MNIST_GAN_Exercise Fixed Updated.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Generative Adversarial Network\n",
    "\n",
    "In this notebook, we'll be building a generative adversarial network (GAN) trained on the MNIST dataset. From this, we'll be able to generate new handwritten digits!\n",
    "\n",
    "GANs were [first reported on](https://arxiv.org/abs/1406.2661) in 2014 from Ian Goodfellow and others in Yoshua Bengio's lab. Since then, GANs have exploded in popularity. Here are a few examples to check out:\n",
    "\n",
    "* [Pix2Pix](https://affinelayer.com/pixsrv/) \n",
    "* [CycleGAN & Pix2Pix in PyTorch, Jun-Yan Zhu](https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix)\n",
    "* [A list of generative models](https://github.com/wiseodd/generative-models)\n",
    "\n",
    "The idea behind GANs is that you have two networks, a generator $G$ and a discriminator $D$, competing against each other. The generator makes \"fake\" data to pass to the discriminator. The discriminator also sees real training data and predicts if the data it's received is real or fake. \n",
    "> * The generator is trained to fool the discriminator, it wants to output data that looks _as close as possible_ to real, training data. \n",
    "* The discriminator is a classifier that is trained to figure out which data is real and which is fake. \n",
    "\n",
    "What ends up happening is that the generator learns to make data that is indistinguishable from real data to the discriminator.\n",
    "\n",
    "<img src='assets/gan_pipeline.png' width=70% />\n",
    "\n",
    "The general structure of a GAN is shown in the diagram above, using MNIST images as data. The latent sample is a random vector that the generator uses to construct its fake images. This is often called a **latent vector** and that vector space is called **latent space**. As the generator trains, it figures out how to map latent vectors to recognizable images that can fool the discriminator.\n",
    "\n",
    "If you're interested in generating only new images, you can throw out the discriminator after training. In this notebook, I'll show you how to define and train these adversarial networks in PyTorch and generate new images!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "\n",
    "import numpy as np\n",
    "import torch\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from torchvision import datasets\n",
    "import torchvision.transforms as transforms\n",
    "\n",
    "# number of subprocesses to use for data loading\n",
    "num_workers = 0\n",
    "# how many samples per batch to load\n",
    "batch_size = 64\n",
    "\n",
    "# convert data to torch.FloatTensor\n",
    "transform = transforms.ToTensor()\n",
    "\n",
    "# get the training datasets\n",
    "train_data = datasets.MNIST(root='data', train=True,\n",
    "                                   download=True, transform=transform)\n",
    "\n",
    "# prepare data loader\n",
    "train_loader = torch.utils.data.DataLoader(train_data, batch_size=batch_size,\n",
    "                                           num_workers=num_workers)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Visualize the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f1aa54f56a0>"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1aa5530f98>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# obtain one batch of training images\n",
    "dataiter = iter(train_loader)\n",
    "images, labels = dataiter.next()\n",
    "images = images.numpy()\n",
    "\n",
    "# get one image from the batch\n",
    "img = np.squeeze(images[1])\n",
    "\n",
    "fig = plt.figure(figsize = (3,3)) \n",
    "ax = fig.add_subplot(111)\n",
    "ax.imshow(img, cmap='gray')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "# Define the Model\n",
    "\n",
    "A GAN is comprised of two adversarial networks, a discriminator and a generator."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Discriminator\n",
    "\n",
    "The discriminator network is going to be a pretty typical linear classifier. To make this network a universal function approximator, we'll need at least one hidden layer, and these hidden layers should have one key attribute:\n",
    "> All hidden layers will have a [Leaky ReLu](https://pytorch.org/docs/stable/nn.html#torch.nn.LeakyReLU) activation function applied to their outputs.\n",
    "\n",
    "<img src='assets/gan_network.png' width=70% />\n",
    "\n",
    "#### Leaky ReLu\n",
    "\n",
    "We should use a leaky ReLU to allow gradients to flow backwards through the layer unimpeded. A leaky ReLU is like a normal ReLU, except that there is a small non-zero output for negative input values.\n",
    "\n",
    "<img src='assets/leaky_relu.png' width=40% />\n",
    "\n",
    "#### Sigmoid Output\n",
    "\n",
    "We'll also take the approach of using a more numerically stable loss function on the outputs. Recall that we want the discriminator to output a value 0-1 indicating whether an image is _real or fake_. \n",
    "> We will ultimately use [BCEWithLogitsLoss](https://pytorch.org/docs/stable/nn.html#bcewithlogitsloss), which combines a `sigmoid` activation function **and** and binary cross entropy loss in one function. \n",
    "\n",
    "So, our final output layer should not have any activation function applied to it."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "\n",
    "class Discriminator(nn.Module):\n",
    "\n",
    "    def __init__(self, input_size, hidden_dim, output_size):\n",
    "        super(Discriminator, self).__init__()\n",
    "        \n",
    "        # define hidden linear layers\n",
    "        self.fc1 = nn.Linear(input_size, hidden_dim*4)\n",
    "        self.fc2 = nn.Linear(hidden_dim*4, hidden_dim*2)\n",
    "        self.fc3 = nn.Linear(hidden_dim*2, hidden_dim)\n",
    "        \n",
    "        # final fully-connected layer\n",
    "        self.fc4 = nn.Linear(hidden_dim, output_size)\n",
    "        \n",
    "        # dropout layer \n",
    "        self.dropout = nn.Dropout(0.3)\n",
    "        \n",
    "        \n",
    "    def forward(self, x):\n",
    "        # flatten image\n",
    "        x = x.view(-1, 28*28)\n",
    "        # all hidden layers\n",
    "        x = F.leaky_relu(self.fc1(x), 0.2) # (input, negative_slope=0.2)\n",
    "        x = self.dropout(x)\n",
    "        x = F.leaky_relu(self.fc2(x), 0.2)\n",
    "        x = self.dropout(x)\n",
    "        x = F.leaky_relu(self.fc3(x), 0.2)\n",
    "        x = self.dropout(x)\n",
    "        # final layer\n",
    "        out = self.fc4(x)\n",
    "\n",
    "        return out\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Generator\n",
    "\n",
    "The generator network will be almost exactly the same as the discriminator network, except that we're applying a [tanh activation function](https://pytorch.org/docs/stable/nn.html#tanh) to our output layer.\n",
    "\n",
    "#### tanh Output\n",
    "The generator has been found to perform the best with $tanh$ for the generator output, which scales the output to be between -1 and 1, instead of 0 and 1. \n",
    "\n",
    "<img src='assets/tanh_fn.png' width=40% />\n",
    "\n",
    "Recall that we also want these outputs to be comparable to the *real* input pixel values, which are read in as normalized values between 0 and 1. \n",
    "> So, we'll also have to **scale our real input images to have pixel values between -1 and 1** when we train the discriminator. \n",
    "\n",
    "I'll do this in the training loop, later on."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Generator(nn.Module):\n",
    "\n",
    "    def __init__(self, input_size, hidden_dim, output_size):\n",
    "        super(Generator, self).__init__()\n",
    "        \n",
    "        # define hidden linear layers\n",
    "        self.fc1 = nn.Linear(input_size, hidden_dim)\n",
    "        self.fc2 = nn.Linear(hidden_dim, hidden_dim*2)\n",
    "        self.fc3 = nn.Linear(hidden_dim*2, hidden_dim*4)\n",
    "        \n",
    "        # final fully-connected layer\n",
    "        self.fc4 = nn.Linear(hidden_dim*4, output_size)\n",
    "        \n",
    "        # dropout layer \n",
    "        self.dropout = nn.Dropout(0.3)\n",
    "\n",
    "    def forward(self, x):\n",
    "        # all hidden layers\n",
    "        x = F.leaky_relu(self.fc1(x), 0.2) # (input, negative_slope=0.2)\n",
    "        x = self.dropout(x)\n",
    "        x = F.leaky_relu(self.fc2(x), 0.2)\n",
    "        x = self.dropout(x)\n",
    "        x = F.leaky_relu(self.fc3(x), 0.2)\n",
    "        x = self.dropout(x)\n",
    "        # final layer with tanh applied\n",
    "        out = F.tanh(self.fc4(x))\n",
    "\n",
    "        return out"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model hyperparameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Discriminator hyperparams\n",
    "\n",
    "# Size of input image to discriminator (28*28)\n",
    "input_size = 784\n",
    "# Size of discriminator output (real or fake)\n",
    "d_output_size = 1\n",
    "# Size of last hidden layer in the discriminator\n",
    "d_hidden_size = 32\n",
    "\n",
    "# Generator hyperparams\n",
    "\n",
    "# Size of latent vector to give to generator\n",
    "z_size = 100\n",
    "# Size of discriminator output (generated image)\n",
    "g_output_size = 784\n",
    "# Size of first hidden layer in the generator\n",
    "g_hidden_size = 32"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Build complete network\n",
    "\n",
    "Now we're instantiating the discriminator and generator from the classes defined above. Make sure you've passed in the correct input arguments."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Discriminator(\n",
      "  (fc1): Linear(in_features=784, out_features=128, bias=True)\n",
      "  (fc2): Linear(in_features=128, out_features=64, bias=True)\n",
      "  (fc3): Linear(in_features=64, out_features=32, bias=True)\n",
      "  (fc4): Linear(in_features=32, out_features=1, bias=True)\n",
      "  (dropout): Dropout(p=0.3)\n",
      ")\n",
      "\n",
      "Generator(\n",
      "  (fc1): Linear(in_features=100, out_features=32, bias=True)\n",
      "  (fc2): Linear(in_features=32, out_features=64, bias=True)\n",
      "  (fc3): Linear(in_features=64, out_features=128, bias=True)\n",
      "  (fc4): Linear(in_features=128, out_features=784, bias=True)\n",
      "  (dropout): Dropout(p=0.3)\n",
      ")\n"
     ]
    }
   ],
   "source": [
    "# instantiate discriminator and generator\n",
    "D = Discriminator(input_size, d_hidden_size, d_output_size)\n",
    "G = Generator(z_size, g_hidden_size, g_output_size)\n",
    "\n",
    "# check that they are as you expect\n",
    "print(D)\n",
    "print()\n",
    "print(G)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "## Discriminator and Generator Losses\n",
    "\n",
    "Now we need to calculate the losses. \n",
    "\n",
    "### Discriminator Losses\n",
    "\n",
    "> * For the discriminator, the total loss is the sum of the losses for real and fake images, `d_loss = d_real_loss + d_fake_loss`. \n",
    "* Remember that we want the discriminator to output 1 for real images and 0 for fake images, so we need to set up the losses to reflect that.\n",
    "\n",
    "<img src='assets/gan_pipeline.png' width=70% />\n",
    "\n",
    "The losses will by binary cross entropy loss with logits, which we can get with [BCEWithLogitsLoss](https://pytorch.org/docs/stable/nn.html#bcewithlogitsloss). This combines a `sigmoid` activation function **and** and binary cross entropy loss in one function.\n",
    "\n",
    "For the real images, we want `D(real_images) = 1`. That is, we want the discriminator to classify the the real images with a label = 1, indicating that these are real. To help the discriminator generalize better, the labels are **reduced a bit from 1.0 to 0.9**. For this, we'll use the parameter `smooth`; if True, then we should smooth our labels. In PyTorch, this looks like `labels = torch.ones(size) * 0.9`\n",
    "\n",
    "The discriminator loss for the fake data is similar. We want `D(fake_images) = 0`, where the fake images are the _generator output_, `fake_images = G(z)`. \n",
    "\n",
    "### Generator Loss\n",
    "\n",
    "The generator loss will look similar only with flipped labels. The generator's goal is to get `D(fake_images) = 1`. In this case, the labels are **flipped** to represent that the generator is trying to fool the discriminator into thinking that the images it generates (fakes) are real!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Calculate losses\n",
    "def real_loss(D_out, smooth=False):\n",
    "    batch_size = D_out.size(0)\n",
    "    # label smoothing\n",
    "    if smooth:\n",
    "        # smooth, real labels = 0.9\n",
    "        labels = torch.ones(batch_size)*0.9\n",
    "    else:\n",
    "        labels = torch.ones(batch_size) # real labels = 1\n",
    "        \n",
    "    # numerically stable loss\n",
    "    criterion = nn.BCEWithLogitsLoss()\n",
    "    # calculate loss\n",
    "    loss = criterion(D_out.squeeze(), labels)\n",
    "    return loss\n",
    "\n",
    "def fake_loss(D_out):\n",
    "    batch_size = D_out.size(0)\n",
    "    labels = torch.zeros(batch_size) # fake labels = 0\n",
    "    criterion = nn.BCEWithLogitsLoss()\n",
    "    # calculate loss\n",
    "    loss = criterion(D_out.squeeze(), labels)\n",
    "    return loss"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Optimizers\n",
    "\n",
    "We want to update the generator and discriminator variables separately. So, we'll define two separate Adam optimizers."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "import torch.optim as optim\n",
    "\n",
    "# Optimizers\n",
    "lr = 0.002\n",
    "\n",
    "# Create optimizers for the discriminator and generator\n",
    "d_optimizer = optim.Adam(D.parameters(), lr)\n",
    "g_optimizer = optim.Adam(G.parameters(), lr)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "## Training\n",
    "\n",
    "Training will involve alternating between training the discriminator and the generator. We'll use our functions `real_loss` and `fake_loss` to help us calculate the discriminator losses in all of the following cases.\n",
    "\n",
    "### Discriminator training\n",
    "1. Compute the discriminator loss on real, training images        \n",
    "2. Generate fake images\n",
    "3. Compute the discriminator loss on fake, generated images     \n",
    "4. Add up real and fake loss\n",
    "5. Perform backpropagation + an optimization step to update the discriminator's weights\n",
    "\n",
    "### Generator training\n",
    "1. Generate fake images\n",
    "2. Compute the discriminator loss on fake images, using **flipped** labels!\n",
    "3. Perform backpropagation + an optimization step to update the generator's weights\n",
    "\n",
    "#### Saving Samples\n",
    "\n",
    "As we train, we'll also print out some loss statistics and save some generated \"fake\" samples."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [    1/  100] | d_loss: 1.3859 | g_loss: 0.7354\n",
      "Epoch [    1/  100] | d_loss: 0.9765 | g_loss: 2.7919\n",
      "Epoch [    1/  100] | d_loss: 1.3439 | g_loss: 0.6600\n",
      "Epoch [    2/  100] | d_loss: 1.3599 | g_loss: 0.7390\n",
      "Epoch [    2/  100] | d_loss: 1.3425 | g_loss: 0.8318\n",
      "Epoch [    2/  100] | d_loss: 1.2664 | g_loss: 1.2607\n",
      "Epoch [    3/  100] | d_loss: 1.3472 | g_loss: 0.8560\n",
      "Epoch [    3/  100] | d_loss: 1.3247 | g_loss: 0.7761\n",
      "Epoch [    3/  100] | d_loss: 1.1462 | g_loss: 1.2789\n",
      "Epoch [    4/  100] | d_loss: 1.1183 | g_loss: 1.8702\n",
      "Epoch [    4/  100] | d_loss: 1.3107 | g_loss: 0.8302\n",
      "Epoch [    4/  100] | d_loss: 1.3450 | g_loss: 0.8098\n",
      "Epoch [    5/  100] | d_loss: 1.4237 | g_loss: 1.0706\n",
      "Epoch [    5/  100] | d_loss: 1.3652 | g_loss: 0.6467\n",
      "Epoch [    5/  100] | d_loss: 1.3006 | g_loss: 0.8369\n",
      "Epoch [    6/  100] | d_loss: 1.2486 | g_loss: 1.8772\n",
      "Epoch [    6/  100] | d_loss: 1.1466 | g_loss: 2.1491\n",
      "Epoch [    6/  100] | d_loss: 1.3540 | g_loss: 1.0715\n",
      "Epoch [    7/  100] | d_loss: 1.2585 | g_loss: 2.9083\n",
      "Epoch [    7/  100] | d_loss: 1.0463 | g_loss: 1.3215\n",
      "Epoch [    7/  100] | d_loss: 1.4342 | g_loss: 1.1247\n",
      "Epoch [    8/  100] | d_loss: 1.3591 | g_loss: 0.7885\n",
      "Epoch [    8/  100] | d_loss: 1.2208 | g_loss: 0.8983\n",
      "Epoch [    8/  100] | d_loss: 1.1589 | g_loss: 1.5397\n",
      "Epoch [    9/  100] | d_loss: 1.2453 | g_loss: 1.1469\n",
      "Epoch [    9/  100] | d_loss: 1.6987 | g_loss: 1.3587\n",
      "Epoch [    9/  100] | d_loss: 1.0271 | g_loss: 1.9565\n",
      "Epoch [   10/  100] | d_loss: 1.2120 | g_loss: 1.4182\n",
      "Epoch [   10/  100] | d_loss: 1.1768 | g_loss: 1.1067\n",
      "Epoch [   10/  100] | d_loss: 1.2838 | g_loss: 1.0492\n",
      "Epoch [   11/  100] | d_loss: 1.1283 | g_loss: 1.8456\n",
      "Epoch [   11/  100] | d_loss: 1.3976 | g_loss: 1.2405\n",
      "Epoch [   11/  100] | d_loss: 1.3824 | g_loss: 1.2381\n",
      "Epoch [   12/  100] | d_loss: 1.2635 | g_loss: 1.4096\n",
      "Epoch [   12/  100] | d_loss: 1.2031 | g_loss: 1.8811\n",
      "Epoch [   12/  100] | d_loss: 1.2095 | g_loss: 1.0614\n",
      "Epoch [   13/  100] | d_loss: 1.2652 | g_loss: 1.1345\n",
      "Epoch [   13/  100] | d_loss: 1.1940 | g_loss: 1.2309\n",
      "Epoch [   13/  100] | d_loss: 1.3746 | g_loss: 1.0958\n",
      "Epoch [   14/  100] | d_loss: 1.2309 | g_loss: 1.4292\n",
      "Epoch [   14/  100] | d_loss: 1.2923 | g_loss: 1.0798\n",
      "Epoch [   14/  100] | d_loss: 1.2820 | g_loss: 1.4911\n",
      "Epoch [   15/  100] | d_loss: 1.2654 | g_loss: 1.0030\n",
      "Epoch [   15/  100] | d_loss: 1.2524 | g_loss: 0.9916\n",
      "Epoch [   15/  100] | d_loss: 1.2806 | g_loss: 0.9805\n",
      "Epoch [   16/  100] | d_loss: 1.2866 | g_loss: 1.0003\n",
      "Epoch [   16/  100] | d_loss: 1.1891 | g_loss: 1.4939\n",
      "Epoch [   16/  100] | d_loss: 1.3251 | g_loss: 1.2005\n",
      "Epoch [   17/  100] | d_loss: 1.2746 | g_loss: 0.8011\n",
      "Epoch [   17/  100] | d_loss: 1.3069 | g_loss: 1.0386\n",
      "Epoch [   17/  100] | d_loss: 1.1499 | g_loss: 1.3550\n",
      "Epoch [   18/  100] | d_loss: 1.2249 | g_loss: 1.2671\n",
      "Epoch [   18/  100] | d_loss: 1.2161 | g_loss: 1.0190\n",
      "Epoch [   18/  100] | d_loss: 1.2829 | g_loss: 1.0240\n",
      "Epoch [   19/  100] | d_loss: 1.3282 | g_loss: 1.1115\n",
      "Epoch [   19/  100] | d_loss: 1.1640 | g_loss: 1.1369\n",
      "Epoch [   19/  100] | d_loss: 1.1953 | g_loss: 1.2755\n",
      "Epoch [   20/  100] | d_loss: 1.2690 | g_loss: 0.9891\n",
      "Epoch [   20/  100] | d_loss: 1.2356 | g_loss: 1.1400\n",
      "Epoch [   20/  100] | d_loss: 1.1966 | g_loss: 1.6823\n",
      "Epoch [   21/  100] | d_loss: 1.2001 | g_loss: 1.1886\n",
      "Epoch [   21/  100] | d_loss: 1.2229 | g_loss: 1.8120\n",
      "Epoch [   21/  100] | d_loss: 1.1959 | g_loss: 0.9895\n",
      "Epoch [   22/  100] | d_loss: 1.2259 | g_loss: 1.3617\n",
      "Epoch [   22/  100] | d_loss: 1.1995 | g_loss: 1.0428\n",
      "Epoch [   22/  100] | d_loss: 1.2408 | g_loss: 1.0687\n",
      "Epoch [   23/  100] | d_loss: 1.2215 | g_loss: 1.3730\n",
      "Epoch [   23/  100] | d_loss: 1.2255 | g_loss: 1.0725\n",
      "Epoch [   23/  100] | d_loss: 1.2840 | g_loss: 0.8882\n",
      "Epoch [   24/  100] | d_loss: 1.1517 | g_loss: 1.5940\n",
      "Epoch [   24/  100] | d_loss: 1.2402 | g_loss: 1.0264\n",
      "Epoch [   24/  100] | d_loss: 1.2993 | g_loss: 1.2158\n",
      "Epoch [   25/  100] | d_loss: 1.2651 | g_loss: 0.9370\n",
      "Epoch [   25/  100] | d_loss: 1.1463 | g_loss: 1.2296\n",
      "Epoch [   25/  100] | d_loss: 1.3957 | g_loss: 1.0510\n",
      "Epoch [   26/  100] | d_loss: 1.2680 | g_loss: 1.3627\n",
      "Epoch [   26/  100] | d_loss: 1.2547 | g_loss: 0.9694\n",
      "Epoch [   26/  100] | d_loss: 1.1992 | g_loss: 1.1708\n",
      "Epoch [   27/  100] | d_loss: 1.1209 | g_loss: 1.4046\n",
      "Epoch [   27/  100] | d_loss: 1.3108 | g_loss: 1.4566\n",
      "Epoch [   27/  100] | d_loss: 1.3223 | g_loss: 0.9555\n",
      "Epoch [   28/  100] | d_loss: 1.3455 | g_loss: 0.9927\n",
      "Epoch [   28/  100] | d_loss: 1.1631 | g_loss: 1.2027\n",
      "Epoch [   28/  100] | d_loss: 1.3669 | g_loss: 1.0367\n",
      "Epoch [   29/  100] | d_loss: 1.2790 | g_loss: 1.1529\n",
      "Epoch [   29/  100] | d_loss: 1.2107 | g_loss: 1.0509\n",
      "Epoch [   29/  100] | d_loss: 1.2707 | g_loss: 1.3266\n",
      "Epoch [   30/  100] | d_loss: 1.3856 | g_loss: 1.2509\n",
      "Epoch [   30/  100] | d_loss: 1.3633 | g_loss: 1.2914\n",
      "Epoch [   30/  100] | d_loss: 1.3099 | g_loss: 1.0131\n",
      "Epoch [   31/  100] | d_loss: 1.2295 | g_loss: 1.0804\n",
      "Epoch [   31/  100] | d_loss: 1.1189 | g_loss: 1.2798\n",
      "Epoch [   31/  100] | d_loss: 1.2833 | g_loss: 1.1630\n",
      "Epoch [   32/  100] | d_loss: 1.2875 | g_loss: 1.3806\n",
      "Epoch [   32/  100] | d_loss: 1.2405 | g_loss: 1.2617\n",
      "Epoch [   32/  100] | d_loss: 1.1488 | g_loss: 1.3961\n",
      "Epoch [   33/  100] | d_loss: 1.2606 | g_loss: 1.0259\n",
      "Epoch [   33/  100] | d_loss: 1.2576 | g_loss: 1.2619\n",
      "Epoch [   33/  100] | d_loss: 1.2904 | g_loss: 1.1252\n",
      "Epoch [   34/  100] | d_loss: 1.2080 | g_loss: 1.3674\n",
      "Epoch [   34/  100] | d_loss: 1.2742 | g_loss: 1.5788\n",
      "Epoch [   34/  100] | d_loss: 1.3250 | g_loss: 0.9812\n",
      "Epoch [   35/  100] | d_loss: 1.2680 | g_loss: 0.9836\n",
      "Epoch [   35/  100] | d_loss: 1.3289 | g_loss: 0.9137\n",
      "Epoch [   35/  100] | d_loss: 1.2752 | g_loss: 0.9772\n",
      "Epoch [   36/  100] | d_loss: 1.2946 | g_loss: 1.1252\n",
      "Epoch [   36/  100] | d_loss: 1.2197 | g_loss: 1.0602\n",
      "Epoch [   36/  100] | d_loss: 1.3579 | g_loss: 1.0587\n",
      "Epoch [   37/  100] | d_loss: 1.2865 | g_loss: 0.8642\n",
      "Epoch [   37/  100] | d_loss: 1.1045 | g_loss: 1.5932\n",
      "Epoch [   37/  100] | d_loss: 1.2242 | g_loss: 1.5443\n",
      "Epoch [   38/  100] | d_loss: 1.3130 | g_loss: 1.0321\n",
      "Epoch [   38/  100] | d_loss: 1.2273 | g_loss: 1.0303\n",
      "Epoch [   38/  100] | d_loss: 1.2576 | g_loss: 1.0434\n",
      "Epoch [   39/  100] | d_loss: 1.2893 | g_loss: 0.9961\n",
      "Epoch [   39/  100] | d_loss: 1.2558 | g_loss: 0.9944\n",
      "Epoch [   39/  100] | d_loss: 1.4051 | g_loss: 1.0955\n",
      "Epoch [   40/  100] | d_loss: 1.2387 | g_loss: 1.1636\n",
      "Epoch [   40/  100] | d_loss: 1.2996 | g_loss: 1.0372\n",
      "Epoch [   40/  100] | d_loss: 1.4264 | g_loss: 1.0341\n",
      "Epoch [   41/  100] | d_loss: 1.2804 | g_loss: 1.0995\n",
      "Epoch [   41/  100] | d_loss: 1.1733 | g_loss: 0.9644\n",
      "Epoch [   41/  100] | d_loss: 1.3015 | g_loss: 1.1262\n",
      "Epoch [   42/  100] | d_loss: 1.3498 | g_loss: 1.8798\n",
      "Epoch [   42/  100] | d_loss: 1.2164 | g_loss: 1.0759\n",
      "Epoch [   42/  100] | d_loss: 1.4100 | g_loss: 0.9173\n",
      "Epoch [   43/  100] | d_loss: 1.3155 | g_loss: 0.9412\n",
      "Epoch [   43/  100] | d_loss: 1.2233 | g_loss: 0.9793\n",
      "Epoch [   43/  100] | d_loss: 1.4250 | g_loss: 1.2246\n",
      "Epoch [   44/  100] | d_loss: 1.2193 | g_loss: 1.8871\n",
      "Epoch [   44/  100] | d_loss: 1.2374 | g_loss: 1.2346\n",
      "Epoch [   44/  100] | d_loss: 1.2401 | g_loss: 1.3843\n",
      "Epoch [   45/  100] | d_loss: 1.3040 | g_loss: 1.0657\n",
      "Epoch [   45/  100] | d_loss: 1.2327 | g_loss: 1.2108\n",
      "Epoch [   45/  100] | d_loss: 1.3048 | g_loss: 1.0463\n",
      "Epoch [   46/  100] | d_loss: 1.3269 | g_loss: 0.9669\n",
      "Epoch [   46/  100] | d_loss: 1.2342 | g_loss: 1.2662\n",
      "Epoch [   46/  100] | d_loss: 1.3249 | g_loss: 1.0152\n",
      "Epoch [   47/  100] | d_loss: 1.2524 | g_loss: 1.2271\n",
      "Epoch [   47/  100] | d_loss: 1.2373 | g_loss: 1.2767\n",
      "Epoch [   47/  100] | d_loss: 1.2353 | g_loss: 0.9954\n",
      "Epoch [   48/  100] | d_loss: 1.3247 | g_loss: 1.2667\n",
      "Epoch [   48/  100] | d_loss: 1.2162 | g_loss: 1.2030\n",
      "Epoch [   48/  100] | d_loss: 1.3547 | g_loss: 1.1066\n",
      "Epoch [   49/  100] | d_loss: 1.2423 | g_loss: 1.0166\n",
      "Epoch [   49/  100] | d_loss: 1.3429 | g_loss: 0.9527\n",
      "Epoch [   49/  100] | d_loss: 1.4049 | g_loss: 1.0012\n",
      "Epoch [   50/  100] | d_loss: 1.2676 | g_loss: 0.9323\n",
      "Epoch [   50/  100] | d_loss: 1.2070 | g_loss: 1.3753\n",
      "Epoch [   50/  100] | d_loss: 1.2962 | g_loss: 1.0743\n",
      "Epoch [   51/  100] | d_loss: 1.3460 | g_loss: 1.2631\n",
      "Epoch [   51/  100] | d_loss: 1.2070 | g_loss: 0.7689\n",
      "Epoch [   51/  100] | d_loss: 1.2901 | g_loss: 1.0768\n",
      "Epoch [   52/  100] | d_loss: 1.3419 | g_loss: 1.0664\n",
      "Epoch [   52/  100] | d_loss: 1.2495 | g_loss: 1.0015\n",
      "Epoch [   52/  100] | d_loss: 1.4008 | g_loss: 1.2223\n",
      "Epoch [   53/  100] | d_loss: 1.3140 | g_loss: 1.1064\n",
      "Epoch [   53/  100] | d_loss: 1.3113 | g_loss: 0.8692\n",
      "Epoch [   53/  100] | d_loss: 1.2323 | g_loss: 1.0176\n",
      "Epoch [   54/  100] | d_loss: 1.3298 | g_loss: 1.0296\n",
      "Epoch [   54/  100] | d_loss: 1.2279 | g_loss: 1.1090\n",
      "Epoch [   54/  100] | d_loss: 1.2943 | g_loss: 0.9608\n",
      "Epoch [   55/  100] | d_loss: 1.2917 | g_loss: 0.8362\n",
      "Epoch [   55/  100] | d_loss: 1.2560 | g_loss: 1.0409\n",
      "Epoch [   55/  100] | d_loss: 1.4085 | g_loss: 1.2019\n",
      "Epoch [   56/  100] | d_loss: 1.3788 | g_loss: 0.8938\n",
      "Epoch [   56/  100] | d_loss: 1.2353 | g_loss: 0.9665\n",
      "Epoch [   56/  100] | d_loss: 1.3364 | g_loss: 1.1218\n",
      "Epoch [   57/  100] | d_loss: 1.2748 | g_loss: 0.9977\n",
      "Epoch [   57/  100] | d_loss: 1.2508 | g_loss: 1.0122\n",
      "Epoch [   57/  100] | d_loss: 1.3322 | g_loss: 1.2624\n",
      "Epoch [   58/  100] | d_loss: 1.1859 | g_loss: 1.5127\n",
      "Epoch [   58/  100] | d_loss: 1.2664 | g_loss: 1.0017\n",
      "Epoch [   58/  100] | d_loss: 1.3991 | g_loss: 0.9734\n",
      "Epoch [   59/  100] | d_loss: 1.2179 | g_loss: 1.1003\n",
      "Epoch [   59/  100] | d_loss: 1.2388 | g_loss: 1.1843\n",
      "Epoch [   59/  100] | d_loss: 1.3712 | g_loss: 1.0272\n",
      "Epoch [   60/  100] | d_loss: 1.1790 | g_loss: 1.2914\n",
      "Epoch [   60/  100] | d_loss: 1.2598 | g_loss: 1.0201\n",
      "Epoch [   60/  100] | d_loss: 1.3238 | g_loss: 0.9984\n",
      "Epoch [   61/  100] | d_loss: 1.3589 | g_loss: 1.0747\n",
      "Epoch [   61/  100] | d_loss: 1.2528 | g_loss: 1.0432\n",
      "Epoch [   61/  100] | d_loss: 1.4136 | g_loss: 0.9747\n",
      "Epoch [   62/  100] | d_loss: 1.3298 | g_loss: 0.9857\n",
      "Epoch [   62/  100] | d_loss: 1.2626 | g_loss: 0.9264\n",
      "Epoch [   62/  100] | d_loss: 1.3097 | g_loss: 1.3326\n",
      "Epoch [   63/  100] | d_loss: 1.2244 | g_loss: 0.9777\n",
      "Epoch [   63/  100] | d_loss: 1.1831 | g_loss: 1.0214\n",
      "Epoch [   63/  100] | d_loss: 1.3703 | g_loss: 1.1684\n",
      "Epoch [   64/  100] | d_loss: 1.2184 | g_loss: 1.1930\n",
      "Epoch [   64/  100] | d_loss: 1.2637 | g_loss: 1.1777\n",
      "Epoch [   64/  100] | d_loss: 1.3456 | g_loss: 1.2615\n",
      "Epoch [   65/  100] | d_loss: 1.2008 | g_loss: 1.2779\n",
      "Epoch [   65/  100] | d_loss: 1.1907 | g_loss: 1.0914\n",
      "Epoch [   65/  100] | d_loss: 1.3528 | g_loss: 1.1487\n",
      "Epoch [   66/  100] | d_loss: 1.3089 | g_loss: 0.9389\n",
      "Epoch [   66/  100] | d_loss: 1.2189 | g_loss: 1.1208\n",
      "Epoch [   66/  100] | d_loss: 1.2925 | g_loss: 0.9490\n",
      "Epoch [   67/  100] | d_loss: 1.2723 | g_loss: 0.9209\n",
      "Epoch [   67/  100] | d_loss: 1.2396 | g_loss: 0.9737\n",
      "Epoch [   67/  100] | d_loss: 1.4041 | g_loss: 1.0134\n",
      "Epoch [   68/  100] | d_loss: 1.2791 | g_loss: 1.0578\n",
      "Epoch [   68/  100] | d_loss: 1.2359 | g_loss: 0.9728\n",
      "Epoch [   68/  100] | d_loss: 1.3327 | g_loss: 1.2511\n",
      "Epoch [   69/  100] | d_loss: 1.2302 | g_loss: 1.5004\n",
      "Epoch [   69/  100] | d_loss: 1.2198 | g_loss: 1.0165\n",
      "Epoch [   69/  100] | d_loss: 1.2520 | g_loss: 1.4796\n",
      "Epoch [   70/  100] | d_loss: 1.2955 | g_loss: 1.0236\n",
      "Epoch [   70/  100] | d_loss: 1.1980 | g_loss: 1.2108\n",
      "Epoch [   70/  100] | d_loss: 1.2542 | g_loss: 0.9408\n",
      "Epoch [   71/  100] | d_loss: 1.1702 | g_loss: 1.3800\n",
      "Epoch [   71/  100] | d_loss: 1.2659 | g_loss: 1.0145\n",
      "Epoch [   71/  100] | d_loss: 1.3043 | g_loss: 1.1091\n",
      "Epoch [   72/  100] | d_loss: 1.3827 | g_loss: 0.9423\n",
      "Epoch [   72/  100] | d_loss: 1.2237 | g_loss: 1.0490\n",
      "Epoch [   72/  100] | d_loss: 1.3535 | g_loss: 1.0753\n",
      "Epoch [   73/  100] | d_loss: 1.3298 | g_loss: 1.0465\n",
      "Epoch [   73/  100] | d_loss: 1.2903 | g_loss: 1.1006\n",
      "Epoch [   73/  100] | d_loss: 1.1679 | g_loss: 1.4344\n",
      "Epoch [   74/  100] | d_loss: 1.2487 | g_loss: 0.9677\n",
      "Epoch [   74/  100] | d_loss: 1.1550 | g_loss: 1.2135\n",
      "Epoch [   74/  100] | d_loss: 1.4275 | g_loss: 1.4524\n",
      "Epoch [   75/  100] | d_loss: 1.2813 | g_loss: 0.9402\n",
      "Epoch [   75/  100] | d_loss: 1.2626 | g_loss: 1.1097\n",
      "Epoch [   75/  100] | d_loss: 1.2438 | g_loss: 0.9771\n",
      "Epoch [   76/  100] | d_loss: 1.3269 | g_loss: 0.9823\n",
      "Epoch [   76/  100] | d_loss: 1.2698 | g_loss: 1.1988\n",
      "Epoch [   76/  100] | d_loss: 1.4380 | g_loss: 1.5318\n",
      "Epoch [   77/  100] | d_loss: 1.2231 | g_loss: 1.2195\n",
      "Epoch [   77/  100] | d_loss: 1.1714 | g_loss: 1.1981\n",
      "Epoch [   77/  100] | d_loss: 1.2050 | g_loss: 1.2612\n",
      "Epoch [   78/  100] | d_loss: 1.1749 | g_loss: 1.3493\n",
      "Epoch [   78/  100] | d_loss: 1.1143 | g_loss: 1.0926\n",
      "Epoch [   78/  100] | d_loss: 1.3606 | g_loss: 0.9840\n",
      "Epoch [   79/  100] | d_loss: 1.2024 | g_loss: 1.2552\n",
      "Epoch [   79/  100] | d_loss: 1.2010 | g_loss: 0.8839\n",
      "Epoch [   79/  100] | d_loss: 1.3169 | g_loss: 0.9148\n",
      "Epoch [   80/  100] | d_loss: 1.2773 | g_loss: 1.0469\n",
      "Epoch [   80/  100] | d_loss: 1.2851 | g_loss: 1.1867\n",
      "Epoch [   80/  100] | d_loss: 1.3819 | g_loss: 1.0378\n",
      "Epoch [   81/  100] | d_loss: 1.3146 | g_loss: 0.8961\n",
      "Epoch [   81/  100] | d_loss: 1.1714 | g_loss: 0.9767\n",
      "Epoch [   81/  100] | d_loss: 1.1360 | g_loss: 0.9984\n",
      "Epoch [   82/  100] | d_loss: 1.2227 | g_loss: 1.6436\n",
      "Epoch [   82/  100] | d_loss: 1.2474 | g_loss: 1.0697\n",
      "Epoch [   82/  100] | d_loss: 1.3744 | g_loss: 1.0271\n",
      "Epoch [   83/  100] | d_loss: 1.2512 | g_loss: 1.3248\n",
      "Epoch [   83/  100] | d_loss: 1.2343 | g_loss: 0.9208\n",
      "Epoch [   83/  100] | d_loss: 1.3537 | g_loss: 0.9883\n",
      "Epoch [   84/  100] | d_loss: 1.1307 | g_loss: 1.3830\n",
      "Epoch [   84/  100] | d_loss: 1.2611 | g_loss: 1.4125\n",
      "Epoch [   84/  100] | d_loss: 1.3066 | g_loss: 1.1363\n",
      "Epoch [   85/  100] | d_loss: 1.2270 | g_loss: 1.6612\n",
      "Epoch [   85/  100] | d_loss: 1.2369 | g_loss: 1.3695\n",
      "Epoch [   85/  100] | d_loss: 1.2754 | g_loss: 1.0202\n",
      "Epoch [   86/  100] | d_loss: 1.2545 | g_loss: 1.0690\n",
      "Epoch [   86/  100] | d_loss: 1.2786 | g_loss: 0.8799\n",
      "Epoch [   86/  100] | d_loss: 1.2489 | g_loss: 1.1471\n",
      "Epoch [   87/  100] | d_loss: 1.2679 | g_loss: 1.0427\n",
      "Epoch [   87/  100] | d_loss: 1.2567 | g_loss: 1.1548\n",
      "Epoch [   87/  100] | d_loss: 1.4202 | g_loss: 1.0865\n",
      "Epoch [   88/  100] | d_loss: 1.3297 | g_loss: 1.1183\n",
      "Epoch [   88/  100] | d_loss: 1.1971 | g_loss: 1.2080\n",
      "Epoch [   88/  100] | d_loss: 1.2942 | g_loss: 0.9439\n",
      "Epoch [   89/  100] | d_loss: 1.2768 | g_loss: 0.9349\n",
      "Epoch [   89/  100] | d_loss: 1.2430 | g_loss: 1.0776\n",
      "Epoch [   89/  100] | d_loss: 1.3476 | g_loss: 0.9630\n",
      "Epoch [   90/  100] | d_loss: 1.1816 | g_loss: 1.3323\n",
      "Epoch [   90/  100] | d_loss: 1.2560 | g_loss: 0.9955\n",
      "Epoch [   90/  100] | d_loss: 1.2772 | g_loss: 1.1294\n",
      "Epoch [   91/  100] | d_loss: 1.3308 | g_loss: 1.1140\n",
      "Epoch [   91/  100] | d_loss: 1.1918 | g_loss: 1.2787\n",
      "Epoch [   91/  100] | d_loss: 1.2469 | g_loss: 1.2314\n",
      "Epoch [   92/  100] | d_loss: 1.2587 | g_loss: 1.3418\n",
      "Epoch [   92/  100] | d_loss: 1.2217 | g_loss: 1.0707\n",
      "Epoch [   92/  100] | d_loss: 1.3858 | g_loss: 1.0324\n",
      "Epoch [   93/  100] | d_loss: 1.2481 | g_loss: 1.2477\n",
      "Epoch [   93/  100] | d_loss: 1.2612 | g_loss: 1.1097\n",
      "Epoch [   93/  100] | d_loss: 1.2986 | g_loss: 1.1736\n",
      "Epoch [   94/  100] | d_loss: 1.3878 | g_loss: 1.0695\n",
      "Epoch [   94/  100] | d_loss: 1.2533 | g_loss: 1.3238\n",
      "Epoch [   94/  100] | d_loss: 1.2736 | g_loss: 1.0195\n",
      "Epoch [   95/  100] | d_loss: 1.2677 | g_loss: 1.2290\n",
      "Epoch [   95/  100] | d_loss: 1.2257 | g_loss: 1.1072\n",
      "Epoch [   95/  100] | d_loss: 1.2010 | g_loss: 1.1713\n",
      "Epoch [   96/  100] | d_loss: 1.1903 | g_loss: 1.1620\n",
      "Epoch [   96/  100] | d_loss: 1.2596 | g_loss: 0.9055\n",
      "Epoch [   96/  100] | d_loss: 1.1941 | g_loss: 1.0331\n",
      "Epoch [   97/  100] | d_loss: 1.2969 | g_loss: 0.8912\n",
      "Epoch [   97/  100] | d_loss: 1.1864 | g_loss: 1.1828\n",
      "Epoch [   97/  100] | d_loss: 1.2355 | g_loss: 1.1226\n",
      "Epoch [   98/  100] | d_loss: 1.2983 | g_loss: 1.0549\n",
      "Epoch [   98/  100] | d_loss: 1.2490 | g_loss: 1.0393\n",
      "Epoch [   98/  100] | d_loss: 1.4249 | g_loss: 1.0447\n",
      "Epoch [   99/  100] | d_loss: 1.2472 | g_loss: 0.8945\n",
      "Epoch [   99/  100] | d_loss: 1.1795 | g_loss: 1.1783\n",
      "Epoch [   99/  100] | d_loss: 1.3617 | g_loss: 1.0648\n",
      "Epoch [  100/  100] | d_loss: 1.2980 | g_loss: 1.0223\n",
      "Epoch [  100/  100] | d_loss: 1.2764 | g_loss: 1.0584\n",
      "Epoch [  100/  100] | d_loss: 1.3193 | g_loss: 0.9065\n"
     ]
    }
   ],
   "source": [
    "import pickle as pkl\n",
    "\n",
    "# training hyperparams\n",
    "num_epochs = 100\n",
    "\n",
    "# keep track of loss and generated, \"fake\" samples\n",
    "samples = []\n",
    "losses = []\n",
    "\n",
    "print_every = 400\n",
    "\n",
    "# Get some fixed data for sampling. These are images that are held\n",
    "# constant throughout training, and allow us to inspect the model's performance\n",
    "sample_size=16\n",
    "fixed_z = np.random.uniform(-1, 1, size=(sample_size, z_size))\n",
    "fixed_z = torch.from_numpy(fixed_z).float()\n",
    "\n",
    "# train the network\n",
    "D.train()\n",
    "G.train()\n",
    "for epoch in range(num_epochs):\n",
    "    \n",
    "    for batch_i, (real_images, _) in enumerate(train_loader):\n",
    "                \n",
    "        batch_size = real_images.size(0)\n",
    "        \n",
    "        ## Important rescaling step ## \n",
    "        real_images = real_images*2 - 1  # rescale input images from [0,1) to [-1, 1)\n",
    "        \n",
    "        # ============================================\n",
    "        #            TRAIN THE DISCRIMINATOR\n",
    "        # ============================================\n",
    "        \n",
    "        d_optimizer.zero_grad()\n",
    "        \n",
    "        # 1. Train with real images\n",
    "\n",
    "        # Compute the discriminator losses on real images \n",
    "        # smooth the real labels\n",
    "        D_real = D(real_images)\n",
    "        d_real_loss = real_loss(D_real, smooth=True)\n",
    "        \n",
    "        # 2. Train with fake images\n",
    "        \n",
    "        # Generate fake images\n",
    "        z = np.random.uniform(-1, 1, size=(batch_size, z_size))\n",
    "        z = torch.from_numpy(z).float()\n",
    "        fake_images = G(z)\n",
    "        \n",
    "        # Compute the discriminator losses on fake images        \n",
    "        D_fake = D(fake_images)\n",
    "        d_fake_loss = fake_loss(D_fake)\n",
    "        \n",
    "        # add up loss and perform backprop\n",
    "        d_loss = d_real_loss + d_fake_loss\n",
    "        d_loss.backward()\n",
    "        d_optimizer.step()\n",
    "        \n",
    "        \n",
    "        # =========================================\n",
    "        #            TRAIN THE GENERATOR\n",
    "        # =========================================\n",
    "        g_optimizer.zero_grad()\n",
    "        \n",
    "        # 1. Train with fake images and flipped labels\n",
    "        \n",
    "        # Generate fake images\n",
    "        z = np.random.uniform(-1, 1, size=(batch_size, z_size))\n",
    "        z = torch.from_numpy(z).float()\n",
    "        fake_images = G(z)\n",
    "        \n",
    "        # Compute the discriminator losses on fake images \n",
    "        # using flipped labels!\n",
    "        D_fake = D(fake_images)\n",
    "        g_loss = real_loss(D_fake) # use real loss to flip labels\n",
    "        \n",
    "        # perform backprop\n",
    "        g_loss.backward()\n",
    "        g_optimizer.step()\n",
    "\n",
    "        # Print some loss stats\n",
    "        if batch_i % print_every == 0:\n",
    "            # print discriminator and generator loss\n",
    "            print('Epoch [{:5d}/{:5d}] | d_loss: {:6.4f} | g_loss: {:6.4f}'.format(\n",
    "                    epoch+1, num_epochs, d_loss.item(), g_loss.item()))\n",
    "\n",
    "    \n",
    "    ## AFTER EACH EPOCH##\n",
    "    # append discriminator loss and generator loss\n",
    "    losses.append((d_loss.item(), g_loss.item()))\n",
    "    \n",
    "    # generate and save sample, fake images\n",
    "    G.eval() # eval mode for generating samples\n",
    "    samples_z = G(fixed_z)\n",
    "    samples.append(samples_z)\n",
    "    G.train() # back to train mode\n",
    "\n",
    "\n",
    "# Save training generator samples\n",
    "with open('train_samples.pkl', 'wb') as f:\n",
    "    pkl.dump(samples, f)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training loss\n",
    "\n",
    "Here we'll plot the training losses for the generator and discriminator, recorded after each epoch."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f1a9c319a90>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1aa5556710>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "losses = np.array(losses)\n",
    "plt.plot(losses.T[0], label='Discriminator')\n",
    "plt.plot(losses.T[1], label='Generator')\n",
    "plt.title(\"Training Losses\")\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Generator samples from training\n",
    "\n",
    "Here we can view samples of images from the generator. First we'll look at the images we saved during training."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# helper function for viewing a list of passed in sample images\n",
    "def view_samples(epoch, samples):\n",
    "    fig, axes = plt.subplots(figsize=(7,7), nrows=4, ncols=4, sharey=True, sharex=True)\n",
    "    for ax, img in zip(axes.flatten(), samples[epoch]):\n",
    "        img = img.detach()\n",
    "        ax.xaxis.set_visible(False)\n",
    "        ax.yaxis.set_visible(False)\n",
    "        im = ax.imshow(img.reshape((28,28)), cmap='Greys_r')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load samples from generator, taken while training\n",
    "with open('train_samples.pkl', 'rb') as f:\n",
    "    samples = pkl.load(f)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "These are samples from the final training epoch. You can see the generator is able to reproduce numbers like 1, 7, 3, 2. Since this is just a sample, it isn't representative of the full range of images this generator can make."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1aa5556400>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# -1 indicates final epoch's samples (the last in the list)\n",
    "view_samples(-1, samples)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Below I'm showing the generated images as the network was training, every 10 epochs."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZwAAAKhCAYAAABkTRjXAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAIABJREFUeJzsnXeUZFXVvp/unoEZgqCAgKIYSKIIKkkFAQmiBMGsoICKoiQV9Ifop5gQRQEVQcwi8ikKCiJJSYKASDQSFAlGwAASh5mp3x98z5yq3X2n+lZ33arB/a7Vq1ele8/d95xz3/fscEZarRaJRCKRSPQbo4NuQCKRSCT+O5APnEQikUg0gnzgJBKJRKIR5AMnkUgkEo0gHziJRCKRaAT5wEkkEolEI8gHTiKRSCQaQT5wEolEItEI8oGTSCQSiUYwo86XR0ZGWiMjI1idYGRkJH4OwPz58wEYHV3488zv+Tv/e/yxsTEA5s6du9DvxdexPaKq3W3tubPVaq2w0EaXY0/KFt3a1HbuhX7ezRbR5pO1xUTfnz9/Pq1Wa+KGTIC6tqi6xvbzt3+vbr8YtC1GR0dr98Wqz6vGUpUt4vcWRVtUoa4t+jlf9GuMVPX9ic6/sM+10bx58yY8XrffT7ZftFot/yZli1oPnNHRUWbPnr3gZA899FDH50sssQQwvmN40bFDzJgxo+PzBx54oONzz+P3Zs6c2fG+5/Fz27PYYot1HPdRj3oUAA8++CAA999/PwCLL754x/nuv//+W+rYYtasWQteV9ki3nDb7LXMmTMHKLbxc9tq26JNbbvH1ybxuPF7sV3awt/NmzdvwbkniypbeM2zZ8+e0Bbx/trmOKHYRm0QJxjvd1W/8Lh+z8/7ZYvZs2cvuEbPLeI5vQZfxza3t6W9jXGMeBzvt+/7O6/d64n9Io4Rx2L7+X2vri1EtzFSNV9E23jtcb7Q5rbZ74toK9vj9+0XSy+9NFBsEfvPnDlzattiZGRkoWPEz+KDIE70XqvX4Oe2J84H3eaL2N/inKgt/Fyb+Lv58+fXGiO5pJZIJBKJRjBSp3jnyMhIa8aMGQuebpEp+PT0/zLLLNPx+X333dfxvseRtfk9jysDso333nsvUNiYT+HIiPyex49P+8gOfHrPmTPnylartf5kbTE2NrbgWLIkryEqk6WWWqrjdbSFbZMtRAYUbRF/rw2ierznnns6rll4vona/dBDDzF//vxaywXttohs2jZH1hSZqvfV4/h+7BeywWiLRz/60eOupd0W9ouooHwdlwseeuihnmzRPkaq+oX/l1122Y7PbaPX4jVXjZF2BQHlfkdbxDEy2X7h7x566CEefPDBWrYYHR1tzZgxY8E5omKpsoXj0fvq+/4+Ko9oC8dY7Fd+P84H2iKuuGiDuBzViy2q5gvb5Dlt+5JLLtnx2vvvPOLv4xjxmqrmC8eex42KKY6Rqv/+/qGHHmLu3LmTtkUqnEQikUg0grpBA7Szt+iojApCJiJD8ftxHdDf+VSXwfpUl6HInF1n9KnuU1ymEtdDo6NOVjDZ4IYqWyy22GKVTtvoR5LJeC22SVvIzvy916BN/vOf/wDFltqgfS21/bwylbimH69Z5tO+dlx3y4oqW1QxRu+bbC3awmuPtpC1RRXgNXr8yNK9xujrif3C80+1X8ycOXPcGIn+hahoVaqRnXvNkblGW8R+4bVGNa8tohqo6hdRGdXF2NjYgmvWFiL6WuJ8YZscO7ZZWzo2vGZtFueROFaq5ouqQI7oo+jFFvYLf+s1eY44jqvmi+jb83j+3n7htdmv/L79wnvRbb5obz8UW7T73erMF6lwEolEItEIaikcePhJ59NRZhBZXFw3juuPYoUVHo5Alr2vscYaAPz73/8GYNVVVwUKY/nrX//a8b1f/OIXQHmq/+1vfwMKO7v77rs7zheZik9zEaNouqHVao2zRfS9xGPLwmQWMkptYZuf8YxnAHDHHXcAsP76D7uWXIP1Wp/61KcCcMUVV3S04/bbbweK7bVxZNayk3ZV6H2rC20RI12i8mmLCuz4vm1bfvnlgcLSnva0pwHwz3/+E4DHP/7xQGFvf/nLXwBYffXVgdIvutlCRKXc3i+magtZeVXfiraIPpkVV1wRgLvuuguAddZZB4A777wTgCc+8YlAGQP2l9VWWw0Ybwv7jeerGiOxX7T39cliZGSE0dHRBddeNV/EfuF995pEHCNrr702AP/4xz8AeMITngCUMeJ8sdZaawFw+eWXA8XGVfNF9E1Hddr+Xh2oXtqvLZ7D1zFaNY6RxzzmMR1tt+//61//AmCjjTYCypzrGHG+uPLKKzvaoS1UOv6Pqw8T9Ys6YyQVTiKRSCQaQW0fzujo6IKnb5WC8Skco458mm6wwQYAPPvZzwbg4osvBgrzfMtb3gIUFifj9bgyGtd6Tz/99AXtg8KIf/Ob3wCMi5n3ODGyow5GRkYYGxsbF5mnLVRltmm55Zbr+L22Wm+99QB4/vOfD8A555wDFEa75557AoWByGhlGrI4bXHaaad1nFfmc+ONNwKFEcV16vY12brszX4RI220e1x7j1FH2kJV99znPheACy64ACiKd4899ui4ZhWw55Pdy3Bjv9AWN9xwQ0f7qmzRC4ut6hdVUUL2C8/l544Nmeq5554LFOb51re+FSjMVdXncf7+978DxdannnoqUPq6rP+3v/1tR/tEVd5HXbSPrRiVGMeIPrnou1PdO2/87Gc/A4oSetOb3gQUBex84XGdL7bccksAfvjDH3Z8vuaaawJw3XXXAd1t0Wu/mDFjxjifWZWvzn6g3X3tfPGc5zwHKLZQ2dgvnC9WXnnljnZoC493xhlndFyTtrj++uuBomziNbePkTr2SIWTSCQSiUZQKw/HuPpYUSAywip2L9t63OMeB8BKK60EwGabbQYUxqKPxu/7NJbNuZ558803A0UNyFD+8Ic/AHDrrbcC5WnvWr7MKea8PPTQQ5POw9EW7dnD7ceKuQEyl8c+9rFAYfUyU9fqX/CCFwCFsTzrWc8Cypqtx/GaXVeWtfu73/3udx22+NOf/tRhA9VA9D3NmTOnVly9tpg5c2alT8wIGs+lbWS0+mJkY/YPVZ8K55nPfGbH7zye1+L5vWZZvv1EG9mP/Dz+vq0/9GyLmAdRlWPmGPH+x36hTTbffPMOW6jiHSPaQgarzb12fxf7xW233QYU1Rj7RVuOWu2cpG62iJUouo0R54stttgCKH1dtWY/cmXFca+/1Gv2dzfddBNQ5g37hb9zhcV+0W6Luv3C/Kw4X0QF6efaRrXmNTk27BcbbrghUFZWXCWwX/i7P//5z0CxdZwvtI3v+31tYb+Ktpg7d27m4SQSiURi+FA7Sg3K0zTm0cR6PTEqTZa+yiqrAEUJyb5do//jH//Y8blr+TJcGY3M5Wtf+xpQWJzn9Xyu7Rq5I6sTMiMZTR3E6KOYYa4KlFnIDGTpkdnKsvTd/P73v++41gsvvBAo69lGnXj8Y489FihKRoakItJG+jNUge0+BL9bBxNFMWmLmI8Tc0C8TzJYbSG72n333YGyruznP/nJT4Cynq0yli3+9Kc/BQqL0xb6Cu0XT3nKU4DC6trrksUorsnCtfmYm1Y1RuxHtlEG6xiwz2oLGam21A+67rrrAsUWfv6lL32p45rtf/oronKS/Ytll112wW/rItoizhcqn1hzzzZqC9m6tnjDG94AlPnCz88//3yg+Du8Js/3xS9+ESj9wP5n/9UW2lCVKJZZZpkF36mLWB0hVh6Idf6EbYwRms5Z+rFUKCocfcJVY+Qb3/gGUJStNtc2jp0nP/nJQJmr231NdeaLVDiJRCKRaAS1fDhjY2OtJZZYomtZb5/GT3rSk4DyVHzRi14ElHVK1yNdh7zlloeLNfsU9ukq85Vt+vrkk08GCpPRV/TZz34WgF//+tdAUWSygfa1WCgM/L777pu0D0dbtMfW/9/7QGFtkaXJ2rfeemtg/Nrs8573PKBED7kmK+NwfTvmpPzgBz8ACsM1Z+Hwww8H4Je//CVQ/FbaNtalmjlzJg888ADz5s2b9Pp0lS1i9J+2sM36DbSFDNfoM1WcDNYIGtuuLXytjY1CMk/D/nLkkUcC4/tFtEV7FeH777+/li1mzJjRWmqppcblJsRq4PZVow4dI9tuuy1QlI1t33jjjYGibKItvHaZt/3plFNOAYov0H5xxBFHTGgLx0i0xdjYWG1bxPmiarsA2XpUmttssw1Q+kW0hSrMFQ/HhKsGzhfaxvlCtu89+MxnPgOMHyMeT4XWPkZ6tUWsfuIY0SbOnd4/x8hWW23VYQvnVseIUaiuAqlYtJn9RNsYneZc6/fi3KkC1hYTzZ0PPPBA+nASiUQiMVyo7cP5v42HOt6Le60I/QOybp/WRmL5VBX6WIRPff0XKhGf3j71fdqeffbZwPjabP5O5hTrQ7VnEE8W+ixi9dRYEdn3VTYyTdsiU5WFySR8X8SaSdrE45q74rVEW6ikZLL6ikS776nuHjBeZ+wXHidWyvbc2kJ2b56EysWck5h7FO+fttBvZe6KOPPMMzt+b3vsX/bLuF9KL/lZrVZrwqoC9otYP84xIuuOtrBfaEt9diL2B++/EXiONc8X+0XMndM/Ehm4yrcu3KCr/ZgxW117x/nCfmJUmv1CW+qHFNE35HGdL7Sp84UsP44R+53nn2i+qGuLVqvFnDlzxlXMtq+oXLw2z+3qjdewySabAGXu9Peqw5gLZ3/wv/PLpptu2vF7x0jVXF5li7GxsczDSSQSicTwoZbCka34RIux4xE+tX1KyyT9vcxVViWr1x+h4jFKTSby+te/HiiM1vVGfQNGlcQ4+lgTKVZmrYv2J3tkEjFiS8Yg+3b92WPot5CZatMf/ehHQKkTpX9C9bbrrrsCJU7+V7/6FVDYYGS8sSZTzKKeM2fOOF9MN3TrF1E5aIuo9myDtrC/uK6tLVzfNmIv9gtZnP3C79sv7DcxyjJW8p07d27tytkiVgGOkXpChaHCcSzYJvu0Sta1eytKyPJ//OMfA0Xp2C+8VvNvVExGP9ovYi5MrCc2Z86cnmzR3i/iPloi7jpqv5BV2y+cL+wXsnrnC9WfUWqe12g2VYK2MFo25uu5OiFiheVebdFexSMq0rjTp2Mnzp22wfkg1kizuobX5hjxfK997WuBMl84RrSt/lJt5b2yXbFOYt0xkgonkUgkEo2gdi219to5USHI1oxsMArEp66sPO5y6PEuvfRSoDxlZcLWlfKprHLxeMaBy6A8vr83wkvmYvRK3EO9F8T6W3GHz7inzwknnACUfIu494drwxdddBFQok+8tph9ry08n2xNVu/xtZXVHPy+tvH4dStmawPrqbVfi34H7e7auOc+6aSTgBK9Fisra1NzTMwxsK1G8HlclY3nUznJ9jxv3DMk7jkUq1v3gti3qvqF6s0IKiO2/G/UkceJY8TjGbllf/C/tlDJqHCq+oXfj2OqV1bfPl/E+nKey37ifXG+8LX9w3nG/uEYiXl7qgLHSKwfZsSWx/V9/9sPHLPOF1O1BTBu7ozXbr/w/n/ve9/raKtjxO/ZDsdInC/sPzH3zOOonBwjnieOEftZ3Cm27upQKpxEIpFINIKe9sOJUV0+pWXn+g9kIq4v6pPRj+HT2XVp/Ro+TX3Kuw5tXL4Myb0+XH+0oq6RPj6l41pwRK87+I2Ojo7bqc9zyLrNe5DZmDPgNWsLPzejV3+D16BNfX/nnXfuOO8ll1wCFF+A/g7zNqItIkNrj0qaSpXkdsh+oi08lzkArsWbg6Q6UKXpb5CBqg5dZ95+++2BYgv3+nj6058OlGxr1Z+2kN1V+azqRuCI0dHRSluo/s2H8Nz6YvRf2S+0hf0lKhvHiL6f7bbbDij90DFitr22MOdNW8QIzohebdFqtcbNF9pCtqyfKkbi+T9GHfq+1+w1qNpUcy9+8YuBMg9dffXVQGHz5513HlDmC/0mcb6IY6UXWzhfxMjHuE+Wfdt+4Rjxmo1G1SejYlXZxCorKl1t4f11jDgna4to0zhGYqThjBkzaq0EpMJJJBKJRCPoqZZa1bpdjHiSlcX1QhWLT0m/r8Ix2/XnP/85UGLOv/nNbwKw9957A2W/GyM2VDqyj6uuugooLEJW4GvXNWWJdTEyMjLO59H+5G+/NhWKiNVYtZ1rqCqTY445BoDLLrsMKH4x60G9+93vBuDaa68FCnvT9t4rbSHDkXV6XlWI6+C9IO7TXmUL2b2Qtct0/X6sCn700UcDZQ8Q2eBXv/pVAN71rncBhdVH1eD5zSiPO5R6L6dqi4X1izhG7NvaTHbtGIn75Kj24hjRptYV3G+//YAStRjHSFSDcY97X+u/WGaZZXpWe1X9QhWosrGNtkFFquq3X6gGHCNWCtC/Jbs//vjjAdh3332BonBU1PYL2+MYirbwXk7VFu3Hjr+PUYwqF8eptnB8x6ixOEbsF84n1kzTFtdccw1Q5gn/awvnC9sbIwyj/3OySIWTSCQSiUZQu5barFmzxuWzxOxTn3qvfOUrgeJbcSdP2b1r9MI19xg54zqj5/vyl78MwI477ggUVue1qJRcm1VFyKRUNLZD1jlnzpxatdRmzZq14JyyoSpbvOIVrwDKGvqBBx4IlJwAr9Frdt06Rq9ZH0zV+KlPfQqAXXbZBSgMR2YiM3JdO65Pu+br5zNmzODBBx+stddH7BfRrxWz2rWF2c2yLpWOysRIKdetta33Ub+E/cJK2S9/+csBOO644zp+5/q0913FHbOrtcXo6GjtPWDGxsZas2fPrsz5ijs6OkasALD//vt32ELWL8s2Asv+ZbSSa/32Q22hrVXEXqtjI9rEfhH3UalbM2siW8R6cr72XDvttBNQ/AnuXmkbVcDawooEsnz9Y+bz2Q+tlG2/+MpXvtLRDpVS3A9HZeZ8oc2sNNDLGKmaL7SRY8R+YX7V29/+dqAoGVWaatxdUe1fzhfawvNpix122AEoY8R2aes4d2pzx4gKfHR0NPfDSSQSicTwoZbCcdc6WbsMwadn3OXStVRzR3x6HnbYYUB5iso09XPI4o202muvvTq+b+SFrDBG6pi384EPfAAoLE22GaPqfGrfddddk1Y4IyMjrbGxsXE1h+J+OEa+uJZqLSRV2Cc/+UmgMAZtIQM26/q73/0uUPwU2kJWZhy+7ZEJyXw+9KEPASU6KfotZMqzZ8/mnnvuYe7cubV2M2y3hX0qVl+ItrBfaIuPfexjQFE2snz7kbYwZ0VbeE3ef9evPZ+2MILP8/i7yDZlyosvvjj33ntvrarAcYwIbWD/iDkm2sI22S+iGjPST3Vn5Ym3ve1tHd9XpZl1bwSXtrbK8CGHHAIUG8XdcB0js2fP5u67767dL9ptoX1jDTXnizhGtIXzhX3dvm+/MNdEW+yzzz5A6Q+OLdVCtIX94sMf/jBQbOgYcWxokyWXXLInW4yNjS24VhF9qfYLfbXmV2mLj3zkI0DJMfK+6cPTFlYJf8c73tHx+zhfeC+0lf0izp2OEVWf/WLWrFm1xkgqnEQikUg0gto+nNmzZy9gaT75I4OVifqUjvWk3P/EiBt/Z/Vn2Zl+jVjh2KetrEAWJwMxGsX1aNdojTLxmmPuzP3331/bh+O1y4qjLcyGNxopVvd94QtfCBQG4udWtnUN1TV612y1kedTqVg7yWsyCknIiMy2jjko8+fP72mvj1mzZo2LOvJaZHXrrLMOUJRsrNbtfkkxW9r3ZVtGLcmy7BexUoFRS7ZHW8i0tXlVv+jVFksssUT7Hksd12i/kEnK0qMC8v7Lwn3fMWIOkvkZnsf+Yv/yWuwXMSfF155H5jxRRfj77rtvSrbwfsVae65IqHCiLeJ84e/dOyjWVvM8KqGoNp0vHCNGZGkrzzOd/WJ0dLQ1a9ascbbwtbZwjKhk4yqB80VUpL6vLZw7PY/fdzXI++6uuCoX+4VjxFUGbRHzcebOnVvL55sKJ5FIJBKNoHYttRkzZixg2ZGJCNmST00jbVQ85gb49HX92miVyy+/HCi1s/ydUSyf//zngbKOrQJy7V8GLHPx/RjZ076zY12MjIyw2GKLLcjlkVH6X7g2qi2M/XcdWVtoU9evVQFXXHEFUHIEjM+36qtr/Ua1yUSMdjPiS1vIjG1n3NnRdes6sPpE7Bdx91MVhbYwSlE25zXKst0J1s9VLObRqBqN6DFiz0guVaDsXzXgeawrZXvtjzE/p64tZsyYscD+sV8YjaSKc2yYX+G1WsVXmzpGVETeT3eG9XdWoPjc5z7XYQvb439toeqz3pi2iGNkOm0Rx5u2MEfIfqFfwvvt/bGvOy+Yo2bEp2Mn2kLfsDZ1XnBM2i+q5ov2nJS6tdS0hceKqz72C+eLWBXa++79dt5RKTsWnDu1mbY0Qu+oo44CSsUT/eG2y37h7+MYaa+YDePnu25IhZNIJBKJRlDLhzM6OtqaMWPGgnVA2bHremZ+WxMr1mVyvdGsVn/v+661ymxjprrnkaH4dDazXMYj03F9UnYm+/e8sZJvnTwcbWE0mUzQY8k4tUXcRVPGoErz87hbqTkoQibk91QDRuIYfeJ6tqxNW3hcfydrbL9XdXNPutki9ouYhe/9if3CiB3/G3HnceN+NvYLbeE+SjJYK1NYZVifj2wxVuCdP39+T7aYOXPmAgXbvrcOlH7hjotea9wnRaaprbxWbawtRFSTqgqVi7X2VFIyWH07bZGaHbZoX7Ovm58VbRHtK2t3V9I4RuwXKh9taVujLaIa8zwqXPu8/UJVcN111wGlXzgfOV84Rtr9FnX7hRF7ttn+4LizXzj3xf1xvKbov7St2sTfx2oJtt2+Hm1hjpNzZ4xi83dxFcCqGunDSSQSicRQoadq0T7l4lPYp2Csvuo6pFEoUdnE6DYZsU91I6pkZTIeGYhPfdf4XQf1fHGnUdstI+pl73qPF/duETJFI+RUFjIJGY3sSxt4za69+z3ZuIxV5RIZjzaUuXovVH/aIkLG0kvlbJjYFtrb+2TEiyzPtXaZrrYw21qbGHHjmr5sXxvrzzCr2n6oDa29pg8p2iLW+JJJ92qL0dHRBX08VtKNtvBa9c3Z9yNb9795FtrO9+0XqjjHiDbSf6HS9do9n/3EsTCdtojKJioPbWEfd4x4n2xDnC/0Q9h/fF/2rn9UW2gjVxXsF0Y52i/1c1b1i+izngzcF8hjxLnT+xSreKtIvUaVq/5ur9koVm3mfOPc6RhROXse+5O2iDlOVWOkfb7IHT8TiUQiMXSo/ageGxtbwGBlYTITn3TmAvi01efj55GVuf+6rNDoIuPzZQVbbLEFUJiLjETlYwy7ET4+hWUTtjvWM4r7t08GRp14Dn8rs/Ac+hNkDLKyWE9KVuX+6yoWGYbr1DIW8zGMMvJaVD4yWhlvZNyxWm27LXrZAXVsbGycLWTv2sL76vf0rcSMc22x2267AWXtX1sYmRNtocK1f1kRV4Vkv4h+FW0R/Si92MJ9gbRBVb/QFvYD1Znnk0VrC2vleR9l5WbJ67/afPPNgZJP4RjRFnGMtOeWtP+PY6TuvifaYnR0dFwNPPuF7ztfqOb9XGi7uP+RY0RfrupeFaBvyGtXNThfOEa0haiySXv0Zd2dLuFhG8bVlTh3Ol+oPL1/Kk8Vjd/bfffdgWJb51RVnVFm5nWpdLSF84Wqz7k1KrEIx9jo6GjldyZCKpxEIpFINILaCmfevHnjsl99+qpkfPpFX4zrg64/GifvuqLfO/jgg4HCdMz8dW3W931aux7pOqZP54n2ZW9/HTOI66DVavHQQw+Ni4iRJcc1Vm2hzWQoMlsZrApIZmn1YI8nI7HNHl9mYkaxqkEWqO8g5pxE/1Wve7VP1C9ElS1kb7GyhHk1Rq15v62RZT8zRymeR1VnZQLviddsv4i2iJFevdii1Wp12CLug2PElv3GVQBfq+L0P2oL+7bHPeigg4BiCyOt4l4+Kh3HSKxMEcdIzBOZ6hiZO3duZS5StIWvZf0qH6/FXBKVse9bed3XRlrpy/N9x44V1M1Ri/tjaSNtGW3RK+bNmzcusjJWUneMqPZsu33ZNlsFXFs4ltwrzOPF6hr2F+cL/Wf6gmO/iPPIRGMkfTiJRCKRGDrUrjSw+OKLj/M/uLbrurLZrVYc0H9hhrBP17h/jceLNbdkQLI4I21sh/kdZhTHHIIq1ubrOr4bMTo6yhJLLDEuuksVFdeljZyT0VpRQAaqerO6gm3TjyV714b+znh5r9WMdPfFcW3Xa4717+I+LXPnzu0pi3rxxRcfx+ajLYwyk23LtrwmlUmMXrKN+m5kb9FfoQ9HNqffyz0/ZGki1rOK+T299otZs2aN88VoC8eIUWmxIrb3T7+D98toR/uZe8GohIw2MhPdMSLsb+6OGfOvqvrFVGwR+0Xc+TX2C5WHCtj7bb+wTfYfESM6VQnawggsr9UcqCOOOKKjPf6PbF7Yr3tRe1YmqbKr/iltZb9w3Jtf4zX5O23mtWkzx5b9SXXnGLE/Wo3anYV933Zoi6q9z+r69VLhJBKJRKIR9OTDEbJnFYhPwehDsRKAfgQVkIrF4xhRYXSSvhv3vZEF+pQ2IqOq+nNkVjFKLq5f14Hr0zHqJNrCz+PuorJ4baF/S5v4ff0WqsDTTjsNKEpJW2i76EuKWf/RFq6re296zbeYN2/euGjAWDsrKmHb6P1X0craZLb6eNzzRVu4X5K2kMmqFiP7ilGLcQfKaIuZM2fWVnv2i5i7IeuOisf7LoywUgHJ5rWFqsDdc/3c3VO1hXsCRebstVZVvhDaon2M9OLDaFcDsf5WXBXQ7vod9KloC8eGfdroNCO1rMnmfKEtrLLg92MkXlS+cX+kicZIr7aI/cL5Io7LmP+iAnW+cG50HvE+2y8cQ6eeeipQ1J9jpGq+iBUQvEe+jlVb6s4XqXASiUQi0Qh62g9HdGOKMR9DheLT3VpGRuC4Hu3v9GfESA3ZWYy88f24L0vM6o0Z8W07+9XaD2f27NnjmGO3tU6jSVxj9X3XsY1a01YqItdgjZ+X4Whro9NkSl57jMSKEVhRlcycObN2zaycHJyUAAAgAElEQVTJ9osYIWW/iHXHtIX9ws/9fcyriHsJuf4dbRHVRWS6USEvvvjiPe1dv8QSS4yz82THiGPANpp1bwSf/cLjuqYfbRF9AbEfRNtMtC9Se/u0RS/74URbVOWAaXf7kgrFNpt1b1Sr/g1x/fXXd/xeJa06UE12my+qVgna54tebKHKbb/WqKain6tqvnA1wDFiv/A4sV/oQxZxzGijWEU+9guPbztmzpyZtdQSiUQiMXyo7bxotVoL1g1lntFPEJ/efu66n6/jOqZ7kpt/4XnMKJcFxjyKyFhjlneMCIl7UfSKhdlCG6hQoi38Xay+KmRr2sKcAG0oczGfJ1ZLiOvTvh/XZOOeRr1E4Hg+r0k2FtmQtog+Hdfq432zreZVmD0t+7IfyGj1f0SWGG0RIwujfyvmGtRFq9VacL9knpMdI9pQxikT9Xv68KwGHseIjNjXUdWJbv0i7kTZ7n+og3ZbdJsv4n3zfnq/9d3Zf6wYoPLxd/onus0XIvaLqry9dj9LL769Vqu14L7GnTe1e9V8oS18HdWZ/kzzcuJ84RiJvhePF3PV4lwZ+0UcO5NFKpxEIpFINILa++HMnDlzwVNYZSITiBE50UcSa1dVxbn7fZmvr+P6ooj+Cn8XKyDIBuMare245557au2H0x7ZE9ehY60kbePruBePbbDNcR+MuLNotF08b9s1LdQWMQ9nbGysp/3a221hm6NiqeoXscZdVJ7RFnEHyVjhuG1/I2B8v4iZ7toi+t2sdFzXFu27Y8YxEvuF903bRFVWVbEg7tkSs/ljRF4cI/4uVkCo6hejo6Pcd999fbVF7CdR9VfljsU9W2J0ZFT9UblUjRFVwkTzRV1buB9OVDCxrl+8dm0W/Y1V84XHjWMkKtiqauD+ztcq5rhDQHu/qOPPSoWTSCQSiUbQU5SaT7e4jihinaC4hh+fqj7FIwsT8ekf135jRd8Y0x73vYn1gdoieGopnFmzZi04ZlQWbd8Dxse5xxwEX8d17Mj2Y25AfD+uT8cd/6rWpWW2iy22WE+RWe22iNcUWVm0RZUNoy2q+kW0UYy8i5Fh0dfn+Tx+e9Z9L7ZoHyNVey7Fvhi/VzVG7LPtyqH99/H9qHxFVE6xX8R6YvaLXqLUYr5VvF/RFralasxEW3SbL+IeLnGlJfovY7+YaL6o2y+q5otYvcC2xAi5GFVom6JSioqoKv8uzo0xRzH6X21PHCMzZszIKLVEIpFIDB9qKZyRkZE7gFu6fnHRxaqtVmuFyXzxEW6LSdsB0hbtSFsUpC0K0hYPo9YDJ5FIJBKJXpFLaolEIpFoBPnASSQSiUQjyAdOIpFIJBpBPnASiUQi0QjygZNIJBKJRpAPnEQikUg0gnzgJBKJRKIR5AMnkUgkEo0gHziJRCKRaAS1diAbGRl5pJcluLNGaZtHtC1arVat0uv9bMugkbYoSFsUpC0KJmuLqW15uYjASqeTKOPzSK11NJSocV/6cl6R5Z0SiWbwX/HAyQllODGo+5L9IZEYDNKHk0gkEolG0BeFU7VU0m0JJW5WFjdW64a4mVFi0UC3fhG3041bD1f9zuP6u7r9KZFITC9S4SQSiUSiEfRF4XRjqjJNvxeVjf/dPjVuwxu3T43nXZSZbbSBqLJpXRXZT/R67tgv4vHi57NnzwbKlr9x6+q4nXI8nv9TCScSzSIVTiKRSCQaQaNRaq69z5o1CyiM9VGPehQA//nPfwB4whOeAMDtt98OwGMf+1gAnvOc5wBw2mmnAbDVVlsB8Oc//xmAq6++GigM1vPce++9fbme6UBk3VGdLb744kBh86ussgoAd999d8f/ZZZZBig21NZRRfZT7XVTNlWqzbaqXFQo6623HlDur69/85vfAKVfPOtZzwJKv/B9lc+1117bcf5ll10WgH/+8591Lm8oMHPmTGD8/bWv20/Ef7Nf87/52ocVqXASiUQi0QgaUTgyS9mZTPbf//43AM985jOBwkSf8pSnALD66qsD8NOf/hSAm2++GYDVVlsNgCuuuAKA9ddfH4Bf/epXHee9//77gcJ0ZTrDmIchGxO22bbKymWw+rdWXHFFAO655x4AllxySQDuuuuujt9HdTEIxLZEH81jHvMYoCjbl7/85QAcf/zxAOy6665AUcAXXnghALfddhsAT3/60wHYeeedgaL2VESqO23TazRlE6hqg8rmsMMOA2DLLbcEyhioivQchmvqN7SB84LX+uhHPxoo9z3aqMrnl5h+pMJJJBKJRCPoi8KRQcyY8fDhZRoyWRXOU5/6VKCwsI022ggofguZ7oknngjAH/7wBwCe9KQnAfCDH/wAgO9///sd53vwwQc72uH5B8lku/lqoq20kUpm6623BmDOnDkd7996661A8YPpy9E2KqJhWMeOzFJfi9ekH+Jd73oXUHwzr33ta4HSL/y9/Ufl7PF///vfA3DQQQcBsNRSSwFFBUbWLwbJ/qO/wf6w8sorA/CnP/0JgBNOOAGAL37xi0Bh9apDr9HjaVPVvjZ0jHieRwLLd76IcCXF+6uNtY3XfvjhhwPw7ne/u6/t7AfiKo5ja+ONNwbgxz/+MVCuWbXnGLrhhhsaaWcqnEQikUg0gr4oHFmUzNWn6CabbALAUUcdBRSfzOmnnw7AE5/4RAB+97vfAfC5z30OgHPOOQeAv//970Bhayqia665BoDnP//5AHzlK18Bxmeky/piXk8/UZVbJBPRVjJU27bbbrsBJfLuvvvuA4rPRuXyuMc9Dih+MI+rQrrjjjsAOPPMM4HxkWGin+w+XntcO1eB/PznP+/43T/+8Y+O1/qn7DcrrPBwYW/7i7bxe/aH5ZZbDoAf/ehHQLn2qHBk+4PM29JvddZZZwHwpje9CYBbbnm4rqz+KRmqDPbNb34zUPqRCvcb3/gGAM94xjOAooD9fYxqW5Txl7/8BYC1114bgN/+9rfA+DGoTydGKQ6zsllppZWAMp7FOuusA8CVV14JlL6t6ncO9vV1110HwNOe9jQAPvzhDwOw/fbbA2We+MAHPgCMX52a6thIhZNIJBKJRjBSh9mOjIy0RkZGutZCi0/DtdZaCyjRRK47+zSVzf3iF78Aynqj69Gyure85S1AYe8vfvGLAXj84x/fcRyZqmzf9qiMxAR+jStbrdb61RYomOz+FtFXE/0YRtztu+++QInUMyLrzjvvBApL1z8hS/OaPe6qq64KFFv6/tlnnw0UxiOzXUgFgynv9RF9ZNGntsQSSwCFfbnu/M1vfhMoSsX7puoTHu+YY47peN+oNpnuJz/5SaDYMkbHxQoW0SZN7HsSfXgqm4MPPhgo/svvfOc7ACy//PIAvOMd7wCKApa1v/e97wVKDtNf//rXju+Jbv0goglb1PWtvupVrwLgjDPOAEpumqsH0S9q9KKK2/tfV+U3uR+OvheVqqsBMbo1wj6tTf/4xz8CZZ7wfRXOL3/5S6CMvcn6diZri1Q4iUQikWgEtX04C2MB0S/hf1n5l770JQA+8YlPAHDZZZcBRfHI7lVGPo2NQvN7KqYLLrgAgCOPPBKAiy++GCgszuNUZV83GaUWo5D0a8nOrIaw5pprAqXtz3ve84Di99IP8eQnPxkoTOT1r389UCJ1XvaylwFw3nnndRw35ihMpw26HSvWxvO+uJ7sa5XMV7/6VaAomL/97W9AsY02db150003Bcq69uc//3mgRCmppFRMVdUYmoxWizbbYIMNAPjMZz4DlPt/8sknA0X9/exnPwPgkksuAUoOkj4a7/c+++wDwKtf/Wqg2HiY83Jim1QmcYXC+2m/Ut15bap5VwPsB8LPFwXsuOOOQFG4Eao0baK69347F9tvYlSb6tAxs//++0/vBfwfUuEkEolEohFMS5Ra9FPENVHXm40ikbW9733vA2CLLbYASrSSFQP8vTkDMhUjcWQ+F110Ucdr18N9ba7KBGvzvV5ybcimzSEyx8R8ik9/+tMAbLjhhkBRKr4v89DPIbPVj+XarlEnm222GVAi/2RG/bRB1bFifpTrx/YHo8lUMPorvG/+jypRH582lZWpcI3o0Ycjq9P2RjcOsuaW9QFVbSpSx8wb3/hGoLTdVQLHTsxRMmdtl112Aco1mqflcVTU0xV91A/o44t+J+E88/Wvfx0o1yKWXnppYPK1FI3s22mnnWq3tV+wHxiN6twXI+yi2osRuKp/51ihbbSVfvV+9YdUOIlEIpFoBLWj1P7vP1CduR/3K/HpKTv36Ssrl2EaiXXppZcCJd9ChiwD9rXne+UrXwkUX8AXvvAFoESr1LjGaY9Si2ulMgwj7q6//nqgrLn+61//AgrTNbJKJuLvZTzazMg+o5v22msvAN72trd1HFdV0c0m/YjAidGDMlj7x+abb+7xgOKH0JfntX/ta18DCqtT7fk78y9Uhf7eiC0hi+vG5pqIRjIKSf+k1RasgO01mXtmvUHHiGNp2223BUpehhGeUeEIlfdk1V2TkVlV0G/pSsn3vvc9oFyj6LdPrglbeA3WA3SVx9qKfu7cGv1Uwvf18QnrVJojaQX+888/v1Y7M0otkUgkEkOFnnw43XafNJpERis7d/3RrGjzJcyml8H69Hb92rV52bkVB1QFZtnL8o0+qapO3ORavexZZuE1GqEX4+FvvPFGoOQsCVmdfhBtYCSXEXtf/vKXAfjoRz/acV6VlBhElFJUEipgWZt+CO+/0YkqHdez9U8YzaZNjj76aKDkMPm+7E+FYzvsB4P04XgfYsScCtbK2OZbqVQdW+ZjOOYcU/qtVEjaVn/IMEepdYPzyQte8AKg+ADNITniiCOA8XXkFkV4n71vjgVhBHCVsrFvm+8XKxWojJ036iqbukiFk0gkEolGMK211Fwf9r9PV6M+zH7fZpttgBKNpiKRsZiDIGM1CsmnvQxYpnrqqad2vJbZxvj9QbC5WA3a1/pWrIWlmvO10UWyOdnac5/7XKDknhip5fvHHnssUNSAdcZkMqrJJhH7g/fJWmj6cP7nf/4HKCrPiJvtttsOgHPPPRcoVRn0X6n2jNDze6oB778KRkVlrbZBsnz7qMpXJatidVXAtmsrr9kKAkYj7rDDDkBhvuZXOMZi/bhFcVdM22xOiX4sccghhwDFVouywnH8Og/EOoTRlxPvo9GMjo2Iyy+/HIC9994bKHmBJ5100vRcQEAqnEQikUg0gmlVODJFn74yW/NkXLM3ysj9J4w+83fvf//7gRJpI5vzcyN0jEJzZ1CVjtEqVsSNVYAHwWhlKK6lqu7MOTGfQhUoq5fByoCtbeTxjCoxikXfz3HHHQeUOnNGag1y7V4/hWxMBRL7xZ577gkUpWJOktV89VvI7lQFqgXZvlFv2sJ+ETPS+1F1YbJwbd59b/RTqvJcU7ft++23H1AqrFtd2r6tv+onP/kJUCpQ+H0xCH/mdME268+wn7hnkLbSlosyYpV3Vyq8f1YDdy40Yld/lr+PCkfFayUKx+Ypp5zSh6soSIWTSCQSiUbQFx+O8KksSzfaTCXiWryfr7vuukBhwLJz94KQpQvrgvl0d83WarBVjDXmxvSD0XruyL7Np3jd614HFJavL8aoJNWZuQayeWuw6b+w7a65aqu4T4a2GQRiRdu4Du19NidA34314ozAURVa+Vol4xq+u6Iakadq8HtGqUV2H/tDk+zfc6l0batRaUbiqf70a3qt+gKt9rv77rsDRdXLcM1ZihF6i3K0mvtlmbOmn9PKFNpgUYYKxHy8X//610BROqr/F73oRUCpluDYcgxEfOQjHwHKnOqY63d9uVQ4iUQikWgE00oBYi5BVDzWDfvUpz4FlEirD33oQ0BhqrI8n86qAF+rBnzfSJ7oO4qZ5DGSo5+szmPra9HvpPqyMrYM9tBDDwWKkjG6xHVoa26plFRAKiaPKwP2GlVY3ptB7GEfdxeVifq+bVfRGJljjT0VsBF8sroDDzwQKJE1RtxoI/MzrCZuLops334UI7iaZPuxUnXMv/rsZz8LFB+eforYdn17+oT0c0Vl466o+kVjf1iUlI73y+hWX3/sYx8DhrtOXDfECFurgZu358qF1TocE1bXcExUYb311gPK/W6qcnYqnEQikUg0gp5qqU3wPlCelioO2ZZ5Dz6tVTpGp1k7TUZ64YUXAoW1mZPg92RpRrm5Dul+3bK/bmvwE1z7tNdSk6msssoqQGGqKh6j1oxOM2JL1qZC8v9BBx0EFJ+M0Seu7erPUiV4PlXEZJXNdNSJiopSW/jffiNLe/vb3w4Uf4bXcssttwCF3Zst7d4x2lIfoGrQfmQUk5Ff/r7KjzGIHT8jXFOPtc5ULLZRtXjTTTcBcNhhhwHFT+rYiGMh+jsni0HYosrnal82S/6tb30rUHLQXAXol8Jpspaa1x79Ulak0G9lfzDS15USbeYqgKsF04WspZZIJBKJocK0+HBiFJK+FZmrGd36YMwlMGrN/z59ZTT6NWTvMlwVlMz0qquuAgqbtz0eJ/qWmkCsMKAvRWbpGqzRJlZH0H9lxJYKSN+MGeRWIjAiywoCcW8i12ZjDkwTEViRRaveZO/6VGyTtdP03RhxFysGqGiMOnPvHyveqhb1h6hobI8RP55f2w8T4pq699Xqz7L2l770pUDZ0VPfTFV2vQzYVQAxzBUHvFbHkrlGwsrojm/vc9wzZlFEVOFx11qVjbDfHHzwwR3vx2jWQSEVTiKRSCQawbQonLhGGtm8bEpG6lPa18bRG13irpfu465ScnfL9773vUBhqConGY3t8X39H67pNhGNFHcr9Rrimrq5JX6uv0sbGU2iOjBn5ZhjjgHKNZmXo5qQDcYqsrGWVj/gOWybisZr0gfntcjK/dzX9h+rKMjC9f25Dq1fSx+eikdfkcpZ35D9xn6qAoqKuEl065N+LkM1CslaafaHLbfcEig28ndx19V4nmFSNu5vZA6abN776X02qtVVgFil/pGEWIm/255kcZx/97vfBeCoo47qazu7IRVOIpFIJBrBlKhAzLNRWcjGZZBx50+ZrPWezHrdY489gLIu6dNcJeT+NyqWWB9MdSBbU/F4ftGEL8dz2BYj7iIL09+g/8HaWLJ3M87N2zEOX7bvNapkPL73QlvL2pvMSbB/xMg4q0TLVK0j5/3TL2UEzite8Qqg1ANToRiV5PFVMu786vFl9VYdN+pNX44YhLIR3frkGmusARQfnhGb55xzDlDyMFTMqvtnP/vZQMnLGWTduG6wzS984Qs73lf92+Y4j+i3UAGZnf/Od76zzy0eHGL+lteuLRwTRgrr+4sKqOn7nwonkUgkEo1gSgpHtly1x4YM0rj4WPPMaBNzBmRnRhcZdWZFAn+/6aabAiUyS5Ugo5bRNlFRoBu0kUpEf4Ztsnaa7N7vmUMiQzEXSX+Dx6nawTHmKgwiez6eU1vIzo2YUoGaJ2NVaPO1tIk+IfN13FdJv5U7gdp/zEXSh6hfI96DYWL5IkYbmmejArY/uLOrO8jqF3NvIW0YjzeM2fdWlBBR1fnaXW7dQ8r7a38axvvZL6j+9GO5CmT1lhNPPBEoymfQc2IqnEQikUg0gmmpNFAF11h9qqpYXJs3P8ensWv0MhdZm79X0RiRZYSOLN+1exlR9B1M4lqnvdJAZJbC1+7QqLpzf5ttt90WGL9r4Tve8Q6g5CTFenHayuNHX9JkMZ1Z1HHdWCWqL8b9jA444ACgKBqjzS6++GKg7BzrHiDmnuiTsZqDv9Ov5b4pfq9q//eq/tHPjPKqPXlk66p1/RtWSPdza++pElU+9icjOqdL0Qyi0kD0Q8b6X1Exb7TRRtNx2q5o0hb2C/uBkZyqe/tJzPvz8+jfnm5kpYFEIpFIDBWmNWA9+nIiu15rrbWAwr703ahoXIM3s1wmLDP1Ke0uiLFmklFJRmwMUwXcGFViJI4VA1x7l73L3qw4YBSbfgjXbmXr8fjDkFdR1R9Uur5WkVpxwkoUKh77jb47/RlWJpDR+l/beP/tb9pmmHa7jPct+t70xVgF2GhFo89UvlYVNxLP/XGG0VczWVTt0eL9c5wvytc4WcTaefYPq80ff/zxwHgfjZXS3R/L/jMopMJJJBKJRCPoqw+naqdH319mmWWAsue8zFam6i52ViCQyaiAYjRaVWx5jWucdh9O2/c72hJtoV9KZnraaacBJe/Ga1btRSZTdfxe1V0/fDhV+yTF+6dNjOBzvdoIHH0/RieKbvXiFuKjWVjzB+K3ENrKseI1matmjpqKOUZ6TTcGaQthXUHzbSKstdjvWmqDtIV+LetM6ttx9aCqunS/kD6cRCKRSAwVelI4ddlzFfsWcR276v0YgRWjn1zXrDr+JNrdN4XTDXENP/oZul1TxFT9Vr2wt6n2i26oUsxVCkmofGKU3GQrCwwDqx8WpC0K0hYFqXASiUQiMVToaYGvLnuuWrvXJ1O197hM1cgM823i96rY/1Tb3U/ENkeWXhVB1W13ykGi17ZEdVelSLspGb9n/6j6fJA10xKJ/2akwkkkEolEI2h044iqePlYky0yXZXNoqBcJou61zKMimaq6BZN2M33Zr+pssl/Q35GIrEoIRVOIpFIJBpBXYVzJ3BLPxoCU8qfmS6sWuO7fbXFgFHHDjBFW0wiD6bW96cZjdpiyJG2KEhbFEzaFrXCohOJRCKR6BW5pJZIJBKJRpAPnEQikUg0gnzgJBKJRKIR5AMnkUgkEo0gHziJRCKRaAT5wEkkEolEI8gHTiKRSCQaQT5wEolEItEI8oGTSCQSiUZQq7TNI30TIeDOVqu1wmS++Ei3RW4uVbAo22KqW41HLMq2mG6kLQoma4tpqRb9rGc9C4Crr756Og43SDxSax0NBOuuuy4A11577YBbMnhM98Q/WdQ9X9xrKCtuJ6YTuaSWSCQSiUZQq3jnyMhIa2Rk5BG1J0vAla1Wa/3JfDElcsHIyEhrdHR03C6lg2L1EVNtx3QunUy1Lf3+fbfPcxmpYFG0xdJLLw3Af/7zn2k97mRtkQonkUgkEo2gtg9n0Gy1HxgbGwNyvXoqaFc308XCl112WQD+9a9/1fr9EkssAcADDzwwrm11MJFqmwzaVwGqdrFd2G8X9r1elcmMGQ8Pda+nl+M/Esf+ogrv9+KLLw7AnnvuCcDnPve5ju9tv/32AJx++ukAPPjggwBceumlALzkJS8B4O677wb6PwemwkkkEolEI5iWKLV+QeXhU3yrrbYC4LLLLgPg3//+NwBz5swBYLHFFut4PVksisrmSU96EgA333zzQNsxEeoyYSOiZG3+f+ITnwiU+/z+978fKNf+m9/8BoBDDz0UgJ///OcAbL755h3t6LVfTMHn0/MxqpRRtM2sWbOAsiZ/xx13ALDDDjsAcM455wBF5c2dO7fj91WIn09WmQ0D7EePecxjALjpppsAeNSjHjWwNvUL3g8Vy9e//nUAvvvd7wJlbBx77LFAUbb2A8dE00iFk0gkEolGUDtKrR+NkJHYlqOPPhqA3/72twC84hWvAOCpT30qUFjb2972NgBOOeWUjuO5hn/vvffWbcrQRqm5Bu+1y2iXX355AGbOnDmt52siAiey6RVWeDjn9h//+AcAe++9N1B8Oe9617uAwupltEIWt8YaawDwhz/8oZdmjUMvtpjuaDLvr9e+3HLLAbDjjjsCsP/++wNwwQUXAHDkkUcCcM899wDwl7/8peN4991334TnibDfyYwHGZmlDfbYYw+gsHqv5e9//zsAK664IlD6h8pWVSicJ/x9XQxDlJr371Of+hQA++yzD9BdwXg/11lnHQCuu+66ns7bpoAzSi2RSCQSw4OB+nBkHFtssQUA3/ve9zo+j+vXQpZ34oknAoWhyPYuuuiiCX+3KKxDR6hgbrvtNqD4tVZaaSWgsPoDDzwQKExnUcDs2bOBcj+XWmopAM477zygKN0PfvCDwHg/RoTvb7PNNgB84QtfAAZz36d6Tn8vS5fdf/KTnwTg1a9+NTCeyboasPXWWwNlLHzlK18Biv9TW+u/VDnHdsuEhwFewxve8AYArrrqKqDY6HGPe1zH9x0bjqGIXpXNMCD6t/0/WR/N7373O6C+shG99u9UOIlEIpFoBANRODLRL3/5ywC87nWv6/hcZlLFZH26+pRXKa2yyiod7y+zzDIA/POf/5y2tjcFr8F16ZgPog30c8jmFyXITI866igAnvGMZwDF3yAL/+Y3vwnALrvsAhTb2D+0zeGHH97x+2GpdDAVaCNVoGvu+h/83Gs0n0K/hUrnpJNOAuAb3/gGUGz50EMP9fcCpoB4/970pjcBxae31157AXD55Zd3/O6aa64B4OSTTwaKH+uRBJWpytcVD5WN0Wo77bQTUMaMtnza054GFFsaCdpvpMJJJBKJRCNoVOG4Zu+avOvNUcn4Wp/OmmuuCcBqq60GjM/beM1rXgOUyB19AC960Yv6cBXNIK7Z77bbbkCJwDEa6bjjjmu+cVPEqquuChS2ZX6V/WO99dbreN9q0yphow9l+bJ3o9LuuusuoPcKA8MA+7Zr8muvvXbH5/fff3/H57feeisAn/nMZwA4+OCDgWKjbbfdFij9xePLlB1T3SoRNInYBtWYbDwqG6/p2c9+NrBo3/8Ir02l4n1XvenTO+CAA4Ci9qMNfa1t6uamTRWpcBKJRCLRCPqShyMz/clPfgLAox/9aKCwrT/+8Y9AYfG2QX+Fvpjoq/H9O++8EyisTLb3vve9DxiftyEmUVGgb3k4f/7znwF4/OMfX+dnC/DYxz4WKDYScQ1/ujAdOQaRNbs/jr61c889F4Dvf//7QMmvMHrI/uF9e8ELXgDAqaeeCpT+dMUVVwDw3Oc+FxjPAqeKOrYYHR1tzZo1a0GfFL36k5ZcckkAzj//fKD4cuzjn/70p4Hi53I14Fe/+lXH913T32+//QD45TADpCEAACAASURBVC9/CRSGq83MdTOKqX31odVq9TX3JPaXurACxQknnACUfbq04XSjyVy1WO9xs802A8pcaGWBqv6lbZ2L//SnPwFwySWXALDlllv20rwFyDycRCKRSAwV+uLDUdkI11z1P5i9LKxw6rpzVCIyHp/mrlvGCB7zL774xS92fO/1r389MD7PRzQRzdSrshHRJtZQUi3K7oYJkanGnT+NqIk5JyKuL7tmb97W3/72t47z2K9i5FVkzv2MzGm1WuPUje/XgX3SaCL3L1EVWjdO5ePnt99+e8fvza8xOs18Lm2rbWTQKpte2z0VTNXnovp//vOf3/H+rrvuChTlsyhB+6vWvU/6KfV3qnBEnNO0rZGfzrlxxaTfSIWTSCQSiUbQF4UTWZOM87TTTuv4nk9f8yaq1r39L4uTyVp767WvfS1QolNUOPoGZIX+TrWwKFTClZGo7oSZxcOobCaLbnvG+P4GG2wAlNpZn/jEJwD461//2vFfxOPFLOymcg6mAv0OXuvGG28MFHV29dVXAyViL6oD/V9e+4033giUVQaZrWNzmPNxusH5xtqLsV7coqhsquB9tSK6uWvOcSqfiVQ2FNuYt2dF7aaQCieRSCQSjaAvCke2pcLRN2P+Rcy7OeaYYzp+J2KUkpFa+mZkafpmzJ6W9b/nPe8BCquru3PkMMDcokUZVT6yKiWsb0d/gwrYCBt9dt5/baRfK2aW669wXxQVzjAqW/v8hhtuCMAmm2wClLEhc33yk58MlMoDRp35Pf8bdbbpppsCJT/jkZCjYr/yGt0/yVykpnNM+omVV14ZKD45oxO1gRGeXrNz6v/7f/8PgJe97GVAmSsdQ00jFU4ikUgkGkFfFI5PXWtkvfGNb3z4ZGFfdRmp++HooxGRobgHjMx0u+22A8rT3B0hjcBQ2ZgX9PGPf7zj/MMMWb9RaBHWTloUEJVErIEWYc0sa+TJ+p/znOcAJUpN5WutPI+rQtIPoqKxavAwK13vq6sC9gP/awP3DNIvqq/GazRHyWoejjVtou9nGFXeZGHbtcUTnvAEoPg7H0kKRz+lPt2q6iz2ff2evv/ud78bgOOPPx4o0Y0vfvGLO77X77kxFU4ikUgkGkFfFI5PYdcZ3/72twPj19jNvr/lllsmddzIxqyZZm022Z3nt76U0WqLgrIRsrXIZLyGpuPnJwP3I9HnUoV4H70mfTTeN7Ogd955Z6D4eL797W8Dpfaafg2PK7ONWfTPfOYzgRKxFTHVTPfpgNdk9JHX7OqAbbz44ouBovKNwPP7qnvVoQzZPB1R5V9blCptR4U7zG2eahv1T+nLcRUgRnhaSUA1b7953vOeB5Taek1H6qbCSSQSiUQj6Esttd///vdAya6XZcnKXZ82+7kuZAmyQTPQf/aznwFl3TruiinTXcjabt9qqdXFGmusAZQd+bxm1+xf+tKX9vP0fa0TJStT8doHZfFea7fckKhIqtij/U/UzTmZii0my2jjHk5GnenX1E9hRJ6VBsxFM0LLNfr/+Z//AeCd73wnAP/7v/8LdN/lMtbsimiiflivsJZiXV+tEVvRh9wNg7CFikVfnP6q008/HSi18txrzMhe/djaZN999wXg2GOPnY5mZS21RCKRSAwX+uLDsZKp+5MYJWQOgNV+pwqrwVpF2nVN16vdJ+cHP/gBUJ/BDBI333wzUBiNkXwf+MAHBtWkKUNFEpWNqFvduWoXVBEr7Xp+fXz9jFabrD8oVtMw2kwl6/0+7LDDgBJt9OEPfxgo0WZek6sH+kkde1WZ51HRTKKi+tDASDyvzd1OY2292K9k/fqzFoV5wfsb+7JKZ//99weKX9saa+KQQw4BYPvttweKrZpGKpxEIpFINIJp9eH49DXj/0Mf+hBQnqauK59xxhnA+N0FJ9sWz3P99dcDJSJHxiOMkvvSl74EVEcntWHgPhyvTfX2i1/8Aii2cQ2/12x54/S75Sg0uddHjG7rNWJG26kSfC3TNULnrLPO6jh/rK0XmfF02iL6dLwf+hv13V144YUd3/e+y8aNYtTnI7MVKpxDDz0UKNFN3fYK6uZzGmYfzmT9ZX5PldCrqmvSFqo4I/IcM64W2KdVMK722F/0m+r3dgXFShVTraWXPpxEIpFIDBWm1YdjNJBs7L3vfS8AX/va14Cy77br0kaTWUHACJoqhiLzvOiii4DylPeprW/ITHTXuX3aLwpwLw8ZrpCVGT/vDo51/R6DzL6ODNT6XubfWAHZPBwhE42M1OPI3r71rW8BJTdFlr/jjjsCsNZaawFw5ZVXAqVfxOrR07VTaDtUMrHWmdfiGr0qXLUug41jQ5+e1xxZ+w033ACUyFBXAbrl4QxzDks3TLbt2labdovMGyS8n6r2WO3bKMW1114bKPfX950zo9pfffXVAbj00ksB2GijjYD+2yAVTiKRSCQaQV/ycGSURqVZD+yFL3whUNYLt956a6DEkuvrcSfHiBe96EVAqZVm9em4c6Tr4TJdFdYkMDAfTtzJL+7D7rXFCtr9Qj/Xp1XARlhdddVVABx00EFAWa+2bpiMVPb2+c9/3vMCpRLuSSed1PG+NlJNGJ2kf+yCCy4AujPjXmzRLUrN+/3jH/8YgMsuu6yj7Y4R1+YdQ+5q+5GPfAQoTNbznXPOOQBcccUVwPjq0CpcX0/WpyeG2YczWUxXRYkmbOH9cazYb2IFkrizp/Ba46668fuOMftB1fGrkD6cRCKRSAwV+qJwhFFB7sng3h6uP3tumehuu+0GFP+En6tsTj75ZKAoIZmya7Hun2IezpZbblmnuTBAheP+JjJao5W0jWu3Zpz3G02wN30v7uXx0Y9+FCi7WsYcFVm/+7erAl2Plq1FBmt/0WfzlKc8peN11c6joh+28NrXXXddoORJqNrtD95/r8n6ck9/+tOBEqX2ox/9CIBf//rXAHzyk58Eig8oRqdVVQdWQVdFLT0SFM50oclIzjPPPBMoc+EEben4vlCxqJQiXFVwnxx3AO02JiY4fyqcRCKRSAwP+qpwxPrrPywazj77bKAonAjZllnWe++9N1Ais1yP1ucjG9Tno5JyLb+HiJuB5+HIfNdcc02g7NPe9J7zTTJZlYrZ9O6fFPOqRMzfkv1HNuZr/R9vfetbgeI3mSz6YYsYVbbrrrsCZZdaq2fEOnFGL+qrOeWUU4Ci8vV/RZ9MZL6i7hgZRoUTKw7EfbdUibE/qCJ7jcwbhC1UqlaDrtoXK/p8jHKLY8RK/dqirb0d3+uGVDiJRCKRGCr0ReHENXSjg1xPjnkPMWJCRuJudAcccAAAL3nJSzq+7/q1DGca9ogZuMKRkUQ0nSPQT1Zf1efsN3vttRdQotGiYpHJahPZm78/8cQTgfH+jF7RhMJxjd01+j333BMovhr3v9FXY31AbaACjqpvuvf2GQaF47WZvxdXTPTZuUoQo16na9+cYbCFPjn/q+709bp6YN1AoyHNvzGPa6pzZyqcRCKRSAwVGvHhtP0egB122AEo688qn6OPPhooe9rbtgMPPBCAE044oeM4fcgIH7jCiVnQE5wX6H9G+DCwN/0ZRi2qjPVzGW12zTXXAHDuuecCsPHGG09rO5q0hQxV9a7/U7+l/gfztdrOC4yPQpvufjLIfuFKiYolqjnh2JHd6/u10vZ0YRjGSDc4n+yzzz5AyU20ov90IRVOIpFIJIYKjSicqvXkqv0qBoiBKxzrivVzr5bJYFFgb9ONqn46SFvEPX3iXi8xenGqiqYqw3zWrFk8+OCDzJ8/f+D9wgrY+vh23nlnoET2Ge0YfcXTjf/GMVKFVDiJRCKRGCo06sNpGj1E6gxc4QwLBhGlNigsynvATBdiVYeqMTPMtmi6fw2zLZpGKpxEIpFIDBWmdT+cYUNVba3EYDBsysZ9lMzr6Sema5fJPkaf9eW4TWK62j6sSvyRgFQ4iUQikWgEj2iFI1LZJCZCE8pGxMroUz2O6Bcb/29eFUhl0z+kwkkkEolEI6ircO4EbulHQ4YEq9b47iPZFnXsAGmLdjRqi36x8QplM9S2aBhpi4JJ26JWWHQikUgkEr0il9QSiUQi0QjygZNIJBKJRpAPnEQikUg0gnzgJBKJRKIR5AMnkUgkEo0gHziJRCKRaAT5wEkkEolEI8gHTiKRSCQaQT5wEolEItEIapW2cROhfhX2q1uIsNd2LOQ8d7ZarRUmeYyhKtEw3fckN5cqSFsUpC0K0hYFk7VF7WrRM2bMYNasWQDcc8894z6DMun5P+7LLrpVznUSFfF4VZNr3P/dB4vv+9/94NsePNNW66jpPTX+G6v6DjNyT5XERFhsscUAmDNnzkDOv9RSSwFw7733Ar33z177dy6pJRKJRKIR1FY48+fPX6BsfMotscQSQNlfpOqpFxXHSiutBMBf//pXoCiauDwUd0tcfPHFAXjwwQc7vieqlJPKxs/jeaaDjf437yMyXahigd6/9dZbD4Crr74aKLauUtiDQL/3rel2vJkzZwKlr8exlMqreYyMjEybsnHOve+++2r9Lq5KTRax3zzmMY8B4B//+Eet46TCSSQSiUQjqLU9wcjISGtsbGycIvDp9+hHPxqAf/3rXx2fy/qjL0X2JTPVp9J2vo7fRVQpoarv+7mI7W+1Wle2Wq31J/zx+GM9oiniIByis2fPBopSlsWpgFdZZRWgKFtZvOvR/cIwOIeHRTkPgy2GBf20RZzLfF31vv1j7ty5dU5TG3HuFpO1RSqcRCKRSDSC2j6cefPmjXvaCpWNiE9j/8encJWyif8ju/O1T10Z7wMPPDDh8bpFvQ3DunaV+tNv5ZrtMLR1uuF9O+mkk4Diq9EGf/nLXwDYbbfdALjggguA0n+q1seH6f7WRWx77MNxbP03wnG/wgoPZzTYTxYlVM2V+jNV+86dzgN1/ZVxzu4W8RvRLbK4G1LhJBKJRKIR1PLhLLvssq1NNtmEM844AyhPXdfQY5SQr4XsTCYqm7cNT3nKUwC4/fbbgcLqVU7+3jV8f7/00ksDheH8+c9/BoovIEa1yRpk1L34cGbMmNFadtlla0dpdIN+DCOwfK3/Qlt7Db/73e8AWGuttQC4/vrrAdhmm206vl/Xz9HEWn2VT837InNVwXgfjbS57rrrANhjjz0A+PWvf93T+buNgSb9FrFN5k3IbCOjdUwsv/zyANx1111A/5TwMPpwtNlRRx0FwPHHHw+Mj2KcbtS1xcjISNf7Ef3dT3ziEwG45ZaHUwSd6/zepptuCsC1114LwJ/+9CegjBXn2njeCfzXk72UCX+XPpxEIpFIDBVq+XDuu+8+rrrqqgVPNVlzfCpHX41KZ9lllwWKglEhyeLe9ra3dXxfRSKLP//884GyRqvCUQW8733vA+DHP/4xAFdccQUAT33qUwH4+c9/DhTlExl2HcybN49//vOfPf8+4uUvfzkAr3nNawBYY401gPG5RtrMtj/zmc/seL3uuusC8Pvf/x6AVVddFSiqYaprsNOJGLW4wQYbAEWpPPvZzwaKmlPRvOxlLwNg7733BuDzn/88ALvvvjsAf/zjHxd6Xm21+uqrA3DDDTcAg40E8756rY4d79ehhx4KlLFywAEHAKWt22+/PVCu6Stf+QoAN998c8fx+h3FNAhoE+eBI444AigK+XnPe95gGtaGmTNnsvzyyy+IuOymLGyz49bVG+/fc57zHABuvfVWADbccEMA7r77bqDY4j//+Q8AJ554IgB/+MMfADj44IOBsnpQd17oVTmnwkkkEolEI6idhzPROmQ3Zqhi8XuPetSjAFhyySWB8nQ+5JBDgLL2utFGGwFlXfrb3/42UBisSul1r3sdUBjyBz/4QQCe8YxnAEUR3XjjjQBcddVVQGELbTHsjeXhyOqf9KQnAfCrX/0KKKzMz3/zm98A5RplJLK6n/zkJ0CxqczJa5Mx12UkTfpwrM333Oc+F4BvfetbAJx33nkAfPOb3wSKQtWHs9xyywHw+Mc/HoAf/ehHQGGHN91000LPq8/Q/lTVf/thi1hVQ/WvevO+uxa/ySabAEUBe62Pe9zjANh6661tK1Bs4xr+F77wBQAOPPBAYPy12t/sL64CRAyTD8exsswyywBwySWXAOW+eo1Pf/rTgXKfp0vlTYctqlaH7KOOc+cJFYo+2h/+8IfA+OocKhwVkP1LBeRYc/VoEu3vaF9E+nASiUQiMVSonYczOjq64CkXKw5U1cByfTDWd5Kt+/T1ab/aaqsBxX8h2zJC473vfS9QmM2KK64IsMCnss466wBw2WWXAWUddP31HxYvqolBVWwFeNazngXARRddBBSWHxnEVlttBcDf//73CY9jdQdtaLTKC17wggmPN4zwPp9wwglAuc9eiyrOfiPbMkLw3//+N1Bs5f2Xzd15550Tnle22CRknvZ9r2XllVcGCiNdc801Adhrr72Acv8f+9jHAvDCF74QKH1cNedqgv3C8+nvMpLL8whtG3PYhhHOE/YTVz6e/OQnd3yujY3wGgb/ZVQKC1HVQPHB/Pa3vwXKPHH66acD5ZqiItUn6Pdj3UtXBSarcKZrHkmFk0gkEolGUFvhQPdq0FU+nbh26tqqzMTIGpnrhRdeCJT1a5/mRmYZqWF7/NzYdNnf0572NKDkqsSaXU3CtXIjq1yjF16LEVpVyibitttuAwoj2nbbbQH44he/CAy+BtdEiFFqsvYrr7wSgCOPPBIY328iO5TJGpmnDVTUroPr+xmk6lNBOEbsszJN/YxWWTjrrLOAcm2HH354x/fMt/F4RmReeumlHef1mlVIUeHE7w0jZOmOjXPOOQcoKyux8ojRqtp4GK4ttqGqNln83HnCVZ077rgD6O5bUeEIx8TFF19cu+0TYTK5Re1IhZNIJBKJRlBb4bRarXFPVSOkjCaTbfk0jTkF+k70pejPcO3d4/3tb38DCnO5/PLLAdh33307jut/8ylcs1XhyB7PPffcjuPJGppct5axyMKFtnzrW98KlCilbnjzm98MlLVZ1duHP/xhAL773e8C9fetaAL2I1mbkXg77bQTUHJKXL/uBv0XRiua26QtVAWD2C8njhn7gYpD9WVfNrLSMaIaN8Iy1tLy+H6u6osRohNU15i2a+w3nvCEJwBlPL/kJS8BSmSffi+jXl/xilcAw+G7iYi7ElchVhixHzhPuFIivK/ObVXVPGJNvrrIHT8TiUQiMdToyYcTn5qy9VjdOUbixMoArr3K8oxKMqtWlm91YCMrfHrLAl3PtMbbd77zHaCs3fu52djm+Rjd1CTLM8PXiDoZjgxX/9OXv/zlhR5HWx533HHA+HviWn/dHQGbhHY32uj5z38+UHx0MtdukK15f800t5+dfPLJQFHAg/BnxT7m2FB52lbZuJF1RjHah6vYuse3skT0Dfo7j7MoKRuh38mVE/OyrB9o3p7Rh7EK/TBhsqrL2mgqHX265qY5xzkHOy+o5iNc5el1VWeqyjgVTiKRSCQaQW2FMzIyUrnTZre1cZ++Pq1lsObHuD6p30GloyLR12OOgbXSrCYtk5UVuO+2auK0004Duu9A2k987GMfA4qqM0PcLGijR+IePmaU68+wSkNUNtr+q1/9KjDceRXaPdYRU8lOVolYd84qwTJb/SE77LADAD/4wQ+A4dg9075nW6PPxfu60korAYXZdmOYRmTGNXpVgWNnstGPw4ioXFQ25ixZacKKI9FfuijBOUl1Z9SZ9zPmmFmHsKo6tHNkr30/7s6cUWqJRCKRGErUUjgjIyOMjY0tYI6yLZ/C3ZSCT0VzRDbbbDMANt54Y6CsV6+99tpAqXzs+vYWW2wBFBXw05/+FCiZ6PpBVEquV0fEJ3KTTFcbuGePPptXvepVQPFXmYNiRJ9+jrjnkPdAf5Tq0e8N81q91+B99b6bS9JNicj2dtllF6DkdcnirUjhDqLRTxJzWAaheGI+jmPLSCv9ETESNML77apBVAEy5FitYZj7Rze4gmGVDaMTVcrOB8OIun1OX41jQ1+ditbIXPfP0Yfn/T722GOBUs2jV8R+lVFqiUQikRhK1FI4rVaLuXPnLniqxR0949MusijXHc2vsaqr71vB1Cgl94743ve+BxRfjLkJrlcarRb3duhWSaDbrnhNQJ/M2WefDcBrX/taoLBxfTUi7jUkS9ffNYgck16h/VUm+ptkqmaKR8SM88MOOwwYH5GnLa1koa9oxx13BOBLX/pSRzs8fz/7QRwT9tXoc/G1iuWUU04BSn04EasM25+M8LTigJUK/D/Za5wOJTTdPlLbZL9xr5eo9odZvdUdn/btd7/73UCZQ52DrSLuXCqsU/j+978fmLrqcz6qu4OwSIWTSCQSiUZQO0pt/vz54/ZcEd32yY7RQ8aKyzB9Gvu5TMXPrXTrurbrmpM9f8wT8nyDZEIyHVm669JVu5HKUIzEkdFGxiTTGeZcBO+zEXpRyUbI+l/84hcD5dplf/E+qlyNWvR+G/Uomqyp160Oofdd316stmBUksexSoP32X1T9F/FXW6nq711MN3Rn9pIv4TXru9m//33Bwr7fyTscqoNzePTP+1Y8dqF983qLbHOZVTG3aAte1U2IhVOIpFIJBpBT/vhiIXs/tbxOkazff/73wfG79lgdFGs+mpWrfkWrmPLWM1pkcl2a1dUQoOM2PHcMlAVTlVbtH83xjrMykbIvlwXdlfK6667bsLvu6fLfvvtB5SKEh/4wAeAaoVrXo+RPNdccw0wXAzYNjsGrIRudOJb3vIWoPgr3R3XnT8dW0Yp6reynuAwVUyeKrwG/VFGaql099xzT6DMG8OCujkrE8H7b2X1uCeZr9/whjcAZV+sU089FSj9yfkj7gnWrdJ/9DXW9UWlwkkkEolEI+ipWnTVUzpWh46I+7i7Zu86tb+zEqr7XXi+mJMgi1cpibqKZZCsTz+UFQdEvAYZiXsGvec97wFKBW3fX5Rw3nnnAUXhen/jLoTawp1gVYEq27iLpjY1J+GXv/xlx3G67bY4CNgm265/IlZWd4xZF06fzdvf/nagjC3HiNFt+kMfCdBG1qGzaoIrHO973/uAovJcCTH3bVCYjnnGPvu1r30NqN4pWD+lPpctt9wSKD5Ax1isOBGrSccxMtVrSIWTSCQSiUYwUueJNTIy0hobGxuXOxB9Ln5u5ITM07V635e9bb755kDZ5VCl41q+jNaqsGbZejxrr8Us7CpfUtXr+fPnX9lqtdZfmA3afjstskhl8sY3vnHCz2NUiGv57l5q7tLtt98OwB577AFMPSO41WpNHCY3Aerawmv2fuqP+PjHP97x337kHi9ec8zOd7dTK0/suuuuQFm3dh8Vq1FHdMv67qctYh80n0pW7v137d0xtvzyywMlulGfT8wwd8+YGJk3WVjDT99BP20xWajyrMagyttmm22AonwOOuggoNhiuqPl6thixowZraWXXnqB0uw1N8m+qh9bdS+cM81RijUZXQ3Sf+mcGft+1VxZVaNtsrZIhZNIJBKJRlDbh9P+RI65Az7tVDDuwHfTTTcBRbmY6b3ccssBJZvaWmpWf5bJyPKM4DG3wJwUM8R92qucPJ9Pc483yMoCwppIsnER98eRpQsr46pktKWRfzHTOEahDBLeh49+9KNA8UMYbaTSsYae993KE1URMkYv+t/7Ktuz4kBklbHK+SAQK5abX+HYsI0yUfNu9GcZreSY8zhmpEd/mOim6jyvymaYEPNurBhhX7ceoSpwGK5h3rx5HSswvaot75c5ieblONcdeuihHd+PVV5i3Utt6PvOR1//+teBMoe7GjHVMZMKJ5FIJBKNoKc8HJ9uVZWLfXrKUI0J9ykrOzM6zTV6Wb1rsOZpyOqOPvpooKzlu3udzKbq6ev7cT/3QebfmBPif21h21UwZ511FlD2ApKByEjMw/Ba9GtVVcoeJGL9t0984hMAHHDAAUC5pp133hkYr2hink3Mmo7qMEY5RlY5TDkpMRPcMaBPxjHl/09/+tPA+AhN77tjz++LbmOkbgb6IGBbjT7UPxV3xTSycxj2P4KHber9mupOvPruVO9G7sWos7j7sn5MV070/ekD/ta3vgUU/5irSVU2zP1wEolEIjGU6CkPx6emPhN31DOS5pJLLgHgda97HVD2epFhqlhcu5ehGCOunyJmhBtHb1SSkTwiKq1Y4aDqe5FJNwFZgYxH9n/++ecDpRK2tpCZbLfddkCxucxGxmT0yjAzVFnYGWec0fG+9b9EZNved1m/bN5dMT/1qU8BJcPcauPdUBUxNB2Z4ZNFlQpTlTsGjNg0/0J1b/TTTjvtBJQ6hX7eLSpqmPtLFVQwsnFtZOX1lVdeGRi8smnHVJWN0PcSK/c7p9o/2nfmhNIv3IPK39nP9INeeeWVQJkb7T8TVWtJhZNIJBKJoUNthQPjfTQRrgvKvo855hgADj/8cKD4coxKc83dmlcqILOl9WNUVXeOr2OeRmQBsf7QdMfnTwbawLYamWeEjTY488wzgfF73cfaRq5fy1QWBaiErRatDVRx5luZa6SiievJ7osjK6uLYWb93m+j1syrsUqwysYcJBWQEVFV9QPFMFxjXXgN9p+4L5crLe635WrBoDE6Olq56jJZ+Ps3v/nNAHzoQx8CSrSqY6IqOtXdUa1LuO+++3Z8Hiv6iyqVmLXUEolEIjGUqF1pAMbXRPMp55qpa+cyj+233x4oGeY+pfVD2AZ3KbQ+mJVur732WqD4eGJUWlv7OtoT3xdGcvkUb1NOjVUakIWZWW5VVxnIEUccMWHbvbbjjz8egDe96U0d36vLnGJ8fpv6G3hGuVhYhAz0n6UPgy1UvNZUixUJVIlGhPq9tnYB3auQd2Osw2ALo1TdIyiOEecJqyT0q3J6L7botc+uvfbaQPFjXnjhhUDZ+dP6cc6hUeE4FzvOXT36F4UwOwAAIABJREFU2Mc+BhSlE9sVK6pXVVjPSgOJRCKRGCr05MMRPuX06biuHLNbVSrWSttss82Awsa+8Y1vdPzO45qJbjSKTGUhT1lgfPRZrP0WfU+DWMdW3fnfbOgddtgBKMxVn4xt18/lNUy17YvCvjkLqXHWcEuah9FI9nmz531t/9G34+pC3byaYYrk6gZ9eVabf+lLXwqUCCzz94axf/TaJtWc2GqrrYAyfp0Pqu5j9PkaYezv4lwp4uup7h2VCieRSCQSjaC2D6c97lr2FaO+2r7f8V/FI2u3yqu5I0ar+dSuqkIdcxB8Ovs717tj9ejYLtvj75r04XTDoHejHIa1+kmcF3hk+nC8NjPJrZ2l709mat05I/wcM92iFXu13TD1C+ePjTfeGIBvf/vb/TzdOEzFFtH+vVaProvJnjfO3VE5OXe2Wi3mzp2bPpxEIpFIDBd6ilKLT0lfmwEe/Q6xDpSKIj5VY20svx9zTyKiUhFVisvjery26xkahTNoDBOTHTSm0xa9KgvX3N0Z9sgjjwRKzlvMu+kXhrFfWMnkpJNOauJ0C1DXFjGa7v+OMa1tGhRS4SQSiURiqNCTD6ftNTC+NpmISiX6esyHcT26at2w6vexHVXXEncerVq3TIVTMIxMdlBIWxSkLQp6sUU3X81C/MuTOk/cTXmq+OEPfwiU6NmFrCalwkkkEonE8KCuwrkDuKV/zRk4Vm21WitM5ouPcFtM2g6QtmhH2qIgbVGQtngYtR44iUQi8f/ZO/NA28r5/7/OvUWGJpqURGnSIJUiKZJ5LEITyZBZ9BMlGUKSqUnJXDRpIJnKkEqSBppEgwhFAzL2jXt+f+R1n70/5zxn7bWnc24+73/O3fvuNTzPetZa7/dnTCT6RZrUEolEIjEW5AsnkUgkEmNBvnASiUQiMRbkCyeRSCQSY0G+cBKJRCIxFuQLJ5FIJBJjQb5wEolEIjEW5AsnkUgkEmNBvnASiUQiMRa0ajEdG7DdA3Fri9I2PZWhF+Oas2E1JeunMOGoGqJlM7rBMZvr4p6KnIuCXuei7QuHxRdffOGNH7vA1Rb1oIvdPjmxKnXtod50nBnOp1Wto5levj4khdVVm+ai9v+xV1BtO48bu6E2zU2c437gsWMlWdFU9buGZZZZBoBbb7216/umdRH7NcXjNf1/v1hsscUWVgOuHaO2hvud/1onR/dnr6p4beK1mK5ni+h3fmKF5F6vT9v9NqE2tgGeF4mW6KsBW8dnAO573/sCpc1AW8RGbfHhLNoujD4e7kNrT1A7dtPNV3uIDvoSj/ttehG1ZW/z5s2rEpCOffa6y2nRa+l1W5A3tVmuIc5VP4223Id/JWlNL/y4Dlwnfo5NDpvWetN6qn2O6FgnI1e+vf6+13ul39/XthOLssIZ9ks02xMkEolEYk6hlUkN7n7bR4agsonsPTKDaIt3+ybmWnsL3//+9wfgn//8JwDLLrssALfddtuM28X99vu2n86k5phtLvfXv/4VqDeBW2WVVQD4zW9+M+051M6pxkhrSihes9jczu0GNenAVLbepICdG3/35z//uev/o0poQk3ZeJxeTcL9+Cs1O9caZ8V7RDV25513dp2D5+Ya/9vf/tY1tpoijojfR1NvNP2JuL4WW2yxvnxo8+fPbzSlxWP2Oue93isRbc3H8XfTqflFDbNlHkyFk0gkEomxYCAfTs0PoNL405/+BNTbkg5qR6y1pI7nV/tedJx/Kx9OJ9NZeumlAfjLX/7S9buo3mS497nPfQD4xz/+0fX7YTGnprldf/31Abjyyiu7ficj7cdv4T683tFvUfMjxetokIBKx/0MGtjwhCc8AYAf/OAHXecV5+ohD3kIADfffDN33XUXCxYsaDUXnWqgxuJVLqo+lUdUolHd+XnFFVcE4JZbbgGm3lsPeMADgHIPun+P7/68VjHIZDplvGDBgqH4LeJ8x3uk9lxZddVVAbjpppu6xtC0f+85x6rics68F7WU1O6dBz7wgQDccccd/Pvf/269Lnr97Wyi3xbV6cNJJBKJxJxCa4Uzk/2y1wiYQe2HtfBaWVzNLh0xDWMeWpRajFLyGMcccwwAe+yxB1BYln8f9KAHAXD77bcDsNRSSwFTQ4Ldnz4AGUlbO/gMdvTWCqemNGvHiCx8+eXvToFyrCqdO+64AygMteajaRtmHX1FEa71QdReE2SUUclEpbPVVlsBxc/1sIc9DIBtt90WKNFrv/rVrwB41rOeBcBPfvITAF7zmtcAReWdddZZQPExer4xxLjzXpucnOwrerHfNTksP0OvkZ69+kMHidi7pyIVTiKRSCTmFAby4XR83/W3poBiRJSfn/vc5wKw9dZbA/Db3/4WKL4g2Z/M9vzzzwdKZNe1114LTLXpxrFpm//1r3/ddb4dvx9Y4UR/lceQkRpttNxyywGF4V5xxRVAUUDbbbdd1/87p86BkXinn346AIcccggAv//974Fme3RTxNYwWH0talGWrf9I27y/u/zyy4ES6bfyyisD8OhHPxqAE044oev/XSf6M5zjeD762fSbHX744QC89rWv7fp9p02/X79FbR2IWk7IGmusARTF+qhHPQqARz7ykQDstttuQFGDzrHrwjE6l17fl770pQBceOGFQJkjfY6exy9/+UsA1lxzza7z/a+6GQqrj36CQfPtRK9Jzk3oVWEtCgrHe0RLiNaEYfqK26yLVDiJRCKRGAuGUmmgxmxj/o12ZiNszj77bKDY1GU+kZ3LxmRv2vSPP/54oEStHHXUUdNuXyt9EaOq7rrrrr4VTq+Z3kYnLbnkkl1jU+HI9nvI+O7a7+9+9zsANtlkE6BEP5nf4TWpRfb4/xtuuCFXX301//jHP3pmb/PmzZtcYoklFs57ZPcqhnguj3nMYwC48cYbgaI83va2t3Wdq4r2F7/4BVAi7D760Y8Cxa+hMtavEf1eHlelpEKW9Xl+nb8fdjSS1zXeI86R6k/fywEHHAAUxbHBBht0nbvbn3baaQDssssuXfu76KKLADj00EMBuOyyy4CyPmoleGKlg36jFzsj9mqqTtR8vTGq1OeJ/k6V0sYbbwxMXTf6qfpFPB/HNEyF09ZvFfO34nZPe9rTgGIBcf8777wzAF/+8pdbHa8JqXASiUQiMafQutLATDkG8S0dlYvM9Q1veANQmIlszM/6J4499lig2K0vvvhioLy1ZTiXXHIJANtssw0A5513HlAY8zT+ia7PtWKTTZg3b950dbcW/h/AQx/6UKD4jVQBsuz9998fmKo83J9/PUfnMqo31d+5554LwHe/+10ATj31VAD+8Ic/AHDVVVd1HScy7ssuu6znYohicnKSf/3rX9ViirIwx7L22msDRe1ZbWHHHXcE4POf/zxQIrOcO1m9c3vBBRcAxTenOjzxxBMBeOYzn9l1fK/NDTfcAMAf//jHrvOKuTD9rouZqi7ITKMK9xy9Ti95yUuAoub9vT6WSy+9FCjRZq51r69+LRWRtnwjA2P+V2dFASjWCJVQvzb/zrWw6aZ3Gw9UbypTz8W5cR05J/o7xeMe9zgAVlppJaBc/xe96EVd+/Xe8jnicaaJNpv2+xpGUbW8dsz4jNWXp1KN5+xcnXzyyUCxNjiXe+65JwBnnHEGMHUdjBqpcBKJRCIxFrRWOJ1v9yblIDuSbRkpo81cfOc73wHg6quvBuCII47o+t2nPvWprt/LGlVEK6ywAlByVx7+8IcDxS9SYzKDYsGCBVV7tCzt+uuvB6aquO233x4o+RHaXA8++OCu36kK9fW8+c1vBko0kyzcuVMBvexlLwNKxJeMONrDB2XzYnJystp2QLYcWbrRhUbYeZ1VHtqZZWnu3/9Xyeyzzz5AmXMVVGxr4DW6+eabq2Po/NvZiqMNJicnpyhQPzsHtWOqMLx3jNDTF6PvTxX/4x//GChz47rxHlDNuV0cT/S3+f/O5aDVHTrXgsrGMfrXY6h4PZb39Qte8AIAXvziFwPwox/9CCg+QM9Z37Dr7UMf+hBQ1plz5f97bfTx1HxN0/mexlVLzfmzyoGfVYtGHQrnSJUnHLOKWTWZCieRSCQS90i0Vjgw1d4ba2bJmvxs9Mg111wDFKYj+9fGauZ3Z2QMFIahHTLmNshQVBGx2VSM0BplpVTPKZ6jPhbHcNhhh3Wdi5UHjDYyN8T9aIM3imndddft2l4/iGrS743UiTXe4vkOOifTVROOrD4eS3/CQQcd1LWdY3N9Gc3oXLhO9Nk4p7I219973/teAN74xjd2bV9D9EH9NxJpxm2mw8TExJR91fogdW4DpVKA5+rvVa7eOzJb17pjVxm5vcxY9aAK0BpQq2rt94Mw+YmJCe5973tPqXCt+tJf5X0rNt98866xeA5e56c+9alAYfHOQedxO//uvffeAHzrW98C4Oc//3nXGLWERJ+eiBF7s1Fp2XtB60BUyvEZ6dx53T1n8wHb+mmHhVQ4iUQikRgL+lI4olb9OfZakX2bWW602fe//31gqrIR2nBjPk4tk1h7pHbKWBm3hn7t1J0stcZ6/F5GKSvXrixz0TdjJJaKxDlcZ511AHje854HFBYvZK6yRv0Yjl02KNusMZzO/iVtMN02tWrAwuulr26jjTYCSlTaeuutB5Q5Ur3JRE855RSgRHRZd87jbrnllkC9rl2seFGLvuwH8RhNfWdiLTXHKqN1bOYsxXwZFbCVJpxr14m11dyv669mw6/1sGmDyclJ7rzzzin70N/oucT8rNgB2LVvBKZ+KSO1XCeu+Sc/+clAiWJ1/RgZutNOOwHFbxqrLYhorYh15cYJz7Gm0j13nx/veMc7APjIRz7S9TvVoM/ScSMVTiKRSCTGglYKZ2JiYsbOf7UseJXNXnvtBRQGG6NVtEPLLFQq+j8iq5M5y5Di90097QeNwIHmygKes4zTscd8mp/+9KdAyUHSvyFbkw3KZIWsy0isDTfcEChjjz1mVFqeT4zE6de2OxPjizWz/K2+N6+7eVTmU+iv0g9hJQH3Y/6FuSbRdh/9FLFnUVQdtaiktuisK9erQvBYT3rSk4CST+F6Ec7FddddB5TrHv2TrguV7Ve/+lWgKB5zWLwG8bpP1xenLaufmJigswKFcD8qW9e6lg77FbnmvV5aRr7xjW8ARZ2Zj+WcuU5irca11loLKD7jmNvmHNSiLWfL7wHNzzJx5plnAsUXHNeyCjj6/MaFVDiJRCKRGAtaKZzJyckZI31iNWjZ1GqrrQaU7HdZkzWx/L1M1RwDmY+ROyohGbOMI+Z5+Lmpz0XMMG+bc9HJgJpqn4nIqmUasnn3Y0axkPk6x/G4q6++OjCVsfj/KqdaRW3zNJp6xPSDmmKI6ks/g0rELHpt7droPUfzcLx+wvVh5QLHrOKJSiYq5UFzLDq3j6or+jk9B8dgTTzvHf0O+i1UJF4v/6oWvJ5aB6w/FyO5HGstWi4q3nvf+95Tas01YXJycoq66YQ+u5gv5/XyeEbmWTMx+pf863bmnD3+8Y/v+v9vfvObQPH9RJ/wuHJrBkGTYlYVOhded58bz3/+84Hx59+IVDiJRCKRGAtaR6nNZJ+ONk7tjkZO+FZ93eteB8CnP/1pAHbYYQeg2GIjK5fFGVOuHdv9G5klm4rx+zWWOWjfjPnz51ftvR5DhlFjh45B9m79rwc/+MFAYWFRPbp/t2/qQR5t8vFa1fJ0esVMnWCbfCMqG1m8c2YNPX07sn6vv1GMce7/3//7f0BhxDH/wwgtv49RSIPmW0zX5dJzr9XM89xkqGbXG+FplJr+C6MSzSlRpTkWFY7WBe+lqHRrqjMqnH79F53rIt5/KpF47Ki+9DsYoWfeTqyKYN1CfTb6P60sYc3FmJ0/G3k1bdEUOen/Wy0++sOdq9jhtbafUc1JKpxEIpFIjAWtFc5Mb75aj3AZpVnvvoX14VjhNuYGyIxk/XY/lMVZVTgyqJgHVGOskRW2xX/+85/Gnh6yNRmuSiRWYbBPhazcubC+mOzO+mL6OcxJUREZoRMZabSlx9yTzqrG/diyJycnG6O7YgSV82+NKxmqY5Gh+tn1ogqI+7UT6Gc+85musYk498Lz1WeoCpiui2mviOugyT+oFcAxG21oLon7MbdIlm8EqHNj7TwZruvE/akqrPIRa+tF1RerhrRF51poqs4cv/c54Dk85znPAeCkk04CSgSe1gH9mKpF1aH3lBF//VYc6VRo4/b3xHP1+mvhMHrV2ouxSrnn25R/M2i0ahNS4SQSiURiLOjLhyNqWdPxs8xCO7Rx8EapyO4iu5LdyXRk/bVeMZFB13ILatu1xXR+i2gDr/kJZF9GIVkx2ez5mIv0jGc8Ayj+CefUiK2nP/3pQIncirkmouZvGzRSZ3Jysud5dA5iR1Dn7FWvehVQ6j5ZO8vcETPPY9dM87xqETiqxFghWXQqG8fUD+bNm1ftWlmrpSbr9hxULssvv3zXdo7ZWnwqXPNyhBFaVrYwR8XjRD+nf50b/6qMh2HTr1kDIvxe9u5ceP2tEu2YtJxEX6/PGxWykXxR4fR6vQf1+Q4TXhcjPPVzOjdRudZy4UT08Y4KqXASiUQiMRYMxYdTe2tGRqB92fpQ2pe1xcq6tO372Sx8cxWMOolsPObf1Ni6+xnUTjnd/j2HWhSQx1TVHXXUUV3bxd/b9dIIPqNMtMXGitlHH300ULqqGuEj64u5B7JEmfU4ECO3zMMwk9zrb9Sa7G3rrbfu2i7a+ptyiKLajN/3m48V0bmuanku0e/oMb03ZONmjltV4fDDDwdK5fUYAagicj2Y3+W95FzFdRoVeVQ2g/izRNt5dh5V+d/73veAEqEnu1fNy+KtNi3M43E/tVy0iFFHbA0Drnl9eLVnnh0+a2OJSndUSIWTSCQSibFgoH44NZtmLVrNt6fsKTKdWC8oZtUbbRJZmexPxqMCqjHeppyVfmBUkGOUnfm9rNoxxx4hsfaZ22mfVokYbaQqkLU99rGPBQrTUQ3o06n5NWI3zEGYbBMjjNWjHXP0H8S5MEPcvjbRH7Lddtt17aeGJpv9dP18+q2SXIsSqnXJdU6sDHDggQcC8MIXvhCA448/HihzERWS1gLXjblpjsnvXT+qR49vZ1HP2+0H9XPCVItCk7KJzwvXg3/1WxndanSh93/czudAkwJeFBRNDc6FlgzH4nXec889Z+W8IlLhJBKJRGIsaKVw5s+fz9JLL72QKTQxWVFjDr6NZVHmrMi29D/o44nMSDv0tttuC8ABBxwAlMx12Zr269gnZxiIiiVGe0Qlo0/GLpaRqcpEZYNG2lhja9dddwXKmGN/d3+njV+bvnNQi8Ov1Z1rg17n1XOJCjYqTxWPzDVGGerz0b9Vi0qMMI9Hpa36GwbDnTdvHve9730X7jNGqdWO4dr3e+u+ea/ZOypGO3oPmcNknbA11lgDKOvCsbo/q0c7F/qMYqfPzsjPfnyendv1Oq81C0pUjbF/kWvdc//Zz3628Byg3FvxHojHq0W5zkWYf2NOYlSHPm9UhW0xjIr6Xfsbyl4SiUQikWhAK4Xzn//8ZyHz6gWx02NkEkZGmRns97EDaK1ng6zO2my+7bVzW3HZnJfIbIbR1VCWZS6JTFK78hZbbAGUSCu/X3fddYFSScAcAvNwjjnmGABOP/10oDBZ/Rj6dLweT3nKU4CigGSqMpvYTydGBNay8AdBE3P0rzXR9Cf5exnrF77wBaAoHtmW6i5GBHpcxxorYfu7uC5j3bt+fDgLFizgb3/72xSlGBWD/++6iRXQvT7WGxSOyevrORv16P87BnNQVIOuA30+qn63cz049+a+9Yt+VEJTTT7hevKeMm/LdaL1wU6hTdFoIvoIo6+5X7U3Cmy22WbA1D5ZQotHrxhWPcEaUuEkEolEYizoK0otoqYQan0r9MnIWIyw6PU4voU/8IEPACUyS3Xx9a9/vWv/vuUjKxnUZj8xMbGQDemrkQ25T/0L1rCSfR166KFA6T2u+jLvQrhffT4yT6HNPuYq2EMmMlSZtBWX3W7QOlEzdYIVNVaosolzd9BBBwFTc5TMuzC/Il4/fT52eDT/y/OrKSLnOuZ3tMV0fjAVqvkxrgOhClORyNatBq2S1acn4tw5V1awUEGZu6Jfw+O5Hlx/zo1qcJg2/Fo33EFZdKykrjq86KKLgKnqsem4cR372conRsvOBdjZM0LrTtvrVqtAMiykwkkkEonEWNBa4UzX36IWlebbMtZnsqqrTCFG6EQ7t9ElsjLZu/XDPI5Vgj/5yU8C8IhHPAIo6kDIiHqt5DsT4hg9d6N+ov9AZbHeeusBxaYuA9199927xvbsZz8bmNoB1OPYH0WGan6O+RxGs9knXsQOkTG6pS06mVG0eTfV0Irs3H4m9oSJeTtevxiVJFwvZu27vlwPRvBFv5sM3Gs1jIg99yHrjpFWjmnfffcFyrrQNu89orJxTvVT2iXXiE7nJHaQ1Wfz7ne/u+v4MXquVhmhc5s2mEn59lrLrNbPyntJJWtVecekb9j7Xz9HzPfrVXHNJWXjGFWs1s5T8fgcmGtIhZNIJBKJsaCVwjHHQDZd650QmYt2aVmXVaJl4UbiuB9ZvWzft7fHk5GqKmSkVlCWwV599dVd5yGGGWESVVL0T8W5kl3FKKOo5ozAs2OjcL/mZRx55JFAYfHmXdhjJnYOldWrvGTGw6qTBWV+e7X/a2OXgcYeMG7vXDl39v6JvYaMzHOM7sceMOZ5qfJizsx0Y2qDefPmTYnyiT6cWF1B390uu+wCFOVrheQYUfemN72p67PHixUN3G7vvffuOp6+n1iLLV6zWkfbXtGpbmqdX/vdt762PfbYAyhz7FxYVVxfX6+19haligP6GX0m6hOeq0iFk0gkEomxYKLNW3xiYmLyv3+B3hmANnfre2lL97N/tVPLZIXHkcVr85fJvvzlLwfgrLPOAopKaMogn+b8L56cnNy0lzFNTExMduZp1CJwVGfnnHMOUDo6Gln3wQ9+ECh+ilh7S0bs/6sGP/zhDwNwyy23AFMz0Jvs5rVsaqPUJicne3ZguC6E18dcIWvbqTj0Z0SfnX4GFchxxx0HFL+E60PGag+Y1772tcBUn1GsTtykxGeonNFqLnrx/dS60zpn1sZT6criVayyec/ZfC1t95tuevcyNofJmno1X11UOLUuuIOsi6ZeLL36Uvx/I/kcozUVVXFvectbAPja174GFBXQ1k85XW3I//Z/6nsuRBx79KnFyDrH5nrxer/+9a8H4Nvf/nbX/saFXuciFU4ikUgkxoK+8nB6rZUW/98oD1mbFQDs6KePJx5H5moF24MPPhgoNn8jsGQH0SY/Klts534jO5OBWO05fr7kkkuAEpW04447AoXBXHjhhUCZIzs8qnj0vWjrj9FxTWPvzJruPP9hMCP3rbLRJ+O51bqk6mfSF2fF5KhojT6zl31ULlG91dRerar5oOtlYmJiCkONTFYVHytmu8btX+J2RxxxBFCUj2Py3lH9ux//X2UsatWqY/WHXrtzNqHTChD3EaNFm3w6zoVj0lKiYna9eQ/p3ximf3IY++uEY9avbe+v2jn4e/3bXi/vBaMXxajWeL9IhZNIJBKJsaAvH8403wNT35rWOJKV21/d3vJuF+uCGXmlbV42WMv8Nh7fyJsextH1ueO8W/twYv2tuM8aw4jnYvSY+1MNiJhDINNR4TR17KyNOdqM/+u/aW2fnj9//hQWVkOTzd7P0fcSIwBrGeRtj98xjimf+/Vn1erI1WqqxSoLoqbOYq6TfkvXoWP0uOauxSoccT9NPr5h+C2m+V3XOddUmP//vve9Dyj3jP4tfX4+B7R86OfQKqAC0k/VFLU63Xodlg9nXBh21eeI9OEkEolEYk5hpAqnrf2w1jOk17dyZC61KsAzjLm1whG16LCYWd4URdax/66/TXPQ6zWQFcY+PdP0IOmbvfXLpmaKnOs8N6Pd9PnEMbSISpzx9x3btZ6LqGim2ScwVYnon3Ctxsi7eK4qG9VezPdRwXhPaHXQWhDzhWZaR8Ni9TW13asv2DVsjtKaa64JlKrxO++8MwA77LADUKLYxLDy8BYlhTNqpMJJJBKJxJxCK4Uzb968ySWWWGJK7SlZXFP/m7Zo6rzX5B/pNZqu4/ueFc5iiy02ef/7338hU4znGivYiph34e/j3A2KqKziecS+OJ129P/85z995Z7E6xAjsWYLTYor/n+n76gtq583b95kZ/2w2HU05rnUcnY8BxF/71+vo2qv5juqVUqO1ywq4EH8Fs5FzOnpV+FMs/9pf9+0fTxOr4p8mFaAexpS4SQSiURiTqGtD+cW4NejO51Zx2qTk5PLN//sHj8XPc8D5Fx0IueiIOeiIOfibrR64SQSiUQi0S/SpJZIJBKJsSBfOIlEIpEYC/KFk0gkEomxIF84iUQikRgL8oWTSCQSibEgXziJRCKRGAvyhZNIJBKJsSBfOIlEIpEYC/KFk0gkEomxoFWL6dkuQNdrCf4BcGuL0jb36BINWZiwIOeiIOeiIOeiYJEs3tnZC346/LdabfV7t2/azwy4p9Y6mpOYP3/+wmrGnYjXb968eVOqLveDAdZFz5g3b97CnjSJRKIbc+qFk0gkEol7LlqZ1EaFpu6I9wTE3vG9oqkn0KKMOKbpesd3fl4UsGDBgoW9aWYD9+T1klj0kQonkUgkEmNBa4XT2dmxX8hcF198caB0Mxx0v732tp8NtFU2YlFkqr12UKx1Ja11chwUbTtBLopYFNdL4n8HqXASiUQiMRa0Vjid7C8ywtrnuG3sny77N7rn73//OzDVli+0U9dUQ405zwaDHfYx9QXJZJdeemkA/vKXv/R0nEc84hEAXHXVVUM5n+kQlU2cA5Vt7fpst912ACy//N0R6q985SsB+Na3vgXAJz/5SQBuvfX5h1FjAAAgAElEQVRWAP75z38CZV383//934znF+cwnkdU3oPgWc96FgBnnHHGwPuCMkeuce+hj33sYwCsu+66AJx66qkAfPSjH+1pv8Mccw2eq9cr8b+HVDiJRCKRGAtatZhum7wUfSmPe9zjAHjUox4FwItf/GIAjjvuOADuuOMOAK699loArrzySgD+8Y9/AHDve98bgEc+8pEAXHbZZUBRRHEsssBemS9w8eTk5KY9jq2vuaipP8/Rc15ppZUAWGuttQB44QtfCMBLXvKSrt//5je/AeBlL3sZAL/4xS8A+OMf/wgUNt+WuY4yqS0qnI033hgoYzn33HMBWHHFFQG4173u1fX722+/HSgs3nVyyy23AHD22WcD8Oc//xkoimuJJZYA4F//+hfQ+9zMpQQ/5+7hD384MFWpOtZPf/rTAOyxxx497de5/sMf/jDj7+bSXMw2ci4KFsnEz0QikUjcczFUhRNZu2xso402AuDBD34wAE996lMBWG655bp+/4Mf/AAoDPeGG24Aiq/mgQ98IFAYqoz1xhtvnPZ8aj6C6GPoyF0YmcKZZvuuz8suuywA6623HgDve9/7AHjoQx/a9f/3ve99u7aX1TtXX/ziFwE4/vjjgaIG2vqxhsneYjUB18XWW28NwDLLLAMU9eYcrLrqqkC5XnfeeWfXflQ2+q9UdY79/PPP79rOMato9OHM5PNbsGDBnGCyKt4ddtgBKH6t9ddfv+t3nrv3yMMe9jCgzE0NqXDaYy7OhfeMzwXXuv4z7wXX/rBy3FLhJBKJRGJOYaiVBmp5FNGWro19ww03BOCSSy4BCiPVV6PPxbeztvx11lkHKG/x3/3ud8DUqKNoo6+x+tnIXYjRRvqvXv7ylwPFP3HzzTcDhXkedNBBQJkD5+rjH/84AGeeeSYA97vf/QC47bbbuo47Gzkmzq9jlmU5Js/117++u5TdBhtsAMDvf/97oKg0tzv55JOBwub+9Kc/AUUFrLDCCgA86EEPAooP8Le//S1Q5szv4/lFpTOb8Jy8F/R7rrHGGgALqxo4NsfunGot0O8Z1Z1oUjazgXg9XNsXXXQRUFT/IYccAhSV/6QnPQmAt73tbQCsvfbaQJkj181DHvIQoKjBRQmeu2v8+9//PlCelU1wTl0/m256t2Hnl7/85VDPMyIVTiKRSCTGgqH4cGoRWNrul1pqKQBe+tKXAiUa6Sc/+UnXfmT5Mg7Z/5JLLgnAaqutBsBTnvIUoDBdo9qiwlLhyA5rTLYDI/fhRN9NVINGpXmusjZt8auvvjpQoo9ksM7lD3/4w67992ujHaV92jkw78q/MlF9OptvvjlQIrGuu+46AP76178CheXJWF/72tcCcM455wDwrne9CyiKRj9GVMIRsR7ZbNrq9XedcsopAGy77bZAUa5GJ7rW9fHstttuQBnjpZdeChTWr7+0rbofx1yoVK+44gqgPD+ciwjXuM8N/Zx+73pxLlxfzsETnvCEfk5zrOvC6+sYLrjgAqAonJVXXtnjtNqv+1tllVUAuOmmm/o6v/ThJBKJRGJOYSg+HN+SMsP4Vwb67W9/GyhvU6OR9OHEGlxu7/632moroNihzb+JLM23vMpIH8BcqEodWXVUXfqjtK3KbLTJm1HuZ3OY3M5oI30/40RTgzzh/xt9aJSZOUT69r73ve91fR8j7pwjqy2okPX1qICMZuv1us+FemTO2ZFHHgnAM57xDKDcI15n/5qT9tWvfhUoEX/+Xhu9fpB9990XgA996EMjHMXM8NxUtvrw9N31ytZVNvovvM4///nPgRK1uM8++3Rtt+WWW/Z97qNCrdq3Plvzr/TVqQaFz0T93VqBvBesNKJq9FnaFMVYw3LLLbfQR98LUuEkEolEYizoS+FEJitT8a0c6zJZT8q3r1nz1sbSVuvb3eghbfuyeZmwb/VYk8nz8u1u9JI2W+3ecyH6SDh3nru1sN761rcChal6ztps9d38+Mc/BsqcqRpmY4y1Sg81ZaESMZ/Kz9dffz1QmKvrQBYmO5PdP/7xjwcKi9MOLWPWpq8Cms3aer3C/Ct9NLHjqQrlhBNOAAq7NxLr85//PADbb789AA94wAOAco+9973vBWZX4TjvrnEj8HxueB/H36t0/avlw+vrulD168Pde++9u/bn82qTTTYB4OKLLx58UAOipq6NHnOOvFde9apXAbDzzjsD8OpXvxoo91CM2LW23u677w6UZ2i/Vh/rGfaKVDiJRCKRGAv6Ujg1RigL8+2rMtFXIzu/5pprgMJgtcGrQNzO42h3lLl897vf7do+5rRYkUDGM8oKuINCZqEqdCyvec1rgFKN4ROf+ARQbPTapY3AktnOpe6YTecS811UOsKae5dffjlQWLrRaObrmG3vHMrm/BsVuGpQzAWfjfBcZbRR2ey6665AqSghYpdR189mm20GlLmLGEeV6Bq87vpW9Euo6lSoKpTnP//5QIlKVc3Hc//Vr37V9TlGywrHrg9pLiiciLimzT361Kc+BRTfnZGcMWJX6PuxQoXryrltqsA/LKTCSSQSicRYMFCUWqxNFqNKnv3sZwMlC1aloi/Gt7YVA2RhKh1t8r51ZcAqJXNUYv0oGeyOO+4IwNFHHz3IMMcCx7jmmmsChXnssssuQLEzm6djprk5BHPRDyFqLNpzVmFos1cRyzz963rYZpttgJKX5brTR6jqk8X5vWrRrPqodOaCT8cxxgjNz372swCceOKJPe3H9fT6178eKLlJwrnxnpoLVgD9E0bS6ZfQzxCrvfu8aIqwUhVGuD8jPeciovr/wAc+AJR7wLF98IMfBKZGufmM/cpXvgJM9Yv57B61shGpcBKJRCIxFgy1lppM1rem0WdPfvKTgZJvo4/FKDLtktpsrSptRYJYMVlWL5M1y14/hm937dy+1X2Lx7e5kWHG7c8GZDJmlB977LEAPPGJTwTK3BqN9NOf/hQorH028m56hey5lo8j29ZGL3tTmXi9XFfaqZ0T862sGu71NjpNhmxUU/T9zeSTHJdPzEg8bfHReqDtva360nYf8aUvfQkYH7OdCUbSOdb99tsPKD4XLR6ydX+nsvH6x6ri9gSywkmEz5+5oGxr8PqoxvVLm5dlzpL3gv5vn5VWJInrwLmymse4kAonkUgkEmPBUPvhyEBifR/f0r6FVTK+ZWPmuQrFv76dte3LiK0r5Vv+wx/+MADPec5zgKKAPvOZz3Qdb4aaaj3XUrvXve41udJKK1V78fQLx/azn/0MKGOOVYOf+9znAiWXadhoWydqYmKiNUN0TOZJveMd7wCKIjFqyUicqGRdZ+5HtWfdOfMwzjvvPKCsI7c3X6dWGaEjim7kNbNUZWZte6+4hvtVWtYLs9uu7F8G3EMX3C6MYi5c8295y1sA+MhHPuL2wNSxN3XPdc6M3Iv9mHzODDq3s1ljzzG9853vBOC0004DikI2kvPggw8GptahU+1rbRoUWUstkUgkEnMKQ/XhyNJkqEajPfrRjwZK9EmsJGAEhozF6q7aI2Vl1sRafvnlgRK9JBM2k9iabdaPktFozzzqqKO6jtcP7rrrrqGrG5havTWyN6v+Osa5gpnmsua78a/sy/wJ7cr2PTn99NOBEs0oKzfaTIaqf8v1pS9QZWO+lz4llbHr1v2O05bv3Bh16Bg8x1jxvFeYt6WyUUXot/AeUgXOJrx+Wigcs2jq2Buvl8+TmMMkVMxzKWetLRzzu9/9bmCq6jNnMc6l0HIybqTCSSQSicRY0JcPp8mGqk3ev7EfjcxThaKtVSUi85TlCxmu0Sl2MTQ2Xcb7ute9DihKKfb39rynYTgj74fTBG2qjjHWIzMvx0oDo8Iw7NOyq1oklOvG2nj6Xvy9UYO1CgFuH2vouR/zdPzsOrDjrAo8RtHFe2KUtnqZ6F577dX11+stW++1GoJ5PKrFyPKNdtOv2VbNjdNvEdd+UzSZ0WpaRFwvsUq81ztWtmiL2fTh1KBC1kcTrUexZ9CwkD6cRCKRSMwpDKWWWvxs/oRvW5mkb1c/q0h8C0ebunk4MmUZi74d4/M//vGPAyU/x+gjFY7nIUucC7kHNVhJOzJTKyKff/75Xd/H7qpzaWy1c4ndC71OdjE0eij2mo9KOtbQ8zq7nvbcc8+u37///e8HSiRYzAMS4/ThOPZXvOIVXedSq8Reg9YEoxvjmBxzv8pmnOjVVxPhc8dqHNGSodXA3jL3RLhOYqUKv7e23mwhFU4ikUgkxoKhRqnJTLSR2ishsrMYZSQT1q4YfT6xFpcM2Iq69rq3irT5GColldBcjkoxou7www/v+t45+NGPftT1vXNtpJ9zM1sKZ2JiospMRez943WMOQIqm+gjjMrI/cUumG9729u6fm8tPZmvkX5GU84m21eFGTVkNXArZcvGjYhUqeivsDuqUW4qHeH60bezKCBejxhFWIOs3mjWqPJiLcd7IlTK0Rrgc2G2K5KkwkkkEonEWNBa4fSSUR6jSWqMN34flY2MJSqb+L2+HaOR7IIZGfUoFU5Td8smWBspzp11peKcqwZjnbjZwuTk5BRFEsfiuWpH9jqrQI1WjFFjXu/4WWVkrb2nPOUpQMm3Mcpthx12AOANb3gDUNZRXG+zASMsVWP6IayZZUSna9x1tt122wGlmsfb3/52oMydc2TElr6iRRG95iDFyE7h9VUd3hOh3/JjH/sYMPWe0ULStkPnsJEKJ5FIJBJjwVB9OLF3g6w7Vnn1b4zAkZ35vdsbfWR0mgz5O9/5Ttfxr7766q7fywrH4bvp9xgyEHuTO2aZbuzl4++33HJLAA477DAAXvSiFwHFP9F0vGH7LTqVb1Q6si2vi74W/Vaes9GFMe8m+nqco4022giALbbYAijVpl1/1peyo6Q+m5i30zaLfxRwTN/4xjeAssbNtzJK0b4m+nBWXXVVoKhEfUD6cuye+b+A6L9yHVptYS5H5vUL1/LnPvc5oDxr433+zGc+cxbObipS4SQSiURiLGitcHphCTHKKPpSZLoyEiMnZMIyWpmubM6IrNgDwv3LiLVb19j8XOp/YVUEKx/Hytkbb7wxUMZ0/fXXA6Xar9nzqrsmjGrM0+03VgPXL2FNPX0thxxyCFD8Gdb32nfffYEyN0YZWUHAvyom++RYR+zyyy8Hytx5Pq6bqGxit8TZgHPx3ve+Fyg9Yex/pJ9Cn8y73vUuANZff30ADjjgAKCsG1XdPRle/wgrkbgO7omwQ/BjH/tYYGp0mpHAc2UdpMJJJBKJxFgwVB+OiDZ8P6tcjLBSkWhzN1fA780xMILLqtDWibKPt2wv5uvEt308v9mE56a6U9HIslVtJ510ElCiS1SH9jkxIi/WJZPRODezmYNUi1ZUubgu7FHv9/ozYtSbc6TfwmrSRuiogFU2KinrS6lsYp+U2VQ2Qj/lDTfcAJR6g47das9aEZwzr7Pr5Otf/zpQ5tZ7zDm4J8H+WhH3ZGXj88F1osUjPnN9Rs4VpMJJJBKJxFgwFIUzQ5Xdrs+xNpa+md122w0oWdfmHsR8G+PoTz75ZKCweNVBVDYxWm0uwbk544wzAPjhD38IwFZbbQVMZSza8vXZ2OPF+mPRf9U2L2jQPKJe9un19Lp98pOfBEqXws9+9rNAYfmxmm+MXjNHySg3O3lGn42sfy5GpwnPzTVu7yYV6yabbAKUfArncqeddgLg1FNPBYpfVF+Pc+1Ya7lRc2kueoXryzGImoVlUYZjNQrtlFNOAYo1wHvGsXrdjzjiiLGeZxNS4SQSiURiLOirH06viP1Qoi1e9mZHUCNujD467rjjgMJ0ZbaxCvAQGcys98Nxzvbff3+g5JCo3nbffXegRC31Gp3WFsPs9VGrECALc8z6I4wuMs/G63vNNdcAcNZZZwFFyeinqPXLGXR9zEbfk3juMYLOe+Ixj3kMAOeccw4wVaHGe25QBTuOudBXqzKN59xrBe3YB8c5i5aWfjHOuVD5qmCtG7nuuuu6f6DMjRGb+gBH7cPNfjiJRCKRmFMYqcKJvT1E7NXg74xes6ZWtP3H/u6DZvdPM/ZZVzhCu3TsUjouDMLe+lUWKp/Yw6O2n1g9elRRZnOxs2OE7D36YobNbEc5F45Bi8e555474+9jZKfXX9avX8O/+kmHlW81jnVhjy+jUs05OvLII4ESwRvHpLXAiiWjRiqcRCKRSMwpjKRatIjsSp+NuSfRrhz7VMQImlgRuV8mPZcrD4hRRQ2Zmaw/ZFiYN2/elMoCTYh2Z9Vc0+9jd9NaderoQ1wUECsdOwaZa5xjx+wYYz27RSkKTd+KEXpGZppTZFSilSRi/y1VgNU4agpmLuRb9Yqzzz4bKFGH+nQOPPBAoHQI1jp06KGHAuNTNm2RCieRSCQSY8FQfTjDVgpt9zcERjtnfDizjbnst4jrYtQKdS7PxbiRc1GQc1GQPpxEIpFIzCm09eHcCkxfuIjhM8y2+xuCrX61Fr+dcS4WcbSZBxjzXDRVtBgy5vRcjBk5FwU5FwU9z0Urk1oikUgkEv0iTWqJRCKRGAvyhZNIJBKJsSBfOIlEIpEYC/KFk0gkEomxIF84iUQikRgL8oWTSCQSibEgXziJRCKRGAvyhZNIJBKJsSBfOIlEIpEYC1qVtpmtAnSWbO+1mVSvv5+m6OOtk5OTy/dyjCzGV3BPm4sHPvCBANx2223AcJrRidmq7DGsVg3/y+siYpRzMduFkNui17lo3Q9nFGiajPjisB+Kf+314WdfOPa6/+c//zntfu2W941vfMPjj7zW0bAu/Fzs4dOE+9znPkD9evSL+9///kDpFNuEJkLii2YYaLo+sZdP7Zx6JVH2RfnHP/7R9X3bF433UtseR/8LWG655fjzn/880mMMuxJ6r9s3dUONXZxXW201fv/73/d8HmlSSyQSicRYMJR+OLW3cJTxTX1M7GlvD/K//vWvQJ3Vte3cucwyywCl37edATu2y344/8UwzAWD9o6vXU9ZVlQFg7LA1VdfHSjrQozSdNK2I2fsADoq5eG9OE0n0TSp/RfDmIva9e9X2bid1p0777yzr/20RfbDSSQSicScQmuFMzExsfBtGe3K2uj9f3uU1xAZcM3B6nGamG382wdLGJrCWRR9LJ0YJpNty+JbHBeYO+wNpt4jKoW77roLGJ7zflRYYoklgKn3rmNKhVMwzLmozXsT2gZUjQqpcBKJRCIxp9AqSm3+/PkstdRS/OlPfwKmvlV9O8vuagrG/1955ZUB+N3vfjft/kTcrqZsmrYbp+pYVJVNv5g3b171+o1K2ay22t2NBvX1GTnUr88oYrHFFutLicyfP3/hdiobEfcX16RMN9revZdUTE984hMBOOuss4Bis3f722+/vWv7eDyj2f7+9793/X+NYf+vredRYKZ7pK2yEaNSNqN6VqbCSSQSicRYMJQotVFB++Ryyy0HwNJLLw2UKCLZXGRpTfsTsoNxRqmttdZaAFx77bUAXHPNNQA8+MEP9hwAePOb3wzAi1/8YgC23HJLoOSIrLLKKsDUKKIa2kaMDSPZcdhJaw960IMAeNrTngbAVVddBcAKK6wAwNVXXw0UVXHDDTcM5TzGkeCn/9P8md122w2AM888E4Bf/vKXANx0000ALL/83fnJ3gO33norUNbVqquuCsC73/1uAI477jigrHnXQ/SHLrnkkgD85S9/Aco90xEVlz6c/2Iu3SPDRtvzSx9OIpFIJOYUWiuc+fPnL2Q7vUZINOXjdOwfKPbopZZaCijKRjvnH/7wh679xeO7n1e84hUAfPrTnwam2q2niRgaucK58cYbgaJQ4jn3Ctn8Ix/5SKDuJ1lnnXUAuO666wB4yEMe0vW5hkEis3plR0avuY7Mj5LFe92f+tSnAnDYYYcB5Xp7HS+44AKg+HQuu+wyAD74wQ8CxUfYr29nmKw+Zmr7d/fdd+/6vMUWWwDwzGc+Eyi5aVZV6Dg3oFRvcO5VSi984QsBOO+887q2i3ka7l9lU8M4FU5THtaiEpkFg8+FvjufiV7vWFXFe8bf+cxcf/31gXJvWR3gj3/8I1CsAj5Hmu7h+P+pcBKJRCIxp9C6llonS4z24Pgb34JNzNLfuZ/tt98egJ///OdAYbK+1T2uCsW3s/v58Y9/DMAXvvAFoKiKZz3rWUCpoSZjNsppHJCJbLjhhgD89Kc/BabOoYiqQay55ppAqZ4gU4lQyThHzoVMyLl917veBcA73/nOVuOJ5xn/Pd25i5gtL/uKNfJUc/o5VEavfe1rAfja174GwLLLLgvAPvvsAxQ14JzH+mKzgRor//KXvwwUFf6rX/0KKL69lVZaqeuz11vm6xgtPPrDH/4QgEsvvRSYek/KfP3s+ojqYddddwXg2GOP7WO0g8ExX3nllUBRY97XH/3oR4FioXAONt98cwC22247oPj49ttvP6DMyWwrJJhaSy8qX6+Hc/GCF7wAgFe96lVd+/n1r+8uA6mP2HWgH9M58h7y2XnOOecAxQd40EEHASyMRK6hX99TKpxEIpFIjAV9Rak15df49pRpqDiE2/m2lblsuund7hMZqizcXIP11lsPgK9//esAbLzxxnSej7b6Zz/72UCJ7JLZ6vtx+z333LPrvEYZpeZYPNfnPOc5ALz1rW8FYI011gDgyCOPBMpYP/KRjwAlIuuUU04BCpPZaKONALj55puBMpdeA6+JDCdm/deqFY/CPi1bW3HFFbvOWcjWVSp33HEHAC996UuBEnHl915/q0QbsaV6lN3J3uJaH3YEDjTXzIp+R6+Tn2M+jX6os88+Gyh+qiOOOAIo95Is/g1veANQ1sFnPvMZYKrPLlbt8G9TntA4IvZUWzXV7/2sSnOu/BvHFHHGGWcA5TnRL0Y5F14fow19NuqPNprQ+zj61f3rM897xHvPuTLq9bTTTgPg85//PFD8oL3Wq0wfTiKRSCTmFPrqh+Pb9GEPexhQmKTQdu5b0rdtjKTys8z2Yx/7GFBYuwzncY97HFDeqi960YuAkk0t49EfYsSNb2+jn2QJMuRhxsA3RXXEvAZt9uZH1CLuhIzEKDPn+JhjjgFKT5+vfOUrXce75ZZbuvYTr0FTtYZhwjlS2URV5bnpn/A6qvZk//qhVA3uRzu3+Tqqwjg2/SSqgFHUNfN6OaY4767teGzPSZZ/4IEHAsV/6Vp3zrTVP+EJTwBKztpee+0FFBUQfTO9+i9GuS5iXpz+xJqyEV4/LSQ+j+Icu16i0jH3bTZzYWpVUJwT148q7DGPeQxQrqdRaP7e+13/lM+Xb37zm0CZC60+z3ve84ByL3ovuf6y0kAikUgkFmm0Vjid+RZG0kQG4VuyqbOj+4kMVvj2Nmbc7GrZmVEq2rMjW/SzEV1Go8VuhsNAr0wg5pA0wTkwOkW15vePfvSjgcJQvve97wFl7mezKnFk1bWq3hF+L3OV1avWHvWoRwHFj+X6MOfEShS12n7OTTy+vqOmCJ1e0NSHpLb2ovJRtclEH/GIRwCFqaoCXRfDjsQbtKfRdPB5oV9zp512Aoqvd+utt552OxVMXD+q/ejH2GabbYBSrUGYuzabWf4eW2uOa9V1Y8Sm/kjnRHV3xRVXACVCU19dU3UNrT7rrrsuADvuuCNQohq1JhkZ6tz2+rxqQiqcRCKRSIwFrRXOdG/O+F2TspGBmEV9yCGHACUCR3ZnRMYBBxwAlFwAI3fOP/98YOrbN7IyfT4f+tCHpv39XICsT9utNnyj2YxeM6pJHHrooQAcf/zxAPzsZz8D5kaNpsjie1WUrg8ZsCxMBaLv7uEPfzgAn/vc54CSV+UcRHUXoypVBbK+YSgb0e/8q/J33nlnAJ773OcCxUcj44x+L+dEX16varIJw1Q2wnOxuvenPvUpoFy/CC0pm222GVCiDmtwbr773e92fXZdnXDCCX2f+7Ch3zlGl/lXpWH+lVDVv+c97wGmViWvwWjY17/+9UB5VqqUzefxOWN020zVYdqsrVQ4iUQikRgLZqVatHZp82Z8m8tIzSx+/OMfD8Amm2wCFFuvuStN0Wa+lX176+eQDYoOX8PIa6lFGE+vzVaF8/3vfx8ouUyyNFm/jEZGoi9H1tgWo8y3aBvx4u9lWeYiveQlLwHKdVehyMJe/epXA3D66acDw6vxN8y5MLIq1j5zDRqFqD/DqLMY0SdUII5FH4Br/C1veQsAJ598ctfv+0U/c2GOUK0ahnBN+zdWUWjrc1UNqmC9//V3WGuxX/QzFzU/YVx73vdWizeyzshgI3pVOk3wmauP1xqLrpe3v/3tQMnD0RcYc6JquZeZh5NIJBKJOYW+8nAGxQ477ABMjW6TAVknyLfn6quvDsAXv/hFYKqyiSwx2q19K6smosKZjZpKnrM2WKNP/N5oEc8tMlMZrezxKU95CgAnnXRSX+czDJ+PY4iRUm33HX9v5QkrKBth49wcffTRQPFf9Ho9VTZx3YwS0b8Zj/nKV74SKPk1kUmag6QVwHVhRJZQ/XnP2F/p6U9/OtDsr6pVoOgHTcpGmFWvCjRCq99jqyZUNu6nycc8StTm3eurqtPioW9WX62+PZ+BtWoKMYJPP3i8d4zo/NKXvgRM9W/H9Tdope5UOIlEIpEYC/pSOP2+5dzOjo2+Nd2PuQYPfehDgRJdJqu75JJLurYTNfui8Pe/+c1vANh2220B+NGPfgRMrUI9DniOseqr0HYqk9GWb7y+rE0GY8UCK2xffvnlozjtGRGVzbCylfVr2PVygw02AApz/vjHPw70HqkTMVNOTL8+j2l6Lc0Ij6MNXT+lc+fcWuncuXA7FY1+rve///0LxwDFx6cKjAo6YpwVKIT3v8+Jpmi0JujfFK5HayjqD23qATRKqGii/0rYwdXvvW72N1IJGa3omPRrGuWo/1u4Lk899VSg5DiqwJoiTL1GyyyzTKv5S4WTSAn9L0AAACAASURBVCQSibFgoCg1baS+4XpVCGb6WvfLCAprr+mX0L5oXSjt1rH/jb4DKwq4nxh/3wNrG1uUmudk3S8rIVtRwPwbVZ05Jyoet/f/ra30iU98AoDXve51g5zerPauX3nllYFy3WX15mkZtfSTn/wEKPZpq4BPc35A/2x9FHNROyf9F15/x2rFdFVAU2SmtnqVbuxNZbUGM9Z7xTjWRayx1tbyoC9IhRP9UE212nrFMOaiVlNNy4X9rvTRPvaxjwXK8+CrX/0qUNS+z+Lf/va3QPHtHXzwwV37V8m4DowYjlanmrUoIqPUEolEIjGnMJI8HO3XsaNjx36AomRUOvbdNm/GGPGLLroIgCc96Uld+9OP4V9rpfVqN5/Gzj72PByhLTfWefPczM/Rli8D+sUvfgEUpRT7rPSL2VA4rgvzbsxBMLow1swyn8Ks6ZhPMyyfXD9zEZmhrL3JJ+T1r1V37vV+9Z7QB7DKKqsARSE/+clPBkomu9FJTfufjXXhXDim2K3Uv7L5/fffHyh5fOJb3/oWUCL1BsUoc9Xi+lHp6MPR16eCNarR/C2fndZqU8k4l1Yh/8AHPtB1vFpUbKyAEJ+xqXASiUQiMafQFw1usonH3vO+Lc2Cdzu70W2//fZAyYI178bt7VJnTwhZW+xn0mTXFtFOORcQo1Qiw429PvRbqRId44c//OFpt4/Z+yoq2eJcmAuvi33V9U/ZxdK8DK+3/qrauffruxlGdF3ctnaO0b/ovTJoRQCPZ92wN77xjUBRifvttx8AL3vZy4B6f565gLiGHYPRi/onVGu162b061xEbf5j3TkrkHhP1J6Ju+++OzC1urzbf+c73wF674/Uto9SDalwEolEIjEWtFY48+bNmxL9FRmIkTUvf/nLgZIJXoN9Toy00B5pJIWs/53vfCcARx11FFCikqLfo6Zo4vdzgdXXUBuD/i39WioWx/K+972v63Pcn98Pu2/KTOi1p0qMpPrBD34AlOx714M5BnHsw8I4OsHqk4v+iLaoRVypfGW8HsfrfvjhhwNTa3vNZhfMJsQKAeacWXVeH7CVtYXWgabn0DgxTS2yrv+vffb6WRvvtNNOA4qPTuuSCsd1oELyuRFzlMaFVDiJRCKRGAsGCmWqMUv9CEaVaWvV7uh2+mj8nfHzxpBbGVUWZ4y5fb7NkhX2jLAyQbSPu38rDoi5yOqiGogRWtZMs0aW5y5zNupItSnDFSqlcfQG6tUf4RiMpJJ9G5EjizN6rZZJXqtEUfPd1bbrVPNt0FmhoLamVO1GFzrWq666qqdjxAgt7yX7nRxzzDFAUTj2XbG22imnnAIUv0bMeI9YfPHF55x/J86tuSRWabAfjs8Nc1bmAmKts17vQ7fzWrjOvBd8llph37wto9nsPVaz+vSKvqvN9HW0RCKRSCRaorXCWbBgwUI2FKPD/KuSsc6PNtZvf/vbQFEYRqfFTp+yNZmuHR3t/aCyiX4JlU3nuXbC/2+yn44S/faG0Q5ttVhrsJl7pE1XW7z7j8rG/Y1C2bQdW8wodzt9M7I2GaoKR7ZmTTyjG10vNSYefXxN59mvb6iX8fsbVbyVId7xjncApYeL/qpaZXTnyrnYY489gJKJLsvfbbfdgKJ4L7jgAqD4RaKyiXPUb526ccLrbgSWUOHOZpVoEefV+7DtvRN/Z6UR61R6L7h+rDYt+u0xJLJadCKRSCTmNFornImJiSm1zKI9z7e29mQVj1Fm2qmtbOrbNnZBNGbcSIzrrruua/+RIfcK2YEZ7JFFjgL6TFQcTf4EmavZ0m9605uAEmmjWnOOjDpyToxW8XhNuUrDQNt9x9pWjj1Gbqnu7NCoD08VoB38jDPOAIoiiox2ho6ePf2uV8y0XawkoBLRb3nxxRcD5XpdeOGFQKmqoDIxV817yjVsrS23dw7MVN97772BEs0YFfCijHPOOQeY6pc48cQTgd778owS0/kJgSlWo179ZT47fZY+73nPA8p1/drXvgbApZdeChTl07ae3LD83KlwEolEIjEWtKqlNm/evMnFFltsYWREkx3PqDD9ClHR+LY1ttz9yd7shyGTdbvYqVEGXFMLHeff9btpxj5wLbUaE/B7Ga1Ves8880yg1DYyM1wFs+mmd5+Oc24Okrbac889F4Att9wSKDWy9Hf1y0wGqR/WLwvy+rgfWZhjs4/72muvDRTle9hhhwGljlRcB7FmXuxeKGboDTOy+mFWSreCuoxUeN1lrP5VMcc1rbK1U6QVBYxWc25j75lebfKzWUV8huMAReVZhcMx+fywCvmwou1m4x7p2B9QfDc+P7z/7ZdkZN6xxx7btb3rYVhVOrKWWiKRSCTmFAaqFu3b1Cgh34oyCO2SURHVlIcKxnwafTYxA13ELOnaWGJWd8d44nZ9K5xatddY2Vr/Qqx86znW/FKOXUYsa1tnnXW6jmNk16AROYMw2V5Zc/ydrH2ttdYCytjMNbLirbZ4K92ac1Lz/QyKYfY98TrH6+MY7QFk9Q3Vvqw9VtXQSnDEEUcAZQ6tL7bFFlsARQ0OymjnosLRh2fEpnAszrVzMywMMhdNlpCOY0y7P+8dI32NXt18882BqT49q7c4F0YQ19Re03qYxlqUCieRSCQScwd9KZxaDbXavqznIwORiapojHqL9Z9uv/32rv0LWb5v6Ygmhh2zqjtqfQ2tH07tHIwq0rYez6mmlGQosbtq9EdEtdBvvs0omWwcox07rUhhno3+Kdm6ylelE7ufNlUB7zc7up+5qB0rjj1G6Fnd1+u25557ArDvvvt2/c7tvM6xmm+vx59hHNP+bi4pHP1VWjh8ngirKdhFd6eddgLguOOOG8rxhzEX+hdjtfBaBLCw2oa5ae9///uBolh8tukz1iLic8YIXZ+xtfOJCmjQdZEKJ5FIJBJjQV8Kp99s+VoMevThCN/2McJmWNWBh+nDGRTPf/7zgZJXY+SeY2+a68jyB42CacveJiYmhpbjE6+LOSSqu2HnzTRhlP6spnOXkfbb+XPYmEsKRzVoJRKrLfgc0a+1zz77APCMZzwDKFU5BsUo7pG4HlQcsbKEyuass84CSkSwvhujHo12jXk3/l4/eVukDyeRSCQScxqtFU6nP6UWkdXE2pr6d9cYbK0vvG/9rbbaCihRczXMwDpnTeGIfv0MEUPIlp8zTHZcmG5dTk5OjiQaqak6c8SgSmlYmMvrwudC7JPUb0XkJgxzLsxRVMXXoO9FH5/RjNajjFVYnBO7JttluYfzBTIPJ5FIJBKLKAbKwxkUTX6H2ltWBhN7jhi1UstBsU+GUU/TdKKcdYUzVzCXmeywUcvTErMxF015Gv1GmTUh3hP92ur/ew6L9LpowijnomlNDssS0iuaFFgqnEQikUjMKbRVOLcAs9MMezxYbXJycvlefngPn4ue5wFyLjqRc1GQc1GQc3E3Wr1wEolEIpHoF2lSSyQSicRYkC+cRCKRSIwF+cJJJBKJxFiQL5xEIpFIjAX5wkkkEonEWJAvnEQikUiMBfnCSSQSicRYkC+cRCKRSIwF+cJJJBKJxFiwWJsfj6oY37hKq/eAW1uUtpmVkx1X0b4s0liQc1GQc1EwF+ai6XkwTYHikaDXuWj1woloW9G2NjlWf671Bmnbb6fXh/I0fXjGXuuo1uPez3EMVpG1I+igVYWnO59hv/jnSi+X2lzHm7Gzivmg5zRbY5uLGFan4EVhrE1oWnvxXqn1DLvvfe8LlAr5cX92y73tttu69tPUffnf//53q/H0ioFeOLULXvu+Nom+aOJk1ya/aaE1NX6LraqH0aCp15tgmnLv0/6u9nD2RSPaXoOIeN6juImbXvxNx5SQDHoTxON4c8Y5GKZ6rI3tXve6F1DGNOj8D3rdxqGcm14cvd4bNq/7v//7v67vY0n/2v0uxl3ivxO1Y97vfvcDSqvoJjJuC/qagvFFI5qeF7VrNKznRPpwEolEIjEWDKRwmtDE+uP38a1fU0SDssBaa+tB0Gs77bZmvlGbD6ZjNHPNVDGosmlqYT7K8dZMJ5Gdt0VknqK2xpvGOA6WX2PJNQVSQ23uYrOyJjY+G8pGVVUb81//+teu/68pl17/P1oH/ByfR7XnU9Mctn1epMJJJBKJxFgwlKCBGosbNnOsKRO/r7VljecXPw+idAZlz02Mp4mpDqqAxuHDmStO3qYgluWWWw6AW2+9FbibDQ6qrmp+ol5R2y76PeK5j/J69ova/Pu9fq1B1d+wMOx1OzExMcU/vfLKKwPwu9/9ruu38b6Pz9qoTGp+ztgqPK6bOPd+bxBCzd/d7/pKhZNIJBKJsaBti+nJmVhf22ii+9znPkB5m3YcB5gaqudb1u2iknngAx8IwB/+8Ieu/bSwZ188OTm5aS/n3m9cfe0cvve97wHw3Oc+FyhRKuussw5QxmSEjrZemYlRLP7+qquuAupz26SU2uYYzJs3r2oT32qrrQA455xzetrf/e9/f6DMwaAYNBqp7VwM0w9WC+V+5jOfCcAVV1wBwAYbbACUKMYf/ehHQLlHhhXmOorck7migJvWyTRRdD3Pxbx58yYXX3zxofnu1l13XQB+/vOfA2VdPPzhDweKsllhhRUAeMtb3gIUBXzaaad1bXfdddcB8Mtf/hLo3dfc9nmRCieRSCQSY0FrH04bpuRb8N73vjcwVZGoiFZffXUAdtllFwCWX/7uZP8TTjgBgJVWWgmARz/60QBce+21APz4xz8Gylte5RNZf/SzRF/P9ttvD8Cpp57a89iaUMsJkFH4+UEPelDXWA488EAAHvawhwGwxRZbAHDZZZcBRbH8/e9/B+Diiy8GigJaccUVgcJ0Pd5vf/tbAP785z93nWdkKkssscTC2P42mIkRRWXjOckYY46BY9h3330B2HvvvYEyJo/lHKj2PvGJTwBw2GGHdY2pX2Wz8sorc8stt7Tebiam3i+bd65WXXVVAPbcc0+grPlPfvKTQJkL762a7b2fJMxh5fr0G502LNTG3pSY7P/vuOOOfPvb3251zMnJyS51E88h+q+cq5iH47P0+uuvB4pVR8X70Y9+tGs79+vnv/zlL0B57tx8880AHH/88V2f43NCRP/3TJaN6ZAKJ5FIJBJjQWsfThum41vat6zMeamllgLK2/qtb30rUPwXlmsQvs3XXnttoDDbo446CoDzzjsPgKWXXhqAX//67go1KiFVmaziMY95DAA//OEP4ym38uG0fbtD8cHIEJxL5+Id73gHAI94xCOAwvb1zfz+978HClNxTM6RykqonH7zm98A8JWvfAWANdZYA4Brrrmm63zEKGz1IrL1P/7xj0BhVc5pjJxpgtd5vfXWA8rYRNO6df04tx3bjbxmVlO0o/fGHnvsARRb/bnnngvA0UcfDRSGasb6TTfdBBTlE9ddDdNFcv631E/fc9Gkqmr/7z3gufgckfX7t1ahIPqExaD1xQbxc9YUZ1RTXoeOYwLlPt9hhx0AeMMb3gCU54TK98YbbwSKZURFpLVAy4fPjze96U1dv1dxeR7+1XKz+OKL8+9//5sFCxakDyeRSCQScwetfTiTk5ONNZA6fwvlLRn9GA94wAMA2GyzzYDyVlYFfOtb3wJg880379q/Bele8pKXALDWWmsBRfksu+yyQLFLXnjhhUCJeb/kkkvaDntaTKdumqLAaoVL/f6kk04CYP311weK30qo2mQuMlevgcz3UY96VNd+Tz/9dAAe/OAHA4X5DrPaQq/wnGRXzpW5CNqXa3Mmw1UVRnb4yle+EoD9998fmFqrL0YKud0dd9wx4MjqiDb1puKrnuv73vc+oCgb175q7owzzgDKmn/e854HFN+fzPVtb3sbUO45o5Fq+RTOufdirZZXG0Q27zlqafB7nwfx/tYyYX0w1dxPfvIToMyFY9OSsuWWWwLFQqJvWL9o22jIfv1ZCxYsmHK/xbyaOO/xmel193d+Nvrs9ttvB+Dqq6/u+uv10x++ySabAGXsW2+9NQAPfehDu77Xh+m69TixDmavSIWTSCQSibGgtQ9nkIPJ8rTJ7rfffgA8/elPBwrDVIH4dvVt/qIXvQgoDNi3sD4AcxK03b///e8Hir3R/A5t9NOwt5Hn4QgZi/4q/8riVHtG4K222mpAmcOHPOQhQPHFyGS13ctUVIlG1agC9XM4Z53sc1Bbfcf3XfuuwSz5l73sZQB88IMf7Pp/fTvvfe97u8agDd7oNBWPUAnJaN/4xjcCRRXWoFr405/+5PkPzYfTa/l5r5/nrtL1dyrcl770pUBZJ6ussgoAe+21F1DmQGYa1YD3Rs23EDHMuXANe45eF+9f1Zjs2nP0uvjZte89pS/OHJQYtWpE34c+9CGgqIII/R2x4rIY5lw0WUYcm2PxOrtWVS76dlR5KmL///nPfz5QnrXud8011wTgO9/5DlB8yCprn52uu060eV6kwkkkEonEWDCSatFN9Z+0yepnUKmceeaZQGGBvqWjj0YbvH9lwN/97ncBOPHEE4GpWfYyKd/6tVjzccC5kEWZd2NUmQwk9v7QB6Micnu/NwpNFfn9738fKHPscS+//HIAllxySaCwy0FyLfqts+R10ecS96P/QWVqZJbZ1jHyyvUj21MZ77777tMeP0ZFyqBHgSbfiIxTJmrUkeckGzfvQj+Y68Pr/NnPfhYoKsHj7bzzzgBccMEFwFTG2m+eThu4b8d0xBFHACWHRDYdFatjNFfN58dxxx3X9Tt9gWbXP/vZzwaKv+OQQw4BylzVxqqyGUXXzKbK2bW1bI6ia1brTrQGuR+tB86lc6DSvfLKKwE4/PDDAXjsYx8LwKWXXtr1+ayzzgKmqv+2SIWTSCQSibGgL4XTtkJyrIlm/oU2XJWGbE0lY30x2Zm2Wj/L5v3d2WefDRTGHM/D72PFg0HYXD8Z252/N6/Gc5PByqZkJLL7Jz7xiUCpleVcxGgXfTZGQ9WuSZM/ow36ZcOOIW6vQvWvrEzmqR1bu3btWhiRUzu/2EV1GKjlWcQ8LL/XZ2dehf4sxyqT3WabbYAyJ3H/3jtGfqkCdtxxR6BEblndYzqbfOd+h4Hoh4j1AFWe+tji9VSJPP7xjwcKK49zGP1iKuNnPetZQLnvHZvqstf8rJqvpw1ivclal1O/d85UKlou/H8rjHjdN9xwQ6BE8JmT6HPGdaafXOWjX+vlL385UKId9QF7XNVj53lmP5xEIpFIzDn0pXCaejDUfq+/4GlPexpQ3pYyHfNkjKCRCZklryI6//zzgVIzy7d8tLH26lMYROG03SZW/XWMQsYhg3XOnvSkJwGFsZrDcMMNNwDwjW98Ayhx9zFOv1eb8fz58we2Vdcy9mvwHC666CIANt307kBBFY0VJWTp2vh32mknYKpCkska1faLX/yi36H0jdq6iPkWMlXHYA6RLF9mqo09svR4POcy9rpXGbtffX7RRzgKeG6ei2P3+eGat5ahtfNUdfrgzJvptbrH61//eqD4sbSkxGocTff/MJSNiM9Kjxl9en4f82aiz9W8POfEZ6Xq3+eB26l8zcMzqtUoNf3bKiGjY30eaYWK598rUuEkEolEYiwYKEqt18rR8S34gQ98AIDtttsOmMpQjYQw49gMYXMJjjnmGKC8peN5zGZ/jV6PHVmeUSdGJVnB2rwcI/Xcv+zPKDSZjgw4MtyIUc5Nr8pG6KdQxTkH73znO4HCSPfZZx+g5B4ZsSMcu+ul5p/oFf3UyqvB+Y6KUyapjdzcEc/9c5/7HFAiOGv5O/F6up5e8IIXAMV6YL6OimYcCkdEde1f/Q/W8TI6VT+EFdGb1mz0FesPi901a37NcSJWzlbZxHwb17iKJVafV715jxi1qs8n7l+15v60Dhm1qC/H58s3v/lNoNxT0/mC21hEUuEkEolEYiwYSR5ORKzSGishazc0w/jFL34xUHJTjKwxe/oHP/gBMJW9R9YQ0bYPRj9oW0nbOk/Gu++2225AYSqxXlhkh2aky4wHZW/DzDVoC89Zv5Rj1/au/ypGMamojH4cVuTdKPq0xLpvQp+KUUXa7s13cC5iBXaVSfThaLNXQamsjPCSKWubj9FSbbv31jCTTzDmtxx77LFAuf5Gn+oT7Kj80PU3VlSW3aveHIO5KjG6Ld4rsROx95jnNQhqEXXRl+f11/dmXp75VyeffDIAG220EVD8W/7e6+y5u73H0X8eo1X10fj3Yx/7GFAiCeOctn1epMJJJBKJxFgwFIXTQ/0loLAyWZwRFE9+8pOBqV0MY2XdX/3qV8DUbNrY76KJwbQ9/17Qdh+ytuc85zlAibiKHfpE9AFo99ZGb5Xo2B1zUYTM0rwqcwJqfXEOPvhgoH3V39mE11Nfm2s7ZpabLxN7N8WurLF3kD4ct9cm73beY7EKca0CQr8VkmdiwB7Tc3BNR5+u94KRmf6/qk+/lP4IlXCsxO4caGExly2eo6pRy8swlE1Eraae56KFI1bV0KdndWdVm//vunH/3kt+r5KxGv1VV13V9Ts/i5/+9KdA8Z++5z3vAboVeubhJBKJRGLOobXC6WQ6kdXX7L4xokJG49tWxSOrkp1F5eJb1lyFWrZuTW1su+22QKkL5PkMI86+baUB2dnPfvYzoFTMtobRxhtv3PV74+u//vWvA8Xmb07TU5/6VKB0DG3b52Y2I/tqiJF5NbzrXe8CSvXftj06rFihkjLiT9U4CJp6R/nXY7kWjdAyYi9GJ8X8DfMp9IvaIVR/5wYbbNB1XjLjOLe1uRtF5QGjxvQnqCysWKxKV9mo5mXj+i3Nmvfeke17fzvn5vlYr1CVp7KKvqG20ZZNmK7jp+fms9O6b47FSDvz7Hx2GjVmTqLrRD+mkZ9G/PmsNd9GeNzoD9Nv6voxUlC0rZcoUuEkEolEYizoq+Nnzf5Yy2r3rWmGuOwqKhsjLsw093cyHRnQ6173OgC+9KUvASUCI/Y2j9An4PkPqmwmJiYWMoReWXVkT0abfO1rX+vaj4xFBitT0U6tspH1OTdG+MWeMjUMs6PjsKA/YpdddgGmsvCa3dsoRxlurBZeg8pGjELZiJp/0Rwz8yG8/h//+MeBqQrHdRd9fcJaekYp6bvROmAuXK/1xAZBrV6YeXTC+9xzVaEYgSWbj9XBHYNz4fVzHfl7rQaeh8omRo5FNTrMuVHVWUFCGFWob9eq3mb6ey5upy8ndrF1Ds877zygKB0tKtaj8763goH5NtarNP/H503tOZcKJ5FIJBJzEq0VTmdcfVPfm9jV0h4sviXNs7GG1gEHHACUt7uKxIgNe4SohHx7qxJqb1vPo8bi+/VfTExMTIm0cW5qfiURKyEbRSK7M6Im7sfjxP42MhujWtzO48ZInBid1DmmfuZhiSWW6FlRNEEbvn4Jr7P5WIceeihQ8jL07dkJ1goF++6771DOpx/0GhkpC5fFawUwuiwyYffrevDesCOoqkDF5Jzp75DVjzOir+aDi/Xkol9Ldt0r3I8qId4zWlSacotq/W8GzU2anJxcqOrimGO/K31u+pu14vh8ULFopfHes0KFith7aPPNNweKX9R1d91113Wdo9FrRqdFFRh9N0svvXSrtZQKJ5FIJBJjQWuF02k/b2Jx/r9vQN/qvnVlb7IvGat9LGQY9rPQrm10ib0bjj/++J7OvamPT1tM3t3Lu+tc47Fi7khNDcYqwNGuHBmxEXd+dg5lyuYQyIBqY49Kp9+K2cNQN1YD/+pXvwqU3h725LAKg2Ox3pzq0LG89a1vBWC//fYD+q8Y0G/uSdM+YWoF5Vg7y9pp0UcT+6Poo5HxmiHufrUGWIH9U5/6FFCY8jiiE2u5OMOqZhAjP1WHXndZv8q4hthTKqKz4kG/a6o2z86R+Vj6qc0xMrfIiLxYdSHm56l09dHYX0kfkj4c14Nj1wolvEb+f+w11jaSLxVOIpFIJMaCvioN1KLURKwAEPujRBup7Fi/hTHj+lxk7/FtG6vI1ioQxPOUEdd6i/SK6SL2tME7Nv9aG8mxORe1Y9ciZGQoRvKp9lSRzplMp4mJDat22sTERHW+e4WKxGgifTfmIEQmbESX9m/nIkYx9QrXqfbyUbD+GAHl9dLXYnSZtfW8jv7e9bXVVlsBJeNc5mumuJW39fFoHTDvaxzRaTX0WxGkhqjWVQNeR9ejn2touhdcj4MgjrWWc3jSSScBxR9pVfh3v/vdQMnTcmw+02I+lj2k9A1pVXL/RgLrW4o+Gp+5URH3i1Q4iUQikRgL+srDETJCWViM0IqMxexYmapv41jrSPuijMXeDb5tPZ5x+aJWUy3+/7CiqTrP3XPRzyCDtLqrYzOKzLFpi435FbGenMzWPApt8x7HqJJYf66GYTPbTn9WkwKOkJ0ZIfPIRz4SKOzLDPEIVWKsvWV0U9sxDjuzHOq+s3hurgPzMVTvKh8Zpsr26KOPBspaVhlps/ceUwXGqMjZUDa150W/cG5dP+ZTWSNNH5+Vk2c712xycnLhOVvPzW603jM+F4w61AdrVXmfJ/qrvM99Xtgd1eeFVhx/Z+SvflFVW+zGGn2MUfFkLbVEIpFIzGkMVC06+mT8GxWGb2NjwGWkRlrJ3tyfkVban80cl7XZJ8O3saxQllhj1qNkddZ7MjPYnCG/tzK20SKeywknnAAU9qWvx9poskGjkmRGMhPrgKkCmuzUo0RbZSO8rl/+8peBwu5kxLI668jJ3vx9rAelT2cu1IeLqq+meJwzfXFGGdrdVPb+mte8Bihzo+3dDHHvHZmt946+w+jfHPXcdEb6eX87ppgD0i/cv4rY54MVth/3uMcBU/vmNKFWMaVfZdY5F/rafBb6rFPlq0ji2n7zm98MlCg2n60+P5xjI/OsYKLvzvUR79HoC6op4Vo1mV6RCieRSCQSY8FQqkXXfDf+v6zNHAFZlz4d6/688IUvBGDXzgHyyQAAIABJREFUXXcFytvbt68dAa32aiXTyN7advzsF9PZMWWcn/jEJ4CiPI477jig2NjddrPNNuv67JhlPjJV/RKO2egVbcBtVcUoGG6/0WmyJitix3VlnlW0c4vIoLV3zxamy9+pVZzo3Aam9smJY1bNuy6+8IUvAEXxmMchY77wwgu7fj/ujq7T3SMqm2FFp6lo9U/ozzCSzzGrjGN+oIh5OLU6eJ1Vn9tgumrRrn2rM5uDqO9OlablI1ZvcX/6bs3b0UdjXs0VV1wBFCuQ68Ht9afHOenVB9krUuEkEolEYiwYKEqtlhXvWzhGofl21U6oHXOvvfYCir9Cu6VvfesJWTNNRVTL8m86d8+39lbvFdO95bWlegzj3mWgBx54IFC6m6pkZChuJ2tze6PfrBodcwyGce6zBcf+6le/GijnJptvgr1DjOAScYy9Muph1Mxq+r8m5ui9IGPVd+faP/300wH44he/CJR7xe1dN6OoeNwGM63PYVsazDXy+bP//vsDJXfNcznooIOAqfd9fI64LqNPsN97rnP/0benoo2VJVTr5lt5T2gZ8XvrxOnnNMrR55F/9c1oSfGz6y2i5oP0c9t7JBVOIpFIJMaCiTYsY2JiYsYfN0V1yBj83u50+maMOT/llFOA0kvcHBeZhW/VGCM+aBQJcPHk5OSmzT9rngvHKrOQUchkVFfOmZWOzddxTDKQpszwyJhruUi9YnJysueU4qa5aAuj0qz/FeGYtNGbg9SUe9QvhjEXTX2HaurL7ayN5di1FojIurXxi+jDMQpOZdQrBpmLUUXGGf1qXTAjsqxYYY0++yt5b5xzzjld+6kpF69BrDfYz1x430drTuxrE+fKzz7j4nqKnUPjZ/3ccUzxudKvj6bXuUiFk0gkEomxoC+FE7PfY+fMmI8RKyaLZZddFij+DW2qqgDfynE/tbdw7TgxMmSGjPjWCifGydf6zkT21MO5dP1/De6/s5ItlKrD9kVpsjtP0++iNXuLmd/9VnSw0oD5ONdeey1Q/Bf9+lb6ZdiDsPqasqmdi+spriOjllwncW5VMP6/itprEStaROZcm5NoNRhkXbhGa6qq1+vjGH0OmbtiX6RXvOIVQPERH3XUUUCpWPKqV72q1yF0YdC5mJiYmPIMi/dl7frX5iZWKIk1+uKzz/+PNRhVxNFKUHs+uX1Hvk4qnEQikUjMHfSlcKKSaGLvNcS3ZpOde9iY5jyH5sPp91wiZiu6aBAmO5ci4KD/CghinP6smo0+RmLJ3mtjihXRh3VtxrkuonUg3iNf+tKXgOKbESqpyMJrVaNrc9jkEx5kLmIfqqb7v6mXV83P1HH8abePOW1NPt8ZujynwkkkEonE3MFQo9REtEPWjhFju9vaLZv2W1Nenp/ssSPvYugKp9fIuaa5mOE8gPpctM0pude97sVdd93FggULZi1KbbYQI4jEKBROWyuA6FWlDaubZsQootRq92v8PlbjMKKrpoBGrbSHORd+rkWPRWyzzTZAqSBgdGutYr7f66up5d3UfDm1815sscX497//nQonkUgkEnMLbRXOLcCvR3c6s47VJicnl+/lh/fwueh5HiDnohM5FwU5FwU5F3ej1QsnkUgkEol+kSa1RCKRSIwF+cJJJBKJxFiQL5xEIpFIjAX5wkkkEonEWJAvnEQikUiMBfnCSSQSicRYkC+cRCKRSIwF+cJJJBKJxFiQL5xEIpFIjAWLtfnxXC3SOGj5+Q7c2qK0zT16LmazxfSwkHMxfORcFORcFPQ6F61eOG3RVBF3WD06en2g1KrPdhx/1modjXsupqv6CndXl72nlDvqdS7sHGn14bmIcfUaGsVxhrXPUVXAjujludXvWIY1F7Eytp+HQLqnxTLLLAOUbq39krk0qSUSiURiLOhL4TT1Ra/1FYnw9zWG6Vt0pZVWAuCmm26advt4XnY7tOPf7bffDsCKK67YtZ9e+7r3gl571Nc6+DnWJZdcEii9xh2Ln2MXxMj64vFjPx6PK2Ox131T58HZwLDZ9gorrADALbfc0rXf7bbbDoATTzxxKMeB5nOP3ShrcF3EDpGx1/xWW20FwPXXXw/ArrvuCsCRRx4JlP4md955Z9d27mcG1T9yeP839YCxV0y8jnFOaveWY+y3B43od27mz58/ZZ6jMmnqguz/e/8ecMABAOy5555d/+9+11xzTQBuvPFGANZee20ArrzySqB0hHV9ud/bbruta6wqG9F5vm1UTiqcRCKR+P/tnXnUduX4/j/P29u3MkWGJUmRMZEmlaIBTSgakNAoIVYUMlQytbCSoRahVEqkgWhVSqRJ80SZolKhMmVYWvLevz/06Xru83mud9/7Hvbz8DuPf+51T3s493XtfRzndCU6wURW/GzCoP4/2bu/VwFFdhZZmSsC+vSOWAyLa73i56iMsMZ0BlWJ2iIqnUEZ66hrlN+/jf+NoE8Fc2GLQVe7feITnwiU8XLHHXcA8JCHPAQoauBnP/sZAH/961+BuvIdgN1PfPXTOCdqxzToyqFxP/H/w64U2sW4UP15neJKnh7rIx7xCKB4c1ZYYQWgjAfvD3pMHvawhwHw+9//vm97fq/yjtegtoJxrviZSCQSiXmFoWI4gyqUmi+1iW1H36pP3cjG3F7MWlHZ1I6zywycpu9lDJ6j57LSSisBcOuttwJlrXF98Kq45ZZbDoDHP/7xAFx//fV9+5PR+r9JZbGMgrmIG0wag55TTWk0MU3HtjE4x4MxQJXxddddB8yMj7q/psyvcVybZZddFoC//OUvs37fpGC23357AC666CKgHLuvnqtqwDEfEcd+vD9pC+MlczEu4z6jVyfaMsZ4VLaOC997biof36twhPfOFVdcEYDbb7+973jaKuKIVDiJRCKR6AStYziLy0EflRH4VDWzwqyy5ZdfHoCrr74aKBkTPr0vvvhiYObTNx7PAMfVKoazcOHCRmbYlB8vQ/HYZbxrrrkmULJGHve4xwHwzne+EyhMxXNS2ZhdYnbS5ZdfDszMUqr5y6dl8mQM5360tUXbzJ3pGNR7YMzmla98JQAbbLABAD/60Y8A+P73vw/AzTff3Le9WtxDNO23rS0GuV/Ec3YOvPjFLwbgtttuA8r94Be/+AVQ2P8zn/lMAG666SYA7rnnHqAoH3/XFNvxOM0glO0vRoENbIsFCxb0llxyyWqtV1N2qN9rG5VsVLzeLzbZZBOgnMs111wDlPvDLbf8p+QwZknG4/D+5H3D9zFWnDGcRCKRSMwrtI7hzPa0H7WFiP5En8a77LILUFic7F8m84xnPAMobH+WDKvqsc52vMMqs+nqpqamakoi1tvIUK0tWHnllQHYddddAXjWs54FFKUj+9J3ryL66U9/CsDqq68OwPnnnw/MZCqxnmecsZ0xthrqg/Eq/dWOj1hX00LRTgTDnPeg8U1jOzvuuCMA73jHO4Ayh4zZ6CUwBlg7ploWXO37tljc/2MNkDbYaaedAHj0o//TZeqHP/whAAcccABQ2PrJJ58MlNoj1YNzS5uIu+++u29/cc7GOTJO9Hq9xXayqF0H74lRWTj/VX3rr78+AGuv/R8HjXNDRXPuuecCZd7H61wbH37ucZgNZ/ZbW6TCSSQSiUQnmEgvtUF7HvmUlbUbd7Aa9qqrrgLgjDPOAGCdddYBCgPxd5deeikweNZcUx3PKBi0pkCWbpW7fml9rxtuuGHf72UUl112GVDiVzIdmayK5znPeQ5QujTceeedAx9fV/3cYoaN7Mv9q+rOOussoNQOCMeFfuuYqfOmN70JKDUosXZgLlFTEtpA28iKff+ud70LgP3337/v99rI8eCr2Y3GNWpwznpc7tf3i4tXtkUcc756fY4//ngAVl11VQDe9773AWXse91VvJ6j5+z9wY4iZq2pDj23qLA8P23tfSLGPZdccsmRbVGLXwmva8xKfeELXwiUuLVen1e96lVAuUfq2TCWpydFxPuHiMcRM4HNalNB3Xfffa0UYSqcRCKRSHSCsfRSi0/F+PSvVavqDzz11FOB4nc2u8SntD5aa1PMuLrhhhuAmf2DRE3BxKf2KIy+yfce4wlm4skQzLjzc/PrZSayN9m7WWcyVtm9tnnNa14DFEVkNps281V01YF3NsxSrQwUNue5yjijjbfbbjugrlAvueQSADbddFOgZDnW6jFiHcYkUeu5Z52F10lWbkxu9913B0rNiT2yjjvuOKB4BWL9Tqz3ieOyds5N3w+DuO/YLy5+rwfjd7/7HVDUuhl4KmGZtq8q4hjP8HPnXsxGa+oaPg5bRFUXx3Cs/FftqUx22GEHADbbbLO+77XNlVdeCcAFF1wAlLqcWM/j/cPx5riJxxEzCKNiGhSpcBKJRCLRCYZSOE3ZX02ZWjJWM2xWW221vs9l52effTZQ+gPJ8szEMJc8Mo7acYjonx5F4bTNSNKvLOvyWKw1MstIW0T/s0wl+vjNUpMBPfWpTwWKWoz/02a1LrXjRE15RsUra1elxSyj2vWsxea0sR11X/aylwElZug4mwSLr6HJFpFhmh0k05Sdy2T13ctoheNFmw57Xbusto8dkj0HYzWOB1Xcox71KKDMKVm6trL7ht4D/+9cmMvOFrVYiXPBee0xx3PceOONgZLB5/aMb/nqvbOpG7WKx8/j/SH2p8z1cBKJRCIxrzGWLLW2Ff0ylze+8Y1AYWGycWsM9E/7dF1llVWAeoaFqPVYs27n0EMP7TvOUVhc2//KvmQqsR7GczVGI9PQ1yqr8/9PetKTgJKxpy2/8Y1vAIUZ1dYMaVtpPgyibayTUL1pE1ndtttuO+vnsvl9990XKCrOWgRtKRxPKiYRawhU2D/+8Y9bnll7RFvE6+/19jp6fXbeeWeg1JLsueeeQOkwEWtLjI+6PZWRNq91HqjFOyeZ4Vero9tqq62AkoFpvZVzwDkSOyl7bt4/jG/stttuQFE+xgCN7TSdozZVIQ2D2v3CY44xHRWv40G1piIRJ510EgA/+MEPgBJjiYra7dhF/MlPfjIAj3zkI4FS4xQVc1yPZ1h1mAonkUgkEp1gJIUT4wGiifVbVe9T2t/bMyl2SDV2Y2+lr33ta33bjyzd7b3oRS8C4IorrgDgk5/8ZN/vxuHDHVYlyWhknHHlPX23siptYvxBRbP11lsDhfm4/29/+9tAYflNKnT6ip+T8m177LU1fqItVXG//vWvgZJ1Zi2Cr7I7lZOwujqyPceLrN/VDyeB2mq2NSURP1ftrbvuugCcfvrpQLGJDDiujmunCueoNjCzS99+zEaKGKeyico2wnM48MADgZl9wxzz1lXFWGz0DhjDUxm/5CUvAeDGG2/s+32tJiZei1GUjdur3S9iPDNW+FtX99znPhco91Cvq3V8cRy5Pbdj1uuWW24JwLOf/WygeEo+/vGPAzMzgce1ImwqnEQikUh0gpEUTm3dCJmsaykIn772RIrxAjvdWlPiU9kMLp+6sWq+dly//OUvgaKkZD5drglTYzIxr1225Xt9qx67eff6sfVvy+JkLtpWv3RTvnytFmYSUIVZV1GDDNcYjfGo6N+22trMPtm9anHvvfcG6nVhxnbscDEJ1Go6jEOq9mKvLJWtStf6rAsvvBCYWadjfNNxIoNVxdmZwJq2uB5OF10YorJx345h+4FZj6eK835gfCF2LnYuxRqzyOadY9/85jeBUocTx5UYd4be9O3EbXou2iR2GLAGyew0Yy7Ob1Wg48Z7rQpbJfP85z8fKGpPxWsHAW0cVwiNdV3DIhVOIpFIJDrBWLLUoh86KhsRq2aFOeMf/ehHgdIl1tiN2Ww+zc0yksnGHlsqGWNC9hvyaS7GUXsy9Mp39+87rj3vGvUy3Te/+c1A8cmbYWMNkr+TtZm11FQtLSbRT66GJmUjvF7GoSL7F2buuV0Vzre+9S2g7nP3HI899ti+910iKs/Yj8rqeeeA52QHCsePvzvssMMAWGONNYDiHTBeodfBuaUC1qa1teonOS7ct8ok9kLzvfGKI444Ytbf1WpL9thjD6Dcd4xbeR/QRrL36XFMaM7gGwa1bcSxHe+pxlic/56rXhzjm2bw2YvR8WMtkzHf733ve0CZM+5HW6hGVc7GOWsZgYMiFU4ikUgkOsFYu0XXMhl8LxtTAenT12cv23e9m7XWWgsoDEUWoG9WVhYZj35P/ZDXXnvtYo93kqixJBnnFltsAcyss9BvLdOQ2Qjra8yXj50I2q49PpdV18Jj9Zw89ziOvK6yOH8vE9Znr+qroZYtNxfw3BzbxjHMsJJ56nv/+c9/DpQ4V1Qoka3LfGX3tay4IVbJHRg1Vhw7VH/3u98Fitq3S7SxXO8fMTvV994/XA1VNel95pRTTgHKfSOuPeNxOKccJ+PMaq19Hlf+9Z64zTbbAEXVe32//vWv9x273+sx0dYqIG1hrMb3dm/w3rnRRhsBpc+lMSJjiUOvaDvUvxKJRCKRaImRFE4TOxK+N0PGLDOfmrI4WbtPV/uDma2ib1Zma1VtzH5R6cR6nklk4jSte1P73uwRWZuMxHOU0ZpxZ+2RHZStijfrxO2ohPTRm4lTO97ZFNhcqZ3YrXefffYBim1k5zLXT3/600Bh+65/86lPfQroNhtxUMRxECv+ZZh77bUXUFi2feBUsPreVXunnXYaACeccAJQ+hRaba+NYkfmWp3QJBA9IB5LjP26aq21R6o+a5K0ifU4zhnjFXYQcE44h8455xygxIy0td/rcTHWEzssT2JV3FrnD1+dv56jr85v76X+zvfGdtyOykS1GOv+rO8xRqjC1vbDdoeOSIWTSCQSiU7QWuEsWLDggbz5qCya4gVmjx199NEAbL755kDJSrIuwqd4VCTmmKtgZCRWwzZ1ExbR3z1s9snU1NQM9RTz3yOTcB/m1cvSZKZm4MV4hucq+5O9+bmMRqYzaEfv2C1iLmM5nrNV1dYUactYNW9GnuPQrDOZbBd9wGbD1NRUdW0dfe2y7Dg2nSNnnnkmUM7ZOKQs28xM4xWxD9xHPvIRAHbddVdgZpwzdsiO4zWuk7TMMstUexc2obYWT4zdWGPmOcfYjcrDOWCMR2Xznve8Byi99dyudVYxEzCu8KmHxftMbS6N4gXQznFcaCM/95hUYSoPr/chhxwClG7RcdVcs9J873hzzsQejMb4PHfXV/Kaedxes2HXEkuFk0gkEolO0FrhLFq0qNoLaZD/QmFrBx98MFAyJozxGIew+lV/dazSlwnrtzTu0VQ9H783s6Ntr6RerzfDLx3XR/fz2OFa5uKKfDISs0n8v7VEKiKZiLVMz3ve84DCWOwbF2NFNUYSmfhcxnC0kUrFYxcy4I997GNAYb76n1W6oqaUJ41er1ddWyeqhNpaUdbVGI8Q9szSJ+/4eMELXgCUOeB+brrpJqDEPeL+YpfiWm81tzsMavaPY9Jjdl8xVqMCUan4f2O6ziGzyox7mpWm4pH1a+v11lsPKLbUg+P7Wkx6GNTmYW3FXbNZ/d54t2uCOX+9nh67dXtxdWVtYEd+FY4xPvej0o4rfM4We2pjj1Q4iUQikegErRTO1NQUSy+99Iw1wMWgWRzW16go7IUlu489laxJ0acr81ABxdzwpphM9CkP0wV24cKFLLfccjP6usXVJuMKjjHjRYbyne98ByhKQ5amEpLVuT3ZvZXjbtdzcfsxy8Ttu10zBeNxzgVkmjG+YBzL7DMVst2ivf7GNbrIuFocFsf6mrJ+VHUqWMe6HQNUujJcsxRjZwHH1XnnnQeUGEDsoRYVecwcm+QqqLX7R4ztxCp4j1UFtM466wDFU+L95JhjjgFKRpfbU0lra+MbqsHaWlE1FTIoFi5cWN1GjDd6DM77zTbbDJhZm2avRWO5dtR33DjPVX920Pe9HpGzzjoLKB4Tx5Hj1eMzY9RxmDGcRCKRSMxLtFI4vV6Pe++9txoHaGIAMgvZuXUTZlhZf2OnZD+3Gt+sERmufse2lcDRpzxMnv19993HXXfdNUPZeG4eq9tW8cim/L2ZNzKZGPPxVZ+qleXGLWQ42kIG5KvM12vk8dRqldraYZyQ9ZuhZS2JvvtYiyDrM94hW2ta42UuoP1ryibGVuyVZ63RpptuCpR+YsYdPUdr1WSmZi+pClSBMQ5aW/lzHLabmppiySWXnLFuTczUi4jZph6Lr/7f2iTrs2TzKhW7djhnPA7HiXNKL0VcMbZ2Pxs2zrlo0aJqZ/24Cq37dh473/VgeD3tjWaXDceZXijP1Ti4sV5h3Nz6Pu+5Hl9cIVav0nSPTZv7RSqcRCKRSHSCobLUIgb1bcq2vvCFLwAzV6O0I6n59QcddBBQ2LeMSF/tuPo9Dcvoe73eA4xCyACMJ+gnlqHIxoy9uL5JrMeIPZU22WQTAHbffXegZKHIjMzEkZH4f20W/eO+ypSmn1PXiOvb2NcpZkY5zmSusUZFVj/quuu1lWwHxfT9xrqFGvydrFsF6hxxtVrHleNCRSOT1bfuq+Mg1qDU1oAZJ3q93qzxtEEz3mIMxTFs3NG4hXPQ3nnaQsWjF8D/+X2c9zGztIZhx9X0/bmPqGwinJ/WLprB67w21uf19r7iuLEmKfaF81i++MUv9n3u/UPba0MVUy0rd1CkwkkkEolEJxiql1rbmEfMuIgrhfoUdbtmscXV73yKywJjrckkM2pmw9TU1AxfuGwqsqB3v/vdAFx99dUAnH/++UDxrcos9DevsMIKQOkObCaOzMYusb7qexVxbfTI2iYRpxm2Y0OMX9QYsGtzyNaE8Y0PfOADwOjnNs5xNKgtYmxFG6i2zj77bKCs9OpcOP3004ESt1DxeA7OlYhB+yCOE019ByOi4rCWxFoU7xv2i9PzoafEmjXXCJLFe7+oKahJ2WR67KdprR2PxXufnSNc6dNzO/7444HiFbIjv/Pf+0jspWcXaOOjtftEbS4Oa5NUOIlEIpHoBEMpnLYMMj7Va+ubmANu9pEMRDVw5JFH9v0/rv0xLF7/+tcDpX/QoOj1etXsHpmJisReV2YT2fHYXlfawAw9s5RkY7I31wqxvkKm28TKays6Rqy44ooPbLMtRmWC8f9efxmt/eYi7Mk3n9F2XSJ/r/KVkTourDVxXJjtGDO6mtZfGTTetfrqqz/AkseFQceLY9e6PKvjZefO21tvvRUocQ/jDt4/HNeyds89rqs1SB3fMFl8s22vyVvkMXmPM8vR+4jHbgcKaxnNYjQDL3piVL61LgqDXptll122VSfpVDiJRCKR6ARDKRwzHyKTGBSRxakG3G6sbrXDaW21ObNPfNK2Zdptlc10eA52cZVpeAx2TXBtl+233x4osRv7wZnNI3PSX20/KHuuyWhjtklkSjWfcO34/Z3+8LZYsGDBjGrpYRWPx2RNk9cn+pdlfaP0+ZoURs3+irazA7bdNqZ3cYaZNS3+3zoufx+7RdfYfKytq62a2wbOU491UJXgsdkrz7iF6s+aN1WeY13vgL3YjH+5Hfevd6CmNsyGc26Mo0bJY1OhxJhd9FjEuLeI3h3jm27Puh3/F7s/N3XliOPDGLX34nvuuSd7qSUSiURi/mEohaNfUNQyXkQtfuDnPqXjU13GYp+x+DSPfu6m7JdJZOK4zdip2M9VZ9GHai2A/mUVjauYmnUmq4o++qY+VPEaxN/HLLZRsWjRotZZi5FRyiTtCv3BD36w71gj9NnPRzSNsXh9auxaxurY15tgVlLs/xXXZVIJx/3W1kEatG5oUCxYsGDGPI3H0mSrqOLN9JRtW5cVbTedhQN8/vOfB4oqjPeb2vgdVvVHTM+Km63DBxS7x8+FNjNG43jw915Xe6kJzy3O+ybPSNxvzHrMXmqJRCKRmJeYavOEmpqa6kFhCIP6zmvdV+O+ZXN77703ACeffDJQ2L9dg0VNOfn0l+G0qIq9stfrrd30o/u32ZvOUmt2rPnytWHs0ivrijUBvsaVQ/2dMaGYlRKPo2aLyGx6vd7AQQjHRVu4z+h3Nn7gsXziE58A4G1vextQ2NuJJ54IlJUeJ4VhbDGsmo5dnGv1EVH5qHhiDHGYPoGLQ5fjItqwdt+p2diYkcoqsvtR623a2mKQ+8W03/f9Lh57vP6OmxiXMi6uh8TP/X+t839bmwxqi1Q4iUQikegEQymcGmuyI3LNb+y+4tNY6H+0T1hc72ZcWEyGRiuFc/8rUM/yqXXjFVZLa4vIbGu+9mnH0fd5U71NE3OxGroLJtsE+88Z2zE7yX5SsrZJ938bpy2aYmvOregViCtFxi4dMftsUhinLcygi3Uc8dwdy95fnAuxC3WTl6GtqmjCKLao3RNjDC0qVWGnAe+1Zv41jYcYD23KuIv3n9ni9W3uF6lwEolEItEJhlI4094DwzOHcfuXx8B0WiscMeq5NGWHiFqFeI3JxBUeY/ZR7fP5oHDmC9IWBfPRFnPRF+7+7Xdmi2Ezb+N9YpZYbd/vmu5fqXASiUQi8V+BtnU4dwO3+KZtJ9yIcXcsHrRP1WKwUovd9dli1HNp6nkV0bS/6Jut1VVUPm9jBwi2+B9D2qJgXtqi1g9swujUFm3vDyLeJ0a9F1dsPbAtWrnUEolEIpEYFulSSyQSiUQnyAdOIpFIJDpBPnASiUQi0QnygZNIJBKJTpAPnEQikUh0gnzgJBKJRKIT5AMnkUgkEp0gHziJRCKR6AT5wEkkEolEJ2jV2uZ/vTEhcHev13v0ID8c1Ba15pjzHfOxSeNcYZK26KrR5LiQ46JgErb4bxsPYlBbtO2l9r+Osfd9ig+acQ+ouL3aGuViUitBJoZDVzeWYbsN/y9g0G7Sc9V1ejrm63UYly3SpZZIJBKJTjAnCucxj3kMAH/605+AsnbLsKsWNq2i2AXiKqbxGB70oAcBM9cQb1IatXPzc9csd713V0kVrv/etM6O71deeWVuv/3vLl9zAAAXH0lEQVT2WY/lfw1dMNimFVhHPZbVV18dgGuvvXbW7Ylhuw2P00aD2mJYNK2+G1fRjBi06/SCBQtG9gwMatdos7hacfzeVVG9z9RWVfZ/2qTJ5T+uOZIKJ5FIJBKdYKQVPxfzu75Xn7quYX7nnXcChdV/6UtfAuC9730vAH/4wx+AOQm0D73iZ4SKxnOQkahEVHORKTXFZGRru+yyCwBXXHEFAC9/+cuBoniOOuooAP7yl78A8Mtf/rJvu03oMji89NJLA8Mr3ElBJtuFLWpKd1yxtlGVStuVHe//z0gXspZwM6hN9KT89a9/BcocXGqppYC6+m/C//3f//Gvf/2LRYsWTTyZpLZCZ/Rw3HXXXX2faxt/7zlrA3+n4vnXv/7V9/tBx1uu+JlIJBKJeYmhYjiXX345AC960YuAwqJFfBrLMP7xj38A8LOf/QyAffbZB4Dtt98egD/+8Y8AHHLIIQDstttuAHzqU58CmhVPZAUxU2sSzLmW2eK5RsiqRBOzdbuqgBVXXBGAZz7zmQB86EMf6vv/N7/5TaD4blWTteOUIf3ud78b4GwnA/3NbRHZ2bjRZQZfbV9tj8HrqjfBcRjnZFuMc+4MOh9r873JJs6du+++u++9/6vNzRpinESVMAqabNB0vfzcc3R7T3rSkwDYeOONATj33HMBuOOOO2bdr14FoaqM94mazduOi1Q4iUQikegEQymcddZZZ6DfrbDCCkB52j7ucY8DYJVVVgHg1FNPBcpTdqONNgIK4/32t7+92O3X/JtdZt6Mez11j01WZVbKox71KAC22moroKi/hz3sYUBhg9tuuy0AH/7whwH429/+BtQVlMpmLmoO2hbFRgXr//5/qCmK42H55ZcHYK211gLguc99LgAvfelLAfjNb34DFMV78cUXA/DjH/8YKLaaC5tNaoxpo5VXXhmAv//970DJ3By2vmYSWXXD2sCx7jk88YlPBEq8yqxF4+CPfexj+/6vLfQOOIfifaIWM8o6nEQikUj8V2AidTg+DWVX3/ve9wDYfPPN+34nI7ntttv6fved73wHgEc+8pF9r/pOfRrXal4iJsnaH/7whwMljtW2kjt+HmM3Mg7jFBtuuGHf99rALJXNNtsMKMqltt/I2sxiue+++yaWHRh94SocX83gu+eeewB49KP/02XIc7rwwguBcv2Nh8nWrOuaL1luwyBmD22xxRYAvOtd7wJgtdVWA0oWpL8Tjhe9C84pbaNCqsUx4nj1msHk6mfawvHi/USb7LnnngD84he/AEr2q3GMP//5z8Dg48M4mEpp2WWXBf4zR9rGgaBkdI0Cr4+ejbe+9a1AUbjf//73gRLrVdGYKex48ZxinDteYz93bjrnmmI7NaTCSSQSiUQnGKvC8al30EEHAYWRvuENbwCKGvApKuM466yzgKKIzGIzXqHCedazngUU1uZTWkYe/dKRTcjWxum39hxqGFR9RYbhe5msDOa8884Dio/+JS95CVCYjF0CBmVSKqphM8XaIMZgnve85wHlevj+t7/9LVBiesYpfJXBaqszzzwTKCx+VMxlTEjbONaPPvpooIwDFYrn7hz4+c9/DswcB9rk8Y9/fN9+br31VqA5BjlOVTOOWiAoccpjjz0WKBmc3gfMdl1uueWAcg4xbtHE0lXSjgf3M2xG5yDnHb0AHqMeCLNKP/nJTwLw4he/uO9Yt956awA+/elPA3DNNdcAxUNivDxmeLr9eB9YXLcFSIWTSCQSiXmKsSocn4aHHXYYUBSKLNyaEBnIm9/8ZgBuuukmoMRofHrqj3z6058OwHOe8xwAHvGIRwBw9tlnA4XlmZnRFLcYR0+nqakpllpqqYGVQY0ReCy1YzaOYX69jOWqq64C4Fe/+hVQmIksrFZ3E1Ve9NUPa5PF+acjk1TpqtY8Zln6qquuCsC6664LlFoD/db6p1/3utcBpeuCPvzrrruu79zaYhRlMzU1xcKFC1vXBsWuHJ7LQx/60L7fXXTRRQAcfPDBAFx55ZVA8bGbpeR4ce688pWvBMocfOMb3wgUZtxF3GvUfajyjz/+eGDmmNXz8fnPfx6Am2++GYCnPe1pQIlrPeUpTwHguOOOA4pNIuL2R61VGySGEzuKxJ54xqu8twrjVF/5ylcA+NGPfgTM7Knm9hyfzj3vqdbrxPtDvM8NG+dNhZNIJBKJTjCRLDUztlQgdrKVrcnKYiaWKsCnsU9fYzhW1+vDlYHI+sxuEjW2Po4OBL1er1XcI/ZGi4rDY/XYzIjZfffdgXJu2lJfrJ/LaGtsraZ43N+ovvpBbOj1jLEWla2fy2S9rrJ21dy+++4LlPFkrcEll1wCwDOe8QygHqeYJHq93lCdD/Spm5FnPY1ZQZ/97GcB+MhHPgLM7DvnnItdPV796lcDRemooPbbbz+gKJyIOF7sHzYXcC587WtfA8oc0gYqGLNd/V5FbPzKmKDvb7zxRgDOOOOMWffr3BhXR4tBxqG/iXEmr5/jw89V/yeeeCJQ4uZxfNQ8HcI5Z89F67Xifcl6wOkdDtrMr1Q4iUQikegEE1E4MlnZliwtrsEQ4dNUhiJL++lPfwoU/6VPb5/W+v5lIE0+eBnLOHoita2Wb8qcM/dfH77f33DDDX3/k6lafxN7pkXE7avOzH4apq6gLWr7iNdLJrvHHnsAcMst/1mI1ew1EbPeHD+veMUrADjiiCOA+joo82G1S4/BcSSTNQ519dVXA3DooYcCM33pNaWsN+DZz3523/df/epXgeaapfj5KHNlWDs7T816VZk4z1/72tcCRdlEj4lxLRWQrNwsR2O+TZgLZRevqzEbsxeFSsT4Va1TQETsYCLMuo33Ee/p0WbZSy2RSCQS8xITUTixOj6uVte0drjvZe8xh7wtK4+rcY5D2YiobAY9t8hArBkwz15bmYGnvznWHGkTf6/fWp9+jFfps5c9zta9etz1J6qxWrxAmL1m9tlTn/pUAN7//vcDJdPKmgHP/Sc/+QlQspf0yTf1gepC2TStbyRksqeffjpQmKvZRrEmJDJQr+caa6wBwGte8xqgjH3nktlM8bpPEjE2W0Ocp7Jq+wZ6zocffjgAp512GjDzOprZqRow63GHHXYA4O1vfzvQnHXWpfKtdXhw7hintI+lv1fZiFr3Fa+B94k3velNQInx6SVyvKmQVYP+L3aXbotUOIlEIpHoBBNVOD4V4+cR8Wkcf2eWmlXU5ox//etfB+or98W+ZJFdToLBNLHoyLpVNjIOK4ftlG1evZXjMtlYgWxsJmbqxXx+1aKvMQY1ier6JmXjsTlerCBX8ahIv/vd7wLFNvqbPWfHgefm9rpk8xFNYzuuGWWXDT/3XKw9MmPLOWBNmx3YrTRX6Wh7Vd/JJ5880HGPK741NTU1cAZk9BZYh2U2otuxjkZPh6zbbvR77703AJtssglQ4l7awv9H9j5Oz0cTap4OEY9p/fXXB4riMR4VO2HHe57zW89JVHvaxM79/s9xKOxGrcdl2LhWKpxEIpFIdIKJKJyYQ+5TN66r3QT9mGuuuSZQ1n6Q6cZc8wj3q083ZjmNE9Gn3sQMZR7GXOzG8OQnPxmAY445pm872s5+YrJ/e6rJiPRLaxv/pzKSEYtJdYaejkHZshkwcY0OWZVxqdjFN3YBdtzI+mVrXTLYiEHHh8zRmqOddtoJKHEtffle1zinZP0yYJWN6yMNqmAnGb+IcaiaArIzQOygbZdobfGyl70MgF133RUo1z3inHPOAcpKxfZuNFZo3KwLNHXlECqaGIM1y9D6O7enEjH+5X1F1ae3SFsY27NzSfR0eO90POZ6OIlEIpH4r8BYFU5TB9Eaw4z1FG5nu+22A+CDH/wgUNj5kUceCTSzc7/XzynGVTk8/Vh9rWWt1WAVvUzVLgxuR/a3/fbbA2U1Q335119/PVAUk8xHBRTXPI+92+LqmVNTU3O2cqYqzVoUq579PGbexVoCx4fsThvMVYX8dNR87DV4bqp7X60tidsxY9NY0Pnnnw+Ubg3aruvao9n2EzMnI/zc6vfYA9Ex/sMf/hAotog1Jf7PXo2HHHIIUGI3KuTYw7Epo3QSiJX/2kiF4xphxu70BujRsCZp0003BWDHHXcEZtrY+4H1WNb3WccXs12N+aiE4no62WkgkUgkEvMSY1U4kRlHxeNTPOaE+xT31a7QrulgJtfnPvc5oFRfN8H9iMjmx4Ha2vBRSUTWFNe5sc5GX6m+VjvaytrNyLrgggv6tiPri1X10dZmbNVUzCTUTRMD0lZmGb3lLW8Bii2OOuoooKg4WbxxKv+vb1+/9qWXXtr3/aArxI4TkWFqX69nbf0RP3esWy8hw4xru6ju7Klldb3/l8XPh64KtTkRPz/ggAMA2GabbYByHb3+junll18emFmv5+/M9NR7EDuaxLhmF4omIqo4x4ndm10XSU+I524mn/0DjfE6V9yOisRefHbi12YqZ7e7wQYbAKU3m/vVpiI7DSQSiURiXqKVwnnwgx/Maqut9gBzrKGWkROrVWMmhRkWrpOjsvHp/J73vAeYmdUSlVSMV9hb66STThrkNEdCzLzxnGJ9jD5Sj9lqeTPpZC76bMWvf/1roPhwrc9xv1tssQUA3/rWt4Bi61qF8FwyXiFL99XeWKeccgpQ/NQ1Bb3lllsCpQpb28gOVYsy2S5jOpHFi6ZO414vFYtxhwMPPBAodTbGMeyyYDcGx08NtVjOJFc7rdmiNgaNS+2zzz4AXH755X2/N77wmc98BpgZjzBecfHFF8+6/6ZzHLctFi5cOLB3JdYKeh9QsWob62v0XDgufO+9+m1vextQMjodf96HzPz0nunvVI96G0ZFKpxEIpFIdIJWCufvf/97o7qB8lSOq8zpJzR7yN+5xoNPbTvlqnBUJvqhY1abefj+z4wKY0Bvfetb+46vbc3MMJApqM70uVsjYHaaimT//fcHis2MyVxxxRVAiW/ZMddXa0xktmbqyO7POussoHntoek2nassNdWd67XXGKbHar8n42D6l13x0e+tVI99p+YzPEfHj750K8RlyjJVffiqPW3nOLNXlohj/gUveAFQYoNdKJ0mGDc488wzgcLa41owxnhit/BXvepVQImPXnbZZcDM+8ig8cxRV8UdRN3EWE7ct4pDxevnKl3vfao9PSeqfrMV9TZ5X1IpeW3MEBXOrRgXbzs+UuEkEolEohNMpNNARGQ0G2+8MVAYjKsbyur1H/p09qktjP1YIewKkNYq+HRWOb373e8GiirYa6+9gPH48mtqKa6kKZNYe+21gcJEYy2JTNZzMZ7h/1ZaaaW+YzcuEZWNfuvYVToeX1SLc6Fu3LdxrqbaJcfNeuutB8ATnvAEYGY/qFhFHTOAusSoKjpmo0XYN0wvguPO8eA4UxFFBm29jra307IrzI6C2K+vST35feyRGMeo2zOuGWtXXAvItaXsIxfVfdOaVoPWTo0TMZPOe6LdELSNnUbM7FXFey+1bs8ejapF7x/eZ6xV1AYqGv/vfvRO1cZhE1LhJBKJRKITtFY4bStLocRiZFVm2tgLSfYlc7V63qwjffD6qfVH+vS15sDtm3Fh/MMqfX24TQx6UCyxxBJVf642UtnYA8lzsS+cDEbVZ4xHBqKf2kw+bSSb+/KXvwyULBKZh9lr8Xji+1o9SJdo6rBdW1NIv7Psz+uvzW+++WagubJdeG1ircE44ByoMcO48qfXOfbQijUiziHnVKy7cC7FcdoUhzAeOo7OBLXuG03b9jrGztmxrs+YjPPe9XP8fewCEjPAmmIrtey6caDJBt4bHRdmveqtMRtVW9ilwzlhZq/z3BiOHhAVsd4A54w2u+aaa/qOb1hlI1LhJBKJRKITtFY4wzAdn6pmk9kXyIpyn94+pV/4whcCheHoTzbuYWxHZXPeeecBJRYkKzjssMOAEtMZdx+pf//731WF4L7i51/84heB4oN1HRzjEa5dLkORacTK4RNOOAEoDCTm1zfFKbT1XCqbJngdI0PV5ma1qXz1W/u/yOKbOvSOQ9lMTU2x1FJLzbBrEzOssWgzN+P1l5naBdqOE44r1xQatO5jEn3DllhiCR7+8Ic/EJuLMZyayopxR5WsLNyaFP9v/NJMO/+vzbVRzMisoWaLkTslz7KabtM2HUfWolmH4zHq+YirnPoaY77WpHlvdYVP79Hub1KqLhVOIpFIJDpBJ1lqMgvXsdl5552B4nuVxcnO9B+rXMw99+lu9okdcV0HQ8ZrNor1O/GpPU7UlE30O8tQ/f0ll1wClMrgjTbaCCi20hZmD5ktFLtIy3xi9lkTItubDx0Haqj57lW8Kh57Zal0BsXilE9bu/R6vcWqxljHENfw8TXGOWSkKhszLV3bxfGiDT7+8Y8DZW41jYt4nu5nlNVS//3vfz8Qb50O54LXtal7u/aMXcGFcU7Hgdffjsiu/TLsWkDjqkka5f9eB1WeikTvjePg3HPPBco90nnuvdFVko3xxrhlUzx1VKTCSSQSiUQn6EThCFmTtSZmahnbqfl6ZT5modjp1EweeyqZjXbaaaf1badL1NiRxyKbc40OYzr2iZLRyki0lezNc4690YZlT/NZ2UQ/spk5qj6VrxlcMftoXPsfBTVFE6+jdRSyddW/9TXGMVwB1Bo010WystyeWTLhYcfFKMomItY/xW3XlI1zxvikcyAi2tgefF/60peGPOJ+zFXnjdmOwXuk3h5tp1fH+KX1Mo6fd7zjHUCpp3EcattJZODNhlQ4iUQikegEU21Y3NTU1Fiooyxf/6FP7djnq9bLSNTWN2nblXYaruz1emsPcApMTU31hvHxC20ga5OpRFXm9lWHnpPM2P811a60Ra/XG5jyjGtczLLdvlcVrJ2RHQ+qQRXQuDGKLeJ1kIGqWOKxb7vttkC5vvYDMzPLOaG6M55hZqfKpu24MMvNmE8N4xgXg47N+Lvaipza5r3vfS9Qeihq20l1B5/LOWIcTFt4/bwnqpjtteg4acrUHPY+MqgtUuEkEolEohN0GsMRsQ+YaFqRs1YxXnv6Tjo+Mcj2awxBGzSxr9g1WGbjq77+CNn/XMSxxoWYweUqhDJds5fspTUfM+5qx+L1VLXZ1duaIn3r+ubN0LIS3K7Re+yxB1DYfFPNiLGjmEmnsplkl+imbsuxG0M8hzj/PVbjWHZQtpea+4m1cnoX/D5uN3YJ6XL9JKFqixmXMSasTY2LqaDtqjJoDVFX99BUOIlEIpHoBBON4YzqF5wDtIrhDLODtmtq1PqIRUQ/d1x5dFAss8wy/POf/2TRokVzHsMRMtTDDz8cgNe//vUAHHHEEQDst99+QMliNCY4xmy1sdsirhGl8rC+Qvi5WWlmscnWZcCDKpK5iO21jdlEdR7HfMxaNKbnWDeTz/3FKnptVVMwTT31ll56ae69995O5ki0Xbx/eA7aatz31CaVZxw7YziJRCKRmFeYkyy1uYJswYweq3KnYeIKJ/rIu1J7bdXmfMhSE8arZJ724rN3nuuy33LLLRPZ/yRZ/Sz/rx1D3/dz5R2Y5LioKRkRVZyxmtVWWw0oas9YzrjXeLIz97TuHhNTvm29Qk1r+oyKptheKpxEIpFIzCv8Vyic2loiE2B7Y1c4sTfWoB2M5wptfbL3/2d+ncSYMR9sMcnssUEwyXExV+c27H5zjsxEKpxEIpFIzCu0rcO5G5iMo3wxqK0lMgF1sFKL3w5ki7ZZYnON+23axg4wR+OiI8wLW8x1P69Jjou5OrcR+w/Oi3ExTzCwLVq51BKJRCKRGBbpUkskEolEJ8gHTiKRSCQ6QT5wEolEItEJ8oGTSCQSiU6QD5xEIpFIdIJ84CQSiUSiE+QDJ5FIJBKdIB84iUQikegE+cBJJBKJRCf4f4NTKbYNTpYsAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1a9c2a3be0>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "rows = 10 # split epochs into 10, so 100/10 = every 10 epochs\n",
    "cols = 6\n",
    "fig, axes = plt.subplots(figsize=(7,12), nrows=rows, ncols=cols, sharex=True, sharey=True)\n",
    "\n",
    "for sample, ax_row in zip(samples[::int(len(samples)/rows)], axes):\n",
    "    for img, ax in zip(sample[::int(len(sample)/cols)], ax_row):\n",
    "        img = img.detach()\n",
    "        ax.imshow(img.reshape((28,28)), cmap='Greys_r')\n",
    "        ax.xaxis.set_visible(False)\n",
    "        ax.yaxis.set_visible(False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It starts out as all noise. Then it learns to make only the center white and the rest black. You can start to see some number like structures appear out of the noise like 1s and 9s."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Sampling from the generator\n",
    "\n",
    "We can also get completely new images from the generator by using the checkpoint we saved after training. **We just need to pass in a new latent vector $z$ and we'll get new samples**!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1a9711cda0>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# randomly generated, new latent vectors\n",
    "sample_size=16\n",
    "rand_z = np.random.uniform(-1, 1, size=(sample_size, z_size))\n",
    "rand_z = torch.from_numpy(rand_z).float()\n",
    "\n",
    "G.eval() # eval mode\n",
    "# generated samples\n",
    "rand_images = G(rand_z)\n",
    "\n",
    "# 0 indicates the first set of samples in the passed in list\n",
    "# and we only have one batch of samples, here\n",
    "view_samples(0, [rand_images])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "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": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}