run_params._asdict().items(): results[k] = v\n",
        "    self.run_data.append(results)\n",
        "    df = pd.DataFrame.from_dict(self.run_data, orient = 'columns')\n",
        "\n",
        "    # display epoch information and show progress\n",
        "    clear_output(wait=True)\n",
        "    display(df)\n",
        "\n",
        "  # accumulate loss of batch into entire epoch loss\n",
        "  def track_loss(self, loss):\n",
        "    # multiply batch size so variety of batch sizes can be compared\n",
        "    self.epoch_loss += loss.item() * self.loader.batch_size\n",
        "\n",
        "  # accumulate number of corrects of batch into entire epoch num_correct\n",
        "  def track_num_correct(self, preds, labels):\n",
        "    self.epoch_num_correct += self._get_num_correct(preds, labels)\n",
        "\n",
        "  @torch.no_grad()\n",
        "  def _get_num_correct(self, preds, labels):\n",
        "    return preds.argmax(dim=1).eq(labels).sum().item()\n",
        "  \n",
        "  # save end results of all runs into csv, json for further a\n",
        "  def save(self, fileName):\n",
        "\n",
        "    pd.DataFrame.from_dict(\n",
        "        self.run_data, \n",
        "        orient = 'columns',\n",
        "    ).to_csv(f'{fileName}.csv')\n",
        "\n",
        "    with open(f'{fileName}.json', 'w', encoding='utf-8') as f:\n",
        "      json.dump(self.run_data, f, ensure_ascii=False, indent=4)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1Rs-jD_Z-C-L",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# put all hyper params into a OrderedDict, easily expandable\n",
        "params = OrderedDict(\n",
        "    lr = [.01, .001],\n",
        "    batch_size = [100, 1000],\n",
        "    shuffle = [False]\n",
        ")\n",
        "epochs = 5"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZL-SM1wgUHql",
        "colab_type": "code",
        "outputId": "521e2125-c967-4c92-c0c8-c9b5696be71f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 648
        }
      },
      "source": [
        "m = RunManager()\n",
        "\n",
        "# get all runs from params using RunBuilder class\n",
        "for run in RunBuilder.get_runs(params):\n",
        "\n",
        "    # if params changes, following line of code should reflect the changes too\n",
        "    network = Network()\n",
        "    loader = torch.utils.data.DataLoader(train_set, batch_size = run.batch_size)\n",
        "    optimizer = optim.Adam(network.parameters(), lr=run.lr)\n",
        "\n",
        "    m.begin_run(run, network, loader)\n",
        "    for epoch in range(epochs):\n",
        "      \n",
        "      m.begin_epoch()\n",
        "      for batch in loader:\n",
        "        \n",
        "        images = batch[0]\n",
        "        labels = batch[1]\n",
        "        preds = network(images)\n",
        "        loss = F.cross_entropy(preds, labels)\n",
        "\n",
        "        optimizer.zero_grad()\n",
        "        loss.backward()\n",
        "        optimizer.step()\n",
        "\n",
        "        m.track_loss(loss)\n",
        "        m.track_num_correct(preds, labels)\n",
        "\n",
        "      m.end_epoch()\n",
        "    m.end_run()\n",
        "\n",
        "# when all runs are done, save results to files\n",
        "m.save('results')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>run</th>\n",
              "      <th>epoch</th>\n",
              "      <th>loss</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>epoch duration</th>\n",
              "      <th>run duration</th>\n",
              "      <th>lr</th>\n",
              "      <th>batch_size</th>\n",
              "      <th>shuffle</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0.536794</td>\n",
              "      <td>0.795683</td>\n",
              "      <td>17.768282</td>\n",
              "      <td>19.271004</td>\n",
              "      <td>0.010</td>\n",
              "      <td>100</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1</td>\n",
              "      <td>2</td>\n",
              "      <td>0.376039</td>\n",
              "      <td>0.860333</td>\n",
              "      <td>17.996781</td>\n",
              "      <td>37.417833</td>\n",
              "      <td>0.010</td>\n",
              "      <td>100</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>1</td>\n",
              "      <td>3</td>\n",
              "      <td>0.354022</td>\n",
              "      <td>0.868283</td>\n",
              "      <td>17.962807</td>\n",
              "      <td>55.510283</td>\n",
              "      <td>0.010</td>\n",
              "      <td>100</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>1</td>\n",
              "      <td>4</td>\n",
              "      <td>0.331763</td>\n",
              "      <td>0.875367</td>\n",
              "      <td>17.973242</td>\n",
              "      <td>73.616401</td>\n",
              "      <td>0.010</td>\n",
              "      <td>100</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>1</td>\n",
              "      <td>5</td>\n",
              "      <td>0.325365</td>\n",
              "      <td>0.880133</td>\n",
              "      <td>17.959134</td>\n",
              "      <td>91.719666</td>\n",
              "      <td>0.010</td>\n",
              "      <td>100</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>2</td>\n",
              "      <td>1</td>\n",
              "      <td>1.023266</td>\n",
              "      <td>0.608533</td>\n",
              "      <td>17.481936</td>\n",
              "      <td>18.437012</td>\n",
              "      <td>0.010</td>\n",
              "      <td>1000</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>2</td>\n",
              "      <td>2</td>\n",
              "      <td>0.518115</td>\n",
              "      <td>0.803583</td>\n",
              "      <td>17.295668</td>\n",
              "      <td>35.884855</td>\n",
              "      <td>0.010</td>\n",
              "      <td>1000</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>2</td>\n",
              "      <td>3</td>\n",
              "      <td>0.430224</td>\n",
              "      <td>0.842900</td>\n",
              "      <td>17.297607</td>\n",
              "      <td>53.324161</td>\n",
              "      <td>0.010</td>\n",
              "      <td>1000</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>2</td>\n",
              "      <td>4</td>\n",
              "      <td>0.385267</td>\n",
              "      <td>0.859567</td>\n",
              "      <td>17.207940</td>\n",
              "      <td>70.658459</td>\n",
              "      <td>0.010</td>\n",
              "      <td>1000</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>2</td>\n",
              "      <td>5</td>\n",
              "      <td>0.350805</td>\n",
              "      <td>0.872050</td>\n",
              "      <td>17.114712</td>\n",
              "      <td>87.922248</td>\n",
              "      <td>0.010</td>\n",
              "      <td>1000</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10</th>\n",
              "      <td>3</td>\n",
              "      <td>1</td>\n",
              "      <td>0.803843</td>\n",
              "      <td>0.695550</td>\n",
              "      <td>18.040418</td>\n",
              "      <td>18.172909</td>\n",
              "      <td>0.001</td>\n",
              "      <td>100</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>11</th>\n",
              "      <td>3</td>\n",
              "      <td>2</td>\n",
              "      <td>0.525285</td>\n",
              "      <td>0.801767</td>\n",
              "      <td>18.150892</td>\n",
              "      <td>36.455813</td>\n",
              "      <td>0.001</td>\n",
              "      <td>100</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12</th>\n",
              "      <td>3</td>\n",
              "      <td>3</td>\n",
              "      <td>0.442718</td>\n",
              "      <td>0.836783</td>\n",
              "      <td>18.343555</td>\n",
              "      <td>54.939637</td>\n",
              "      <td>0.001</td>\n",
              "      <td>100</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13</th>\n",
              "      <td>3</td>\n",
              "      <td>4</td>\n",
              "      <td>0.399096</td>\n",
              "      <td>0.853450</td>\n",
              "      <td>18.032484</td>\n",
              "      <td>73.114547</td>\n",
              "      <td>0.001</td>\n",
              "      <td>100</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14</th>\n",
              "      <td>3</td>\n",
              "      <td>5</td>\n",
              "      <td>0.372009</td>\n",
              "      <td>0.863767</td>\n",
              "      <td>18.014012</td>\n",
              "      <td>91.263999</td>\n",
              "      <td>0.001</td>\n",
              "      <td>100</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>15</th>\n",
              "      <td>4</td>\n",
              "      <td>1</td>\n",
              "      <td>1.535063</td>\n",
              "      <td>0.480167</td>\n",
              "      <td>17.057506</td>\n",
              "      <td>17.950979</td>\n",
              "      <td>0.001</td>\n",
              "      <td>1000</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16</th>\n",
              "      <td>4</td>\n",
              "      <td>2</td>\n",
              "      <td>0.759446</td>\n",
              "      <td>0.711767</td>\n",
              "      <td>17.094348</td>\n",
              "      <td>35.176463</td>\n",
              "      <td>0.001</td>\n",
              "      <td>1000</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>17</th>\n",
              "      <td>4</td>\n",
              "      <td>3</td>\n",
              "      <td>0.646758</td>\n",
              "      <td>0.752467</td>\n",
              "      <td>17.009808</td>\n",
              "      <td>52.331167</td>\n",
              "      <td>0.001</td>\n",
              "      <td>1000</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>18</th>\n",
              "      <td>4</td>\n",
              "      <td>4</td>\n",
              "      <td>0.588217</td>\n",
              "      <td>0.776717</td>\n",
              "      <td>17.077423</td>\n",
              "      <td>69.553246</td>\n",
              "      <td>0.001</td>\n",
              "      <td>1000</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>19</th>\n",
              "      <td>4</td>\n",
              "      <td>5</td>\n",
              "      <td>0.551008</td>\n",
              "      <td>0.795183</td>\n",
              "      <td>16.956005</td>\n",
              "      <td>86.645293</td>\n",
              "      <td>0.001</td>\n",
              "      <td>1000</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "    run  epoch      loss  accuracy  ...  run duration     lr  batch_size  shuffle\n",
              "0     1      1  0.536794  0.795683  ...     19.271004  0.010         100    False\n",
              "1     1      2  0.376039  0.860333  ...     37.417833  0.010         100    False\n",
              "2     1      3  0.354022  0.868283  ...     55.510283  0.010         100    False\n",
              "3     1      4  0.331763  0.875367  ...     73.616401  0.010         100    False\n",
              "4     1      5  0.325365  0.880133  ...     91.719666  0.010         100    False\n",
              "5     2      1  1.023266  0.608533  ...     18.437012  0.010        1000    False\n",
              "6     2      2  0.518115  0.803583  ...     35.884855  0.010        1000    False\n",
              "7     2      3  0.430224  0.842900  ...     53.324161  0.010        1000    False\n",
              "8     2      4  0.385267  0.859567  ...     70.658459  0.010        1000    False\n",
              "9     2      5  0.350805  0.872050  ...     87.922248  0.010        1000    False\n",
              "10    3      1  0.803843  0.695550  ...     18.172909  0.001         100    False\n",
              "11    3      2  0.525285  0.801767  ...     36.455813  0.001         100    False\n",
              "12    3      3  0.442718  0.836783  ...     54.939637  0.001         100    False\n",
              "13    3      4  0.399096  0.853450  ...     73.114547  0.001         100    False\n",
              "14    3      5  0.372009  0.863767  ...     91.263999  0.001         100    False\n",
              "15    4      1  1.535063  0.480167  ...     17.950979  0.001        1000    False\n",
              "16    4      2  0.759446  0.711767  ...     35.176463  0.001        1000    False\n",
              "17    4      3  0.646758  0.752467  ...     52.331167  0.001        1000    False\n",
              "18    4      4  0.588217  0.776717  ...     69.553246  0.001        1000    False\n",
              "19    4      5  0.551008  0.795183  ...     86.645293  0.001        1000    False\n",
              "\n",
              "[20 rows x 9 columns]"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3WJ2b7t-Ibdv",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# helper function to calculate all predictions of train set\n",
        "def get_all_preds(model, loader):\n",
        "  all_preds = torch.tensor([])\n",
        "  for batch in loader:\n",
        "    images, labels = batch\n",
        "\n",
        "    preds = model(images)\n",
        "    all_preds = torch.cat(\n",
        "        (all_preds, preds),\n",
        "        dim = 0\n",
        "    )\n",
        "  return all_preds"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SfSculOmI5Y3",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# bigger batch size since we only do FP\n",
        "prediction_loader = torch.utils.data.DataLoader(train_set, batch_size=1000)\n",
        "train_preds = get_all_preds(network, prediction_loader)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3Nfb8Z3aKz47",
        "colab_type": "code",
        "outputId": "d7432c9d-25ba-4691-8546-cc6c1d90eec2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 235
        }
      },
      "source": [
        "!pip install scikit-plot"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: scikit-plot in /usr/local/lib/python3.6/dist-packages (0.3.7)\n",
            "Requirement already satisfied: matplotlib>=1.4.0 in /usr/local/lib/python3.6/dist-packages (from scikit-plot) (3.0.3)\n",
            "Requirement already satisfied: scipy>=0.9 in /usr/local/lib/python3.6/dist-packages (from scikit-plot) (1.3.1)\n",
            "Requirement already satisfied: scikit-learn>=0.18 in /usr/local/lib/python3.6/dist-packages (from scikit-plot) (0.21.3)\n",
            "Requirement already satisfied: joblib>=0.10 in /usr/local/lib/python3.6/dist-packages (from scikit-plot) (0.14.0)\n",
            "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib>=1.4.0->scikit-plot) (2.4.2)\n",
            "Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.6/dist-packages (from matplotlib>=1.4.0->scikit-plot) (0.10.0)\n",
            "Requirement already satisfied: kiwisolver>=1.0.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib>=1.4.0->scikit-plot) (1.1.0)\n",
            "Requirement already satisfied: python-dateutil>=2.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib>=1.4.0->scikit-plot) (2.5.3)\n",
            "Requirement already satisfied: numpy>=1.10.0 in /usr/local/lib/python3.6/dist-packages (from matplotlib>=1.4.0->scikit-plot) (1.16.5)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from cycler>=0.10->matplotlib>=1.4.0->scikit-plot) (1.12.0)\n",
            "Requirement already satisfied: setuptools in /usr/local/lib/python3.6/dist-packages (from kiwisolver>=1.0.1->matplotlib>=1.4.0->scikit-plot) (41.4.0)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Cjmpnd7jeSvc",
        "colab_type": "code",
        "outputId": "a7d62f30-ea10-4016-eeae-a17fcaee0184",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 199
        }
      },
      "source": [
        "# use scikitplot to plot the confusion matrix\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "from sklearn.metrics import confusion_matrix\n",
        "import scikitplot as skplt\n",
        "\n",
        "cm = confusion_matrix(train_set.targets, train_preds.argmax(dim=1))\n",
        "cm"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[5075,   51,   87,  400,   52,    8,  269,    1,   57,    0],\n",
              "       [   4, 5805,   34,  104,   30,    2,   19,    0,    2,    0],\n",
              "       [ 114,   15, 3567,   43, 1184,    6, 1003,    0,   68,    0],\n",
              "       [ 294,  141,   14, 5136,  186,    3,  210,    0,   15,    1],\n",
              "       [  21,   50,  640,  310, 4558,    1,  387,    0,   32,    1],\n",
              "       [   1,    2,    1,    3,    0, 5373,    1,  474,   38,  107],\n",
              "       [1435,   52,  824,  239, 1046,    1, 2305,    0,   96,    2],\n",
              "       [   0,    0,    0,    0,    0,  234,    0, 5336,   10,  420],\n",
              "       [  54,    4,   62,   62,   38,   41,  117,   23, 5590,    9],\n",
              "       [   1,    2,    0,    1,    2,   55,    0,  304,    4, 5631]])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 13
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_iMsrhFiHyLu",
        "colab_type": "code",
        "outputId": "112e1d80-00cf-49d4-eb2c-6d2dd3576741",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 313
        }
      },
      "source": [
        "skplt.metrics.plot_confusion_matrix(train_set.targets,train_preds.argmax(dim=1), normalize=True)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7f4147301ba8>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 14
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAATUAAAEWCAYAAAAHJwCcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzsnXd8FFX3h58LAUJooZMCKfQSCAlR\n6UWqdAIIFuy++qJg/YmIgCi+NlBErAgoSA+hN5WigJTQewmGkoQWIISQujm/P2az2U3dwA4kYR4+\n98Pu3Dvnnj2ze3Nn7sz5KhHBwMDAoKhQ7F47YGBgYOBIjEHNwMCgSGEMagYGBkUKY1AzMDAoUhiD\nmoGBQZHCGNQMDAyKFMagVgBQSo1XSs0xv66llLqplCru4D4ilFKdHWkzH31/pJS6opS6cAc2dInL\n3UYpNVopNf1e+1GUuS8GNfMP+pJSqozVtueVUpvuoVvZIiJnRaSsiJjuZr9KqQeUUquVUteVUleV\nUjuVUs84wG4t4E2gkYjUuF07esZFKSXm74eT1bYS5m123ciplOqglDqfVzsR+VhEnr8Tfw1y574Y\n1MwUB0beqRGlUaTippRqCWwANgN1gMrAy0APB5ivBcSIyCUH2NKTa9h+3h7mbQ7DetA00I8i9ePM\ng8+Bt5RSrtlVKqVaKaV2KaVizf+3sqrbpJSaqJTaCtwCfM3bPlJKbTOfFq1QSlVWSv2mlLphtuFt\nZWOKUuqcuW63UqptDn54m2cOTkqplmbb6SVRKRVhbldMKTVKKRWulIpRSi1USlWysvOkUuqMue49\nO2Lzi4h8KiJXRGO3iAy2sveCUuqUeRa3XCnlblUnSqmXlFInzTO9aebBvzPwO+Bu9n9WdjMa61Nj\n84wxzByni0qpyZnjYn7vbvbjqtmvF6zsjTfH41elVJxS6rBSqkUeMZgNDLN6Pwz4NZOfzyiljppt\nnlZK/ce8vQywxupz3jT7N14ptVgpNUcpdQN4WtleanhUKfWvUqq8+X0PpdQFpVTVPHw1yA0RKfIF\niAA6A0uAj8zbngc2mV9XQvur/CTgBAw1v69srt8EnAUam+tLmLedAmoDFYAjwAlzP05oP4iZVj48\ngTYDckI7HbsAOJvrxgNzzK+9AQGcMn2GEmgzqf+Z348EtgOeQCngB2Ceua4RcBNoZ66bDKQCnbOJ\njQtgAjrmEr9OwBUgwGxvKvCXVb0AKwFXtJnZZaC7ua4DcN6qrc176+Njfv0P8KT5dVngoeziAvwF\nfAs4A/7mPjtZxTMReARthv4/YHsun0+AJsBF82eoaH7dBBCrdj3Nx1sB7dH+wAXk8rnGAylAP7QJ\nRGnrY21u8xswy/zdiAJ63evfS2Ev99NMDWAs8Go2fwl7AidFZLaIpIrIPOAY0NuqzSwROWyuTzFv\nmyki4SISi/aXOlxE/hCRVGAR0Dx9ZxGZIyIx5v0noQ0O9fPh+9dAHJA+63oJeE9EzotIEtqPZaB5\nJjMQWCkif5nr3gfScrBbEe0HF51L348DM0Rkj9neu0BL65ko8ImIXBeRs8BGtIHmdkgB6iilqojI\nTRHZnrmBUqom0Bp4R0QSRWQfMB3bmdYWEVkt2jW42UCzPPpNBFYAj5rLcvM2CyKyyny8RUQ2A+uB\nbGfcVvwjIktFJE1EErKpH472R2MTsEJEVuZhzyAP7qtBTUQOoc0oRmWqcgfOZNp2BvCwen8uG5MX\nrV4nZPO+bPobpdRb5lOXWKXUdbTZXRV7/Daf5nQAHhOR9MHJCwg1n+5dB46izbiqmz+PxV8RiQdi\ncjB/DW3Ac8vFBZv4iMhNsz3r+FivbN7C6rPnk+eAesAx8yl8rxz8uSoicVbbMh+vzP4423FN61e0\ngTHLqSdYTg+3m095r6PNBPM6htl9byyIyHW0P4BNgEl52DKwg/tqUDMzDngB2x9AFNogYU0tINLq\n/W2nMzFfP/s/YDBQUURcgVi00xh79v0Q6CsiN6yqzgE9RMTVqjiLSCTarKumlQ0XtNObLIjILbRT\nvuBc3LCJj/kaUmVs42Mv8WinvOm2igOWmbOInBSRoUA14FNgsbJatbbyp5JSqpzVtszH63b4G21w\nrw5ssa5QSpUCQoAvgOrmY7iajGOY0/cj1++NUsofeBaYhzYbN7hD7rtBTUROAQuAEVabVwP1lFKP\nmS/QP4p2XcpRpwLl0K5pXQaclFJjgfJ57WQ+zVoIDBORE5mqvwcmKqW8zG2rKqX6musWA72UUm2U\nUiWBCeR+rP8P7SL220qpymZ7zZRS883184BnlFL+5h/3x8AOEYnI85Nn5QTarKmnUqoEMAbtVDz9\nMz+hlKpqnpFeN2+2OXUWkXPANuB/SilnpVRTtBnenNvwx9quoF1y6GN+bU1Js5+XgVSlVA+gq1X9\nRaCyUqqCvf0ppZzNPo8GngE8lFL/vYOPYMB9OKiZmQBY/vqLSAzQC+0Cfgzaj7yXiFxxUH/rgLVo\nP+gzaNdqcj0tMfMw2qxhsdWq2mFz3RS06z7rlVJxaIsGD5o/z2G0azVz0WZt14Ac76ESkW1o13U6\nAaeVUleBH9EGe0TkD7TrciFme7WBIfZ++Ex9xQL/RbsGFok2c7P2rTtwWCl10/wZh+RwLWoo2uJB\nFBAKjDP7eUeYr5sezmZ7HNofwoVo8XwMLf7p9cfQBv/T5ksC7pltZMP/gHMi8p35WuUTwEdKqbp3\n+jnuZ1TWP0gGBgYGhZf7daZmYGBQRDEGNQMDgyKFMagZGBgUKYxBzcDAoEhRoB6wVaXKiXLJ9naq\nO8Lf1657XAsEpjR9Fm6KF8vzlrjbQh+roFMY0CMMei216RHbM2ciuHLlyh2ZLl7eSyQ1uwXprEjC\n5XUi0v1O+ssvBWtQc6lMqQ55PXudf7YsKDyZXuISU3WxW85Zn0NdTKfBMjFZn8xLziUdn44tTacR\nWI/Ytn4wr+f680ZSEyhVf3DeDYHEfdPu+oyiQA1qBgYGhQEFBTj7ljGoGRgY5A8FFCu4CYiNQc3A\nwCD/KL2upt45xqBmYGCQT4zTTwMDg6JGAZ6pFbjhtktzT/Z/M4hD3w7mrQFZ8/rVrFKGtRN68s+k\n/uz8cgDdArQMO7WqluXq/GfYPnkA2ycP4OuX2tjst37dWvybNMCvYV2++PyTLHaTkpIY9vgQ/BrW\npX2bhzgTEQFATEwMPbp2olqlcrwx8hXdbQJs+GMdrQMb85B/Q6ZO/ixbuy8+/RgP+TekR6fWnD2j\n2Q1ZOJeH27SwFDfXUhw6sE93fzPHpGnj+jRuUIfPP8u+jycee5TGDerQttWDlj6y44/1awnyb0SA\nX32+/OLTbG09O2woAX716dy+pSUO6Zw7dxbPahWY+lX2acoc6Wthi+0dodBmavaUe4CuvSqluiul\njptzyGdOzJjVmWLF+OrF1vT9cC3NRyxmUJvaNPC0lRR4Z1BzQraepuWboQybtIEp/2ltqTt98QYP\nvbGEh95YwojvM9JhmUwm3hj5CqHLV7N7/2EWLZjP0aNHbOz+MvNnXF1dOXj0JK+MeI3339PcdXZ2\n5v1xE/j4k89t2uthM93uu2+OZO7iFfy1cz+hIQs4fszW7txfZ+LqWpHt+47yn/+O4KNxowEIHvwY\nf24J488tYXzzw0xqefnQpKm/rv5m9v21EcNZtmINew8cYdH8eRw9YtvHrBk/U9G1IoePneLVka/z\n3uh3crT19hsjWBS6ku27DxKyaAHHMvk7+5cZVHCtyJ6Dx3n5ldcY//67NvVjRr1F567Z3yLlaF8L\nU2zvHKXN1Owp9wDdBjVz8r9paKo8jYChSqlGue3zQKA/4dE3iLgYR0pqGou2hNPrAdvcjSJQ3qUk\nABXKlCT66q08fQnbtRPf2nXw8fWlZMmSDBz8KCtXLLNps3LFch5/8ikA+g8YyKaNfyIilClThlat\n21DK2Vl3mwB7d+/Cx7c2Xj6a3X4DBrNu1QqbNutWr2DwY08C0KtfMFs2byRztpXQxQvoFzxId3+t\n2bVzJ7Wt+hj06JBs+lhm6WNA8EA2bfgzi+8Au8N24utbG29zHAYMHMzqlctt2qxZuZyhj2tx6Ns/\nmM2bNlhsrVqxjFpe3jRomP1XzpG+FrbYOoRixe0r9wA9Z2oPAKdE5LSIJAPzgb657eDhVp3zV25a\n3kfGxONR2Tbp6cQFuxnSvg6nfhpK6JjuvPHTNkudd7Vy/DOpP+s/6kXrhhkSk1FRkXjW9Mzox8OT\n6EjbJKlRUZF4emqnsk5OTpQvX4GYmJwyYOtjEyA6KhJ3jwy7bh4eREdH2baJzmjj5OREufIVuHrV\n1u6yJYvpN/BR3f3Naf/0PiKz66OmVR8Vsu8jOioKDytb7h6eWeIQZdUm3d+rMTHcvHmTKZM/453R\nY++Kr4UttneOum9PPz2wTYR4HtsU2gAopV40S6KFSUrej14MbluHORtOUOeFefT/aC0/v9YBpeDC\ntVvUe3EeLd8M5Z0Z25n1RkfKlS7huE9TiNgTtpPSLqVp2KjJvXblnvDpxA94+ZXXKFv2dmUSDHJF\ncX+eftqLiPwoIi1EpEXUlRt4Vsn4InpULkNkTLxN+6cerk/I1tMA7Dh+CecSxalS3pnk1DSuxiUB\nsPf0FU5fuEFddy2zsru7B+fPZSRXjYw8j5uH7fjq7u7B+fPaGJyamsqNG7FUrpzzc6h62ARwc/cg\nKjLDbnRkJG5utklU3dwy2qSmphJ3I5ZKlTLsLg1ZSP/gR2320cvfnPZP78Mjuz7OWfURm30fbu7u\nRFrZioo8nyUO7lZt0v2tVLkyYWE7GTdmFE0b1ua7aV8z+YtP+PH7abr5Wthi6xDu05laJFbiH2j6\nlLkKY+zas586buXxqlaOEk7FGNSmNqt2nbVpc+7KTTo01b7c9T1dcS5ZnMuxiVQp72x5Vs67ejnq\nuFXg34ua2FBgiyDCT50k4t9/SU5OZvHCBfTs1cfGbs9evflt9i8AhC5ZTPsOnVC5/KXRwyaAf0AL\nToef4kyEZnfpkoV0fcRWUKnrI71YOHc2ACuXhtC6XQeL3bS0NJaHLqZfsO2zeXr5a02LoCBOWfWx\naMH8bProY+ljSchi2nfMvo+AwCDCreKwZPFCevTsbdOme8/ezPtNi8Oy0BDate+IUoo1v2/mwNFw\nDhwN5+XhI3jjrVG8+NJw3XwtbLG9cwr26adugqJo98CdBnzQRCv2A41z20e5eknfCWvkROR1CY+O\nlbFzdopzvx9l4oLdEjxxrTj3+1H8X1ko245Ey/7TV2Tf6SvSc9wqce73owz5ZL0cPnNV9p2+IntO\nXZYBH2ntnfv9KPFJaRKydKXUqVNXfHx8ZdwHH0p8UpqMGj1GFi5eKvFJaRITe0v6Dxgovr61JbBF\nkBw6ekrik9IkPilNanl5ScWKFaVMmTLi7uEhYfsO6WbzQmyyzFm0THxr1xEvb18ZNeYDuRCbLK//\n32j5ZV6IXIhNloiLN6RX3wHi7VNb/ANayI59x+RCbLJciE2WkJW/S0CLByzv04te/iakiE0JXb5K\n6tStKz6+vjJ+wkeSkCLy7nvvy6IlyyQhReRaXIL0Dx4ovrW1Po4cD89iIyFF5Fp8qiwIWS6169QV\nbx9feW/cBLkWnypvj3pPflsYKtfiUyU65qb07R8sPr61JSCwhew9dEKuxafalHdGvy8TJn5qea+H\nr4UptgEBgXKnv21Vzl2cO35kVwHC9BpjcvRPT40CpdQjwFdoKtkzRGRibu2LVfQWPbJ0xBhZOows\nHWaMLB0t2L077I4MFyvvIaWChufdEEjc8N5uEbnz1CD5QNcnCkRkNWZFIgMDg6KC8ZiUgYFBUaMA\nPyZlDGoGBgb5x5ipGRgYFBnu4T1o9mAMagYGBvnHSBJpYGBQdDAWCuymuW8Vti56weF2KwblntLl\ndonZMdXhNsuU0ucvoF63XuiFHrde6EVhi61DME4/DQwMigzp+dQKKMagZmBgkE+M008DA4OihrFQ\nYGBgUKQwrqkZGBgUGZRx+mlgYFDUKMAztYI73OJY9ZwurRqyP/R9Di0bx1vPdMlSX8utIqu/f5Wd\nC95l3U8j8aimCb60a1GX7fNHWcq17V/Su0NTi396KAj9vn4tzf0a0qxRPSZ9nr2K0lNPDKFZo3p0\nbNvSYnfDH7/TtmUQDwY2o23LIDZv3JBjPPRSJipMdguTr3ravR2UUnaVe4KO+dRmAJeAQ/buExAQ\naMn7dDMxVXx8feXI8XCJjU8SP7+msmf/YZvcUF99PU2ef+E/kpAi8suceRI8aHC2+a5cAl6R8LOX\npEHPsVKuxQjZf/yc+A/4UJz9h1tKyPrd8tz7v4qz/3Dp9sIU+W3FDpt6Z//h4tbubYm5flMqPvSa\nlGr2svj4+Mqho6fkWlyiNPFrasmDlV6+nPKNPPf8ixKflCazZs+V4IGDJT4pTS5djZPfN/wlU6Z+\nK/956b82+1yPTxYfH185cOSkxNxIkCZ+TWXX3oMSl2iylMlTvpFnn39R4hJNMvPX32TAwEESl2iS\nLdvD5MTpcxKXaJIdu/eLm7u7ZR/reDgytoXVbmHy1ZF2HZFPrZirl7gEz7CrcA/yqek5U5sFZK9P\nZgeOVM8JauJN+LkrRETGkJJqYtG6PfQyz7bSaeDrxuadxwHYvOsEvTr4ZbHTv3Nz1m89QkJiCnLr\nki4KQpoyUW2L3eBBj7Jyha2K0qoVy3jsiWEA9BswkE0bNRWlZv7NcXPXsgI3bNSYxIQEkpKSdI1t\nYbVbmHzV0+5toRSqmH3lXqDboCYifwFXb3d/R6rnuFerwPmL1yzvIy9ew6NqBZs2B09E0reTppHZ\nt1MzypctTaUKtkpWg7oFsHDtbu3zpdzUTU3Kw+ZzexAdldlulI3dCtnYXRYaQjP/AEqVKpWlD72U\niQqT3cLkq552b5eCfPp5z6+pWatJXb5y+Z758e6XobQNrMM/896hbWAdIi9ew2RKs9TXqFKexnXd\n+f2fI7lYKRgcPXKYse+9y5RvvrvXrhgUUYxBLRfESk2qapWqlu2OVM+JuhSLZ/WKlvce1SsSeTnW\npk305ViGvDWdlkM/Zdw3mnhw7M0Myb7gLgEs33CA1FRtoFMlyuqmJhVp87kjcXPPbNfdxm6sld3I\n8+cZOjiYH36ehW/t2tn2oZcyUWGyW5h81dPu7WIMareBI9Vzwg6foU6tqni5V6aEU3EGdQtg1aYD\nNm0qu5ax7Pv2s934Zdl2m/rB3QNZuDbM8l65VNNFQUhTJjplsRuyaAE9e9mqKD3Sqw9z5/wKwNIl\ni2nfQVNRun79OgP79+aDjz6mZavWOfahlzJRYbJbmHzV0+5tofJR8jKlVHel1HGl1Cml1Khs6msp\npTYqpfYqpQ6YdU9yR89VCMCb21z9dKTaj7P/cOn7yjQ5EXFRws9ekrFTl4uz/3CZ+MNqCR75vTj7\nD5ehb/0kJ89clBMRF2XGkq1SPmikZdWzXo/3JfLiNSnd/BXbFVMdFITiEk2yeOkKqW22O3b8hxKX\naJJ33h0j8xeHSlyiSS5fj5d+A4Itdg8cOSlxiSZ5f9wEcXFxEb+mzSzl9NnoLKufjoxtYbZbmHx1\nlF1HrH4Wr+QjFR6bbVchl9VPNEGmcMCXDMW5Rpna/Ai8bH7dCIjIyz/d1KSUUvOADkAV4CIwTkR+\nzm2fwMAWsnVHWG5NbovClHooTafj4VS8wE7KDe4ijlCTcqrsK+Uf+ciuttfmPJ6jmpRSqiUwXkS6\nmd+/CyAi/7Nq8wNwWkQ+NbefJCKtcvXPvo+Rf0RkqF62DQwM7i35OK2topSynqn8KCI/ml97AOes\n6s4DD2bafzywXin1KlAG6JxXh8ZjUgYGBvnDzutlZq7kNFOzk6HALBGZZJ6pzVZKNRGRtJx2MAY1\nAwODfOOgBYhIoKbVe0/zNmuew3wTv4j8o5RyRrukdSkno8aFFgMDg3yhsO92DjsGvl1AXaWUj1Kq\nJDAEWJ6pzVngYQClVEPAGcj1hlZjpmZgYJBvHPEIlIikKqVeAdahrYTOEJHDSqkJaKumy4E3gZ+U\nUq8DAjwteaxuGoOagYFB/lAOO/1ERFYDqzNtG2v1+giQ802X2VCgBjUBUk05Xv+7ba7t+sbhNgH6\n/bjD4TaXPP+Aw20CHI28oYvdhh7ldbEbfvGmLnZrVy/rcJtpafrchlOQVaruWVohOyhQg5qBgUHh\nwBjUDAwMigzpCwUFFWNQMzAwyD8Fd0wzBjUDA4N8oqBYsYJ7N5gxqBkYGOQb4/TTwMCgaFFwx7SC\n90RBYVJSCqxZgelDmzLj8WYMbu6Wpb5L/SrMfyaAaYObMG1wE7o3zEiC+exDNfn+UT++f9SPdnUq\nZfFPD5Uqa7Zu+oP+nQLp096fmd9OzlI/Z/o3BHd+gMHdW/Gfx3oTdf5srvasfXdEbLds/J1e7ZrT\no3Uzpn8zKUt92PYtDOrehmZerqxfudSyfefWvwju2spSAmpX4c+1K3T1Nd2W3sfMUJOyD90GNaVU\nTXNytyNKqcNKqZF57SMivDnyVZYsW8WufYdYvHA+x47aps/+ddYMXF0rsv/ICYa/OpKxY7S8cpWr\nVGFhyDJ27N7PD9Nn8sJzT2Xbh8lk4rURw1m2Yg17Dxxh0fx5HD1i28esGT9T0bUih4+d4tWRr/Pe\n6HeytTW8nTdjVh3nxXkH6FC3MrUqls7S5q9TMQxfeIjhCw+x9qj2dMcDXq7UqVqG/y48yMiQwwz0\nd8OlRHGLf2+MfIXQ5avZvf8wixbM52imGPwy82dcXV05ePQkr4x4jfff02Lg7OzM++Mm8PEnn+ca\nZ5PJxKdj32TqrMWE/L6TtctDOH3ymE2b+o2aMmfFJhau3UbnHn2Z8r+xOVizteuI2JpMJj4a8ybf\nzV7C8o27WL1sMeEnbP1z86jJR5O/55F+g222P9C6HSHrtxGyfhszFqzE2dmFVu0f1s3XdFt345g5\nyt87xd4BrcgNakAq8KaINAIeAoYrpRrltsOt+PhCo6SkgOjYRC7cSCI1Tdh86iotfSpiD7UqluZQ\n1A3SBJJS0/g35haBtTQhGE1NyvEqVdYc2rcbTy9fPGv5UKJkSbr1HsCm9ats2gS1akfp0i4A+DUP\n4tKFqDw/l6Niu2vnTmp5+1LTS/OvR99gNqxfadPGo6YX9Rs1yfUG1fWrltK2YxfL59DDV7g7x6xA\nqUlxn87URCRaRPaYX8cBR9HyJ+VIckpKoVFSUgou30y2vL9yM5nKZUpk6a+NbyW+e9SP97rVpUrZ\nkgCcjrlFYC1XSjkVo7yzE03dy1PVXKf17XiVKmsuX4yihpXuQTU3Dy5djM6x/dKFs2ndIasAdGYc\nFduoqEhquGX4V72GB5eic/YvJ9YsD6FHv4G6+prRTt9jVuDUpAqwRN5dWShQSnkDzYEszxUppV4E\nXgSoXLnKHfeVrqS0dOXaO7Z1p2yPuM6mkzGkpAmPNKrGW518GbX8GHvOxVKvWhkmD2hEbEIqRy/e\nRKcnbe6YVaELOHJgL9MXrM67cQHi8sULnDx2mNbt88wpaHAbFOTVT90XCpRSZYEQ4DURyfIAorWa\nVOXKlQuNkpIIltkVQJWyJYmJT7FpE5eUSop5tFp79BJ1q2boiM7fHcXwhYcYveIYCoi8nmDVt+NV\nqqypWt2dC1Yz4EvRkVSrnnWhY8eWjfz8zRd8NX0+JbOZ9WbGUbF1d/fgQnSGfxcvRFLNLat/ubF2\nxRIe7t6bEiWyzp4d6WtGO32PWYFSk1L36ekngFKqBNqA9puILMmrvUuZMoVGSUkA9wrOVC9XCqdi\nivZ1KrH932s2bSq5ZPygHvKuyNlriQAUU1CulDZJ9qlcGp/KLuw+p0n2aWpSjlepsqZxswDORYQT\neS6ClORk1q1YQvsutiI9xw7tZ+Lo1/hq+nwqWUkX5oajYtsiKIiz/4Zz/qzm35plIXTs0tPuzwew\nZtkiHuk7SHdf4e4cs4KkJqXQLr/YU+4JOipJKeBX4Ct792keEFiolJTGrDgq567dksjrCTJz+1np\nNm27zNl5TsatOibdpm2X+bsjJSImXsIv35R952Plud/2Sbdp26XX9zskIuaWRMTckiPRcfLy/APS\nbdp26TZtu8QnpemiUrUnItamfD1zkdTyqS2etbzlv2+NkT0RsfLCiP+TL3+aJ3siYuWB1u2lUpWq\nUq+hn9Rr6CftOvfIYmNPRKxusf32l8Xi5VNbPL185NX/GyuHzsfJSyPfkakz5suh83Eyb+UmqV7D\nXUqXdpEKrhWldr0Gcuh8nBw6Hyfr/jkk1aq7yYGzsZZt6UUPX/U6ZgVVTapU9bpS5601dhVyUZPS\nq+ipJtUG+Bs4CKTnExptzp+ULQGBLeSvbTsd7oteSkqFKfXQ8eg4XewaqYcKV+ohR6hJOdeoJ15P\n2aekduKz7jmqSemFnmpSWyjQ9x0bGBjcFvfy1NIOjMekDAwM8oWiYCewNAY1AwODfGPM1AwMDIoU\nBfk+NWNQMzAwyB/GNTX7SRMhKdXxwit6sfTFBx1us2Lw9w63CXBxwYu62NULj2ySAxRUCvL1JT1Q\nKCNJpIGBQdHCmKkZGBgUKYxragYGBkUH45qagYFBUUJ79rPgjmrGoGZgYJBvCvCYZgxqBgYG+acg\nr/gWuHXZP39fx4PNGxPUtAFTJn2WpT4pKYnnhj1GUNMGdO3QirNnIgA4eyYCzyrl6NAykA4tA3lz\nxH9t9itMgi5dmtdk/7dDOPT9UN4K9s9SX7NKWdZ+1Jt/vhzIzimD6BZYy1LXxKsSmz7tx+6pg9k1\nZRClzNoHAH+sX0tg04b4N67H5Bxi8PQTQ/BvXI9ObVtyxhzb3bt20ubBANo8GEDrB5qzYlmo7jFI\n9zfIvxEBfvX58ovs/X122FAC/OrTuX1Ly3chnXPnzuJZrQJTv8oq3OJoXwuz3XxTwPOp6Zl6yBnY\nCewHDgMf5LVPU//m4u3jK2EHj0vU1Xhp3MRPtu7aL1dupljKZ5O/lqeefUGu3EyRH2fOkb4DBsmV\nmymy5/BJadCwsU3b9HI9Pll8fHzlwJGTEnMjQZr4NZVdew9KXKLJUiZP+Uaeff5FiUs0ycxff5MB\nAwdJXKJJtmwPkxOnz0lcoklm1PNYAAAgAElEQVR27N4vbu7uln2sU7rcTEwVH19fOXI8XGLjk8TP\nr6ns2X/Yps1XX0+T51/4jySkiPwyZ54EDxqcJTWMS7/vJTzqujR4YY6UG/CD7D99WfyHzxfnPt9Z\nyvS1h+XVbzeLc5/vxH/4fIm4cEOc+3wnZfp9Lwf+vSJBIxaKc5/vxP3xGeLS73tx7vOdXL2ZLN4+\nvrLvyEm5HKvFYMeegxKbYLKUL776Rp55/kWJTTDJz7/8Jv2DB0lsgkmiY+IkJi5JYhNMcvz0ealS\ntarlvR4xSEgRuXIjSbx9fGXvoRNy8dotadykqfwTdkCuxadayudfTpWnn3tRrsWnyvRZmr/W9X36\nDZC+/YNlwsRPLdv08FWvGOhh1xGph8p41JMH/7fJrsI9SD2k50wtCegkIs0Af6C7Uuqh3Ha4dSse\nH9/aePto4hL9Bz7KmlW28mZrVq1gyONPAtCnfzB/b9qQPojmiCaMUTgEXYLqViP8wg0iLsaRkprG\nor/D6fWAt00bESjvomXdreBSkuhr8QB0bl6TQxExHIzQ8tJfjUuypMXZnR4Dc2wHDHqUVSttY7B6\n5TIeezwjBpvNsXVxccHJSbtSkZiUmONfYEeKg+wO24mv1XdhwMDBrM7k75qVyxlq/i707R9s8Re0\n41nLy5sGDbPX+tFLyKSw2b097lM1KdFIT4pVwlxyjXBKcgrunhkCFu7ZCK9ER0VZxFnSxSWumsUl\nzp75l46tWtC7Wyf+2brFap/IQiPo4l65DOevZOQSi4y5iUflMjZtJs4PY0j7upz6+QlCxz7CGz9q\nn7WuewVEYPn4nmybHMwb/TNOXaOyi0FkHrEtnxHbsJ07eDDAj1YtmvHl199aBjk9YpDZFwB3D0+i\no20VraJy8PfmzZtMmfwZ74zOWdZPLyGTwmb3dinImW91XShQShUHdgN1gGkikqvwSqU7EF6pXsON\nfUdPU6lyZfbt3c2wIQPZums/5co7JolhQRJ0Gdy2DnM2HGfKsgM8WL86P7/eicBXF+JUvBitGtWg\nzZtLuJWUypoPe7En/DKbDkTmbTQPWjzwIDv2HOT4saO89PwzdOnWA+c8ZN3uFZ9O/ICXX3mNsmUd\nnxDSAFD38UKBiJhExB/wBB5QSjXJpo1FeKVS5UpEnc8QsIjKRnjFzd3dIs6SLi5RqXJlSpUqRSWz\nyIR/80C8fXw5deqEeR+PQiPoEhUTj2eVjB+jR+WyRMbE27R5qksDQraGA7Dj+EWcSzhRpbwzkTE3\n2XI4mpi4RBKSU1m7+yzNa1ex9J0lBh55xPZGrCWm6dRv0JAyZcty5PAh3WKQ2ReAqMjzuLm5Z7KV\nvb9hYTsZN2YUTRvW5rtpXzP5i0/48ftpuvlamO3eDun3qd13p5/WiMh1YCPQPbd2Li5lOB1+ijMR\nmrhE6OIFdH+kl02b7o/0Yv5vswFYHhpC2/aa8MqVy5cxmUwARPx7mtPhp/D29gXShTEKh6BL2MlL\n1HGrgFe1cpRwKsagtrVZtTPCps25yzfp0FQ7Ta/v6YpzyeJcjk3k9z3naOxVidIlnSheTNG2iTtH\nz2piMAHpMTDHdsmiBTzSM1MMevZh7m8ZMWhnjm1ExL+kpqYCcPbMGU4eP4aXl7duMQAICAwi3Oq7\nsGTxQnpk8rd7z97MM38XloWGWPxd8/tmDhwN58DRcF4ePoI33hrFiy8N183Xwmz3dinIg5qeq59V\nAVfz69JoegW9ctunWfMAmReyXHzr1BVvH18ZPXaCXLmZIm++857MXrBErtxMkfNX4qRPv2Dx8a0t\nzQNbSNjB43LlZorMnLNA6jdoJE38mopfM3+ZszDUsvoZl2gqNIIuzn2+k74frJIT569JeNR1GTt7\nhzj3+U4mzt8lwR+ttqx4bjsSLftPX5Z94Zel59gVlpXRpyf9IYfPxMihiBiZFLLXsj02wSSLQrUY\nePv4ypjxH0psgkn+790xMm9RqMQmmOTitXjp21+LbUBgkOw7clJiE0zyw8+zpEHDRuLXtJk09W8u\nvy0IsayY6iW8ci0+VRaELLf4+964CXItPlXeHvWe/LYwVK7Fp0p0zE0rf1vI3kMnbFY/r8Wnyjuj\n38929dORvuoVg4IqvFLWs760m7zFrkIRE15pCvwCFEebES4UkQm57eMfECh//u14MZNSTvpMSPUQ\ndClsqYdK6hTbxGSTLnadSxbPu1ERxhHCK+VqNpDAN2fY1Xbz661zFV5RSnUHpqCNE9NFJMsNeEqp\nwcB4tIXG/SLyWG596im8cgBNld3AwKAo4aCVTfNC4jSgC3Ae2KWUWi4iR6za1AXeBVqLyDWlVLW8\n7BqPSRkYGOQLLUmkQ66XPQCcEpHTAEqp+UBf4IhVmxfQ7py4BiAil/IyWuAekzIwMCj4FFPKrgJU\nUUqFWRXr6yAewDmr9+fN26ypB9RTSm1VSm03n67mijFTMzAwyDf5OP28kts1NTtwAuoCHdBuDftL\nKeVnvqMiW4yZmoGBQb5QjnugPRKoafXe07zNmvPAchFJEZF/gRNog1yO5DioKaXK51by8tbAwKDo\nUkzZV/JgF1BXKeWjlCoJDAGWZ2qzFG2WhlKqCtrp6OncjOZ2+nkYbQnV2rX09wLUym6nO6G4UpQp\n5fgz4vSHuh3NxdhEh9s8P/d5h9sEaPj6srwb3QbhU/vrYjdVp2OmB3p9vwr0o0gO8E1EUpVSrwDr\n0G7pmCEih5VSE9Dub1turuuqlDoCmIC3RSTXB1pzHEFEpGZOdQYGBvcvCm0F1BGIyGpgdaZtY61e\nC/CGudiFXdfUlFJDlFKjza89lVKB9nZgYGBQ9HDQ6ac+vuXVQCn1DdAReNK86Ragz23vBgYGBR87\nFwnu1bOf9lzAaiUiAUqpvQAictV8Uc/AwOA+pbALr6QopYphTvColKoMpOnqlYGBQYFFQfqNtQUS\newa1aUAIUFUp9QEwGPhAV68MDAwKNAV6ZTavBiLyKzAG+AK4CgwSkfl6OwaOVc9Zv24t/k0a4New\nLl98nr2tYY8Pwa9hXdq3echiKyYmhh5dO1GtUjneGPmKzT6b/lxPpweb0j6oMd9O+TyLzR3bttCz\nY0tqVy/L6uVLbOqGDe6Dn28Nnh06IMt+eilqdWhUjb/Gd2bLB10Y3rVeFrvjB/qxfnRH1o/uyN/j\nu3BkUk9LnXvF0sx9tRWbxnZm49iH8azkkmV/cNwx+/P3dTzUvDFBzXKOwfNPPUZQswZ069jKRknq\n8KED9OjUhjZBzWj3oD+JidnfelPQv196+nsn2JvK+15N5ux9oqA4kAIk52MfQHsSXym1Vym1Mj/7\nmUwmXhsxnGUr1rD3wBEWzZ/H0SNHbNrMmvEzFV0rcvjYKV4d+TrvjX4nR1tvjHyF0OWr2b3/MIsW\nzOfoUVtbv8z8GVdXVw4ePckrI17j/fdGAeDs7Mz74ybw8SefZ7E59p3XmLVgGb9v3cvyJYs4efyo\nTRt3z5p88c2P9A1+NItP/3nldb789udsfX3njREsWLKCrWEHWLJoPscz+frbLzNwdXVl14FjvDR8\nJB+8P9pS5+1Tm03/7GbTP7uZ9PW3lu3FFEwc0ownvtlGxwl/0C/Ik7o1ytnYHb/4IF0/3kjXjzcy\nY1M4a/ZFW+qmPB3Id7+fpMOEP+j56SauxGUVn3HUMTOZTIx6cwTzl6xg664DhC6ez/FjmWLwqzkG\n+7UYTBirxSA1NZX/Pv8Un0+ZxpZd+1m6+k9KlCihm6/pthz9/dLTX0eQj2c/7zr2rH6+B8wD3NEe\nY5irlHo3H32MBI7m2SoTjlTP0dSkMmwNHPxoNraWW2z1HzCQTRs1W2XKlKFV6zaUypSPP2zXTrx8\nalPL24eSJUvSu/8g1q+xHbdr1vKiYWM/VLGsYW7driNlypbLsn1P2E5dFLWae1ci4nI8Z6/cIsUk\nLAs7T7dmbjm279fCk6W7tGeN69Yoh1OxYvx97DIAt5JMJKZkzXfmqGO2a+dOvK1i0C/4UdaszBqD\nRx/TYtC7X0YMNv75O42a+NHErxkAlSpXpnjxrDnUCvr3S09/HYGys9wL7Jl1DQOCRGSMiLyHli7k\naXuMK6U8gZ7A9Pw65kj1HK1dhkqVh4dnFiUl6/7SlYlyU+KJiorE3T3Dppu7Bxej71zgJDoqShdF\nrRquzkRdS8iwcS2BGq7Z/5A8KpWmZpUybD2uDWK+1ctyIyGFn158kHWjOzJmQJNs70Fy1DGLiorE\nwyNTDDLF9kIOMQg/dQKlFIP6PUKnNkFM/fKLbD9jQf9+6emvIyjst3REZ2rnZN5mD18B/wdknZKY\nsVaTqlnL4U9e3Vc4SlGrbwtPVu2JJP3pH6diigfqVKbbxxuIvJrAd88HMbilF/O3ndHhU9wZplQT\nO/7ZxvpN/1DaxYXgXl1p1jyAdh063WvXigza6ue99iJncnug/Uul1GS0xYHDSqnpSqmfgIPAlbwM\nK6V6AZdEZHdu7azVpKpWqWrZ7kj1HK1dhkpVZOT5LEpK1v2lKxPlpsTj7u5BVFSGzeioSKq7ZU4F\nlX/c3N11UdS6cD0R94qlM2xULM2F69lfQO/bwpNlYVaf7XoCh8/FcvbKLUxpwrp90fjVcs2yn6OO\nmbu7B5GRmWKQKbY1coiBu4cHD7VqQ+UqVXBxcaFztx4c2LdXN18z2jn2+6Wnv3eM0pJE2lPuBbmd\nfh5Ce6h9FVp+8H+A7cAEYI0dtlsDfZRSEcB8oJNSao69jjlSPUdTk8qwtXjhgmxs9bbYCl2ymPYd\nclfiCWwRRMTpU5w7E0FycjIrQhfRpXvPHNvbS/PAIF0UtfaduYZPtbLUrOxCieKKvi08WX8g64S7\ndvWyVHApQdjpq5Zt+yKuUcGlBJXKavdct65flRPRN7Ls66hj1iIoiH+tYrA0ZAHde2aNwYK5WgxW\nLA2hjTkGHR/uytEjh7h16xapqals2/IX9Ro01M1X0Of7pae/jqAgn37eFXUXtNQhK/NqFxAQqIsq\nT3xSmoQsXSl1zGpS4z74UOKT0mTU6DGycPFSiU9Kk5jYW9J/wECLmtSho6ckPilN4pPSpJaXl1Ss\nWFHKlCkj7h4eErbvkMQnpcnMeaHi41tHann7yFujx0vElQQZ8ea78tPsRRJxJUGWrf9bari5S2kX\nF3GtWEnq1m8oEVcSJOJKggQ91EoqVa4ipZydpYabu/yycLlEXEmQKzdTdFHUcn9piTwxdauEX4iT\nfy/FySdLD4n7S0tk8sqj8tS328T9pSXi/tIS+WLFEZm69rjlfXp59Ku/5fC563Lk/HVZsC1CvIaH\nivtLS3RTUpq7eLn41tZi8O7YCXI5zhyD+UvkclyKnLscJ737BYu3OQa7DhyXy3EpcjkuRb79aZbU\nb9BIGjRsLK+MfNOy/XJcSqH6fhVUNanKPo3kqbn77SoURDUppVRtYCLQCLBcWRaRrDc65WyjA/CW\niPTKrV1gYAvZuiPMXrN2o1dqmMvZ3NZwp5R11icZcdO3VuTd6DbQK/XQzcRUXezqEd/ClHrIEWpS\nVXwbS++P7btVddbQprmqSemBPaufs4CZaNcHewALgQX56URENuU1oBkYGBQeCvstHS4isg5ARMJF\nZAza4GZgYHAfohQUL6bsKvcCe+biSeYH2sOVUi+h5RDP8RYNAwODos89WwSwA3sGtdeBMsAItGtr\nFYBn9XTKwMCgYFOAx7S8BzUR2WF+GUdGokgDA4P7FMW9e67THnIc1JRSoZhzqGWHiGRNL2FgYFD0\nuYcZOOwht5naN3fNC53R687m6hVyfwi5IKHXrRcVu/1PF7vX1uUnZ0I+7MYnO9xmhdJZs4A4glST\n43OxOurmk0J5TU1E/rybjhgYGBQOFJqcZUFFnzs9DQwMijQF+YF2Y1AzMDDIN0ViUFNKlRIRxz8X\nZGBgUKjQUnUX3FHNnsy3DyilDgInze+bKaWm6u6ZgYFBgaVQixkDXwO9gBgAEdmPJm5sYGBwn1LY\nhVeKiUjmFKdZE9TrgF7qOYXJbmHyFaBLkC/7Z73IoV9f4q0hD2Wpr1WtPKs/H8rOn55j3aTH8Khi\n3xN3jvJ34x/raNOiCa2aN2Tql1nFTpKSkvjPM4/TqnlDej7chnNmlaqUlBRGvvQcnVoF0O6Bpkyd\nbKtwpZea1O/r19LcryHNGtVj0uefZmv3qSeG0KxRPTq2bWmxu+GP32nbMogHA5vRtmUQmzduyDYe\nt4MCnJSyq9wL7BnUzimlHgDErAz1GnDCHuNKqQil1EGl1D6lVL5yCumlnlOY7BYmX0G7H/CrEV3p\n++5Cmj/7I4M6NaKBl23m1f+91Inffj/EAy/8zMeztzLh+Q53zV+TycTot0by2+LlbNqxn2WLF3Di\nmK0m0LzZM3F1dWXb3qO88N8RfDT+PUBLRJmUnMSGbXtYu2k7s2dOtwx4eqlJmUwm3hz5KkuWrWLX\nvkMsXjifY5ns/jprBq6uFdl/5ATDXx3J2DGa3cpVqrAwZBk7du/nh+kzeeG5p/KMc34o7DO1l4E3\ngFrAReAh8zZ76Sgi/vnNqaSXek5hsluYfAUIauBOeOQ1IqKvk5KaxqKNR+nVyjbtXgOvKmzeGwHA\n5n1n6NWqbq42HelvukqVl7dmp2/wYNatts0zt271CgYN1Z4G7NV3AFs2b9QSDyrFrfh4UlNTSUxM\noGTJEpQ1az/opSal2a1tsRs86FFWrlhu02bVimU89sQwAPoNGMimjZqqVjP/5ri5uwPQsFFjEhMS\nSEpyzDqfslMer8BK5InIJREZIiJVzGWIiOSpUXCn6KWeU5jsFiZfAdyrlOX85Yw035GX47KcXh4M\nv0TftvUB6NumHuXLlKJS+dLkhiNVqtw9Muy4uWejUhUdhbtZyUpTfSrP1asx9Oo7AJcyZfCv70VQ\nkzq89OrrVKxYyapvx6tJRUdFWhSzNLtZlcWioqJs7FbIxu6y0BCa+QdQqlSpXPvLDwV5ppbnLR1m\nsZUsf6JF5EU77AuwXiklwA8i8mM29g01qfuId3/YwJevduWJrn5sPXiOyMs3MOnwOJCj2bt7F8WL\nF2fvsQhir1+jX49OtO3QCS+zBkRB5eiRw4x9712WrlzrULsF+T41e04//wD+NJetQDXA3nlsGxEJ\nQEsqOVwp1S5zg7uhJpVln0JitzD5ChB15SaeVTPk+DyqliPySpxNm+iYmwwZv4SWL81k3M+bAYiN\nz/3r5EiVqqjIDDvRUdmoVLm5E2VWstJUn25QqVJlQhfPp+PDXSlRogRVqlYj6MFW7N+7x6pvx6tJ\nubl7WBSzNLtZlcXc3d1t7MZa2Y08f56hg4P54edZ+NaunWtf+UFRsJNE2nP6ucCq/AIMAALtMS4i\nkeb/LwGhaELIdqGXek5hsluYfAUIOxZFHY+KeNWoQAmnYgzq2JBV207atKlcvrTltOTtx1ryy9oD\nudp0pL/pKlVnzSpVy0IW0rWHbZb5rj16sWieplK1ctkS2rTrgFIKD89abPlrEwC34uPZE7aDOnW1\n02i91KQ0u6csdkMWLaBnr942bR7p1Ye5c34FYOmSxbTvoKlqXb9+nYH9e/PBRx/TslXrXPvJN3be\no3bPZnO3oQxVGwi3o10ZoJzV621A93uhJqWX4tHdsFvQfXXu9LFN6TtqgZw4FyPhkVdl7PRN4tzp\nY5n4698S/N5Cce70sQwdHyInz8XIiXMxMmPVXinf7dMsNpw7faybv7MXLhXf2nXEy9tH3hnzgURd\nT5LX3h4tM+culqjrSXL6Qqz06jtAvH18xT+ghfyz76hEXU+Sk+djpFffAVKvQUOpW7+BjJnwsURd\nT5Ko60m6qUnFJZpk8dIVUttsd+z4DyUu0STvvDtG5i8OlbhEk1y+Hi/9BgRb7B44clLiEk3y/rgJ\n4uLiIn5Nm1nK6bPR0twBalLu9ZrIxD9O2VUooGpS18i4plYMTdx4lIgszGM/X7TZGWjX7uaKyMTc\n9tFLTcpAP4zUQ/qlHkrL47d5O7Rr9QB77lBNyrO+n7zy3VK72r77cJ27riaV60KB0ubHzdB0CQDS\nJK9R0IyInDbva2BgUMRw1KmlUqo7MAUoDkwXkax3LmvtgoHFQJCI5DrzyfWamnkAWy0iJnPRR+DQ\nwMCgUOEIhXalVHFgGtpCYiNgqFKqUTbtygEjgR2Z67LDntXPfUqp5vYYMzAwKPpoEnn2lTx4ADgl\nIqdFJBmYD/TNpt2HwKdAoj3+5ditUir91LQ5sEspdVwptUcptVcptcce4wYGBkWTfDxRUEUpFWZV\nrO9v9QDOWb0/b95mQSkVANQUkVX2+pbbNbWdQADQJ5c2BgYG9xmKfF1Tu3K7CwVmveHJwNP52S+3\nQU2Bpsp+Ow4ZGBgUXRz0CFQkUNPqvScZi5KgiaY3ATaZr8/VAJYrpfrktliQ26BWVSn1Rk6VIjLZ\nHq/zw61kE3sjrjvaLM1qVXC4TYDIawkOt1m1nOOez7NGD19Bv1svZu2K0MXu00HeDreph+oTgJMd\nF6Xyi2PGIkUxx1jaBdRVSvmgDWZDgMfSK0UkFqhi6VWpTcBbea1+5jaoFQfK4qg4GBgYFAkUjpmp\niUiqUuoVYB3aeDNDRA4rpSag3bS7PHcL2ZPboBYtIhNux6iBgUERRoGTg25UE5HVwOpM28bm0LaD\nPTbzvKZmYGBgYI2jZmp6kdug9vBd88LAwKBQca8SQNpDbgrtV++mIwYGBoWHAjymGWLGBgYG+UNh\n36NI94oC59v2v/5gSNcgBj0cwK8/fJmlft6MaTzW/SGe7NWaV4f1JTryrE19fNwN+rZpzKQP3rbZ\nrofaz18b1tOttT+dH/Ljh6lfZLG5658t9OvSioYe5Vm7IjRL/c24G7RtXpcP3rW9c+aP9WsJ8m9E\ngF99vvwiewWhZ4cNJcCvPp3bt+SsWQAknXPnzuJZrQJTv5pks33Lxt/p1a45PVo3Y/o3tnUAYdu3\nMKh7G5p5ubJ+ZUYWhp1b/yK4aytLCahdhT/XrsiyPzhWperw9s2MG9KJsYM6sO7X77LU/xX6Gx8+\n0Z2JTz3CFy8NIvpfLXebKTWFWR++yYdPdOeDoZ1Z++u3uvt6N1Sf9FIAyzcqX08U3HV0HdSUUq5K\nqcVKqWNKqaNKqZa5tRcRvhj/NpOmL2Lumu38sTKEf08es2lTr1FTZoRuYPbKrXTs1odvPxtvU//j\nVx/jH2TbjR5qPyaTiQ/efYOf5oay+q/drAxdxKnjtspEbh41+WTKD/TqPzjbz/vVpxMIesg2gZ/J\nZOLtN0awKHQl23cfJGTRgiwKQrN/mUEF14rsOXicl195jfHv294rNmbUW3Tu2j2L3Y/GvMl3s5ew\nfOMuVi9bTPgJ29i6edTko8nf80g/W38faN2OkPXbCFm/jRkLVuLs7EKr9lkvuTpSpSrNZGL+F2N5\nZdIsxs5dz64/llsGrXSCuvbh/Tlree+X1XR5/EUWf/0RALs3rCY1OZn356zl3Zkr+HvpXGKiz9vs\n60hf74bqk14KYLeD9kTBfTqooaUUWSsiDdDSEB3NrXFiwi08vXzxqOVNiZIl6dxzAH//abPaS+BD\nbXEu7QJAY/8gLl3IuAH52KF9XI25xANtOtnso4faT9iunXj5+FLLy4eSJUvSs99A/li30qaNZy0v\nGjTyo1ixrGE+tH8vVy5fpk2mwWF32E58fWvj7aP5OmDgYFavtL1dZ83K5Qx9XFM86ts/mM2bNliU\nk1atWEYtL28aNLRNdnBwXxi1vH2p6eVDiZIl6dE3mA3rbf31qOlF/UZNKJbLcv36VUtp27ELpc3H\nwBpHqlRFHNlPVU8vqnrUwqlESVp07s3+v3+3aVO6TIaoS3JCguVCj0KRnHgLU2oqyUmJOJUogXOZ\nsrr5ejdUn/RSALtdlJ3lXqDboKaUqgC0A34GEJFkEcn1cYGUlBSqW+WMr1rDncsXo3Nsv3LxbB5q\n1wWAtLQ0pv5vDK++82GWdnqo/URFRVLDPcNmDTcPLkbn7Ks1aWlpfDL+XUaN+zhLXXRUlI2CkLuH\nJ9HRUZn6zmiT7uvVmBhu3rzJlMmf8c7orLf5XIqOpoZVbKvX8OCSnf5as2Z5CD36Dcy2zpEqVdcv\nX6BidTfL+4pVa3D98oUs7TaF/Mr7A9sT+u0nPPr6OAACOvWgpLMLo/o8yHv9W9N56AuUKe+qm693\nQ/VJLwWw26Ugq0npOVPzAS4DM82ZPaYrpcpkbqSUejH9Cf74mzeyWsmBtcsWcOzgPh5//lUAlvw2\nnZbtu1Atk5BGQeS3mT/S/uGu1HB3rK+fTvyAl195jbJly+bd+Da4fPECJ48dpnX7zrrYvx06BA/j\nw8Wb6fffd1g96xtAm+UVK16cT5Zv58PFf/HH/OlcznTttaCRrvo05Zus1w4LHvblUssrn5pe6Ln6\n6YSW5eNVEdmhlJoCjALet25kls37EcC7dj25aKXDePlCFFWt/lqns2vrJn75djLT5q6kpPmv2qG9\nu9gf9g9L5v5Mwq14UpJTKO1Shv++PT5faj8enp52qf24u3twISrD5oXoSKq7ZfU1O/bt3kHYjm3M\nnfUT8bfiSUlOxqVMGd4e8yFu7u42CkJRkedxc3PP1LfWxsMjw9dKlSsTFraTZUuXMG7MKGJjr1Os\nWDFKOTvz4kvDqebmxgWr2F68EEk1O/1NZ+2KJTzcvTclSmSfvjo/qk+e6XHOQaXKtWoNrlnN0q9d\nvoBr1Ro5+taic2/mfa59tXauX0bjB9tR3KkE5StVobZfC84eO0BVjwwJRkf6mh/Vp/TvV35Vnxzp\n751yP69+ngfOi0h6tsrFaINcjjiXduF8RDhR586QkpzMH6uW0ObhHjZtjh8+wKfvv85nP8ylUuUM\nSb3xk38i9K9DLNl0gFfe+ZAe/R/lv2+PB/RR+wlsEUTE6XDOnYkgOTmZVUsX83DXnnYFZtK3M9m8\n+zgbw44yauxE+g16jBy9/OgAAB8ESURBVLfHaKfNAYFBhIef4oxZ8WjJ4oX06GmrINS9Z2/m/aYp\nHi0LDaFde01BaM3vmzlwNJwDR8N5efgI3nhrFC++NByAJs0COftvOOfPRpCSnMyaZSF07GKfv+ms\nWbaIR/oOyrHekSpVXg2bcul8BFeizpGakkzYHyto2sZ2hnjp3L+W14e2baBaTW8AKlX34PjufwBI\nSrjFv4f3Ut3LdrBwpK93Q/VJLwWw26UgLxToNlMTkQtKqXNKqfoichztCYUjue2jlOKNcZ/x+rPB\nmEwmeg18HN+6Dfnpq49p4OdP24cfYdpnY0m4Fc+YV58GoLq7J5/9MC9XX5ycnJj01VT69uqOyWRi\n2NPP0KhRYz78YCwBAS3o2bsPTz3zHM8/Mwy/hnWpWKkSv8zOsNmwng9xN26QnJzMihXLWL5qHQ0b\nNmLsx5N4bmhfTCYTA4cOo26DRkz59EOa+AfwcLeeHNi7m+HPDuHG9ets/H0NX38+kdV/5S4s4+Tk\nxGeTphDc9xFMJhOPD3uaho0a8/GH4/APaMEjPXvz5FPP8tLzTxHgV5+KFSvy8y9z8zweTk5OjP7w\nC/7zeD9MaWn0f/RJ6tRvyDeff0TjZs3p2LUnB/ft5rXnH+NG7HU2/b6GaZMnsmzDLgAiz53hQlQk\nLVq2ybWPL6d8Q++e3TCZTDz19LM0atyYCePHEhDYgl69+/D0s8/x7NNP0rhBHSpWrMTs3+Zna6u4\nkxND3viAqa8PI82URqteg3D3rceKnyZTq4Efzdp2YdPiXzkWtpXiTk64lKvAU2O022raBz/J7Ilv\nM+HxrogILXsOxLNOQ918dXJy4ouvvqZf7x6kmUw8+dQzNGzUmI8+GEfzwEB69urDsKef5YVnh9Gs\nUT0qVqrEzF+1Y/bjd9M4HX6KTz/+iE8/1lZvl61cS9Vq1XTz945R3LNTS3vIU03qjowr5Q9MB0oC\np4FnRORaTu0b+jWXGaEbHe6HkXpIv9RDtavrc/3OSD2kT+qh1g+2YPcdqknVadxMPptrn+J7sL97\nwVKTulNEZB9wVz+QgYGB/hTkmZrxmJSBgUG+KbhDmjGoGRgY5BMFFDdmagYGBkWJAjymGYOagYFB\nflGoAnwCagxqBgYG+caYqdmJS8niNPd2zbthAaFm5awPdRdU9Lr1IjHZpItdPW69AKj4SNYUUXfK\ntdVvOdwmQOytFIfbTE2781u4tCcKCu6oVqAGNQMDg0LAPXxY3R6MQc3AwCDfFEqNAgMDA4Ps0JJE\n3msvcsYY1AwMDPKNsfppYGBQpCjAZ58FOi2SbkIThcluYfBVL6EYPfzt0sKb/T8/y6GZz/HWow9k\nqa9ZtRxrPxvMP98+yc7vn6JbkA8AJZyK8cOb3dn1w1Ps+G4YbZvWzLKvo30F2PDHOloHNuYh/4ZM\nnfxZtrZefPoxHvJvSI9OrS2xDVk4l4fbtLAUN9dSHDqwL8d+8ouy8989QUR0KUB9YJ9VuQG8lts+\nAQGBkpAikpAicjMxVXx8feXI8XCJjU8SP7+msmf/YUt9QorIV19Pk+df+I8kpIj8MmeeBA8abFOf\nXSlMdguDr1duJIm3j6/sPXRCLl67JY2bNJV/wg7ItfhUS/n8y6ny9HMvyrX4VJk+6zfpHzzIpr5P\nvwHSt3+wTJj4qWWbXv6GR16TBk/+KOV6TJL9py6K/3MzxLnL55YyfdU+eXXKenHu8rn4PzdDIqKv\ni3OXz2Xk17/LL2sPinOXz6XmoGmy+0S0lO6q7aOXr5FXE8TL21d27DsmZy/flEZN/GTzjn1yITbZ\nUv73xdcy7JkX5EJssnz/82zp03+gTf2F2GTZuG23eHn7yoXYZGnqHyB3+tuu37iZbD4eY1cBwvQa\nY3Iqus3UROS4iPiLiD8QCNwCsurE5YBeQhOFyW5h8FUvoRg9/FVAeNQ1Ii7EkpKaxqLNx+jVyjZ5\npAiUd9HSP1UoU5LomJsANPCqzKZ9Wkrwy9dvEXszicB6WTPxOjK2e3fvwse3Nl7m2PYbMJh1q2yl\nCdetXsHgx7TY9uoXzJbNG7PYCv3/9s48PIoy28PvkS0ksi9KEpYEUECWbIiALG6MC7IYVJxBnRkU\nZxwdZxxhEETUGe94LzN6HXeuqCCo7IqiAuOCigqESGQTSJAli5KwhBC2LOf+UZUmnYSkm1RBd/O9\nz1MP3V1f/eqk0pxU1Vfn/BbMZUTyqZt7+o2PDSJD1U2qjKuADFXd5esGbhlNBJNuMMTqllGMG/GK\nQGZuged9Vu5holo08hrz5JtfM/qqrqTPuYfFf0/mwRctH84NO3IZ2rcjdc4T2l/YhPjOFxDdyntb\nJ2MFy9AlMuqkuU+bqKhKxzYn5+SYunXr0qhxE/bvr2DosmgBI0bdWkm/NgSym9SZmigYDVTZnlZE\nxgHjANq2a1fVEEOI4rZRzOlwyxVdmL18E88uTKFP1zbMmHA9ieNeZ+bHG+jSrjmrXrid3T8f4tvN\n2ZQ48HS+26SmrKFheEO6duvumGaZ72eg4vqZmojUB4YB86tar6rTVTVJVZNatTzpOeCP0QTgs9FE\nMOkGQ6z+GMV4tMoZxUx9ZCI9u3bkpRf+zdP/fIrpL7/gWryqeJ1dRbU6n6x9BV5j7vxFDxZ+sRWA\n1VtyCKtfh5ZNwikpVSa8/DmX/X4Wtzz2Lk0jGrA9s3ITZ2ePbRTZWSfNfXKysiod2zZtTo4pLi6m\n4FA+zZuf1Hp34TxGJjt7lgaBfaZ2Ji4/rwNSVfVnfzZyy2gimHSDIVa3jGLciFeBTlHNaH9hE+rV\nPY+bB3Vh6TcZXmP25BYwOM66Yri4bXPC6tcl9+ARGjaoS3iY5aJ1ZUJ7iktL+WF35UtGJ49tXEIS\nO8od23cXzWPI9UO9xgy5fijz3rKO7QfvLqT/wMEerdLSUpYsXsCI5FsqadeaQM5qbs9EAO9geRPU\nOLb87OfRItXFS5Zqp86dNSY2Vh974u96tEj14clTdP6i9/RokeqBgqM6MnmUxnbsqIlJvXXz1owa\nZ/6CTTfQYz1QWKxzFy7Rjp06a4eYWJ089Qk9UFis4ydO1jnzFuuBwmLN2XdYh49M1pjYjpqQmKTf\nbdzmNft5oLBY/zppyilnP52Md/ikBbptzz7NyDqgj772hYZdM02ffPNrTZ6yyDPj+fXGTE1L/1nX\np/+sN0ycp2HXTNOLxryiW3fv0y278vSTdTv1ol+94pkxdSvWn/JP6Oz572lsx07avkOsTnzkcf0p\n/4T+ecIknfn2Qv0p/4Tu/PmQDh1+k3aI6ahxCUm6ev0PnlnPhR+s0ISkS71mQp2Y/ezSPU5XZxz0\naeEszH66bbwSAewGYlU1v6bxiYlJump19W5LhsDCrS4dYfXruKJ7rnfpGDLoMtK+W1erc6iuPeJ1\n1nuf+zT20o5NQ854pRBw3k3VYDCcXQJ3nsCUSRkMBv+wbpcFblYL6DIpg8EQgNj91HxZapQSuVZE\ntopIuohMrGL9gyKyWUS+F5FPRKR9TZomqRkMBr9xYvJTROoAL2A9IdENuE1EKpaWfAckqWpPYAFQ\nuQC2AiapGQwGPxFEfFtq4FIgXVV3qOoJrCclhpcfoKqfqeoR++23QDQ1YJKawWDwGz8uP1uKSEq5\nZVw5mShgT7n3mfZnp2Is8FFNsQXURIECpS6UnpznUptON2INNurXdefvYnFJqSu6bjx+0WzQZMc1\nAQ6sfNJxzboO/F/w87naPCce6RCRMUASMKimsQGV1AwGQ5DgzHlCFlC+MV20/Zn3rkSuBiYDg1T1\neE2i5vLTYDD4jUNNItcCnUUkxq4RHw149a0SkXjgFWCYqu71JTZzpmYwGPzGiSYdqlosIvcBy4A6\nwGuquklEnsAqr1oCTAPOB+bbEw+7VXXYKUUxSc1gMPiLg76fqvoh8GGFzx4t9/pqfzVNUjMYDH4T\nyBUFJqkZDAa/EIyblF8sX/Yxcd270KNrZ/45rWonnjt+NZoeXTsz6PLLPE48+/bt47ohV9K6eSMe\nfOC+GvfhhNuPW7EaXVix/GPie3SlV7eL+Ne0qh2q7hwzml7dLuKKAX09mp/+ZwUD+vamT2IvBvTt\nzcrPPq20bfm4nXJ9uqZPZ9Le/hMb5z7IQ2MGVlrf7oKmfPjsb1kz836WPTeWqFaNPZ9//dof+PaN\n+1g3+4/cNaKyw5Ub8daWQG6n5nYvtT8Dm4CNWO28w6obHxefoDExsbpxS7oeKDim3Xv01JT1G7Xw\neKlneebZ53XsXeO08HipvvHmW5o86hYtPF6qe/cX6IpPv9Bnn3tR7/ndvV7buOH2c/hYsSuxHjpS\ndM7rlml+v3m77jt0VLv36Klrv9ugBcdKPMvTzz6vv71rnBYcK9HXZ83Rm0bdrAXHSvSrb1N02449\nWnCsRFevS9M2kZFe27nh+hR++WTNyMzTLqOmaaOBUzRtW7bG/fIZDes3ybMs/OR7Hfu3+RrWb5L+\n4r5Xdc5HqRrWb5I2GjhFGw+aomH9JmmLqx7Tndn7NWbYPzSs3yRXvrcJCYm17qd2Sc943Zx92KeF\nUHKTEpEo4I9YdVvdsWY3Rle3zZHCQmLLOfGMuuXWKpx4lniceEbeNIrPP7OceCIiIujX/3IahIVV\nG5dTbj9r16xxJdaUtUbX0uzo0Uy++VY+eN/boWrp++/xyzF3ADDiplF8/pnlUNUrLp42kVbL667d\nLuHY0aMcP1750SYnXZ96d40mI3M/O7MPUFRcwvxPvmfogK5eY7rEtGbluh0ArEzd4VlfVFzCiSKr\nJ12DenVO2fvfLWex0+VcdpOqCzQUkbpAOJBd3eATRUVEtz1Z2hUVFU1OVU48FZyJanJPOtX2Zfs4\nHbcfa4zzsRpda3yU1+8oipzsiprZXppNqtB8b/FCesUl0KBBg6rjdsj1KbJVYzL3nuyBmrX3EFGt\nmniN2bD9J4YPsmq1hw/qRuOIMJo3bghAdOsmrJl5P9sXT+Bfc74gJ8/bN8HpeJ0gkC8/3fT9zAL+\nidX5NgfIV9XlFceJyLiyurCCgkNuhWM4x9iyeROPTn6YZ59/6WyHAsDDL3zEgPgYvnn9DwyIiyFr\nb77HjSpzbz6X3vkc3W99mjHXJdC6WcRZjtYHAjiruXn52Qyr4j4GiAQi7PotL7Scm1SL5i3I3HPS\nPScrK5M2VTnxVHAmqsk96VTbl+3jdNx+rDHOx2p0rfFZXr+jLNpEVtSM9NLML6eZlZnJbbck88qM\nN4jt6G1WXFVMZXGfrutTdu4holufPDOLat2YrFzv7vU5eQWMnvQWfX/zAlOnrwAg//CxSmM27fiZ\n/r06uBpvbSlrEulARYEruHn5eTXwo6rmqmoRsAjoV90G4RERZJRz4lkwb24VTjw3epx4Fi9awKDB\nNbsnlccpt5+k3r1diTUxyehamukezYXz53LDUG+HquuHDuOt2bMAeHfRAgYNthyqDh48yKiRN/L4\n3/+Lvv36n3IfTro+pfyQRafoFrRv04x6detw81U9WfrVD15jWjQJ92w7/vZBzFy6DoCoVo0Jq289\nWdW0URj9erZn2+48V+OtNQ42iXQFF2c++2DNfIZbh4GZwP3VbROfkKgL3/1AO3XqrDExsTr18b9p\n4fFSnTjpEZ234F0tPF6q+/KP6MibRmlsrOXEs3FLumfWrF379tqsWTONiIjQyKgozyycW24/bsRa\neLzU6B4v1QXvvq8dbc1HH/ubFhwr0b8+/Ii+s2CxFhwr0dyDhTripmSP5vebt2vBsRKdMvUJDQ8P\n1x49e3mWHbtzKs1+Ovk9COs3SYf/5Q3dtitXMzLz9NGXl2tYv0n65GufaPKEWRrWb5LeNmmObt+d\nq9t25eprS9Z6Zjyvf2CGfr89R9O2Zev323P03qcWe2ZM3fjeOjH72b1XvKb/fMSnhRB0k3ocuBUo\nxupgeZdWU2WfkJikX32z1vE4TOuh4KPUpe9l3TrOX5wEU+uh/n2SWLcupVb/IXrEJep7K1b5NLZj\n64Yh5yY1FZjq5j4MBsOZJ5ArCkyZlMFg8IuzWi3gAyapGQwG/wngrGaSmsFg8BvTpcNgMIQU5p6a\nwWAIHQRceqDAEQIqqQnuPX7hBsEUq1u49ViLG49eABQeL3Zc041HLwCaXT7Bcc3jWzNrHuQTgfvd\nD6ikZjAYAp9AbxJpkprBYPCbAM5pJqkZDAb/MWdqBoMhpHClUN4hTFIzGAx+E7gpzSQ1g8HgJ2e1\nrZAPBJybVHnccs8JJt1giDWYHMA+WbGMPvGX0LtnF5791/9UqTP2jl/Su2cXhgzux+5dls7uXTuJ\nbtmIwX0TGdw3kb/88V7XYwW45rKLSJs7no3zJ/DQ7YMrrW93YVM+fO5u1sz+M8tevKdSG/FG4Q1I\nXzKJZ/4y/JT7OB0CuUmkq32NgAewnKQ2AX+qaXxCQqIrbj9uuPKcCd1giNUtlyq3HMA6xMRqyoat\nmr2/UC/p3kNXrU3TvMNFnuV/nv633vnbuzXvcJFOf322Dr/pZs07XKSpm7Zrl66XeI0tW9w6tuF9\nJ2jGnjztMvIf2qj/RE3blqVxt07TsD7jPcvC/6Tp2Mff0bA+4/UX976scz5M8Vr//Dtf6jsfp+pL\n877SsD7jVSIuqHU/tV7xCbq3oMinhRBzk+oO3A1cCvQChopIJ1+3d8s9J5h0gyFWt1yq3Ih37Zo1\nxMR2pEOMpTNy1K18tPR9rzEfLX2f0b+6HYBhI5P58vNPazyebsQK0LtbWzIy89iZvd9yqVqRxtCB\nl3iN6RLTmpUp6QCsXJfhtT7+4ihaNz+f/6zZ5nP8vhLAFgWuXn52BVar6hFVLQZWAjf5urFb7jnB\npBsMsbrlUuVGvNnZWURGn4w1sgqXqpzsbI+TVZnOfltn964fuaJfEjf+4kq+WfWVq7ECRLZqUsGl\nKt9jglzGhu05DB/cHYDhg7vbLlVW6/CnHhjKw/9eWmWctcM3e7yzZZHn5kTBRuBJEWkBHAWuB1Iq\nDhKRccA4gLbt2rkYjsFw+lxwYRvWb9lB8xYtWP/dOu4YPYpVa9No1LhxzRu7yMPPLeWZh4Yz5oYk\nVq3/kay9BykpLeWe5L4s+/qHSgYwTnDOVhSo6hYR+W9gOVAIrAdKqhg3HZgOkJiY5DkH98c9Jzo6\n2mf3nGDSDYZY/XGTiirTctEBrLp4IyOjyM48GWt2FS5VbSIjycrcQ2TUSZ3mLVogIh7/0Lj4RDrE\nxJKevo34hKRK+3Dq2Gbn5ldwqWpCVq63jWRO3iFGT3wTgIiG9RlxRXfyDx+jT4/29O/VgXHJfYlo\n2ID69epw+OgJJtw7q9J+Qg1XZz9VdYaqJqrqQOAA4PPFvVvuOcGkGwyxuuVS5Ua8Sb17syMjnV07\nLZ3FC+Zy7fVDvcZce/1Q3pljJYklixcyYJDlUpWXm0tJifU3eeePO9iRkU6HDrGuxQqQsiWTTm1b\nnnSpuqYXS7/c7DXGy6XqziuY+b51MfSbqW9z0Yh/0GXkUzz83Ae89eE6prz4UTVH2T/OSTcp+8Zn\na/vfdsAPQFNfZz+ddPtxy03qTOgGeqxuuUm5Fe/bC5dobKfO2iEmVic9+oTmHS7Sv/x1sr45d5Hm\nHS7SzLwCHTYiWWNiO2p8YpKmbNiqeYeL9PXZc/XiLt20e4+e2qNXnM6et7jK2U8nYw3rM16H/2mG\nbtu1VzP25OmjL32kYX3G65OvrtDkh17XsD7j9baJs2yXqr362nurtfHlE71mP8P6jNe7nnjH0dnP\nuPhEPXikxKeFEHST+hJoARQBD6rqJ9WNT0xM0lWrK912MwQwbrUecqutkxuthyIauHMXx5XWQxtn\nUXr4p1od3PjEJF25ao1PY5s0rBNyblID3NQ3GAxnnnN2osBgMIQuxqPAYDCEFIF8phbQtZ8GgyEw\ncaqiQESuFZGtIpIuIhOrWN9AROba61eLSIeaNE1SMxgM/uNAVhOROsALwHVAN+A2EelWYdhY4ICq\ndgKeAf67ptBMUjMYDH4h4FSZ1KVAuqruUNUTwDtAxXYiw4GZ9usFwFVSw0OOAXVPLTV1XV7DerLL\nh6EtgTwXQjC6wRVrsOkGQqzta7uz1NR1yxrWk5Y+Dg8TkfLPaU23q4gAooA95dZlAn0qbO8Zo6rF\nIpKP9ZjYKX/egEpqqtrKl3EikuLGsy9GN7hiDTbdYIq1OlT12jO1r9PBXH4aDIazRRbQttz7aPuz\nKseISF2gCVBtixeT1AwGw9liLdBZRGJEpD4wGlhSYcwS4E779SjgU62hDCqgLj/9YHrNQ4xuAGka\nXfc03dR1Ffse2X3AMqAO8JqqbhKRJ7BqRpcAM4A3RSQd2I+V+KrF1dpPg8FgONOYy0+DwRBSmKRm\nMBhCiqBLajWVVZym5msisldENjqhZ2u2FZHPRGSziGwSkQcc0g0TkTUikmbrPu6Ebjn9OiLynYh8\n4KDmThHZICLrKzyzVBvNpiKyQER+EJEtItLXAc2L7RjLlkMi8ieH4v2z/fvaKCJvi0j1bjO+6z5g\na25yKtag50w3cKtl08k6QAYQC9QH0oBuDugOBBKAjQ7G2gZIsF83wur660SsApxvv64HrAYuczDu\nB4G3gA8c1NwJtHT4uzATuMt+XZ8aGpCe5nftJ6C9A1pRwI9AQ/v9PODXDuh2x/ICCcea9PsP0MnJ\n4xCMS7CdqflSVuE3qvoF1syKY6hqjqqm2q8LgC1YX+7a6qqqHrbf1rMXR2Z7RCQauAF41Qk9txCR\nJlh/iGYAqOoJVT3o8G6uAjJU1ZcKF1+oCzS0n7UKB7Id0KyVY1uoEmxJraqyilonCrexOwvEY51V\nOaFXR0TWA3uBFarqiC7wv8AEoNQhvTIUWC4i62z3sNoSA+QCr9uXyq+KSIQDuuUZDbzthJCqZgH/\nBHYDOUC+qi53QHojMEBEWohIOJZjW9satgl5gi2pBR0icj6wEMuh/lBN431BVUtUNQ7rCexLbePo\nWiEiQ4G9qrqu1gFW5nJVTcDqxvAHERlYS726WLcLXlLVeCy3MkfurwLYD4IOA+Y7pNcM64oiBogE\nIkRkTG11VXULVteK5cDHnMKx7Vwj2JKaL2UVAYOI1MNKaHNUdZHT+vYl12eAE7V4/YFhIrIT67L+\nShGZ7YBu2ZkKqroXWIx1G6E2ZAKZ5c5QF2AlOae4DkhV1Z8d0rsa+FFVc1W1CFgE9HNCWGvh2Baq\nBFtS86WsIiCw26PMALao6tMO6rYSkab264bANVhOXbVCVR9W1WhV7YB1XD9V1VqfTYhIhIg0KnsN\nDMG6bKpNrD8Be0TkYvujq4DN1WziL7fh0KWnzW7gMhEJt78XV2HdY601ItLa/rcd1v20t5zQDWaC\nqkxKT1FWUVtdEXkbGAy0FJFMYKqqzqilbH/gdmCDff8LYJKqflhL3TbATLvB3nnAPFV17PELF7gA\nWGy3wKoLvKWqHzugez8wx/7jtgP4jQOaZYn3GuAeJ/QAVHW1iCwAUoFi4DucK21aKCJljm1/cGHC\nJOgwZVIGgyGkCLbLT4PBYKgWk9QMBkNIYZKawWAIKUxSMxgMIYVJagaDIaQwSS2IEJESu3vERhGZ\nb5fGnK7W4LJOHCIyrLqOJ3ZHjHtPYx+PichDvn5eYcwbIjLKj311cLLLiiF4MUktuDiqqnGq2h04\nAfyu/Eqx8Pt3qqpLVPWpaoY0BfxOagbD2cAkteDlS6CTfYayVURmYT2p31ZEhojINyKSap/RnQ+e\nXnQ/iEgq5bo5iMivReR5+/UFIrLY7teWJiL9gKeAjvZZ4jR73HgRWSsi35fv6SYik0Vkm4h8BVxM\nDYjI3bZOmogsrHD2ebWIpNh6Q+3xdURkWrl9O/aQrCE0MEktCLHb11wHbLA/6gy8qKqXYBV3PwJc\nbReRpwAP2k0J/w+4EUgELjyF/L+BlaraC6uechNWsXiGfZY4XkSG2Pu8FIgDEkVkoIgkYpVYxWF1\njOjtw4+zSFV72/vbAowtt66DvY8bgJftn2EsVpeL3rb+3SIS48N+DOcIQVUmZaBhuZKrL7FqSyOB\nXar6rf35ZUA3YJVdmlQf+AboglVUvR3ALlavqg3QlcAdYHUDAfLtLhPlGWIv39nvz8dKco2Axap6\nxN6HL3W53UXk71iXuOdjlcCVMU9VS4HtIrLD/hmGAD3L3W9rYu/7nC/kNliYpBZcHLVbDnmwE1dh\n+Y+weqzdVmGc13a1RIB/qOorFfZxOu2k3wBGqGqaiPwaqwa3jIo1fGrv+35VLZ/8ynrWGQzm8jME\n+RboLyKdwNMl4yKsTh4dRKSjPe62U2z/CfB7e9s6YnWZLcA6CytjGfDbcvfqouxuEV8AI0Skod2Z\n40Yf4m0E5Nhtmn5VYd3NInKeHXMssNXe9+/t8YjIReJ8g0hDEGPO1EIMVc21z3jeFpEG9sePqOo2\nsbrOLhWRI1iXr42qkHgAmC4iY7EaDv5eVb8RkVX2IxMf2ffVugLf2GeKh4ExqpoqInOxvCP2YrWK\nqokpWB2Bc+1/y8e0G1gDNAZ+p6rHRORVrHttqXYbn1xghG9Hx3AuYLp0GAyGkMJcfhoMhpDCJDWD\nwRBSmKRmMBhCCpPUDAZDSGGSmsFgCClMUjMYDCGFSWoGgyGk+H8Ory2ACfNL/gAAAABJRU5ErkJg\ngg==\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "guThaIpBMl8J",
        "colab_type": "code",
        "outputId": "69654f5b-aaeb-4ae7-8cfa-ab663c5b7ed7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 254
        }
      },
      "source": [
        "# use ngrok to display TensorBoard on Colab\n",
        "!wget https://bin.equinox.io/c/4VmDzA7iaHb/ngrok-stable-linux-amd64.zip\n",
        "!unzip ngrok-stable-linux-amd64.zip\n"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2019-10-23 20:25:18--  https://bin.equinox.io/c/4VmDzA7iaHb/ngrok-stable-linux-amd64.zip\n",
            "Resolving bin.equinox.io (bin.equinox.io)... 3.226.36.176, 34.202.125.55, 52.4.192.223, ...\n",
            "Connecting to bin.equinox.io (bin.equinox.io)|3.226.36.176|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 13773305 (13M) [application/octet-stream]\n",
            "Saving to: ‘ngrok-stable-linux-amd64.zip.1’\n",
            "\n",
            "ngrok-stable-linux- 100%[===================>]  13.13M  13.4MB/s    in 1.0s    \n",
            "\n",
            "2019-10-23 20:25:19 (13.4 MB/s) - ‘ngrok-stable-linux-amd64.zip.1’ saved [13773305/13773305]\n",
            "\n",
            "Archive:  ngrok-stable-linux-amd64.zip\n",
            "replace ngrok? [y]es, [n]o, [A]ll, [N]one, [r]ename: n\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zXIf3ARdMsFm",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "LOG_DIR = './runs'\n",
        "get_ipython().system_raw(\n",
        "    'tensorboard --logdir {} --host 0.0.0.0 --port 6006 &'\n",
        "    .format(LOG_DIR)\n",
        ")"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YerFMXaZU2WT",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "get_ipython().system_raw('./ngrok http 6006 &')\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ui9bGasKU7Qj",
        "colab_type": "code",
        "outputId": "83898702-48b1-4b22-89a9-e39a89f48368",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "source": [
        "! curl -s http://localhost:4040/api/tunnels | python3 -c \\\n",
        "    \"import sys, json; print(json.load(sys.stdin)['tunnels'][0]['public_url'])\""
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "https://1eeb52d2.ngrok.io\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OIm37f1QVBIo",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}