PytorchBasics.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true
   },
   "source": [
    "# Pytorch Cheatsheet"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true,
    "hidden": true
   },
   "source": [
    "## Overview"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false,
    "hidden": true
   },
   "outputs": [
    {
     "ename": "SyntaxError",
     "evalue": "invalid syntax (<ipython-input-13-5efeee96745a>, line 5)",
     "output_type": "error",
     "traceback": [
      "\u001b[0;36m  File \u001b[0;32m\"<ipython-input-13-5efeee96745a>\"\u001b[0;36m, line \u001b[0;32m5\u001b[0m\n\u001b[0;31m    Normal Convolution\u001b[0m\n\u001b[0m                     ^\u001b[0m\n\u001b[0;31mSyntaxError\u001b[0m\u001b[0;31m:\u001b[0m invalid syntax\n"
     ]
    }
   ],
   "source": [
    "* How does a Conv layer impact the size of output feature maps?\n",
    "* How does a Conv Transpose layer impact the size of the output feature maps?\n",
    "* How does padding / stride impact both of the above?\n",
    "\n",
    "Normal Convolution\n",
    "* W2 = (W1 - KS + 2P)/S + 1\n",
    "\n",
    "Convolution Transpose\n",
    "* W2 = S(W1 - 1) + KS - 2P"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true,
    "hidden": true
   },
   "source": [
    "## Resources"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false,
    "hidden": true
   },
   "outputs": [
    {
     "ename": "SyntaxError",
     "evalue": "can't use starred expression here (<ipython-input-14-5a53398567fd>, line 1)",
     "output_type": "error",
     "traceback": [
      "\u001b[0;36m  File \u001b[0;32m\"<ipython-input-14-5a53398567fd>\"\u001b[0;36m, line \u001b[0;32m1\u001b[0m\n\u001b[0;31m    * ss\u001b[0m\n\u001b[0m        ^\u001b[0m\n\u001b[0;31mSyntaxError\u001b[0m\u001b[0;31m:\u001b[0m can't use starred expression here\n"
     ]
    }
   ],
   "source": [
    "* ss\n",
    "* ss\n",
    "* ss"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true,
    "hidden": true
   },
   "source": [
    "## Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "collapsed": false,
    "hidden": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/bfortuner/anaconda3/lib/python3.6/site-packages/matplotlib/__init__.py:1401: UserWarning:  This call to matplotlib.use() has no effect\n",
      "because the backend has already been chosen;\n",
      "matplotlib.use() must be called *before* pylab, matplotlib.pyplot,\n",
      "or matplotlib.backends is imported for the first time.\n",
      "\n",
      "  warnings.warn(_use_error_msg)\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.init as init\n",
    "import torch.optim as optim\n",
    "import torch.nn.functional as F\n",
    "from torch.autograd import Variable\n",
    "import torchvision.datasets as datasets\n",
    "import torchvision.transforms as transforms\n",
    "import torchvision.utils as torch_img_utils\n",
    "from torch.utils.data import DataLoader\n",
    "import torchvision.models as models\n",
    "import torch.backends.cudnn as cudnn\n",
    "import torchvision\n",
    "import torch.autograd as autograd\n",
    "from PIL import Image\n",
    "import imp\n",
    "import os\n",
    "import sys\n",
    "import math\n",
    "import time\n",
    "import random\n",
    "import shutil\n",
    "import cv2\n",
    "import scipy.misc\n",
    "from glob import glob\n",
    "import sklearn\n",
    "\n",
    "from tqdm import tqdm\n",
    "import numpy as np\n",
    "import matplotlib as mpl\n",
    "mpl.use('Agg')\n",
    "import matplotlib.pyplot as plt\n",
    "plt.style.use('bmh')\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "heading_collapsed": true
   },
   "source": [
    "# Basics"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "hidden": true
   },
   "source": [
    "## Tutorials"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "hidden": true
   },
   "source": [
    "* [Official Pytorch Tutorials](http://pytorch.org/tutorials/)\n",
    "* [Pytorch Documentation](http://pytorch.org/docs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Datasets"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true
   },
   "source": [
    "## File Management"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false,
    "hidden": true
   },
   "outputs": [],
   "source": [
    "random.seed(1)\n",
    "torch.manual_seed(1)\n",
    "DATA_PATH='/media/bfortuner/bigguy/data/'\n",
    "CUB_PATH=DATA_PATH+'cub/'\n",
    "CUB_IMAGES_PATH=CUB_PATH+'images/'\n",
    "MNIST_PATH=DATA_PATH+'mnist/'\n",
    "MNIST_WEIGHTS_PATH=MNIST_PATH+'weights/'\n",
    "MNIST_RESULTS_PATH='results/'\n",
    "CIFAR10_PATH=DATA_PATH+'cifar10/'\n",
    "CIFAR10_IMGS_PATH=CIFAR10_PATH+'images/'\n",
    "CIFAR10_RESULTS_PATH=CIFAR10_PATH+'results/'\n",
    "CIFAR10_WEIGHTS_PATH=CIFAR10_PATH+'weights/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true,
    "hidden": true
   },
   "outputs": [],
   "source": [
    "def get_paths_to_files(dir_path):\n",
    "    filepaths = []\n",
    "    fnames = []\n",
    "    for (dirpath, dirnames, filenames) in os.walk(dir_path):\n",
    "        filepaths.extend(os.path.join(dirpath, f) for f in filenames if not f[0] == '.')\n",
    "        fnames.extend([f for f in filenames if not f[0] == '.'])\n",
    "    return filepaths, fnames\n",
    "\n",
    "def get_random_image_path(dir_path):\n",
    "    filepaths = get_paths_to_files(dir_path)[0]\n",
    "    return filepaths[random.randrange(len(filepaths))]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true
   },
   "source": [
    "## MNIST"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false,
    "hidden": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MNIST Train Samples: 60000\n",
      "MNIST Test Samples: 10000\n"
     ]
    }
   ],
   "source": [
    "MNIST_BATCH_SIZE = 128\n",
    "MNIST_MEAN = np.array([0.1307,])\n",
    "MNIST_STD = np.array([0.3081,])\n",
    "mnist_train_loader = torch.utils.data.DataLoader(\n",
    "    datasets.MNIST(MNIST_PATH, train=True, download=True,\n",
    "                   transform=transforms.Compose([\n",
    "                       transforms.ToTensor(),\n",
    "                       transforms.Normalize(MNIST_MEAN, MNIST_STD)\n",
    "                   ])),\n",
    "    batch_size=MNIST_BATCH_SIZE, shuffle=True)\n",
    "mnist_test_loader = torch.utils.data.DataLoader(\n",
    "    datasets.MNIST(MNIST_PATH, train=False, transform=transforms.Compose([\n",
    "                       transforms.ToTensor(),\n",
    "                       transforms.Normalize(MNIST_MEAN, MNIST_STD)\n",
    "                   ])),\n",
    "    batch_size=MNIST_BATCH_SIZE*8, shuffle=True)\n",
    "\n",
    "mnist_train_labels = mnist_train_loader.dataset.train_labels\n",
    "MNIST_CLASSES = np.unique(mnist_train_labels.numpy())\n",
    "print(\"MNIST Train Samples:\", len(mnist_train_loader.dataset))\n",
    "print(\"MNIST Test Samples:\", len(mnist_test_loader.dataset))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## CIFAR10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Files already downloaded and verified\n",
      "Files already downloaded and verified\n",
      "CIFAR Train Samples: 50000\n",
      "CIFAR Test Samples: 10000\n"
     ]
    }
   ],
   "source": [
    "CIFAR_BATCH_SIZE = 64\n",
    "CIFAR_MEAN = np.array([0.49139968, 0.48215827, 0.44653124])\n",
    "CIFAR_STD = np.array([0.24703233, 0.24348505, 0.26158768])\n",
    "normTransform = transforms.Normalize(CIFAR_MEAN, CIFAR_STD)\n",
    "\n",
    "trainTransform = transforms.Compose([\n",
    "    transforms.RandomCrop(32, padding=4),\n",
    "    transforms.RandomHorizontalFlip(),\n",
    "    transforms.ToTensor(),\n",
    "    normTransform\n",
    "])\n",
    "testTransform = transforms.Compose([\n",
    "    transforms.ToTensor(),\n",
    "    normTransform\n",
    "])\n",
    "\n",
    "cifar_train_loader = torch.utils.data.DataLoader(\n",
    "    datasets.CIFAR10(CIFAR10_IMGS_PATH, train=True, download=True,\n",
    "                 transform=trainTransform),\n",
    "    batch_size=CIFAR_BATCH_SIZE, shuffle=True)\n",
    "cifar_test_loader = torch.utils.data.DataLoader(\n",
    "    datasets.CIFAR10(CIFAR10_IMGS_PATH, train=False, download=True,\n",
    "                 transform=testTransform),\n",
    "    batch_size=CIFAR_BATCH_SIZE*16, shuffle=False)\n",
    "\n",
    "cifar_classes = cifar_train_loader.dataset.train_labels\n",
    "CIFAR_CLASSES = np.unique(np.array(cifar_classes))\n",
    "CIFAR_CLASS_NAMES = np.array(['airplane','automobile','bird','cat',\n",
    "                          'deer','dog','frog', 'horse','ship','truck'])\n",
    "print(\"CIFAR Train Samples:\", len(cifar_train_loader.dataset))\n",
    "print(\"CIFAR Test Samples:\", len(cifar_test_loader.dataset))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Swiss Roll"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      " 11.2859  -4.4224\n",
      "  9.7413   9.7744\n",
      " 10.7755  -4.8743\n",
      "  8.2207  10.6784\n",
      " -0.8344 -11.3071\n",
      " -2.7668 -10.5984\n",
      " 11.7795  -3.4875\n",
      " -6.6013  -7.8201\n",
      " -4.7877  -9.2370\n",
      " -6.1006  -8.7008\n",
      " -8.1754  -6.1561\n",
      " -1.8350   7.9860\n",
      "  7.1470  -8.8236\n",
      "  3.5312   6.1914\n",
      " -8.8901  -3.9086\n",
      "  4.7486  -2.7880\n",
      " 10.5335  -6.0866\n",
      "  4.6021  -3.2873\n",
      " -4.6778   7.1029\n",
      " -7.9958   3.1143\n",
      " -9.6130   1.5783\n",
      "  2.0332   7.3880\n",
      " -6.6942  -7.8496\n",
      "  6.5717  -0.2127\n",
      " -0.8685 -11.0825\n",
      "  0.7146  -4.9460\n",
      " -5.1389   6.8356\n",
      "  4.0975  -3.3304\n",
      " -1.4073   7.8709\n",
      "  6.1093   2.4540\n",
      "  2.9354  -4.6179\n",
      " 12.6135   3.1774\n",
      " -9.0066  -3.7590\n",
      "  1.0331 -11.0076\n",
      "  2.1664 -10.9575\n",
      " -8.8992  -4.8085\n",
      "  1.0081  -4.5886\n",
      " -9.3248   0.9309\n",
      " 12.6142   0.2867\n",
      " 11.2504  -4.4437\n",
      "  1.6806 -10.9212\n",
      "  2.0674 -10.7526\n",
      " -0.3842 -10.9054\n",
      " 12.3477  -2.2549\n",
      " -7.3852  -6.8657\n",
      "  2.8458  -4.2350\n",
      " -9.5080  -1.0432\n",
      "  0.1223   7.5927\n",
      " 11.3272   7.3686\n",
      "  9.0065   9.6114\n",
      "  1.5104   7.4256\n",
      " -3.5433  -9.8809\n",
      "  1.4911  13.8762\n",
      "  6.5093   0.4793\n",
      "  1.4910  13.9619\n",
      "  1.9456  -4.5431\n",
      "  0.4149   7.7717\n",
      " -5.3088   6.4350\n",
      " -7.4506   5.2326\n",
      "  9.6120  -7.3555\n",
      " -3.8838   7.3569\n",
      " 11.4464   6.3178\n",
      "  9.7236  -7.0900\n",
      " -5.3181   6.5565\n",
      "  2.8181  -4.5720\n",
      "  3.3429   6.8511\n",
      "  5.1245 -10.2428\n",
      " -7.3599  -7.0429\n",
      "  7.5721  -9.0072\n",
      " -9.5100  -3.0249\n",
      "  0.6974   7.9001\n",
      " -7.5929   4.9205\n",
      " 12.6528  -0.8977\n",
      " -0.3310   8.0990\n",
      "  9.1408   9.6647\n",
      "  4.6480  -3.4762\n",
      "  3.4730  13.5688\n",
      "  0.5727  -4.7259\n",
      "  3.4732   6.5489\n",
      " -2.6653   7.5605\n",
      "  0.3234   7.6561\n",
      "  7.3296  -9.2732\n",
      " 12.1787   3.4578\n",
      " 12.4633   0.0099\n",
      "  7.4134  11.3983\n",
      "  3.9959  -3.8084\n",
      " -0.4861 -10.6528\n",
      "  8.3799  10.2960\n",
      " -8.6356   1.3203\n",
      " -8.6509   2.9220\n",
      "  4.6836  -3.4492\n",
      "  4.3992   5.7833\n",
      "  5.3000  12.8673\n",
      " -0.8171   8.0852\n",
      "  6.0090   3.6254\n",
      "  5.9488   3.5750\n",
      " 11.7524  -2.9543\n",
      " 12.2179   3.6586\n",
      "  0.0826   7.4973\n",
      "  4.9475   4.6845\n",
      "[torch.FloatTensor of size 100x2]\n",
      " \n",
      " 12.2212\n",
      " 13.3307\n",
      " 12.1420\n",
      " 13.4639\n",
      " 10.9256\n",
      " 10.7777\n",
      " 12.2977\n",
      " 10.2653\n",
      " 10.4963\n",
      " 10.3937\n",
      " 10.0933\n",
      "  8.1106\n",
      " 11.6424\n",
      "  7.3446\n",
      "  9.8275\n",
      "  5.7567\n",
      " 12.0565\n",
      "  5.6685\n",
      "  8.4224\n",
      "  9.0444\n",
      "  9.2889\n",
      "  7.5718\n",
      " 10.2733\n",
      "  6.2284\n",
      " 10.8970\n",
      "  4.8360\n",
      "  8.4888\n",
      "  5.5941\n",
      "  8.0269\n",
      "  6.6753\n",
      "  5.2711\n",
      " 12.8198\n",
      "  9.7724\n",
      " 11.0686\n",
      " 11.1698\n",
      "  9.9414\n",
      "  4.8505\n",
      "  9.3045\n",
      " 12.5774\n",
      " 12.2032\n",
      " 11.1461\n",
      " 11.1986\n",
      " 10.9699\n",
      " 12.3941\n",
      " 10.1878\n",
      "  5.2989\n",
      "  9.5466\n",
      "  7.8334\n",
      " 13.1535\n",
      " 13.3903\n",
      "  7.6266\n",
      " 10.6547\n",
      " 14.0359\n",
      "  6.3397\n",
      " 14.0446\n",
      "  5.1886\n",
      "  7.7479\n",
      "  8.5359\n",
      "  8.8331\n",
      " 11.9284\n",
      "  8.3655\n",
      " 13.0527\n",
      " 11.9363\n",
      "  8.5110\n",
      "  5.2547\n",
      "  7.4107\n",
      " 11.4514\n",
      " 10.1761\n",
      " 11.6960\n",
      "  9.7376\n",
      "  7.7544\n",
      "  8.8595\n",
      " 12.4969\n",
      "  7.8691\n",
      " 13.3897\n",
      "  5.6415\n",
      " 13.8964\n",
      "  4.8549\n",
      "  7.3483\n",
      "  8.1738\n",
      "  7.8044\n",
      " 11.7011\n",
      " 12.8480\n",
      " 12.5646\n",
      " 13.5710\n",
      "  5.5411\n",
      " 10.9290\n",
      " 13.4515\n",
      "  9.2675\n",
      "  9.1191\n",
      "  5.6661\n",
      "  7.2119\n",
      " 13.7411\n",
      "  7.9247\n",
      "  6.8484\n",
      "  6.8595\n",
      " 12.3164\n",
      " 12.8608\n",
      "  7.8407\n",
      "  6.9989\n",
      "[torch.DoubleTensor of size 100]\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXUAAAD1CAYAAACvOmWkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f13241b7400>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.datasets.samples_generator import make_swiss_roll\n",
    "\n",
    "def get_swiss_roll(n_samples=100):\n",
    "    noise = 0.2\n",
    "    X, _ = make_swiss_roll(n_samples, noise)\n",
    "    X = X.astype('float32')[:, [0, 2]]\n",
    "    return X, _\n",
    "\n",
    "def plot_roll(data):\n",
    "    # data.shape = (N, 2)\n",
    "    x = data[:,0]\n",
    "    y = data[:,1]\n",
    "    plt.scatter(x,y)\n",
    "\n",
    "SWISS_ROLL_BATCH_SIZE = 100\n",
    "X, _ = get_swiss_roll(100)\n",
    "swiss_roll_dataset = torch.utils.data.TensorDataset(torch.FloatTensor(X), torch.FloatTensor(_))\n",
    "swiss_roll_loader = torch.utils.data.DataLoader(swiss_roll_dataset, batch_size=SWISS_ROLL_BATCH_SIZE, shuffle=True)\n",
    "    \n",
    "# Test\n",
    "data = get_swiss_roll(1000)[0]\n",
    "plot_roll(data)\n",
    "\n",
    "inputs,targets = next(iter(swiss_roll_loader))\n",
    "print(inputs,targets)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true
   },
   "source": [
    "# Image Handling"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true,
    "hidden": true
   },
   "source": [
    "## Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true,
    "hidden": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true,
    "hidden": true
   },
   "source": [
    "## Viewing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": false,
    "hidden": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAakAAAFHCAYAAAAbYrOpAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nWf8Hb1Sf8pdYURAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1341142278>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def imshow(inp, mean_arr, std_arr, title=None):\n",
    "    \"\"\"Imshow for Tensor.\"\"\"\n",
    "    if inp.size(0) == 1:\n",
    "        inp = np.squeeze(inp.numpy())\n",
    "        kwargs = {'cmap':'gray'}\n",
    "    else:\n",
    "        inp = inp.numpy().transpose((1, 2, 0))\n",
    "        kwargs = {}\n",
    "    inp = std_arr * inp + mean_arr\n",
    "    plt.imshow(inp, **kwargs)\n",
    "    if title is not None:\n",
    "        plt.title(title)\n",
    "    plt.pause(0.001)  # pause a bit so that plots are updated\n",
    "\n",
    "def visualize_preds(model, data_loader, class_names, mean_arr, std_arr, num_images=6):\n",
    "    images_so_far = 0\n",
    "    fig = plt.figure()\n",
    "\n",
    "    for i, data in enumerate(data_loader):\n",
    "        inputs, labels = data\n",
    "        inputs, labels = Variable(inputs.cuda()), Variable(labels.cuda())\n",
    "\n",
    "        outputs = model(inputs)\n",
    "        _, preds = torch.max(outputs.data, 1)\n",
    "        preds = preds.cpu().numpy()\n",
    "        labels = labels.data.cpu().numpy()\n",
    "        for j in range(inputs.size()[0]):\n",
    "            images_so_far += 1\n",
    "            ax = plt.subplot(num_images//2, 2, images_so_far)\n",
    "            ax.axis('off')\n",
    "            ax.set_title('P: {}, A:{}'.format(class_names[preds[j][0]], \n",
    "                                              class_names[labels[j]]))\n",
    "            imshow(inputs.cpu().data[j], mean_arr, std_arr)\n",
    "\n",
    "            if images_so_far == num_images:\n",
    "                return\n",
    "        plt.tight_layout()\n",
    "\n",
    "def plot_mnist_samples(arr, dim=(4,4), figsize=(6,6)):\n",
    "    if type(arr) is not np.ndarray:\n",
    "        arr = arr.numpy()\n",
    "    bs = arr.shape[0]\n",
    "    arr = arr.reshape(bs, 28, 28)\n",
    "    plt.figure(figsize=figsize)\n",
    "    for i,img in enumerate(arr):\n",
    "        plt.subplot(*dim, i+1)\n",
    "        plt.imshow(img, cmap='gray')\n",
    "        plt.axis('off')\n",
    "    plt.tight_layout()\n",
    "\n",
    "def denorm_meanstd(t, mean, std):\n",
    "    return (t * std) + mean\n",
    "\n",
    "def plot_cifar_samples(arr, dim=(4,4), figsize=(6,6)):\n",
    "    if type(arr) is not np.ndarray:\n",
    "        arr = arr.numpy().transpose((0, 2, 3, 1))\n",
    "    arr = denorm_meanstd(arr, CIFAR_MEAN, CIFAR_STD)\n",
    "    plt.figure(figsize=figsize)\n",
    "    for i,img in enumerate(arr):\n",
    "        plt.subplot(*dim, i+1)\n",
    "        plt.imshow(img, cmap='gray')\n",
    "        plt.axis('off')\n",
    "    plt.tight_layout()\n",
    "    \n",
    "def get_cifar_batch_of_size(bs):\n",
    "    # Return batchsize smaller than 64 for testing\n",
    "    inputs,targets = next(iter(cifar_test_loader))\n",
    "    return inputs[:bs], targets[:bs]\n",
    "\n",
    "def get_mnist_batch(bs):\n",
    "    train_loader = torch.utils.data.DataLoader(\n",
    "        datasets.MNIST(MNIST_PATH, train=True, download=True,\n",
    "                       transform=transforms.Compose([\n",
    "                           transforms.ToTensor(),\n",
    "                           transforms.Normalize(MNIST_MEAN, MNIST_STD)\n",
    "                       ])),\n",
    "        batch_size=bs, shuffle=True)\n",
    "    inputs,targets = next(iter(train_loader))\n",
    "    return inputs,targets\n",
    "\n",
    "inps,targs = get_mnist_batch(12)\n",
    "plot_mnist_samples(inps)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": false,
    "hidden": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAakAAAFHCAYAAAAbYrOpAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzsvVmTZFlyHuZ3jT0iI/esfe/qdbp7pmcBZh+AGgIkjQRh\nBIwPotFMLzKTmUxGSS/6CXqXyWQiIFIbDRREUARACAIGmBnMPr1OV3VXd3XtVblHZmTsd9WD+7n+\n5WT2dEPqDAbMzvdSUSfves6597r7+fxzJ89zsrCwsLCwmEW4/6EvwMLCwsLC4sNgP1IWFhYWFjML\n+5GysLCwsJhZ2I+UhYWFhcXMwn6kLCwsLCxmFvYjZWFhYWExs7AfKQsLCwuLmYX9SFlYWFhYzCz8\nkzz47/4XL+VERE6eFW2hHxARkePq9zGKx0RElKSJbhfwdinsm2e57Ktt5jB5UivaHOLjBMGkaPPI\nk301eTnNUiIiilM9XpY5chCvaEtSbpvIv3wO2Z4yaHPkfvQ+0lTOC0nTLvF5Izk/EdFAdhlG2vbf\n/t5dPeEJYmdnJyciShK9boc+2VM7zl/zeMfkmGNT7pg26Nf8OJsrk/PDPJJ7c8BG+0VJ7cddO26/\nsrpy4uP0u39xl08Iz8ju9gYREU3G46Lt0uUrREQ012oWbYHLlxeG+riHHt97CM+h78gp4lHRVq8H\nh45BROT5/Nt39RnZ2+sQEVGj3tDzhryvT7odyXHSLNKmY4bNvB9Gg6Fen8/HKZcrRVsU8XGSSJ/1\nSqkix9DztluVEx+j//Gf/3c5EVF98ZpeixcSEVGzUS/aehN+xgcHu0WbK++0DOazL31Q8UtFW9mT\nMYT3WPGoQpN5t2FbJm2HzuH7cn7tq+Pmu+Mcffdm8P76+X1LJb3m0JXfmbY5JT7fcPedou0r3/zN\nY8fIelIWFhYWFjOLE/WkIrGg8lwtPcr4S1witSxc2c739ctcWFfwsXZ8bpzEcdGWZLIvWNGeWBu+\nq9aVk4kFmui1GI8my9SKiBy2wlJXv/pR5su/4EnJfTiZXktZrs939Fpc31inajmSw/vkcHPGuveO\nMytPGJ5nznmC0+Gv60kB3MICxEY+Xo4en5kDOYxT0Z+H/DBzEGj6BfJgx137lOXE6lWej26uYzQZ\ncFsWqbdRDvlaaxXdzvxySb2wks8WfiWEuSo9PEl1Xpa8MhERhYe2k+MG+tyE4uW4LkYW5FkvBUWb\nJ105GOnzYJxctL4zeSdMoO8DT7y6QI8XTyZyj3p9ZXOcqcQhFFnOfZV47aItDjjCk3rwvgt4HAaj\nftGWpwMiIoJbo0kukR7o07F5x2hXFZEo19PxGA3ZG/agzfRbFOk7y3XlXYSerfGyw7BoS5JUttPz\nOo55b+tca7f53ksVuF8Zmgy8P0fmRNrX7T4M1pOysLCwsJhZ2I+UhYWFhcXM4kTDfbkJhUG4L5fQ\nmgMhtkxCYV5Zv5kmVADearFQF4J7meTsNmaJd2Q746ISKXnDhe+yIyG9HEJ7o4xd9o2Our+DiN3U\nfl/DJZ644o0yhDwkbtGs6sJuJZRQlQsutgntoSsu/8bZ9FXpDQngRBXxP+axi0Vb2D6T0Oohkyo3\nfaeN45j7OghgWid8HM897vzZMW3HXNMxbdOuHuBLiNiFaw49/h24EJ5zeY6WPW0LJMY2GSkhwpM+\nLXs6V+OJhI0gLJin3JbD85VKuDQMq0WbKwvrJpxPpM9wmunxhmMOTe5ubxdty4scInKAnBEIycOD\nMK0n5whgQHzZZwLnMK+HJMaFfYijnRDcnK8hzfW8qbwTUtJ3YLnOF7hwfkX37e4REVF9qCHAaMyh\nzLReLtqy1hwRETWgE8x4uUhGm/D7K4XxMGFQnM9mHiNZwpCmXE+PlwgZLMNHRnYJPe3birz7nEOb\n8dzNYHkjM1t8jGUA60lZWFhYWMwsTtST8hPxRnygCYt3VfIS2FC+pmAJFF9xMFgT42UcsrjYmltd\nVdrnwf4OERHt7OqCciALwG6uVkmU8u2PcrUm37nPFl5eWijaYo8XPyOwaPpdptw+3twr2uplWQxc\n7xZt51a5baGhi5Bl39DS1bKQ9W5Kc+iXKcFYUX9tmviH4BPxMuBSDGU2hxSARDzjGKzl9+/eISKi\nldVl3VcsyqUFXcw2C/n/f67zk+qrjwvjpWewwO2JhRrAwnogbW6qXlMoq+wOPHOGUh6AN5+5fGw3\nU8JRMuZ9Sq4ucI+F9l2tqiflGUYErqxL/w7Gei2vvvoaERHFI302281XiIioHOq1mMffOXQ8vhYH\n0z5yY+HDcyMpJXk23WcpIabfu6TPeibe7gRo+CYdpgbsh2ZVIkKv/aRoi3bYq1p77qmizdnmd9DE\n0ZSbuvR9bzQo2sryzJRy9XLcBR5DF4gTpp8nVX23+bFEH2Kd470aj3mpq+82/+wzREQ0nGvp/Qox\nLYV3dDmTtCO8lpTv3Us/2k+ynpSFhYWFxczCfqQsLCwsLGYWJxruMzEbx5vTFgmTJODGuxKGiFJ1\nQ0PJ1E4hZyM3+0CUxuRvfPZXfrVoe+37PyAioid7mtE9kNBekqqbfP/RFhER3X38qGgrzZ0iIqIz\nyxf1vGV24yNP3fOgvsTHG+tC5+7WEyIiqrbni7ZHfVYFGMOK40qD3d4q5JmkMYc/jl3fP2EcR5z4\npMNZH6XacMwexS9X8jWQUzLqcfhhv6shjs0dDsFWGjrGC6KA4GLumhw7+6hbnHJI7xchlIXyHEPd\nZrKkqKwiITFoCyS8FCfaZhQfvKaGphwTSoSQYpb45kfR1u/tExFRvabPgytjmUz0HCaPan+g4b7O\nAc/zig+L/In5V591XwY7h9BeKmobSaLvCaM4gWQq857IkqOKCCcLed9ByN7N+VrTBBKbJDznZHrN\nY4fnaQDvJ2eRw9bDno5HfPc9IiJKSJcohOtFgwDuV943IZw3eiDvm/gouWUMSxnemNt87WaayLLF\naKNTtDUcfgc6zcWizSgExfAiC1wJZcI73xOyj/8xXnjWk7KwsLCwmFmcqCc18dg66I50gTUVa65d\nV2ujab6qSDsWa8mBD62htKPO1HDAX/a/+MN/W7Rt7vE5Ngf6Db7/mLe7v/6gaPPKvJCYeqpzVhOr\nIKjpQrEvWmAlsMbLLls8O5FaiWtnzhER0RgWMO/eZU+q01UKqufwsS8sgs6ZLwoW6bStPyLXMZ7F\n/wfP4RjdsOJPcDz3mGOnxqMB7URPPIYo0vmxvXtAREQHA+3DkdE/G6rl7paq0qaWZ70qFj5cX7F8\n+zFv150Bh6rk8P2mDpIfxGqdoIoK/z3PoM0oA7hA05ab8hw1l1V1QDsrkWcuJd2u3+PxeIDnFc8o\nB2v5bJOfh90dpZu/+eZbRET0wnPPFm2ZociDh1TOxXsGEsdoIEQRT8+RiNqGF+g7Jha69GSi5IwG\n6eL+ScFEfXA+58YPgDBAJISpNNDtWn1RflhWWnpl+TwRESW5khVIqPn54mrRNAr42P6GRo5M7s4A\ndA7zFSaDBRmkbUi6Tg2iD1GP+23i6cT3KzweHjyD/jx7ek4IlHsZtwY8MyaNIHH0c+O4QfHXj4L1\npCwsLCwsZhb2I2VhYWFhMbM40XDf9ohduU6sxInvfP8viYjo6asaTvvaM+yGtiG2Z1xmzHp2XVO+\nQ8MCJgJ3997doq0zFiWJqubGeHXJEWj3iraK8PsjKHUQGdWItrq/zTr/3lrfKNoO9qU0AZQ/KFfY\ntX6wt1O0BQ1237c3NMxY3+BrWG1qiKIirnAy5dwOIqLBSO4fQhK+hAtyaDMKGR4QPkw8FkkIXnbU\n9ilaIOzXl0V2JFBUAl7AHUO4b13CfVt7B0WbyViPIY437DGJZWtHF3cfPVknIqJnrl4q2i5fOM3X\nhDlp5hqw3IdcKt6bWed1p6xe6kn+SRZrCMuVPMRRV/uFJMSVQxjPE7HZEMSQjSCsE2loOjXhsRTK\nNvgmFAwKLAMOP21u6rXUmvx85ZDrmAuZIeqjAC4/m9v7+0Xba29zCLBW0vNeucTj5UPocTLk56YS\nwHtCwu0phNhSE0maQL/QGp04TK4jlLDIitxO3cyEBQMHlEJuv09ERONXv1u0Ja9IqBMUcfKc3xlh\nT8dyTNy/9Sfap15ZxIdrqLojZDQQ6G4s8Ls5eAyhwj4/R8GKLkfQQ/6734R8uW0eN6+qbdk1zp0a\nB0rIcWUuhpAT5ctzi2lwHwbrSVlYWFhYzCxOVnGiyTTuYUe/hXHIi22dIZAfYvZAmgFQX423BDRN\nT3TGxrF6INtipO2Arl51jq2m9tK5om2QsVW1CNRNTxYVI1/POxZrbdxTK+y8LDgOwWvaEgvOCdTK\n6XbkYsCSGvXZUvVCPe9Wj1Uq1rtKuji/IMXHTjgp4Djsj9hiq1fVe3R9Q9kHq9AMGXpN8ts7VB7j\nw0tbIJliY/0xERHNzytlv1JmC2wyVuu7WuK21SWlupoSHYOhesE1oapHoHDgCbmgD9ToRGyzAApb\n5oUXf7Sw5aG7KYotThdlU3QOi2eKJ1UC5ZK6uH0tWJB2haZfgnlZNga+r33lSp+HDtClZbziA31G\nGkI9b8O43X3EUYY7DzXa8N7tPycior0dtfD7Yz7OML5RtPnEbfFACQLPPcUKMn/v179ZtJ2W53BS\n1vsYD/jeooGet5kzNZpGGjUheopOGqaUiAt9byJCWKbClBWp76kXmzzk9JUmvE96j/meooqSPnLi\nSIOzvlW01U6LIk4rh+34uaj0dN6He/JuAxJMssORhhBKpyQ9HodSRwll8VBSICAisX/nIe8LnlRj\njckenjLaKRdiHKpuJI4p/PrRrpT1pCwsLCwsZhb2I2VhYWFhMbM40eDSU5/6LBERPfrBraKt3mJX\n/LOf/2zRVvXuExFRNNQQmyuKE06gfmOaMxGisXy2aHvjrdtyXBWEPX2eczCwBEcgocRsoguEUSTk\nDB8rh3KX3HjrzaKtKYKk1bqGw2pSefLJ5mbRZgRwPXDZ200O83X31cXe63Bo8u6GhjdOrXCI0id1\nz6cFv8l9l8Kid2xy0XwsO8C/kXBgomRejqGGozDRQLSKkmgix8BMeT72XFP7uiBHQEmAqihJDCBv\nwymEVDVAVxLRX2xLHJPTc/Q6D0cqpRTGMfcz7cq8D+/dIyKiGCo89w44fJPGOmceP+YQ6h5Uwx30\n+blaXtDwXL3Gz5UH4xtJfpEPoWlXKvgOIPw6Np0EVYIfPGGy0N2HSloZTHjfcksFf50a9ynWY62J\nasz6/feKtidP+Ln67ne+V7Q9fY1DTUtzGoYa9TmUODjQ5zp+mkN7/a6KP3/x2S/TSaMUcp/mME/J\n5HkB+8aV3/1An4b+K58iIqKm/+mibdjj8Y09rGgrfR5BKZYKn3eQQnVdE6YFskIgz/QIqiybANwI\nwvrDPp+3VtF371j2KdV15OYb/D5O4f3Zr8hvILdUYt43gVC/6Y74YzxH1pOysLCwsJhZnKgnVRXv\n5vxlLaMxEiP83KUrRduiSMPv37lftMWF5pVa1J/9yt+XfT9TtF18/h4REb36uno+7TpnYz/ZUiq4\nL/TLUgBWjnzE+wNdwOzusSXYrgU/vxmlQMdeXGKPcJKoV7Gzx56RA7T5hnhfvqddHY35fHcePC7a\nlubYer16Wq3EaeF3/uX/QkREDtxfINZRvaHW1JWLTER55YVnijYjwYZU9cxQ0MF7oUKzUS02s/Ae\nlvQchhARhOqNLrTZ3stg4TUQkkSIVTFDQ6zRMdkXkso+lBjoddn6jodKGjAu4cKCpktcvcKWewCU\ne3Ob3pR1/b77vR8SEZEDZA9T3HMECif3NngBHj1CM0btJiiriIJACbYLhJbul0CTz+exGY7VSvdb\nfJwctCw3OkxbjiH9oNowfanjEQ1EoxLoKGN5HpoNvb7Pf/p5IiIadDuwHXvNDx6oh/TBBx8QEdEI\nUhHu7/K4juC5/uLfpxNHrcbzLykr/To2JVOAbp4IWcABj7WywuSIg4F6rNtd7lMH5ng0FDo3qDdE\n+7wP6qGWQn5+D+C5LJtioMDOMnNoAsothiHVHcG4yZ+rvp6jcYYjWh5yH4Qg4qD/Y8quYIxFPKjs\nY3DQrSdlYWFhYTGzsB8pCwsLC4uZxYmG+7ySkAs23inaXvw0V+GstaCqZ4/DXpg17ktY7s4jJVN8\nsS3lM6pnirZGjV3dsq8LehWp1luGkJHJXTp9SjPPb0qoIAw13HQgi5UXzlwt2q49zeGtTgeq8DY5\nlPFkU/MVHFmYnINSHV1ZvPUgBFgRJYxRSV372w8lmz6cvt0wklyjCHKOAsmT6mmKC1WlLX36uu6b\ncxzAS9WVD6X/cU00NeVAIEzWmueQ6SHxWYlTxah0IaELD0Ndxb+63b37XJn38ZaOSWeXF9RHIw3t\npRMpDQO5IUaM9MxZFfg8d5bnWS3QsIwJWeRTVpx44zbfW7WiKgC5EFgmic6jVptD7CUgHEUTDntt\n91HkmK+/UdZweiKlchxY0PdcycvxdbvSgMcjivTZ7OyZsByWe5HzQ4mQnoRYo5G2nV3i52VhTkVT\njapFZ0/FaRfm+Fo+8ykVp330hN8d3bHOjXcf8ZijEPU04Eu/Ver6buuPOGTnQ2mSVEJivgPlinIh\ndpHOSVNJ2Yf7ML/iSMe8Iu9KH8J4JnQbwL6pLE1EY8wZFHJQBUgNUgE7hHlgRGmDBEKPwobCSsll\n8x5IQc3FKLdg+Z3iTx/9HFlPysLCwsJiZnGinlRQ5oXQ8QQtVpHbD9XaqNZ4u1pJ20oeb1f31fr7\nn/6Hf05ERH/3t/4zPYdkmocl1Pjjr/jFS6eLtq1dXlAe93UxdXWZFQxMITYioomsEF66rMSOy1eY\n+NF9/bWibSA6cbjQmYgnOBrpNc+1eUE03dPs9+YcWz5JpLR01+XzPlpXy3Fa+Ee/8RtERDRB9Yay\nWNCwXUXor2AAFjToDMosBKJi4AOFNRdPcpRgaQhR2cBCfpJ64ANZwRAXHBdKIIhXFYPlPhZtuhro\ni7Xn2ONNoa/Lolyyv6tkikeP7xER0ZWLOu6eWKYpuITGA8mmy0CnXqF1BsQEyfSvQHr/mbOXiYgo\nnuj9bm/yM7KzqzTtlRWmhZcWNSox2Oe/ZzDArXn2LEsl1cEcy6GHiXpS5So/w2msRUA9IQuEZUgF\nEX2+uKzj+9mX2TO6dv6UnkM0Be9+oPf2wa2bRET0hVeeL9rOnuN9HrwJpCux5rN0ujqYYcj3FFaA\nmEB87xVISzFqCz1Q8UiFHFFuaRRmpSZeM5ALjCePHognvoYHpYSwCOTPI4eIlfGkUqC5m3IrLqhB\nhMaHgzSVibxnMRDiy74pgWagc7TIY6FU4320n2Q9KQsLCwuLmYX9SFlYWFhYzCxONNznSOb1EEJs\nY1k4DVBIcVdcQ08XqANZQFyb07DA+++wnP2TR7f1JEMO491/eK9oemmV1SxOn9eF2FNbHLYY3Naw\nwHyJQ0GNlgqX3rnDJT/WTmuocP+AwxoxuMmbWxIagUxyR3KhhhDuc0RcEcNmNaNckalKRujIgvLu\nJk0bmeSpeWCzmF6vl3TBvCL5M6OxhnmGMd/fvTv3irZQ8j/OXTxftN0VAc0//JM/L9pih+dHuaR5\nJdUKnwOz3VuS3zPXUtLASy+9QERES4ua13RZxsx1cOGf7wQX6k0ZktGyhlZOrfJxTp3ROWOIPEPI\nF6mJKPHHiFJ8ogiEhLS0pCGxsoS4d3YeFW0DUQtAhQOTN9Za0ns7LWHNRkvDeM1FDgHuAkEolRBN\nDKIgoyE/z8ORhvai2BBToByI5OWUYQ4FQhBYXtacqKU2/y7DQv1Sm5/XZqj5irv3udzN/Q/uFW2r\n8/zsdrd+qOcQQk7knejr7Qh8edY9KGtSlnfg/pb2aafPoq7b6zpu7Qa/C557WkOZgcy1CRATYlGG\ncJFYJM8ths3NbweeBRPGSw+VnjH1RTB+zW0uKmdIyA4rL/tyDS7kcbnyTAe+PtPBMdW7zT7pxyh7\nbT0pCwsLC4uZxcmaGkbLDr6+a4tsMVRhMfVbbzEVvJ3odlfbYmWHasWGPlvD21v39BRjtlDOXblY\ntJmCX9WmWomLK7xAvNtR66/bZdIDyFbR0hJbkz54eqYAXwRKBiMp/ZBA+QNT1gKJIknCdsDCIuiX\nibURknpcJVmETHP1JqeFP/h3f0pERBkSOcSTrQOZpSEezYWruti+tMAW/sKalkWZl3st19Qb2n+H\nPdi333lYtI2EkOADU9gUuWvAvlfOskf2hc+9XLQtyKJyzQVlEDHeJnAfqRBhhl3l0sdCsKiAtzYn\nRS43N5S+viPFEys1HZOVVb63alXnxyJ4eCeFuTn2GDxYEJ9MeP5gdn+nw/d5cAAEBpnLXqYdff8x\ne+xNKBfTEpKJB0SMidD0HUfnfsmUrMmhaGfO53B9sIzlua9VdbtAlGTOLKh3VRXCweBAxygZitoC\nWN8XRaXmnXfuFG3XnpISHECSWBdaeglSQaYBQxDwwbPIhGjQ6ylxanubiSz7HVWcee+tHxMR0btv\n/KBou3KVU18uXHm6aGsvSooEeCCpKZQKBTsdiYUc9vj5P76PaQJCBELFmOKFCOQModDnoPpS1Ak9\nhkWUgwJIUqRtKMx8wuKmHwbrSVlYWFhYzCzsR8rCwsLCYmZxsnlS4va2IFwy1+DfTqZu3kHOrv/O\nnrqXi3XetwYLp6nLoYd7T+4VbStznId0/oqKnpo8jh+/qkoXj9c5LNioawgwkKz8G7cfwFXzdzuD\n7/dEXNL+QEMjcyKOmkBF2nVRn6g1tJKmL/kHVQh5hCaUGGveSjrgUMfK8smHjn4eP339bSIiKge6\n2BlNmBwRgALG5z7PaiH3H2vIbpfXgOm5Z1UFIJQw2hDCnoGEYA3hgYhoLJnvmNNx9TKHbZ99Wiup\nnhJyRLOq8ygTwdOHkFe2tS8Vj3e0bSCknf19DSVFUtoigErLRuQ2hTBFLOHd6pyOyXPE99mCEN+l\n1SU6aZiQ3RBUMjyJhXk+lLOR0gy+p7liRvA3DPWaFxdZeaVe1z4ty7i1QPDXlzmBSiG5LLInkPPW\nktw018WwEV+rn0PF7UlfzgHHE0WKNFVySyTKBiMQtq3Kc3V/U5+bm3c4VD2Z6LMZj3nccm+6ihMG\nHsSvy5JveP0pVWm58jQTfIY9rSZ84zXOwXz9p0oA+e53OET+zs23i7ZrT79IRERXn9IQ4Fybnw+T\np0VE5BWkESQmZMe08XjFsGyBOY9Fm7AtUnjfZUXO1i+GEa7OQf3CqIEkHyPh0HpSFhYWFhYzi5PV\n7pPFvdUVpb76xlMB/ai1M2w9/xQ8pH2HrdPcU/p6a5EtgVZDvatAtMwugCdlCiD+7u/8z0XbUM53\nMOxAGx87gF5YFTWIcUep6oMSWxktKMT37i2mw29uqtV+ICoUcy09YLPNFqaXgyJDzOf1hD5PRLRU\nY4uxVZ6uJhwR0fYjvtf5tnqZp88wQeCZF1TDMBDr98brPy7aVsRSrEMpgq0ddq9qTfUoF5q83d/7\n21p8zlDFWy3dbnGBx64D6gh373Nfd/eV+n7Q5YXoHqiF7IvX1OmpkkQSi8IJlGgJxTt3YVW51eJ7\nm5tTSntbvNoSesEV/t0HdY5pYGGRn6EsVnJRXQrMZal6EYFo7S2vKFXdkbIrYVm9JuM5lsugAmAW\nx8FrKopFQptRNhgOlJzhCkmiBA9TLl7VsKtj+fgej2Un0OPNiULDCpRJKZe5n3FhPZeiln5V6evb\nj/gZOrum3mxDipkeTKarOJGJQgNqBuamUCgQHYyXM7egBKQvfpWftytAAPurv/xLIiK6e08JFoPX\n5T0GJJPnX+CCiWfPaDFYX/o3BUKJKTWUQRQrN3SGHMkU4iFhpR0hgCBJxzhBros091zOBbI0jiFO\nwL7i3aNn9mGwnpSFhYWFxczCfqQsLCwsLGYWJxruMwSBZlvDfUnKpyz5mmdyTSq+/vRVXdg9CDgn\nInM0v2DlFIctbr6ruQS/9JV/SkREP/i+LjgOBqIQEWll3q0Ns9iv3+V+LIvMkCXf9njx/XRFQ0vd\nbQ5RJJ6Gw1aW+XcKlXmNsOx4rCGowYDvM8k0NBKPOdN82dftTom8/ySBarFTwqNbN4iI6ACEWf/O\n3/pPiYjom9/8RtH2Z9/iReplIBIs1zgEWoH8mLIIlK60NCzTEKJBGcgPJn8CS6WYvLONWxrieLDF\nOT1RrCEJX0pMNBqaC7MsIaI4OrrwGwABx4haorhlo8HX12w2YDu+p35fx2lzk+cUjjG98qkj5/uk\nYUSYYyBOVOrcb3NNzcHLhPjhhyFsx/eUQ6kTV0JOWQ5t5tkA09Wk3uSgepDIHE1S7YODXe4XfKEE\nEu7rdzUkvv6Ew3MrbZ0bczXOARtGeo5MQo8JHNEQNk5DWOupq1w9+cVnLhVt793hZ/31nylxahow\npXpcqJrrBkLSgdykVP6OJS5cCUdfvaaKE5nkWK5v/H7RtrfD/ff+REPam49vERHR5atKznj6WSYo\nLa9oaSJfVCCSSJ8Fo2CRQsVsM0+c49QgDondHv17UcIGSRKO2RUlJ0TVAvIcPwzWk7KwsLCwmFmc\nqCdVa7C1215UbbxErIixq5ZeuS7abHO6gP7gEdMzv/iKUpvHff6KVxuqCrD+iL2S27fe03NkbG3C\neh4NZDG9saCWhVGcaIG6wVPXniMiop+8+W7R9tq7rOf3xa/8WtFmSo3c+eB9PV6PLUykr49F3+z8\nilrolRrvOz+v1mQuBc6S/wBmw3jI/fDCp9SK+/o3vk5ERAtQOuCXP8ekB/cLahE1jLdcV1KJJ8Um\nfSjHYhbRM1LCTFcK5TXbOhcymZKXnnquaFs+w6VSOnvq3TaE4BCD5pgjZj8WejOL2eOxeqj9ARMs\ncqDd9kXh4OH6etE2HnG/xECSSMXyrNagoOYUMJBraFTR0+O+2tpWYsKBUO0zKOlxRVQZ5ub1OfRM\n+RNQEDClZqJIx2goJTPGEyhJI8UOnRTKrogCC6aMzM3x3KmESmrwZRF9DgoDthr8O5rAeeX6Iyg5\n4opKQRsaL8gjAAAgAElEQVQ89KroFz56qEQnU3Xi2aeU9DMNmOKd3iGSiUQLwOnIzPwEz8IoOeD9\nnjl7gYiILpy/ULT9ZJPnZwKpEttbPObbO0rEeuedt4iI6CKUnrl8mftjZUV1SRsmXcbRcRtPxLuC\n6wvEM8+BYGEo6FjcNMc6PnB3REQOkCTML88WPbSwsLCw+JsM+5GysLCwsJhZnGi4L4slnDavC/KD\nEbuSQwjTmAXsc2c1b+C9tyU3Zgh5ITUmWEjxUSIiuv8eu/mPn6ir+4XPc6kOLCXQqLOLO39a8xAe\ndDikN4IF27DGIYrmoi7OvvQKX9f2joZV7t1/U+5HF7L3u3y+pUUNb7Ryds/P13S75aaEpRwNoUTS\nVzVn+nlSl5/mhf/f+o//k6JtmLL7f+u2hlYzWeQsN3Q8YykJ0dkHld6MQ2tpAiVLJJqA4T5T1dfb\n1BDHky0+3wTCHpkoCNSqGlK88z6Hee8+ULUQkw80v6glUEwIqdvVhebdbV7kx3CfK2EKrP5bqzDJ\nY66i5y2LcsaoP12CS0nCaLs7Oh4f7PF9oFLDnOS6ra2tFG2RKAjEkY5HJgvlB0PNQxxJeDNN9Hie\nhGlDKKNhQnplUJKpSH7UeKjPXCbEgFpd54sJhYUeqiPI8wChwrEQkhzYzhAN4lifpUe7THQa9nV8\nfSFlrZ7S98k04Mkc8jDkZYhVDpB5cpM3BM+M8RcgdmbUKoywMxGQGVABJDfhNCVx9fZ4nry+A6oW\nb/6EiIjmF5QAtrrC77nVUxfgvBwCXIClkaUVnk84Hpk8PwmQLhJTDuRQnpT8AyHoXMLmeXZcePAw\nrCdlYWFhYTGzOFFPqtdhL6ICZS8mosWF9e5NhvPivFrA73ksx7+1q5bermcUJ5TSfv05/urfua96\ncqZA2z6oEVy9youGVy+qG3b/CVtfN27+TM+xzYu4YUmtv7bQkx/dUDLFxi7v6wABxCvzdmtn1Vs7\nL1bEuYaSM8pSlmMyAcptxlZkHE83S56I6B/+439MRETtVbU833ybPZUI6NyRWD0pyvWLdYQLoI4s\nlKZQJNJktruHzCLRDYMSLTs7TDdPgIpvnJu5lioSmMX9DswPEitvZ1s9BkPpT5D8IJa4B+oI1QqP\nYwlo6V5iCiYipZ0nVwXINtPA/h578euPNWJQFQLO9WeU8GLKpFSB6j8ecR/t7anaSiz6hUPQ1atW\nRbuvqc9rTQpdVsDL8R2jZqD9kiSRHFet6rHMc6QqGzWGFCIpJrPA9/RZyjNJ55jouBkPeGdXU0tM\nCYw90GY0Hnepqe+TacARLwIVAw2RAL0cx3hLhyQd+HcABUBHUsByY0PJPOtC7Dno6hgZjdRGTcko\nNfHCqlB80JB+HkOxxffv8Xt2NPqWbpeZ8kKqWvL8c6zoc/WqRphMWaNmUwk5JVEDyQmeD3kPJBgk\nEpp7ZIkTFhYWFhZ/k2E/UhYWFhYWM4sTDffduc2u5LmrKitflnIbWaThHL9sxC7VRWzIYmsdciKu\nX+d8jz/7v/+4aBse8MJgdV6z7m8/5kXDs6e1WuzFp7iqawnKM1w6z3/f39sr2m6+y4SNDBYDH+/x\nNR+MIZSR8rV29zWkuCLhsge72jZ/lsORuyXIq5E8rv0ExTM5PDPJNPwyLbz+xk+JiOitn71RtDnE\n1+N5EOYxFV59DHXx3z3ITfIld6UMJR+MwGsIFZndUM6R6zmaJV7UdctAzvCk4jGIZZoIoRF8JSKK\nRxLCGmg+VSQkACeGkJ3EHCMPVAAGPGaDTEkDVZkrS6DE4UuYL/zoRPlPFPNLvHDdhgrPpgKsD89N\nr8/EhX5f+6BUMqFkKJkhZIpTK0ryKZU5NORBuY1cxEgHkGc2FsLLPoQPd3dZVWI00vDr08/w8xqA\naG+RHwNqBoYkMRmouswjUYjZhrIrUcTXPxzoObr7HHYPPX2uTR/8+bc0hPXf/Ff/OZ04JI8rA/JD\nHoviDNSlzQyRwMfwpgkV6jvmzddeJSKi/p72wbzknj6CEjVNeUcGUFE5kzB3sw45W5IbF/pQKbnE\nx/Nc7dOO9On9ezeKtv09DhG+9iqUtxGlmLNnVO3j1Cl+p66d0rDgqRVuq0GZJKcigrXuR+cbWk/K\nwsLCwmJmcaKe1Bvvs0dz7rnPFm0Z8RfbAS/CZF4f9NSS2u/y4ujC/ItF26/9R18jIqIXP6UF8X7v\n//wDPh7okrVa/MU+DRTUeoOtOS9Vi2F+lW9/7aJa2V0p/Pb6G+pVrA/YGskD9epaq7wou3hZVTKM\nh4Hy87ekoOPtTbWQQtGEG411UXgo3ZFk0y/U9t1v/xlfT08XnwPx7CpVLMLIFrmfg56a2DkOUJRN\nSY9yCQrqiQcVltWK86vch+VQ+zAUmrsP5pMj5UscsEZjsaonI+1D4ylkSAEWUo4P+xa6YiV1h1q1\nQP7Ve6sbMkWgxwuESuwA7XsaiMU6L4Mn6gvlPgU9NaNw4AMBxDgt5TIsyg+4r0ZdfeZG8tOHQpeu\njGsOXuytd24SEdGDe/eKtkQKHOYQgTh1iglO81CKZSTqJuZfIqL9Ds+73X1N8RhJEcMUykoMZZ/u\ngXqJroxrFQpnbgi5YH1D6dfTQCxeO6pGOKJVatQyiIgMEzuHd6AZrn5fKfxjiQw8dU0jUS+/+Bki\nInr1LS2E+KOfMrW8O9A+TYXIsrym5IcvfvGLRETkwxy6d49TeH74I9VDffZpJkk0Ydw217kvN0VH\nk0ift9VVpapfvHiBzw+kKaP2k8MzGPj8Xhwfo7P587CelIWFhYXFzMJ+pCwsLCwsZhYnGu57r8vh\nnp1UQ0Z5wOEZN9IM8VxCXFjR8tQaLxB/6QsvF23lkEMJF85pGO/Xf/O3iYjo//g3f1S07Wzwsde7\n6nKOx7eJiCgkdbE7EmO7/QDCAuJ+5ksqe99e4RBVBu6qIxIKGYSvMofDKSh62hXlhnIAgrpS1mIA\nihOxEAvy7KPd308aq1KBdn2kagaJhLNaC5pr4ss9d7eVaNKTasRJAgoRKY9xDv1QABa4wzKPcQbh\nvlSmpAulPyoiVFuDCrmZySfDc5TF5gpxXz5eBUIc8yKGe6au8/LMKud6VIETMhlz/MvNNaToS6h2\nrqmhzGngvfe47MSzz2oF6oqE7zBp3xVqQgZqGpui4jEA1Y3J2KiCQOVWCeldunKhaFta5n5BBYFA\nyDQtIDUVpAuIVo8lnP3uu7eKNiPki/lPcXKUcDAQcsYICBtDUceIJkoAMZWAD7Y0d2p/f0/u55j5\nd4LICyUJGBCTEoWVeeVn5sKSh7RVgAj0pa9+Q/6k+/rS99de1CWU5z79ChERuRjRln0W4fm9dIlz\nRH0I+164yiU9Tp3XJZSKKK1gxWxzb52OhmRNSG95SfNWjWCtB+FX11ThBVJSLP2ROR89RtaTsrCw\nsLCYWZywJ8XfwH/7V6ro8OJ5tsxWQ9VDq4o1tLaqX+S1RbbSLl1WWXlD3V7f1a/57/wr9qBee/1m\n0TYRSwu5GaZ6W56qFZaW+BwpFN7yhXqdABEjcbmtjL0l5IhxBHpUYh34QNH2xALNx0CfFksrAC0r\nzxFadDx97b7caCxC+YkDsWCjVBdyn7rOZVPyNS3fYfQMN4EqPNgX2vJQreBM1AlgHZxqPltd119Q\nFZDHYkFvH6i3NpqwBW205YiIfLGvQijuVxdvdA405RaF/rx2SufWlVNM514u6xgbynano/fhCZW+\nCtTZeoOPvQD6Z9NAIl7dGMgtriErgIdvihmm4Nm+/z6Xsel3dd9QnrkA0gQMpT1LQNPQcP3BK1lY\n4PHHmnjDMc+TEehlPpQyOridY0gDID0yFE3BLqhGDHbY6wtAFSSRNIIEvETjHSZAfU+KRfuP1oX7\nJDEa8Xz3DrQPfFOCA5Q9EqGZJ6C0YtQgMvBYjWOZAGnFMekT0AenzonCTaYd7chvN9d+vvuAUwZQ\nq9Qcr9FSlRxzDXtdPa8vnlGteUFvWN6Bna4+5082O3IMnS8lUeWBR5Wcumg97qlH/WGwnpSFhYWF\nxczCfqQsLCwsLGYWJxru64ub9+evadXc9z9gFYpvfloXgC+f4rDP3Tta5fbLr3Bl1jJkZfdidp3/\n9Z/8pGh7/SYLbg5TyFyWcJsLuTvG/cR8BROeS8F1nkgILk61zZHcmAlpWNAsJPqQ0GNKDhixUiKi\nQFx78H4plerEmEuQCBEgrGt2/rSw+/iRXIMubI4kVDKCUhgLEkpaBDWIQEJxVU/vZeSbBeSjwqwE\noanhiEOFX/qMVl9+VsRSHzzQSqu7shBuxImJqOhQH8g2FVk5XgSSxFytJmfX8dzY5Xu6taPCnY4s\nJjeXdKG50mJiRbWhoWlTBqQOi8rTQFnmWQThtHIg+WPQB67MQRfCZE1RzCgHul1dxEg9UEKpinJF\nArkr77/LosrdjqpLdEUZIoWcqCAU9QvIzypJfMeBFf2hhJG3YQF+KCQKVC1pi5hwhLmEEvJMYhBm\nLp6ho2KtSMSaBr7znb8gIqJu8lbRVhN1h3QMJCkJp8WgLmPEerHybSwhW3w/GUKCqZ7L+5pSHXq/\ngZQrmW+p+Gtd3i0xLDOY95IDYrfmt+u40CbzCgS1fXk3H7cdVus1aYuOA+/UmhxvpOH1D4P1pCws\nLCwsZhYn6kktSPG/zp5+Vtf3eHH0+29q2Ys0Pi+/9Cu9JDp4rqeW3k9e5SzrP/qWZkdPMqFswnau\nd/TbmwqZIgeXxtB00XoxahEBUCiLQl9QSsAsMntAqW5IMUAXFoW9XAgDsICZGY8MFk5XV9gyb0zZ\nQiciWj3FC+GP7quEfxqZYm3qed69xVTibqg0WXNXA1QGkGz3/BBzheGBwTuZsGX82vf/tGj7qmg2\nPgd9OGqyR5NVwRKTY49jtbS7Qpvf6igd+f4tzpDfGalKwVg8kMqyEkDaK2xlllp6b554xFWkWksZ\nCMc70UfnCLTEBSx6C7kng/IYhjaPxImKzGU30EjASPTvJh0t/fFQ6OEZjJspKxHAvkZZJQDiiSvd\nEUW6b29vJNcEKgriUSA9qCxjHUMB0VgIB0hBH4m6CHoW5llLwGvKUm4rBdMlIZXFa4pdmEOizlIq\n6xzKTCQFSBKuRHWQBJPJM+WAp2IUPbIcVC2kN/McvSFDgtHrc4n71wfq+0RK3jhYQ0cOk8C70pQQ\n8mA7V96B6IW57tF3byQlR3IoQzQ2oi/e7pHtfx7Wk7KwsLCwmFnYj5SFhYWFxcziRGMWZlE7gMq8\n8ZhDKPc2NfwyGXA2/Zdfvla0VeZYtLAL1Wu//SMuKTGGKpdmcbEE+R6G5z8EEUsDz8GKwPIDFvlK\nEsZxYDujkuCU1I03Wdk+hAVjCXWMJ5qzkYrLPIHQXqvNi5krp2BRU5KwRgcq+DktnL3KUvoHUN5h\n8OioGz6ScEEH7iWQHLMYlDKS/BfkqWBIQv69/aYSYR72+DhLruY6mXBsCqGEvpTr3QA1iNvS748S\nJYAMq9yvjbMqtLlykcPLZQjjFUoYECquS+ix2lRlClfmcu5M177r7XMIE0WAt57wszQZ6/2mcu9Y\nlsOEajDU7UrcNQg0dGZIQB7IRpjyDlhENpZF/vFAzzGRvu8daHjOPKY1qEptwkU5kB8mIoyKqiVd\nERA2uUdERIkJf0GwEMkbeh8Slp+yeksm4eb+QHP8qlKKAqL9lIpvEENYNYpNH0DekITlcgjtFSLK\nCeSPCXEiTZDsxedAFQ/z+OS5zpdCeQSIYmYfnC+qopFCmxA2nKPPNJ7XEyIOju9wjp+91bNKwvow\nWE/KwsLCwmJmcaKeVGa+zmBG5FKYKyK11rb6/GV/7ZYu4v7akL/EvVw9i8cd/l2q69c3GfJxxhMo\nVlcVLwdouObvh+i6sriIJIlcVoBz+H6brPw+LFBHCVuOxqMiUosfvaa+0KYbc+o1zYnWVQTW7q17\nTCQJsqOW4Umj2WYCwdKqFtRbF08Kl56NbTSG0hDGqk7Aa8p+QaY/ltswB49gcXywzZRUt6xUfE/+\n/gSsuDeIx/O2DyUBGmxB186oGsTSKfagFqRoIBFRSTQAI7gWU3Su5IMX4RtyDLYJCcGbLr154z6n\nZ+TZUZUCNDX9ktCC4fKMpRtCRKMq5Be0gk0EIgELP+7z+CIhIssNRRmsb1G/CEFJZln6ftBXzcAD\nSSdAmrsh2KCHNIiOeleG15wj21z28VEbT+bJYDDdqMTDh0zsur2uz3VVaPg+eHxpcZ9KRjG6iRl4\nf4EonmCb6Y8UOUkyhkhqcEQTz0W5D/eop2xSc6JI35+G1o96g65Elhy4ZuNo5fA+MA4UvjdiSUVJ\n2zo3Tj3H5UdaH+1IWU/KwsLCwmJ2YT9SFhYWFhYzi5NN9ij8QagcKlLzGVSgTaXC6L0tdc9/5/f+\nmIiIvv7VzxRtd9c5FDRMMeeIfwcgP+9JWYxqQ7cLpeLuqKekhmJBGYkAJV+O4R3Z7jg3eTTsH2nD\n7doSSltY1uqVO7ucvb+3oyVCuhLOuXJJhR6nhUqZ3fASqA8EUu4ijXHxlJGCgkBqQnsY4cuPYaSQ\n2UwDAUbxYwC5P+9KmKcV6mL7u2MuNXEj0bHrSKmM+XPaX2sXWIx4blXzn0o1yV0D8c1Y5qMX4Jzh\ne/dBBdOEO9JD6iNHs/GnAS8byrVAR5swGYRlUldCTbmGZUyXT6CacCIL9RnEzvA+DQwxKIB+Mc+w\nfyi/UCoHl/SVUqpwn+7t6nkHUtolgHhkIa4MIfskP6pQomEt3deRsSxDyL4v4sTDoYYZpwE3l/wx\naHOkVM8hBQYTYvVgXgk5wsfQsixHIDfEqEq4SOxyTJ7U0fFDwVrzXkph7sYmjxOUJHIJFeI1m2cV\n3+Umh+5QHpcQchJ4fzZE0PnM80qM8x2eG/vvqfj4h8F6UhYWFhYWM4sT9aTmRX9rPFYPqT9kS6/s\nK+HASNa7sLD7nR+z/tXdx0qm6A74q7/bB1qqGI71GpApzCI4eAbGQi5XQAdLLDIfLGpDD8Vsaycz\nlgUsfsrCb5ToImlFtM8WoNDY/CJ7UBGQRyZC6BiV9LwdyegfAIlgWohlFXYA41Sf4/GZDJQSa/ok\ndXGxXX6k2F/FryPnwqz4XPq/D5v9lRTDvD+E4pRV7jt/9WzRtnqaiSgXl5aKtoUW97sLc2EglvgY\niqv5YuWVQeOvLEoSPnhwZSlAVyprGyovTBNG2+2weSuKKeAl5uL5Zuj9mR+ohCJWPHqT5nk5pCog\nY4Q+cS7knhTo0qmUUYl87dORlM8Y9DXaYJQwwlAt7bEUM0TlF2M+43mNB4JOe2iuL9Jr6eyw5x1H\nR1NQThKJ0P9TuJa4xHeQpKg7aQp7QpN4KFi4MM6OqksYzwjJFKEogKBzn+VHvWzz9xRIK2RIEtDT\nhR4mElQ8+TvoDQZywBjnn5T7aT+l5XdOX+DndryxWbTducXpROVY58aHwXpSFhYWFhYzC/uRsrCw\nsLCYWZxouM9kM5fgUzgRNzUAMdlEvEus1ulKOYj76yrl7kreSgKL+SZUOAZJ/76IZ2LYwoQyaqGG\nayoV4yZrAKEiIaBKRUNGkRAndqBcQUbchqU62lLSYXVec3xWZRF/f6CLwr09XtjFSqnted5uZ1vF\nUaeFWBbUvVD7dX5ZcolG2l9JxH+Pce3eVB4+JtyHeS+uCdWgdIEsdvs+kBqq3P+TluY6XWrxwmu7\nrQoR9RbvW69q2Kgki/ZjyLyPJGcmR4FUkz+H1yK/DxEEfKOYoo+JWXzOjyGFnCTGEqLxIY8rNyQO\naPOFhOR6er9G9NM7pqQHwUK9ERnFXCxTFTbNUOVFiCdQfiLqibAtxLBqUoIDBWs9ueYJKEkcqmMj\nSPOjbSafygfhWEPi2N3Q90Q84RCSczTafLIwOtRwfa48Uz4mdxniF7S5sjN6DZnkoXnQFaWQ+7fd\n1OfDJUPw0X424+WBonNJlheSBJ5V2RcJFkbAtwcVhnOTxwWlkw7kQfcX9VrOXWVyRHte80Ifv3ub\niIh2b98t2ny5vnLw0X6S9aQsLCwsLGYWJ+tJibR+CRbvREqNslgtKWFaHlIqML8zUKYwljylsFBn\nstDB8iosQfCkOh32XnYTPW+rzp5Pq62U5aZYmOWKLpanGXsaPmTYe1KmAHXTTGkA3C4RGmwy1O36\nXVZzSEFxolxii3A8ZSUDIrX8WvPqPRoPJQXtROO1JkCDzskUmFPL3RHbxz1Gwh/LqPhiZVbAEzDl\nTlbqWrKkHjKJo1ZSsk0oHnEEPIZ+yMceoUUpi7tl8KRC7yituig7AHPGzKkIFprDUBb+g+mOU1CQ\nGsAbkt9YHsFoCqITUXBGUIvNkICAlm50JtHzSWIp+Al9MBaShCFLEBHF8qzXwIutNJnIguoSsSiw\nOMe4OYfaDEEAeSLivZbhGRl0ObpxcLCHGxIRkeeebIbNz8MzIaEI32MTuSR91o0ihgdk9YIUkqHX\naSjeUL5jwn05HKIHbH6Drp7MXSzjMpaisRjhMCU9jpOWSbFoqVxDCs9vY5nHd+mapoG48t6+9eMf\nFW2TbX7feehRyxhmOdJgjof1pCwsLCwsZhb2I2VhYWFhMbNw8mMWKC0sLCwsLGYB1pOysLCwsJhZ\n2I+UhYWFhcXMwn6kLCwsLCxmFvYjZWFhYWExs7AfKQsLCwuLmYX9SFlYWFhYzCzsR8rCwsLCYmZh\nP1IWFhYWFjML+5GysLCwsJhZ2I+UhYWFhcXMwn6kLCwsLCxmFvYjZWFhYWExs7AfKQsLCwuLmYX9\nSFlYWFhYzCzsR8rCwsLCYmZhP1IWFhYWFjML+5GysLCwsJhZ2I+UhYWFhcXMwn6kLCwsLCxmFvYj\nZWFhYWExs7AfKQsLCwuLmYX9SFlYWFhYzCzsR8rCwsLCYmZhP1IWFhYWFjML+5GysLCwsJhZ2I+U\nhYWFhcXMwn6kLCwsLCxmFvYjZWFhYWExs/BP8uBf/9WXcyKioNEq2rb2domIqNPZL9rGvTEREc2v\nNvXC5heIiMgJPD2gx9/UuBcXTfdfvUlERGErKNrOXl0jIqKK7xRtWcx/TxNtay9ViIho7cKCnsLn\nLsnSRK8lCImI6GBPz7u9scXXkunxvvDZ60RElE903z/9028TEdHpC2eKtkpQIiKiJw839LyVOhER\nNWqNou3P/tWf68FPEH/4X/9mTkT07vZm0fadN+8QEVGtWinaXrl6joiI5nK1bfLBiIiI4lzvOajz\nPq6r2x0c9IiIqFQq6Ynl793huGjqTvh36ut25TqPT2cQFW0b5lpHadHWlH4l6LWEMiIiGsS6b6lc\n5r+lWdGWxXyceqlctC3NLxIR0YNNHadBlPz8Kej/euPmiY/Tv7jWzImI1up6zatlfjbKjvZ9o5wT\nEVGrrpfku3zviaP7evK4xLor9Ua872ii+6Y5//ZgLGPi7fYGuvMg4u1y+RsRUdq8xOd9+TNF28F3\n/oKIiLZ8Pd5mxBczP9gu2u52+JlLGvpOoDo/I5vDQdHUGg+JiKgEbUOX79PL9Fr+xW584mP0m//k\nH+VEh/sgz/l3ksD7xOFL8cFHCMMaERE56Dc4/L7xAn3vhGUey9FoUrTFE+/Qv0REWSb9l+q+44T7\nKE0nsB1fH3QV5fJIJYk2JvJ8xIkeL5ZnKoZJlMX89zzTuebK/U4ifQYTeY4SON7W+vaxY2Q9KQsL\nCwuLmcWJelK+WNSVpXbRVp/wV7yzt1e0Layw97B6ea1o2x+brzh8XD2+3OGoVzRlOX+dq0311paW\n2fryc/WuDrr8Zc883be+UCUiojhVa3wy5t8pWt5Vcw1qqcQTtgD8UD2NhRafd9jvFm2DA7b0tp/s\nFm2VkK/LA4+k1pwjIqIIPINpIRBrNXrwoGj79PULREQ036oXbQ0zW/p6jXmF+3Cupv2QpexdpdCv\nlRLv7IA1n4x5LjQDHSdKebvBRK09z2ML0BmrxxXKkIzBajU2GRyNSCzZAOyx/n5XrlOvpdXgOVgF\nT8/J+e818K78IJPDTnecvsZOLDUDfR68gK3RPljVbi5zFCIGkVi14wisW5e3m4C1fCCHGcRgQcsu\nrq/7pnLo3kjPMYjNabVtOOA5f+ff/1nR1sr7RESUwzkcedQSiHzUG+w9365rZOFnexy9aMG4zcku\nIZjbxmP0pmyDO65cDHgljnPUOchlXuWH2ng+uRA4SuXdlkTqJZLP2wUwD5JYBs6FyFFm5gTMF8f0\nR6jnzfgcAXRVnPNgmuiCXCD/k+p7kTKJKmTqDTlZai5At5M+cKDNl2sNQoisfAisJ2VhYWFhMbOw\nHykLCwsLi5nFyYb7JPwVlNS9rEtYpVbRtpUz7NpX6hpa6kYcFvB9CN64fLkphH1MhKBWU7cxlkVK\nN9MQ1HjQ4X8nB0Vblixx24Eer7PBYUg31PMun+Xj+OBiTwbsTpcl3EVEVC4JOWOsPvt4yK5wNFTn\nfmWB77fc1PuNxV5Yv/+Epo1EYgwL7fmibXWN+yYaa6gh6nGotD8ZFm2eLPimGGqQRdHyIVeeXf0U\nFpDNLjEcryrhCR8W1kOP+y72NfywHfE+g7G2eQ7Pj6CkY1fxeXwaNR2TRpmvuQzbmcVdE9YgIprI\norwHERtXQhuuM1377lTA8y1L9H5HEjAaTiAUJ3+ONAJIaWIIEdpmbjNK9eb6QgIaQJQnlkN7MB6p\nwzv3gYswlvDixIFQYSQhrExDRAdlIahkOh6h7LPt6OvocZNDrDcPdP7d2ePxuAzn8EtCHsmBBGNY\nABBxmgYyCWflwEIwJIoM2rLULBWUYF9+Lg4OOkVbWObj+SUgP0w4lF6v1Yq2xhzP497BqGiL+/Lb\n1fesm/LvFFkSqSFOwByKeF8Ht5O4bw7LICa050Lo25OH2oP3diC/87K+j83j43+M58h6UhYWFhYW\nM4sT9aRai8tERNTrqnVQbohl21YvYm6NLfg+WHqBw1/fMhATYrFU4rFaDGHI1oGT6ld/b4O9sDJ8\ngvNsNuUAACAASURBVCd9IUw4+tWv+mytNap6LVksNHdY8Cxo6UCXdMW8DgK1VDzxSCqw0L529hQR\nEZ05e0HbTnG/TEI9x6P7j4iIaDjSvpoWYqHbLy+vFm3lEvdD4Om9ZCPxOKEPKxWxksBa9qVvKhW1\nFFOxHkMfLGihgvf7PdiO/w1g3HsH7N02XFhslwXcXl/ngiPTOYAlaUdcBj/XazEkDyREpGLJJmAV\n7nfZ607Ae5mrc3QA6fXTQGef+28MKQ+pePujHB5jT9IlehoxSMXjGQO1PBZqeQIL6yMh8gwzJE7I\nPPeAgi6U9wl4Q8aTQpp7LgSHCnRVT7qyG+s1u/Ksj2CuPRKyQLqvUY4VIYW0XT1HQy4/AK+plPIJ\n0yl7u4YkkYOnZ6aiB/1nqN3YZjykJxtKXrp8hYlktar21VAIReMI0g4kbQW4Y0SezJeB9l8acf8l\nkY5bbshlmc5xR54BJDoYYkVYUQ/Jc3m8XGB7BJI64jnaZqIUh73JYwgWHwLrSVlYWFhYzCzsR8rC\nwsLCYmZxouG+koR2HAjxLK9y+OtgslO0ORJOm3Q13hd67DYGmX5HMwndRJGG3YyzuL+jeVeVGofv\nxmV1iecWOFer3tCQQk9c3GEKLrGEqBzIoh51ecE2LOm1OAFfS7WmxImSy2Gk5rK2XX/xaf7hgYtd\n5X1dcPerZXajX/6lT9HUUahFqCu/t88huCDU65Z1cKpUNBRXr/DfMcveE8JKnmkf1iXEhmkjScz9\njiGE8TD6uWsiWm5xOCOIdX6cP82hkJ3Jll6fyevA1BSZM70uEGbGHBIrNeGahYGDUbySECswn8Xw\nB7wpm3dPHF4o73tAQgp5nkVAnBhInt2wD6ogEoobR0B0kJAU5jVFEu6bQAgmFzs2BJJEIovjkXt0\nX4x0xRIW9oGM4kuYb7J0qmgrLbAaS3d9Xc/bYZUPDUAT9eTg56vaB4GoaRAQmNy+kHRg/k0DiQlh\n5fmRvx2XL5Uk0ZHfAark5DzfewMNaY/GRqlH34G9Pj8Dtbo+l66QjUpV7QPP4XfbZKzncHKOEXqe\nPoMtydl0gUDjy989eI8ZoLqEZ94huZ7jONUNo1YRjfU+PgzWk7KwsLCwmFmcqCfV63JmPy7APXxw\nn4iIaoEuZA87vF0aqyUQuvxFHuyrh+SKl5ODBWIW5UKgEy+eY+ugNjdXtFUbQo4AUzmNhNoMjA1H\nLID+lhIYutvs9T3zyvWibWFVVTQMSgF7aXNNpYfW5nmhfYSaV+L/tet6fe1zfG+9Xv/IcU8aoXhG\nE7C0N7fYkzq1rLR0o3mHShIGPljVhdXoQvpAsRgL7BgzdoF6tyORHzgA6nt7mSn7Cxlkyje4vxJH\n23a2eR6dXVAtxlDULHa3dB4FYlEmMXrkPC9ycJEMAaYcAnlEpnLoH9K1OHGsuzx/B6Ciku+yJzo+\nQEub+xntXU8IB2NQapgIwSEFMzUXokGW695OQbAATT4z1EhMMHp0jl5fJm8XD+Z+zef7KD//UtH2\ngcP9uw1WdVvUFg66qtSyUOMDnmvp81WXa81dfZWNJ/wMOR9tpH+iSGU+ZeBJFY+Ci2QFSR0YauqF\nIRa15lRho9fnOZu7+iy4XiTHg2dQJuUA9AuJ2GvxwLWtlPl9s7K6UrSVfCZxuQ6mYwjZCMY8M2k9\nqOEo0aYESBwuGY1UmGtC9kggLyIStZlJpFGsD4P1pCwsLCwsZhb2I2VhYWFhMbM42XCfuLMx5DXc\ne+NnRER0+rwunDZqRqRUFz9lzZC6++DCisuZTtTVrdfZ9b/wqXNF29JlDvdglrwjoYnN+1oi5OE7\nnJs039CQ1rPPPk9ERD+9ca9o29/h8EGtrokIZmFyAtdSFVe9XNK8q1qNQxmVXNtMTtfi3FLR9rMb\nrxER0a2bt2jaqLW4v9bv3C/aIiGsoKKGEd3Nq9pGkkeDi8AVURNJINs9lPBSNtBwZhjIcYBYYxLf\no6ESHboSJiiBIsG8JMF9+vxi0bYnpBgUL819Ca2UdJwiwwCBEOVArsuFheFKtSZtOo88uVbfO7oQ\nfpJ4Z52vL54gM0HUAqCsjCv3lEEoLpSQ3QTCUK6wS1zILzS5fx70syv7YNjIkdCah6qkRd6L7ms4\nTwEs8ufzHCa/CwK4P7pzl4iIDjoa2ntqgce1meu+F2VoapCn543l3icaNspzfu+47nTHyKhGoHqD\nUZ/AMJnpyt6uPgudDpeeKWu0j9pnJCztg8KO6UvIGTThuWQCothCeGqUUHFCQoWhnrde51Cx7+oz\n3R9y/yaY+yg5nSjUbOZiAhH8SBRZsKRHIiG9FMJ9acJtzscQaraelIWFhYXFzOJEPamhKENEQAWd\niGVUW9PF7UrGFkMKFGNXMpbrQHfe7vBC4niklsWV588TEdHFl07DOdgCwHXd3mP2oN77/tvati8e\n0nW1NlKS8hHLsLjomn91u1h+Nk7r9W1NmGzRgPICNfFEfFj0NzpYaawXeOfWQyIi2ritlOppYSLU\n7fsPNdv9/LkL/DcoSOjKAq3rIL1UVAVgnHwpxpeD1V8ycv3Yh2S8MLWmaqJnNsnUsstkLuSekm2M\nRecBrdUTavLdx0plDut8HFgXprHR5Mu00Xj9WJQxlN9IJDAlEtJ0ulb6ZkeuDygR4syTA15dSR7p\nFBe9DSkEVADMs+GB1W901NBJ9GTcykBMMIUkc/CAJ4EQNqDsiluowWif7ohe5TtAN//g/XeIiCiY\nKAGknPL74aqn11cTYkAEkZlEPKkgRRq0eO10lC59kjDkh8N0cylxgeMRm+cfSt7IMzDESMNE3otA\nnPDlPgOgeLsyhiF4/I5EkVzwbCuS6jOINP1nr8t9Xq2d130DJnuFEIlypU/7u1qGKDGlWlIlFhX3\nifqKMqF8qKeSi2dp+uIXwXpSFhYWFhYzC/uRsrCwsLCYWZxouK8ioZb+jrqXq6c5u/zClUtFW7vC\ni6kPbt8t2p7c5UX8+SUNnQUSiovWNL/ozHVWHnBhEdeVUhkOLM7eeZVJEv1ddZ2vf+oy//vZp4u2\n9YccdmuWdN/rrzzFx21BCYg5DkcEVRCJjDikuNnRUJUjVTA9WMRNJWei19PwxpbkYqEI47Tw4AGH\nXlaXtTKyCZQM+prLUZc+ziB8G4grn0CbJ9PKIwjzynECyJnLAqm0DGEek2sSQS5WJMfrxRp6bIlC\nRxUiOg0JOc4vKhGmNs9kl6Gri/KdAYeNUyAczM3zPhjuM/ksPix658eoCUwDqfSBA2E3k8fiQE0K\nEwLEqwwkjIchJ1NqBMV4zX3iXE2kxEmyoHmBFQnRlMraV31Z0PdBKSSWJKwRhGl7MR9va0fHw5G5\n0/D1WtZGPF+Wc1hsl0X2DOKRJt8Lol9FiCud8lC5x5zPiE4fCiOLYOyFM0oe29/l98k7t39atJkq\nuHjcWoXnsyk3Q0SUC1khhH4xP0cTFW82uVVBGSqMp9y//aGSpkLJpwpcqEgt8y4oQ76cDE3Z0/dd\nKOHIGEL4ZtwIQsuuEDr8sq3Ma2FhYWHxNxgn60m12QsK93SxzRUbvV5u6nZN/hJfeuapom3jIWt3\nbWyqxbVa5y/7iy+o53NWtABz0PhLXLYs3r9xu2jbfsieyuql5aLt+mefJSKixqJaAiMhZTSb+oUv\nrbCV7Qa46M/W3+btbb2Wa0y2GCXqGfjGDELCgNC2d7a0wOGeWJYVoIJOC7mwCjxXreC+qIUst3Sc\nQqMr5im9NBA6cA8WfBPxNuqBWmzVBlt+caKWWC/l805AEzET0kulqd5QOpGCcLvqkcddHqeVpnra\nXmrKp6gFGJS5P8sN0GLMmeBSAb2yojAnFm8Ur9YBWnosNF/PO9FH5wg8mT9oVaoiAFioQkbBgnXG\nEUSP0LgeSIIJhPRQa+uYj+tSKK+lxBijdJEeKrbI/TsA7zQTdYlJWef0fmIIUeqZXbzAf68kmnbg\ny3G6EWguioadD+VKEvHgUqDNO/lRz3EaaIhmKJKImg3uy2ZT+7QhijTtls7d13/yQyIiCu7rXDPv\njhzG13f5vdRqauqFISaUSqCzKSoP/a6OR+rwuKFaheFa5A7oCGayXV6F7eQ5gveBI+9cdwjHM9eK\nlHtRbnEIvSYZm2PUa34e1pOysLCwsJhZ2I+UhYWFhcXM4kRjFmUR4QzQPZfcgAzyGhwJsVRqGqa5\nLKG/V7/7w6Lt3cdMfnj+l58r2iZSMiPY1+Mt5HycHinB4tlrV4iIaPGqiv8HdXY/UZhx6QLvE7ZA\n9FSiW/MVdac/eJ3DkY8ebhZtX7zOahWZqwv8JuqSu6o4EaccSstiJSXk4vbmznTLCxAR7XSY8LH1\n6E7R9qmnuf9RXDWJJKcCxHxJwjJzEHYjEXANPV30nsiCehfSInaJwx5eRfetCBFlfk3z1IIeh0KH\nQJzoSXg0GGm4YCQ5eAmQC/Z7vM/eQK9l+4DH+0xLQ7B9CVemsLgbiAgylp8IhezhHlN64SRheDxY\n9sLMxgxsTefIDyV7IOcjl1BSCvsmUqm672tYZksqH5cDDf0MfQ5nlSEs2DzHpJvzFzXfZu3sM3yd\n85oTOfzu94iIaLKrY7kpZKXHN18t2jZW+Dk88PUc/haHe+cONLRsFuVzB/OGJJdtyuG+p6/xM1Ov\no+IMz3EkTniiBgHTlPaklEwOOXklCb8OJvpO2JF7b9W0/xqSe+ZDpenchKiHepJcxF9zCMVlIhfh\n+Rh2M8KxGtZPicOBCeSteUYsuKxtdSFCOJCflcpLMIF3fiplkpKP4SdZT8rCwsLCYmZxop7UilAT\n74FqgcnKjkFnKhW6ogsLf2eeukBEROv3lRq5sSPUV1B52JXF1uUDtQQaKdM02xW1aK587VeIiGj+\nlC7Id0fs0fQdLcsxSdlyDJ/o8TIpotavqucTSPb+lZe0fEd5kT2C3V0tCzEUym0digeWPLZUgAla\nZKn3BkoZnRb+n299h4iITs0rrbXV4HvZ2VIFjKFc27kzSj5piveLVnomVlSnq/sm4nz5UOzu7KkX\n+bhQ7PKJpCEkA7XiGqKhV6rpuB/0+HdWVS9sLAvmKZTg6GzxGL/9vnqJY0lNiIFcYLx55DIn4iUm\nCao8GMLBR2fKf5IoFwUEkWbM/XzIQzqmGqNh/WfgEppfOagKHIimYQLah7WrLxAR0fWv/62ibUHS\nSNy6zpeSLKhjHCBJ+fnfBQ/40uc+T0REXzp3pWi78cMfERHRf/9jjZp8TyjRzabqZX7lIntm+QMd\ny3SHyUcp3JubGS9xup7U2ipHaQJU3fAMrV/nkAksIUXeEy82ApWWQApdNir6mh5IgdYs0/eT8ci2\nOkriKklZIxc8uEQ8stDRNkeuK0vg3eYLrR/c8YGUzknKOl8C6ecylI8JfYlOoLqJ/JthuRfxGBOy\nRQ8tLCwsLP4Gw36kLCwsLCxmFica7uvvc3ho0FdighG27HY0/8GQBpbPKqnBrbD7+9wXPlW0PT9m\nhQgP8nRGUo11BSr9Vk3oZk8XWDfu3JZ9VYi2Kdx/DwQSJ5JNH3Y0BBX6vN3OYw3jXRER2Qlw/8ey\nSO9D1daDAS/wT6DkwOocHw/FFf2Qh+L0quY/TAtvP+BrPH1Oy520peyIB5V0a1cuEhFRs6lh1N4B\n98lkDMoAskC6M4ayF2XeZ25Ox7he5xDRcPde0eZ73Ievv/ZG0ba7y2GMC6d1AX4iagY+5Cs1JU+l\nB+HWvZGUsyANFWYyFhswL+dEwaJyyGyT6welYnNvWT7dsq+l3CgroEis5ETlKAQqYp5IiBAZitzB\nfDQ+3hBIErVLV4mIaPEFfeZKF1gZZsvXsNvP3mMC09aGhnNH+/wc9npaCqezzyGk/YGGkj7z+c8Q\nEdEv/bOvFm31L/F1vfqFzxdtv/+Xf0JERLsHG0XbUoPH/3PnrhZtwy6PoRvrs+6bhf98uiFZT8gK\nDoTJTGXjGEgIpjI3COLQ6gqHUG+CykMy5n5ZXNSSPmvLIuhc03GrSw7iJNWw6ijifgkg/GuIE0Gg\nZJRU8hKx8jJJKNHDKuYJHy+LMAzPz162r+MbxfxuKwX6jtCUKB2PkRDo+iObJ2VhYWFh8TcYJ+pJ\nObJ4t3pG6cQT0WlLY11ijUZshe9t6MLf8oWzRETUXgAdtg5f7uSBKjWcLrFVELuq8hA5/LU/taYW\nQyzlMeKHav1tR2JlQ8mBRpUtAKORRUTkixqBC6oRTVlA3NlRNY3oHv/O59Uaqkq5Ag9NdFlYxSJ0\nF6+zl3Lx3BmaNlYW2EItgR7YpnioAVh79TnukwlYU6Z8RlDRvtk74D6ewPRaFcJE6KtH033MpUGi\njpZtmBMxvuuXLxdtb8r5Fla1bwytegLlXQIpmjnaVmWKA5lbEazoT8x/wFKsSpG2EpYnkBXpSYR6\ndGJlTllxwpe+jNCsNI4elA0xxA4sUrEvJI8ALPxYFs9b1zWdIz7Hff7jbZ3T+/eYzJCFOm437nxA\nRET35V8iopro7y3NK5FlfYcJSWNYqP/SV75CRESDIRTJrHH04Mt/9x8WbT+4eYOIiO490HPcEKp6\nqQKqB2XRsgP9x7YopyRTFu+byDuGoJCf4XOkUFLGEAiQSLAsz+CFMxeLtrv33iMiIh9Gc/ks369T\n1mhNLioubYhwbEtBRSfTZ9UojrgQ6UmMOgcQMVIScgbQYAzDP4UKh64o0KRApjgQVZCqq/NlFPN2\n/bFeS0+2Gw7VC/swWE/KwsLCwmJmYT9SFhYWFhYzi5NVnGhz+Cjc1VBQpcmhsNCHkgPye++xLpIu\nr/ECewry88kBu5rxnrqIWxmHDQKQfDe5O2UohlsVocfxEMpHDNitzUHksN9nskc/0EVIbyjuMUrS\n11kg82xLw5GZXMvtdx8Vbe1VzimagNhqfyQipdD9FVFxiHINg0wLX3qBc70aVb2/V9+8RUREz1xT\nMsVKJCFTqCg6lnspgahmWXKsVkFUc35+UfZVl/9Awn3pQMNLrUXur8WLZ4u2xdMcLm6AyOnBARNv\nwlAHeXeTw8UOLBarcKzeb1VKyGCuky9xzXodlEZkUTfCUIiEbYIpl+xw5BlxgAxgFrZx/ppcqAxI\nEiMJV2aQqxNIOLXT0LD2jbe5avX+npKa5he475N53S6Vee5BeZyBKFMsVfR5CFocfrr+3ItF2+d+\nhcN9Y5hDfp+v+YWXlTjx1a//GhER/d7//i+LtlzCvm++/27R1pR+WQKxYKNmUPGmS5w4EOUa55Aq\niITYYDvD6wqg/yoiOvuFz32uaGvIO21nV1Vt3n6d8whrbb3fM2f5eQtKEFLMJN8TzuELucyFd1Hx\n/EBfpbIMQlB+xzFlYSC3sC+5U2FJlX26e/z+HKV6fZOYf2NporEI4GI48sNgPSkLCwsLi5nFiXpS\npmAeakCZ5H0skpfKwp9fVUt5eMBf5HJLFwN90Yf7pa99pWj70auvERHR9159vWh7/ipTVFfauojb\n2xHNq7ZahGdWWG9sBGUmdvdlsRcWdk0luc1d9fSqTbZAzl/R8iLOmO/pIhT2u9fZkmtXpYXBmI99\n731dFL77HluHaxd+maaNi/Ps8axvKeGgsL5JPRVXLPEQ6P5DYusIVTbqC2xZ1UCRIBANwJKvx2uf\nZyLE7iZY+KIu4YNOoi/U8jjRMWk16nJNYM2L67x2StMMuiK8WAYv0dBoo7GOe0VIIadPKzmje8Dz\n98ETtWQNkGY8DRSVaIAA4hbqF9oWiScVNfS5mV/mexpP1HvZX+K5/+rdB0WbiW7Mg9e0uMhz41Gq\nz3AkJJMGlG1wRQdz8dyFou1rL79CRETf+ObfKdqWzrC2XwRlPnwZt/FYIyRhi+fB888pHX7j/ZtE\nRLQz0nHrtzmi8dzzn9FzyGL83s9+TNNEKhRwD7xYxzmaKlEWD6kEGpixED9aczpPv/aNLxER0bvv\n3izadr73XSIiSvo6ls0Sky7STNVqHKF7w+NGZUnrCYFgYbJgkAc0OuDnLIJXoClX4wLZYyCFKT3Q\nKpy4vNOoryQ4Svi9jpGjOSFaBR5c4IfAelIWFhYWFjML+5GysLCwsJhZnGi4LxqzC1urgnspgoIZ\nyLtXJKRXrWlmtan0mcGi8OMuKyNcBVHRz77wMhERvfq6usRDCVVVqhq2KBt5fIiNPJEwDrrd5y9c\nICKiHMqLBLLQeBYUCtYf8763b+p5rz33EhERXZ5/tmjr/JDd3g6QPWLJe9jtanZ+q80u+yXID5oW\n6uLCr0GIaFNIKkMQBx6LqgRmjpvSK519JT94Mp4LMO7lMrv8vY6GBUPJsfIgpyIa8diV5iAXaywE\nFywdIIvTKOa53OZF+wyqNBvB3iGEkjYlt60S6hhXpaRCGcodNM8x2ePRNqoo8O/FBlYZPXnkRlgW\n5qWuz+v9xhLS6cxp5dv5qyzmOoSqyLd3uV9Wnn6+aHsoeTmw5l1UbB1C3z/7HO/zzW9+s2i7evEC\nERGdPqPh0vllJj9hKRFTFoYCffWYEjD/6+/+btH2V3/w+0RE9Pyy5g2NEz7OHii1PPMMX8sXf+Vv\nF23+Bj+b37vxJk0TJt2yEmoHlkOeJ2WIu/ly75g7tbfH77atzcdF2zPPXCMiotMXVKXl16u/SkRE\nnT0VxW6I+kTu6Huxs885ZXmmz6/J8ctjjeOZqtyHKCbyjkSFEsdUhnZ1y7GE0qNUj+dWTEKVHm7O\nlxA+JvkJgWoASy0fButJWVhYWFjMLE7Uk/LEQq/CwlpzgX9PQBMuDEW94aEqD9SW2Co+eKJtZbFQ\nfnhDKai//CIvzv6D3/gHRduje/eIiCidqPVnaNG43t2oi0UDNMgnjzbkmkDrTRaK/YpaSCtn2evr\n7qp3tbPB1PPbXaXwrq1d4Gtav1e05Q22Ss5d1wJx924ytXTjkZIXpgVDp25X1DsoV5j8MN9Qeqmx\n5oNQt2vNcZ/cX9dx6g64T54CCvrNt35GREQ766r48exVpr67oCXW7/D9bx3cKNoc0Zer1/RajB5k\nCp52T2it7z/Rc9y9z8SAjV0dk5FY4m6u1m2WHaUrl4Qg0lxUzcCH2+wZh4OPzpT/JGGIGlhsMZZx\ni2HVe6/EVuvPRjqnBzfeJyKiypz2X3OerfMDuI/7GzyGUK+OKnvs+Qz2dJ7/s/+SlSF+67d/u2ib\nGAUZUHkY9GQBfqLWvC9/9h0dtz/+/X9DREQ//N/+tZ53lz2LUU8vZm2Vn5e10y8VbZ/70teIiGh5\nea1oC2s8n0pzWlJmGmiINx7CeASiWuIDwScUunwF6P8t0cocTpT8sLDK75jri9eKtnff4DSB1SWN\nOt16j9NFLlxUclYo5TbWD7SsiUlPMPRvIiJP3r2HPSl2CQModJl4PE+wRAi5kqIBKhSBaHQSlLcp\ni/cfD5WC3tlib3e/pxGYD4P1pCwsLCwsZhb2I2VhYWFhMbM40XBfVRbLExDAbIvygAuhuHHE4YCt\nJ6DUIGGBJFb3t3KK3fdOoC7n99/g/Khf/4ZWDs2FsPHgg9tFW0kW8SdA/j8lZTFKJe2G/R4v5JUh\npOVIjsjmnobi0hJ/3ys1UCgYcEgphkX6b7/OoZZ7Q72Peovd/da8hhTPPMULzosrKsY7LVQlhwkr\nme51+XodV0NdJVHtiFIoAyFZ52NYWH94m8fx+WdUaaDfYbd+saGLu/OLHNJ9dOdh0fbaW7zY3VpR\n5YLdLQ79rEBV3x3JwXsAYrJdIXk8eQQlJCTfrVzVcSJZ/G3VQahUFrGbTc3toirPgTaUSojSd/hc\nkYY4pgIJo2H1iVgUM7KWhvHWXvkCERH9bHO3aOtt8FhGXUh8CXme33n/naJpImSZHMJViy0mYARt\nDf20WvzcrG9qCLUjJTpGI+0XE9FrQz5VXfKpsJzw6irntWFO1HCPCTbLF7Usx+I1Dg83l3RuGHGR\nXl+fr7aMddbWfpkGSh6ftwLVcJtS0mehrde8KkLJ7QUty1Ot8jthcUXb3r3NIfLV0xq2XFjmeyr7\nOk/ffofJW6inWxFVFW+kYbcoMxV3FblMKIgiky8qFBmqFAvpJkqg7JIQyuJM55VED2l0oO/A7V3+\nHe1puG8keWGO+9H5htaTsrCwsLCYWZyoJ1WZYwvK1LMnInJdthie3L9btEVV/ppmvn5VNx+yNX7m\nvHoW0Ygt5fkzalnc/AEXx6t9+7tF20vPs/U1hsz0UDypxVW15KMhW/cRWMWL8+w5ZA5S1ZlMkSKF\nMuK/J7BdoRkGooEPt9iqd8Fq6uwyfTQGCvqnv8xKE6uL0/ekzKJuFxc2O2yJL070uiPDOqm2j+zb\nmlNL8d/90beJiOjq+etF2+WLTINOB2p9d0XdY6+j2elzoon45V/61aLt4ftMjX731ntF25MdPs7t\nbaW0R0LtT1Lt/9V5tjwroMm3vv//tvdmsZadV3rY2tOZp3vuPFTdWwOrOIqkJIqaW1K3291oG4bt\nthE7CQzDMNBwEAR+D/KapyB5ChwYSRAgAQynYchq2G71YFkkRVGixElkjayqO8/3nvnsc/bZQx7W\n+vdat+tSrLR4Tx8Z63tgXf5n2sO/9/7Xt771LfxMweMxI4lxxOqxtoyRWyvky8So71sDFgOMBUYg\nEjKL0KcI79W/91+lY7kvovfbD/7ff5OOdR9itBkLybNHc7QrSgdGdG4yeb5GCpQIn55nablD4owD\n4TLSJfcBUZ0AUyQMGEb8u+0DnPOlIrMIL38LxQ+ZCgsJtvdQiu3VeGxoHDaE/2NMrEUsHDG2SMB0\n4HN0NQ5cv4buM/MzfH+apRYclQpHda6L824w5AjEzLuXXmTnjI83kYW59fHddKxCU7FY4+jetMnZ\n3ucWRovLeHxNM1UAgEEqVuOTZIQQthgzXqqOiHIcUrxEfSG6cPGzo4A/O6TX+0JQZp/g+7yRbNiJ\n0ablyrjufGgkpVAoFIqJhT6kFAqFQjGxuFi6j5J3nQFTFI/uopihJ0QIxTyGriIahJ6P4aKT7Iul\nLAAAIABJREFUYUrm4TrWvLRPOIxffgFppH//50z3dYZIW3zpc1xNP6SukAWRQM9Q5XerybSboRTz\nwi3B9jB0zuaFSwYll2UbB9OxVtIbl66ig0RXdMNs2VSXNMchO1Cy9WDACe9xwbS2KOQ5OX75Mrbo\nyDmcBDZGwXZGtIagfTU0LgDA9g6e23/xf/2rdOxv/g6aAs9UmUrKHyAd29rh4w8d/I32I667Wq4i\nZXJU5M/eeUTUhrD/r88TzVLkpHKe2ATPYlrBIXq3I6iuaBb3XZrnlqj9yKKgl+tzSEceiS7S44AR\ntQxFR+uV30Cx0Cv/+A/Ssbc3keqqzHLdkFdE2ihJRIdh6mjc7whKLMFjPxoyTX6fuu9euv5MOmZT\n+5NBKExniYLLi+7OvS5+z5/8u++lYx98iIbQs/M89//6b/8eAABcu8Fdgt15pFo7TaYU+0Ok9oZC\nmGR0UP02b/OPXvtPAACwvcvuDePAl175KgCcnUO2EYgI94Y+iaje+ukb6Vji4vGrTvM12BrgHGu0\nWAg0X0DasNlm2typkCtIn49Vj8yYXZtv8RmygUiEViEmdYstBr348fYio/R1vo4SatsRh3xfTKi9\nTUlQ6UOHjHdFh2EnMUIg4Zj8CdBISqFQKBQTiwuNpLLkFLB3yBLjjTvoFvHCF3nV5JCUsXPCT+RS\nFROmsmWGkXFubm+kY4s3sQr9yufZL+/jDVxNXr3CDfuureL7BsJ/LyTN5twCt3bYpe9utHmFmaGk\nYiicKRqn+Hq2wKsmsypIRHI7k8UVSK/NEdIKbdfqs+zTt9PAKLE7GH/TwxxFDKIHIPjU+E6uUEck\nL46AI5DWEYpKNre4fMCIKY4bHCH963/7fQDgynoAgHk6n7PCrt+mqLbf5dVyZQ4FOEc9PidxFueM\nbBJ5RNF5ItQPJpJaFC1aZmgbEvG7I4pQOl2OzGYpsV3I8WUyRW1NGsLVYhzoj0hcVGCBSn4VnQi+\n/5N30rH9Fq6ma0LynM3hqtaKeN/2aJ4PAj6/aTuVHLMIBZK3S5cRm45vIMo5QtMMUrAhf/Td7wIA\nwP/9f/zLdCyx8H2Wy+tj40byT/+bf56O3biJ9wdLrL6N4MgX4psRNcx8/c++n4598NabAABQf4Kk\n/GcJjqA4KjECLNfl/egHOI9f/8mfpGMnTfIRrYpmlSHuZ6HEx35AkWV/yNdWLyYfP5cv4P1D8rsc\ncqSSKeJ8F0EdRKZOIBKtcSiSisX59Sm6DgRLFJrvHvD+ZogOqwgRTCfA/fA7fE2bKiI7+vQ4SSMp\nhUKhUEws9CGlUCgUionFhdJ9rSaGed02h6Ylqqy2RMIsm8WwvF4TtSwnSE30RA3T2nWkyapzXKfz\n4B4mdp9eY+rMJmNESQX1yRmhUuAEfyfEkDgYcc1LgeoZjpvCtYDaS8gQtuBRclEYZU6VqB1FxBRK\nkWirmqhCr1KC/2jIyfduSFRW8umdKj9rOKa1wEAk1qkOyBL0TfcEz2NcZdqzTWa6J0fcvfa5K2he\nWp1mwcE2GQUfnzJVuEHmpkNhQDxLlFM/xz98ZwupqQdCrGCRm0lbbJ9pDZMID8wjouxGokZoeRrn\njzT9HBH1+/Ahd6qdmcPkvVXheTlVxrklulmMBQFtf26O5+Ab72KN4B/97/9POva5z6PLx/UX2b0h\nS1RdKGq7+kSZuYIadchU+fkvfCkdW72OtW75HCf0HfpMKOuuqGXK0SHX6nz/3yHdlxMcYJ3Mev2A\nqccH97EO6N/+IQtt/tbf+QcAANARwpiTJomtIt6PH//gTwEA4P2fvJmO5RK8Z+SLPK/Ggch+nF40\nBsjNJlOUj9Yx/RELii2bRQraFZx7t4X7e3rKbTnCANMGIfBxsYhKzMTs7NE7wGM0FMKY5av4uieN\nlR28loWmBiyilmXawqI5VMzwZz26f4Z9/rBFFGA2LwQbs3i97Yn5F9H933HVcUKhUCgUv8a40Eiq\n36XEn2gq+NXf+g4AADz9zNV0bOsEo6HtOV5Z+PdwpeD7LHToUHJ7tsR+cicxrjZuf8jtO75JHmAz\nwputQ9b/lTonlC2ymG8JcQZYeEhEby8okv+W8SIE4ORtNitWJRauFPpZXjUVevhFVxdZnHHi4vsa\nLZbhe9QmI/RFGDAmhLTaazVkxIurUE+snDrUsM4V/f4SkqSuiWZ3N1ZxbG+X9y9HbTuemeYGbg6J\nShLhIFCj9x2KsoCPtkic0eQINUnwdVs0PfRILi9lt21K9HZPWLjSpRXdXI53pLCC23V8wqvWR3dw\nhX/lWZ6ry1O4KrwrFQJjQETNQgfCJ21zex0AAFyxgu/QyjkjzluNPOzu3VpPx0YkH8+IUotCHWXh\n5SpfXz2KdutTPDY393gLDBOR3fvo3XSs2cJjOVVhsUyDmvWFojVKhaLTD99lAciNp1DyvnCJj73Z\np4d32YHh7i1sXZETF+wseUwW88KvcQzoDPFedXTIEf+jdXTW2djgCL1Lc7uUZzeXPLl8mCaTAACn\nMZ7L9UcsFAsz+N1ORjBRDh6/uRJfW7N0Lu8d8H3xww8xgquviPIaKquRjRorOYzWs3mO2o3GKAqE\n8My0/OjyPcuhNjixx/tRyGNZQln4YjZODKOijhMKhUKh+DWGPqQUCoVCMbG4ULqvvojU2uIN7iz5\n0g2sV5qa4VCyUsdQMyOa0rolpIJODjisjSn83VxnN4JaAb/Hm+NQ95CMJS8J5wGHrOajAQsxwsDU\n/QgXCnKSyFiyXgG3YVH+BukquqJ2p0mV5ANRve038bNHPtcRJWQia4n2FllK8tpZHhsXTshMtiHo\nvpUlpO+qNRapbJCYpCm68K5eQcHK7BXuMny8ge0fdu5yG4jVKh47R7R9NTTwyOFj3aZaClmjUa8h\nLdJP+DyN6PVhIOYHJXz7spbDRZ7CzvBvHBCFNV9h2sMiqvDokPctGeL25wq8zfM1pL1uXGcaahzo\n0npyKGpN4lmc01cucz1gRHU5iaBR8lQHJ7sYm3YMlWl2fqjPr9FnxfylLssrl5jOtUlw0u/ztlj0\nuwcHLKBxydGlKNqzmC7d3R7TRi2iADsdpmQ/vos03uLqmvgN3Kct6rwNABCSse1UjumqnBGDROOt\nk/o330VT34P9/XRsQF2JZednl3pgxAPRPbmPKYLBQJhi23jeLk1fT8ceHePcHojWP3kSbJVn+Ppw\nqR5tcUWkRigLYYsaNVP+5gm6L0N1crbH10cM1PImJ5wpinh8T/b5XBqxRb/L+2Ho96k630sCum67\nom71k6CRlEKhUCgmFhcaSfnUhG67yx5awQhXWqtXrqRjK9To6+bSzXTMoadvPsOJ7OEQn9LDDksZ\n201cjXzuJkdrOZKZNw95ZTZL9vjboknezjG+nmQ44rq6gFFOucAiCYtWZr5Y3bs2tToQEs+QBADz\nZdFKpI++aR+tP0zHrqyREEOsXkbkGbi1zgnWccGmtcrinHApsHFfpVNGlqLL1ilHXAeA25u5zF5x\npSWUbq++9Ll0bG4KV+ynu5xU3t/CcyF9vqp5qnYv8orNprYSJZvFCm3ynjsWK/K+iaqEVyRQe4K8\nxefTJQHMSBz/PTqPh8csegli3IbBe5x8vkzOJauXLsE4cUhy5WAoWnWQ3DfJcsQQk+vAQMh9yWIN\nAhG5u5TMrs3weVtZw+hwRogkLIrI8jlRHrKHMnPj3QYAkCURSiR6ddjkV+kIZ49KDUUcYczzICQn\nE7/Dx37jEV4313dZNNCl+8nOFl8jI0rej0QTxT751kFmvMKJD99F0YgtmAGHos6REAcNqCVOKBpE\neiQjz3r82ZyJdmc4ii2X8PidnnDEn/NIRi5YhR7gfTZT5O8rxNQewxOlA3QNeJ44R+RW4uWY7Wp3\n8T4w9Hle5cltZ2aZBSCdDRREJCO+V3Y7OFar87yqTuN3d/oaSSkUCoXi1xj6kFIoFArFxOJC6b6T\nPaRzQpGwvXUHw/crB0wBfvUrrwAAwEyNE3Wr0xjiOoLi2aLE/aVnmE473EY3iI/vv52O1eqYpK8I\nCsAUrm9usoDh7gbWDcwJZ4SZAoa9szUOTaeow/DWHtMMFaIDa9PccbPXw3qFoxZTlKckrGi1REsE\nSjL7oqJ7/yG2MMnH4032InCtkggHBtMFFUSLi2miagqi3mH7GM/Jj9/kY/OFL2N30dBluuXnH90C\nAICSsLAIHfwN2bKkQEldpyVaApiOrIkQWFCNW7XMTghxmtDnavw+iSSKRemYgNN+FHDif9hDemJ+\nlimOZTIenl9iSuzWrY8AAGBRJIHHgbahgaRDBJkRhyWeR4mF9E2/y7RMiWiWpStMp1dmkdJ56ia3\n4Lh5E02aVxa4O7TJsUsj5SxRREks3ALoHBULLJKwLbyWIrEWXlzGYzo7zyKk2x98AAAAPSEa2N9H\nOuverQ/SsR6d1yNx7wioI283EHVrpoYtM941eEgtRIYiLWAEPlI4kSPBUD7P20caE7BF+5MRCVM6\nvmhNMqSWQ0Jf1TpC+r2R4d/NzeL9KVdkGi9L08QHphkjEnFIsYzj4PbJrr7g4PU4AJ5XxqlHlIqm\nHbDjFu/viN7X7QrT7gLeQ4qlT3cF0UhKoVAoFBOLi3WcoJVeJcur2PvrmDDdfMRS1W4Lk2evfO3Z\ndKxOlf0LMyyvLVIF9GZjPR2LV/CJ3M1xAq7dwwgpFMneDq36/Ble6bkuJr8bQi4ZmgWZiMLaJM2e\nnucVpt/FZGBDNM6zKfm4c8rijHc+xorzmZfYW9DI27fvcVRXyuNnM8n4JeghRXSJEDAcNKgViVjG\nXKniObFFtFfO4oqtEXLDtfU76wAAMCWcCbZ7mKwNxeI7R4l1WxiH2SQQmHJY6HAake9ikVfzdQ+j\n20is5gfk3TcQgghrCt9XqXKEFFGjyp4vHCxonzyHL4kyrUKLQrJbpO+OReO9sSBL2yW83bwebn9l\nRfhR0vwN2nw+ThskfhGmhj5Jf+/d4TKB/S28bkyzRwAAj86RlxftVGjVHQsmwIy1jtnzMqbkeUas\nyO/fu4fvFwzJ4SHeCwKRbG9Tsv3tN7iZ6YD8/gIRWRj/RVn2Ydbe0nlkHDhNRVl8jrIkfigIB5Cs\nR8yAzdd6SHNxKFrj+NQuSLbL8Sg6rAvnnJiEQMddjpAGFMnkLOFMMTIth+RW4+t94ZG462Okmq/z\n7w4TfH0oZPMWnX/5fR6xZslQ3sdIeNbh7zPVAcVyET4NGkkpFAqFYmKhDymFQqFQTCwuNB7OG4PH\niMN4OzSV6Swu+LPvvQEAAJUaUwBPvYBV1gWXTWJXyphgzwozybsxUmYW57YhM8RYMhmKBGaOaphm\nmIKaC/FDvVOuz+jQZ0oJJ/n6ASb+XEF5FKn1RkPQgo+2sBbqzvp93pgCmT8uca3DBz98CwAAfuOL\nr6Rjr3zjKwAA8Pp/5G6d44LpzBsIUUODnA1qeaaSTI1EW7ReMcnQqRyH7dYIj8mDW2wEWs1ignRV\nuHb0e/g9iTBNjRNyHxG02xQdw8DjMY8EAr0WUwiGpHLLgq6izxRE3Ztp2xGI8xlRYjsWlFiXDJIf\n3mYKa55qiNaEuGAcMDVHOY/3o0e00v7mx+mYT/Nyd4tru/YPcPu7UrxDtTxSpmOcJM5Y55o54fA8\nsImutsXcN/S4BSLxT6KBq5fZXNkiA+fjY66/W17COXH7Njs1JBGZPzeZOjcuGrYQ86TOGqI2Kab7\nQ2J9ehuIzxJOBn8vY/OxMnS5HbLgYEi0dDDkFIVPLW+CDo9ZVPfkCiqzkEXxkm0L5we6RkvCkaWY\nxb+jQ1GHRCIiV9CRcYZcd4Sx7bGFn/EafI/O0TWVFR2GrQjnWtBjqtBvk5hiwOcoR9saC+qxF+J1\n6xWV7lMoFArFrzEuNJLyqNpZ9EYDr46r4tUpXolu3cbE6Rt/+l46Vqjgk7tQFKsDkmzOVdk3zSvg\nynbjmFeT7T61NcjzqrjRQsFGJ+BK98EhJmcLPeEJF2NCspnn1UaGGpIFAY81urjK2OmycOI0Q/6A\nZf6+hWmSpa9z5bxL33P5OssvHRdXlrUSJ/jHhcMDlPtmi/zbs3T8F2ZYih9Qdb8nIq4pajEAQlyQ\nJa82sbiFLCWxc2IVbJZIiRADDGgl7or1U57cDKxYVO3TcR/1eIVaIWl8Tkh7LYoOcnIFSOIHXyR3\njf5iJJpx+oCrvWkhupihavxSZrzNKT3ydswIWX9M4oPhkBPm+xQtdUNeGZvk/dwSR7HGvSGMhZ+f\nfSaGItD5iqWLByXHhVza+ALKqDihY/nhhywjv3kDZe6LQoK+uYnMw0CIUazH/mBxhhyzqIzBFk4N\nGZLLW+541+AFiv6l28KAZOR+h+8TAbUfSkI+b2CcOoQKwXRgiUXEakRLtjgdxhWkKAVDpjyhJUQm\nNLVlSVBMrYGkdx8AjsVDZilicvaJRURtEfsgGauki+8TUwNGdA1KbcuAfFMHZ1Uc50IjKYVCoVBM\nLPQhpVAoFIqJxYXSfUmMoWbzlJN3ezuYCH3mS2vpWNDDcLV5wuHlD76PDhKhiGuDGxivLgkjxWnq\nInpz/rl0rEFU0GGfk64OhbAFm2u2hhnqWPrOLd6+I0wyLy5zXdPpQ+wcHAw4QWgR55CfY8eJy89i\nRf+UaJ1gquilPf70Ioo3kjzvR5MSps02/8a4UCBhQqXEFJYx2M1kmV46bVDdi8vTxqHasDjhsD0h\nU9eZKidF8/Q+byR4CjokXSGsOaZWKqEwiS2TwWwsKCzHpeNf5e1LDPUjRBcWJW1lEj1nksqChjYt\nLkLR3qFAFFucCMEGUStBf7x1UkOiS6UzQGAo1DrPwcVlFBpZOT6XJeqMm4i6pvVHWL/nC4otl8d5\nYEtXC6Ji7RG/LyKz1FhQNeHQOCswhTUiSqcvzG5vU1ddS3ZPJipe0kGu6eYdiXW0OYdiyFCU0h3B\nM+d3zOYtHRKo+G0WYo36eP3L1jOkr0jdVQAAbLqmAiFliYkaly1WEjIJjn1xL6J7pBQ1tJu4DY6Y\n4y4dEEfMA5doPsnCG+FT1BP1iz3c6EAIQMyvebYUU+D7IpHjGRpRhs1zMiTaUpoefxI0klIoFArF\nxOJCI6nmIVa93/n5vXTMp0S3I9wgpi+hk0EgrOu376OQ4C1gMYWRWrZnWRpZaeAqcmmOxRS1ElrH\nZ8RKpQDkyVdkW/nZNRJxVNmF4oc//hkAADzqsxz2uIcy9+ka69yXL2PLhpUVHru0hA4Wxydc7d9N\nva6ES0MZ93cYC3loRFL1ZV5xjQtZkqCXCizkcEk72/Z5G7dJet5uspR5pojHriKiJmeInz0QbT4K\nlMzOSt0EuUuMXF5hBbRib4rVaBLisSlk2XHCyOZHYjVvmuJlhJlYQklnV0R/pkGfI5QdA0p2l8Rv\nlGiOBkI04JD8OgnHe55C2q4EOHHtTmFEPr/CDScLcyj3HonlZ48cGppC9p0hkUxpiksyPIq+Eilv\npmPkCkFJYhLmoi3HiKKlYMDzxTTwE2+DbIaYDDEYkctKMOJ5ZYImx5bn0nxUtJqgRH5eJP5NdJVE\n43Vv2SVxlC2iiCwdS0dsc5ZYhUS0XQmo7Urs8rGP6LOhKIuISGxhxY83AHXzzBLZMYnWhuK8EXVg\nnSkdoGMU8phDr8exiJCMx584lzFFQ3K+EGEFIviDIe27EbkAACRgvk+FEwqFQqH4NYY+pBQKhUIx\nsbCSZMzZRYVCoVAonhAaSSkUCoViYqEPKYVCoVBMLPQhpVAoFIqJhT6kFAqFQjGx0IeUQqFQKCYW\n+pBSKBQKxcRCH1IKhUKhmFjoQ0qhUCgUEwt9SCkUCoViYqEPKYVCoVBMLPQhpVAoFIqJhT6kFAqF\nQjGx0IeUQqFQKCYW+pBSKBQKxcRCH1IKhUKhmFjoQ0qhUCgUEwt9SCkUCoViYqEPKYVCoVBMLPQh\npVAoFIqJhT6kFAqFQjGx0IeUQqFQKCYW+pBSKBQKxcRCH1IKhUKhmFjoQ0qhUCgUEwt9SCkUCoVi\nYqEPKYVCoVBMLPQhpVAoFIqJhXuRXz4/vZAAAIS2l445oz4AADy/NpeO/ZPf/10AAFio8TNzBAH+\nO8qmY4NhBAAAtt1PxypZB/+Ik3TM/JnwEJg/LYd/I4pwtNkapmP3tw4AAODBwWk6Vp1bwm2P+bPP\nPbUCAADPXJ1Pxyy/AwAAGctKx0LaCMvmMcfC70mimLcvimjbeeyrf/A/8ocuEPfufJAAAAyHfBxG\n0AMAgCDw07Hp8jIAAIQBf3Y4xPedNg7SsTgJAQCg0+Q3DpoDAABwck46dhLhZ90Cz492B4+hI87T\nwsIs/ju7kI6t0DlxgT8bxjgvwoj3o1ioAgBAkkTpWDDE7dra30jHWp0mAACUiqV0rFwuAADA5v5u\nOuYPcD+sXjoE//U//GcXfp7+y//+f03+4phlPf6zNo3JeX7e+z360xHzzbxs27YYw79tR/48/p2c\n2aJE/BcR02dzYtAD/JFAbEtkjQAAwIWQv40u4iSW+4jfF8nrmq6rMOLzG5nrSnz0X/4P/+3FX0vJ\nCO93cSwHAQAgPuftZ86QeYM4qPE5W8znRh5pi75PfOCXhR+x/POxaQW2/WSHKv0ace+98/4vAADg\npMn3zy9//WsAAOB5GfHp9I7Mv2t55/6wRlIKhUKhmFhcaCSVLs3kkovGThrtdKjXx5Vv5TKvlHtB\nFwAARgk/fYOQVmFiiRGNcBVWKeTSMcfB1XosVjRRiKu02OPIzM7hqjlf4PcVfXw9OBikY482MUpY\nm5tJx5aXcCVfKhZ411zcrowlIkKbIiSbj4FFf8Yhr/4Ss33J4yubi4ZXwOPl5nlfEsDjORqO0rE4\nxONULHG0EY0w0uq3u+mYlcf9r89O8WcLOCYjxct1/J5sgc+J4+CULNR4Wwol3JaMI6arCcjEijwe\n4W8MfY7WkgSjJld8NkPR1dzUSjpWr9ZxfyI+78VyEQAAGqeddKyWrQEAQGW2DOOETdtviZWv9UtW\nvDIa4g+IVSt91rZ++fvMfDx7CZuQS4zR/1gWvzFL2+rK24z5PfG+hK7xOOao2HPwesh4PP9KBfye\nGp0rAICQJsL69n461qNAOrEv9vb2F2EiqCjiiPDe7bsAADAzN5uOTc8hixQEzDS8/5OfAgDAwGfm\n4tXf+CYAALgu70dI3+2Kgx9b9LtiWywTXZ0bNT0ehUnE5v4q51d8zn2JNsEO+Uc2bt8DAIAHDx+m\nY5//8qsAwNf22W0Qc9KBc6GRlEKhUCgmFvqQUigUCsXEYizxsC0pCgr5mn0Oa3dPMMn2orWYjgUU\nQvYD/uzIQZoGirV0rNXZAwCAYZ/fV6vm6cdEUtjEvVmmlqwsvm9K0FfPFzEs7wz5+376858DAEBO\nUIpTs9MAAOBk+H0u0X1nc8xIV8QWUwCG6Ugskey1iKIUVMG4UCogfZLIpC2lRa2ipCmJznTz6dhg\niPRYMOKw3R8RjRHxOR76SN8V8rwu8nv4vl6Xj8PCPJ5bL+Fj3WsS9VNj6tfv4mcTl3/DsCJRwsfQ\n7+H2FbPFdCxPmfyZOtNGcVwBAIC2f8jbPEKab7rG78s4uO/OOTTJRSJl5c5k4I24QLzPvGLJN1pn\nXgNgGu8c7cXZ5D1RWOe974zAwtCHUjRA5yE8k9DHv2ObaTwjjohjvh0ZoUYxw59dW8Trf3aGxUrr\n23S+pJonIWHSObt2kRB3m/Sv4z2kIbOCepxdQLpve+tROnb3HbzHSErsmeeeBwCAUr2Sjtl5nH/t\nIxYmZPJ4Xbji/pRuw3nMnqDxRgO6fgSd5xWKtD/i2jdzLZTzCv+2M8zT1et4jg63eD+6rQYAAOQX\nlsRHn/zsaCSlUCgUionFWCKpM8leWhIORMCwfXACAAA9n6XD/R7KiQciQsqbRHyZk+pxDpOtjUOW\nCedoFV4r8+rZc3D14Gblcxm/20tEEpKk77UcH5qVBRRMrF2/ko6VaRXuAifakwBXFGEgdi6iFWbC\nqw3XLEtdXpXElPZMbJn+HA9Ypv346iYW0m0jb07E+SyWMfJpdfhYd0nKPuzxPu/u4Yp35RKvptot\nFM/4AYtolo7ws1G8l44VSvg9V6+w0GHQw5V4bYlX0LGNK7agK6TWgHOllBVlATH+hiNKI2ISWEQx\nlzf0+yhLnxECC8/GY9XtsFBkHDACh/gvpMcBRJQF58vSzwMFxWdWy+eubdNzfu6L/FeSLtPTsRhM\nRCM2kOZTOGymQw7dhjyHGY1iFsdWFllwUK/h66eNRjq2vYvzRErQzQEZb6wLkBj2QYw5pAZonByl\nY9t38Gi+9r0/SsdaO3j/mlnm6+Odt34EAACFKot0XvkGyrnf/MF/TMeu37wJAAA3XnwhHQvNPHEE\nw9HFuolshuf9g/sfAwDAsM/3sVe+8lUcG/F9zKMIb+9wOx07PsH79tWbV9OxgyM8H0cbD9Kxn//p\n9wEA4Du//1+kY0DMkStLJVQ4oVAoFIpfN+hDSqFQKBQTiwul+xyiCqJEUgpEH9icBN8hd4eTZisd\nGw2RdvG7TIlVlzBMzVU4kW1bSAH0epyI3T9COqDT5qR6iZKL09McOueJ0gtFvULfJysBUdc0t4yJ\nzqWrl9Mxq4jf54lEp03UUjjgbQHz3SGHzhGF0bEco2p7U/f1V4HkTKLU1NFwDG6oJElnHh0ijXf7\n9kfp2GkbxSn5IruKnJxirdn+EZ/3Ibk3+P5xOvbxPdz/wYApwEIJf/cXMydiY1HUsLjKNOOVp3F+\nSNZgKovnJDvF7xv08btDWyTbPfzUwGe6b0Q1YEHEdVLZXJa2iSnnccAIE5hWE68Jiu88uu98Gpf+\nfWJ5gVzPGqGD3IZEvEK/a24vFl8jFolpGgfr6ViOKKm1y8+lY9fXkFqfFaKBgJL8D7d5vjR8og8t\neSsjl4wxL8FT0Yo43uEAqeXX/+wH6ViZ2LbciCnjsIf058N7PO8bm0idZSo8d2++gNQmK/8PAAAg\nAElEQVRe0hOfHZDTiqhpHNJ5lYfgzgfvAQBApcTHdNjFz/Y7bKFi7ofRgK+PxMPj223xtdCn+sG9\nB1vp2Ec/eQsAAGYELb3zzs/wtaucLrl8/RkAAChXWASXkYYUAhpJKRQKhWJicaGRlFnVuXJ1Z1bj\nYoV+cIqRz84Br5BmSW7sCClt0Men/XSWJdDGPaBY4lX23hYl73xelTgxrv7nl9g1olKhRG3E29cg\n5wS7zKuNpQV0wijUqrzNTdzmcp4PYTGHgg0vx9sXOrhsynj8PrPnoyGvVBzyGwtHYnU/ZljnpJrl\nqtCIXiIRAeYpUskXxMrpNiaB63PsOGG7GFWNRjIywxViSawUXRcjlcTmJHqng6u9TosjY5si3Y1N\nTkgf7OGx/tZvraZjlRqeC9vhZHHGwrGBWMkOB7iS9bss3jmmlaIMbt0VPMf1Mv/GOGCupfMipU+L\npM6HcZJ40kjqcSPMxHr85TNDRpgiREMeiQumhF9j2EcmZbbK71uYNWIevm52DzAa3z0VpQ2A88Wy\nHheA22MWob/9PrIJu5s76dhoF4UGjUf307EKXRfTdZ7jDrmqHLRZwNDv4vyU0e4P/vB7AADgb/O8\nb+wjm/GLn32QjqWeocIPcZfEDAVxLyrTfa7n8/t27qFrhCUccbwKXue9LkdcPWKshCEGVGI8N3NT\nzFiNSLb+4ff/OB072MZ7xDd+92/Cp0EjKYVCoVBMLPQhpVAoFIqJxYXSfS49AkPJl1DltSO4gq6P\nFNepoAWeWsUkW7YjDE4dDD8dURhi9A1l4RqRJ9PXxgm7B+SpNuDohCkeU61eKXJo2vMx3K4LQ8hn\nrz2NY9NMFXb79N0xZ/ucBMN4Sxjg2sZ9XtQmALUmcDym9kwReP6TXBbHgPOoH0kfjQLc7tOjpngd\n93/t6hfSsQ8+ehcAABYW2TC4Po0J0mkhYDDGwgNBARZrSD+MBO05oOTzQAhhBgOcCxmPt7nZRGrl\nwX1O7q5idxGIbaYz3CLSfVaL59bpISZ/Q+FwMmwhTXEaMm1creLf9eJfjcDlDJn3xNTe40jS6+/J\nvkPSaXCeiCMx9VSiFQ4dShuYQnUAj9vcvKiXO8BzHY/4OCeUeN87YXrp/g7Sgn4iTKJpu3K2pPvw\nh4Pz+mNcIP7DD94EAIDjfabibmRwf5+e4XlfpvZC4ZDnZJFo7qkMU2xbZt6HfExb1ApDuozkenjv\nKwiqOoqNYEuIxwo47zMxvy/pI2XX7fK2tHt4zIdCTDG1gm5AMwt8X1x/iDVWuZiv1dUFFLW1+vwb\nWaJ9nRbfjzeTDwEAYPTtv8Y7wtmUM9BISqFQKBQTiwuNpL71LVxdf/DBL9Kx01NchWc8Fhf8xm9+\nHQAAPv+Nb6ZjZZIdd0csbxyRk8FIRGaWg0uKWo2FDteuP4W/4fIKPYrwad/vssS408WVmW0LPz+K\nZKh/GQDwSiAjZPMxbUOpMp2OLS9g9DcSjQJtqpwfRkIiT/JbT/T4ytMqxxPHZVzgCOqcSEp4jrXb\nuNI9PmKRSm0Klz+joCQ+g8dw7Sqfk2euY0hTLfH+eR6ukd69w9+3s0+rN0e02yCpyfQcR7x9ir4t\n4Rpx89k1/I4dlsO/9jp6on35C59PxyoFjOr8AUfVlTqexyjhxLVXxu3b3WP3i9376MU2leMV5cz0\nU3DhSM78AwDnx0BPLIRI3/aE61TphJLKzYV3X+oVJ9ro0DXnCan/VBHnkyvmVa6Ac6fd5hKUo1O8\nNu9v8NxoU9NTz+XrMEPCgBuXOTIz9nJ3N9iFZhzY+HgdAABO28w0XJ7FfZsTDEmPhAvWDDMzDkUt\nWcESzZOrTbHO7zMOL1aGffo8atS5vMzOKLk83tOkEMu0MLKEy6ARbz0tzltsyoSkb6pLbVcSjpC2\n75DAos2CtyFtfsvmEo0KsUjFId97LZK5B71Pd27RSEqhUCgUEwt9SCkUCoViYnGhdN/f/4e/AwAA\n3/oOJ9U//PAWAADkshwOvvo1NE0sV0THVzIdHQjT2Z1tpP4KJQ5rF5bRtt8T5cozFEYX80zjnZKI\n4uiAk5pR9Dg1Yrpg2oJuahHNlZ8SLgPUZiNb4ENo2g/EFofYoxDD2ZMGh8QnDdqPAlNfpcpl2g8h\nsBgbHk+ipzl5maElc9CsOK4OHfeDfW4dYL6necrU2Y/e2sRXhC7k1c/jeey0OUHbOsW/8yWmM0yS\nPxTnpJgn404hpggipAOrdZ4fLarlaBwyBZNUyVg4z79Rr2IdV6fJ52lvHetdhk2eg3GM67rjPZ5H\nV6/Bf/aIhY+HofbOUo/4f0YYAQCQIcHE5Xmul1tdQHp489776ZgRWDVaTP3cu4dJ+e5QCJMsvDZK\nDl9fT5Pp8PwCi3TuPCQRzJiFEwebKNx5tLGejl26jtu1Kjo55yykwb06txzJUf1dPs8CiyhHtWLW\n42IUR9hpBESj2eLekaEu22XRZsY1B/ocojiKH68zc0WN2ojSKi1h5D1NYrWwd5CO2VRD6WUe75Tu\nCsNap4P3iwfvvpWOLV29+dh2AWgkpVAoFIoJxoVGUlNT+KQtFTlimJnDJ3tOuEZk8/g+W9i22yRt\njiKR5Bvhyqzvs8S428OVR8aTn8W1QCHPkc+QGnk1s/zZVhdXytUq+0eZpOFQVFsPSXo9HPCqfXoG\nP1Op8cpnRElFryClubh66A54hR4muPLpiiix0aKmgEv8feNCEpt9FWtj4y4hWiCY89Ps8IroHq1a\nj5ucFLVJzOBmOOJqkcTVF7Lv2/dO6DXhEUbbYOTuAAAeCWCExSLkTeQsVsshfSaXYS1ry0HpbGxx\nlD49hyIOGRGm+17lS6JWw0R+rczNOKvkNeYVxy9w+VVxjkHE/w+IEDiVlvPccIlZEB1uYIXKDp67\nwe4cGWptsxNxlG2RqKjX54h6GCHz4RRYoFIu4DXy/BpHIGvkq7l1xO07dvZwZR8/obz+s8LW5gYA\nAAwHfC28t4XR1eUrX07HXnr5VQAAyMzyvBqQEKsd8D3B+P7F4h4Yhvg+VwqLyE0nFNFQi9rg7O7e\nSceM4CwQv2HufcUi33dOjpAlGInGrxE1dJ3O8zFtdvFeagmBRdmjazBm5sIO8L4R5gRLRM1df/z6\n6+nQN/7uP4LzoJGUQqFQKCYW+pBSKBQKxcTiQuk+U/OTxEzdmFy15zLVks2SMavonmrcBQJB+8zO\nYS1LqcxUS0KJ/Vjo/G2qiB8MhFljH0NwEwYDADhUbyGpD0NgxILmMknFjujGWiTj0kyWQ1jzfSCE\nEybR7hVETUn/8Yxuq0eGtT2mKEUZxYUi7OFxsrKie7BHBr8iQTvo4379+A2ue4sSPA6tJh/rUUgO\nIicc8pv1UDxiquHhQzwnsTgBNtWiDX3eliyZ+Epz2l4Hj+FIdFXutPH7KhVxTnJ4ELcO+NyVykhT\nzDCTBFN1pD0ckfB9/vNI0ZwhjejnntiX9bMCbcSnGcjaT9qfInncYPbcuivzvWJ/HaqTyjk8j+vU\numZZmApfo666tTJf6ydkIh2HwoWCNtmKmUYekTl0pcx1iEtzeI5mK3z9txqYgF/f3E/H+sbJwRmv\nCGkYGOcM3o/FNeyWa688nY71skhHhz7f24xxayjqKafq+D5Jz42oTU4o7lqxETeJWqw/+WPshvvG\na0ynmbYYA2Fs/bWvomjt2WefTcfefAPFDD1x7zX312++wt1/Fy+jaMUVtHmBhGSNhD/rkvVHps/X\n4MDcI0QT80+CRlIKhUKhmFhcbNNDWtXl87zysag52SjgSMWhpmeif2CasM/nOLLI5VF0URM28LaD\nK984FhXxtMhoiSaKh9Scr9HgBGs2h8nCSo2FE34fI7izK1KHtplXju0mrgoCIbDIZ6jJo1iTmlfz\nBU7cG++zSESYgwC/b+9wMx27OqZuEI1dTDQX53jValdxFWrLBm7k+HF9jZsZhjEe/9u3WYLeOMZk\n8bs/49XtwiK6MuRzj1v4S5cCx8NVWSymZrdPq/5QSPspCWyicACAfhfPba3M5zNHAoeGkIx3GxhJ\nfeObvLqdmTWRnnRWeFxqPebuDynOk32nr/0lPPx+WeuPs7+LcMQvG5eHmSJHKkt1PM4LVZaMT1EZ\ngZUI8Q1dVtPCB9P3ce4PfY7GuxRl1LL8uyViKAY+CywOGvjZ4w6LmoAYDfdX8Db8S4HuBfkCz8kX\nv4jlN9UKu68MiNXxynwtmKaWkbgv7uyg3FuKl4xHqfQvdUhYdHDAUvDXXnsNAABe/eKX0rEr17BW\n4uSUr9WFBRShzM3z+fj6b34LAABcj8+vYZNcIW6LQoyU3cs30rHYiKaEZ2C0g4ISX8jXI4oYj075\nHv1J0EhKoVAoFBMLfUgpFAqFYmJxwZ15yYFBJl1d6kAbMtUVkM4+FA4QESUfZ6Y5uz0gvX23y+IH\nl6zt+z1hXkgUlay7ylBdTUl03C1QInFhiesV9veRoiqJugGbeMhwJMkWj/ZDbLP5UxjbutQFtlzi\nyu9KBcPtdofNMw1dGUZPkEn8jNFuYMht5WQLBNwHX9SG7VHH0Uqeac+YktNF4RARhNTRtiuODVF1\nkoCpVTHx2u0x/dBpoZmrpG9NZ1JbFEXlqftxTRjWhjFSegd7t9KxSg3pkakp3rfpeaQnCqKrsqEf\n/jLU2XhwngmwoQDPET+c6ah83vd98n5a5/yP7OY6R52Un7rMc7p3jJTOO2/+PB0rffu3AQBgaorr\n1jJUO1ef5VqnXguv00qRz1Gvj9dBryvqGg/wN+J4OR077uI8GVlMM5rrXtZxjQMBCSeW5nn7XKKv\nB6LIz1B2Z8yA6SQdC8ou/VsIIq7fuI7fW+Q0iEMG2e/89L10zCeR0/wSu6+8/iMURHzwHnfw/d3f\n+V0AAOiLe9v2AbXUeLwZ89m6VbqxZ7Pi2iKz25Jw4pkq4vnfFoKNzAD/doVZ8CdBIymFQqFQTCwu\ntulhBp+Bg0B6n5FkXERSLkUeR9s76VjUw9cXL11JxzbIH25vj9/X7ZvIg9d/S0tk2y8iMyNVXlzg\nlcX0LEZpgfAby5G8tVBmoUOPErWiuwAsZsgrTKwsIKbEr2h66Dm4yqgUhZR2Hj/j+6IRn4fHYKY+\nJt25wICiwXCPI7uwh+ds/4ijnKNtjHKaItlpZzH5awM3NHv5xa8AAMAIeF/yVHsQ+CxDNee9PsWt\nLow4xe+zQ0eOVt810RalRsepLsQe3fbH9B0s2AiHuK2jAa889/dR2PH+LzgBb1q9TFV4m5cv4Yo4\nI1aKSTJmQziCQ44OliWFHST2EHPfRBEll+e+m+CqNSv8LWOzPhUJ+KxpISH20bh9zFY4yT9TxWM5\nLxr5rZMe6eSY58HW1iMAAKhWn+Pvo3leEB515QKxHHlO1JtAemuD26S8+w5GANsf3E7HVp57BffR\nzonP4ofHrXExzi2zs8z+jIbkEFHi/TWsTiwcIhK6jyRCrFCYQTm/jGI9+juxRHshurc1hVDMdfA3\nOm2+3h49wPPROuFr2qJtyGZlJIrn6Fi4eBg3kMGQxS0ZEsTN1bnsYGYOo6aZGt8/hz5u890W3+9m\niVnJT3NE/UnQSEqhUCgUEwt9SCkUCoViYnHBwgkMXRNbdNI1LEMkRA0xvq+/xw4FPrVsuPHUi+lY\nfQY/U60xvdHpYdgo7ezrU0gByRYQ/W3U6B/vs1Z/YQ5DzUR0yAWHKrqFzn+KwllLVNiHlAj1HEHt\nUf1IFPO2JBSK5z2mm6IihfE5DpOLlIwuyHqqcYESoBnhttClTqG7Wxvp2Mh00xTJ7IRMMK9zY1Tw\nY9yH+5v8fVmyLPAHTP12uncBAMB1mEryByiKGQqK2B/g/PGFIe+eqbm4KzqKGsFGwLSCRQKQbJbn\n4MEe7tvOBiekp6bwMw4wvfTS5/H3XvnyM/A4xiuwiKnA6KxIgjrkijqkPHXQrYh2FnMlvL6WV1gg\nZGdwPnqecH4xtJKoyzFdXGV6ezjA49c64iR/ZFo0CGp0feMBAABcXuXJUaHEeiLMny0bf8NyxbVE\nf88I6myRBE5t4PuERQbOtugYa9FtbTTmNXiOKO0rVzlFYdP9QdyeUlGQFHYZS4/LeT6/FQvncyJo\nQT9BOq0HfM2YGraFRT6/d+7epb/4+4wIJRY9TAKi3TqCKmyetOk7Pk7HTF2irGWzyW2nN8vCmOtl\nvKZ6Fl+D/YhEZgnPokGA5/D0lOfQJ0EjKYVCoVBMLC40kgJaRYyEQMBUzluyLQdFHkWHo4hmF5+0\ndsKJxKk6Oh34Pq8S8yX8bF884Y+PMenuOPzkvnQZ7RvaWU7sNk4wwT57mZPlNUrithu8WlsjIUZv\nyDL30yP8jY2PeaVy9TquoDy5SvRIlCBWes02Ji5LFX5fuUr7/lfgaFAgz8SaaDGyeR/FKe/8/Gfp\n2JCcN27M88r4+nMo586IhO8f/jF+Rvr5tdr4fYd73Dqg38MVmy0k++USrhAdm89dSCtPVyhXQloN\njkbShYJW5CLIKZrvc/mzvQ6e990dXvXXaiiSKGV5Dg762Ibk6jUW28zM4feN27svcDBJ7QAfUyfC\nfa8LKb2/dw8AAPbbLDxZe+VlAACYFh56Hjl1ZEU7FYuW+5ZoA2FKEVwhphj28VwPR6IVSwt/T8aX\nxi/z+IjdPrIufjaWpRYUyY9EOUdM4iNbNEd9/mUUScQ1jna3T1AYEDvn3MrGfI6MZ+jMDDuezC/g\nvaXb57IZsPD1UETAMYlbyqfMXOQ38DoaiR3xXyaxV4bnqUOOLWciH7o+nAzfZ2OKikPga3WXhC7Z\nGkdmRyfUKLTJ90DjR2iJeiKLxG+dDI+dnOJnjoVHKlTxGDiixMUNcJuP2nxP/SRoJKVQKBSKiYU+\npBQKhUIxsbhYg1miWGLBjbQpDLQG/Hx0KSlcLXDNy26MCbWTE07o1a4RZddh6uz0FGmGfEFUYBN9\n1BWtNcolrIMpLXNy8Z2foo295bKT6+IyiilORM3W3uY6foeonD8hZ4of/6c/ScduPoPU11e/9e10\nbOESUmP+gOmNxgnSHznh0pDL4jEIgvE7TvRaeJwae0yjdMiw86AhjCLJEHc0x7UNGaJ0N7a5Nunw\nBN8XhkytTldw/2avv8SfLSDtsLcnTCaJLxoMuGbr4YO3AQCg32MhjDEllg4dS0t4/A11AgBw5zZS\nJhlP0Ickirl8+eV0rFa6hD9f4O8zSeKTIxaKzM3jPArj8boZGIFANma67xnqULs6zXO/mUMqOZ9j\n14NsAanC430+R6b2q5DjOVgo4fyW7Uo8j/4W9LxLAouMoEaNA0ilytfIiM6NND51yT0hGT0+z5ui\npqdLlL68HAZE5+6eMkXklqjFinBlMMYk1pj5Ppu2oVjk41Kt4nzp+kx/pW2ABC9t/rTE3A3pGnAy\nogs0GdDaohYzodYpfZ+Pi2lx5LhM7dXnkCrst3nuenTBdcW11aWWGpUK13Zlic7PC3q4T+KmUpnf\nt0tduxPh9mEFuB9dn+8li7T9jrbqUCgUCsWvMy5Wgk6rL1c0Bmx3cFUa9US1NUnA50qccLx2E5tw\ndXr89M3T6qA+zZ5h5Qqu3GQ7kKMjk8TlimnTeuOMPJlknzsbnKxcnsdoLiMywOEQt2Gmejkd85u4\nep0SkvHb73+E2ywivac/h/LlfIFX8gElvOeXF9Ixj9YLRjI9TpTquM/dDK/IK1U8AFMz99OxLEns\nSyV+36NNdG/4V9/7bjoWWWsAADA3x8drYZYa5RU5Qbt4GX3IpqY4gstSonx39x6P2RjxzEzz/KhM\n0Qq1z6vHl17CxHoiHER+8pMf0ffyVDdzoFK6mo4tr6DoJZvn418p4Tk72OeI8PoNjF4cd7zruwLJ\n65+9xHP/ay/g8W3uPErH+rSqzgpHhxGJjwaiyd5U1jTt5OumQGyEJQQqJnru9fizCUUC2Ry/z7AX\nZeHU0mhhhLy3xxFcngQbQ7Fy393BSOv2fd6PIZWoXH3mc+mYR2xItsqMS0ytf4SqOo2fxm3D+Myz\nawAAkBFCgpAEJ7K1RkLNHT1xrRvNSJDjSDQ/h9dHJJo3RhY5Uwgfy4R2NCvus6YlinF6AQD4LWrB\nsbm2nY6VqJXN5jaX5myso2AoKyLRkCLqtmDFAhKDtEX7nV0br2Wprs8XcD+fKvI90CXhR6X46aGU\nRlIKhUKhmFjoQ0qhUCgUE4uLpfso5syK5GyOkrh90QLCI/rIyjBFMUN29s1Y1Bd1kL6TJqxlMm7M\nidokk7gs5DkM7ZCDQs/iMHluGZPLm/fZsNJQO7IS36XkYyA6giZkO//UFaaMpqlq++CYhQAbdx4C\nAECtzi1ChiNMfkcDTpKWKJyW4oxxIU/UmS/McptU79ATdFpELgA50cbk0RYKTG7fZ3pueRnD+lzu\n+XTMD47p/e+mY7c+RipxeooprJUVrEmKI+Zqfu9v/B4AAMwuMN3XbOF2yXq7chmPXRjxnPntv46f\nlTRqFOHxr1b4d0+oY/P9+0z9bm8jrXV0wOfuhRexu+n0LI+NA3kyjJ0V9Synh0itvPfOO+nY9jYm\nqZ96kd1MppeQoizI+qcc1SvlmJ5LjEhCdEAGovuk6aypa5NtTay0QorHfBKeBCM+H4ckorh7i6+5\n3S0c2zlgev60j787d43pvloWz+9IUnvEPiWifseYACdj5vuW5/EeVMkIg94Q990SNWBDcm+Q2xeS\nY0eryDWIJ1fxPudZYt8cvPYyNlN7joOvV4QJ8Nw8iiTqwvy1SnSp7P4b0/n91lOvpmNPrVBdk8/i\npVwOKdawyNdMAKbdBm/LbBmPQd3i+3uGpt12yPfod974MwAA8O1Pj5M0klIoFArFxOJiHSfoKemI\navD6NAkTEpablmq4Qk5AtESglUetwKv2fXJqODxgeXiOVlfZHCeATXuMgmgM5tI2+EP+3eU1TJbH\nIct6Dw9wxX9pbY1/g8QRx4ecQG/TyrsqRAR5ku5OlXmFVCvgvhUdjiYjSmDvP2LBQJf8stZucmQ2\nM/tlGAcSqlj3XNkSAJc/shHlzike//1j0b6DjkOpzNJ+03Ryd4dFF6YVSavJq/TBEJO1u5ucyL1/\nB/3eLl3isoBsAVeA7/zsYTr2aGMdAAAKIqqrVo0jCUe84YhWdDJJTZXyX/7yF9KxrQ1czf/sLXbY\nqFTx3LnXbqZjPfKKnJkf7yo9oNYMt9dZhJD4eOyPjnj+Bi6unDfavH0HMc6tgssr/GwGz3W1yudj\nsY7zvOqJ5pLkV5nEfPwsklr7fV4tm7YTspGfkUTLFiEtYht2dkS7nR5u/1Cs8GvUwsEtMLMwACOS\nEFEdJffl75qWGWO2V4RCFa/xalU4yZTxvnTS5GtrSNdU4Eu3FNN0k6OSkEQhfHYBbLpW7T67Qbgk\nrKjXWVBSoeauCwsszipSK5QZ4bV3fIL3uyurfL298jzO9/79X6RjDpUW7Hkcrd1/hKxDJMoJYmpi\n2BYOQBDgPPGF5+fiAl6ro9ynlwloJKVQKBSKiYU+pBQKhUIxsbhQus+mcNVzmcYzrSj6Hoe6AVWm\nZ3JMKQzIFcASdU01otaGfQ4ljXHrqMNj7RaGlTNzHOrmSQhRr3DCu1TDJOXiDCcD73+E3T9Lomrc\n0BYnx2zaOaC2EQVBR7oehvbzCxx2F6n2ZCjMaftdSiSLhGgn7NPvs0nkF74KY0FMx9gR9THFCu6/\nNGbd2UNK7O33+JwYgUm9xsfaUGy7W++nY1MzawAAkBfHNQHjsiGogQSP4fbWZjr0+us/BgCAapWN\nXitVPHexSOg3yawyFF2fI6rpcYWJrefib3z4IR/rXhc/mxFCAlPaEgo62NRYjbsGJ4jxPBy0RauE\nGKkVd5pbQzhEF7VD3t82iUzcRLTMobnnnTL9vbOHO/zcJZ6/a3N4vSSCsjfmpf0ufxbIfHjvkJ0V\nTrt4rF56njsvXyaa58rqWjrWG+LBvPVAdFR2kTor1Tjx79OplgSR0UvEUjiR/j3ek1R+CuvWhsKV\nobGH94xGh6mu1tE6AAAUM9LtA+8jXl4Itqhbri2uS4dSKHLPEvq/bI5p1cNDdLV57bU307G1VaJQ\nhUCtUMBtaN15Ox377o8+BACAzS7fo7tkNnva4v1ok6gqjkQNHV3ThioHAJgq4fV2TbQhefF3qKNy\nXTaBOR8aSSkUCoViYnGhkZRZ5NrCGj5LrgZuVvh0UUIt7/IqNkcRyKDPSfA8JeXKU7zSG5Esen+D\nk++NYxQkuDlhU09iijjhXR6QNtJ1+Wm+QG05Rj5v3+EuVmB3RJV82XiUCV8tj/6WknufpPZtsZIK\nQlzd54R1vREv7G+yOGNcGNA2hqJZo0MS11qVI8/lRYxkThscUZrAKCeS46dUid7rcYuGyMIIICci\nlakp9MubqnIUViji7+3vs6T9/fffAACA6WmOpEyE2hfJ+wE1R3Q8jiKM/HlthRPD/ghXgLe22eFg\nfgFfv3b1Rjrm0jEIhiyNHqQR8XhX6TYJAxIhAImpfYf0xkzS7RItFajsIpafpfcNhdo8IGeVVdbA\nQELHwBKS55gcU2RIE5Kwoh/ynIYMXqcLy9fSoetXls78FgBAK6DmeTm+hk/b1KBPSt/p9wQBkQom\npHDCtPkAe7znKCF/uygR848EYPs77Ojww3//xwAA4CW8fcYP0Rb3hCJJymsiKqnXkfVZmGf/TBOF\nvfc2i3465MdZEt/XruJ8ibpb6dj0MkZ/2QbP8ZP7twAA4NThUp+IpPRZ4X6xRKVATobvqcbZxxLS\ncpfcMVzRCDGg73NFK6ZPgkZSCoVCoZhY6ENKoVAoFBOLC6X7TNJaROzgGZv/HId5jWMynS1xnUSh\ngiFuXnTmNN0hbWHWaFNNRF4kF/PUliESle6GJolC/o1OC0NcW3TINN01Wy2uQzg4QPqwUOb6glLZ\ndJAVNEhyTsKWMuySkslSm4Rej8UUHaI3/MGnd6r8rBFSsj0SyecRHZJOhxPhBcAIGs4AABC1SURB\nVOoAuzL/TDp25yGKD9KWDgBQJqNgf8DH0HTQ9QfctmFhARP+cwtMKzz3AhoL//AHTI88eoDGvVdX\nuRr/G99AJ4KtHaaITk6QQqpWOdne7eL2f+Fldi5YXMTt+5/+5/8lHfN9rP26cnUuHTs8OKF/j8X7\nqA3JiOdg5tNzv78ymMQTLRrSF8+jtQTdd+5nTW8IOUb7JK4vxzGJer6ITXuUSKxxG30cy9X4+M1V\n6dosMmVskSAiHDL1s0+dYA3FBwAwSt1P+Hfj5JPpu+Tcv8dL93kW7q/j8P2kPoX3iYFoU1Elg+xR\nl0Uwpq2QJfjXB9RpNy/cdMz90xFUpkMG3e02pyMiOlZHByxGWch9BwAArlZ4wnaaOMeH4tgaAUYG\nRG0cpWmkQCWlYsUNPiI7kHDExyAiYYW/yvfPI3NvbsoqsPOhkZRCoVAoJhZjEU6EQqJo03OxkOen\nuWmtIBvYxfS+jJBpWh4lSUXDuaBH1fQ2JwgXjLdfiX/DrAoc4d0HJEuOQ/7dAVV+NxqyzQceJtmG\nwCVptlwxmORtQTRgTMgbKyfEFKY631TfAwAktP4riVYW48IwwP0KA448fRIINITk9N49rED/2qtf\nSceWFlDMYHu8WjY+ZN0+CyeMOMYWK/cHD1Huv7e3zu/rYaS1Lxr02eQ5ZzzPAABcitzywp9xfgFX\nahWxmreP8Hscj6f6CSWVB0PhjkAR3p//+X9IxzoUhU2VONIzrigjITiAIkw4zokozhmS4gODVPIs\njr0RTgyFCOmwQ2OWaMdg4zwYiQZ9ETEP/ZDH9o8x4g6Fd6S5/qPoHFHImchxvI0NfxnsBK+Zg1N2\nRrGGOF9qRb7+a+RV2bb42kqIzSgLB5VGw7wuPP7oeHS6HDUZ1w3J6gBdM802R3B/+H0s5aiJ0CQh\nAdjqPLMPjYDcYfp8DzSRbSDcJcypGY34N4DubZ6I9J55GksQVl9gAc1xQOf8Cc6fRlIKhUKhmFjo\nQ0qhUCgUE4sLpftMYi0S9FwUE09i85jphhpbwnCRPhMNRJU8JVFtEOKHQ0x479x9kI7NX8JCj5Iw\nUhxRLZAjbe+pAj8RwokeCQUiQT0YE1PpvmAorb4wUjQOG4MBj5lEoyPaJHSoZionumZmqZ5hNOT9\nHReaLTzusial18dj0u/z9nT7uN3v/uK9dOzlF78OAADXb7yUjt2j1id5aQ5Kx3845GMzJGFFp820\nwvEhCiHksc5SVfz9j/kcHx39awAAuHyFu//W6/h7iaj3ylAd3d177C7xzs/ep21ietMY4O7vcasO\niwQE0zUuHBoMiAphphaETuPCkHzWtNYTmjIY8VM8ZHopTvC4tAPepuMuzpPIFioSmk8D0aYiJFp+\nr8ECoWNyComFYAOIYhe3jpTuO2+TrfPEI/F4qcCbJRL2CIqyUMM5Odjh/c09iy1sbt3lWsBoilwj\nBC0Nd+8AAEAY8UGw6L4kW8+kXX9FbVKtipRiRgjPTqiF0JHF894+xGswK1Iyv//3/jYAAPyL/+3/\nTMe2tpE2j0Vck6euvk89zY4iBbqn3b31Ef/uKQqPTk+YhidzHnCf4HankZRCoVAoJhYXK5wwvloi\nKonIa28w4JVZJoerCONKgO8L6Dv4OToiy3cr4ojr3gdYHX3v5+wT9+q3vwYAAJVl9uQzMshQrJ7N\nescX8tBej5wmxMpsROKIWLQIGA7xe0IhujCigKMjFgwkcUjvk5l2er9Y+RgJqi+aQY4LbZLCitOU\nRiMvvcgGgsaPyxPbbbziInFOjAtEpTKbjo0C045Btncgz0Bxjk3EEImoLqSkfbPL4cvcHFbc//1/\n8HfSsdU18vMT8vD9PTwXtmiOuLm1DgAAWzt30jEvetxnsj6FTTGLJSFp7+B5b7fFEh8+vWr+V8av\nEBT8sqDpvK8dCZGEiXwTkTAfJrha3j/lEoO+8TS0hatFhH+fdvhav/XxOgAAfLzBJQYB3YZswTaE\nyeMiCeOSYcualnMcJ9KxMUvQb1ZwTsoI3SefUTdgtmBpAaP/j+7yZ69cWwMAgGKG59Jbr70OAABB\nyPvrkNjLEftm3B0yZxq14tiVq8w0GDn/L259kI7k6DgP+7zNH97CFju9vv/YZyEW9zsSzpSyoqFo\nQPftEZ+Pd3+Kv9c95vnyz/+7P8D9LX76I0gjKYVCoVBMLPQhpVAoFIqJxcV25iXEidDWE+0WCNpn\nGJDjhM0hYpSg6WjGEzVHFOIGIjR1qQYjL+ozHNP5UoTdRhAx6ssKZ/xsR3SM9Ml+vifqEAz9Id0q\njCDC7/F+ZHO4re0WuzT0SWxQyMs6iRp9hxRnFOg7hEHnmBAEWHVuielQKmHd06tf+Vo6tryCXYMf\nPuCOu/2+cWPg4zAK8Hjl81w7NT+/BgAAQ5GAP+ig0WUsqAtDmbqObFmAf1crbOabzeP6qt/jc3d0\nSNXz4rwPBiTeEecuCHAbBn3+7MDH8zk9wya2JaIrXUGjjEgEMBoJyuk/GxDtLhL1QzqXtnAfaPl4\nPR+cth/7rCWVDkRD7dN5AQDYP0D61Zedfp2M/Ar6Hvo+RwzGpk5S/CxdQzKlwO4u4z1HPlH/HZE+\naDSJ5uPdhdsfIv316J6gm8nkWho6mzkrIwlD80VCZGKOh9SOmJrOfp+Nsq9cWQMAgL1dToPsb2GH\n5DtNPpe/IJGRpF8TswPid/0ezo2333wnHQuJ9q/VeD++/Zt4D/nc555Px+ZX0VzaeYInkEZSCoVC\noZhYXGwkRSuaWKzMTBI8CETTwxFGN4FYIvmUsC1kRZKeLPBDsVpbXMXk9kyZn9zTK+gQcHrKnmvd\nDq5oEtFE0R8YEQdHVwOSgB+dsPihTD59o1BGhLRCl6tO+h7f52jBNOATP5H+XkU0YDTtPYIhRwHj\nwmCIx7UuGpC55O7Rbp2mY0GAUWZiiQZ4dBy8jJAZR7gPslmbS4KEjGhtYjwMpa+hQ60AXJcTyMaL\nb36OpeD5PK4Uf/jn3KwtQ7rWivBYnKWWBiPh45j6mUn/M1r1e54QQdCqUUqAjSvCMJDCicnBeVLs\nXyYfsM443ZmVuxAhWGblzufymHwt/UCUh5BgIpHNB+n4DcXxsy1TbiIiKUren+msYZwu5KbAeSIJ\ni8aEByGbGsI40aHmje0Bb58f43zKF/hany3h9f/yNY7ad/fWAQDg3kcsDpIOPAYxibjkIXDo2pJx\no2naur7O7WiML2ilxq0/Tqg5YrXGTM/8PLJYbeHbub+3T98hymuM4EWEiTEJ3lYusc/mP/mn/wgA\nAHKioaNxIQqf4HankZRCoVAoJhb6kFIoFArFxOJiHSfIhTMQYoURhXmJqDQ2FJcMb41jfShZFaIP\nYlGX4s0hFVeYYWPGLtFtphUHAIBP4ghL0AIjov4GPsecjQ6G27JeaXoa6cMwkOE3JfiFM0KS0pWi\nLox2wMpyqGsSkrJGJaJ9l+H0uFAik0nZUfToFI9hv8chfxQjXVCb4uPfbuO+Pnq4k44N6HzarhAr\n0Jis5TBGvNKRxCHBhKQKA6KID0T9WTFPrT98PoqmRYgDLLYp5IzBL48Zqk5SJp73eL+NAYk8DN0L\nAOCa7qvReJ1BntRx4lznhfPeR/864nsNs56RlCedj+6Qz9EBmS/L9h0Jna/oDI1HdD/IWifz+6LO\nx7x2hrIzbSB4+2wjkhDvi427yBkKEMfGXSdl9tMSagCb/i6IdhtLy0hb16e4/m61gbT6oRCZNNfw\nfb6oVwrontoTXcJ9qkmSRr5ALjpPP/NsOmTMonf22Lx5SPed73ydBVKrlzCF0mjwtvRJrPb22z9P\nxzYeoTtMKByAMiT8ev75p9OxiERybZEGcYkedq1Pj5M0klIoFArFxOJCI6l2G1fhqQwTABJ66noZ\n/mkjPmi1WC7Zr+LTNyhx9OJQUzFbVLVHVFntC8GBT0nyoWggZmTJnsMrZp8iqJ6QkZ9SFf15K2vZ\npNAk2KX/nlkdylVvlpKFZ9p30KrPePjJseCvwLsvpOxlq8u/bYrcg8gT7zNRkGiZQSvFBw84QWsq\n5DOiLMA4Tsh1UY7EDwNx7lwSTkg3DuOhJ0UqpoShx1MmdfdIQIpj8HwaST0A+5rt72+mYyaSFVUB\nMPTx/GTzq+mYl8GIve+POZIykYKIGJ40ajoPNrWsybr8HR5F0kcNjp4DEjBJ/71uj86ljJrSuQ9i\njIQOT7idUhBhyjPkmImuzookjJhCfg/9O+Y2HhlSeYRCNm9TA8GSy21+rJgaP4bM/izM4vyMri6n\nY8EQ5/jZ8gnDIDDj0qf7V1PcPz94cAgAAN1OXnwWj1+7IYwnSejy7s9YRn7rPWzJUyjyZ4sl3Nac\n8BbMuOZA8/bdWEOHi7kq79veBkraCwUeK5jWRbY6TigUCoXi1xj6kFIoFArFxOJC6b7DU0x0t5ps\nLJjLUwdKj8Nf07W2v8uJ8QNK7s2QaSMAQDZjkqn8GyktJ2qYjo6wPmpnndsuJJRwzIo6HYsEDH3R\nPmJEofW0SGqaqvtE1HF5JN5whHGpEWDkc/wbXppo55C4TR15XZGgzhNtWCzxZ8eFtumqKuteiGKT\ndEsmZ1qWCOrSRlrhtMnnzrQHkI1CwxCP8SiUDh04B2ybKVObPuR+Sim66djsZphSNAIMv890VUw1\nW4MhUxw5Su7Kbskmf2sMPAGY4rhylZPAXhbPTzAaL5Vk6o9+BYbvDByiaKRxrHFM2DlmmnyfaFfZ\nUXlE2xKLseRJe3+cg+Qck9hfOiZrsRJ47H2pk8tndbCeEHlyXfE8IbDyaGLFojM3HWdZdxmOqAY0\nEjWIOWFTQTAUbyjOm2nv0xcuOUVKL7x3d0u8D7dvvsTbYqi6pMf3aHP2o45wl/DwdzOibvVz17H1\nRqUgumPPTwMAgCuucy/Ea0a6AuWI5st6KpxQKBQKxa8xLjSSOjpGCaOUMhaKFDUJOaJDK4+BkKrv\ndlHSvDx7KR2rVHAFHJzxZkMxw+nRQTq2/ggjqL0tXkX4XUwqFkv81K/UMFoajnjlWKDXpWzWp9dd\nKZul9iKBkKWbFZxMaDepOl+ufIy7RLnIEYmR/co2CeNCSAnzXJ5XOkVKmtqiOWVIkVZXeB3OTKO/\nnWVx2OT7GMn4A45oTGRWEs4gTqoe5s+6bp7G+BiGMf6udIOokWw+L6TlYYSfcaQEmJLZsVh9e3k8\nT2vX2EtsROKNbJ4j/BdewEaOV1ZZOOEAtV5wxxxJnSOcSF0ZrMeFBBLnCSwMKxAL+X9CQohQNC4M\nzTo2frz84oxknCIp6zM6LGY/pL+lOYeyl6EZO2+/zxu7SBi9hJsVFEJqIygcWehfV4iDEprbsgxn\nNMJ5bEmfPsPmZEXkmMNzWC3z3J2Zw4jm+WfW0rFOG++BgwHf7wzDY5xx8G8qmxHzxqWLVTJHGWIx\nHNFexLAU2awsNSHRhRBOZLLGgeZxgdpfhEZSCoVCoZhY6ENKoVAoFBMLa9whsUKhUCgUTwqNpBQK\nhUIxsdCHlEKhUCgmFvqQUigUCsXEQh9SCoVCoZhY6ENKoVAoFBMLfUgpFAqFYmKhDymFQqFQTCz0\nIaVQKBSKiYU+pBQKhUIxsdCHlEKhUCgmFvqQUigUCsXEQh9SCoVCoZhY6ENKoVAoFBMLfUgpFAqF\nYmKhDymFQqFQTCz0IaVQKBSKiYU+pBQKhUIxsdCHlEKhUCgmFvqQUigUCsXEQh9SCoVCoZhY/H8m\ndwgso2iLYQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f133dd9bef0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "inputs,targets = get_cifar_batch_of_size(12)\n",
    "plot_cifar_samples(inputs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true
   },
   "source": [
    "# Models"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true,
    "hidden": true
   },
   "source": [
    "## Linear"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": true,
    "hidden": true
   },
   "outputs": [],
   "source": [
    "class LinearNetMNIST(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(LinearNetMNIST, self).__init__()\n",
    "        #Let's start with an input of (bs, 784)\n",
    "        self.layer1 = nn.Linear(784, 500)\n",
    "        self.layer2 = nn.Linear(500, 250)\n",
    "        self.layer3 = nn.Linear(250, 10)\n",
    "        self.softmax = nn.Softmax()\n",
    "        \n",
    "    def forward(self, x):\n",
    "        x = x.view(x.size(0), -1)\n",
    "        x = self.layer1(x)\n",
    "        x = self.layer2(x)\n",
    "        x = self.layer3(x)\n",
    "        x = self.softmax(x)\n",
    "        return x\n",
    "    \n",
    "class LinearNetCIFAR(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(LinearNetCIFAR, self).__init__()\n",
    "        #Let's start with an input of (bs, 3, 32, 32)\n",
    "        #We flatten it to fit the first linear layer\n",
    "        #(bs,32*32*3)\n",
    "        self.layer1 = nn.Linear(3072, 512)\n",
    "        self.relu1 = nn.ReLU()\n",
    "        self.layer2 = nn.Linear(512, 256)\n",
    "        self.relu2 = nn.ReLU()\n",
    "        self.layer3 = nn.Linear(256, 10)\n",
    "        self.softmax = nn.Softmax()\n",
    "        \n",
    "    def forward(self, x):\n",
    "        x = x.view(x.size(0), -1) #(bs,32*32*3)\n",
    "        x = self.layer1(x)\n",
    "        x = self.relu1(x)\n",
    "        x = self.layer2(x)\n",
    "        x = self.relu2(x)\n",
    "        x = self.layer3(x)\n",
    "        x = self.softmax(x)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": false,
    "hidden": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Variable containing:\n",
      " 0.0856  0.0971  0.0903  0.1015  0.0910  0.0767  0.1473  0.1231  0.1022  0.0853\n",
      " 0.0841  0.1002  0.0872  0.0988  0.1089  0.0776  0.1262  0.1015  0.0779  0.1376\n",
      "[torch.FloatTensor of size 2x10]\n",
      "\n",
      "Variable containing:\n",
      " 0.1025  0.1128  0.1012  0.0917  0.1088  0.0957  0.1047  0.0971  0.0925  0.0930\n",
      " 0.1092  0.1095  0.1064  0.0849  0.0987  0.1018  0.0940  0.0921  0.0937  0.1096\n",
      "[torch.FloatTensor of size 2x10]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#MNIST\n",
    "inputs, targets = get_mnist_batch(2)\n",
    "net = LinearNetMNIST()\n",
    "print(net(Variable(inputs)))\n",
    "\n",
    "#CIFAR\n",
    "inputs, targets = get_cifar_batch_of_size(2)\n",
    "net2 = LinearNetCIFAR()\n",
    "print(net2(Variable(inputs)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "hidden": true
   },
   "source": [
    "## CNN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "collapsed": true,
    "hidden": true
   },
   "outputs": [],
   "source": [
    "class BasicCNN(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(BasicNet, self).__init__()\n",
    "        #Let's start with an input of (bs, 1, 28, 28)\n",
    "        self.conv1 = nn.Conv2d(in_channels=1, out_channels=100, kernel_size=2, stride=1, padding=0)\n",
    "        self.conv2 = nn.Conv2d(in_channels=100, out_channels=10, kernel_size=2, stride=1, padding=0)\n",
    "        \n",
    "    def forward(self, x):\n",
    "        out = self.conv1(x)\n",
    "        out = self.conv2(out)\n",
    "        return out"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true,
    "hidden": true
   },
   "source": [
    "## GAN"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true
   },
   "source": [
    "# Training"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true,
    "hidden": true
   },
   "source": [
    "## Helpers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "collapsed": true,
    "hidden": true
   },
   "outputs": [],
   "source": [
    "def train(net, dataloader, criterion, optimizer, epoch=1):\n",
    "    n_batches = len(dataloader)\n",
    "    total_loss = 0\n",
    "    for inputs,targets in dataloader:\n",
    "        inputs = Variable(inputs.cuda())\n",
    "        targets = Variable(targets.cuda())\n",
    "        \n",
    "        ## Forward Pass\n",
    "        out = net(inputs)\n",
    "        \n",
    "        ## Clear Gradients\n",
    "        net.zero_grad()\n",
    "        \n",
    "        ## Get Loss\n",
    "        loss = criterion(out, targets)\n",
    "    \n",
    "        ## Backprop\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "        \n",
    "        total_loss += loss.data[0]\n",
    "    \n",
    "    return total_loss / n_batches\n",
    "\n",
    "def predict(net, inputs):\n",
    "    out = net(inputs)\n",
    "    val,idx = torch.max(out,dim=1)\n",
    "    return idx.data.cpu().view(-1)\n",
    "\n",
    "def get_metrics(net, test_loader, criterion):\n",
    "    data,targets = next(iter(test_loader)) #1000??\n",
    "    inputs,labels = Variable(data.cuda()), Variable(targets.cuda())\n",
    "    \n",
    "    outputs = net(inputs)\n",
    "    loss = criterion(outputs, labels).data[0]\n",
    "    \n",
    "    #Flatten and Get ArgMax to compute accuracy\n",
    "    val,idx = torch.max(outputs,dim=1)\n",
    "    predictions = idx.data.cpu().view(-1).numpy()    \n",
    "    targets = targets.numpy()\n",
    "    correct = np.sum(predictions==targets)\n",
    "    accuracy = correct / len(targets)\n",
    "    \n",
    "    return loss, accuracy"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true,
    "hidden": true
   },
   "source": [
    "## MNIST"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "collapsed": true,
    "hidden": true
   },
   "outputs": [],
   "source": [
    "N_EPOCHS = 5\n",
    "mnistnet = LinearNetMNIST().cuda()\n",
    "criterion = nn.CrossEntropyLoss().cuda()\n",
    "optimizer = optim.Adam(mnistnet.parameters(), lr=1e-4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "collapsed": false,
    "hidden": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1, TrainLoss: 1.663, TestLoss: 1.560, TestAcc: 0.911\n",
      "Epoch 2, TrainLoss: 1.556, TestLoss: 1.540, TestAcc: 0.926\n",
      "Epoch 3, TrainLoss: 1.546, TestLoss: 1.529, TestAcc: 0.939\n",
      "Epoch 4, TrainLoss: 1.541, TestLoss: 1.536, TestAcc: 0.928\n",
      "Epoch 5, TrainLoss: 1.538, TestLoss: 1.538, TestAcc: 0.926\n"
     ]
    }
   ],
   "source": [
    "train_loss_history = []\n",
    "test_loss_history = []\n",
    "test_acc_history = []\n",
    "for epoch in range(1, N_EPOCHS+1):\n",
    "    train_loss = train(mnistnet, mnist_train_loader, criterion, optimizer, epoch)\n",
    "    test_loss, test_acc = get_metrics(mnistnet, mnist_test_loader, criterion)\n",
    "    print('Epoch %d, TrainLoss: %.3f, TestLoss: %.3f, TestAcc: %.3f' % (\n",
    "        epoch, train_loss, test_loss, test_acc))\n",
    "    train_loss_history.append(train_loss)\n",
    "    test_loss_history.append(test_loss)\n",
    "    test_acc_history.append(test_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "collapsed": false,
    "hidden": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f133ff83a20>,\n",
       " <matplotlib.lines.Line2D at 0x7f133ff83da0>]"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD1CAYAAACrz7WZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJztvXt0XNWZ4Pv76iGp9H75Kcny20IYbIgbAoRAojx4xU4Q\nBEwgq+f6zgzdSTrdM1m5ye3pzp3ue3tmrdzb02QymZ5ZgTAEYgKYxGCegYQ4pAUk+IVt2fgpS7KR\nLMkqWe967PtHlWRZUkl1VFU6p6q+31parjp7n3N++lT+atc+X+0jxhgURVGUzMJlt4CiKIqSfDS5\nK4qiZCCa3BVFUTIQTe6KoigZiCZ3RVGUDESTu6IoSgbisVsA4K233jK5ublz2jcYDOLxOOLXuAz1\nsoZ6WcepbupljUS8BgcHuxoaGhZM1+aI3zQ3N5e6uro57dvb20tpaWmSjRJHvayhXtZxqpt6WSMR\nrz179rTEakv7aRkn/rFAvayiXtZxqpt6WSNVXmmf3IPBoN0K06Je1lAv6zjVTb2skSqvtE/u7e3t\nditMi3pZQ72s41Q39bJGqrzSPrkriqIoU9HkriiKkoFoclcURclA0jq5D4yG2N02jC5brCiKcjmO\nqHOfC8YY/mLnUVr9I6xb1sd1NSV2K11GSYmzfMZQL2s41Quc66Ze1kiV16wjdxF5TEQ6ReTgDH1u\nFZF9InJIRH47YXupiDwnIkdEpFlEbkiWuIjw+XUVAGzf1+G40Xu21dQminpZx6lu6mUNO+vcHwdu\ni9UoIqXAj4DNxpgrgXsnND8CvGqMqQM2AM1zV53KXXWVFOa4OdQxwAcf9Sfz0AmTbTW1iaJe1nGq\nm3pZw7Y6d2PMbqBnhi4PAM8bY85E+3cCiEgJ8Eng0ej2UWNMb8LGE8jPcfOp6hwAfravI5mHTphs\nq6lNFPWyjlPd1MsaTq5zXwuUichbIvK+iHw1un0FcB74iYjsFZEfi0hBEs53GQ3LcvF5Xexpv8jR\n8wPJPryiKEpakowLqh7gY0AD4AOaROSd6PZrgW8YY94VkUeA7wB/M/kAnZ2dbNu2bfx5Y2MjjY2N\nVFVV4fF46O3txe/3TzlxVVUVBV4Xn11RxAsf+nn0X07xtY1Fl7XPtn+q2v1+//hqb3acP1a73++n\npaXF9vhM1z7mZWd8JrcDjonPZMLhsCP9xv6Wdsdncvvk15hT/IaHh6e89uPdfyYknguRIrIc2GWM\nWT9N23cAnzHme9HnjwKvAr8D3jHGLI9uvxn4jjHmzsnHaGpqMnNdFbKlpYWiBUt56OeHCIQM/+Pu\nOlaU++Z0rGTS0tJCbW2t3RpTUC9rONULnOumXtZIxGvPnj3vNzQ0bJquLRnTMjuBT4iIR0TygeuB\nZmPMR0CriKyL9msADifhfFMoz/dye7Ry5un9zpp7VxRFsYN4SiG3A03AOhFpE5FtIvKwiDwMYIxp\nJjJSPwC8B/zYGDNWNvkN4CkROQBsBP4h2b/AWI3ol69ehFvgtycv0O4fSfZpLJNtNbWJol7Wcaqb\nelkjVV5xTcukmkSmZSby/+1u4bUPe7h9XQV/dfOyJJgpiqI4l1RPy9jKxBrR+zYswiXwq2M9dPaP\n2miVfTW1iaJe1nGqm3pZQ9dzj8HEGtHqkjxuXlFKMGx47oNOG62yr6Y2UdTLOk51Uy9rOLnO3VFs\n3bAYgFeOdHFhKGCzjaIoij1kXHJfWeHj48uKGQkZnj943m4dRVEUW8i45A6wdWNk9P7i4fNcHHHm\nPJuiKEoqycjkfsXCAq5ZWshgIMzOw1126yiKosw7aZ/cY9WIjo3ef3Gwk6FAaD6VgOyrqU0U9bKO\nU93Uyxq2refudGKthbxhSSH1Cwu4OBLipeb5H71n29rRiaJe1nGqm3pZw8713B1NrBpREWHrxkUA\nPHewk9FgeD61sq6mNlHUyzpOdVMva2idewxmqhG9rqaYleU+egaDvH5spiXpk0+21dQminpZx6lu\n6mUNrXOfAxNH7z/f30EwbP9SC4qiKPNBRid3gE8sL6W6JJeO/lF+c2J+R++Koih2kfHJ3e0S7t8Q\nGb0/va+DsAMWSlMURUk1GZ/cAT69upxFhTm0+kd4+3RSb+OqKIriSNI+ucdTI+pxCfdevRCIjN7n\nY5njbKupTRT1so5T3dTLGlrnHoN4a0RvW1tBuc/D8e4h/tDWl2Kr7KupTRT1so5T3dTLGlrnHoN4\na0RzPC4ar4qM3n+2N/Wj92yrqU0U9bKOU93Uyxpa5x4DKzWid11RSVGum8OdAxw4159Cq+yrqU0U\n9bKOU93Uyxpa554EfF43X7pyAQA/26c30lYUJXPJquQOsOXKBeR7Xew9e5EjnQN26yiKoqSErEvu\nRbkevnBFJQDb9+voXVGUzGTW5C4ij4lIp4gcnKHPrSKyT0QOichvJ7W5RWSviOxKhnAyuHv9QnLc\nQlOLn1M9Q3brKIqiJJ14Ru6PA7fFahSRUuBHwGZjzJXAvZO6fBNonqvgbMylRrQs38vt6yoAeDpF\no/dsq6lNFPWyjlPd1MsattW5G2N2AzMtyvIA8Lwx5ky0f+dYg4hUA3cCP07QMyZzrRG99+pFuAV+\ne/IC7f6RJFtlX01toqiXdZzqpl7WSJWXJwnHWAt4ReQtoAh4xBjzRLTtn4BvR7fHpLOzk23bto0/\nb2xspLGxkaqqKjweD729vfj9/in7VVVVAdDf3x+zfab9G1aX8fqxC/z0D2fYuibH8v4ztYfDYWpq\naua8f6raw+EwLpfLtvPHar9w4cK4l53xmdwOc399pbp98eLF5ObmOuLvN7F97DVmd3wmt0987dsZ\nn8nt3d3d9PdPLc2O9/UZC4nnyzwishzYZYxZP03bD4FNQAPgA5qIjNbXAncYY/5cRG4FvmWMuWu6\n4zc1NZm6urpZPaajpaWF2traOe3b7h9m23PNuER4/Mv1LCycmuDnSiJeqUS9rOFUL3Cum3pZIxGv\nPXv2vN/Q0LBpurZkVMu0Aa8ZYwaMMV3AbmADcBOwWUROA08DnxaRJ5NwvqRRVZLHJ1eUEgwbnj3Q\nOfsOiqIoaUIykvtO4BMi4hGRfOB6oNkY811jTLUxZjlwP/BrY8yDSThfUhm7kfYrR7u4MBiw2UZR\nFCU5xFMKuZ3IVMs6EWkTkW0i8rCIPAxgjGkGXgUOAO8BPzbGxCybdBoryn3csKyE0ZDh+YM6elcU\nJTOY9YKqMWZrHH2+D3x/hva3gLesiM0nWzcuoumMnxebu/jyhkUU5SbjOrOiKIp9pP03VJNRI1q3\nsIBrlhYxGAiz89D5JFhlX01toqiXdZzqpl7W0PXcY5CsGtEHojfS/sWh8wwFQgkfL9tqahNFvazj\nVDf1soau5x6DZK2FfPWSQuoXFnBxJMSu5q6Ej5dta0cninpZx6lu6mUNXc89BslaC1lEeOCayOh9\nxwedjAbDCR0v29aOThT1so5T3dTLGrqe+zzwJ9XFrKrw0TMU5LUPu+3WURRFmTOa3CcgImzdEBm9\nP3Ogk2A49TfSVhRFSQWa3Cdx0/JSqkty6egf5dfHZ1ovTVEUxblocp+E2yXcHx29P72/g5CO3hVF\nSUPSPrmnokb006vLWVSYQ5t/hN+f7p3TMbKtpjZR1Ms6TnVTL2tonXsMUlEj6nEJX756IRC5kXY8\nK2dOJttqahNFvazjVDf1sobWuccgVTWin19bQXm+h5M9Q7zX2md5/2yrqU0U9bKOU93Uyxpa5x6D\nVNWI5nhc3LM+MnrfPofRe7bV1CaKelnHqW7qZQ2tc7eBO6+opCjXzeHOAfafm3qnFEVRFKeiyX0G\nfF43XxofvX9ks42iKEr8aHKfhS31leR7Xew9209z54DdOoqiKHGhyX0WinI9fKF+AQBP7+uw2UZR\nFCU+0j65z0ft6t3rF5DrFprO+DnVMxTXPtlWU5so6mUdp7qplzW0zj0G81G7WubzcntdJRD/3Hu2\n1dQminpZx6lu6mUNrXOPwXzVrt5z1UI8LmH3qV7a/cOz9s+2mtpEUS/rONVNvayhde4xmK/a1YWF\nOXxmdTlhE1lzZjayraY2UdTLOk51Uy9r2FbnLiKPiUiniBycoc+tIrJPRA6JyG+j22pE5Dcicji6\n/ZvJFLeD+zYswiXwxrEeOvtH7dZRFEWJSTwj98eB22I1ikgp8CNgszHmSuDeaFMQ+PfGmHrg48DX\nRKQ+MV17qSrJ5ZaVZYQMPHtAK2cURXEusyZ3Y8xuYKaFzR8AnjfGnIn274z+e84Ysyf6+CLQDFQl\nbGwzY8sBv3K0mwuDAZttFEVRpicZc+5rgTIReUtE3heRr07uICLLgWuAd5NwPltZUe7jhtoSRkOG\nHQc77dZRFEWZFk+SjvExoAHwAU0i8o4x5kMAESkEdgB/aYyZdnnFzs5Otm3bNv68sbGRxsZGqqqq\n8Hg89Pb24vf7p+xXVVVFSUnJjO2z7T+X9k8tMjS1wIuHu/jy1YsID/dP2X94eJhgMJiS8yfSPjw8\nTEtLi23nj9U+0cvO+Exut+P1FW97YWEhgOP8xv6Wdsdncvvk15hT/Nxu95TXfrz7z4TEs9phdOS9\nyxizfpq27wA+Y8z3os8fBV41xjwrIl5gF/CaMeYfYx2/qanJ1NXVzerhJL7zynH2tF/koWsX89C1\nS+zWURQlC9mzZ8/7DQ0Nm6ZrS8a0zE7gEyLiEZF84HqgWUQEeBRonimxJ4pdtasPbIzMvf/y0HkG\nR0NT2rOtpjZR1Ms6TnVTL2vYVucuItuBJmCdiLSJyDYReVhEHgYwxjQDrwIHgPeAHxtjDgI3AQ8B\nn46WSe4TkTuS/QvYVbt61eJCrlxUwMWREC8d6ZrSnm01tYmiXtZxqpt6WSNVXrPOuRtjtsbR5/vA\n9ydtexuQuas5GxFh68ZF/IfXTrLjg0621C8gx5P23wlTFCVD0GyUAH9SXczqCh89Q0Fe/bDbbh1F\nUZRxNLknQGT0vhiAZw90Egxbv5G2oihKKtDkniA3LS+hpiSXjv5Rfn18pu96KYqizB9pn9ztXqPZ\nJcL90cqZp/d3EIqO3u32ioV6WcOpXuBcN/Wyhq7nHgMnrNH8qVXlLCrMoc0/wtunewFneE2HelnD\nqV7gXDf1soau5x4DJ9SuelzCfdE1Z7bv68AY4wiv6VAvazjVC5zrpl7W0PXcY+CU2tXPrSmnPN/D\nyZ4h3m3tc4zXZNTLGk71Aue6qZc1bFvPXYmPHI+Le64aG71/RDzLOiiKoqQKTe5J5M66Copz3TR3\nDnLkgjM/AiqKkh1ock8iPq+bL61fCMDLp4ZstlEUJZvR5J5kttRXku91caQnSHPngN06iqJkKWmf\n3J1Wu1qY62Fz/QIgMvfuNJwWrzHUyzpOdVMva2idewycWLt69/oF5LqFd870cbLbWdMzTowXqNdc\ncKqbellD69xj4MTa1VKfl9vWlgOwfb+zRu9OjBeo11xwqpt6WUPr3GPg1NrVGytDeFzC7pO9tPmH\n7dYZx6nxUi/rONVNvayhde5pRnmei8+uKccAP9/fYbeOoihZhib3FPLlqxfhEnjjWA8dF0ft1lEU\nJYvQ5J5CqkpyuWVlGSEDz36go3dFUeYPTe4p5v7ogmKvHO2mZzBgs42iKNlC2id3p9eurij3cWNt\nCYGQYccHnTZbOT9eTsOpXuBcN/Wyhta5xyAdalcfiN6Kb9eRLvqG7S3HSod4OQmneoFz3dTLGrbV\nuYvIYyLSKSIHZ+hzq4jsE5FDIvLbCdtvE5GjInJcRL6TLOmJpEPt6toF+XysqoihQJhfHjpvo1V6\nxMtJONULnOumXtaws879ceC2WI0iUgr8CNhsjLkSuDe63Q38N+B2oB7YKiL1iQpPJl1qV8dupL3z\n8HkGR0N2KAHpEy+n4FQvcK6belnDtjp3Y8xuYKY7Pz8APG+MORPtPzaxfB1w3Bhz0hgzCjwNbEnQ\nN225ekkh6xcVcHEkxK4jXXbrKIqS4XiScIy1gFdE3gKKgEeMMU8AVUDrhH5twPXTHaCzs5Nt27aN\nP29sbKSxsZGqqio8Hg+9vb34/f4p+1VVVQHM2D7b/qlq9/v9BIPBy9oblgoHO+CZfee4pnCIFcuq\n593P7/fT0tJie3ymax/zsuv807WDM19fAOFw2JF+Y39Lu+MzuX3ya8wpfsPDw1Ne+/HuPxMSzx2D\nRGQ5sMsYs36ath8Cm4AGwAc0AXcCVwO3GWP+92i/h4DrjTFfn3yMpqYmU1dXN6vHdLS0tFBbWzun\nfVPJdF7GGL6+8yjHuob4+o3V46tH2u3lBNTLOk51Uy9rJOK1Z8+e9xsaGjZN15aMapk24DVjzIAx\npgvYDWwA2oGaCf2qo9uyFhFh64bI3PszBzoIhvVWfIqipIZkJPedwCdExCMi+USmXpqBPwBrRGSF\niOQA9wMvJOF8l5Futas3Li+htjSPzv4Abx6f6VJGaki3eNmNU73AuW7qZQ3b6txFZDuRqZZ1ItIm\nIttE5GEReRjAGNMMvAocAN4DfmyMOWiMCQJfB14jkuyfMcYcSvYvkG61qy4R7ot+a/XpfR2E5nn0\nnm7xshuneoFz3dTLGqnymvWCqjFmaxx9vg98f5rtLwMvz00tPsYuWjqNmbw+taqMJ/aco71vhN+d\n6uXWVWWO8LIT9bKOU93Uyxqp8kr7b6imY+2q2yV8+ero6H3/R8RzUTtZpGO87MSpXuBcN/Wyhq7n\nnmF8bm05FfleTvYM886ZPrt1FEXJMDS520SO28U9Vy0EIjfSns/Ru6IomY8mdxu5o66CkjwPR84P\nsu9sv906iqJkEJrcbcTndfOlKyNfZPrZPmfdSFtRlPQm7ZN7uteubq6vJN/rYv+5fg53DKTYKv3j\nNd841Quc66Ze1tD13GOQ7rWrhbketkSXIdg+D6P3dI/XfONUL3Cum3pZw7b13J1OJqzR/KX1C8j1\nuHi3tY8T3YMptMqMeM0nTvUC57qplzXsXM/d0WRC7Wqpz8sddRVA5FurqSQT4jWfONULnOumXtbQ\nOvcM596rFuJ1CbtP9dLaO2y3jqIoaY4md4dQWZDDZ9eWY4Cf70/t6F1RlMxHk7uDuO/qRbgE3jze\nQ8fFUbt1FEVJYzS5O4glxbncurKMkIms964oijJX0j65Z1rt6v0bIwuKvfphN92DgWQqAZkXr1Tj\nVC9wrpt6WUPr3GOQabWry8t83FRbQiBk2PFB5+w7WCTT4pVqnOoFznVTL2tonXsMMrF2des1kVvx\n7Wruom84ub9fJsYrlTjVC5zrpl7W0Dr3GGRi7eraynw2VRcxHAzzy0Pnk2iVmfFKJU71Aue6qZc1\ntM49y9i6MTJ6/+Wh8wyOhmy2URQl3dDk7lCuWlzI+sUF9I+G2NXcZbeOoihphiZ3B/NAdPS+42An\nI8GwzTaKoqQTsyZ3EXlMRDpF5GCM9ltFxC8i+6I/fzuh7a9E5JCIHBSR7SKSl0z5TOdjVUWsqfRx\nYSjIq0e77dZRFCWNiGfk/jhw2yx9fmeM2Rj9+TsAEakC/gLYZIxZD7iB+xORnY5Mrl0VkfG592cO\ndBAIJT56z+R4pQKneoFz3dTLGrbVuRtjdgM9czy+B/CJiAfIB87O8TgxyfTa1RtrS6gty+P8QIA3\nj19I+HiZHq9k41QvcK6belnD6XXuN4rIARF5RUSuBDDGtAP/L3AGOAf4jTGvJ+l842R67apLhPs3\nRL61+vP9HYTCid1IO9PjlWyc6gXOdVMva6TKy5OEY+wBlhlj+kXkDuCXwBoRKQO2ACuAXuBZEXnQ\nGPPk5AN0dnaybdu28eeNjY00NjZSVVWFx+Oht7cXv98/5cRVVVW0t7dTUlISs322/VPV7vf7qa+v\nT8rxN5YLC3wu2vtGeP4Px7huce6c/fx+//jHQDvjM7n9zJkzUz6eOsHPqa8vgHA4zIoVKxzx95vY\nPvYaszs+k9snvvbtjM/k9hMnTpCXN/VyZDz7z4QYM/tIUESWA7uic+ez9T0NbAI+BdxmjNkW3f5V\n4OPGmD+fvE9TU5Opq6ub1WM6WlpaqK2tndO+qSTZXi8d6eKRt1tZUZbHf7+7DpeII7yShXpZx6lu\n6mWNRLz27NnzfkNDw6bp2hKelhGRxSKRTCMi10WP2U1kOubjIpIfbW8AmhM9X7by2TXlVOZ7OXVh\nmHfP9NmtoyiKw4mnFHI70ASsE5E2EdkmIg+LyMPRLvcAB0VkP/AD4H4T4V3gOSLTNh9Ez/U/U/Jb\nZAE5bhf3XL0QgJ/t+4h4PnEpipK9zDrnbozZOkv7D4Efxmj7HvC9uakpk7l9XQXb93Vw9Pwge89e\n5NqqYruVFEVxKGn/DdVsql31ed3cvX4BANvneCPtbIpXMnCqFzjXTb2soeu5xyDbalc31y+gIMfN\n/nP9HOrot7x/tsUrUZzqBc51Uy9rOL3O3TayrXa1IMfN5vpKAJ6ew+g92+KVKE71Aue6qZc1dD33\nGGTbGs0Ad69fSK7HxbutfZzoHrS0bzbGKxGc6gXOdVMva+h67so4JXke7qyrAOY+964oSmajyT1N\nufeqRXhdwu9O9XKmd9huHUVRHIYm9zSlosDL59aWY4isOaMoijIRTe5pzJc3LMIl8ObxHj66OGK3\njqIoDiLtk3u21a5OZElRLp9eVUbYwDMHOuPaJ5vjNRec6gXOdVMva2idewyyrXZ1MvdFlwN+7cNu\nugcDs/bP9nhZxale4Fw39bKG1rnHINtqVydTW+bjE8tLCIQMOz6YffSe7fGyilO9wLlu6mUNrXOP\nQbbVrk7H/dFb8e1q7qJveOYXisbLGk71Aue6qZc1tM5dicnaynw2VRcxHAzzi0Pn7dZRFMUBaHLP\nEB6Ijt53HjrPwGjIZhtFUexGk3uGsH5xIVctLqR/NMSu5i67dRRFsRlN7hnE1o2RypkdH3QyEgzb\nbKMoip2kfXLPttrVmfhYVRFrK/PpHQ7yytHuaftovKzhVC9wrpt6WUPr3GOQbbWrMyEi46P3Zw50\nEAhNHb1rvKzhVC9wrpt6WUPr3GOQbbWrs3FDbQm1ZXl0DQR44/iFKe0aL2s41Quc66Ze1tA69xhk\nW+3qbLhE2Br91urP93cQCl9+I22NlzWc6gXOdVMva2iduxI3t6wsY2lxDmf7Rth9auroXVGUzGfW\n5C4ij4lIp4gcjNF+q4j4RWRf9OdvJ7SVishzInJERJpF5IZkyivT43YJ910dGb1v39dB2JhZ9lAU\nJdOIZ+T+OHDbLH1+Z4zZGP35uwnbHwFeNcbUARuA5rlpTo/RpBWThjXlVOZ7OX1hmHfO+O3WURRl\nnpk1uRtjdgM9Vg8sIiXAJ4FHo8cZNcb0WjacgZP/9aec+Yv/xEcv/prw6OwrImYTOW4X9169EIiM\n3vWNUFGyi2TNud8oIgdE5BURuTK6bQVwHviJiOwVkR+LSEGSzocxhnPPvcbAuwfY96//A29d+0WO\n/v1/Y+Bka7JOkRBOqKm9va6SkjwPR88Psqf9IuAMr+lQL+s41U29rJEqL4lnRCciy4Fdxpj107QV\nA2FjTL+I3AE8YoxZIyKbgHeAm4wx74rII0CfMeZvJh9j586d5rvf/e7488bGRhobG6mqqsLj8dDb\n24vfP3VqYWF+ER0736Dlf/2CoWMt49vzP1ZP6RcbqHtgCzkF+TH3n+34mdD+7MEufvLHc6wr8/Ct\nTcWO89N2bdf2ubcfOHDg/YaGhk1TGklCcp+m72lgE+AB3jHGLI9uvxn4jjHmzsn7NDU1mbq6ulk9\npiMYDOJ2u/HvOUTrT3dybucbhIcit5zzlpdQde/tVD+4mcI1y+d0/LkSDAbxeDzzes7pGBgN8eDT\nhxgYDfFfvrCGdRV5jvCajFPiNRmneoFz3dTLGol47dmzJ2ZyT3haRkQWi4hEH18XPWa3MeYjoFVE\n1kW7NgCHEz3fZNrb2xERSj+2nqv+6a/51P4Xqf/P36Jo/RoCPX5O/4+nefvmB3h3y5/R/uwrhIbm\n516jTqmpLchxs6W+EojMvTvFazLqZR2nuqmXNWyrcxeR7UATsE5E2kRkm4g8LCIPR7vcAxwUkf3A\nD4D7zaWPA98AnhKRA8BG4B+S/ytcjre4kGV/ejc3/upxbnj1Uaof3Iy7IJ8L7+7ng2/8Pb/ZuJnD\nf/2PXGw+kWoVx/Cl9QvJ9bh4r7WPM33O/JaeoijJZdbPAsaYrbO0/xD4YYy2fUSmaOYdEaFk4xWU\nbLyCuv/rG5zb+SZtP92Jf18zZx59jjOPPkfJx66k5sEtLN7cgKfAZ4fmvFCS5+Guugp2HDzPy6eG\nuPkqu40URUk1WfENVU9hATVf2cwNrz7KjW88zrI/vRtPUQH+9w9x8K/+gbc2bubQt79P3wdH7VZN\nGfdctQivS9jTGeCRt8/w0pEujp4fYFSXBlaUjMR5VxdSTPH6tdT/52+x9m++xkcv/pq2J3fS+8eD\ntD7xC1qf+AXFV9dR89Bmlnzps3gKk1a5aTsVBV6+eOUCnv2gk5eOdAORJYFdArWleayuzGd1hY/V\nlfmsLPdRkOO2V1hRlISIq1om1SRSLdPb25vwkpkXm0/Q9rMXOfvsKwR6I/Xg7nwfS770GWoe3ELx\nxiuIXjOeV69kY4yh6fhHnB12c7xrkBPdQ7T6hwlP8xJYWpzLmgofqyp9rK6IJP5Snzdlbk6MFzjX\nC5zrpl7WSMRrpmqZtE/uySQ0NELHy2/R+tOdXHhn3/j2oivXUP2VzSxt/BzekiIbDZPPcDDMqZ4h\njncNcrx7iOPdg5zuGSYwTcavLPBGRvcV+ayOJv0FBV7Lb3yKoiSHjE7uqapd7T92mranXqT9mZcJ\n9ES+QODy5bL4Cw3UPLSF0k3rZ0xq6VxTGwiFOdM7HEn2XUOc6B7kRM8QQ4Gp8/PFuW5WVVya0lld\n4aOqJBeXxYSfzvGyC6e6qZc1UlXnnvbJvaWlhdra2iQbXSI8MkrHK7tpe+oFun/3x/HthWtXUP3Q\nZpbeczs5ZcVT9ku111yZq1fYGM72jXAsmuwjiX+QvpHQlL4+r4uV5b7LEv6y0jy87tjX7zMtXvOB\nU93UyxrqREtYAAAVLUlEQVSJeM2U3J33NuYwXLk5LPniZ1jyxc8wcKqNtp+9SPvTL9H/4SmO/M0j\nfPh//3cW3XUrNV/ZQtkNGzN2isIlQnVJHtUleXxqVRkQmcM/PxDgePdgdIQ/xLHuQboGAhzqGOBQ\nx8D4/l6XUFuWd9mUzoryPHxevXCrKKlAk7sFClZUs+6v/4w13/7XdL7+Nm1P7qTrrfc4t+N1zu14\nnYLVy6h+YDNVX77dbtV5QURYWJjDwsIcbqy9dEGodyjA8e5Ish+by2/vG4nO6Q/Bh5F+LoHqkjxW\nVfiodA2zyXuR1RU+inL1ZakoiaL/i+aAy+th8Z23svjOWxk8c4727S/Stn0XA8fPcPTvfsiH/+mf\nKbxlE4X/9gHKb7oWcWXF1wnGKfV52VTtZVP1pemqgdEQJydcuD3RPUjLhWHO9EZ+AJ49dhyARYU5\nrKn0XTaXX5GfukodRclENLknSP6yJaz5P/4Nq/79/0bXm020PvkC599s4uIb7/CHN97BV7uUmgc3\nU3XfneQurLBb1zYKctxctbiQqxYXjm8bDYY5fWGY492D7D3dSceIm5M9Q3T0j9LRP8rbpy+thFfm\n84yXZK6q9LGmIp/FRTkZOw2mKImS9hdUnVi7Ony2k+M/eY6u53/FcHsHAOJxs/DzN1P9lc1U3vIn\niNueuWYnxgsueYXChlb/MMe7ImWZJ6JTOQOjUy/cFuS4I8l+QnlmTUkeblfyEr5T4wXOdVMva2id\nexpiQiG63nqP1id3cv7132NCkQSVV72Y6ge+QPXWu8hbssBmS+djjOGji6PjFTpj9fgXhqYugpbj\nlmilTv74CH95WR45nuyaGlOyg4xO7ulSuzrc0UX70y/R9tSLDJ05G9nocrHgMzdS8+AWKj99Pa55\n+D3SJV7x0D0YuGwO/1hXZEpnMm6B2rK8y+bwV5X7yI9jiQWnxguc66Ze1tA69xikW+2qCYfp/t0f\naXvyBTpe3Y0JREafuUsWUL31C1Q/cBe+6sXz7mU3yfLqGw5yomeIE+Mj/CHaYiyxUFWceynZV/im\nXWLBqfEC57qplzW0zj1DEJeLyluuo/KW6xg538PZZ16h9cmdDJ5q48Q/PsaJ//ITKj/1cWoe3MyC\nz96Ey6t/IisU53m4ZmkR1yy9tEzEUCDEqZ7h8Xr849FKnfa+Edr7RvjtqUv3bZ+8xELeUIiqsMGT\nxHl8RZkPNHPYSO6CclZ87Sss//MH6PmXvbQ9uZOPXnqLrl830fXrJnIXVlB1/51Uf+UL5NdW2a2b\ntvi8buoXFVC/6NIqn5cvsTA4XpffNRCgayDAO2f6Lh3g7X2U5Hko83ko83kpz4/8O93z4jyP5aUX\nFCUVaHJ3ACJCxU3XUnHTtVzR3cvZ516l9ckXGDh2mpM/eIKTP3iCik/+CTUPbmHhbTfjytGa70Tx\nul2sqshnVUU+n18bKVENhSNLLExM+Ce7+ukbNfiHg/iHg5y+MDzjcV3CeKIvz7/0BnDp+aXH+V6X\nlnIqKUOTu8PIqShl+b+9n9p/cx+97x2g9ckX+OjFN+ne/Qe6d/+BnIpSqu6LjOYLVi2zWzejcLuE\nmtI8akovLbHQ0tJCdc0y/MNBLgwF6BmM/HthKEjPUIALg5HHkZ8AF0dCdA8G6B4MQPfQjOfLccul\nTwD5XsrHPwl4KfV5KPd5KYt+KsjTah/FImmf3EtKSuxWmJZEvUSEsus3UHb9Bq74+29ydsfrtD65\nk/7mE5z60VOc+tFTlN94LdUPbmbRHbfgzsudF69U4WQvt0soz48k3VWzfA9tNBSmd+jyN4KeoeCE\nN4HAeNtwMDz+ha3ZyPe6Im8E+dGk7/PiE8OS4e7Im0L0zaHU57X9+oCT/5ZOJFVeaV8tk00YY/Dv\nPUzrT3fy0S/fIDQUmSLwlhWz9N7bqfnKZgrXrbDZUomXoUAokvAHo28AY58IBgPjjy8MBbgwGJx2\nff1YFOe6Y38SmDBdpNcH0p+MLoXMxNrVuI5/cYCzz79O21Mv0Hfg0r1fS6+7OnLT77s+hTs/b969\n5op6xcYYQ/9oiAuD0amgaNLv7h+ldyR06Y1gMEDvcHDass/pcAkTkn7kwnCpb+KbwqXnBTnuuK8P\nOCFm05GJXgkldxF5DLgL6DTGrJ+m/VZgJ3Aquul5Y8zfTWh3A38E2o0xd013jmyqc08F/v1HaHvq\nBc7ueJ3QwCAAnuJClt5zGzUPbqaofrUtXlZQL+tM5xYKG/qGg5euCURH/j2TPgn0RK8PxIvXLeMj\n/8unhyJTQmXRN4lSn4eO9laWL1+e5N82cZz6t7Szzv1x4IfAEzP0+V2sxA18E2gGpt7RQkkKJRvq\nKNlQx7rvfZ2Pdr5J60934t97mDOPPceZx56j5Jp6ah7awuItDXarKinG7ZJIss33shLfjH0DoTC9\nw8FLyX/CdYGeCdNFvUMBBgPxXx/wCBTlXaQgx01hjpvCXDcFOW6KcjwU5Ea2TWyb2Kcwxz3jTV2U\n+Jk1uRtjdovI8rkcXESqgTuB/wf4d3M5hhI/noL8yJo1D3yBvkPHaHvyBc7ueA3/3sP49x6m+W8f\noajheryfv4WSa+rJX16lpXhZjNftYkFBDgsKcmbtOxQI0Ts04RPA4MSKocsriAIhM15BNBdyPa5I\nwh97E5jmDSDy3BPpk+u+rH8yF45LZ5I1AXWjiBwA2oFvGWMORbf/E/BtYMa7Snd2drJt27bx542N\njTQ2NlJVVYXH46G3txe/3z9lv6qqyBd7Zmqfbf9Utfv9/vG5NFv8CnNY+/ffZN3ffI2TP9/F2e0v\nMXTgKL07f0Pvzt8A4CkpouSaK8i7YiWyqhpf/So8FaXxHT8F7S0tLfMXnzjbwZmvL4BwODzvfoVA\noUBNAVStnX7/rgu9ePOLKSpfwFBI6Ojxc97fz1DQMBA0DAUMg0GD8eYxGAjjHxzh4nCQwaBhKGgY\nCYYZCYYj5aRzIM8dqS4q9uVQmOsmTwxegrjDAUpPDJLvcZHvFaoXVlDiy8GMDhIeHsDnEfI8Mn6R\neb7+vsPDw1Ne+/HuPxNxXVCNjtx3xZhzLwbCxph+EbkDeMQYs0ZE7gLuMMb8eXRe/ls6524vF4+c\n5MOnX0BOncW/t5mRzu4pffKqFlGy8QpKrqmP/GxYh6ewYJqjJRcnxguc6wXOdUvEyxjDcDBM/2iI\n/pEQA6Oh8cf90ccDI8HLto31GYhuS6RERIgsJV2Q46Yo1z1lamns08J0nyoKc9zkeax/Mc2xa8sY\nY/omPH5ZRH4kIpXATcDmaMLPA4pF5EljzIOJnnMi2Va7mghFdStZ9Zd/SmlpKcYYRs6dpzc6ZePf\n14x/XzPD7R0Mt3fQ8dJbkZ1EKFyznJJrogl/4xUU1a9O+rdknRgvcK4XONctES8Rwed14/O6WTCH\nMUXYGIYC4WjiDzIwGuJi9E3ivL+fkCtnwpvE2JtCMNJnJMjI0CiBoUEuBgMMBYO4gwE8wQCeYBB3\nIPLYHQxG/41uDwbwBCLbPKEAvnCIPBMiNxQkJxwkJxTEGwziCUX2dwWCuAIBXIFRGA1QeMt11P7o\ne3OOWSwSTu4ishjoMMYYEbkOcAHdxpjvAt+N9rmVyMg9qYkdcOTi++B8LxEhb+lCFi9dyOI7bwUi\nK1YOHD8zPkfv39dM36Fj9H94iv4PT9H+85cj++Z4KV6/NpLwo6P8gpU1Cd1O0OnxciJOdZurlzGG\n8Mjo+E9oeJTwyMik55e2XXoe2Xb580v7+0ZGyR0ZpWSW/uGR2S8WJ5Nw9N/TbRe4MQXHnzW5i8h2\n4FagUkTagO8BXgBjzD8D9wB/JiJBYAi438xj8Xwm1q6mkpm8xOWicO1yCtcup+q+OwAIj4zSd+j4\nhNH9YQaOteDfcwj/nkPj+3qKCyOJfuMV46P8vMXx34gkHeNlN05wM8Yw0tHF4Mk2Bk61MtR6jkD/\nIIwGIwl0LHFGE2loQrK+tC2asIfnN7lOhys357Ifd14Ortzc6OPIv668aFtutG3Cc5PjJejxEnB7\nGHV5GHF7GHG5GRIPQy43g+JmEDf9uBjAzUXjon5ZeUp+l7T/ElMmzjumkmR4Bfr66dt/hN69h+nb\n10zv3sOMnDs/pV/ukgXjI/vSa+op3lCHt7hwmiNmdrxSxXy5GWMY6exm8FTbeBIfPNkaeX6qbfyb\n0slgpuR66fnYT+6k5zP1z+V87wWW1tbE3j/Ha8vN7B07565kH97iQipu3kTFzZdeU8PnzuPfFx3d\n743M34+cO0/nufN0vrJ7vF/BmlpKNly6YFt85WpcubOX4impxRjDaNcFBk+2MnCqjcFTrZcS+an2\n8S/HTYe3vIT8FdUUrKjBV7uUi6EAlYsXjY9y3ROSbeT52Ih3fpPrYEsLpQ59o04FmtyVpJC3ZAF5\nS25h0e23ANH5+5Ot4yN7/95m+g5+yMCxFgaOtXD2uVcBEK+HovrVuFfX4Pnk9ZH5+9XLbBlBZTrG\nGALdvZHkfbKVwdNtDERH4AMnWwn1z5DAS4vIX1FD/spIEs9fWUP+8moKVlbjLb38+4ktLS0sy6Ik\n6lQ0uSspQVwuClfXUri6lqX33AZAeDTAxeYTly7Y7m2m/9hp+vYfgf1HuLDjVwC4C/Mj37odK8fc\neAV5SxfqF67iZLTHz+Cp1kjiPtl2WRIP9vXH3M9TXBgZga+sIX9FDQUrqyNJfEUNOWX6BfN0Q5O7\nMm+4crzjSyXwp3cDkQXQ/AeO0vKb3yOnzo2XY/b8fg89v98zvm/uworLyjFLNl4xZcSYTQR6+xg6\ndJyzfzwynrgHo9Mpgd6LMffzFBVERuArqsaT+Nho3Fteom+gGUTaJ/dMrPVNJU7z8hQVUHHTtbiv\nXDleQjfS2R2duz88ftF2pLObztfepvO1t8f3zV9Zc1nCL75yLW5ffOvax4ud8Qr09UfnwKMj8FOt\nDERH4oGeqd9YHMNdkB8ZdUeTeGQUXkP+impyKstSnsCd9hobI9u80r5aRsl8jDEMnm6/vP7+g6NT\nSufE46aofjUlG+vHSzIL1y5H3G6bzGcneHHg0hz4WPKO/hvo6Y25nzvfR/6K6gnTKJce5ywo1xF4\nlpDR1TJOqPWdDvWyxoz19yIUrKimYEU1S+/+HADhQJD+Iyfo3ds8ftG2/+gp+g4cpe/AUVqf+AUQ\nSYLFG+oo2XgFpdfUR9bSqV48r2uTBwcGo8m7fbyMcCyhj3ZdiLmfy5cbvWhZMyWJ5y6qJBQKpd3f\n0k6yzct5v6lF2tvbHVmHrF7WsOrl8noovmodxVetg69+EYgk0b4DR8dLMf17DzPUeo4LTXu50LR3\nfN+cyrLL18/ZeAU55dN/NI7XKzgwxODptgmlhJfKCadbw2f898jLIX959XgpYX50OqVgZQ25iypm\nrBrKlL/lfJFtXmmf3BVlDE9BPuU3XEP5DdeMbxs53zO+bk4k6R9mtOsC59/4F86/8S/j/fKXV1E8\nPrqvp3j92il3sgoNDjPY0h6tQomWEEaT+MhHXTG9XLk55NdWRRL38kgFyticeN6SBVr2qaQETe5K\nRpO7oJyFn72JhZ+9CYjM3w+diayKOb6kwoEjDJ5uZ/B0Ox/98g0AxO2msG4lUr2QjouRUfnw2c6Y\n55EcL/m1SyOj7ujUSf7KyOO8pQsdPe+vZCaa3JWsQkQio+jaKpZ88TMAhINBBj48fWmFzL3N9B85\nycVDx+DQsUv7etz4aqsiyXvlhCS+ogZf9SJN4Iqj0OSuZD0uT+RbskX1q6n5ymYgMgXTd/BDzry7\nh6Xr6yIj8OrFuBx4QU5RpiPtX6nZVruaKOoVH+78PMquuxpZu8yxS+s6LWZjqJc1UuWV9ldynPof\nT72soV7WcaqbelkjVV5pn9yDwbndhDfVqJc11Ms6TnVTL2ukyivtk3t7e7vdCtOiXtZQL+s41U29\nrJEqr7RP7oqiKMpU0j6579ixw26FaVEva6iXdZzqpl7WSJWXJvcUoV7WUC/rONVNvayhyV1RFEWJ\nG0cs+fvmm2+eB1rmsm9PT09leXl57IU9bEK9rKFe1nGqm3pZI0Gv2oaGhgXTNTgiuSuKoijJRadl\nFEVRMhBN7oqiKBlI2iR3EblNRI6KyHER+c407SIiP4i2HxCRax3idauI+EVkX/Tnb+fJ6zER6RSR\ngzHa7YrXbF7zHi8RqRGR34jIYRE5JCLfnKbPvMcrTi874pUnIu+JyP6o13+cpo8d8YrHy5b/j9Fz\nu0Vkr4jsmqYt+fEyxjj+B3ADJ4CVQA6wH6if1OcO4BVAgI8D7zrE61Zglw0x+yRwLXAwRvu8xytO\nr3mPF7AEuDb6uAj40CGvr3i87IiXAIXRx17gXeDjDohXPF62/H+MnvvfAT+b7vypiFe6jNyvA44b\nY04aY0aBp4Etk/psAZ4wEd4BSkVkiQO8bMEYsxvomaGLHfGKx2veMcacM8bsiT6+CDQDVZO6zXu8\n4vSad6Ix6I8+9UZ/Jldm2BGveLxsQUSqgTuBH8fokvR4pUtyrwJaJzxvY+qLPJ4+dngB3Bj9qPWK\niFyZYqd4sSNe8WJbvERkOXANkVHfRGyN1wxeYEO8olMM+4BO4FfGGEfEKw4vsOf19U/At4FwjPak\nxytdkns6swdYZoy5GvivwC9t9nE6tsVLRAqBHcBfGmP65uu8szGLly3xMsaEjDEbgWrgOhFZPx/n\nnY04vOY9XiJyF9BpjHk/1eeaSLok93agZsLz6ug2q33m3csY0zf2UdEY8zLgFZHKFHvFgx3xmhW7\n4iUiXiIJ9CljzPPTdLElXrN52f36Msb0Ar8BbpvUZOvrK5aXTfG6CdgsIqeJTN1+WkSenNQn6fFK\nl+T+B2CNiKwQkRzgfuCFSX1eAL4aver8ccBvjDlnt5eILBYRiT6+jkjMu1PsFQ92xGtW7IhX9HyP\nAs3GmH+M0W3e4xWPl03xWiAipdHHPuCzwJFJ3eyI16xedsTLGPNdY0y1MWY5kRzxa2PMg5O6JT1e\naXGbPWNMUES+DrxGpELlMWPMIRF5ONr+z8DLRK44HwcGgX/lEK97gD8TkSAwBNxvopfHU4mIbCdS\nGVApIm3A94hcYLItXnF62RGvm4CHgA+i87UA/yewbIKXHfGKx8uOeC0B/peIuIkkx2eMMbvs/v8Y\np5ct/x+nI9Xx0uUHFEVRMpB0mZZRFEVRLKDJXVEUJQPR5K4oipKBaHJXFEXJQDS5K4qiZCCa3BVF\nUTIQTe6KoigZiCZ3RVGUDOT/B/caJko1MKhmAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1335a54860>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(np.stack([train_loss_history, test_loss_history],1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "collapsed": false,
    "hidden": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f133fce6ac8>]"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX0AAAD1CAYAAAC87SVQAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f133dd0c3c8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(test_acc_history)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "collapsed": false,
    "hidden": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAARUAAAEGCAYAAABLrAHXAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nlYqIeKVKRUS8UqUiIl6pUhERr/4fy0Y6J/1YzBYAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f13401b66a0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "visualize_preds(mnistnet, mnist_test_loader, MNIST_CLASSES, MNIST_MEAN, MNIST_STD, 6)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true,
    "hidden": true
   },
   "source": [
    "## CIFAR"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "collapsed": false,
    "hidden": true
   },
   "outputs": [],
   "source": [
    "N_EPOCHS = 5\n",
    "cifarnet = LinearNetCIFAR().cuda()\n",
    "criterion = nn.CrossEntropyLoss().cuda()\n",
    "optimizer = optim.Adam(cifarnet.parameters(), lr=1e-4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "collapsed": false,
    "hidden": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1, TrainLoss: 2.147, TestLoss: 2.077, TestAcc: 0.383\n",
      "Epoch 2, TrainLoss: 2.092, TestLoss: 2.065, TestAcc: 0.400\n",
      "Epoch 3, TrainLoss: 2.069, TestLoss: 2.020, TestAcc: 0.439\n",
      "Epoch 4, TrainLoss: 2.053, TestLoss: 2.016, TestAcc: 0.442\n",
      "Epoch 5, TrainLoss: 2.041, TestLoss: 2.010, TestAcc: 0.445\n"
     ]
    }
   ],
   "source": [
    "train_loss_history = []\n",
    "test_loss_history = []\n",
    "test_acc_history = []\n",
    "for epoch in range(1, N_EPOCHS+1):\n",
    "    train_loss = train(cifarnet, cifar_train_loader, criterion, optimizer, epoch)\n",
    "    test_loss, test_acc = get_metrics(cifarnet, cifar_test_loader, criterion)\n",
    "    print('Epoch %d, TrainLoss: %.3f, TestLoss: %.3f, TestAcc: %.3f' % (\n",
    "        epoch, train_loss, test_loss, test_acc))\n",
    "    train_loss_history.append(train_loss)\n",
    "    test_loss_history.append(test_loss)\n",
    "    test_acc_history.append(test_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "collapsed": false,
    "hidden": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f133ff07a20>,\n",
       " <matplotlib.lines.Line2D at 0x7f1340004ef0>]"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD1CAYAAACrz7WZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJztnXd4XNWZuN8zRdKojmyrjporstywARcghODQDBuKEtIg\nIWuS0AIEstRsICQQWtiQTRYIkOQHJGGTAKGsaQklOLjLvctF3ajOqPfz+2PGsiyrzJU0unek732e\neTQz55yZd85cfXPn3G++q7TWCIIgCOMLm9kCgiAIwugjwV0QBGEcIsFdEARhHCLBXRAEYRwiwV0Q\nBGEcIsFdEARhHOIwWwDgww8/1JGRkcMa29nZicNhiZdxHOJlDPEyjlXdxMsYI/Fqbm6uXr58eVJ/\nbZZ4pZGRkeTm5g5rrNfrxe12j7LRyBEvY4iXcazqJl7GGIlXQUFB0UBtYb8sY8U3C8TLKOJlHKu6\niZcxQuUV9sG9s7PTbIV+ES9jiJdxrOomXsYIlVfYB/eysjKzFfpFvIwhXsaxqpt4GSNUXmEf3AVB\nEIQTkeAuCIIwDpHgLgiCMA4J6+Be39rJ6rI2szUEQRAshyXy3IdDV7fmzrcKKaxpISL2U740P8Vs\npeNISEgwW6FfxMsYVvUC67qJlzFC5RW2e+52m+LSOf4fZj2zvpzXdlaZbHQ8Ey2ndqSIl3Gs6iZe\nxpA89344b9ZkblyWDsCv15Syak+1yUbHmGg5tSNFvIxjVTfxMobkuQ/AgthWrlvqAeCJ1SX8fX+t\nyUZ+JlpO7UgRL+NY1U28jCF57oNw2dxkVp6WjgYe+2cRHx2sM1tJEATBVMZFcAf48oIUrlqUSreG\nn31wmH8d9pqtJAiCYBrjJrgDXLkwlS8vSKFbwwPvH2Z9ic9sJUEQBFMYMrgrpTKVUh8opXYppXYq\npW7up0+uUmqNUqpNKfWDftrtSqnNSqk3R0t8AFf+/dQ0LpubRGe35sd/P8TmsoZQPqUgCIIlCWbP\nvRO4TWudBywFblBK5fXpUwvcBDw2wGPcDOwetuUg9M0RVUpx7RIPF8+eQkeX5kfvHWT7kcZQPLUh\nL6sgXsawqhdY1028jGFanrvWukJrXRC43oA/SHv69KnUWm8AOvqOV0plABcBz46KcR/6yxFVSnHj\n6RmcP2sSbZ3d/PCdA+yubArF0xvysgLiZQyreoF13cTLGKHyMvQLVaVUDrAQWGdg2C+A24G4gTpU\nVlaycuXKntv5+fnk5+fj8XhwOBx4vV58vhPXzz0e/2dMY2Njv+03Lk2nvUvzwYE67ly1n9tOiSM7\n3nHc+KEef7jt3d3dZGZmhuzxh9ve3d2NzWYz7fkHaq+rq+vxMnN++rbDwNuX2X6pqalERkZa4v3r\n3X50GzN7fvq29972zZyfvu01NTU0Np64uhDs9jkQSms9aIeejkrFAh8BD2itXxmgz31Ao9b6scDt\ni4EVWuvrlVJnAz/QWl/cd9yaNWv0cE+zV1RURHZ29oDtXd2aB94/zOrDXuIi7Tx20UymTnIN67lG\n08ssxMsYVvUC67qJlzFG4lVQULBp+fLlp/bXFlS2jFLKCbwM/GGgwD4AZwBfUEodBl4CzlFKvWhg\n/Iix2xR3fS6bJZnxNLR1cfuqQoq9rWOpIAiCMOYEky2jgOeA3Vrrx408uNb6Lq11htY6B/gK8L7W\n+sphmY4Ap93Gfy6fyimeOHytndy+aj9lPqkmKQjC+CWYPfczgKvw73VvCVxWKKWuVUpdC6CUSlVK\nlQK3Aj9USpUqpeJD6G2YCIeNe8+dxoK0WGqbO7njrf182tButpYgCEJIGPKAqtZ6NaCG6HMEyBii\nz4fAhwbcRp0oh437z5vGXW8dYFdlE7ev2s9jF88kKSbCTC1BEIRRJ+x/oWo0R9TltPPABdOZNSWa\nioZ27lhVSG3zCRmcY+41VoiXMazqBdZ1Ey9jSD33ARhOjmhMhJ0HL5jOtEkuSn1t3PFWIb7W0S27\nOdFyakeKeBnHqm7iZQyp5z4Aw62FHB/l4KELp5PtjqKorpU73yqkoW30AvxEqx09UsTLOFZ1Ey9j\nSD33ARhJLWS3y8nDK2aQkRDJgZoW7n77AE3tXaZ7hRLxMoZVvcC6buJlDKnnHiImRfsDfGpcBHur\nmvnhOwdo6RidAC8IgmAWEz64AyTFRPDIihkkxTjZ+WkTP3r3IK2d3WZrCYIgDBsJ7gFS4yJ5ZMUM\nJkU72FrRyI/fO0h7lwR4QRDCEwnuvfAkRPHIhTNJiHKwqayBn/7jEB0S4AVBCEPCPriPdo5oVmIU\nD184g7hIO2uL63nowyK6uoMrrhZKr9FCvIxhVS+wrpt4GUPy3AcgFDmi0ya7+NmFM4iJsPPxIS+P\nfGQ8wE+0nNqRIl7GsaqbeBlD8twHIFQ5orOmRPPA+dNxOW18cKCOX6wupjvI8sih9Bop4mUMq3qB\ndd3EyxiS5z4AocxdzUuJ4SfnTSfSrnhnXy2/+qSUYOvfT7Sc2pEiXsaxqpt4GUPy3E1iflosPz5v\nGk674s3d1Ty9rizoAC8IgmAWEtyDYJEnnns/PxWHTfHKjip+t7FCArwgCJZGgnuQLM5M4O5zcrAp\neGnrp/xhy6dmKwmCIAyIBHcDnJnj5s6z/QH++U0V/HmrBHhBEKxJ2Af3sc5dPXt6IredlYUCnt1Q\nzqs7Ki3hFSziZQyreoF13cTLGJLnPgBm5K6eO3MyN52ZCcCTa8v4vz3VJ/SZaDm1I0W8jGNVN/Ey\nhuS5D4BZuasX5U7h+mX+Mwv+cnUJ7+6rOa59ouXUjhTxMo5V3cTLGJLnPgBm5q5eOieJaxano4HH\nPy7mgwN1lvAaDPEyhlW9wLpu4mUM0/LclVKZSqkPlFK7lFI7lVI399MnVym1RinVppT6gZGx4c4V\n81P4xilpdGt4+MPDrD7sNVtJEAQhqD33TuA2rXUesBS4QSmV16dPLXAT8NgwxoY9Xz85ha8uSKFb\nw4PvH2Zdsc9sJUEQJjhDBnetdYXWuiBwvQHYDXj69KnUWm8AOoyOHQ8opbj61DTy5ybR2a25/x+H\n2FXTMfRAQRCEEOEw0lkplQMsBNYZfaLBxlZWVrJy5cqe2/n5+eTn5+PxeHA4HHi9Xny+E/eGPR7/\n58Rg7UONH83281M1Nd5IPixt41dbGkhN9rEwI8EyfgA+n4+ioiLTnn+w9qNeZs5P33awzvbVl+7u\nbkv6HX0vzZ6fvu19tzGr+LW2tp6w7Qc7fjBUsD+jV0rFAh8BD2itXxmgz31Ao9b6MSNj16xZo3Nz\nc4Py6IvX67VUilO31vzi4xLe3leDy2njZxfMIC8lxmytHqw2X0cRL+NY1U28jDESr4KCgk3Lly8/\ntb+2oLJllFJO4GXgDwMF9lCMDQarvVk2pbj5zEzOmZ5IS0c3d79dyL6qZrO1erDafB1FvIxjVTfx\nMoZpee5KKQU8B+zWWj9u5MFHMjZYrJi7arcpvn+Gh89MddPc0c1dbxdyoMYaAd6K8wXiNRys6iZe\nxjAzz/0M4CrgHKXUlsBlhVLqWqXUtQBKqVSlVClwK/BDpVSpUip+oLGj+QKsmrt6pKKcuz6Xw7Ks\nBBraurjzrQMU1bWYrWXZ+RIv41jVTbyMESqvIQ+oaq1XA2qIPkeAjH6ahhw7nnHYFPcsz+G+9w6y\nsbSBO94q5OcXzcSTEGW2miAI45yw/4Wq1Ymw27j389NYkBZLbXMn/7GqkIqGNrO1BEEY50hwHwMi\nHTbuP28ac1JiqG7q4I5VhVQ2tputJQjCOEaC+xjhctr56fnTOSkpmiMN7dyxqpCaZvmhkyAIoSHs\ng3s41WiOibDz4AXTmT7ZRVl9G3euKsTbMrYBPpzmywpY1Qus6yZexpB67gMQbrmrcZEOHrpwBtmJ\nURR5W7nzrQPUt45dila4zZfZWNULrOsmXsaQeu4DEI65qwlRDh65cAYZCZEcrG3h7rcP0NTeZbqX\nmYiXcazqJl7GkHruAxCuuauJ0U4eWTGDtLgI9lU3c8/bB2gegwAfrvNlFlb1Auu6iZcxTKvnLoSO\nKTERPLJiJsmxTnZVNvGjdw/S2tlttpYgCOMACe4mkxIXwcMXzmRytJNtRxq5772DtEuAFwRhhEhw\ntwCehEgeXjEDd5SDgrIGfvKPQ3R0SYAXBGH4SHC3CFnuKB5eMYP4SDvrSur52QeH6eoOrhyzIAhC\nX8I+uI+n3NWpk1z87MIZxETYWX3YxyMfFY16gB9P8zUWWNULrOsmXsaQPPcBGG+5qzOnRPPgBdNx\nOW18cKCO//q4mO4gT6gSSq9QI17GsaqbeBlD8twHYDzmrs5OjuGn508n0mHj3f21/OpfpQR7xqxQ\neoUS8TKOVd3EyxiS5z4A4zV3dV5qLPefN40Iu+LNPdU8tbZsVAL8eJ2vUGFVL7Cum3gZQ/LcJyAL\n0+O49/PTcNoUr+6s4rcbykdtD14QhPGNBHeLc1pmPPcsz8Gu4H+3VfLi5iNmKwmCEAZIcA8DTs92\nc+fncrApeKHgCC9tlQAvCMLgSHAPEz47LZEfnJWNAn67oYJXdlSarSQIgoUZMrgrpTKVUh8opXYp\npXYqpW7up0+uUmqNUqpNKfWDPm0XKKX2KqUKlVJ3jqY8TKzc1c/PnMQtZ2YC8NTaMt7cXW34MSbS\nfI0GVvUC67qJlzFC5TXkCbKBTuA2rXWBUioO2KSUek9rvatXn1rgJuDS3gOVUnbg18C5QCmwQSn1\nep+xI2Ki5a5emDuFjm7Nrz4p5Zf/KsFpV5w/a7LpXiNFvIxjVTfxMoZpee5a6wqtdUHgegOwG/D0\n6VOptd4A9D2t0GKgUGt9UGvdDrwEXDIq5gEmWu4qwBfykvjO4nQAHv9nMe8X1gY9diLO10iwqhdY\n1028jGGJPHelVA6wEFgX5BAPUNLrdil9PhhGykTLXT3KF+encPUpaWjgkY+K+PiQ1xJew0W8jGNV\nN/EyRqi8glmWAUApFQu8DNyita4fTYnKykpWrlzZczs/P5/8/Hw8Hg8OhwOv14vP5zthnMfj/5wY\nrH2o8aFq9/l8dHZ2hvT5r5jnob2rmz9u+ZQH3z/EdQtiWZAUMeh4n89HUVGR6fPTX/tRL7Oev792\nsOb2BdDd3W1Jv6Pvpdnz07e97zZmFb/W1tYTtv1gxw+GCuZHMUopJ/Am8I7W+vFB+t0HNGqtHwvc\nXgbcp7U+P3D7LgCt9c96j1uzZo3Ozc0d0qM/ioqKyM7OHtbYUDJWXlprnllfzl+3V+K0KX583jRO\nzYg33cso4mUcq7qJlzFG4lVQULBp+fLlp/bXFky2jAKeA3YPFtgHYAMwUyk1VSkVAXwFeN3gYwiD\noJTi24vTuSTPf6D1x+8dZGt5g9lagiCYTDBr7mcAVwHnKKW2BC4rlFLXKqWuBVBKpSqlSoFbgR8q\npUqVUvFa607gRuAd/Adi/6y13hmi1zJhUUpx3bIMLjxpMm1dmv989yA7P200W0sQBBMZcs1da70a\nUEP0OQJkDNC2Clg1LLsgmGi5qwNhU4qbzsiko6ubvxfWcc/bB3h4xQxOSoox1StYxMs4VnUTL2NI\nPfcBmGi5q4NhtyluOyubz05109zRzd1vH+BATbPpXsEgXsaxqpt4GUPquQ/ARMtdHQq7TXHH53JY\nlp1AQ1sXd751gKK6FtO9hkK8jGNVN/EyhiXy3K3IRMtdDQaHTXHPOTmclhGPr7WTO1YVUuprNd1r\nMMTLOFZ1Ey9jSD13wRARdhs/+vxUTk6Ppbalk9v/r5CK+jaztQRBGCMkuI9jIh02fnzuNOamxFDd\n3MHtqwqpaekyW0sQhDFAgvs4x+W085Pzp5ObFM2nje08vqmBzeUNckYnQRjnSHCfAMRE2HnwgunM\nmOyisqWbO1YVct2re3hnXw3tnd1m6wmCEALCPrhPtNzV4RIb6eDRi2ZyRZ6bRJeDg7Wt/PyfxVz5\n0k6e31RBXXPfgp5ji9Xm6yhW9QLruomXMULlFVRtmVAzktoygnHau7r56GAdr+yo4kCNP03SaVN8\nbnoil81NYvrkaJMNBUEIhsFqywRdFdKqHK28aDWs7BXhcHDuzMl8fsYkth9p5JUdVawp8vHu/lre\n3V/LgrRYLp+bzOLMeOy2QX+cPKpeVp0vK3qBdd3Eyxih8gr7ZZmJlrs6Unp7KaWYnxbHfedO4/dX\n5HHZnCRcThtbKxq5972DrPzrLl7dUUlze+gzbMJhvqyGVd3EyxiS5y6ElLT4SK5blsEfvzqX7y7x\nkBIbQXl9O0+uLeNrf9rB02tLOdIgefKCEC5Y7zuKYCoxEXby5yVz6Zwk1hT7eHVHFduPNPLyjipe\n3VnF6dluLp+bxJyUGPzVoAVBsCIS3IV+sdsUZ+a4OTPHzb7qZv62o5IPD3pZfdh/mTUlmsvmJnHW\nVDdOu3wBFASrIf+VwpDMmhLN7Wfn8MKX5/C1k1NIiHKwr7qZhz8s4hv/u4s/bTmCr9WaRZkEYaIS\n9nvuEy13daSMxGtyjJOrT03nqyen8n5hLa/srKKorpXfbazgD5uPsHzGJC6bm0ROomtMvUKJVb3A\num7iZQzJcxcsh9aazeUNvLKjivUlx86ZfoonjsvmJnFqRjw2WZcXhJAhee4mMBG8lFIs8sSzyBNP\nibeVv+2s4t39tWwqa2BTWQOZCZFcNjeZ5TMScTntY+Y1mljVC6zrJl7GkDz3AZhouasjJVReme4o\nvndGJn/4yhyuOS2dKTFOSnxt/PJfJVz50k6eW19GVVP7mHuNFKt6gXXdxMsYofKy3seYENbERzm4\nYkEKl89LZvUhL6/urGR3ZTP/u62Sv2yv5Kypbi6bm8zs5JihH0wQhGEzZHBXSmUCzwMpgAZ+o7V+\nok8fBTwBrACagau11gWBtu8D1wTGbge+pbVuHc0XIVgPh01x9vREzp6eyO7KJl7ZUcnHh7x8eNB/\nyUuO4bK5SZyZ4x6zEgeCMJEIZs+9E7hNa12glIoDNiml3tNa7+rV50JgZuCyBHgSWKKU8gA3AXla\n6xal1J+BrwC/H80XIVib2ckx3HPOVCob23ljVxWr9tawq7KJXe83kRTj5JI5ScyNltLDgjCaDBnc\ntdYVQEXgeoNSajfgAXoH90uA57U/9WatUsqtlErr9RwupVQHEA2Uj+YLEMKH5NgIVi728LWFqfx9\nfy2v7qyi1NfGs+vLibTD+ZUlXDoniYyEKLNVBSHsMbTmrpTKARYC6/o0eYCSXrdLAY/WeqNS6jGg\nGGgB3tVavzts236YaLmrI8UKXi6nnX/LS+Ki2VPYWFrPKzuqKChr4PVd1by+q5olmfFcPjeZk9Nj\nTS9xYIX5GgiruomXMULlFXRwV0rFAi8Dt2it64fqHxiTiH+vfirgBf6ilLpSa/1i736VlZWsXLmy\n53Z+fj75+fl4PB4cDgderxefz3fC43s8Htxu96DtQ40PZXtsbKypzz9Q+9HrZs+Px+NhcWYCs+I0\nu8oU/yhuY01FG+tK6llXUk+2O5LL56VwyhRFS2ODKX5W3r48Hg+AJf18Pp8l5qdve+9+VvEDKCoq\nGtb4wQjqR0xKKSfwJvCO1vrxftqfBj7UWv8pcHsvcDZwJnCB1npl4P5vAEu11tf3Hj+SHzFNtNzV\nkWJ1L29LB/+3p4Y3dlVR2+IvaZAQ5eDi2VO4ePYUJkc7TfGyIlZ1Ey9jjMRrsB8xDZnnHsiEeQ7Y\n3V9gD/A68A3lZyngC6zVFwNLlVLRgcdZDuwe1qvohyNvvM/2x57Bt2U33Z3Wqm0y0XJqR8pRL7fL\nydcXpvLCV+Zw+2ezmTHZha+1kz9sPsJVL+3kkY+KKKxuHnMvK2JVN/Eyhpl57mcAVwHblVJbAvfd\nDWQBaK2fAlbhT4MsxJ8K+a1A2zql1F+BAvxZN5uB34yWfMmLr1Hz0QY+/cUL2KNduE+Zg3vxfBKX\nLMC9KA9HrORShytOu43Pz5zE8hmJ7Pi0iVd3VPJJkY+/76/l7/trmZcay+Vzk1ialSCplILQD8Fk\ny6wGBv3vCWTJ3DBA273AvcOyGwLPly+iMy6ajl0HaT5YQs3HG6n5eCMAym4nbs4MEhfPJ3HxAtxL\n5hOVMiUUGkIIUUoxLzWWeamxVDS08drOKt7eW8P2I41sP9JIalwEl85J4vxZk4mJGLzEgSBMJKy3\nAGWA9MvPo+OUk8jOzqatqpa69duoW78V77pt1G/fR/22vdRv20vRs38BwJWdTuLiBSQu8Qf8mJnZ\npmdjCMGTFhfJtUszuGpRGu/uq+FvO6uoaGjnqbVlPL+pgvNPmsyleUmkxUearSoIphPWwb03kUmT\nSL3obFIvOhuAzqYWfJt3UrcuEPA37qSlqJyWonLK//IWAM5JCSSeNq9nzz5hfi62iLE9YCcYJybC\nzmVzk/lCXhLrSny8sr2KbUcaeXVHFa/trGJZVgKXzU1mXqqcLUqYuIR9cB8oR9QR42Lymacy+Uz/\ngeTuzk4adh3o2bOvW7+Ntk+rqXxnNZXvrAbAFhVBwsl5PXv27lPn4kyIG1UvsxlPXnab4vRsN6dn\nuymsbubVnVV8cKCOfxX5+FeRjxmTXVw+N5nPThv+2aKsOl9gXTfxMobUcx9ltNa0FFdQt34rdeu3\n4V23jcZ9h47vpBRxs6eTuHg+7kDAd3lSxtRTMEZtcwdv7q7mjd3VPWeHmuRycHFeEhfnTsbtkm9m\nwvhhsFTIsA/uo5m72l7rw7txO3Xr/AHft2U3uuP4FMsoTwqJSxb4D9QuWUDsSVNRthP3CsdjTm0o\nGW2v9s5u3j9Qx6s7KjlU569T57Qrlk/3ny1q6qTgzhZl1fkC67qJlzFClece9sG9qKiI7OzsUTby\n09XShm/r7sCe/VbqNmyns77xuD6O+Fjcp84LLOXMJ+HkPOyuyJB6jYSJ5qW1ZktFI69sr2Rdr7NF\nLUyP4/K5SZyWOfjZoqw6X2BdN/Eyxki8xvWZmEKJ3RXJpKUnM2npyQDo7m4a9x7q2bOvW7eV1rJP\nqX5/DdXvrwFAOR0kLMjFlpuD69zP4D5tPhGTrLnWNxFQSrEwPY6F6XGU+vxni3pnXy2byxvYXN5A\nRkIkl85J4tyZk4Y8W5QghBMS3A2gbDbiZk8nbvZ0sq6+HICW0iPUbdiGd/126tZvo2FXId6NO2Dj\nDmpffBOAmJk5PQdpE5fMx5WVLlkcJpCREMWNp2fyzVPSeHtvDa/t8lel/NUnpfx+YwUXnjSZS+Yk\nkRwbYbaqIIwYCe4jxJWRiisjlfTLzgOgw9eAd9NODr/3T7r3FOHbvJOm/Ydp2n+Y0hdfByAyZcpx\nB2nj8qZjs+Ba4HglLtLBl+ancPncZP512MsrO6rYVdnEX7ZX8vKOSj6T4+byeXK2KCG8kYgyyjgT\n4kg6ZynN09PIzs6mu72D+u17e/Ltj6ZgHnnjfY688T4A9pho3KfO6dmzT1g4B0dMcAf8hOFjtynO\nmpbIWdMS2VPZxKs7q/jnwTo+OuTlo0NecpOiOTNFkerpJtIR9qcbFiYYYX9A1ev14na7R9lo5Azk\npbWmqbAosGa/De/6rTQfPr5wkLLbiZ83q2fPPnHxfCKTJoXUy2ys4lXV1M7ru6pZtaeahrYuACLt\nikUZ8SzLSmBJZjyJY1yZciCsMmd9ES9jjMRrXGfLjAdaP63Gu35bT8Cv37EPuo8/7Vz01Iye9Ev3\n4vnETM+SdfsQ0trZzd/31/L23hr29apCqfCfNnBZdgLLshLIdEfK+yCYxrgO7uMxd7WzsQlvwS7q\n1m3Fu34b3k076WpuOa5PxGS3vwJmIODHz50VVOmE8ThfoaSzsxNvWzdri+tZW+xjc3kDHV3H/mfS\n4yNZlhXPsmw3c1JixrRCpZXnTLyCR/LcB2A85q72pbuzk4adhf41+3X+FMz2qtrj+thckbgXziFx\nyXx/0D91Ho64Ew8IToT5Gk36erV0dLGptIE1xT7WFfuoDyzdAMRF2lmSGc/S7ARO9cQTHeIqleEy\nZ1ZhPHpJnnuYY3P4c+cTFuSS8+0v+0snFJUdd5C2aX8RtZ8UUPtJQWCQjbi86T1r9olLFhCVlmTu\nCxkHuJx2zpzq5sypbrq6Nbsqm1hT5GNNkY+y+jb+XljH3wvrcNoUC9JjWZaVwNLsBJJiJL1SGFsk\nuIchSimiczKIzsnA8+UVALRX11G3cXtPwK/ftpeGHftp2LGf4t/+FQBXZhoRC2aR8pNbJdCPAnbb\nsVrz31niocTb6g/0xT52fdrExtIGNpY28N+flDJziotlWQksy05g2iSXrNMLIUeC+zghYkoiKRec\nRcoFZwHQ1dyKb8vunqUc78bttJRU0FJSwZqNOznlhUeJn3eSydbji0x3FJnuKK5YkIK3pYP1JfV8\nUuRjU1kD+6tb2F/dwvMFR0iOdfr36LMSmJ8WO+yKlYIwGBLcxyn26Cgmnb6QSacvBEB3ddGw+wCb\nf/AQLVv2sO6S61nw1I9JPu9Mk03HJ26Xk/NmTea8WZNp6+xmS7l/nX5tkY/Kxg5e21XNa7uqiXba\nOC0jnmXZCZyWGU9cpPxLCqOD/b777jPbgdLS0vumTBn+KfCioqJG0Wb0sJKXstmITJ5M/Hmno6vq\nqN+6h4rX/oEjIRb3ojlm6wHWmq/ejNTLYVNkJESxNCuBy+clszgzHrfLQVNbF1VNHRR5W1l92Mdf\nt1eytaKRxvYu3FGOoAL9eJ2zUDHevCoqKiqmTZvW73mpwz5bRjCO1poDP/8thY89B0DWv3+R3Ptv\nkhIIJlBR38baYv86/baKRrp7/TvmJEb1HJA9KSl60OqVwsRkRKmQSqlM4HkgBdDAb7TWT/Tpo4An\ngBVAM3C11rog0OYGngXmBsb/u9Z6Te/xkuc+dvT2Kn/5HbZ//0F0ewdJy5ex4On7ccSaU08lHOYr\n1DS0dbKhpJ41xT42lNTT3HHsh2yTXA6WBA7ILkyPI9JhkzkzyHj0Giy4B3MkpxO4TWudBywFblBK\n5fXpcyEwM3D5DvBkr7YngLe11rnAAmC3Qf9BKSsrG7qTCYSDV3r++Sz+yy9xTkqg6h9rWHfJ9bSW\nV5ruZSXWGG1KAAAb4ElEQVTG0isu0sE5MyZxzzlT+cuV83jowulckjeF5FgntS2dvLW3hh+9e5Av\nvrCNe987yJ/XH6SupWPM/IJF3ktjhMpryI8LrXUFUBG43qCU2g14gF29ul0CPK/9XwPWKqXcSqk0\n/HvxZwFXB8a3A+2j+gqEEZG4ZAFL/+8ZNl35Axp27mfNhdew6IVHSZgvmTRm4rTbWOSJZ5EnnuuX\nZXCwtoU1xfWsLfKxr7rZn3IJ/L9dO46VQ8hOIDNByiEIfgx9F1BK5QALgXV9mjxASa/bpYH7OoEq\n4HdKqQXAJuBmrXXTMH2FEBAzNYOlb/6Gzf9+F3VrNrP+kuv8mTTnf8ZsNQH/7xqmT45m+uRorlyY\nSlVTO+uK63l/zxH21nWyq7KJXZVNPLehHE98JMuy/WmWY10OQbAWQQd3pVQs8DJwi9a6fqj+vR5/\nEfA9rfU6pdQTwJ3Af/buVFlZycqVK3tu5+fnk5+fj8fjweFw4PV68fl8Jzy4x+MBGLR9qPGhavf5\nfD1raVby8/l8FBUVndhe7yPl0VvpfvA3+FZ9TMHVdzLr3huY+t2v4vP5xsTvqJeZ89O3Hay5fc2L\nhjknx5DiyeKf+46wpsjLtuoOyurb+Ov2Sv66vZL4SDuLsxKYP9nBVFcHUQ4V9OOPpP3oe2mF9693\ne99tzCp+ra2tJ2z7wY4fjKCyZZRSTuBN4B2t9eP9tD8NfKi1/lPg9l7gbPwHUNdqrXMC938GuFNr\nfVHv8VJbZuwYyktrzcFf/J79Dz8DQNbVl5P701tCnkkTrvNlJn3depdD+KTIR3l9W0+b06Y4OT0u\nsFcfz5QQlkOw6pyNR68R1ZYJZMI8B+zuL7AHeB24USn1ErAE8AXW6lFKlSilTtJa7wWWc/xa/YhJ\nSLDm+UnD1UspxfTvfwtXjocdtzxI8e9fobm4gpOfvr/fQmRj5WUWVvWCE916l0P49uJ0SnxtrA0E\n+t2VTWworWdDaT2//Bf+cgjZbpZlxY96OQSrztlE8womFfJM4GNgO3A0N+tuIAtAa/1U4APgV8AF\n+A+ifktrvTEw/mT8qZARwMFAW13v55A8d2tSt34bBVffSUetl9jZ0znlxcdweVLM1hKGQV2vcggF\npfW09SpbfLQcwrLsBOalSjmEcGJcl/wdj7mrocSoV/PhUjZd+QOaCouJTJnCoucfIWHB6H8Qj5f5\nGkuG69bW2c3m8gbWFPnLFte2dPa0RTttnJbpP+vU4sx4YodRDsGqczYevUaa525pJlru6kgx6hWd\n48+kmXT6Ito+rWb9pdfz6dv/NN1rrLCqFwzfLdJhY2lWAt//TBZ//NpcnvjCLL66IIWcxCiaO7r5\n6KCXhz4s4ksvbuf2Vft5dUclFQ1tQz/wCL1CzUTzst7HmGA5nO54Tn3pv9j5Hw9T9r+r2Pytuzjp\n3hvJ+e5XJKc6zLEpxezkGGYnx/Ct09IpP1oOocjH9iONbCn3X55cW8bUxCiWBk4vOEvKIVgeCe5C\nUNginMz9xT1ET8tk/8+eZu99/03zwVJmP/h9qUkzjkiPj+TyuclcPjf5WDmEIh8bSus5VNfKobpW\n/rTlUyZFO1ia5Q/0JwfKIQjWQv4rhaBRSjH95m8Sne1h+80/peT5V2kpKefk3/w0pJk0gjkcLYdw\nzoxJdHR1s62ikbXF/uybqqYOVu2pYdWeGiIdNk7x+NMsl2TGm60tBJDgLhgm7dLPE+VJoeCbd1D9\nwTrWfuFaTnnhUVwZqWarCSHCabdxSkY8p2T0KocQSLMsrGnhk8B1BaTF2JlXVMTs5Bhyk2LIToyS\nX8qaQNgH94mWuzpSRssr8bR5LHvLX5OmcfcB1q74tj+T5uTZpnqNNlb1AvPcjiuHsCiNqqZ21gZO\nL7i1vJHypi7K99Xyzj7/SdyjHDZOSoomNymak5JjmJ0Uw+QY55h7W/W9NC3PfSyQPPfwpcNbz+Zr\n7qF29SZsrkgW/Po+UlZ81mwtwSTaO7sprGlhT1UTeyqb2FPVzJGGE2sFJsU4yU2OITcpmtnJMcyY\nEk2UrNsbRvLcTWAieXW3d7Dzjkcp+9OboBQn/ecN5Fz3VUOZNBNpvkYLq7r19apr6WBPZXNPwN9b\n1XxcrXoAm4Jpk1zkJsWQmxxNbnIMGQmRo5qREy7zZYQRlR+wOmVlZZasFzGRvGwRTuY+fhfRUzPY\n/+BT7L3/VzQdKiHvwduwOYPbxCbSfI0WVnXr65XocvaUJAbo1poSbyt7qprZXdnEnspmDte1UFjj\nv7y5xz8uNsLuX84J7OHnJseQEDX8kBUu8zVahH1wF6yBUorpN33Dn0lz008ofeE1WorLOfmZB3DG\nx5qtJ1gIm1JkJ7rITnRx/qzJALR0dLG/urlnD393ZTM1zR1sKmtgU1lDz9j0+AhOSjq2nDNtsosI\nKZfQLxLchVEl7ZLluDJSKPjG7dR8tIF1//ZdFr3wGNFZaWarCRbG5bQzPy2O+WlxPfdVNbX7g31l\nE7urmthf1Ux5fTvl9e18cMBfnsppU0yf7CI3OYbZydHkJsWQGhchP65DgrsQAtynzGXpqmf9mTR7\nD7F2xTUsev5R3Iv6np1REAYmKSaCpKkRfGaqG/CXND5c18LuQMDfU9VMcWB5Z09VM3/b6R+XEOXo\nWcY5+jcmwm7iKzEHCe5CSIjOTmfpm0+z5Zp7qPl4I+svv575v7qX1Is/Z7aaEKbYbcdSMC+ePQWA\nxrZO9gaC+9GA72vtZF1JPetK/OcUUkCmO4qM6G5Oba5mdnI0OYmucZ97H/bBfaLlro6UsfRyJsRx\nyh8fZ9cdj1L6xzfYcs09zPrh9Uy94esnfG2W+TKOVd3G0is20tHz4yrwn2zmSEN7IDPHf8D2QE0L\nxd5Wir3wSbn/bKCRDhszp7iYnRTj38NPjiYphCcwGQzJcxfCFq01h371IvseeBKAjK//G3kP/UfQ\nmTSCMBLau7o5WNPSk52zt6qJ8voTc+8nRzt71u1zk6OZOSUal9PayznjOhVyPOauhhIzvJRSTPve\nVUTneNj2vfsp/cMbtJQc4eRnfoozIc40r2CwqhdY181qXhF2G7nJMcyYFMmlc5IA8LZ0nLCcU9Pc\nwerDPlYf9p+v1KYgJ9FFbnJ0oJRCNJnuqFGvhhmq+bLOOzBMJlru6kgx0yv1387x16T5xu3U/HMD\n6/7tWha98CjR2ekyX8PAqm7h4OV2OVmSlcCSrGO596W+tp5Av6eyiYO1LT2XVXtqAP/JTI7l3vv3\n8BNdIyulIHnuwrjAvWgOS1c9S8GVP6Bx3yHWXuSvScNkyYUXzMOmFFnuKLLcUZwXyL1v7eymsPpo\nKqb/b1VTB5vLG9lc3tgzNjUuold2TgwzJruIsEApBQnuwpgTnZXGkjeeYsu3f0jNPzew/vIbSPvR\n9fDv1tvbEyYuUQ4bc1NjmZt6bMejpqmD3VVN7A3s4e8N1M450tDOhwe9ADiO5t73Cvjp8WOfey/B\nXTAFZ0Icp/zh5+y66zFKX3ydsrt/QUxTK1NvvEp+gCJYlskxTs6McXNmzrHc+6K61mPZOVVNFNe1\nsjcQ+F/bVQ1AfKT9uLz7k5KiiRvG+WmNMOSjK6UygeeBFEADv9FaP9GnjwKeAFYAzcDVWuuCXu12\nYCNQprW+ePT0hXDG5nQw59E7iJmayd6f/Jp9DzxF86Ey8h6WTBohPLDbFNMmu5g22cWKQMJfU3sX\n+wLLOUfLKdS1dLK+pJ71gdx7gIyESHKTY8iJbCMUhyiC+Q/qBG7TWhcopeKATUqp97TWu3r1uRCY\nGbgsAZ4M/D3KzcBuYNRP0yK5vsawmpdSiqk3fJ3uKQkcuOMxSv/4Bs3F5Sx87sGeTBozsdp89caq\nbhPdKybCzsL0OBam+7dfrTWVjR3+ImmBPfzCmmZKfW2U+tpYmhHDFSHwGDK4a60rgIrA9Qal1G7A\nA/QO7pcAz2t/0vxapZRbKZWmta5QSmUAFwEPALeO9gtwu92j/ZCjgngZY/qXL2byzGkUfPN2aldv\nYu3F3+WUFx8jOjvdVC+rzhdY1028jkcpRUpcBClxEZw9PRGAjq5uDtX6l3NS40Lz4ylDh3SVUjnA\nQmBdnyYPUNLrdmngPoBfALcD3YSAzs7OUDzsiBEvY3R2duJelMeyVc8Qe9JUmvYfZu2Ka6jbuN10\nL6tiVTfxGhqn3caspGi+kJfEorTQnH846IVNpVQs8DJwi9a6fqj+gTEXA5Va601KqbMH6ldZWcnK\nlSt7bufn55Ofn4/H48HhcOD1evH5fCeM83g8lJWVkZCQMGD7UOND1e7z+cjLyzPt+Qdq9/l8PV9P\nzZyfvu3FxcU9Xun/80PK7n6CpnXb2JD/PaY/dCuO0xeY4mfV7Qugu7ubqVOnWuL9691+dBsze376\ntvfe9s2cn77tBw4cICoqaljjByOo8gNKKSfwJvCO1vrxftqfBj7UWv8pcHsvcDZwE3AV/nX7KPxr\n7q9ora/sPX4k5QeKioos+YMJ8TJGX6/ujk523/M4Jc//DYCZd32XaTd9Y8wzaaw6X2BdN/Eyxki8\nBis/MOSyTCAT5jlgd3+BPcDrwDeUn6WAT2tdobW+S2udobXOAb4CvN83sAtCf9icDvIe/g9OuvdG\nUIr9P3uaHbc8QHd7h9lqghAWBLMscwb+ve/tSqktgfvuBrIAtNZPAavwp0EW4k+F/NboqwoTDaUU\nU6/7GtHZHrbecB9l/7uKltIj/kwa96gnXgnCuCKYbJnV+EsiD9ZHAzcM0edD4EMDboIAQMqKz7Lk\nlV9T8M07qP1XAWsv/o4/kyYnw2w1QbAs5hdAGCETPafWKOHqlbAwj6WrniE2dxpNhcWsWfEd6tZv\nM93LTKzqJl7GCJVX2Ad3yak1Rjh7uTJSWfrG00z53FI6ar1s+NJNlL/6ruleZmFVN/EyRqi8wj64\nWyl3tTfiZYxgvRxxMSx64REyv3kZ3W3tbLvuPg781+8I1UlnrDpfYF038TJGqLzCPriXlZWZrdAv\n4mUMI142h4O8h35A7o9v8mfSPPwM228OTSaNVecLrOsmXsYIlVfYB3dhYqKUIue7X2Hh736G3RVF\n+Z9XseHLt9BeF9Tv6wRh3CPBXQhrUi44i8V/+x8iU6ZQt2Yzay/+Dk2HSs3WEgTTkeAuhD0JC3JZ\nuuoZ4vJm0HygmLUXfZu6dVvN1hIEU5HgLowLXJ4Ulrz+JFPOWUZHrY/1X7qJ8ldCm0kjCFYm7IP7\nRMtdHSnj2csRG8Oi5x8m61v56PYOtl1/H4U//+2IMmmsOl9gXTfxMobkuQ/ARMtdHSnj3cvmcDD7\nwVvJ/cnNoBSFjz7L9u/9hO62dlO9QoFV3cTLGJLnPgATLXd1pEwEL6UUOd/+Mot+/5A/k+avb/sz\naWpPLJs6ll6jjVXdxMsYkuc+ABMtd3WkTCSv5PM/w+LXniQydQp1a7f4M2kOlgw9MMReo4VV3cTL\nGJLnLgjDIGH+SSxb9Sxxc2bSfLCEtRd9m9q1W4YeKAhhjgR3YdwTlZ7Mktf+h6TPn05HXT0brriZ\n8r++bbaWIIQUCe7ChMARG8PC3z9E1sov+jNpbryf/Y8+G7KaNIJgNhLchQmDzeEg74Fbyf3pLWCz\nceDnv2X79+4fdiaNIFiZsA/uEy13daSIF+RccwWLfv8w9mgX5X99hw1X3Ex7jdd0L6NY1U28jCF5\n7gMw0XJXR4p4+Uk+7wyWvPY//kyadVtZe9G3aTpQbLqXEazqJl7GkDz3AZhouasjRbyOET/vJJa9\n9Rxxc2fSfLjMn0nzyWbTvYLFqm7iZQzJcx+AiZa7OlLE63ii0pJY8tqTJJ17Bh3eBjZ8+WbK/vyW\n6V7BYFU38TKGaXnuSqlMpdQHSqldSqmdSqmb++mjlFK/VEoVKqW2KaUWBTtWEMzGERPNot8/RPY1\nX0J3dLL9pp+w/+FnJJNGCGuC2XPvBG7TWucBS4EblFJ5ffpcCMwMXL4DPGlgrCCYjrLbmf3T7zP7\ngVv9mTT/9Tu2XX+fZNIIYYtjqA5a6wqgInC9QSm1G/AAu3p1uwR4Xvt3ddYqpdxKqbQgxwqCZche\n+UVcWWlsvfZeKl59D2/hYdQVK3AmxON0x+FMiMOREIczMR5nQhz2qEizlQWhX4YM7r1RSuUAC4F1\nfZo8QO+iHaWB+yqCGEtlZSUrV67suZ2fn09+fj4ejweHw4HX68XnO7Hok8fjARi0fajxoWr3+Xx0\ndnaa9vwDtft8PoqKikyfn/7aj3qZOT8AzMog66kfUf4fP6dl+372bH/ihLFHsUVFYI+LRcW6sMfF\nYk+IwR4Xgy0uhkRPGhGTEuhw2mlz2LDHx2KPi8Ee72/PnJozIv/u7m7Aetv/0ffSattX323MKn6t\nra0nbPvBjh8MFey6olIqFvgIeEBr/UqftjeBh7TWqwO3/wHcobXeONRYgDVr1ujc3NygPPri9Xot\nmeIkXsawolfrp9UU/vYv2Oqb6fA10OFtoMNXT+fR6956dGfXsB/f5orEmRCH0937W8Gx685e3xAc\nR+9z+2/bIpyWnDOw5nsJ49OroKBg0/Lly0/try2oPXellBN4GfhDf8EZKAMye93OCNwXzNgRYcU3\nC8TLKFb0ikqZwty7rhuwXWtNV3MLHd6GYwHfV98T+Dt8DXR6G3p9MAQudf4PiO6WNtpa2mg7Um3Y\nze6KwpkY718iSog79uHgjiPCHX/8h4Tbv5Tkvz8Om9PQF3bDWPG9hInnNeS7rJRSwHPAbq314wN0\nex24USn1ErAE8GmtK4IcOyKOLn1YDfEyRjh6KaVwxETjiIkGT4qhx9Va09XUfCzo93xA1B//IeFr\nOO6bgv92I10trXS1tEJ5peHXZI+JPv74gfvYN4LeHxI9xxkCbY6EWGxBvEfh+F6aSai8gnnEM4Cr\ngO1KqaO1Uu8GsgC01k8Bq4AVQCHQDHxrsLFa61Wjo+/PEc3Ozh6thxs1xMsYE81LKYUjNgZHbAyu\njFRDY7XWdDU2c3jnbpJj4wf5plB/3DcF/zeHRrqamulqaqa17FPD3vbY6OOXknoC/7EPiQanIvvU\nk3FlpmF3WeeA80TbxoLJllkNqCH6aOCG4YwVBMEYSikccTE405KINxgUdHc3nY3NgW8Kxy8h9ftN\nofdxBl8jXY3NdDUO/cFwNLsiMnkyrqw0XFnpuLLSiA78dWWmE+VJDuqbgDA8ZGYFYQKhbDac8bE4\n42OBNENjdXc3nQ1N/XxTOLaE1FHro67wELqyjpbSI7RV1tBWWYN3444TXex2otKTe4J/dM+HgP8D\nIDJ5Mv6VXWE4SHAXBCEolM3Wk8VDdvqA/YqKisjOzkZ3ddFaUUVLcQXNxeW0FFfQUlxOS4n/dtuR\nalpKKmgpqYB/FZzwOLaoCFyZ/r386F57/0c/CJzu+FC+3LBHgrsgCCFB2e24MlJxZaQy6fSFJ7R3\nt7XTUnrk+MB/9IOgpIKOWh9N+4to2n9iDjiAIz722FJPZp+ln8w07NFRoX6Jlibsg/tEq9E8UsTL\nGFb1Auu6Betli4wgZnoWMdOz+m3vbGjq2cs/Gvybj34IFJXTWd9Iw479NOzY3+/4iKRJx63zM8VN\nTW4Vrqw0otJTQp4SGiyheh+D/hFTKBnJj5gEQZh4aK1pr67zL+v0DvpH/5Z9iu4YuJSustuJTEs6\ndoC395p/dnrYrPeP+EdMVmai5a6OFPEyhlW9wLpuY+GllCIyaRKRSZNwL5pzQnt/6/1NRWW0BZaB\n2o5U01p6hNbSI/DJiY8/luv9Zua5W5qJlrs6UsTLGFb1Auu6WcGrv/X+owd6wVrr/abluQuCIIw3\nhlzvb2w6Icun99KP0fV+/x5/+piu90twFwRB6IMjNoa4vBnE5c04oU1rTUeN91iwL+mz5l96hPaq\nWtqravFt2nnig9tsRKUn9wT77hkZZN/4zdF/DaP+iIIgCOMYpRQRUxKJmJKIe9GJ5x7SXV20Hqk+\nfqmnuIKWEv/f1oqq49b7489dBjeOvqcEd0EQhFFE2e24PCm4PCmwbID8/rJPe5Z6GiJCk5UT9sE9\n3HN9xxrxMoZVvcC6buI1OLbICGKmZRIzzV8l3ev1huZ5QvKoY8hEq9E8UsTLGFb1Auu6iZcxQuUV\n9sG9s3PgHyqYiXgZQ7yMY1U38TJGqLzCPriXlZWZrdAv4mUM8TKOVd3Eyxih8gr74C4IgiCcSNgH\n95dfftlshX4RL2OIl3Gs6iZexgiVlwT3ECFexhAv41jVTbyMIcFdEARBCBpLlPz9xz/+UQX0X6Fn\nCGpra6dMmjSpepSVRox4GUO8jGNVN/Eyxgi9spcvX57UX4MlgrsgCIIwusiyjCAIwjhEgrsgCMI4\nJGyCu1LqAqXUXqVUoVLqzn7alVLql4H2bUqpRRbxOlsp5VNKbQlcfjRGXr9VSlUqpXYM0G7WfA3l\nNebzpZTKVEp9oJTapZTaqZS6uZ8+Yz5fQXqZMV9RSqn1SqmtAa8f99PHjPkKxsuU/8fAc9uVUpuV\nUm/20zb686W1tvwFsAMHgGlABLAVyOvTZwXwFqCApcA6i3idDbxpwpydBSwCdgzQPubzFaTXmM8X\nkAYsClyPA/ZZZPsKxsuM+VJAbOC6E1gHLLXAfAXjZcr/Y+C5bwX+2N/zh2K+wmXPfTFQqLU+qLVu\nB14CLunT5xLgee1nLeBWSqVZwMsUtNb/BGoH6WLGfAXjNeZorSu01gWB6w3AbsDTp9uYz1eQXmNO\nYA4aAzedgUvfzAwz5isYL1NQSmUAFwHPDtBl1OcrXIK7ByjpdbuUEzfyYPqY4QVweuCr1ltKqRPP\n5msOZsxXsJg2X0qpHGAh/r2+3pg6X4N4gQnzFVhi2AJUAu9prS0xX0F4gTnb1y+A24HuAdpHfb7C\nJbiHMwVAltZ6PvDfwN9M9rE6ps2XUioWeBm4RWtdP1bPOxRDeJkyX1rrLq31yUAGsFgpNXcsnnco\ngvAa8/lSSl0MVGqtN4X6uXoTLsG9DMjsdTsjcJ/RPmPupbWuP/pVUWu9CnAqpaaE2CsYzJivITFr\nvpRSTvwB9A9a61f66WLKfA3lZfb2pbX2Ah8AF/RpMnX7GsjLpPk6A/iCUuow/qXbc5RSL/bpM+rz\nFS7BfQMwUyk1VSkVAXwFeL1Pn9eBbwSOOi8FfFrrCrO9lFKpSikVuL4Y/5zXhNgrGMyYryExY74C\nz/ccsFtr/fgA3cZ8voLxMmm+kpRS7sB1F3AusKdPNzPma0gvM+ZLa32X1jpDa52DP0a8r7W+sk+3\nUZ+vsDjNnta6Uyl1I/AO/gyV32qtdyqlrg20PwWswn/EuRBoBr5lEa8vAtcppTqBFuArOnB4PJQo\npf6EPzNgilKqFLgX/wEm0+YrSC8z5usM4Cpge2C9FuBuIKuXlxnzFYyXGfOVBvw/pZQdf3D8s9b6\nTbP/H4P0MuX/sT9CPV9SfkAQBGEcEi7LMoIgCIIBJLgLgiCMQyS4C4IgjEMkuAuCIIxDJLgLgiCM\nQyS4C4IgjEMkuAuCIIxDJLgLgiCMQ/4/3WMdgxxyU64AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1335f39a90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(np.stack([train_loss_history, test_loss_history],1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "collapsed": false,
    "hidden": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f133fcf79b0>]"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD1CAYAAACrz7WZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\no97ZbycAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f133fd5cc50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(test_acc_history)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "collapsed": false,
    "hidden": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAATkAAAEGCAYAAAAezeKJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ncgqFwtf4XxXpBwl8UBIvAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1335d924e0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "visualize_preds(cifarnet, cifar_test_loader, CIFAR_CLASS_NAMES, CIFAR_MEAN, CIFAR_STD, 6)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true
   },
   "source": [
    "# Experiments"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true,
    "hidden": true
   },
   "source": [
    "## Load/Save Weights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true,
    "hidden": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true,
    "hidden": true
   },
   "source": [
    "## Visdom Web Server"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true,
    "hidden": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true,
    "hidden": true
   },
   "source": [
    "## Experiment Class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "collapsed": true,
    "hidden": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import os\n",
    "import torch\n",
    "import visdom\n",
    "import shutil\n",
    "import sys\n",
    "from pathlib import Path\n",
    "\n",
    "\n",
    "class Experiment():\n",
    "    def __init__(self, name, root):\n",
    "        self.name = name\n",
    "        self.root = os.path.join(root,name)\n",
    "        self.epoch = 1\n",
    "        self.best_val_loss = sys.maxsize\n",
    "        self.best_val_loss_epoch = 1\n",
    "        self.weights_dir = os.path.join(self.root, 'weights')\n",
    "        self.history_dir = os.path.join(self.root, 'history')\n",
    "        self.results_dir = os.path.join(self.root, 'results')\n",
    "        self.latest_weights = os.path.join(self.weights_dir, 'latest_weights.pth')\n",
    "        self.latest_optimizer = os.path.join(self.weights_dir, 'latest_optim.pth')\n",
    "        self.best_weights_path = self.latest_weights\n",
    "        self.best_optimizer_path = self.latest_optimizer\n",
    "        self.train_history_fpath = os.path.join(self.history_dir, 'train.csv')\n",
    "        self.val_history_fpath = os.path.join(self.history_dir, 'val.csv')\n",
    "        self.test_history_fpath = os.path.join(self.history_dir, 'test.csv')\n",
    "        self.loss_history = {\n",
    "            'train': np.array([]),\n",
    "            'val': np.array([]),\n",
    "            'test': np.array([])\n",
    "        }\n",
    "        self.error_history = {\n",
    "            'train': np.array([]),\n",
    "            'val': np.array([]),\n",
    "            'test': np.array([])\n",
    "        }\n",
    "        self.viz = visdom.Visdom()\n",
    "        self.visdom_plots = self.init_visdom_plots()\n",
    "\n",
    "    def init(self):\n",
    "        print(\"Creating new experiment\")\n",
    "        self.init_dirs()\n",
    "        self.init_history_files()\n",
    "\n",
    "    def resume(self, model, optim, weights_fpath=None, optim_path=None):\n",
    "        print(\"Resuming existing experiment\")\n",
    "        if weights_fpath is None:\n",
    "            weights_fpath = self.latest_weights\n",
    "        if optim_path is None:\n",
    "            optim_path = self.latest_optimizer\n",
    "\n",
    "        model, state = self.load_weights(model, weights_fpath)\n",
    "        optim = self.load_optimizer(optim, optim_path)\n",
    "\n",
    "        self.best_val_loss = state['best_val_loss']\n",
    "        self.best_val_loss_epoch = state['best_val_loss_epoch']\n",
    "        self.epoch = state['last_epoch']+1\n",
    "        self.load_history_from_file('train')\n",
    "        self.load_history_from_file('val')\n",
    "\n",
    "        return model, optim\n",
    "\n",
    "    def init_dirs(self):\n",
    "        os.makedirs(self.weights_dir)\n",
    "        os.makedirs(self.history_dir)\n",
    "        os.makedirs(self.results_dir)\n",
    "\n",
    "    def init_history_files(self):\n",
    "        Path(self.train_history_fpath).touch()\n",
    "        Path(self.val_history_fpath).touch()\n",
    "        Path(self.test_history_fpath).touch()\n",
    "\n",
    "    def init_visdom_plots(self):\n",
    "        loss = self.init_viz_train_plot('loss')\n",
    "        error = self.init_viz_train_plot('error')\n",
    "        summary = self.init_viz_txt_plot('summary')\n",
    "        return {\n",
    "            'loss':loss,\n",
    "            'error':error,\n",
    "            'summary':summary\n",
    "        }\n",
    "\n",
    "    def init_viz_train_plot(self, title):\n",
    "        return self.viz.line(\n",
    "            X=np.array([1]),\n",
    "            Y=np.array([[1, 1]]),\n",
    "            opts=dict(\n",
    "                xlabel='epoch',\n",
    "                ylabel=title,\n",
    "                title=self.name+' '+title,\n",
    "                legend=['Train', 'Validation']\n",
    "            ),\n",
    "            env=self.name\n",
    "        )\n",
    "\n",
    "    def init_viz_txt_plot(self, title):\n",
    "        return self.viz.text(\n",
    "            \"Initializing.. \" + title,\n",
    "            env=self.name\n",
    "        )\n",
    "\n",
    "    def viz_epochs(self):\n",
    "        epochs = np.arange(1,self.epoch+1)\n",
    "        return np.stack([epochs, epochs],1)\n",
    "\n",
    "    def update_viz_loss_plot(self):\n",
    "        loss = np.stack([self.loss_history['train'],\n",
    "                         self.loss_history['val']],1)\n",
    "        window = self.visdom_plots['loss']\n",
    "        return self.viz.line(\n",
    "            X=self.viz_epochs(),\n",
    "            Y=loss,\n",
    "            win=window,\n",
    "            env=self.name,\n",
    "            opts=dict(\n",
    "                xlabel='epoch',\n",
    "                ylabel='loss',\n",
    "                title=self.name+' '+'loss',\n",
    "                legend=['Train', 'Validation']\n",
    "            ),\n",
    "        )\n",
    "\n",
    "    def update_viz_error_plot(self):\n",
    "        error = np.stack([self.error_history['train'],\n",
    "                         self.error_history['val']], 1)\n",
    "        window = self.visdom_plots['error']\n",
    "        return self.viz.line(\n",
    "            X=self.viz_epochs(),\n",
    "            Y=error,\n",
    "            win=window,\n",
    "            env=self.name,\n",
    "            opts=dict(\n",
    "                xlabel='epoch',\n",
    "                ylabel='error',\n",
    "                title=self.name+' '+'error',\n",
    "                legend=['Train', 'Validation']\n",
    "            )\n",
    "        )\n",
    "\n",
    "    def update_viz_summary_plot(self):\n",
    "        trn_loss = self.loss_history['train'][-1]\n",
    "        val_loss = self.loss_history['val'][-1]\n",
    "        trn_err = self.error_history['train'][-1]\n",
    "        val_err = self.error_history['val'][-1]\n",
    "        txt = (\"\"\"Epoch: %d\n",
    "            Train - Loss: %.3f Err: %.3f\n",
    "            Test - Loss: %.3f Err: %.3f\"\"\" % (self.epoch,\n",
    "            trn_loss, trn_err, tst_loss, tst_err))\n",
    "        window = self.visdom_plots['summary']\n",
    "        return self.viz.text(\n",
    "            txt,\n",
    "            win=window,\n",
    "            env=self.name\n",
    "        )\n",
    "\n",
    "    def load_history_from_file(self, dset_type):\n",
    "        fpath = os.path.join(self.history_dir, dset_type+'.csv')\n",
    "        data = np.loadtxt(fpath, delimiter=',').reshape(-1, 3)\n",
    "        self.loss_history[dset_type] = data[:,1]\n",
    "        self.error_history[dset_type] = data[:,2]\n",
    "\n",
    "    def append_history_to_file(self, dset_type, loss, error):\n",
    "        fpath = os.path.join(self.history_dir, dset_type+'.csv')\n",
    "        with open(fpath, 'a') as f:\n",
    "            f.write('{},{},{}\\n'.format(self.epoch, loss, error))\n",
    "\n",
    "    def save_history(self, dset_type, loss, error):\n",
    "        self.loss_history[dset_type] = np.append(\n",
    "            self.loss_history[dset_type], loss)\n",
    "        self.error_history[dset_type] = np.append(\n",
    "            self.error_history[dset_type], error)\n",
    "        self.append_history_to_file(dset_type, loss, error)\n",
    "\n",
    "        if dset_type == 'val' and self.is_best_loss(loss):\n",
    "            self.best_val_loss = loss\n",
    "            self.best_val_loss_epoch = self.epoch\n",
    "\n",
    "    def is_best_loss(self, loss):\n",
    "        return loss < self.best_val_loss\n",
    "\n",
    "    def save_weights(self, model, trn_loss, val_loss, trn_err, val_err):\n",
    "        weights_fname = self.name+'-weights-%d-%.3f-%.3f-%.3f-%.3f.pth' % (\n",
    "            epoch, trn_loss, trn_err, val_loss, val_err)\n",
    "        weights_fpath = os.path.join(self.weights_dir, weights_fname)\n",
    "        torch.save({\n",
    "                'last_epoch': self.epoch,\n",
    "                'trn_loss': trn_loss,\n",
    "                'val_loss': val_loss,\n",
    "                'trn_err': trn_err,\n",
    "                'val_err': val_err,\n",
    "                'best_val_loss': self.best_val_loss,\n",
    "                'best_val_loss_epoch': self.best_val_loss_epoch,\n",
    "                'experiment': self.name,\n",
    "                'state_dict': model.state_dict()\n",
    "            }, weights_fpath )\n",
    "        shutil.copyfile(weights_fpath, self.latest_weights)\n",
    "        if self.is_best_loss(val_loss):\n",
    "            self.best_weights_path = weights_fpath\n",
    "\n",
    "    def load_weights(self, model, fpath):\n",
    "        print(\"loading weights '{}'\".format(fpath))\n",
    "        state = torch.load(fpath)\n",
    "        model.load_state_dict(state['state_dict'])\n",
    "        print (state['trn_err'], state['val_loss'], state['val_err'])\n",
    "        print(\"loaded weights from experiment %s (last_epoch %d, trn_loss %s, trn_err %s, val_loss %s, val_err %s)\" % (\n",
    "                  self.name, state['last_epoch'], state['trn_loss'],\n",
    "                    state['trn_err'], state['val_loss'], state['val_err']))\n",
    "        return model, state\n",
    "\n",
    "    def save_optimizer(self, optimizer, val_loss):\n",
    "        optim_fname = self.name+'-optim-%d.pth' % (epoch)\n",
    "        optim_fpath = os.path.join(self.weights_dir, optim_fname)\n",
    "        torch.save({\n",
    "                'last_epoch': self.epoch,\n",
    "                'experiment': self.name,\n",
    "                'state_dict': optimizer.state_dict()\n",
    "            }, optim_fpath )\n",
    "        shutil.copyfile(optim_fpath, self.latest_optimizer)\n",
    "        if self.is_best_loss(val_loss):\n",
    "            self.best_optimizer_path = optim_path\n",
    "\n",
    "    def load_optimizer(self, optimizer, fpath):\n",
    "        print(\"loading optimizer '{}'\".format(fpath))\n",
    "        optim = torch.load(fpath)\n",
    "        optimizer.load_state_dict(optim['state_dict'])\n",
    "        print(\"loaded optimizer from session {}, last_epoch {}\"\n",
    "              .format(optim['experiment'], optim['last_epoch']))\n",
    "        return optim\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true
   },
   "source": [
    "# Examples\n",
    "\n",
    "Links To Pytorch Implementations of cool new models"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true,
    "hidden": true
   },
   "source": [
    "## CNNs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "hidden": true
   },
   "source": [
    " * [VGG](https://github.com/pytorch/vision/blob/master/torchvision/models/vgg.py)\n",
    " * [ResNet](https://github.com/pytorch/vision/blob/master/torchvision/models/resnet.py)\n",
    " * [InceptionNet](https://github.com/pytorch/vision/blob/master/torchvision/models/inception.py)\n",
    " * [SqeezeNet](https://github.com/pytorch/vision/blob/master/torchvision/models/squeezenet.py)\n",
    " * [DenseNet](https://github.com/pytorch/vision/blob/master/torchvision/models/densenet.py)\n",
    " * [FCDenseNet (Tiramisu)](https://github.com/bfortuner/pytorch_tiramisu/blob/master/tiramisu-pytorch.ipynb)\n",
    " * [Sub-pixel CNN (superresolution)](https://github.com/pytorch/examples/tree/master/super_resolution)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "hidden": true
   },
   "source": [
    "## GANs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "hidden": true
   },
   "source": [
    "* [DCGAN](https://github.com/pytorch/examples/tree/master/dcgan)\n",
    "* [Wasserstein GAN](https://github.com/martinarjovsky/WassersteinGAN)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true,
    "hidden": true
   },
   "source": [
    "## RNNs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true,
    "hidden": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true,
    "hidden": true
   },
   "source": [
    "## Other"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "hidden": true
   },
   "source": [
    "* [Pytorch Examples](https://github.com/pytorch/examples)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true,
    "hidden": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.0"
  },
  "latex_envs": {
   "LaTeX_envs_menu_present": true,
   "autocomplete": true,
   "bibliofile": "biblio.bib",
   "cite_by": "apalike",
   "current_citInitial": 1,
   "eqLabelWithNumbers": true,
   "eqNumInitial": 1,
   "hotkeys": {
    "equation": "Ctrl-E",
    "itemize": "Ctrl-I"
   },
   "labels_anchors": false,
   "latex_user_defs": false,
   "report_style_numbering": false,
   "user_envs_cfg": false
  },
  "toc": {
   "colors": {
    "hover_highlight": "#DAA520",
    "running_highlight": "#FF0000",
    "selected_highlight": "#FFD700"
   },
   "moveMenuLeft": true,
   "nav_menu": {
    "height": "192px",
    "width": "254px"
   },
   "navigate_menu": true,
   "number_sections": true,
   "sideBar": true,
   "threshold": 4,
   "toc_cell": false,
   "toc_section_display": "block",
   "toc_window_display": false,
   "widenNotebook": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}