Notebook2-AutoEncoder-MSELoss-Compressionfactor2.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Numpy\n",
    "import numpy as np\n",
    "\n",
    "# OS\n",
    "import os\n",
    "\n",
    "# Pytorch related libraries\n",
    "import torch # for pytorch\n",
    "import torch.nn as nn # for Core neural network class in pytorch\n",
    "import torch.optim as optim # for using the inbuilt pytorch optimzation functions\n",
    "from torch.autograd import Variable \n",
    "import torch.nn.functional as F #for using different inbuilt functions of pytorch\n",
    "\n",
    "# Torchvision\n",
    "import torchvision \n",
    "from torchvision import datasets # for various datasets\n",
    "import torchvision.transforms as transforms # for dataset transformation\n",
    "\n",
    "# Matplotlib\n",
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import cv2 # for image manipulation\n",
    "\n",
    "# cpickle\n",
    "import _pickle as cPickle\n",
    "\n",
    "# Set random seed for reproducibility\n",
    "SEED = 21\n",
    "np.random.seed(SEED)\n",
    "torch.manual_seed(SEED)\n",
    "if torch.cuda.is_available():\n",
    "    torch.cuda.manual_seed(SEED)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Preprocessing and transforming data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_transform = transforms.Compose([transforms.RandomHorizontalFlip()\n",
    "                                      ,transforms.RandomRotation(10)\n",
    "                                      ,transforms.ToTensor()\n",
    "                                     ])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "batch_size_num = 64"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Files already downloaded and verified\n"
     ]
    }
   ],
   "source": [
    "dataset = torchvision.datasets.CIFAR10(root='./data', train=True, download=True, transform=data_transform)\n",
    "loader = torch.utils.data.DataLoader(dataset, batch_size=batch_size_num,num_workers=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "50000"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(dataset)"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "1. Calculate mean and std of the pixel values in images and nomalize it while transforming"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def calculate_img_stats(loader):\n",
    "    mean = 0.\n",
    "    std = 0.\n",
    "    nb_samples = 0.\n",
    "    for imgs,_ in loader:\n",
    "        batch_samples = imgs.size(0)\n",
    "        imgs = imgs.view(batch_samples, imgs.size(1), -1)\n",
    "        mean += imgs.mean(2).sum(0)\n",
    "        std += imgs.std(2).sum(0)\n",
    "        nb_samples += batch_samples\n",
    "\n",
    "    mean /= nb_samples\n",
    "    std /= nb_samples\n",
    "    return mean,std"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "cifar10_mean,cifar10_std = calculate_img_stats(loader) # Note : only calculated on train dataset as will be simialr for test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([ 0.4735,  0.4640,  0.4295]) tensor([ 0.2200,  0.2165,  0.2151])\n"
     ]
    }
   ],
   "source": [
    "print(cifar10_mean,cifar10_std)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_transform = transforms.Compose([transforms.RandomHorizontalFlip(),\n",
    "                                      transforms.RandomRotation(10),\n",
    "                                      transforms.ToTensor(),\n",
    "                                      transforms.Normalize(cifar10_mean, cifar10_std)\n",
    "                                     ])\n",
    "\n",
    "test_transform = transforms.Compose([transforms.ToTensor(),\n",
    "                                     transforms.Normalize(cifar10_mean, cifar10_std)\n",
    "                                    ])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Files already downloaded and verified\n",
      "Files already downloaded and verified\n"
     ]
    }
   ],
   "source": [
    "trainset = torchvision.datasets.CIFAR10(root='./data', train=True,download=True, transform=train_transform)\n",
    "testset = torchvision.datasets.CIFAR10(root='./data', train=False, download=True, transform=test_transform)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50000, 32, 32, 3)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "trainset.train_data.shape"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "2. Explore the classes and images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'label_names': ['airplane', 'automobile', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck'], 'num_vis': 3072, 'num_cases_per_batch': 10000}\n",
      "0 : airplane\n",
      "1 : automobile\n",
      "2 : bird\n",
      "3 : cat\n",
      "4 : deer\n",
      "5 : dog\n",
      "6 : frog\n",
      "7 : horse\n",
      "8 : ship\n",
      "9 : truck\n"
     ]
    }
   ],
   "source": [
    "def load_classes():\n",
    "    path = './data/cifar-10-batches-py/'\n",
    "    file = 'batches.meta'\n",
    "    \n",
    "    with open(path + file, 'rb') as fin:\n",
    "        dict = cPickle.load(fin)\n",
    "        print(dict)\n",
    "        return dict['label_names']\n",
    "\n",
    "def print_classes(label_names):\n",
    "    for i in range(10):\n",
    "        print(str(i) + \" : \" + label_names[i])\n",
    "\n",
    "classes = load_classes()\n",
    "print_classes(classes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "def show_images(images, cols = 1, titles = None):\n",
    "    assert((titles is None)or (len(images) == len(titles)))\n",
    "    n_images = len(images)\n",
    "    if titles is None: titles = ['Image (%d)' % i for i in range(1,n_images + 1)]\n",
    "    fig = plt.figure()\n",
    "    for n, (image, title) in enumerate(zip(images, titles)):\n",
    "        a = fig.add_subplot(cols, np.ceil(n_images/float(cols)), n + 1)\n",
    "        if image.ndim == 2:\n",
    "            plt.gray()\n",
    "        image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)\n",
    "        plt.imshow(image)\n",
    "        a.set_title(title)\n",
    "    fig.set_size_inches(np.array(fig.get_size_inches()) * n_images)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 4320x2880 with 10 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_images(trainset.train_data[0:10])"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "3. Create the data split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainloader = torch.utils.data.DataLoader(trainset, batch_size=batch_size_num,shuffle=True, num_workers=2)\n",
    "testloader = torch.utils.data.DataLoader(testset, batch_size=4,shuffle=False, num_workers=2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Define the Autoencoder model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Autoencoder(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(Autoencoder, self).__init__()\n",
    "        \n",
    "        # batch_size : 64\n",
    "        # Input size to the Autoencoder : [batch_size, 3, 32, 32]\n",
    "        # Output size from the Autoencoder : [batch_size, 3, 32, 32]\n",
    "        self.encoder = nn.Sequential(\n",
    "            # 3 input channels(RGB), 12 output channels, 3x3 square convolution\n",
    "            nn.Conv2d(in_channels = 3, out_channels = 12, kernel_size = 3, stride=1, padding=1), # Output size : [batch_size, 12, 32, 32]\n",
    "            nn.ReLU(),\n",
    "            nn.Conv2d(in_channels = 12, out_channels = 24, kernel_size = 3, stride=1, padding=1), # Output size : [batch_size, 24, 32, 32]\n",
    "            nn.ReLU(),\n",
    "            nn.Conv2d(in_channels = 24, out_channels = 48, kernel_size = 3, stride=2, padding=1), # Output size : [batch_size, 48, 16, 16]\n",
    "            \n",
    "        )\n",
    "        self.decoder = nn.Sequential(\n",
    "            # For Upscaling :\n",
    "            # 2x2 kernels can only learn nearest pixel upscaling.\n",
    "            # 3x3 kernels can do bilinear but will require asymmetric padding.\n",
    "            # But 4x4 can do bilinear again without asymmetrical padding.\n",
    "          \n",
    "            nn.ConvTranspose2d(48, 24, 4, stride=2, padding=1), # Output size: [batch_size, 24, 32, 32]\n",
    "            nn.ReLU(),  #TODO : try with LeakyReLU while decoding\n",
    "            nn.ConvTranspose2d(24, 12, 3, stride=1, padding=1), # Output size: [batch_size, 12, 32, 32]\n",
    "            nn.ReLU(),\n",
    "            nn.ConvTranspose2d(12, 3, 3, stride=1, padding=1), # Output size: [batch_size, 3, 32, 32]\n",
    "            nn.Sigmoid()\n",
    "        )\n",
    "\n",
    "    def forward(self, x):\n",
    "        encoded = self.encoder(x)\n",
    "        decoded = self.decoder(encoded)\n",
    "        return encoded, decoded"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Define the train method and start training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "# moves variables to CUDA\n",
    "def get_torch_vars(x):\n",
    "    if torch.cuda.is_available():\n",
    "        x = x.cuda()\n",
    "    return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train(model, num_epochs=5, batch_size=batch_size_num, learning_rate=1e-3):\n",
    " \n",
    "    criterion = nn.MSELoss() \n",
    "    optimizer = optim.Adam(model.parameters(),lr=learning_rate,weight_decay=1e-5)\n",
    "    \n",
    "    outputs_store = []\n",
    "    for epoch in range(num_epochs):\n",
    "        running_loss = 0.0\n",
    "        for i, data in enumerate(trainloader, 0):\n",
    "            inputs, labels = data\n",
    "            inputs = get_torch_vars(inputs)\n",
    "\n",
    "            # ============ Forward ============\n",
    "            encoded, outputs = model(inputs)\n",
    "            loss = criterion(outputs, inputs)\n",
    "            \n",
    "            # ============ Backward ============\n",
    "            loss.backward()\n",
    "            optimizer.step()\n",
    "            optimizer.zero_grad()\n",
    "        print('Epoch:{}, Loss:{:.4f}'.format(epoch+1, float(loss)))\n",
    "        outputs_store.append((epoch, inputs, outputs),)\n",
    "\n",
    "    print('Finished Training')\n",
    "    print('Saving Model...')\n",
    "    if not os.path.exists('./weights'):\n",
    "        os.mkdir('./weights')\n",
    "    torch.save(model.state_dict(), \"./weights/autoencoder_16*16.pkl\")\n",
    "    return outputs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_model():\n",
    "    autoencoder = Autoencoder()\n",
    "    desc_model(autoencoder.encoder, autoencoder.decoder)\n",
    "    if torch.cuda.is_available():\n",
    "        autoencoder = autoencoder.cuda()\n",
    "        print(\"Model moved to GPU.\")\n",
    "    return autoencoder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "def desc_model(encoder, decoder):\n",
    "    print(\"==================== Encoder Network ====================\")\n",
    "    print(encoder)\n",
    "    print(\"==================== Decoder Network ====================\")\n",
    "    print(decoder)\n",
    "    print(\"\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "==================== Encoder Network ====================\n",
      "Sequential(\n",
      "  (0): Conv2d(3, 12, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
      "  (1): ReLU()\n",
      "  (2): Conv2d(12, 24, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
      "  (3): ReLU()\n",
      "  (4): Conv2d(24, 48, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1))\n",
      ")\n",
      "==================== Decoder Network ====================\n",
      "Sequential(\n",
      "  (0): ConvTranspose2d(48, 24, kernel_size=(4, 4), stride=(2, 2), padding=(1, 1))\n",
      "  (1): ReLU()\n",
      "  (2): ConvTranspose2d(24, 12, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
      "  (3): ReLU()\n",
      "  (4): ConvTranspose2d(12, 3, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
      "  (5): Sigmoid()\n",
      ")\n",
      "\n",
      "\n",
      "Model moved to GPU.\n",
      "Epoch:1, Loss:0.8401\n",
      "Epoch:2, Loss:1.0056\n",
      "Epoch:3, Loss:0.7173\n",
      "Epoch:4, Loss:0.7093\n",
      "Epoch:5, Loss:0.7552\n",
      "Epoch:6, Loss:0.8635\n",
      "Epoch:7, Loss:0.7572\n",
      "Epoch:8, Loss:0.9590\n",
      "Epoch:9, Loss:1.1580\n",
      "Epoch:10, Loss:0.8998\n",
      "Epoch:11, Loss:0.6794\n",
      "Epoch:12, Loss:0.6477\n",
      "Epoch:13, Loss:1.0175\n",
      "Epoch:14, Loss:0.7755\n",
      "Epoch:15, Loss:0.7225\n",
      "Epoch:16, Loss:0.7136\n",
      "Epoch:17, Loss:0.6250\n",
      "Epoch:18, Loss:0.8309\n",
      "Epoch:19, Loss:0.8848\n",
      "Epoch:20, Loss:0.8012\n",
      "Epoch:21, Loss:0.8820\n",
      "Epoch:22, Loss:0.6927\n",
      "Epoch:23, Loss:0.9335\n",
      "Epoch:24, Loss:0.9821\n",
      "Epoch:25, Loss:0.8047\n",
      "Epoch:26, Loss:0.7202\n",
      "Epoch:27, Loss:0.5993\n",
      "Epoch:28, Loss:0.8688\n",
      "Epoch:29, Loss:0.8886\n",
      "Epoch:30, Loss:0.9679\n",
      "Epoch:31, Loss:0.8313\n",
      "Epoch:32, Loss:1.0187\n",
      "Epoch:33, Loss:1.1861\n",
      "Epoch:34, Loss:0.7124\n",
      "Epoch:35, Loss:0.9257\n",
      "Epoch:36, Loss:0.7864\n",
      "Epoch:37, Loss:0.8225\n",
      "Epoch:38, Loss:0.9162\n",
      "Epoch:39, Loss:0.7878\n",
      "Epoch:40, Loss:0.7891\n",
      "Epoch:41, Loss:0.9798\n",
      "Epoch:42, Loss:0.6735\n",
      "Epoch:43, Loss:0.8454\n",
      "Epoch:44, Loss:0.8830\n",
      "Epoch:45, Loss:0.8585\n",
      "Epoch:46, Loss:0.8830\n",
      "Epoch:47, Loss:0.9909\n",
      "Epoch:48, Loss:0.7967\n",
      "Epoch:49, Loss:0.7694\n",
      "Epoch:50, Loss:0.8404\n",
      "Epoch:51, Loss:0.8990\n",
      "Epoch:52, Loss:0.8395\n",
      "Epoch:53, Loss:0.8025\n",
      "Epoch:54, Loss:0.8917\n",
      "Epoch:55, Loss:0.8099\n",
      "Epoch:56, Loss:0.8058\n",
      "Epoch:57, Loss:0.6722\n",
      "Epoch:58, Loss:0.7622\n",
      "Epoch:59, Loss:1.0971\n",
      "Epoch:60, Loss:0.9139\n",
      "Epoch:61, Loss:0.8043\n",
      "Epoch:62, Loss:1.0237\n",
      "Epoch:63, Loss:0.9093\n",
      "Epoch:64, Loss:0.8170\n",
      "Epoch:65, Loss:0.6474\n",
      "Epoch:66, Loss:0.8180\n",
      "Epoch:67, Loss:0.9911\n",
      "Epoch:68, Loss:0.7682\n",
      "Epoch:69, Loss:0.8193\n",
      "Epoch:70, Loss:0.7621\n",
      "Epoch:79, Loss:0.7341\n",
      "Epoch:80, Loss:0.5804\n",
      "Epoch:81, Loss:0.8517\n",
      "Epoch:82, Loss:0.7776\n",
      "Epoch:83, Loss:0.8608\n",
      "Epoch:84, Loss:0.8270\n",
      "Epoch:85, Loss:0.9743\n",
      "Epoch:86, Loss:0.8546\n",
      "Epoch:87, Loss:0.8543\n",
      "Epoch:88, Loss:0.9510\n",
      "Epoch:89, Loss:0.9007\n",
      "Epoch:90, Loss:0.8076\n",
      "Epoch:91, Loss:0.8629\n",
      "Epoch:92, Loss:0.9437\n",
      "Epoch:93, Loss:0.9451\n",
      "Epoch:94, Loss:0.7810\n",
      "Epoch:95, Loss:0.5591\n",
      "Epoch:96, Loss:0.6913\n",
      "Epoch:97, Loss:0.9505\n",
      "Epoch:98, Loss:0.7832\n",
      "Epoch:99, Loss:0.7820\n",
      "Epoch:100, Loss:0.8645\n",
      "Finished Training\n",
      "Saving Model...\n"
     ]
    }
   ],
   "source": [
    "# Create model\n",
    "autoencoder = create_model()\n",
    "\n",
    "#train\n",
    "max_epochs = 100\n",
    "outputs = train(autoencoder, max_epochs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "def imshow(img):\n",
    "    npimg = img.cpu().numpy()\n",
    "    plt.axis('off')\n",
    "    plt.imshow(np.transpose(npimg, (1, 2, 0)))\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loading checkpoint...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GroundTruth:    cat  ship  ship airplane\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(\"Loading checkpoint...\")\n",
    "autoencoder.load_state_dict(torch.load(\"./weights/autoencoder_16*16.pkl\"))\n",
    "dataiter = iter(testloader)\n",
    "images, labels = dataiter.next()\n",
    "print('GroundTruth: ', ' '.join('%5s' % classes[labels[j]] for j in range(4)))\n",
    "imshow(torchvision.utils.make_grid(images))\n",
    "\n",
    "images = Variable(images.cuda())\n",
    "\n",
    "encodings, decoded_imgs = autoencoder(images)\n",
    "imshow(torchvision.utils.make_grid(decoded_imgs.data))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GroundTruth:    dog  bird  deer airplane\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GroundTruth:  truck  frog  frog   dog\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAB6CAYAAACvHqiXAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJztnX90lNWZx5+UTNuhJqdMKmNNTp2sSWvSEmushCNxiVuwQgWOsosU7QIVXaWLbdGqZ0tXPCtbpbVWXJUWrFCFo7biEbokK3gIW/EYW7AmHkIbNKOb2A42Y88EmdaEZv+4z/Peb2ZuJjMhv3h5Pv/cm+e9877vzLxz89zn183r6+sjRVEU5dTnQ2N9A4qiKMrwoBO6oiiKT9AJXVEUxSfohK4oiuITdEJXFEXxCTqhK4qi+ASd0BVFUXyCTuiKoig+QSd0RVEUn5A/mhfLy8vTtFRFUZQc6evry8tmnGroiqIoPkEndEVRFJ+gE7qiKIpP0AldURTFJ4yqU9SFWPrvXPx3niweSxIR0cJF3/Rk0xdcSEREr7T9NxERTZ36FTjLRdz+FWQfyek+4u8dIyKi6//pfCtMxomIqKb2Uk8ULjf3MblkGhERVdd+0Tu2b9c+c9+zLrHnmDS0/5l5eVn5QCwXQL+C2/8C2W5u7wRZAbe13L4IxyZyWwyyZMq5iIgi3C7gdgcce9Vxn0u4bQPZSyljPgf973N7AGTSj4Ds/pRzJKHfyO2TINvF7buOexwnVHFbAbIYty0gq+Y233Gsw3HeALc9Ga4ddvSbM4x3UXKR7RcWmnZykZV1J/g+kunjCvjZ7O21x+Jdpp1SZWXFJaY9As9TNNp/fHCiPVZWztcJ2t/lkba/mfEJuHn+kAqCpg2H7CH53JJw30fj/Y8REQX4CynmDzAYsMei/MV0wznk+2tL/T3kgGroiqIoPmHMNfSZ55q2M/6mJ2tmVSBY+CNP9nTTH4iIaPNz5u+SKquSLVzGr4P/0rPnXElERMunbs/qPh7ZeA8REXUnrE4Ti35ARET3/uy7dmDYqMJP7X2biIh6u+yh4gtnmM6krC45vKBqcKtpquA+mvdy57CVVfIi5vJfm3YrnEI0QfoUCBdx2wgy0ZamcfvMIPfJGhXFQSarC9HoX4dj/8ZtBGRf5fZxkJ2Rcj9PwbHj3NaAbD63V2W+3ZEAlFTvByiKKCiCNNvx2rqzTRt8x8rkI53MbRDGi1JYALJYSuu6tyrHsZOhi7/vBGjBopkHQHPtlu/qjwOfK9pp+wk+x1GQiWae5A81CddseY2v2fs3T5aU3w7ch6zwknyuOHxYvTzetcIJwoef5PO18ryE77MnwzlOBtXQFUVRfIJO6IqiKD5hzE0uu9/gzhtWVsXLyquW1Xqy1o4oERGt32TsA4desOPXcD8PzAPLl83J6T62PWm8b2WFH3iyI+ww2/yk9RZWLzL2gXiRuVgnLKPmnpN+XrEshNIPDS+wJGwrX0xERGVU6MlaEhuIiCh6uR1Xx8vDrc+mncLyNvTlveJTU58yHj/21ymd57ldBrImbl1OVJHhzfE18r5hRX2BlHFLHedCx/Esx/ERRtzcYKXzvqGE45iYwNC/W82mFvwK5LViccRnbQmbolqPWZmYd8ph3BRuI2yGOwLxBRtpaPSAQ7PXZVtwPU+O13oyPkcneHo98wp8SI6Xph3sgfsJ8ZeQH0gb5plrcLw4N/NB5plTHOdwvZe016VcY6iohq4oiuITxlxDd1FmIgOpq7fVk82dv8J0Vvx6wNcFAx/2+uUll6Udj3KMWoTOJCKiB59e4x079KrRzMtAg5054xNERBSOTPNkC8pEDRevTal37KZJ5sZrZ9uVxdz5XyYiooUL7fKhm9spNIxA+N2mgm1ERFQGkZvL+TOdgh459J4NwDzoxx8w7YsXglBCCEVTvwOOXcktau3yWnzyJAxSNGgMaZSYvR+BjMPA+o6DTGLqxOl7NhwTVRRVVwy9HCVchYwSDpkgfj58Xa63fZQ181aQdTlHGl5hzfxAhjHZEoDvOMDPGmqkQr9JKAstFbX97gyav2sBUCjhiPAshPA3wUzkF8W47YQVooQyBuEc3S6nb7D//eJ9y2oAwzJVQ1cURVE8dEJXFEXxCePG5FIJ2YFVbLFoabNetYa9xuRSynHr7eBEFSKV1qHZM+FXREQU9aJ0iUpDdeb8NcaUcuTQn9LO0QrL/bKIaW++dqrjjk3K4+0HN3uSal6CLXQ4R5HPrjGerUN3fckIzl7gHevrXJP5xVkgy+t14Ni6n7PPboRxxRebNtMyHhNFQ7x879xnZe3SkTjw1+AFskyNgEy8gI85bvhBbjElUbI7nwOZmFBqQXYXt0+YZhYsfSWGuJ8ZoY5bhyOWT+GZxois3zYKMsmgxJD3kwXjv2/iFrM993OLJhSnM5sRdz5aOuQr6HaMG07C6PDjfhLMCmKCQKevwyLjITHe/cLF+cX9rBUZnJBlPB3UVFnzbD7bgWKx9z1ZUaE5HuMgiQR8yFX8xV8+/zOebM/zvyMion1Ndpy8rxDfdxI/cCaQ0YObO6qhK4qi+IQx19AnGv8kTYYYqnxOays+6xOeLFJhvA5lNSbDayvUDGli7e3QTiurDF1rOmVwsfdM09xgNPNKW4aFDvF/z15wFO47mK7Be0wyutQ9Fww8ZCDuWWM8dvWlpupGrL3BO5ZJ28rIcuhvMs374BS9hLX1m/E1LzlkjJS/wITOQsc4L+xOwuLwBbI4WggyUUH/E2SyDJDVUSMcEx/4WSDjBU3heVaUkO+NM2J3o1orjlL0BmaIIxVnZB3IpF4KOjGPcDucGjouTkRDvxpkS7mdevbHPVm80PyMDx42zytqvPwo9PMzi1I40rvNYIZmV7L/tQd8TYrGHUTHKj+UCXCGO7Vefo2EFxbAg7t8/hVERDRz3tfsOaJmnXkkaj+lSMRMSPlB8yaWzI/CMePZD55nowPqqkw87g9/cpcnO8gvKWfHfoc9hQc6RVv48pmc1oOhGrqiKIpP0AldURTFJ+T19Y3eNp+uPUXzuIBUNRROmsvBz+Eiu6w8J2zW6rNn/IAlUJuTzHo/Dq6drXtM5abG5l94su23fEAnz8dMM8lUwOqLr8n5DFLe1BWHnlfM+YS4tj/mGJgKejY5C7IJzEFT2ZGT10BZIct8tFjt4fYoyNopA3J9tCOJJ/EgyE5w+8/con2Al8sTIbP0eCV3MI5eTCxyrUVwTD6bb4PMlcWawk3Qv89xSSn7toBGj1Xc3rfwOpAaO1c8YT64xib75jaxmTE1oXc0mHiu7R/n5xmLQouTMwLlbc9hU1gJm+uwVG4kYuaD4iIb6JDk7zsW+33a9UvCphR2dZVNhMi/4N+5l6su+zfof8ghEyORtfVFXzFmmGSvsQt1J2zFsY5YlIiIjsbsw16/13xvOxy/Ud1TVFEU5TRjzJ2ifaxBSFF6IqJO9krNvMyqWfE2ychEzVwwMXMhsmmeK2dezq0tCntLkQk/3LfXeNoObMn9fosm3EBERIXB5CAjB6Ygw7Gl3/h7IiLqhTK+T6x9c6DhFtiXQzaRqIH317fDfG6FDTbTNlOWovgpXe8SVxaZNPQLHSGBB6TsK+ob4tiSjD3U6Fn97UFn60+4xfg1cXKWUjriZcL6LdJP3RgDeMQhexgyUJPvOAaMMD/kNvT0o55MFFzxFeIKSr4//FgyrqqGkR74fuTrxme/jldaty260pNNrzO/LzpHlmFYv9nFn7lFL7g8DGfT8IG6L0cYnIBfxwSxJthlSWQqrvGI7L0STX3HBNHGojbOd3KBqT29o2HoO1yohq4oiuITxtyGLuB2Vau+vpKIiBYvsYVBDr9hbE0zzp1xMtc3HfmnD1pf3RxjkwyF/9GTbX9Q7G3p9WNKzzQVDd88ujXt2GBIUkiF8+gLfK8zczsplsQTxQHiEUVXuHu93eqv6ObBNX/cZEHqumBY1TZuD2U4B25nlnTIFvJ3H+fQ1VZQfBolUQOTiKDKosfLKSe+Bo7JJhbPgmw9t46YTdFmXZosvk9ZWNQ4xo0mcn3XphTyXb0FsuGo05INgTNtfzZH+K289gpPNrNWQgcjdmCUf5QzIKY4I6L1Ykab2NjPpYEB+/dbv0o/XCTLRXbi4D55AVdgIY8743OOY5mAjTZe/SkREU2svj5tlNrQFUVRTjN0QlcURfEJY+4UFXBn7vMqLiUiot4TNj2w6UVTYSNXk8v2Bofn6u10UePTsunmo+kHHbS/2zb4oAHInC2X7VIzBVxHOzyZ4uAL/+DNNFmq6wbBcDdJ5o2ATNxZErTl+lRc2a9ochE/bQW3WLL3ajaPbYL3d0D2GcV4wSi3rn1NZByaXH7gGMfIe8ENIOR9VYJssO1TR4umlBYdoPIDP0KjzzVgJlv/3TuJiKjggq9a4RvGLR970m4Oe7TNvIspNezY/KgrFft3tvsWhyvG4ckLR017tvwQXGYQcKI2c1xrDDYyrb2Ez8VPAaa9SmGXXkhZPc4TWDl4gj9qa70MjNWpg5HzM4zLDtXQFUVRfML40dAhY6NhlwnIDwStPrTloR8TEdFtS27L6nxrVxvn4uq12ToXswgN7If5798Ckmw3rJC3KpXzpud4ZScbshu2BlYnz8w1FeXKd5qEq2zXHC43cEVKS2QXClGQyTWwLowEmUmu0RNwrJbv9zcQgbafFaRbHrCyJlGMRL3GxCLxkWHRFdHQHau1jnSRk9FMKMqF0QpLHIzH7gBPfYi9ojtftjIuNXi0zS6/Og4ZjXtKZ9QIimG5+ZhJEnzwvoc8UU3QnCPZa52L1bVmWVdQblb6VAsPw8WOHVZ4H7vo3r2eKFLC65wAP6nHQRuPi7YO5xANHhIZqZTXoWF5AGGS6+KHOA5PWwjT1oaGauiKoig+QSd0RVEUnzBuTC4tUDc0mTBLn5mX3eDJpk6blvqSNFavWuP1197/Pe59Ekb8YUj3dqDXhs+vu9NkiT211sTQVuXf7h1r7L2XiPqbXvb8xbSYCRvmlZiUCm2BVeWMTKGzw8ytbGr5BXs2V4PT0FX7Q7L88P1t5lYWzejQnMvtfpCJQxj3AJEYaTG9oDlGqvOcD77t1/jzevE9KxPjXJuUyrXlPqwX92GQDbw1bUZnMTpzh1zqeBzDRaf7mb2GzIWX2P5eE8Hf2xb1RPlFxk5W/8z/erJWrls0m8e37Pqxd6zq2f/J7rqHxY5mUqVnkU2Zfn6z+Y1SOeSstptvsiQMD42YUGJswOqCpJVenjYj4CIXc0mXNVzGHjd1vd9qNmam/B7rWI0lzW+vJWpPUTfLlgsfKqqhK4qi+IRxkymKlJxnshlnz/sXT7b1CaNxF4eM1l5R9XXv2I6nvmk6J6yKeftCEyYVgzVIa2uUiIiaXhVHxAvZ3/yA2NCkefVGPbzNlpTxNFLcIusIa5H5/E89DhFRUsR/1Wch1fGv22gkEX/iVSBzbdog+WtYgXE1t64NyyXsD8eLhotb24ljWT4GrEWSKQMVnX8RbmVlcQd8B83ivUxPwHPSx0uLvJ3px9ZDFN3Njlo1pyp13O5Y/3nTKSzyjs1Zan4nuW5T1xeFdV4za64dNssy3my+3Vs2/NyTiWNcnpntNLyIDr53rg1lTBw391Q9zW5YQbXcz+dfcBtkosouOLVftrII5+l2WE2+877vExHRg1vMygIDBpYsN47b3nzrbO3hXVwmbkgvBaqZooqiKKcZOqEriqL4hHHjFEU6DpuY8I2Hb0871vbOOtO+DvGsHOlcOOliT7Jw0VeIiCgWsku8ipBZcEWmmOXOtOvt9vVNm+qGeLfWMbKD46LDYE+YyYeDYFap4NXsVC5etGKTPdYrK93SS63w8MiaXMRcgmYWSdwFy4W3DMbYe9kIQ7Z4xZK8bSktkV3ao8lFnKxi+sl2mX0D9HdxK8XE6mCTgAXcz3aTh4O8CsaiW5KFCfsR0N38/a1+N8sTj2Ok0Fj93t8SEdHcRSu9Y7fdaEwjL27IMaigFQxmvKED2hePNJlPFRObpX+YRgYxtj6y05o1qjnHofpScMdLxEIXvwe5fyLqTphpsyACMeRh/qGfZTerLa76tGnJmFz6mSULIkRElF9iTVv5bSezm6hBNXRFURSfMC419Oz4ntcL32g0841QJqKalfVO2L4t1IVbRhHNjtgwpSYaKvAR8kmaYXu17feYtmsb1BKlP3ErKwosaM+bnX0OK4lkwVzoO5x5uSKuGtxDQj4jKLtDkoc72XFp0bYw87KRW9TyI9zKCiCa5T3irnuf5pY/bloIx3ZxPZgNkBV6E299SBD6KNzAvvJqkMl7fxaep3Oy2RrwFEGihq/muIKJzz7oHXt48cfTX5AFvc02AzT6GrtUu62e2vSq0fhfgddIMKHouZmc4kPh3knmYVi5wBb9eeWAWd8lYzYQNRjglUQiatrjdmVREOC7i8GT3cH9Egh9vNQ4VmsuN9dsbrAP4J4nzOdRd6ldiXd1RHN7Mw5UQ1cURfEJOqEriqL4hHEZh54dO6A/b8BRo8pHzNvrt0P9BtkF8pYsT/Il00z4Bys6ke4cTuNO6N+V5aUyIBv5TP+IldXzVoo1sM3jFLa/7Gcv1h44R6fjvFFujzuOCRMdskzjXWC2qbj3rgaZbCnqx2zP4UY+y0x70Lo4MNc+KK/sNOYGNL+J3xorPw/1Wtki518DumzoDGOKnVFrd/OKlHPp7gSX1A1A0etiNoeGwSxawn00uch+UbtN5nvyUNQ7Ej1kvOsBMNmWVRoTTd6W9IxYjUNXFEU5zRhzp2gN7yfZlKG2hpvdgw8ZFTbbLv9zPt6Kx3PVAdl5VPE1T9LUYjT/mrwM/6SHQStH5K4XQmzgdPaT3QLOxR52Loo2jiFo4tiKgiziGCfHZfk2mDa+itstIEsN+EINb21Kq+TGULXlfc/bB6WRW4gXcJYpHmkNXc67GfbynMnO7Z4GW0K7/JDpywQZttGFFOkxd5lfCNp4gjNED2HNF/NUxttEG7dafsUc3v0jBu+0BLcmGRqqoSuKoviEMdfQ9z66mIiIdu625Ra3PmOC/rECY7sXIma2aCtdbIvWt2+zIVajDxS0Dy0x7T4Mj8w1WeB9IiIqKbThXTGuNFh65rc8Wfu79w9+Ktx5K708REYkyehJW/Ofink1FYH9zLayOh1mW3vZX+0x0cZQExMNDLfhk7L+orUP5miRPCwM7Ez9lLFSooTF4cM+9DDVdMTmn6ud3+80wbMgj8xgm4fIZi8SwjpSqyqolOQ9pxg+OZkXF1yhhSpgVZpImh9TdQhqb5aytt4FTyKHMgZcm06KqBKeYqifM1RUQ1cURfEJOqEriqL4hDE3uYgnsTNq7SvBoCn0vmSFzaJas86U2AyfGyEiorISa5IY2z0Ud9kul3PIq7T/J/u2PJfj+Uzo1P/tvybtyPZl/+7129dlYXJxmVkmQF+8lhlqkRwEn2417zGyco615Vxz689MJ2gMJoldtszoputWEBFRL5xfMgGxlkuEW8kexQ0xXGYMcSOhY1WCQ+WBjsIxGReikUHOqyaX/mBsQKaMT3T1S/2clYuNKfatbbaO0XBsuvHiDGOyLYaHYf+zJjUYn7vJ1J9+NYo4RDcUtgGXkUuk9C5UbIkZB2l+j3kqgyVgohE7YxcGTWBe9tBQDV1RFMUnjLmGPrFq3cAHX/h5mij2xqNERLR73aMjdUuDINqp7Jpg4wVLC40z9M319v9k3hOs1p74ZVZnn/etX6TJtkrpucosa2rIJgz47Uo0FWwz55U5rON2H6WxCbTrhzmisjNhVf/EHKNhVIhTqPIL3rGK795NRESRdrtKKQ6Y+NSaWnveGG8Nt4czTQbTdCVBCDfOaORWnG+oHY506pzL0aeO0v7fQSZmQn/GReb3lT/NPCCP19qaKzfWm1Dl2p0YsDo4By6+wutXr5hvOs12u44g/ybwOZGfi/z03oJjUoE0HAaNu4edobDNXG/C9D29Ox9S5nrMk9Gd6PZEBUnXNjG5oRq6oiiKT9AJXVEUxSeMucnl1MOYG0rPvoOIiNphN/r2x9hpuf5uT1Y0zyzxZs//EZzDuOkWsnNRNgwnItr6uDHbrIOiKD0cs7p56SM0EH19S+AvU6lkxV+u9SSPXMcdSGTznKFi9foiHOO4f7zifVxqdj+YZlbVzKaBKOXY9Jk3fNiTRblt+Q877hHOEs604MT9TiPcYvbBSGUWZoNUEsIsSNk/crzkM48Fru8TI63lM8InqJpNLVTIGQvFNntyeuU3iIioq9hmaG7Z8BAREcXoz54sRMY0uXK5SXMOzppmLxBmIxh4Rev4OW2EuHlxVcp+JlhKOcJtENNHJeMzZisYdXUbM8zEoPmBFyTgKS00svxeOwXHYRONoaIauqIoik9QDX2InMMpZAla78nu27Iybdw9dywnIqLyqVY2I8N5p9808P/Y1bQiwyvRUWSckFVPOYZB5idJ4qnsXr8Yjjl2vZPNK7Ld/b2dNZ6ND36Q5SsMWClxDbeNIBPNfLw4HMU5Ox1kV3GhwZshwzDbLfBOJTBbt23AUV6Zo36arnzPqMkfaTNPV1kVa+aBs+xB1nBDF0Y80eLFJrS5u9tmaJZVnc8X4KVvAgKbCwP9WyKaO//zREQUfOa3nmzzCdMe5b/r4B4jZ3AnCRo3e8aTsT96oiQ7SgNB8x564/Ye85NGuw8G7dOeiGf6BLNDNXRFURSfoBO6oiiKTxhzk4tkiY3eNhvDQ3m5LJVsbOkSDqjF3K/lbGpxlOdxyuQLWQsWlFVLHANTeHqV7S+8zDiIml+GAXK7uKoTr5SYXBxmlqXQF3/qnvRhw4JsOoFvV8w8O2j8IlmqaOEq48TBcp+bXFzPsIugQyYGCIxX72TzSFmMM47Riyo2trbfeKJ4zGRrlmBMeD6bPTp+b1prBSGK8QkhKTPI72LmLLshx+QD5ouLc+BAGDKsy9jO1HscjX7pLuDClGJbyaTNbc5nc0yw0BqtwiEZ/4e0c2WLauiKoig+Ycw19N9vNvF0bW3WwZAk40RY98BPPVnTsfe5JyFwuTnaho0JptZKa5upPdMWtf91W/9yJRERVXzUDt/O2u+CK2yQ3eZtK/m1JjZw7bf/1Tt20zKzf90j6+o82eqlg9/WHRttv5ijtFpxN6zV3KL6myHWT3SFuSCTaMWTCREUPQpXMVK/Yzm3WFNjpFYDIwJs1yca4FgWdh4NsOZJpppKksGL9XTkl3MUZDteMr/zaIdZLlaU2izj/OPmFd1xW55aSqEkS+xSqJidi50dZqUah4dNlObisM26Tvb09GuJiEoiZp6prjIrcHTUU8BMmz2gcQcL0zdOjMXMhZMJMy6ZtNOtOEMr+mnxJz8dq4auKIriE3RCVxRF8Ql5fX2j547My8tLu1hf72bTmVAJUrM1TnKf9Qx+rG4pERGFPsU7c79t9//LdU+gkSZwgY1H72ljx8yx9J28q2Z8h4iIVq6w8eXXX12cNi5X7mWzytNgcjnAlT7zZllZ35Pcceznej23eDdShizXJwYXlRKHjBmUsruQ5NiNRweoPJ2ZysDOg/5z95o27/YRuqFxQrZx6AJaAcX8FneMk+cu4pChg1WMFGgSkb6cF12Xcr6SM6ysrOJDPN6acuJcM2tq+SeJiCgA5pgEm1oCIXvVYBHvMxqEu+sx4zo7jemls81ms7I1hsrKPuHJKoqM2SavATzpTF9fX4YNhS2qoSuKoviEsdfQc7z+/KtMbOCRRpvy2PWe+c8ac75iDJhg/+vSiT+N+uUlIisCiasJVodiD8DAd/uPx8Ar0TY7QXaAhgZUyvUcnuM5TFX2kcSQPNEE8RkTl5jLSSwrnI2OY34C924duNKQG6nSUgAy+UxFe6+AY8UpY4istt7Pacn0pIzBceicDZ9n2ih8ufLaCr5ocdDWIyrkOjOilRMR9QbNryi/EiwN5fwk8d6inbvtnFX/vLEw4JYWFfyma9IVdNXQFUVRTjfGPGwxV57bbgLZWvZs9WRPPWEssi3NNnCqM2b0qwPvvDRq93b9laYYym233urJyi42u03k5WX1D3ZYEO2iDNSWelYOroINK2STOyneD7lJXu2U4ahiiFr+uNHM2cZNDhu3aG+oHZZzzslB0J5kteH6jPyomaP9u47bCse4bHGFOcqvRD5T9OFI37WVIGrhBZIEdCL9mDyLuClJnHexwBWZXCPABvjyy2xNmWC+OWN3zHrvekLG/h3ssLIgW/E7WUOvb7d+v3quc4Qhm+TQzHNFNXRFURSfoBO6oiiKTzjlnKKZaGmw5pX9LxsXXmObDTQ7EjVLn+Zm0/Ycw0Cr92mscH4GJ2yI05zPm8Ig9a+/mT4uAyXfsv0OjhMM2O1AvWWl7Go4XsrRjhcuhP7lvIxvOGFlmZzEYp5w7Td6qtIIfXlfaKZYOgLXrIG+hLxGJlkHZew9kzEehForNVWfISKiHs7Q7OywtoytbOqIwnnFF4pGUdk39Co+7/Raq/se7TBBGG0Qbxlnh2ZZrQ2IKI+YJ2j/gSYiIqpvtL/pRt4sJluTpjpFFUVRTjN8paEPxv6XjAa/8znjIWwFJ+r+JvNftOu919NfOELcvsTs+XbP5twrlmTlZMXaIqJKnXANVFxgdQ6JHshWo7qbQ+FWH8487lRFgvPuAdk818ARABPVxAWJYYuzOWmoh7dRbIBj2a5CJXS1jltMfhKnawy8rQ18LfzNyQq4A7a2GyqqoSuKopxm6ISuKIriE04rk0smOt/6HRERrbjOxpDveOGXw3b+uvNs8fzl15vdyK9ZxnVuJ30x5/NlZXKBehXemhS3JY1weye3b+R8G75EMhgxSePkd3s8tcAMxvStGyxQGqhffR6/4HJuj8WmPGpyURRFOc1QDT0DyXdNWuWmhx7yZK0cBtndbdxj5xTb/Lnp00yYUnX1FzxZeMpFI3JvWWno34G+lMXDcoHi4dswPPeknPqII3gmyMZj9UsPCVfE9NF3+w8JgKOyZxgclGOBauiKoiinGTqhK4qi+AQ1uZyiZGVyWQ99Cdi9yzVQGQ4ws9QVrx5KaYlsvLPrdZk2gEAnbX1VHq+lAAABEElEQVRWd5cdUgIXzSySJYFFrnIu2iYxAcNQgKofWZhcCLJIJQ8jD8ww3gx0EuaYIs47iIP3tO+Ye+xQUJOLoijKacaYa+iKoihKZlRDVxRFOc3QCV1RFMUn6ISuKIriE3RCVxRF8Qmj6hRVFEVRRg7V0BVFUXyCTuiKoig+QSd0RVEUn6ATuqIoik/QCV1RFMUn6ISuKIriE3RCVxRF8Qk6oSuKovgEndAVRVF8gk7oiqIoPkEndEVRFJ+gE7qiKIpP0AldURTFJ+iEriiK4hN0QlcURfEJOqEriqL4BJ3QFUVRfIJO6IqiKD5BJ3RFURSfoBO6oiiKT9AJXVEUxSfohK4oiuITdEJXFEXxCf8PgjtLqlhOiW8AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GroundTruth:   deer   dog truck  bird\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GroundTruth:   deer automobile truck   dog\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAB6CAYAAACvHqiXAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJztnX90VNd1778qo9pDYVYYGcux9OyhSC3iIfkZasRDpIhGchF54Bq34GdggV8wARpw7LyAG0iBBmKb2FDDKzg2sfGKcYxTYIEacI38QDF4IVzEi0QRjSCMvSRq2dGQNaIaOzNU7499zt1bmjvDiF/Cl/355x6de+beO3fuPdpn/8zq6uqCoiiK8sXn9/r6AhRFUZSrg07oiqIoHkEndEVRFI+gE7qiKIpH0AldURTFI+iEriiK4hF0QlcURfEIOqEriqJ4BJ3QFUVRPILvep4sKytLw1IVRVF6SVdXV1Ym41RCVxRF8Qg6oSuKongEndAVRVE8gk7oiqIoHuG6GkXdqDu4AgDQHo05ffUNJwEA0x962OmLtLQBAAZm07hEpM3ZF+1sBwCEhhc6fXlFQ+hY9e85fQlfNgCguHAUAMAfyOELiXXS1owBgHDTMfocuC+UXwIAaD5L5/+kqZmvI0rX7c8XX9AXpO9Zx+Naz4YBAIGcAHoSaWmh75ngvkScfqaWKHfufP8/u31O0yBfHllZybamrsFfokYgyJ3RqOkTz4x4ZgnxOsUitO1oF/uT5adYNW13NtA2LE7ZSo8pgqO5zz7h8sXNRnfWvsbtY7tpO6SS+243z2fFBO4rGUDbaUlXmDlu91K5vqiEriiK4hH6XEJvjvQHANS/d8DpS3SQ9OuLswTU3hYGALRFPgYAFBeyGJxXQHJLdk5/PnDsLACgKBhxuhqPHaFdMZJp/KXjxJUY6ffW4U5P6F4ad7r2iNN38BhJ0Gteoutd+fBXnH3BOEno2eK2xrPpmnKDfL39QX0+My5spHIA8GeTiJYtxC67sohFpbT3OyjXiJARjRPR5H3RuPjDrLACZitF5ah5Fj8T42+9Jelw/sm0nTE5+VQdR2m7bQv3NYVoW1bBfSPN1r4txUXiIEakbx7IXfeaR3xuP+4LJZ9e+QKiErqiKIpH0AldURTFI/S5yqV03CQAgD/e6fTlgJa6jYcOOX0Ro2qJt7cCAEIBNhCGCo3qJODnAw8opeMOYsvP6GG2bU1LUoVhb0VE9OXSJsbjTp+lazu4658AAIdCbNgcl2/GZbPqJxG/g7oSvB4P+ug6i4bT2reocIizL2aMb3WH2JjrN98r5OdxeP/foFwj7hlD23ZWhSFhftt2oUKxzUSH+VuoaM6Zz971l9z3rz+i7eA7M7qMgcYYOn908r7TwvBZbU579yLaTi4Rl20e52bxauQaVYvoUvqM+0TbzhHvX/bRVEJXFEXxCH0uofujJMlMrWRJ+ujB/QCA5tZjTl9eLrmLJawhSkj0OGl8vqRpZ4D9bxcWZ8szW+PyeOrnvGvYA6YhpKyLZOQMBdk4O/Asux8CwOl6XkUsm0WrjdZ2PsaaZ34KAOgQLofTZtGKIhr9FQAgIFzh/HES+wJINsgFpcuccs0IW6MoePV1+sD/BQBUxIU/acC8PjlmJZebx/uMcRvH/4n7/vhPaDtuDPflmvaqh2h759CMrrFgtmiv3UWNKnqXqmMsoueNGw8AKBNSe6nxTdSnqQeD/hu3z4dN47fX+KQfXNWjqYSuKIriEXRCVxRF8QhZ1zPC0C197v7nHwQA5Oaz/3e4jQxQsRgbpWLtYQDAQGOJCvjZyFiUT0vjvAnCATdoFpThk9znp3GNB6jvdqHCyM0Pmc/dwePvMEvqwfeIKz5pvstyAMCCsfw/cdNhUqHsXP5nTt9Daz5CT5r30XeOG7VRtp+NqLEIfffwIVY3xeL0XeubePn3bG33Y2qk6OXhFt0YT9C9bOLQCNTtPQMA6Aizys1n1IX+aBgAkN1xlj9w6g0AgNCMMEP/mNsRo9bxm987KrSgE75B2xc3cJ+bPXXLOtqGjQW0U0QgN5HqribMfYGFSwAAoxe5Xdzlc3NHitp54D/TjrpcNH2uoijKTUbfG0WNFB7j1CzIMzk0pLEw4iPDZDCXDFDhNpaUwsaQGGsIO30FlUbSvneSOBsdL9iSn3T8mr1kIC0fxdKyL8cm1uDbtGrq8m7Xv+zxh8RfJL0/7CKVSwomGgPwp/SlE51s4I0lyKUyGOMVSJ6Pvl99wy/SHre37Kw5BwBobSVX0KaTvJqZ8yjJlgOFsBdpJ+mjrNjbcoDPuPUVi2jM4gprrHQxWl6kzek67lpe9g4AYDZ+k/ZcNSaCuCJiDKsX2MiO6o20bedIZZQYN8jox9wXM9J9jjHYFo3ifQGS0CtivHroWPEvAIBVZ3/i9K1YlxzFeuXcJtrp78ONwe9zs5/RGFz8fxl+1s4Vffs9vf1mKoqi3ETohK4oiuIR+twoGt1PBppoB/tdxzpJveJPcFRewCxNO03U5IfGEAUAsRiN8/k59q1ssjVkCgdckxQLZ0wUZjDX2dMRpuMNDLCOIRY3KpGAuI68te5f7hKsXj3CaS9bZiIGPyMVR+1Btr6FTaToaPA5i0xer427Wc20eO2vux0/09/xnrK5Trvh/R+b1pfNlu9f5YP/EwAwf+FCp29qRbJFrvWCubYXyBd6/ASOJ6ga+6WMrqkvcU2fezXeiQ/NNiRUNCO+Rtsf/b3T1eojmapuDxldIw3vOPsqfBSBGsoW0anjTFR0icjRvN+oTk4adU1CxIBWGz/4wV92uiKf0jOWA448PtvVSOdK+6XSk3wv/9Bl1K9d+v5AtP/jCq4gc6beyfej8tF5AIDiUfzsFhVR0r2dRhW7cy+/o/vete1M1TFXjhpFFUVRbjL6XkJv2EQNkS92oDF8IiBcCG3qWCutSHPurVYalwUjTBTomSbuyjNujTEyAmKQTJ9rDaTiwBdJIpbpbYeEHuv5FdKyYP6fAwA2bf4b0Wu+12dkqNry8j84e/adJAPX1DEsgU0pIonrUIIlqkllIpkHMpcqb7t7otNu/+ifM/oMQ///173MeWbCYfoOG9bMdBlvJPRbQk7Pnn8mCXTy+MG9PPeVc0cB56h9/Q2K4K0sHZg07mq+E8uF1LrSrISa7prn9MXzKUJ55OF7kz6blfWkab0pru3cJc95uHiO0x53gp4TYWqFk413xYdO36aVdwG4svwu6d0WH6HNAH6u82eSgbfAz04BTevLAfCbLOJyIZxCM8K+3cXCTfTJb/9vOucY4YocN3PLaH43MuGtJ9c47enrrbPEtVltqISuKIpyk9Hnbovbd7wAAMizgT0Aqh4yrllGrw0AsU7S+/mdMnOs/wZIusAW1m+vf2wpAEBmXik1w0Y/QP/HiqYIl8YJs2jbTxTh6kf6z6MNosJABjx2F7c3bX7btGTOBiMRGPGjMJellqJxpMcrCrGsNDCbdPmhOJfYA7pL6Jniy+5ZsAyYs4hWSUePsN/dyQ/cjk9ui0/MHev0HD1F7S0vvQQA+I9POOKp7TPa1uw/4/RV9IFkbmk7w3lV3CTzq0l4y6cAgDWibxf+HQAQ/Gi/0/fd4Q8gNVau/l7SnnkbT/AxFpF95hPjPtk4QawGjYS+BV91uqJG1t0xTjyo15oR5Eq5eN4sp2vRInoWanZ96vQ11P8pAKCpllx0Q+IQ4822XvQ1IJkq43b6/HoKyCoqF+95vnnXBl25m+a0dcucts9oDh5+nl1B4xetL/b1sQsAKqEriqJ4Bp3QFUVRPEKfq1zm/jWlD40dEm5B3yXDY0BoB0Ih+t+TFzPlywv/l7Mv8gIZJDau4gWu/WKl4lxxE8B5dCOpDho38hLcD2pXrX6FjzGFcmlMn5KZIdTG57WIQNHE+3RtvrEhMdK4aLaEAQDjQ6LQhl0Syuy5/U2U7MetGV1HOior73farxsVRG4uGWl/8tOtzr5RBVblwtF+XV28NLaMHkbbvPz8pH25t9J2xuTMUsJeTWqPf+60x99Ly+uTrWzsnPCnZCCVaph0LJ9Dbpmrtz6Y0fgt+637IbtunnSeSlYX5u9LNoZaurqOptznE56MTca2+YNnqOjJijFsBHx5Ln3njUIdWYQfAgAOV/K5E4OeBgCMX2MMg6IuBy5bS8bPTr7JtzS+hA8WN/bdgXEuKjN3FaksAhFT//fAc86+pt2ktqwX75eN6556F59r7uPfAQAUlRuz6BDxbA64FhGxwNRHyc33QA6/uI+/sBcAcOzc9StGoxK6oiiKR+hzt8WuLptJ7iXurDEGnwqRcP5TY94c/Hemg4sEjMkqA9BdGi9zOb8VaiIu+6xz42HR52Zw6YmU3a3R5inRt8zYD+cffl30GmPJZ1aKE85Zt9qsk7KcGUkrrTv4P33+4u7XkenvuHxttdNes3RKj70ilwV+BwDYe/i805MuUOiBGbQ62r1tWcox15NJ01c67b3bqX1aLDDWPUcrkM1r5yR9trfvRMTcopzgANGbqSGM3FoPJsh4Xu6T3mm3mevhC58whTIr+nPZhfXVl2nVsGolibwLJnEA2MxH6JlJCNfA5058EwBQJUqdnTbb7WbLGY2AifgrAEBoxF87fW0/oKc9xJ6gyW6Lwx5xmqNMYODdIvhvxixaBgRyg05fxcTU5fmOTqHjz+RH2MmgsmMnv19Bk4nVn2dckYvZIOyOiY7rNjNYg6ZdVUmX6DQrzjPH+Xp37AYAlC5ddYnzXxp1W1QURbnJ0AldURTFI/S5URTHje5AGgGt1qHRJVfC4DAA4PGsbztd1ntalLdwFkwiTtSJgguZrVxgWaVHJmoWicj66xyvSvS1Oxcgl2zmU06Eq/QNt2oYUXfyTro5eSIr6uUifc2T+V1SzymRUrdq7Nik/TVHabm6bx8ZgNo+Y5XLJ+ZrFt99GRd6hUyZwv7dbhGMOXf9+VU7V3CQbbEaBCOMQe7Ei0nj80d0OO2WMN23hmNJw2BTsW4TttGD1d9OGnXHFltGg9RIkSiryUrmkYF0mgjkrfrvRm35EecHLjBba3Y/zcMxDOQ3X3SCn/bQ8B4VVlxYt/xbTjvPlOXNDbBB2G/yMiEo343UFJT/DwBAoJoN2TMm0vfLGykiP21Rm+EybqM7+6YUO22bJXt0YYgHGLVN2EwM85bucnblGbvqq599Io5ojL0BNsCOLhqO641K6IqiKB6h7yX0XJOFsIUj32BTp7wsxlm7Qi0ZCEVRLsddUOahsBK3jCe1RtE8Y7tKXOB9mToE9swxsUfss+2pou92Z+WxW/SazosmYrBZVBYf5nLScxRhiJMu+wz/dfQcpx2LkQGsrY0lqlyTH+fsiZ+lPogLTz7G5uXtb1LNMn9/vtMHq7tnn1z7zFtOe90qktAK7+UsdgWFJDVJ90mbD8Zm2YzFYs6+pib60j4fr2I6TEbK9gi7e1r3wwVLtgIAag+I+nFOfg02VPY2j01WFhnCurrOJO27f2Vy5fb+JSSddZ5I2oWo9PA8Qc9F1ehpyQMNJbkpdxlauv01cggbrwMh2kZEMpdZH5GL4hLxGSuvLjP5ZuLg5WDAyXPExstMZo7iQl4zV5SYl+5W3h82tzIn79LHAoDgOFp1jR7AEnqVfY5kRsqQWSn1S46Erb6bXrApHyW7Em4eyvdx/ov/BwAQMFlg/WAJfavxiH112z/yh2csgBnIBJKjsq81KqEriqJ4BJ3QFUVRPELfq1yaaQmeaOK1qc+uuKVN405a7teu+Af0ZPYwWmK2nmLVhf1iI11Oae0yUuUSdxnnhl1RtaUZs1O0t5e4jTBevv3MVbqtzM6LZXzMZX8Pgjm8HI6ayvGzH+Vo2rnzKAIwIFylJ5TTMvFYbbLhzo26dzem2UvFDKw6BADyh5Fht/k4W/wCpoCIVAdF2iPd+qzKCADGl5O6ZtcOsbw1SNVM7lAymCVMfVlfNj/a0+eSEVDWTE3ESWnmnoQsGTdViyXoPG2sO+h84zn3wQCib39D/PUGXduH21KOD6d72AD0LEtRIN7qxjBtfcLa/zreBdBdUXNg8AoAQF7MPJtB8fJlG0VjVBgvbYasNAbvgiH8sJ02YSR5Qt0U6m0A8ShSr0yeyX7ludawGhMuDoHUOqo6N1XLfXS8jR+86/QtqCQr8soHvw4A2J3g2ISOTUb9ki09OSj6HPK5DqfRkV4jVEJXFEXxCH0uoWeV03+7KhEglm+uqk3kbJhxlqTDp1wy2X7bSOblos/Kq27Gzqj5J5ovc1QkpylxkEJ2b90ap71hS13J5BjWmGdc64ZK9yZjjBokIg0HmdwcQSFd9Egv896+dU47YtKoBvulv7Z/ObgZALBwKUnSblGTmTJ9LkURvrSOS9zBtKtr+eZaqVpK18Eg/Vo+YwCVeWFWL6FQxHXPPM3H/ZzcWZeu5rVQcQn9SvXHaDXw/Hou81YxWkZwEms3k9vd0gwl9FVLacW04tn7kvYFmknWzXaMh0DcMZ/LggeyoAqRD3I5zGv4PGmfZfJo+dcfmHOxW+ux1+mlWP51MgaOFI/TLmMMHS/smTYKVLr5YoIxLr5lIrYvyDWrldaFRNqSfB96EhCzS8R4EfgHuY/NCGPkDOSwBB6N0Ms8sE2Uv2g3K7zBxuHiM16KjzTvUtdbwkmhgvoWuLi3rtxFZRpX9OOJZ+Aik89no3AIqKXjxdrYUN/4S1df1GuKSuiKoigeQSd0RVEUj9DnKhdLjTD8xC8m7281S0e3uoJ2cbhf9NnIN2kUtfF5MbMCC4pVpV2RypxXdreMY7OKAlkJKS2FtlqKtGz+kdla49+vxD67lP6l6DNL3QyXq5dStfRk07Ozu22vNpeuH2rTvY5PPeTz5KjhqDDAJhK0pg+bKldSzfLwY6SO2r4lOcoyUwpykw1tMZOH6eU3KMlZDjiZnP21O7sl6SLf9zj4Pvtg1EtTKNLRNS1YFn+XLud4bMDDTLIu7jZPav13RIWeWZQiNyHeL/usd7PFv/U10/gtkrGpaUWNX1MKGItSpxPeKGISVjyV2s8+c8jw2NbGv3vAVDILHBF1SSOkVBptozZFTeCQTalbwSmGLYXgSOhmJ3FZ6oR0bXE+Z813TJ3YAM8WYaMOsoqc65EGUSV0RVEUj3DDSOgICOPReZJCcgbxf1FfrnFPOuGS38WFsNnKvCrWsclK3oeFHaqoxxYAbNVJmfPFRoE+63JOa4uaKATSnS+QYWTL8z92+irNSSpK6L9/bAgvFUYvNNJVRBig4ibHyp0y/0jvIh0tL27j9J7zZ6QurvBFYPPaeU679ZG/BQAU30PG0Y2vsdtnaSmterb3rjRsN2Y86VJ/M9r9z3Zk9mz6el0P9lKpeH/d7a/sD15w2sHHtwJgMzwAtBk5rsW62gEAbDSvkWonc6pcTDfv4Uxx3aHQJa4JiIpI3kxXl+loO05G80gLuyY0xchzokl4KzSZ1UgsShK0X6xF8iaIHEk9KB3BrprNJ0hCH4XU+WBaxXEPf0B5d8Im/w4ABMxK+Vd1jQCAzTvYSLruChwQ0qESuqIoike4YST0+HmWQnKMyjBynnVfgf7JLl89kU5HNhfcatG3r8d4mXbBSuFSX25zqUlBzErt9v92hdg3w2wLfsgri7UnSfe67xyPazXtsFG0trWxhF7WQtn3mg6JzIdm3OTK1FK5W0ZBG2wDAG1nrNafzzV/RupgmXQcbmTJrqz42soEW95ySYbiwPdozxtU6m/3NtJUSouF/Z2ffUbej8xKz6XDn0bl35dsFiuA2tcfBQA8Mo4vNhezAAALcIQ/NPmntK2iVIK3LeTx7dUkiY4Tevv3QpzRMRXllX/SyytPz64d9G7sO8D+zH6zypbvqH2Xw82Ut6iskFf/fjPjnV7MNqotb1oXRn775wwme0heiFYsNWv4nhYMp1VgJMEziC2MI92aN8ykwiAFo8l98nmzBYDq/SStNx9PF6zXe1RCVxRF8Qg6oSuKoniEvle52CLcwkAZNVqBLmEMsqlV01Eg2lbB0OEyzvaJcohoNFuphrE3RwTZOZGn/h5bAAiaFBO1cWFIyafovdWbuPhBYSG1t73+CgAgHmDj0VN7yajSKSyxc8ZRrc/vbUouQGGJijqYAaN+cVcr/GHKY2TKtVazSNY/Z3Ki9OPlKi4mq2HGTXwCAKc1lr+LrSUaifBNrXyQCnHs37XmKlyldVdMbxR1TevTS2zGF/lcz4ep3XkXqSVrPlro7Gt++/vUmMIqlGl3kuG94px4sqtfMY2vAADauymtfgEAqINw9fteanc+y8gJmX3jnUf55X/oQeM+ec6qW+V1iNBxQ1qXQHPYKlHHtKme/J/31fKxrCr21dlfcfri5gk6XEdKlLq9HFkaaTHRqYWiWMcAUuuMy2ZD6aKt7LbZk1/VUwLw+xdzIZb9GytSDc8YldAVRVE8Qp9L6Fnmv6jvFu6Lu6S1qK/7xSWPlSuq1jcbg5nMfGjlhQZnPGMlbxlsYeVsGURkK8rZz0pJ8BPj8dUc5yOHmylo6Ggzm0s6JlHejIN1lJtFCBAsmQsjalMzfZegXCqITJEAG2sBltYDLobSISO4cEGmOV+uNb8fpCIa8fM2mIN/xynGHfFfj251+n5vICU36brAromH3l4PAHhoBuVT2b2NS+EdPvRvPY4P7N/VvSjElWHz9CRL6Dn95jvtiouZZbV0MBkQY59yVNBMJB9j/hN/Qw0jzBa9yAFOzbfQ750QLoqN594EACwURRtsnFBtNRmOs8H5dOImt2juUHYThci9lIpwC1+330fvRFuUjfId2eT98IM3X+EPnRMBUyn5stN6Zv4PAQCHt//E6dtznpwHysdT0FPVo1xgZaBxt8zbzRJ3UR2VZcwL8cq65iAZLQuLaNa4eziXuCsqoXsamsZZH6uevLygvB0b+BiBV80xLvTWrZVRCV1RFMUj6ISuKIriEbK6uq5HhgFzsqyslCeTygE7qL9QBdjC4C1p3F9LRN6FBpecFHYRbhdWIbHPKkSkysWakZaLvmqzLTfbYrEvNI2WgjU+XrqZ4uHwR3n5WTSGlmx1x8gPWJToREMzRf3JogYlQfpexTm8DN78dnfD4PZ9HzrtaRMpqnHhUl66Xas8Lb3lauRVYcPur5P2PPYEFaItvoeXyOGzlAFo3arpGR0943fCWNKzSmyVh2SjXc6I7U77459RPhOfW93YNPyXrKVOuwVrk/ZbczvfjUVOqylMxrciUYgiYTQtC6dyvdgtJncJf3NWawBU0/bZrbx3icvj1DMWYsESLgpcNYEMjvUnORvTtj3vAACaa3eITyXfw56ULzrstA9sGJtmZIbY3FHpVI8XRFTtgGsjB79lgrinj0xWlXZ1dSV3uqASuqIoikfoc6OoRcpENgG/vLpsa/VLI6G39RMfcMnYaB3UbB6WMvEfeYPLeGuD3Jm8y6nAVSGOEQmS21j9Ia44X2Aqfy+ZN8vpy8kjA1H4l5SbIhFLOPtKcsg1rKOFJZlw2MTBxXokDxFYqRwAGo2wXjVpUorR15dHF2922m6S+YatRwEAC2YnF02YNJXWR93dC5Mlc8vIUbT6WfB1Lr8XyMlJNbzXnF7H0bUFlZTl8NBWuuELv8lruYYLPwcARE/ws1D3DMnSZcPFc7r0z0wjeUVpXRNbusldNvaZz2WfFBtzmAc+Z6l5ZBau5Bw+m1e5FWbsyb8nd3VyoZKmM5SwqChNGbkXN3G5yFguOQJs3SGWo++bFKoZSOWSlTOvglQuycQp4BpJ5ZJpJrVSZutId1RCVxRF8Qg6oSuKoniEG8Yo6oosBWn8rq06prPnWABTH2Fj0M43Uie9sSl4Fog+t3S46bCL+EViyXk4n6IZ23ysGsk2apJiUSDBVr5vOEYJgiIRVrkgTsvxQI6jeILPqGg+aWXf6ZOnuqdUtWoLAFjkorroS9wShzV/wo9C4ZeNadklAvTZTQcB8D0DgAUzU6sMht9H1rqCQjZM28Rd3bFGv2TVQrp3Ytsc9t0uryQ/54h5GL79Ta7ruv8M1easvOvvnL4lhTSw6F2uL5vnWlCCeNxsN4i+xbPp2ja8JvPRpj7G1WTU0Cec9vyF9L2qHuCXNL8gjd3uQfN87hFVZS5+yzQ+EANtIq3klMHl00j5eWB76qIaXsDtfVGjqKIoyk3GDWMUdeVCcpebZF5utkVjOMVuafbHAIC6136WNN7Kz02iz0rc7UiPLWJhXRrbhI0n0kkSdKI/SyGtUZI0WtpYirK7gzkkjQQL/8jZ13SEpNSSQnZRbIrSccPtqQsd1LzzjtPO9tHPGk+w5N8XUrubpJE1gK5j7dOi2oSLZG5ZurAcAFDXlH5xVzL26wA4X4svOzvdcFz6l3Zn3mvPOe3O12wRCFoNjBo8SoykByNyD6+qHqgmI2GnkKhtyQO75mgEIyVzy87XrHH42krl0oxs79TtLZwLp/UQRVe2jvsqMmKveT4vyvhs+/LI/EL2ntrfj5/rg+309rmlRlYIldAVRVE8gk7oiqIoHqHvVS7WpiJsJXBJzgVbp/PT5F22DuiHDf/o9DWGm5MH9kBWJ5o8lhIahUo5ze3K9buQCrtEPprN/xODAZOAKMbL+WiM1CSBbE44FY1Ssq1wA+2bMZOXoWUlpXTcvVyDNGbWmLliHXy2x33Y9CNORrXm+xQhWigMg27qj/TQ98odyr7sM2aSL33F/fc7fYk4/XCrVlASrWO16RNQPb/+7wEATz42Ie24npQW8fW/vJ0UE49N5zjdyQ9QGtLXXn0VANDY8Euko2QsfZeG93+cdpyl1mRC9d/C1z354acAAMWj6Jkpzx/u7Dv4NBkNl1enj4i1Sppl/ahebGvlHbzz7eQkTS1wM/BePaqMsbjGxVhc8zlfz+FdFEOR99DRpHGufP60aTwgOq0KR6rHbNu+nfw+4l1Sv/RfwGl8Z1JBJjw3modJpc7NhkroiqIoHuHGdltMw3DRtk5MNcLuV2ctni6GVUulcIvcb1YI+WyLREuakps2S8Xpr97m9NVEyNUwLtY9oTGUfjMUZPF682qSdBImOrVsPOfNCProAopz2PRz+iRJQ23ZbAJLomvuAAAHUUlEQVQ6eLy7gfTYab61owrccpu6RP71AQ1hus6SUG9XDEzcPLPZYtWRPYiiB8vGkWH8YHVyzhNJ4E4y5kVd0rU674R4dtZ/920AQF7JRKev4mHarn2GjLrPrpGZfW585C9QZmQ7v8lwtD9NNC5BLrrPbmcz7tLpmfymd4n2x2YrC7fYfExWzpbmWSvdP85d8ynv9lYORsbVzFrUYd7R+gZWG/gCdM6ReTzOf+uVn8um8M5Xt0VFURSl73Xol8nt8g9THKPxtOhLI5lbYi56+7Y0UrnE/jetOfQbp6/JxAIVlXDWx/ojVKm8Udxqo2pHsRnXGmad+6FTJEkH7mOdeyiXpPad76bOefGXfzFN/HVjSONutLSSu13hvY84fc3H37ji49riFQer37/ESMJNMk9CrOCe2GAkc1F4pJZ+WoSPHMnonH1JvnANbDPSt6yXEjTPp9+E7klxMM/o1VvEc1VoLVeX8g5N4lJ5W37bYyuxUvvDTs/MeSTx3+4yutq8y0ER/FfmMs7y4hZ+dhY/T4Fh8VPWLidXxKQKKH2QC34c2Tk3zZFTI0IKkX/3tJTjMkUldEVRFI+gE7qiKIpH+MKqXORy8W2jLumU7o52uZxG9RKUllWyO7rWM3XD1p+QCW07TWrfY7ViuWhTc16Uhh+i0UfjYh1Ju7D9Ax5fNZiWqdLNsmec49kTyRGxNyKTyigHiTTGZ2X1TuVSf6p353x9D+niwuGw07d8sU1n3Ev1lLA3jzeByQ37Se+wU8hHcVHD83ojzYhlQ0k9EOEgT8TPk8pF2P8xedjXAAAJE3sZP8Ux2RHzlBcJg2bZg5SPpkAYBq891o045PS0tVD+nHr20OVoW+NDsOQSRz36NuWSWTD/W9x58YMUowGbe6ZuF7tGHz5PKpeyQa4fSEm3hNgfHUo1LGNUQlcURfEIN7aEPli0zX/bXGNTCYldruE/GRhF9xxP7hsnXB8PpfknHTZbmVfCdVXgUjjDGnHbjWQuvBGRE6T8LnnZbITZYyRSKaF/0ZHGoAVLtgIANq+dk3L8yVaW6Ifn9c7lsXQMWcVmTGbr2LJFZN3MOODK/o5h7uo4SQ/j0b3GgHYdpXJpi7S2fft8yNVr4xl6iPMGjHD6ikxGw7I7OWgn4aOj1J8gA2+rWLkUGaNo2VAOCuowJ8vu9gJcL9Y7rf0vkBHXH1nt9M02fouhAUiNKJSz8LlXqHExfTBaMrwSX/U8ubW+s3piqsGuzFwg/C3RlnJcpqiEriiK4hF0QlcURfEIN3akqKz110N1Ievubcf1wy5rrUHppNyZxhCbJb7L7RQ8iraw6XCpkxrgFTKiqbPLeoKWDnos8sz9qz3OlumCQtJP5Q+U+rffoDeMm0iFGd7bty5pn2sxgZh5TIU6oaOOrqn+0G6nb+UaMqIdvAp+/9bDWqpSbKyD9LEuAsUnNIroSmsgt3ZBtyxGQ4Ts1mlUQ/3FfqumsV95GDgOorTfVwAA9Rc5FXBoIvlgVz31pNNXXn750b+Xj/Wv/1vuGloJAMgeTubhlxfzruZ6ygez5vucBhkXbBJjqT/qrfrMxJ4MYOtsaSm96Ln5lB57yXo2z774ci0A4PWl5RkdXSNFFUVRbjJubAk9DVK6cBys0kj03T5rvK86W0WnFZGkP6CLC6OVoKwd0y8ML45JIwODbEpuSX3udMjf0coZ0c94f8DkmnArCLB2M0kLO3fscPqWPEWZBJuauAyILZyRewdnBAyfpZrzMZMSMleU2rNFJuznAKDd+M+dOslrm+ISkmSWLBiPVEyYstRpt7aQpGjztgBcom72ozMAAHnCnS43TZ4NNwm9ZR/dy1xhhbaFR1au+IbTt/NCZlGpl4tdBZYP4Mhj+3juu8AGOXvHnQyg4hh2vHRRtM+pXA3YjKV3GxkvINwOalzyulTcS6ueisefdvpmzkmX0MTmK5KGP3vlVzuy2RbdeM9sk12Gu2Pl2kylcjte+pRc6hwABojSeRd+nvnnoBK6oijKTYdO6IqiKB7hC6tyuSJsNJcM0wqZrbSLiERMqQgINU/URc1jVUOd0qbnUqTjSrmev6OXcFO5nN1D9zIgVC71h943WzaK1ux/k7YXyR/9av8CrHLhFM1NF8ggfEyMs0Uy7jZbGV5x1mzd/NZlIQh7LquSC4t91hQ6HKz6GTbsIQBASdUsp2/l+vKeX8GFL4m2DdW+tqorL6AqF0VRlJuMm1NCVxRF+QKhErqiKMpNhk7oiqIoHkEndEVRFI+gE7qiKIpHuK5GUUVRFOXaoRK6oiiKR9AJXVEUxSPohK4oiuIRdEJXFEXxCDqhK4qieASd0BVFUTyCTuiKoigeQSd0RVEUj6ATuqIoikfQCV1RFMUj6ISuKIriEXRCVxRF8Qg6oSuKongEndAVRVE8gk7oiqIoHkEndEVRFI+gE7qiKIpH0AldURTFI+iEriiK4hF0QlcURfEIOqEriqJ4BJ3QFUVRPIJO6IqiKB7h/wOLPmSgSE9BGgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAB6CAYAAACvHqiXAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJztnWeAFFW2x//dk2cYhszAgAxhiAICIkhOSlAUxKyriIriiq75qU8xoyui60NFXFxd4yqi4KpIFESyIEkZcgYJApNTd70P59577kxX9/QoMtqc35eqPlV961bo6nNPuh7HcSAIgiD8+fFWdgcEQRCEk4O80AVBECIEeaELgiBECPJCFwRBiBDkhS4IghAhyAtdEAQhQpAXuiAIQoQgL3RBEIQIQV7ogiAIEUL0qTyYx+ORtFRBEIQK4jiOJ5z9REMXBEGIEOSFLgiCECHIC10QBCFCkBe6IAhChHBKnaJurN/2PwCA3dvyjWzNkq0AgOuu+ouR/bx3JwCgmqcEAFB4YrfZdijnCACgbYdWRpaS0RoA8NP3s4ysJDYGANCmZVcAQFRsPe6I7ygAwBNVxYiO/PgtAMCPOCOr0aQtAGBnZhYA4MDGtWbbLyc20LEbxhpZoTeZzmnlLiPbv4vOJblGipL4zbbj+w4DAKqUsP/YcaLomLn8/zt9Xh5s/A634QH5Txw4FZTZfpdQ/utw9wv1XZfv/dqm3HBry6XbXm+gTuNrRvfFExfPwj0FtKxam2XZOWolSjdmtZ9Ny337WZZgtaco/JyWH2fS/duZzG0c7EzLKh14/wx1Dj6rjRi11N985X3etmo6LVPPY1kj9dgPOJ9l7VTXLlGfowJ6CpR3392upXBqkTsgCIIQIVS+hr6pKgBgzTffGVn+sQMAAK/vuJEdPUwa7sGjPwMAzs5g7Tq9BWnmMXWSjMxbshEA0LLWL0b2w+rVAAB/Huk0sV17m22OhzQkT3SakdU5sxcAYM/ixUb23TI6/lOT5wMAxl3T3Wyr7WymfkSzfhMbR+dXvy73t4oaBcQojWfbnj28fwxti7buTH4JafyF3hMITqC66vkNMrf23PT4sp+coFvKtuUJ2M81MMv9oOXu57Mmboky37UaCdGeJ7UprWRbNyFFPUe5RSzz1qJldTWqi+JRErISaGmJeGzCB48fSstrhwbqVnn0uOKjd1m2rQEte/ThNlqo1WL1eUkLPnf/XbTcmsxtpLamL9xinV6Dsgd3uT6l9XP1KSImyAkxavyTIRq6IAhChCAvdEEQhAih0k0ufXtdAQBIyM02supR5PBb8+23Rnb82CEAQNHhvQCAo9WKzbY6GWcAADxJ7NBEXA8AgLdZPyPq3GydWmuhlkfMNo9xA+UYGbw1AQC+QjbbbN1FztAF078EAAxM57Fsl0bk0PTGJBpZSUIqACCmOIHPL5ocwC1bnwkAaJ6RbrYVZtN1WLpwgZElViUnXcMkbhfYDhtP2MaOX48nDPuHez8CZaH3c3G+hTsaVvtF2dYVT3DzkSutlAfxKDuy0bi6kvFzB5/Sh4rU81GYy9u2KjNam4v4mMuV7aRunYBDslWI+5rUkXo5smNgFzM/5DNYqB7ZhtfR8rwzeb/jymq5nf36qKN+9ckVfGacEJ/+3FTmuXSy1rV+vfJXtyYauiAIQoRQ6Rp6fNYOAMCQIRxXtXr+1wCArXs5NLFh/WoAgEKHlkXFBWZbyY8/AABi4ptyw3WUxu3sYJlXOzxJ28e26byt6WVq5SjLfOQATavFWnvKnp9L9T9z1UKzfvd1FwIAjh1j5+W4Z6cBALKPlxjZldd1AwDk5lB4ZkJyitmW4COnW1UUcrdBala15FoIhpuT0X0/pqJ6vGO068D23Bx+oXsUriwUgU7Ok+HeOnIGOcOLvauMLHPuIgBAdz8HDMboEMMa9WnZ1FKD42rQctU8lrVQI8M+vYzIqaM0tP+5mZZNrFBahdsIo+WVfIYtXvgP9XcYjSjmF3Q229J79AUAnHs2f7eHcsQmurQbNk7kOBINSe15PXezWsl33fXk8f1JbU00dEEQhAhBXuiCIAgRgsc5hXGkbuVz50wgp1Ht9NZGtucImTXyc9i8UXiUTCdVVEZkchL/F7VuQLHeqd0zuOH6ZJ7wb/2RZYnkwMxclAkAqGeZMFLSKBLXUz2V9z9D5eAlWal6vvUAAG/MkwCAv/bimPP/+2YTAGDm4/2NbNjjbDbSbPyaxrzRJcqJFsfO3KIsctTu+maNkRUoa83SzWxmemFuGReVlSlaUaOLE67j0Zg1QhtdwutH8L7ZZpuQcc6e4P0IN2zd62J2KCyga7lpCR979dfkhD62NdPI4rPJQR+nzIZxJ3ib76eZAIC/2IHommbNeD1b5U5Eqe9mWTmgfW+i5b8nsSzFxez1zrO0so2c8sivzhs3kPlv4c4aRhQ75kEAQNfbeTf3XARzBLVT6Pvy588U9bisu9w/V/S5h7t/xZDyuYIgCKcZle4UjS6g8K7cn/kfPzWFwgVTLGfhL9GkOddIpVoae37eZrZtVXVPcjftNbLGaeSo8ra92DoaaSm19lJYUHwSay0LZ5Ej9txOnG0aV1fV7UiKMbJnrni6VP8funMof3BoNHC5i1Zu/3O27EsavD/vIACgJJvD3fx5pL3VyWUHW/U4UtFXbeSM1bKE6xT1W4rVzG/I2XrgIPV73QYezVw/aiQduzr3/Mgh0h67tuDHRm9l5e3XO9oct++6Oe60whjiu79lnBCjSvG07cPn3raP1qot7Vodxa8iGY/woApju5KWfC3YQe5RV8s5zqO6ZR4Kue3iJ2eok8NO9qgvp9DKMCuMrdVw2i9rJ8ty1H2rpbJY21txjtXJyd+7YLkRHRu3FADw1N5pRvbws/S8uWt4oRzYoYZ1SdZ6btC9/jjwc+2JaQMAcIp/CPO7un5TQci9fm9EQxcEQYgQ5IUuCIIQIVS6UzR33t0AgLysLCM7UUTZknGF7GBIqUUmlGKVDbotdyu3kUf7RSdy2dquA2jYGRXdzjoafde3h4a13hoc81uwjUw4sdXYzOPL30fLFO5HUv2XQp1iUP4+nkv73vvA6wAAfzEVEFu8gLNCd+fRdWjv52Oe2YBMLpM/ZafbX5/bV6p9t/K5NvrCd+hzjZGtXajrrJKJC17OZh007GoAwK2332JkF/VpHNDu0WJq+fkXafjer99As+38s6vqDoXsW9leOiGdrixzc846oZyzYTry2MH8G8xHu9SxmpzBwvaUp4DX+Bkq8NAzO/tLureH13Dc+gVR9DtIjeGcBPQ+R7VlldOao+7jRlXgLs/KKFamRNThsr+7DpKprTE4iOCAn0o/1zWScgqZhXUt06x1/du08kJMwzGWzCp+9jsyIo3P9LyR5Hxu24WDGdq0pkzzz778CgDwn5lzzbbZ8+n36lPX7FQgTlFBEITTjErX0PM2k7bqWH8t8amkOXtiWatwCo/RMqpQtcVf8MTpsMKq3EjxpwAA37bNRhTVpCW1UUSOWE+VnlZPlJYK1vKdEhoFHDzM5W3T0m5BRbj99gEAgJf+8SD310O1PPzF5MR9581/mG1frCfP2iXdWbMb1owcVot9nAk7sNvbpY7jt+5jqL/yOukDzPrhXfNC7OkGOfNefJO/t2M3jRRefuwal/2Voy2G+/3p3GUAgKG9uAZOOEGW4enstsAJ2NagxRAj++jtjwAAPbtVRVn0tfwtFXEc5X1+NIqf0/uUU35b09uMrLgB9anzgq4BbXi9d+ject9KyJHusYrVOGX6u7QDTwzTbe17JLPanazbepaDCKbeT9q0cQuGG/dpXfnQYYtX0SKRtfYzrqfs7IxkzsRe9zw9nzXUOdn6uv4V2pN7hAou1L/uthktjOy+++4BADTsyGHS/mKqxRPd5cIQ/Q9k+r2Pm/VLJz5VpkcAUIKThWjogiAIpxmVHrb43jsTAABpjZsY2cDLaCKskv1bjMzJpfCvmIyWSmLXNVG28HfHG8mkkY8CADKtv+4uSuntOoJOu9lQrsSI7qpUXczVRuSJphC1ZesmoyKMscymL7+sp8BjHclxyIbpdegfvGlttofedQvZ8do3Znt2Qhxpwe2LmltHKa2hhxuoFx0TOLnYDXe8CgBYvpyrW/64/AOXb9Mxbh/JtUh+2Ekxe1MmvQwAyDrI55mjVKnZC1gTPL97FdXDilVbdLOruyYgqTBHW2PTWsuBLTwdYc9u1RCccGfVCM6e9yhc8ClLNk3V5Kmzje2x97e6IEQrOlzxMRapkxk5aZ0RPT6GpkU8qhTCDX3v5f2Vhv5P9DGiE8qO/fE59Y2MnwqdZFbeZCAVG9l7zqJR9B03jzay0aNpdLR4FtdPara8CwBg20IayXEP+T7a1U/crNgXxFCHJ0x8EQDQvA+PzDyNG9KK5W/z/sr7fMmEcWZ9Riz9Di59/kMjK/bpMUUxThWioQuCIEQI8kIXBEGIECrd5HLjvWRCKZq/wsjmPUShVknWSKh5U+pqjTyaw9PT8q9mW+6zDwAAXnx0opHpIWQ361j5KoFzxYs0Nl374myzrUoUrfd7eiq3MYiKXYwYzE6sUGjX7B4rUbRkOTlDo89hx4zHr2pu7N8JAOh+Btes8TRSQ8ICy6ESTY7ghH2lQxVLE3oSB711wJBBRvbOy3TOqankpH3743+bbZ3P0CYXdhr6/MpUZGVvdmpMQ9cG6WQ2snyAqKYcd5cPbGh1KtBc4lZ6l/cvUxe31IkFD01csp4z9nq0pSntt2ezO63j2VRD6HjmF0GPabf+xE1UovbRf17OR1TXwS2wYOrML9Uam842KSfxJnC47GufdtGNleo/APj8S9WZBV6XmFj+6WaSnxRPjv8JAPBYdy4DO/UGug4vvcW/jdYOmTlX9eMaRfnVKAO654tknoi62jqnGJd7EFamKIdPNqxLmbP92lnZo+pnUK34oBHd/CRNApKaRYaVrNlsRt30KWVtruEYBeh86uHpXL9mzF33AQAy+p1LPW1iPX+x2tRycieBuXA0TdSzoCbPg3zXS3MAACv3bXH9zu+BaOiCIAgRQqWHLfr9b9GKY9VIWaj+0XpZSUG5KqkmWe3ncL2KrlHk3LS18e4ux9f6mXbB2J3ZqJbfWbJ1KJ/R1rpOS7jPkv2vip26aeFbllRpJMVfBfYkWjl9Hf6nRz6FMma/y/VrUsaUDtRySyxyrHa1bNxEntTjiXtGlDkbe8BGI4SvFx8zkvO6pSAYw66n6pOfvf0IC120TqtDLh9cnJFlnJ1B2yvDRVdzP2a+T33bk8ff+8djbwEAJk64MeC7+lramrHb70Rr6NlZtK1qijUFoqnpUV71PZqA4tt8CgXtmWDrWOS49ft5CsQBlzwPAEiqwUlBk18fBgB4fDxNcHHHBY3Mtiuvpuqgjj/PyCZuGQsAOA88gYcO7v1MnXOS9ez0B41mGre9w8iyJ9BvrhbPSxMQtuhtyuGT7VMpLLhJImvo115Hz19yKmvQ/frXRGm4H6uGk3Z9zQweaenJ/GZ8xkECiUm0X0wqjS6j2gxGIPb91JNYnLA26+kH1W/CscJbvXZwgt6ffi/OnrVG9MO0TwAAne591uX4FUPCFgVBEE4z5IUuCIIQIVS6UxQrRtLSrq6pZy3fbBk9dChnWxoc3hfHsbbancq5X2bWUKy3ZHqw16zMPgDndIVjZgHYKGDPMKp8PBhoy0xJVS7V61EmFyemg/rMt8EB1YnweNik5CRRfZfks0IM3x17NdA8oGUrliwP2MYEZrZt3MIldc/r1i2g/QUrabj69UwyHx0rZFPHflVut01DF73BCfrB5TPCMrPYDBt+kVn3erWLnNtIajAEwQk+unWb3zPJJL1yLoW3A2V++tf8M2D/uh3Zq/fzeoo1X7c+YDdAxa1/tIxNDPM/eyBgr8/fVlm6Poo5P37siNl25miKUb/WqiJ93hAyQWEzmyK0EWGPukY/We23BJmD0tdzu7XaWTkcQfj7kzyDRno+Pbu1q3NkeXQiOYyjqyYjOHy9W/Uj+07VGZxPcM0gCjao0YnNs74dNNFMVIuzAptT5rT5w3hbci167ju1YFMVEug3edRHx7/+nv+aTQ3iSDYl25p/OIrMkZ7qbD5q29quI3VqEA1dEAQhQqh8Db2pcu6ssEJ7dHTeK9Z+j6l/6sWkj79gKZM6SMuqMWdqQNhzqGslv56KdSqyitjtRXjoMCmtM82wtun1yyxZnome+9CSquGIn8IznR1WUfwmOkTMqliRq8YBrloc0a73KD5mNjl39h84YGT16tKV2LHu0+CNuHD3Dexe/vhDCt9MiOcsu/kzSlefHG9lyk0YR8635mezNte0Cd3vQUN5HLN1804AQGEh3ZDcHB6uZWZSLmC0hx/VrDyqQnjkADsLD22nEMwx99OkEAu+5WxMfsw5Yy9371cIigmttFqIIk2wxMcVL7W2ft7jeozII6iENqQZ51qTXmh8NXnyEhRTeOigDsOCdqd508Ds3tINlg5n7dqMNd6kdFoe3cjbr91MgQX/a32npdKEH1b1Fn1oY7bVBE0uE2/XSgqjTMlZzc40673OUtq4pUIe3E3Xq1paeHplUneaTObsBNbQBw4hpzLi+N5GtVCZ19H2r5/4uhlVPR28IzCU8J/N+Pdy/es07V9inspQ93J469QCejImf/KukXkvV6ORWK4cGVX11OvLoqELgiBECPJCFwRBiBAq3eTi30Bx1/4feQgUrU0hLa0d08YAABY/MSWgjVtb0HBuR2a+kemBT2drP+XfQFVVPKuE58hASZg+N91uYYh9PrbWp5lkvFhLqswpHtVajGVy0cP9Amsuw2y1DDHyrl6dY8S9ajz88CgeVN9yMzkBq1h3vM+gkQCAFfM4QzQUS79+NfhGLzmDsk/wRCVpZ5ITdctqtjskKQ/iwd2HjezoYXK2HT1IspwCNrn07X0+AODjaVwszKNqLefn811IzaAM2OISGsbHRsebbZfd+AQAYMdWdvXl+2iIvnGxSxEyZUopVQjVtzlgN/3IVDVZm+zwy333qYD9NUfmjLU+UV7A5hBJwIf3lPNwetJKdahpFA/716r5JLzr2UbyPmhuWs7PBObWpozm1GyV2ZrIce6IV89WgVXQTD+e9hwWZWjbku/Bns10X+o3Zh0yNb2C+mQ7MksNH/kfI6qbrPqbww5b1KyDYCxXphb7p/Rq194AgBeW8XyuN/WnGPrHR5DDeUY+X7/cN8ls6Ym2ckUcZfI5xA7vvJ2bQp3N74Jo6IIgCBFCpWeKai1ocCrL0pSCcWA/y0bdQv8997xK//T2RFZaPxpgybQbzp6+QB8rRbVfx1Kajyul0HZoardkG0tm+ZbCwr+DtDZPo9ctqdZOtQbDtTcc6FDGIkumwuFythuZt8y8DD5ryrpcpUzY2rhbuJ1mzCNvAgAmPxWYNRkul46m+iAfTb47YNuXSzkDLy+HRlEFhazJHzhELulNG0iDrt+Qywk/eRfF20UntDUyfyHdhQfG88QP7dtRGNqK76kGyNDBPFlB30727PPEhFfIaXr/2PMDtvn1tbSu2dOPUHjhw0+cHbD/rdd+BgB484P3jazE0eGe2609aVqzaHDdkTpKV/xqNtUQan9+YF/91lSM3jhy/SfgHCPbMG0+AOCOqygj8fUFD5ltd1EVWgw9wm3c/Cw9GPZULS9drrTTj/QI2Arh8yjnpv0ITVI1bW5lnbBspmhOLjv2f9miNPQzo6z9UUGovSWPconrJtWokdSufY3MiaeXiacjOVFRzFmyX9Sn39cFKosTANCDni1vTPAO+X1WyLC6DsWvcXZ7TDtythb9zKGMy5bNBAD0meBSL6iCSKaoIAjCaYa80AVBECKESneKelS91dlHrbKhxYFxwAe+o0+2qUWj3RVfWzKds2cPkLULI1/5L1KsUZR2Kd5u7a8jW23rhjaEhF0QM41KdjlWDLRHeXsd6KGg7TzRJcbsoHNlnohHUOzxGJtawiup+9qTFMP+6pOjAncsd27JEDsq0RBrdiCPKjjlWBkCHrRQMp51vWz72sxik3WcnaeOnxzLu7aRp7tXRzZdXH4rmYOmTbk/1JkE6z4AIKN+g4DthSqteMr7NMNWIjgzUGfT+sAOb4/KuDyBK4wswas89OfTsN8O7zbaVhzbBv1mj0W846XpAIDPlSln1Z1sfPSMJtNM4R4+G33lS9k/PxqoZAWqr4zjULsex7oGk9VDduulCMaLj3OWxgOPk0PT3czi9pC5ZDsrh/eRg2zCS8ynZ6DKas4r+PEw3ftzUpTHdi97nOv2UXO39mYTnibDe65Z32JKF8cH7WJOCT9/s++9iVaq8dti52EyrUarL5dUcJanX4No6IIgCBFCpWvofjU7utfKPkQJhaOlVOcylfF1dHZd8HRJ+/9vp1raFTsOozSLrWRMndNm14PRWrtd80VPb6ALYtrH1FGWQ63ors9eJTXun8+9Y2SD1cH6nkX//rlN+O+/802kXTn52UYGv9Luq9oa7DyUD/dOK9JTP+Hrd+MI0lI8LmF6oeaRLDU5RRlnq+MyBnBrKlxZKJ3mtee4ePG+vRSi2a4jaclvfLDYbOvZjYJXpwVGvLri5j++Ykxg1iFyS9fWySunEpC+Ngl21nCZ8jzuGlZ5aZl7Sy2TvucU6xrVSUM/EsUHOqh+9vtLtUvOYY+OQxw6xmzxjFRhAVe9wbvXSy+nT8DxXzgwMjouxI5hzod74ie6vkf38mitZDtp39u5ai1+VH7J/BPkgE+0ghQb97eLbJema/vGZn3LGtLQO0efGWx3bLVCkRetpPH/bj+HMlZVwRfbl1G/X5rJ5YonPmONhk8ioqELgiBECJWuoWvVsSSPk0SqqGpmOSd2siyxj1oLrqHbJ7NALZ+wZJ+rpf6/TrC26aogdt03XUnxmCXTWruu2GiHSl6p/h7bTuKMqHEbqDbGV1wmArvUUGGLqglx4ABrT313kydg3XcctuhVprpLewXXysuGjAFA/eY83dyBrTQ5hmOFQ44sIY9EuA+B1vK/3cz+gJ7N6dusqQdOLfdbeHvm2hBbuR8z36MZ2Ke/Q9ey2NJ81Ux4GP8MVxc8uDl4LRcnxMjCJu7ckzuN2cni/8CjwUVv3gAAuLZ/byOrC7J7j8Vq/tJAFXJ5FYVFJo/kGj65n5Oq2Q886pnb1B63utPz/MAQz/IpPSazonHx8TQKNf1qHtvEY9UgI9v6mtaRt2+gBKDebXh4EKce9p/+xu+RKR/oEEZOzhuZSqHEjRrRyGz++LfMtox2NGLJKuI3yBLVT/tpff3q4QCABueQlv/8OaztfzGP4kkzl4c5bAwT0dAFQRAiBHmhC4IgRAiVbnLR2Vl+a4xcoJyVfmtIXZCfU25bTax1XXkj35LpI+hgo4usbdqdZUcGatNMLUumB3t6sGXvn6zSU7/2sXPF05DC0p564yYja964KQDg3fcoO7Aomd21d39FIVlFVj3f6/tTT+57w/LiliGrkB1cVdW4cv/mWQH7ebx8lXjaBxqvus0uX/rLtNBmFlvGwYvhObjccFxiJSc+T7O+e2NaGZm/2J5+geg9mDJUfaqJOEtV2XWChL9YWXz9hj8MAJj/qTWXrTl68FBMdyetDlcM7RTtEHJreGjjgDX7JUZBzQ3btBcAYN62O822LbPIFOUZscDILq5JWbQ9jlo1T+Yph2csOd7zYP/eyAm5yP6FvVAD5dFzUGBooI2+lv/9gUM7hw26gLYdUWnijhUc4Fip4wp9m92mftGxBBfGs2kkczVl/M5atMvIvlRffm9UTyMr9NH8sItXkRFl6SzOLD2+l/pRpSWntycm0pugRww7Sm/+l5WNitJP049LJwMAhtzNuelfv3QefiuioQuCIEQIla6he1TYYnQM/3/5XEofrli8tNy26kfx6Wz2kcZqVz7spJa69p+teet1uwqbruGy05JtUEut29jzvB9Lo3PY6uN/7l2bqZbH91vYCZN1IWlBC1ZQEkySNTNHoXKeOlbxuMytpJnXsjtcZsBSxapDkatGO0kutSmatOe6MTmm5oueVIP3c9NRjQYfUvEOTyu3awjF1u4CACg5ulJJ+C5cfD1Nabf2Ww71S0ih/QuzVxrZwq8mAgAuv44cW5++w1PhfbuQNPqiEyuMbMEMq9Rm2b65OEWdgBXew+sh55ffCdTQ60bzyGxISeB0dCFp+CgAoHAvD9cuc94M2G3Uvap2yzG638228VhgaxRdK8fLOuyPR2mqur+CHcMvqmdh9efkSI9FutlWqAJ+655xGx80ZBgisXYHa9edmlHf9pzgkeSJKGrkyffeNjLnEI8kgsPOywm3UQXQhZ9MNbIZP1Ntm779SNsffCPX64lvQFVBG3zGSU+tVtK0jHUac4XJBfPpLdGyJY0yGrbkpLFWnSgMtuEw1qjPH+sShlg2/tV6dvQo8LOJHIqcMJkqPKLgHfxaREMXBEGIEOSFLgiCECFUevlcTZQ1OjGOLcsgVFv5NfZaPpKytIth58e6YnKH2oMePQjX+ad27p+upmKbXHQJXrsCiK4X00MtbUdXwyso5nxOLA/d4lWDSVmcNdeqW0cAwIqVZAKIS+D/1fVbKDZ8qzX7wDm16Lxa12Sn1KtfboDNB1/8bNZHDKoNALjz0beM7B9PjAQARLtZRDwBK3A3ugSXhXKo2jf98lueAwBMm/I/QfcvD08U1RRxfIEzwd54DzmbOnbimN8dW3YDACaMuzpgfzccR5snAs/T/rn4qFnEpOv7sjOgrdrt2Oi3/30yzUS15vutfxG6OKrHSt7UspqxdxnZMecfAcdoqpY6f7IEfzPbCreSKaq65c8soGq/+NslnYzs38oQye5J291P0gn/5hyGu66lH6ffuh4xUaX1wzH3c8noC/qReeL7TM7yfHf6HADAloUfWd+yp91wp+9tHA8/b1LwzM+wKVHBBnYuh7e0t98ptGoxxfMEIiwMWCn9sITBJ+oldFnrQD1byucKgiCcZlS6U1Rj/9PHKq3WY6nLcWpOAI/S0N3++36G/c+ZH7Cfzhp9QS17WH9nk5RSZmvo2rX5mcuxdFWGC6xDHqus2lJBAAAOFElEQVRBevv3S9mx0yKZHD833nKNkVVJpbHBjjUUXGlnybavQZUHs/duM7JtO0gzKvEHz84bMbC2Wd+gJjMYOvgSI3PVzDXuHr/gslKKq8deuO523V2TjMxNM3/1XZqU4sYrabxjB2cOv4KmRvv6k+f4kC6auaZLZ6pdMmYkT9YRXyMl2O6uuDpF1WXYPYmzFBsOpLoni94jx/fY0Q+abWtzaVKDo+u+NLIVE8khd24H1vacsdqxVqCOzejJzI7Zepf3cVr6xxmRflL+pfuF+WZbD/VojXmWHcivP9QF5VMQIPHl8aQk2w5S2GKz1IDdDFNemWzW8+tRFaS3PrFqly7WmvbPCA+6I+NuLEcrL1ObqNQT7PasR4eY21GPnEJq5dYHN63ctSOBXNQy9PZwEA1dEAQhQpAXuiAIQoRQ+U5Rl+GIMbVYBiFH+WN0+GuhS7cv/cutZn3aO5MDd1Do2PE7LNkzQfd2R88KeXcGj7sWNaBsxkMx7LmNyqXSXu3rsgu2SjKVAPtxHTk2Dx2y5oxUTrGUmhzoG5NGQ/u9h3cbWeYPpYfEL03l0pxjbyCna6kY8hBzirplioZ8LjzhOX70WUW7FA7beZjNDk0aUmy8v+DHgP1emEIFyZKqcsT/rVcGNxm06notAKB50xZGNuO9RwJ39Kihv5MVsMnvd8s7JD4exdl/5w6iSRuOVqMH9v7bJphtc7ZT0aVBDTkT9Z4WdMyWc9kUllYql7k02hk/wZLdMYr2f/lN24xUhFNBxwye/mXsWDKBXXgxByLUblT2PvNz4r2UAgD8M6zntvivasWO39ftaTMk34v+V9AcnXM+uBDhwM98YPbyqXzvhWty0ZujXH4v4hQVBEE4zfjDOEVtHO0Vs71jSmsvdiln0l/9d7Xq2c/IukWTA3HJv6YH7K+TLO1JzaqU2RYMHQzXSy33W1meRwt2q26zVrEvlzSSg+s5MzG+iP6La9Skug+1MzjM8afl5Cg9M4NrZWRmUbu7tgc6qjQLF7DzLTGJHDglPj6bW686N+A7BhftPZRGH+q7fkvzcdPMY1LIaTn+ubf4Oy6aueae0ZRJt3x7cE0WADp0pyy7Q1lU5SQ21sWJZeOEbi8AdUv/8tYLRlT0Fo0IvWpalA51OQvXoyYu/KU9O3CHf0Hu9RxLK9dZy/qb9mSEtmau+fDD8froFep+Ranq4XuXpcI46+79xcj2fvMdAGBP1wEIB/8MClFEse1F1cWqMyyZnlhDH5/rES06QU5l+zdqZ2ozpVViWxnnx9U1D/j3IczmT0YxZtHQBUEQIgR5oQuCIEQIlW9yiXWJ33QbTabQfv5jOnWLN7VQf0s7v+cCTuu2U3RuqIFVFevvbEQ3Kr7T5GwufTvupRkoizamnK0a/s4qKlYzuT4A4FAeV9bKyiHnTrJVzzXvGHk+d++nE73iah6Gdm9PDr/lX3CBnvw8sjPVTeM2dmWVdtxNfuNhs/7IuH8DAFq3TjcytxmNQkM2rhoZbMa68drrAAD9B/BMrX51hR995F4AwKr5gcWjbP7+AmUu3nVTnwr1pksTdr699QkVUxo5gp2jw4ZTnPPUqeSMXLd+JULRrvdI2m/hGyH30yz8nO5BlTie+WfA1RRT375TOgCgd0OeA3fReHKzP/Tfe0K2qzONH4sn08WOrpxPgG8+CNj/UN4TAbKTyQXK3T/LOR6wbXY+92fh9EwAQO1L+gfsx1i/uGIVP49h1nZdzrh0CThChx2cwU3Morj1lFvZtHWl8qu+ZlXqrepiajE9OpXO0EpANHRBEIQIofLDFkN+Ifim1lZLlyqH6eyz+QvL1tIOUZyEaWrEaM63HJrz1aggLZ3b2LU1eJTlVLWyZQA3Mvs4OcJKvJx5lt6ZtIkmtasZ2aSnaP5GHR3XoydrZSkxpJG0r86uny2babRxKJbbnb/COjEAyw5yEZCu9XSdXTu88Bj+CKzfR/1sm/brB4eFPrpwcVbtkFg1u32vXpStO3fGxJBtJDQgjTh/79yAbTps0W/5oF99gDI/UzsMNLJBV9A5PPUMlUb++9N2ZZ/goY9/FOz8yG7qUxxohDoXwcsLA4BXhQc8/QHPovngVSEyLg11rXWtoVsFbEwNmfou+1+slmNZdBuN3P71MouuU4/FyXAyFqjo2hUb2WwQXYPue8d6/Py5JZJWFD3orhYtYYuCIAinPZVvQw9FCH0+zd5NKQZrN1pfUApsWa3cJtdSDErU+r7toQcReus+tTLnmzyzbWMCrZ/ZMcnI1qycDQDYYIX3VVGzY3RsR/vt3s3a88KNNJlAchf+y29Ui7SU6XO5jkhZrhpyrdXJQPvnH4W9OynpqsXZnFyTuSowtLSiFB2jxJW5M1aUsyfhppmXxWsVHLx9Ik2WUGwNdJbMIrv6jhV68pU/nlaun7oGnkZGdsCh6dfs+VKqqT0TPfRKiHZY10tVE0rsA598hpcSt5wKv0HKq9uih0XbyywB02MPT9t29S00oqjn5d9XsfqVzlHdrcODY3QOoee++e5Csz7maZqMpDhTV8vkRDh4aCTWdfhoI1oyjdeZ4FMZauy5fGo1vibofuEiGrogCEKEIC90QRCECOGPbXIJQbK1/pXyV+RbJhSPGi6HSgisaYU6RalyEr5i933Loqb+RLY1ZCpQJVxWfZPL/dBX2MWSsyaa9stzsZB8tJw7MrgumVrirJFbUZn2dqz+EH8GBnenDNgiH6f8xkZVTK9Yvamw/J0sPviCnMo79/Pw/aHRFObohOssVma9mOos6tWLhBu/pIzf6ZZ+5KtE80uStd6/CYV2njjEz1NRDplcGlj7DWtDc4n6/eTkL/qJn+EslZvZChxe22cYmT0yzjgZrsdw+ZwWTlMjObyfnN/fN+cfxFrtm1WPyZByurh8DoW43nSDNZWNL0TYq7MaALBsOk88suQEmVy6VaxSs23IQcmeORX7sguioQuCIEQIf0wNXf+jJlgiFR1YVeXsNLP+dWfqP2dLKQqnVMfna3hdR2/2PIdl34bwr+1UywK7ToQqkOhYCqSjRg0e+69TXfVjKmor2qr5X6M2aXtp0RwmNVMV+Ig/lcrQ7wZdMHtCk9EPkANqynM3Bf3WmgOsy3SoV7EYsa490gEAVyRz0tg9o+jihz06UPfR2cMdL9hE47TvZr0G4NRo5foRsAME9cBU/1xszXvDdtI0z4jj6pOt1YPao3Y6txFFrazaSNUtD4LrtrQEefF7NL3YyLKVJurJqYxEnZfM2tyJdNZxx7mq5WjlM62nBhRuP5uSPB4h3v68mirPt8Zlz1DwD/2JiVRz5ovHBhmZV4W/OiqM2evSkUvveN369NuDGURDFwRBiBDkhS4IghAh/LEzRe2/mzKj2Sut4cuHJ/MUQhwT4Ak29CTqP7lttBPf1MguyrIS1GxHBzn0kzqA7eNTokSetB5568vp85+cvfl00vXV9Vu0gc1NLZqSCapeFatQqsOx/+HQ+4LbAAALPp8UsM2txo2/UN0E6z4WLKE+rfiGY+YfG0+FRL7x//Ys3JpqGWfJDqulXSK2oxq+r/SzyUDnFGtX4TZrf/3TSLce7Fz1kNnOU3183VZryxrbI5qKRS8r2WlkzQfRZBcDHrzTyPr0DidT1CbMmR9CoieO4Tln0ZSyeWPb0JV7+85Ys2njGsr4febhF43MX7hErdm1uUOZzzxlloC+c554rjPTpRdNElonlcxdD77ygNn2ymTKXXj3vu4hjsNIpqggCMJpxh9bQw+B7RYz7jJbQXCZCEMTp7xGhQcsoS6nYitbLtFxWm/R/4TV+c8fh1TlRSfPOk23My6rmNh/q/rEKhaZh8IS1iiKVbs51jx9SSrmMdb6n9fJq6/+awEA4IOPPzbbHn7wMQDADxt+MLLEeDrZWrU5fG3XLgqBy8kl3S6tHtfeiI6hk4mztOBDv9AF3vQTj23ankWVLu8d1TfgvPQZDBjOIWIH9+8HAHQ7t4eR1U6kcMjLbx4BAKhXj3XduiG8yW4a+ok59EQlVuEHavNKmi7w4YduNrLPcpYHbbei6F7Yj4u+yr0TOGX1hJ/OZVYhe/31FdeVZJZZbZxQS9tRul8t7dFAK7VspH5ESR7+xgKVWWpPrzK0E03g2Ov2541s5A12i2XRgcZ2tIKOBvgFJ5eeaqlDD4NPDEPo+2xr5aFeVXp/+9kJFe+snr/YC1hUpCfuKAnY2w3R0AVBEE4z5IUuCIIQIfxpTS6/Ba+KafdboyRvUxpGOce5i87BQMdIWWtJghXJn69HRVZFsGg1OvNbaZ7+fNquTR6lboFbPR8nqIjbtGeqr2i8+h/irvxGfuU5u5lcDi2lLMkqJbxt1aKdAIBl371rZF/NpRjiRUWUHBHCyheUst22b4UunnVRIrtF1+aTQ3i1w/e7s1rqKPvvrDa0VdFOONGPvZX+AD01R7zq0RZr22HVqwwrMaRZO8q07TCYHX1PPNcawXFJLoFO1T55pqvwcXtgwv0heMssgYrf/Yr96MTkIgiCcJpxWmrogiAIfyZEQxcEQTjNkBe6IAhChCAvdEEQhAhBXuiCIAgRwil1igqCIAi/H6KhC4IgRAjyQhcEQYgQ5IUuCIIQIcgLXRAEIUKQF7ogCEKEIC90QRCECEFe6IIgCBGCvNAFQRAiBHmhC4IgRAjyQhcEQYgQ5IUuCIIQIcgLXRAEIUKQF7ogCEKEIC90QRCECEFe6IIgCBGCvNAFQRAiBHmhC4IgRAjyQhcEQYgQ5IUuCIIQIcgLXRAEIUKQF7ogCEKEIC90QRCECEFe6IIgCBHC/wOSws7+Ijqw7gAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GroundTruth:   deer  frog   dog  frog\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot some five set of reconstructions\n",
    "for i in range(5):\n",
    "    images, labels = dataiter.next()\n",
    "    print('GroundTruth: ', ' '.join('%5s' % classes[labels[j]] for j in range(4)))\n",
    "    imshow(torchvision.utils.make_grid(images))\n",
    "\n",
    "    images = Variable(images.cuda())\n",
    "\n",
    "    decoded_imgs = autoencoder(images)[1]\n",
    "    imshow(torchvision.utils.make_grid(decoded_imgs.data))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [default]",
   "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": "3.5.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}