MNIST_demo.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# MNIST With SET\n",
    "This is an example of training an SET network on the MNIST dataset using synapses, pytorch, and torchvision."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Import torch libraries and get SETLayer from synapses\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "from torchvision import datasets, transforms\n",
    "\n",
    "from synapses import SETLayer\n",
    "\n",
    "#Some extras for visualizations\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from IPython.display import clear_output\n",
    "\n",
    "import seaborn as sns\n",
    "\n",
    "from scipy.stats import chisquare\n",
    "\n",
    "sns.set()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## SET Layer\n",
    "The SET layer is a pytorch module that works with a similar API to a standard fully connected layer; to initialize, specify input and output dimensions.<br><br>\n",
    "NOTE: one condition mentioned in the paper is that epsilon (a hyperparameter controlling layer sparsity) be much less than the input dimension and much less than the output dimension. The default value of epsilon is 11. Keep dimensions much bigger than epsilon! (epsilon can be passed in as an init argument to the layer)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Input batch shape:  (2, 128)\n",
      "Output batch shape:  (2, 256)\n"
     ]
    }
   ],
   "source": [
    "#initialize the layer\n",
    "sprs = SETLayer(128, 256)\n",
    "\n",
    "#We can see the layer transforms inputs as we expect\n",
    "inp = torch.randn((2, 128))\n",
    "print('Input batch shape: ', tuple(inp.shape))\n",
    "out = sprs(inp)\n",
    "print('Output batch shape: ', tuple(out.shape))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In terms of behavior, the SETLayer transforms an input vector into the output space as would a fcl.\n",
    "## Initial Connection Distribution\n",
    "The intialized layer has randomly assigned connections between input nodes and output nodes; each connection is associated with a weight, drawn from a normal distribution."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Inspect init weight distribution\n",
    "plt.hist(np.array(sprs.weight.data), bins=40)\n",
    "plt.title('Weights distribution on initialization')\n",
    "plt.xlabel('Weight Value')\n",
    "plt.ylabel('Number of weights')\n",
    "plt.show()\n",
    "\n",
    "vec = sprs.connections[:, 0]\n",
    "vec = np.array(vec)\n",
    "values, counts = np.unique(vec, return_counts=True)\n",
    "plt.title('Connections to inputs')\n",
    "plt.bar(values, counts)\n",
    "plt.xlabel('Input vector index')\n",
    "plt.ylabel('Number of connections')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The weights are sampled from a normal distribution, as is done with a standard fcl. The connections to the inputs are uniformly distributed.<br><br>\n",
    "\n",
    "## Killing Connections\n",
    "When connections are reassigned in SET, some proportion (defined by hyperparameter zeta) of the weights closest to zero are removed. We can set these to zero using the zero_connections method on the layer. (This method leaves the connections unchanged.)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sprs.zero_connections()\n",
    "\n",
    "#Inspect init weight distribution\n",
    "plt.hist(np.array(sprs.weight.data), bins=40)\n",
    "plt.title('Weights distribution after zeroing connections')\n",
    "plt.xlabel('Weight Value')\n",
    "plt.ylabel('Number of weights')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Evolving Connections\n",
    "\n",
    "The evolve_connections() method will reassign these weights to new connections between input and output nodes. By default, these weights are initialized by sampling from the same distribution as the init function. Optionally, these weights can be set at zero (with init=False argument)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sprs.evolve_connections()\n",
    "\n",
    "plt.hist(np.array(sprs.weight.data), bins=40)\n",
    "plt.title('Weights distribution after evolving connections')\n",
    "plt.show()\n",
    "\n",
    "plt.title('Connections to inputs')\n",
    "plt.bar(values, counts)\n",
    "plt.xlabel('Input vector index')\n",
    "plt.ylabel('Number of connections')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see these weight values have been re-distributed; the new connections conform to the same uniform distribution as before. (We see in the SET paper, and here later on, that the adaptive algorithm learns to allocate these connections to more important input values.)\n",
    "\n",
    "## A Simple SET Model\n",
    "\n",
    "The following is a simple sparsely-connected model using SETLayers with default hyperparameters."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "number of params:  23226\n"
     ]
    }
   ],
   "source": [
    "class SparseNet(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(SparseNet, self).__init__()\n",
    "        self.set_layers = []\n",
    "        self.set1 = SETLayer(784, 512)\n",
    "        self.set_layers.append(self.set1)\n",
    "        #self.set2 = SETLayer(512, 512)\n",
    "        #self.set_layers.append(self.set2)\n",
    "        self.set2 = SETLayer(512, 128)\n",
    "        self.set_layers.append(self.set2)\n",
    "        #Use a dense layer for output because of low output dimensionality\n",
    "        self.fc1 = nn.Linear(128, 10)\n",
    "        \n",
    "    def zero_connections(self):\n",
    "        \"\"\"Sets connections to zero for inferences.\"\"\"\n",
    "        for layer in self.set_layers:\n",
    "            layer.zero_connections()\n",
    "        \n",
    "    def evolve_connections(self):\n",
    "        \"\"\"Evolves connections.\"\"\"\n",
    "        for layer in self.set_layers:\n",
    "            layer.evolve_connections()\n",
    "        \n",
    "    def forward(self, x):\n",
    "        x = x.reshape(-1, 784)\n",
    "        x = F.relu(self.set1(x))\n",
    "        x = F.relu(self.set2(x))\n",
    "        #x = F.relu(self.set3(x))\n",
    "        x = self.fc1(x)\n",
    "        return F.log_softmax(x, dim=1)\n",
    "    \n",
    "def count_params(model):\n",
    "\n",
    "    prms = 0\n",
    "    for parameter in model.parameters():\n",
    "        n_params = 1\n",
    "        for prm in parameter.shape:\n",
    "            n_params *= prm\n",
    "        prms += n_params\n",
    "\n",
    "    return prms\n",
    "    \n",
    "device = \"cpu\"\n",
    "sparse_net = SparseNet().to(device)\n",
    "print('number of params: ', count_params(sparse_net))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Consider a fully-connected model with the same architecture: It would contain more than 20 times the number of parameters!<br>\n",
    "## Training on MNIST\n",
    "This code was adapted directly from the [pytorch mnist tutorial](https://github.com/pytorch/examples/blob/master/mnist/main.py)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "class History(object):\n",
    "    \"\"\"Tracks and plots training history\"\"\"\n",
    "    def __init__(self):\n",
    "        self.train_loss = []\n",
    "        self.val_loss = []\n",
    "        self.train_acc = []\n",
    "        self.val_acc = []\n",
    "        self.p_value = []\n",
    "        self.p_avg = []\n",
    "        \n",
    "    def add_p_value(self, p):\n",
    "        mem = 10\n",
    "        self.p_value.append(p)\n",
    "        if len(self.p_value) < mem:\n",
    "            new_avg = sum(self.p_value)/len(self.p_value)\n",
    "        else:\n",
    "            new_avg = sum(self.p_value[-mem:])/mem\n",
    "        self.p_avg.append(new_avg)\n",
    "        \n",
    "    def plot(self):\n",
    "        clear_output()\n",
    "        plt.plot(self.train_loss, label='train loss')\n",
    "        plt.plot(self.train_acc, label='train acc')\n",
    "        plt.plot(self.val_loss, label='val loss')\n",
    "        plt.plot(self.val_acc, label='val acc')\n",
    "        plt.plot(self.p_avg, label='p-value')\n",
    "        plt.ylim(top=1)\n",
    "        plt.legend()\n",
    "        plt.show()\n",
    "\n",
    "def train(log_interval, model, device, train_loader, optimizer, epoch, history):\n",
    "    model.train()\n",
    "    correct = 0\n",
    "    loss_ = []\n",
    "    for batch_idx, (data, target) in enumerate(train_loader):\n",
    "        data, target = data.to(device), target.to(device)\n",
    "        optimizer.zero_grad()\n",
    "        output = model(data)\n",
    "        pred = output.max(1, keepdim=True)[1] # get the index of the max log-probability\n",
    "        correct += pred.eq(target.view_as(pred)).sum().item()\n",
    "        loss = F.nll_loss(output, target)\n",
    "        loss.backward()\n",
    "        loss_.append(loss.item())\n",
    "        optimizer.step()\n",
    "        if batch_idx % log_interval == 0:\n",
    "            print('Train Epoch: {} [{}/{} ({:.0f}%)]\\tLoss: {:.6f}'.format(\n",
    "                epoch, batch_idx * len(data), len(train_loader.dataset),\n",
    "                100. * batch_idx / len(train_loader), loss.item()))\n",
    "    history.train_loss.append(np.array(loss_).mean())\n",
    "    history.train_acc.append(correct/len(train_loader.dataset))\n",
    "    return history\n",
    "\n",
    "def test(model, device, test_loader, history):\n",
    "    model.eval()\n",
    "    test_loss = 0\n",
    "    correct = 0\n",
    "    with torch.no_grad():\n",
    "        for data, target in test_loader:\n",
    "            data, target = data.to(device), target.to(device)\n",
    "            output = model(data)\n",
    "            test_loss += F.nll_loss(output, target, reduction='sum').item() # sum up batch loss\n",
    "            pred = output.max(1, keepdim=True)[1] # get the index of the max log-probability\n",
    "            correct += pred.eq(target.view_as(pred)).sum().item()\n",
    "    acc = correct / len(test_loader.dataset)\n",
    "    test_loss /= len(test_loader.dataset)\n",
    "    print('\\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\\n'.format(\n",
    "        test_loss, correct, len(test_loader.dataset), 100. * acc))\n",
    "    history.val_loss.append(test_loss)\n",
    "    history.val_acc.append(acc)\n",
    "    #Check p-value for connections distribution\n",
    "    vec = model.set1.connections[:, 0]\n",
    "    vec = np.array(vec)\n",
    "    _, counts = np.unique(vec, return_counts=True)\n",
    "    history.add_p_value(chisquare(counts)[1])\n",
    "    return history"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.manual_seed(0)\n",
    "#Optimizer settings\n",
    "lr = .01\n",
    "momentum = .9\n",
    "decay = .0002\n",
    "epochs = 50\n",
    "batch_size=512\n",
    "\n",
    "log_interval = 16\n",
    "\n",
    "test_batch_size=batch_size\n",
    "\n",
    "train_loader = torch.utils.data.DataLoader(\n",
    "    datasets.MNIST('../data', train=True, download=True,\n",
    "                   transform=transforms.Compose([\n",
    "                       transforms.ToTensor(),\n",
    "                       transforms.Normalize((0.1307,), (0.3081,))\n",
    "                   ])),\n",
    "    batch_size=batch_size, shuffle=True)\n",
    "test_loader = torch.utils.data.DataLoader(\n",
    "    datasets.MNIST('../data', train=False, transform=transforms.Compose([\n",
    "                       transforms.ToTensor(),\n",
    "                       transforms.Normalize((0.1307,), (0.3081,))\n",
    "                   ])),\n",
    "    batch_size=test_batch_size, shuffle=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Dealing with Optimizer Buffers\n",
    "Synapses recycles parameters. When connections are broken and reassigned, its parameter gets set to zero.<br><br>\n",
    "This system is designed to be computationally efficient, but it comes with a nasty side-effect. Often, we use optimizers with some sort of buffer; the simplest example is momentum in SGD. When we reset a parameter, the information about the overwritten parameter in the optimizer buffer is not useful. We need to overwrite specific values in the buffer also. To do this in pytorch, we need to pass the optimizer to each SETLayer to let synapses do this for us. <br><br>\n",
    "<b>Notice: I'm still working out the best way to initialize adaptive optimizers (current version makes a naive attempt to pick good values); SGD with momentum works fine</b>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "optimizer = optim.SGD(sparse_net.parameters(), lr=lr, momentum=momentum, weight_decay=decay)\n",
    "for layer in sparse_net.set_layers:\n",
    "    #here we tell our set layers about \n",
    "    layer.optimizer = optimizer\n",
    "\n",
    "#This guy will keep track of optimization metrics.\n",
    "set_history = History()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def show_MNIST_connections(model):\n",
    "    vec = model.set1.connections[:, 0]\n",
    "    vec = np.array(vec)\n",
    "    _, counts = np.unique(vec, return_counts=True)\n",
    "    sns.heatmap(counts.reshape(28, 28), cmap='viridis', xticklabels=[], yticklabels=[], square=True);\n",
    "    plt.title('Connections per input pixel');\n",
    "    plt.show();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "epochs = 500\n",
    "for epoch in range(1, epochs + 1):\n",
    "    #In the paper, evolutions occur on each epoch\n",
    "    show_MNIST_connections(sparse_net)\n",
    "    if epoch != 1:\n",
    "        sparse_net.evolve_connections()\n",
    "    set_history = train(log_interval, sparse_net, device, train_loader, optimizer, epoch, set_history)\n",
    "    #And smallest connections are removed during inference.\n",
    "    sparse_net.zero_connections()\n",
    "    set_history = test(sparse_net, device, test_loader, set_history)\n",
    "    set_history.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Test set: Average loss: 0.0757, Accuracy: 9772/10000 (98%)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "set_history = test(sparse_net, device, test_loader, set_history)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Node Connections Solution\n",
    "SET provides an elegant search algorithm for optimizing the connections between nodes. To demonstrate this, we can look at the number of connections per input pixel on the MNIST data."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "While the effect isn't super obvious yet, more connections start to appear in the center of the pixel map over many epochs. The researchers show in the paper that after 150 epochs, and especially after 5000 epochs, there is a clear preferences for those pixels where the handwritten digits appear (in the center of the input images).<br><br>\n",
    "From the paper:<img src='pixel_connections.png'>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "One problem, if you look closely, is that the algorithm takes many, many epochs to find an optimal connection solution. The choice to reset connections every epoch makes the search frequency depend on the size of the dataset.<br><br>\n",
    "One of the reasons I wanted to build this library was to experiment with other ways to apply this idea; PyTorch makes it easy.\n",
    "\n",
    "## Evolving Connections with an Alternative Search Mechanism\n",
    "An alternative idea would be to reset connections for each layer with some probability on each pass. The following is a quick experiment of this idea:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "class EvolNet(nn.Module):\n",
    "    \"\"\"\n",
    "    Resets connections with weighted coin toss on\n",
    "    each forward pass\n",
    "    \"\"\"\n",
    "    def __init__(self, evolution_proba=.05):\n",
    "        super(EvolNet, self).__init__()\n",
    "        self.set_layers = []\n",
    "        self.set1 = SETLayer(784, 512)\n",
    "        self.set_layers.append(self.set1)\n",
    "        self.set2 = SETLayer(512, 128)\n",
    "        self.set_layers.append(self.set2)\n",
    "        #divide probability by the number of set layers.\n",
    "        self.evolution_proba = evolution_proba/len(self.set_layers)\n",
    "        self.fc1 = nn.Linear(128, 10)\n",
    "        \n",
    "    def zero_connections(self):\n",
    "        \"\"\"Sets connections to zero for inferences.\"\"\"\n",
    "        for layer in self.set_layers:\n",
    "            layer.zero_connections()\n",
    "        \n",
    "    def evolve_connections(self):\n",
    "        \"\"\"Evolves connections.\"\"\"\n",
    "        for layer in self.set_layers:\n",
    "            layer.evolve_connections()\n",
    "            \n",
    "    def sample_evolutions(self):\n",
    "        \"\"\"\n",
    "        Each layer will evolve connections with\n",
    "        self.evolution_proba likelihood.\n",
    "        \"\"\"\n",
    "        for layer in self.set_layers:\n",
    "            #Flip a weighted coin to evolve connections.\n",
    "            if torch.rand(1) < self.evolution_proba:\n",
    "                layer.evolve_connections()\n",
    "        \n",
    "    def forward(self, x):\n",
    "        #Chance to evolve connections on each training pass.\n",
    "        if self.training:\n",
    "            self.sample_evolutions()\n",
    "        x = x.reshape(-1, 784)\n",
    "        x = F.relu(self.set1(x))\n",
    "        x = F.relu(self.set2(x))\n",
    "        x = self.fc1(x)\n",
    "        return F.log_softmax(x, dim=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "evol_net = EvolNet(evolution_proba=.2).to(device)\n",
    "print('number of params: ', count_params(sparse_net))\n",
    "\n",
    "optimizer = optim.SGD(evol_net.parameters(), lr=lr, momentum=momentum, weight_decay=decay)\n",
    "\n",
    "evolnet_history = History()\n",
    "\n",
    "for epoch in range(1, epochs + 1):\n",
    "    evolnet_history = train(log_interval, evol_net, device, train_loader, optimizer, epoch, evolnet_history)\n",
    "    evolnet_history = test(evol_net, device, test_loader, evolnet_history)\n",
    "    evolnet_history.plot()\n",
    "    show_MNIST_connections(evol_net)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This algorithm consistently shows stable convergence, converges slightly faster in terms of supervised objective loss, and appears to learn a network topology solution faster as well."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy.stats import chisquare"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2.975539612351204e-06"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vec = sparse_net.set1.connections[:, 0]\n",
    "vec = np.array(vec)\n",
    "_, counts = np.unique(vec, return_counts=True)\n",
    "f_obs = counts\n",
    "chisquare(f_obs)[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_MNIST_connections(evol_net)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX4AAAEBCAYAAAB/rs7oAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJzsvXl4HNWV9/+5VdXVq9Ta98Xy1sYr2AbMvhNClh+TmYQQEjIhM8lkkiGZLW9mMiH7m9n4JUNWQsgyZIUkhAQIAUJmAAMBY2O84LZly5K1763eu7b3j2rJkq2lJcuS3K7P8/ixVHW76t5S9bdOnXvuOcKyLBwcHBwczh6kxe6Ag4ODg8PC4gi/g4ODw1mGI/wODg4OZxmO8Ds4ODicZTjC7+Dg4HCW4Qi/g4ODw1mGI/wODg4OZxmO8Ds4ODicZSgzNQiFQv8J/CmwDNgQDof3TtJGBu4GbgAs4F/D4fB35rerDg4ODg7zQS4W/6+Ay4HWadrcCqwEVgEXAZ8JhULLTrl3Dg4ODg7zzozCHw6HnwuHw8dmaHYzcG84HDbD4XAf9sPi7fPRQQcHBweH+WVGV0+ONDDxjaANqJ/F593A+UAXYMxTnxwcHBzyHRmoBl4G0rl+aL6E/1Q5H3h2sTvh4ODgcIZyGfBcro3nS/jbgEbspw6c/AYwE10AQ0NxTHN22UIty+KJX+1naCTK/pX/Q2rPZXz0T8+lttw/q+OciZSWBhgYiC12NxaUs3HMcHaO+2wcM8xu3JIkKC72Q1ZDc2W+hP9B4C9DodAvgVLgJuwnUK4YAKZpzVr4AbZevIyf3vcyustH73CceFKb03HORM6WcY7nbBwznJ3jHh2zZejYAYOAECBkhBD2PssCLYWlpxEuLygqQojs9iRWcgRL18DUwTSwtBRWOo6VTiAkGdx+hNuHkBT7DJZpb1fcCEUFLKxMEiuTBEMHWbH/mSZWKoqVioKWts/r8oAk28dPxbAyCYSsgKwiFBdIsv1PyGDqWJrtnVHPuRLhCZw07lkwKxd5LuGcdwNvA6qAp0Kh0EA4HF4XCoUeA+4Mh8M7gPuBC4FD2Y99LhwOt8yu33Nn1TkVeEqhvHMlXZaFZpgLdWoHh3nHMg1byMzj32UjJWEZGkgKGNqYcGEZIMkIScbS0rbIJUewTB2huBGqB2Q1KzbZWA5TtwXM0LFMHQxt3M+6LUiWCaYJlgVY9v+WaZ83OWKLnWWNCZlQ3KB6beHDAl3DMjKgZ7D0DGhpLCxbUCUZLMsej56ZeBxZGfu5SxGkh/ox44OgpSZeJCGOj0tLgTXuOy8rCNWHlUnY41lMFPfYA2dKZBWl9hzkccJ/2rs1U4NwOHwHcMck228c97MBfGh+u5Y7QgjKNglST/soExqa7gi/w+RYho6VjGAlIuOENStspmELiGVlrUoJ9DRWOoGVjtn7ZRUUF0JItnhh2QJpGrY1aWqQSdkWop4B7ENZ46xDKx0HIWwRHLX8DFuALS11ssgB8QW8RtMh3AGErxDhKbCvj2nYD5z4kD1mLQlCQsgukF0Il9u2hBW3fc1MHUvPILCF25J9SLIEjF5Dez+mgelSkIqqkevWHT8f2H8jQ8u20xEuL8LtB5cbK5PETEYxI0MItx8lWIIUKLb7INkPFUtWMTMWZsYAYSEpAkm27MMLAQiwDPThYVJt7ViahruuDldVNZJLRY9FyXR2Y6YzqDV1qHWNSB4/RnSETHcHWl8vRjyNEU9gZjS8K1bgXXMOSsCHpWVId3SQ6ewASUby+JACBUhlTQv6d1wqk7unTKBKIh7ooyZWTCazyE95h0kZe/U2jazlp9hWpJ61DI0MZF/JLS2NlUnYIpl9bTaTUXokjXRKyx7QtK1GQ7PF1+2zhUlRMWMDWCO9mLEBwLLPhQAtefoHKmRQPbYVbHcUC4Ghu8lEBVoMlIAbtciDq0DFNASGYWDqBsLjRSrxI/v9WCaYGQ0ro+PxyKTSGkIykX1+lJJS5GAxSDJmPI4+EkUfSaCPpNCGRrB0HcnnRfZ5ELKEpWtYWvY6yQpCcSGUrJUtsm8DijJmkZupFGYqhZFMYiaTmImE/X86jZlOYGUiqJWVeJqW42lqwkyn0Xp70Yb7AJC8XiSfFzOeRh8aQh8awkwlwbTsh6CWwUylwDQRbg/+DRsInLcFtaqK9LE2UkePIkWH0RAINY7kMXEVF6MUFSM8HjI9HaTbe9D6epEDAZRgEZLPR6ark9TRFsz48UelUFWEJGXvQbDSJz9YR/ssFxQiFxZiRCJofb0T/6xuN5LHgxGJTPygLCN7fRix6MT2ioJwuYj8z9MAuMrL0YeGsPST9an2Y3+Hf/3GnG+xUyVvhF+WZAaqWmhoLmFkYAG+3GcJY24HLW27FSwLS9cwR7oxh7qwRnqwLPO4eJB1CZimLfKjop2IYsRjWLqJkEByZY0rssa2ln3jl2zdhOwb8qiu6xKmpYJQsCxx3DBDRggJSwgk0YcQGlgaluXF0BX0dCHoBqZhgmFgGj4s3cDSDTwNdRRs3oA/tBLhctn9lyT7wNm3AKGotjUpu9EGh0h3tpPp6MgKmS2OpqbZ7mfLwjIM29KLxzDTw0iqivB4QNcxYoOALTBmsvfkiz1LhMtlX2djohtBDhbZAhWNYibGvSsIYY/PyM0dLHk8SF4fks+H5PWiBINIHg/C7UEoCpnODiLPPcPw00+N9cdVVg6ShJlMYCYSCFVFKS7BVV6O7PXZ55cEkstlH9vjQevvI/bqLmI7Xj5+bq8Xb3UVWiqNmclgJpITxwK4yspxVVZiJhIkOjsx4jHUyioC523BXW9Hk5vJJGYygTXOZy77fMgFBcgFhWCaGLEYRjyGEY1iREfQR0ZQ6+oIXnkV3hUrkdxuUm2tpFtbMVMp1Npa3LV1yH4/me5uMl2dGLEorvIKXBWVqBWVKMXFSH4/WBbp9mMk9u4hdbSFwOatuBsbcdfW2w9kTQMEal3dbP70p0zeCL8kBEm//SSODTvCfyKWZWElhjH7WzEGWjGHu7FiA7ZlrKWQPAXgKcTQJdASkI5ipWLocQ09BWbGFmvZbXs7RrURdwAjJdDiGnrcwEibmBkLI23ZFj5ZdT9xskoIZL8HEBjxZNaXPB0mkDoudqP+Z0k6bs1NsKQyyMEgrrJyJLcbJWt9Saob4XEjhCD26i56H/wNckEhrrIyTM22ioUsIbk9CLcbM5VCHxrEGBmZ0Ec5UGALos+HpKp2P4QAWUGtqkEOBJDcblu00imwGHvld5WVY8SipNvayHR1Ift9KEXFyMEglq7b1nXKHqvk8SB5PBQX+RjoGcbSMrY4Za1oZBklWIRSVISrtBRXVTWy13v8767rmJmM3UdZPj7paZoT5hCwyG4zAQvJ4x27rtPeV4ZBpqcbyeNFKSrK6TOTUXHrbaSOHEYfGsJd34CrooKKyiB9fcetaDOdRh8exkwmcFVWTRjn6cZd3wCXnByv4mlaPv0HhcDT0IinofE09Wxu5JHwS2hqEh2L+PDkr3L5gmWZmEOdJKIZtL4B2yJPJ7AyCcxUDGNkBG1oBG04hj4Sx4gnMZIZMC17PkwF2e8Dlw8UH6bmJdM3hDbcg2XMNprgeNiZUlyMUlaGWlSEEiyyregsktuN5PMh+/yYmTT64CDa4AACYX+uuAThcmWtaPvBLRcWohQUIhcUoASLkAsLqawrmyAG4zG1DGY8gZlOoRQVI7ndk7YbpfydtxLfu4foC9sxEglkVUVSFCzTtPuRTiP7fLjr6nGV2FarWl2DWlWN5PHM8jpNRCkoRFm3Hv+69Tm195cXkPCXzPo8QlGQlYlfc/sBJSNkedbHO+n4soy7pvbUjyNJeFeumraN5HajVlae8rkc8kz4EZAQBsmRnBewLVksU8eM9GAOd0Mmgaml0QcGSBzYT/JoO5lhg/FzkxP+H48kkL0qsj+IkF1oiTT6UAzMBJAAbN+lp6ER33nLUKuqbJeFZlvPrgr79VUpLsYYiaANDNiWJpY9WScJXOUVqNU1yD7fab4q0yO5VKQiNef2QpIIbNxEYOOm09grB4elR34JP5CQDNLRNIZhIstLM+u0ZVlYmoaRSqK1HiBzaBda51G0aBIjrqEnNISlMeo2N9KgJ49HrEkeF96VywjUVpPW7IlLoajgciNkGaWgALW6BldVtf36PepMHz2/aWKm0whZtl/NZfmkNpMhe72olVWn45I4ODgsIPkj/FlfckIywIDhgQSlFQsXFzsZlmlgdIXJNO8ivv8giaM9ZIZSE0KOJyBA8btt37cosKPWdAtXVQm+mhrU2jp8obWodXUIISgvL5jS7TEdQpIW1D/q4OCwtMgf4Zdsf2VKNkCD/t74ggi/mU6TOPA68d07Sbe22JEB8QRmJnO8UdYFowRU/KtrkVwSQgKhyKi1y3Gt2IhaWWv7uefB7+rg4OAwHfkj/FmLX3dZkBH090QJrT/1iSA9EiF5KIzk9uBdHUJyu7FMk/irOxh+4hESR46BaS/+cPlBdoErCJIqIworkIqqUSoaCZy3FbWmJieXioODg8PpJG+EX2R9/LICkkehv2fuyZ20gX6G//A08T2vkeloP75DlnFXlaIPDWMkMkgu8Neo+FY04l27EaW8cWxVo/AG7eXnDg4ODkuMvFEmOSv8igzCo9DfE8eyrFlZ2KmWIwz+7nFiO3cA4KmroHB9NS7Rj5XRSEcM0kO9yG6Fom2bCF75ZuTqlY4V7+DgcEaRN8I/ZvHLAsslkxlMEo2kKCyaeRLT1DL0//IXDD/5OySPh4LVFXg93chqN6KgDKX+MuSK5YjCSqTCctuad8TewcHhDCVvhF/KCrGigCHbHv/+ntikwq9HhjFTaYQsoQ8P0/Pf3yfT2YG/qYhAyTCSN4J6zo24QpchglWOyDs4OOQVeSP88pjFD7oioQpb+JeHysfaWJbF4KO/YeDhhyYsv5fcEsWrwVsrUDfcjGvNlQjVCXd0cHDIT/JG+AWjPn6BZppUlvknTPCa6TTd3/sOsR0v418fwl2gY/S2gqnjbazAe/6bUFZe5EzIOjg45D15o3LSOItfS1uUVQToaB0CQOvro/Mbd5Nub6fo/NV4CCMUP65LLkVZeSFydWhsjsDBwcEh38k74Zdk0HSTssoAB/f10L9rD8Pf/xZYFmWXh3AlD+Bafx3uC292rHsHB4ezkrxRvtHJXVkWpA2T8mp71e7rP/gF9cEgJet8iMgB1AvejrrpRmfC1sHB4awlj4Q/6+qRbIs/0HcExUgzUr+Jyssk9FcfwnPF+3GFZlMD3sHBwSH/yBvH9nhXj26YxHftoDTTS79agbb/CeSGTY7oOzg4OJCPwi+BrunEX9tNbYWLRFxnKOHGvfVti9xDBwcHh6VB/gg/oz5+KI92YcbjLFtn547vLtiGXLa0Sp85ODg4LBZ55OO30xlLEjQOtyEUhQLRSrEcoNNctridc3BwcFhC5I/Fn43SkYTFilgb3lUrMZqfob5K0NOdJJ3SZziCg4ODw9lBHgm/PZSC6AjFWhR3XSkYOo2b12FZ0H50aJF76ODg4LA0yDvhr2jvBMBd4QegevUyVLdM25HBReubg4ODw1Ii74S/7FgHne5ShBFD+IqQXSr1TSUcOzKINS4xm4ODg8PZSt4IvxACX9KgsG+QZn89ZqwfqcDOzNmwvIR4LMNgX3yRe+ng4OCw+OSN8MtCor5bQwDN/jqI9SMKygAoq7TTNwwPJhexhw4ODg5Lg7wRfoGEL2UCMKL4EIkhpKzw+wIqAMl4ZtH65+Dg4LBUyBvhl4SEO2NiAX4lg8Aac/V4vC6EgLgj/A4O80I8mnbmzM5g8kb4ZSHh1iwMt4sS2fbli0Jb+CVJ4PWrJGKO8Ds4nCqxaJoffvOP7HyhbbG74jBHclq5GwqFVgM/AEqBAeC2cDh86IQ2FcD3gHrABfwBuCMcDi/IyikhBO6Mie52USrZlbekQNnYfp9fdVw9Dg7zQEfrMKZp8crzbaxaW0lhkWexu+QwS3K1+L8FfD0cDq8Gvg7cM0mbfwZeD4fDG4GNwBZgwTKjSULCkxm1+GNYQkIESsb2+wIqCUf4HRxOmc62YVS3jBDw/NOHF7s7DnNgRuHPWvKbgZ9kN/0E2BwKjatibmMBBaFQSALcgAp0zGNfZ8StWehulVIphu4OIiR5bJ/PcfU4OMwLnW3D1DYWs+XiRloO9juLI89AcrH464GOcDhsAGT/78xuH8/ngdVAF9AN/C4cDm+fx77OiDtjYbgVSqQYGXfxhH2jFr8zIeXgMHeikRQjwylqGoJsOr+OYLGX555qxjDMxe6awyyYz+ycbwdeA64BCoDfhkKhPwuHwz/P9QClpYE5n7y8vABPxiTtd1MqD2P6mygvLxjbX1FZgGWB3+fGH3DP+TxLjfFjPFs4G8cMS2Pcna0RANZvqqWyupA3/dkGfnzvS+z+YzvXv3XtvJc0XQpjXgxO97hzEf5jQG0oFJLD4bARCoVkoCa7fTx/A9weDodNIBIKhR4GrgJyFv6BgRimOXuLvLy8gL6+KGrGYkSWCEpJOoS9bRQza+kfax2ktGLuD5ilxOi4zybOxjHD0hl3eG83bo+CUKCvL0qw1MuGLbX88ZkWLMtiyyXzV/diqYx5oZnNuCVJzMlgntHVEw6He4FXgVuym24BdoXD4b4TmrYANwCEQiEVuBbYO+sezRFL13EZFoa9VoukUjRhv89v73AmeB0c5k5H2zA1DUUTLPtLrl3B6vWVvPTsUfbsWNBpPYc5kmtUz18BfxMKhQ5iW/Z/BRAKhR4LhUJbs20+BlwWCoX2YD8oDgL3znN/p8RIJADQs+8wCSU4Yf/o6l1ngtfBYW5EIymiEdu/Px4hBFfdGKJpVSnPPdVMy8H+ReqhQ67k5OMPh8MHgAsn2X7juJ8PA9fNX9dmhzkq/LIBGkSlE4TfsfgdHE6JzrZhAGobik7aJ0mC6/6/tfzk3pfZv7uLptVlJ7VxWDrkzcrdUYtfU0x0SyIu/BP2u1QZlyo7Fr+DwxzpaBvG41UoKfdPul9WJBqWl9B1LOJE+Sxx8kb4zYSdpkGTdIasAJpx8iSxz+8s4nJwmCudbRGq64umjdypbSxCyxj0dp19k7JnEnkk/LbFnxEaQ1YB+iQWh7OIy8Fhbhw+0Ec0kprUzTOe2kZ7f4dT6nRJkzfCbySzwi/pRChA0ycRfidtg4PDrDB0k+eebOaJX+2nvCrAqnUV07b3eF2UVQZobx1eoB46zIX5XMC1qJhx29WTkQxGROGUFv8xR/gdHHKiuyPCc08epq87yoYttVx01XJkZWZbsW5ZEa+93IGWMXCp8oztHRaevBF+I5HAkOxwzqhUOKXFn0kb6JqB4nJuSAeHyejuiLDjuVaOtQzh8bp4w5+sY3ko9yid2sZiXv1jO13tERqWl8z8AYcFJ2+E30wm0FQZU0jE5SD6FJO7YId0FhZ5F7qLDnmIYZj0dceIjaSIjaTxeF2s2Vi12N2aM63NAzz28714vC62XbWc9efVzNpqr64LIkmCjtYhR/iXKHkj/EY8gaZKWALSSgG6bpzUxut3AY7wT0VvV5RjRwbZdGE9Sg6v9Gc7mmbw6AN76DoWmbC9uj5IsHh295dlWbQfHaKmvignd8rp4tWXjlEQ9HDz+7fO2U3jUmUqawtpP+r4+ZcqefPtti1+CROwFHXScM7R5GyJmLbAvVvaWJbFnh0dPHT/Ll569ii/+clukgnnGk2HoZv87pf76G6PcOl1K3nH7Vt451+ejxAQ3tM96+N1tA7zyM/28NqO9tPQ29wY6I3R2RZh3ebZW/knUtdYRH9PjFTSuY+WIvkj/Ims8AsBsntyH/+Yqye90N1bsmTSOk8+/DrPPdVMfVMxV90Yoq8nxkP372J4MLHY3VuSmKbFU795nWMtQ1xxw2o2bKmltCJAcamPumXFHNzbM+v033tfsXPcvL67e9FSh+/d2YmsSJwzD66q2mV2WvQOJ7pnSZI3wm8k4uguCVOArCiTRvV4fHbRdSeW/zg7trdyJNzHtiubeOOfrWfNxireestG0imdh374KumUY7GdyMvPHeVIuJ+Lr1nBOZuqJ+xbvb6S6Ej6JPfPdEQjKY42D1Bc6iMylBxLjbCQpFM6B/f1sGptBR6v65SPV1FdgEuVaXfi+ZckeSP8ZiKBrkqYQkJR5EktfkkSeH0qibgjZqO0Ng9Qt6yY87Y1jK3IrKoNcsPb1pFKaLQedqornUjLwX7qlhWz6fy6k/Y1rS7DpcqE9/TkfLz9r3YB8Ia3rUN1K+zfPXtX0Ww5uK+Hxx7cQ3+PXZ/6wJ5udM1k/eaaeTm+LEs0rijhSLjPSd+wBMkL4bcsCyORQHdJWELCpUiTCj+Mrt51XD0AI8NJhgeTk0ZeVNUV4vW7aG0eWISeLV3SKZ2h/gTV9cFJ97tcMivWlHM43IeWOTnA4EQM3WT/7i4aV5RSXOpj9boKjoT7TqtvfM+ODn7/mwO0HRnkFz/YyUvPtLBvZydVtYWUV81fAZBVaytIJXWOtThW/1IjL4TfzGTAMNBVgSlJKLI0qasHwBtwORZ/lrYj9heyfhLhF0LQuLyUtiNDjsU2jt6uEQCqagunbBNaX4mWMWg52I9lWXS3R9j90jGMSYyRw+E+UgmN9VtsS3vtudWYhkV4b+5vDLliWRY7trfy3FPNNK0q5T1/vY1Vayt45fk2IkNJ1m+pndfz1S8vwe1ROLS/d16PO1dM0+LRB/bQ/Pqp92egL86uF9vIpPV56NnCkxfhnHrMfl3VFYEpxIwW/2CfM2kJcOzIIAVBD0Ulk4ceNq4s4cCebno6RqiZIUfL2UJPp518bDrLuLo+SEHQw84X29j9cvuYO6WrfYTrb1qLJB1PcrZ3ZyfBYi912cnQ0ooAFdUFvL67i41ba+etlKFpWmx/qpm9OztZvb6Sq24MIUmCq9+8hpVrK2g/OjSrRVq5IMsSK84p5+DeniWxirezbZi2I4MMDSRYHiqf8HeYLX/83xZamwfY/XI7265YTmhD5byXnTyd5IXFb2TTNRgugSUErmksfl9AJekUXccwTNpbh6hfXjzlDVu3rBhJErQedtw9o/R0jFBc5sPtmdpmEkKwZmMVQ/0JDMPk8jesYttVy2k52M8fHgtjWRappMZzTzbT0zHCus01E/4G55xbzVB/gt89tJ9HH9zDwz9+lcMnFbzLnXRK57Gf72Hvzk42XVDH1W8KTRC9huUlXHz1CmR5/uVg1doKdM1cEsVZRt88opEUR07hembSOu0tgyxbWUph0MMfHgvzm5++NqeysYtFflj88ePVt0whULIWv2VZJ4maz69imhappI7Xd+rRC2cKI8MpMmmdskq7PmfXsQi6Zk67slJ1K9Q0FNHaPMhFV61YqK4uWSzLoqdzJCfL+Lxt9SxbWUpphX/sHjR1k5eePUoykaG3M0omrbP23GrWnTdxQnXVORXsfaWD/p4Ybo9CMpHhoR/v4ub3b511xM3IcIrHfr6HyGCSK25Yzdpzq2f+0DxSXRckUOjm4P4eVq+vXNBzj8fQTY6E+1m1toLeriiv/rGdFWvK52Sltx0ZxDAsNl1YR3VdkB3PtbJjeytD/fEzpp53Xlj8o64ew2ULv0sWWIAxyRP4bKzElUxk+NWPdvHQ/bsY6LWv1bGWQSRJzJhmt3FlCUMDCSJDyYXo6pImMpQkndKprJnavz+KLEuUVQYmCMvmixs498J6jh0ZorwqwNvft4Urblh90ipplyrzjtu38u4PXcjb37eFN719A6mExvO/Pzyr/hqGyeO/2Es8muFN79iw4KIP9tvPqrUVtLcMLep3ru3IIJm0zur1lWy6oI6+7iidbbmH3I7nSLgfr99FVW0QIcTYA627Y2Q+u3xayQ/hH3X1yGABLsX2JU6VqA0geZYIv2VZ/P6RA6QSGi5V5ncP7SeT1mk7MkRVXRDVPf1LX+OKUgDH3cPxL3Yuwj8ZQgi2XdnErX91IW++eWPO1mFpRYCLr1pBeG/PrOLiX9neykBfnGveHBqbQ1gMVq2rxLLsnP6LxaH9vXi8Lmobiwitr8Tjc/HqS8dmfRxdM2g9PEDT6rIxd1lhkQevz+UI/0Iz6uM3sxa/Itt/kMn8/IECO21DNJJauA4uIrtePMaxI0NcfM1Krr9pLSPDSX730D4G++I0rJg5gVaw2EtRqY/WZieev6dzBNUtU1zmm/MxhBAUFnlm7WK4/LpVBEu8/O/jB9G0mcNEe7tG2PlCG6H1lSxbtbj1b0vL/ZSU+zmySMKfSescbR5gxTnlyLKE4pLZsKWWtsODDPbFZ3WsYy1D6JrJilD52DYhBFV1QXoc4V9Y9FhW+GUTE3BlX50ns/gDhR4kWTA8mH+ui0xa57mnmnn8l/t48X+OsOuPx3jpmRZWrCln3XnV1DQUceGVy8eSZzU05WYFNq4oobNt+IwNXZsvejpGqKguXJToDcUlc+UNqxkZTrH7j9Nbqrpu8vQjYXwBlUuuXblAPZyehuUldHeM5LS2Yb5pOTSAoZusWnu8iMz6zTUoLoldM1zLEzkS7sPtUU5ax1FVW0hkKHnGuJDzQ/jjcYSqAhaWACUbnTCZxS9JgmCxl+GB/Arp7Ggd4mf37WDvKx0M9sfZ/VI7L/7hCIVFXq584+oxsTr3gjqaVpcRLPZOWTT7RBpXlGKa1lmdd0XLGAz2xamsmb8FTrOlpqGImoYgh8PTR8jsfL6VoYEEV74xNG300UJS31SMaVp0Hlv4e6h5fy8Fhe4Jay88Xhdrz63m0L4eRoZzMwINw+Ro8wDLVpWeFAE1euwzxepfGnfFKaLH48g+P5Kpz2jxAxSX+hiY5SveUmbHc0d5+blWgsVebnr3eVTVFmKaFtFICn+Be8LkoRCCN/zJWgzj5IinqaiqK8SlyrQdGaRp9am5DSzLIhaLkEzGMM3ZW3+9vRKmufALynTd5JqbqvH6oLu7dcHPPzru8y4pJJ3S6ew8OmkcumVBYVmGzZeUU5/jG91CUFUXRFYkjrUMjc0bLQTJhMaxlkHOvbD+pPv93Avq2buzk10vHuOKG1bPeKz2o0M4LD1kAAAgAElEQVRk0gbLx7l5RimrKkCSBd0dkVP+jiwE+SH8sTiS14NE1Bb+MYt/8rjaolIfRw8NYBjmaYldXkg0zeCV59tYtqqUa99yztgimdE3m8kQQqAoubsrZFmirrGItiODk4bIzoahoT6EEJSUVCLLyqyPpSgS+hQP9FyxTIvIUBJZkfAHVKQc7oF4LI0qZyir9CNJC3/PjI5b1w0G+xIUFLrxZiPUxqNldCQrRnFJkqGhPkpKpq+Ru1AoikR1XXDBk7YdCfdhWbDynJOvg7/AzZqNVRx4rZstFzcSKHRPe6yjhwZwqTL1k0yUK4pEeVXBGTPBe2arXhYjnhV+i9ws/hIfpmkxMnzmT/D2dIxgmhZrz60+rSsj65eXEBtJn7KLLJNJUVRUiqK4Fm2lYyqlk8kYJBMaA/0JEjMs6DNNi3RKR1akRRH98ciyhCyLKedbtIyJEDKlZRVkMkvr/q5vKmaoP0EsunC5sppf76OoxEtpxeRuzfMubMAyrZwifDqPRaipD05ZKKeqtpC+ruikqTmWGnkh/Ho8juTxIAAT67jwT7F6t6jUtoTzwc/f2TaMEPZCmdPJ6EKv0fw+c8dCiMW77SzLIpnIoCgSJWU+XC6J2Eia/t440UiKTEaf8BAws28HumbiD5xsYS80QghUj0ImY2BNsk4lk9ZRFDtDrR3cvHQYDSmdrdU/MpzktR3ts15tH4+l6WwbZsU5FVMaGYVFHlavr+T1V7umnZhNJjIMDySomuZ7VlUbxDAs+rIpOnLBNC0OH8gtod98kh/CH4sjedSsxW+NTe5OZfEXldjheEN5IfwRyqsKZozHP1UKgh6KS30ca8ktrHOoP74oERwzoWsGumbi9btQXDJFJT6KSry43TKppMbwQJLBvjjRkRTplM7wYAJdMygs8sxLnvr5QHUrWBZkTri+pmmhaQbqEpnQPZHSCj9en2tS4bcse+L3xDoGmbTOow/uZftThxnqn9339cgBexJ85Tkn++THs/mihmwk1AF6OkcmfcB0t9sunKmyssLxCd7u9twWhqWSGo8+8BpP/Go/vV3RnD4zX+SH8MfjSKqKhJVdwDW98KtuBX9APeMrTGmaQU/XyLQ343xSv7yYzrbhGePIe7ui/Oy+HezYvvCToDORTGgICdye4yJ+/w/vw+tXKK0IUBj0oCgyqYRGZCiJoZsEi71Tiv6BA/v57Gf/Zdb9+OIXP8MvfvGzOY1BVWWE4CR3j5bRsSx7/1JECEHdsmLaW4bGxDUylGT775u5/xsv8vCPdvOrH73KK9tbsSwLy7L4w2PhsTfz2UYENR/opaTcT0nZ9NFrRSU+tl3ZRHfHCL/871384gc7T3o4dbVHkGVBxTTJ+XwBlcIiT05+/v6eGD///k46j0W48o2rqW1c2CSIZ7zwW6aJkUgguV1IgGlZ04ZzjlJU6jvjXT0drUOYhjVj2oX5omF5CYZhTVshytDNbCIyxtISLxUMwySd0vF6XRMiYr73vXvRNA1JEnh8LoIlXkorAgSLPRQUqdO+Ta1Zs5ZPf/oLC9H9MYQQqKpCJj3RLZVJGwiJRc+COR11y4rtuZXeOM2v9/LAd3ewd2cn5ZUFXPOWNaxeX8lLzx7lmScOsf3pwxwJ93PRVcvxF6izSrEQG0nR3T4yo7U/ynnbGrjtw9u47PpVpBLaWDK9UbraI1RUF07p3x+lqraQno7J3xrGH+uh+3dhmiY33XruSVXcFoKl+U44C8xUEiwLSVWyrh5zRosfbOE/tK/3lKNUFpOjhwcQgmn9jvNJdX0RiiJx7MjUIXmvPN/KYF+cplWltBwaYHggQfEMFtdCkUpoWJZdgnOUu+76NwA+9KHbEULiq1+9h7vvvgtZlmlrayWRSPD97/+Yz372X2hra0XTMtTW1vNP/3QnhYWF7Ny5g69//b+477776erq5C/+4j289a1v48UXt5NKpfjEJ+5k06Zzp+1XIpHgK1/5D15/fR8AN9zwJm699b0AfPe73+app36H221HnNx99z24XC7+9T/upOXIEdxuFw2Ny/jc575EOq2jqrOPlFpI6rIhpr9/5ACDfXGqagu57qa1YyvqV62twB9Q2fXiMfbv6mLFmnI2XVBHf0+M9tahnL+vo+khJovmmQrVrbB+cw1uj8JTv36dzrZhahuL0TIG/d0xNl14csW1E6lpKOLgvl6efjTMZdetPMloGB5M8Nuf78Vf6OamW88dyx220Jz5wp+wrXbJJSNMezpLlqZO2TBKUYmXTFonmdAW7eKfKq2HBymtCCzYIh1FkajJhnVORl93lJ0vtLF6fSUXXt5Ey6EBWg4NTCv82/d08dxrXTn3QQg7Vj0XLt1YzSUbbGvKsiySSQ3VLWcnPm3+/u//Dw899CDf/OZ38fmOp2I4dOggX/vat/F67UCAj370Hygqst+svv3tb/CjH/2AD33ob046ZyQSYf36jXzwgx/miSd+y7e+dTff/OZ3p+3n97//HUzT5L//+2ckEnE++MHbWb58JevWreeBB37Mww8/jt/vY2Qkiqq62b79WVKpBHd/5fv4AiqmlULXTUzDQg0s7a90oMBNcZmPwb44G7bUctHVyyeEVNv5jJYTKPTQ1xXl0utWIoSgpiHIof29RIaSY3N009H8eh9llYEpQ5qnY9mqUlyqzMG9vdQ2FtPbZUfO5RJAEdpQRTSSYucLbXQdi3D1m0PU1Nv3TSKe4dEH9iAkwZvevmFRdWdp3yU5YIwJv4SUshVBzo5qOou/uNS+eYYHEmek8Ou6SUfr0EkpfU83DU0lPHe4mchQcsKXyjRtf6zXr3LptStwe1yUVwU4eqifzRc1LGgfJyMZz2AaFr5gbn/rK6+8Zkz0AR5//BGeeOJxdF0jmUxRXz/5mLxeH5dcchkA69Zt4Gtf+8qM59qx4yU++tF/QAiB3x/g2muvZ8eOl7jggm3U1tbz+c9/mm3bLmLbtkvx+fysXLmK1taj3Pe9/+KcNZu48IKLsXQ7RFJ1L103zyhX3RgindKnTQm+fnMN5eUF9PXZk57VWfHsbIvMKPwjw0l6u6Jsu7JpTv1zuWSWh8o4HO7j0utXjk04V9XOLPySJLjg8iYalpfw1G8O8PCPduMLqJRXBoiNpEnEMrz1XZvm9ECaT8544R+1+IVLIGXDlkcNiOkt/qzwDybOyOpSvZ0j6LpJdcPCuHlGaVxZwnNPwdHmgQnFxrvbIwz0xrn6zWvGJk6XrSrj5WePkohlxrKinsglG45b5bkwlwVchmESj2Vwu5Wco598vuNfzN27d/GrX/2Cb37zuxQXF/PEE4/z61//ctLPqepxN5IkSRjG3PMbybLMPfd8jz17dvPqqzt4//vfzV13fZWVK1fxwx8+wMsvv8Tzzz/Hj37yHb78n9/F5/ecEQsS55LdtKjEi9fvovPY8IzppUcLrszGzXMiofWVhPf0cPTQAF3tI5SW+2f1Zl1VF+Qdt28hvKeH3q4ofT1RkvEM1771nDlnd51PchpJKBRaDfwAKAUGgNvC4fChSdq9A/gUILC9LteGw+H5Lx46jjGLXxFIWReAlDV6prP4A4VuFJd0xoZ0dh6LgICaBYroGaWwyM7x03Kwf4LwtxzsR5YFy8ctV29aVcrLzx7laPPAouSCHyWeXTDkn2Jlps/nJx6PTXD1jCcajeL3BwgGg2QyGR599Nfz2r+tWy/g0UcfZuPGTSSTCX7/+yf48Ic/RiIRJ5FIct55Wzj//PN57bXXOHLkMIWFhRQWBrniiqu48MKLuOmmG5BcGQqLzjwDJleEENTUF9HZFpnWz29ZFof29VJdZ5e/nCs1DUUECt0ceK2bns4RQnMoIqO6FTZsPV7HeCnNJ+b6CPsW8PVwOPzDUCj0buAe4OrxDUKh0FbgM8DV4XC4OxQKBYHTvkRvzMcvgcguWBECJCGmXMBltxEUlfgYHjgzs3R2tg1TVV04ISxxoWhaVcrOF9pIJjJ4fSqWZdFyaIC6ZcUTIkpKyv0UBD0cPdQ/L8KfTGRIxjWCJd6cLVsto5NK6vgC6kkFT0Z55ztv5Y47/gq328NXv3rPSfu3bbuYJ574Lbfc8jaCwSLOPfc89u/fd0pjGc+f//lf8OUv/zu33XYzAG94w41s23Yxvb09fPKTHyeTSWNZFqtWhbjiiqvYuXMH3/rW1wAwTYN3v/vPqalZvAfrQlHTEOTwgT6ikRSFRZO7SgZ64wwNJLj8DatO6VyjBWR2vWiv6J2PAIqlIvqQg/CHQqEKYDNwXXbTT4CvhUKh8nB4QuHKvwX+MxwOdwOEw+G5lbeZJWba9u8IxUIStuiYljltwfVRikq8Y8WzzyQOvNZN17EIF1w2Nx/mqdK0uoxXnm+jtXmQNRurGMiuet1y8US/txCCZatK2b+r85SLbVumRTyawTQtYiPpnHyklmURHUkjyWLaeZzbb/8At9/+gbHfP/nJz0zYrygKn/vclyb97ObNW7nvvvsBqK6u4dFHfz+278TfxzP+HD6f76RzAlRUVHLvvT/I9uG4i+uiiy7hoosumXI8+UrNOD//VMJ/cF8PkiRYsSa3MM7pWL2+ckz4T/fK+IUmF4u/HugIh8MGQDgcNkKhUGd2+3jhXwu0hEKhZ4AA8Evgi+FwOOdA7tLS2derDL7hKopryhGJA8jZWd2SEh+qS8KlKpSXT73gora+mOYDfRQX+VBcS39SzDRMnnzkdf74TAtNq8q47NqVeH0LPzFdVmZHS3QcHeaya1ax75VOhIAt2xrxBya6U849v549OzqIDCQ5Z2M1vb3SlJb3dMRjtuh7vAqppI6uGTOupE0mNHTNpLjEu2QXNc2G2V43SZKmvf/PBMb3v6w0gNfnYrA3Pum4LNPiSLifFWvKqW+YuchQLueurguSSmo0rVjYjJun++82n5O7MrAR+81ABR4H2oD/zvUAAwOxOVSqVyi//DLaHtyByFr8vf1RZEkwEk2NRQVMhuq1azU2H+xd8kWSTdPi8V/spfXwIBu21HLxNSvw+tRpx3c6aVhewv5XO+nsGGbfq51U1QZJJDMkkhPznfgKVNwehddeaaesOoBpmrOenLUsi9hICpcqU1zqp687yvBQkhJZmjQ18SixaBpZFiiqfMoZPRebuUxqm6a5aPfHfDA+qmeUqrogRw71TzqujtZhopEUF17RNG/jvupNIXTNWNDrONm4p0KSxJwM5lxMiGNAbSgUkgGy/9dkt4+nDfh5OBxOh8PhKPAwcMGsezRX9DSyZD/HTMtAkaVpo3rgzMrZ0/x6L62HB7n46hVcet3KaQVvIWhaXYZhWOzd1clAX3zKHOSSJGhYXkLrkcE5PNRtUkkdw7Dw+VWEsPMGmYY1Nmk7GVrGQMsYeP3qkvKtOpwaNQ1BopEUza/3nrTv0P5eFJfEspXzl+8/WOxd8kbhXJhR+MPhcC/wKnBLdtMtwK4T/PsAPwauD4VCIhQKuYBrgN3z2dnpsPQMSlb4NVPLzcdf6kNWpCVfNcc0LV7Z3kpJuZ+N59fO/IEFoLo+iMersOPZowA0rZ76y9a4spRUQptTIirLskjEMyguaSxG3aXK+PwukgltyrxBiXgGIbFkEqs5zA9rNlRRVVfIkw+/zu6X28e2J2IZDh/oo2l12ZJOWbFUyNXV81fAD0Kh0J3AEHAbQCgUegy4MxwO7wB+CmwF9mOnxf8dcN+893gqtDQBSQU0YloclyxNWYhlFEWRqKkPcqxlYYtDzJbm13sZHkxy/U1rl4z1KkmCxpWlhPf0UFrhn3KyDaBheTFCQOvhARpXz85PnU7pGLp5UoFyX8BNKqkTj6ZPWtBj6HZOHp9fXfQ3I4f5RXUrvOXmjTz1mwM8//vD9HdHiccydLYNY1mwdhHy3pyJ5CT84XD4AHDhJNtvHPezCfxd9t+CY+kZAh4ViBPNxLMW/8xpgeubinn+6SNEI6lTivs9XZimxY7trZSW+1keWlol3ZpWlRHe00PTqun75fa4qK4L0npogMbVs4u2SKd0JFmctHhGkuxInVg0TSatT1iYlUxkEAK8fsfaz0cUl8z1N63l+d8fZs8rHQSLvWy+qIFVayuWTF6opc7SX+aXK3qagGwLdywTy8nVA1DfZM/+L1Wrv3l/L5HBJFsvbVwy1v4oDStK2HJxA+s2z5w2onFlKQN98Vn5+S3LQssY2TTEJ4/d63MhyYJ4ND2WDdEwTJIJDbdHyTnW/7777kHTtJz7NZ65pmV2ODUkSXDpdSu5/WMXc8sHzueCy5sc0Z8FeSP8lp4hoNjCH9XiKLKENoOrB6C4zIe/QM25wMhCMmbtV/iXZAFnWZa44PKmnHIdNWYn3HQt98gUwzAxTWtKn62QBP6AiqaZZNI6qaQ2Vqxjsnq0UzGalnkydH36lAuLkZbZ4Thuz+KV8DyTOeNz9Yyhp1FdXtymSkyL4VLKcrL4hRDUN5VwJNyHaVpLyifc3R4hMpTk2reec8bf3EUlXgqLPLMKSRyt4DXdZJ3H6yIR1xgZTmFZoLgkCoPenNdlLFZa5ieeeJwHH/wJum4/cD784Y+xdasdBHf0aAv/9V//yeDgAJZlccst7+Etb3krfX29fOUr/0F7ux1Qd+21b+A973lfztfTwWGUvBF+S8+A4iZgBYhl4jmFc45S31TMgde66e0cWbDc9rlwtHnAnkRdceqLURYbIQTLVpZi6OaEB6x2cDta+JlJP6PrJm7TIq3KZMYd58QiF27TwtBNJFlCkgUZIAO4QpfjWj39CtfFSst84YXbuO66NyCEoK3tKB/96F/z0EOPoes6n/jE3/OBD/w1V199bfaYduGbz33uU1x00SV88Yv/AcDw8OwqUjk4jJIXwm9ZFuhphKISEH6imRjeHH38YFcFEgLaWoaWlPC3Hh6kpqHotNfTXSgaV5aSTPehZfSccgxZpoWQBDO960iSQKjyjO1mw+lOy9zR0c5nPvNJ+vr6UBSFwcEBBgb6iUQiGIYxJvoAwWARiUSCvXtf48tf/vrY9qI8TsrmcHrJC0WxDM2uzqG4KZACDKcjFMzC4vd4XZRXF3CsZZALLlt2ejubI5GhJMMDCdadlz/hadX1QY4c7iOdOi78rtWXTGqVG7pJrC9OoNA9YQ5hLitY58LpTsv8mc98ko985G+5/PIrMU2Ta6+9lEwmM2lbB4f5Ji8mdy3NXsEpFJWA6rfj+Gdh8YMd3dPXFSWVnFt0x3zT2jwAMK+rEBcbWZZwuWRSSX3Mfz8VowuzFmIxzmha5qk4HWmZY7EY1dV2NNSjj/56TPQbGhqRZZmnn35qrG0kMozP52P9+o088MCPx7Y7rh6HuZJXwo/LTYErQDQTQ5FFzhY/2IuMLAvajy6NsM6jzQMUl/mmXRh1JuL2KEiyIDaSmuCrtyxrwu9axkBIYk4J3WbLaFrmP//zdxGNnry6eNu2i6mtreOWW97GRz7yAUKh0Cmf8447/o5//ud/4Pbbb6Wzs4Ng0HYxKorCv/7rXTz88C+47babee97b+GFF7YDcOedn2fPnt285z3v4L3vvYVHHvnVKffD4exETFcNfgFZBrTMLUkbBKUR2r91B56rP8j/qhkean6UK+T38bsXuvjO/7kqp4gY07S4/+svUlrh5803b5z9COaRTFrne//1PBvPr+Oiq5ZP2W42yZyWCt3drRQFaxgZthfMeX0udM0gMmyn1y7K5tof6IujyILgCatyF8rVs9SYy7i7u1upqmo8TT06/ZyJ9/d8MMckbU3A0VzPkR8WfyZr8Su2xQ9gymkswMjxQSJJgg1baznWMkR/z9Sv/QvBsZYhTNOiceWZH80zGW6PgkuViUfTJOMZhgYSWKaFaZhEBpPomoGhm7jUvJiCcnBYcuSH8OsTffwApmRvm42ff915NbhUmVdfOjHx6MJytHkAt0fJqbjzmYgQgkChG9O0C6W4XDLFZT6CxV4Mw2Ro0K6K5iTbcnA4PeSF8JuZ0Spcxy1+XdjbZuPnd3sUztlURfP+XqKR1Px3NAdM06Lt8CANK0qW1GKy+cblkgkUuvEXqGOlFFW3QmGxFywLIezFWA4ODvNPXnyzLC0bBjfO4jeywj8bix9g41a7gPhrO9pnaDm/6JrB/lc7eeC+HXbFnxkSn+UDPr+KP+CeMAfjdisEi70UBD1n/GplB4elSl44Uc0xV4+bQNbi10QKcM3K4ge7yMeqtRXsf7WLrZc0nvZi5pm0zp5XOnjt5Q5SSY2yygDXvGXNksvEuZDky4I1B4elSl58w6ysqweXG1V2ocoqGZJAwawtfoBNF9RzcF8vr+3o4PxLl81rX0cxTYtdL7Sx++V20imdhuUlnLetnur6oGPpOjg4nFbyQ/h129UjFHuFZ4ErgIb9MEikp8+uOBlllQFWrCln1wttrFxTflrSvTa/3stLzx6lcUUJWy9tpKK6cN7P4eDg4DAZeeHjH53cRXEDEFD9mMJ2//QNJ+d0zEuvW4nikvnDbw/OuVbsdPT3xJBkwQ1/ut4R/TOQj3zkA2zf/uxJ27u6OnnTm65ZhB45OOROXgi/paVBSCDZ4X8FLj8ZkggBvUNzE36fX+XS61bS0zHCnlc65rO7AAz1Jygu9eV15I6Dg8PSJC9cPaaWBuV4dEhADdAe66K00DNn4QdYtbaC5v29vPS/LciyoK87Rld7hNqGIq64YfUp9XmwP071EsoEerby/e9/h5GRCHfc8feAnRfnXe/6U37+80fYt28P9977TTKZNIZhcNttt3PttW+Y1fFffPF57rnna5imSVFRMf/4j/9MXV09bW1H+eIXP0sqlcI0Dd74xrfwrne9h2ef/R/uvfebSJKMYej87d9+nM2bt56OoTucxeSF8Ftaesy/D7aPP5aJUVfsoecUhF8IweU3rOZn33mZZ59oxu1RKAh62P9qF40rS+ecQC2T1omNpCku883cOM/5Y9crvND1cs7thbATsebCRdXnc2H1lmnb3HDDm/ngB9/LX//1R1EUhSeffJxLLrkcr9fL6tVr+MY3voMsywwODvD+97+HCy64iMLC3FxzQ0ODfOELd/LVr36bpqblPPLIr/jsZ/+Fe+/9Ab/85c+59NLLxwqpjIyMAPCd79zDxz/+Sdav34hhGKRSc79/HRymIm+EH5d77PeA6ke3DEqLXex6/dSSrgUK3Pzpezdj6CYl5X5M0+LB773CM787RE19cE6hh4P9cQBKyp0aoYtNVVUVy5at4MUXt3PppVfw2GOPcMcdfwfA8PAQX/rS52hvb0OWFUZGIrS1tbJ+/Yacjr1v315WrFhNU5Odb+nGG9/KXXf9G4lEnHPPPY9vfONuUqkUmzdvHbPqt2zZyt13//9ceeXVbNt2McuXrzw9A3c4q8kL4TdPsPgDLltQg4UW8ZROLKkR8M49Hr9oXKIwWRZc+cYQD92/iz/+bwuXXb9q1scbzNaFLXGKQ3Nh9ZYZrfLxnI4kbTfe+GZ++9tHqK6uJR6PsWnTeQDcdde/cskll/N//+9/IITgne98G5nRvFCnyJVXXsP69Rt56aUX+eEPv8+jj/6aO+/8PHfc8fccPtzMK6+8zKc+9QluvvlW3vrWP5mXczo4jJI/k7vjXT2qvYjLH7AFYq6RPVNRVVvIhi217N3ZSXdHZNafH+qLoygShUWeee2Xw9y44oqr2b17Fz/96Q954xvfPDZXFI1Gqa6uRgjByy+/SEfH7HI4rVu3gcOHD9LaehSA3/72EVatCuHz+WlvP0ZJSSk33vgW3ve+v2T//n0AtLUdZcWKlbzjHbdw/fVv5PXX98/rWB0cIK8s/nGunqzF7/bZxTx6hhI0zXPI5AWXL6PlUD8v/k8LN916cjHt6Rjsj1Nc5nMWai0RPB5P1s3zGx544HiRlQ996CPcdde/cd993+acc9ayYsXs3u6Ki4v5l3/5HJ/97CcxDIOiomLuvPPzADz99JM88cTjuFwKQgg++lF7cvmb3/zamGspEAjwT/905/wN1MEhS14Iv6WlwX1c2EfTNkiqXU2r7xQmeKdCdSusWFPOnlc6MHQTeRYFQwb7E9QvK573PjnMnU984lN84hOfmrDt/PO38dOfPjRp+6997duTbq+uruHRR38/9vu2bRezbdvFJ7W77bbbue2220/a/qUv/edsuu3gMCfyxtUz3uIvyCZqSxkJigvcpxTSOR2VNYWYhkV/b+75+1NJjUQsQ7Ezsevg4LBI5IXwj8bxj6LKKqqsEtPiVBR56ZlnH/8olbX2W0ZP50jOnxmL6HFCOR0cHBaJvBB+S0sjXOqEbQUuP9FMnIpi72mz+AMFdj75ns7cy8MNORE9Dg4Oi0z+CP84ix/s1bsxLUZFsZeReIbkHJK15UJlTSE9HbOw+PviuFS7CImDg4PDYnDGC79lmliGNsHVA7bFH8vEqCy2XSrzHdI5SmVNIdFIikQ8k1P7wf44JU5Ej4ODwyJyxgs/4+rtjifgChDVbFcPzD1Z20xU1BTYx8/Rzz/YnzgtaZ4dHBwccuWMF/7RQuvjUzaAnbYhpsUpC9qLpHqGEqfl/OVVBQhBTn7+RDxDKqE5qRrygKnSMjs4nAnkFMcfCoVWAz8ASoEB4LZwOHxoirYhYBfwjXA4/A/z1dEpOaEIyygFagDd1BGKQaFfPW0Wv8slU1oRyCmyZ8iJ6HFwcFgC5LqA61vA18Ph8A9DodC7gXuAq09sFAqF5Oy+X81fF6dnzOI/ydVjW9Wx0xzZA7af/+C+HkzTmja//kCvk5xtqTHfaZl1XefjH/8YkUiEdDrN2rXr+Md//GdcLjtX1P33f48nn3wcISS8Xi/f+MZ3kCSJRx55mAcf/CkALpeLf//3L1NSMrfsrw4OMzGj8IdCoQpgM3BddtNPgK+FQqHycDjcd0LzTwCPAIHsv9PPmMU/0dVT5rW/NB3xLiqLvOxvPTESEokAACAASURBVLUsndNRWVPAvl2dDA0kKJ1G1Hs6R/AXuPEHnIieUUae307kuWdybi+EwMoxL3Pw0sspvPiSadvMd1pmWZb59Ke/QDBYhGVZfOELn+bRRx/mppv+jN/+9hGee+4ZvvWt/9fenYfXUd6HHv/OzNkXnSPpaN9tSyPvtmxjDDZbEmLAJISQAE7xA2mTQG/T5j43SW97m6W0T0rS0NzQksLNgglbmqZpQ2PAYJYAxgavgLexbEmWZFn7evZzZub+cY6FvEu2ZC3n/TwPjzmjOXPen87oN++88y6/wOVyMzDQjyzL7N69k6eeeoKf/ORn5OYGCIfDKIoy6t+JIIzVaGr8ZcBxTdN0AE3TdFVV29LbhxO/qqqLgU8C1wPfOtuBJsLJ9XZPr/FXZZXjtDj4sPsA+dkr2LqvnXhCx2Yd/z+okwO5OtsGz5/4jw9SWCKWWZxKxntaZsMweO65p9m+/R0MQ2doaAiHI/WcaevWt7jtts/icqVnj/X5Adi2bStr195Cbm4AAJdLNAUKE2tc5upRVdUK/D/gvvSF4aKOk5s79puEUL9MBMjOy8GR5z3lZ0uLF7Cv4xBfKEvdrCRlmZLT9hkPgYAHh9PKQG+EvHMcf2gwytBgjCuvyzvnPhdjPI91OXR2ylhGzGuUc80acq5ZM4klgnXrbuWllzZRWlpKOBxk2bJlSJLEP/3TQ6xZcy0/+MHDSJLE5z53G7qewGKRkSQJRZFOiQXglVde4MMP9/L44z/H7XazcePPaW5uxmKRkWUJRZHPeI8sgyyfeazTXejnp5NledqdH6eb7uW/WBMd92gSfwtQoqqqkk7qClCc3n5SETAbeCGd9P2ApKpqlqZpXx5tYXp6gmNe2DzRk5oWuX8oidJ1as+aGk817zTvJGi0A7Bz3wlcysT0n88r8tBwuJvOzsGz9tFv0LoB8PhsdHWNfqTveT8zzztux7pcDMO4pPn0J2I+/jVrrufHP36YZ555irVr16HrJmAyODhEfn4hum6yY8c2Wltb0PVU+U3TRNfNM8oyMDBIVpYfu91Jf/8gmze/SG3tPJJJg1WrVvMf//HvrF597XBTj8/nZ+XKq/n+9/+eW2/9DDk5ucNNPXb7R02CFxO3YRjT7vwYaTqe3+NhLHHLsnRRFeYLJn5N0zpVVd0L3A08nf53z8j2fU3TmoHAydeqqn4X8FyeXj3pfvynTdkAMD9XRZZkOpJNBHw+dmldXLO4eEKKMac2n9df0Ghu6KVi9pkP5TraBpAVibyCzKzBTGXjOS3z2rXreOutN1m//rNkZ+ewePFSYrFY+me30NXVyZe/fB8WiwWn08mjj/6Uurrl3HPPvXzta3+KJMnYbFa+//0fnZL4BWE8SaN5UKaqai2p7pzZQB+p7pyaqqovAN/WNG3naft/l7El/kqg8WJq/PH9W4htfRr3PY8gO89sP/+/ux8jlAgzJ3QrW3a28uM/X4PLMf6zUeu6wTOPvYvP7+TTZ5mf/z+f3oNpmNy+oW7cPnM61oja249RWFhx0e+fiBr/dHAxcV/q73qyTcfzezxcZI2/Cmga7WeMKgNqmnYIWHmW7TefY//vjrYAl0yxIlntSNazr2a1MDCP3x75PWurbGx+z+T9o92sml84/sVQZBavKOWd1xroaBukoPiji5CuG3S1B5m/tGjcP1cQBGGspv3IXWv1VZR+6Z/OGMB10sLAXAAGLS34PDZ2n9EDdfzMXVyEzW5hz/ZTl+jr6QyiJ41TLgaCIAiTZdonfkmxYs0+dw0+35VHgSuPfd0HqavJ48OGHmIJfULKYrNbWLCsmMbD3fT1fDRFRHt69k7RlVMQhKlg2if+0VgQmEt9fwML52QRTxrsa+iZsM9auKwExSKzd0Stv6NtCLfXhidLLK4uCMLky4jEvziwAN3UGbIdw+2wsOvwxDX3uNw25i0u4tCH7ezf0wakBm6JZh5BEKaKGbHY+oXM8lVQ6inmjda3WVJ9M7sPd5HUDSzKxFz3rrx+FoP9Ed7cXE8oGGdoIMrCZRPTjVQQBGGsMqLGL0kSHyu/hvZwJ4HyQSIxnd0TWOu3WGQ++Zn5VFbnsmvrMQBR4xcEYcrIiMQPsCx/Mdl2P0fjeyjN8/DclnqCkcSEfZ5ikbnxtnnMUgO43DYChWLgVqYSc/cLU03GJH5FVri+bDVHBxq5+YYsgpEEz245PLGfqaRq/l94YOWY51kRBEGYKBmVja4uvgKnxcG+0E7WXVXJ9v0d7JrAfv0niaQ/ta1evZyf//xx7r13PXfffTtvvPHqWffbuPFnPPLIw8OvBwb6ueWWjxGJRNi58z2+8pX7uO++9WzYcCdbtmw+6zFOr/2PfN3d3c3f/M03+dKXNrBhw5388pe/GMcoBeEjGfFw9ySHxcHq4ivZ0vwHvrn8BvYc9vDU5kPUlPnwus4+AEyYWNqH7Rz6oH3U+49lPv7aRYWoC0c3SluWZTZufJbm5ibuv/+PWbx4KdnZOafsM95z95/u7//+29x775+wZEkdiUSCv/iLB5g7dx4rVlw56mMIwmhkVOIHuL5sNdtO7OCf9z7Op6+7gyd/E+LZLfV85VPzJ7towiRat+7TAJSXV1JTo7J//4esXn3tKfuM99z9I0UiEfbs2UV/f//wtnA4RFNTk0j8wrjLuMTvs2fxjeV/xr9+sJF/O/YMS1au4d1tHayozaeuJm+yi5dx1IWjr5XD5Z2kbdOm54eXQ1y//h5uvPEmbr55HS+++HuKikoIhYIsXrwUgIcffoirr76G733vH5Ekibvuup14PHbGMRXFgml+VP54PLWQkGkaSJLEz372SyyWjPuzFC6zjGx8Djhz+fqyP6U2u5oD+h/Ir+rll5u1Ce3lI0xtmzalpmNuaWmmvl5j/vyF3HLLp9i48Vk2bnyWG2+8CYBrr72B99/fw69+9TQ33bRueO2FoaEhioqKkCSJHTu2c/x4y1k/p7S0lIMHDwDQ2NjAkSOpDgYul5vFi5fy9NMbh/ft6Ginp6d7okIWMlhGJn4Ap8XJ/YvupcCVj7e8jVAkwXNb6ie7WMIk0XWd++5bzze/+TW+8Y2/PqN9/6STc/dv3vwCa9euG97+wAN/xqOP/ph7713Pa69tOefc/evXb2Dbtq1s2HAnzzzzJNXVH61W9+1v/x2NjQ1s2HAnGzbcyXe+89cEg8HxDVQQGOV8/JdBJRc5Hz9c2rzdW5r/wH8e2cSVyud5fdsgX/vcIhbNDlz4jVPAdJyvfCrOx7969XJefvnNKb3WrZiPP3Ncjvn4M7bGf9LKwmUokoKjsI2iXBfPvHKYRHJiZu8UBEGYCjI+8XttHhYF5rGjczd3fXwWXf1RXnq3ebKLJVxGb7+9c0rX9gVhvGV84ge4qvgKQokwcdcJlqt5bNp2jO6ByGQXSxAEYUKIxA/U5lSTbffzTtt73HlD6qHcr187MsmlmqmkU7ozChMj9TuWJrsYwhQlEj8gSzJXFa/gUF89pi3MLasq2Kl1TeiCLZnKZnPQ399NMpkY9QhcYfRM0ySZTNDf343NJhb+Ec5OjBRJW1W0ghcat/BS06vcufJ2tu3v4F9/t4+v37WUqiIxpfJ4yc7OIxgcoLe3A8MY+0N0WZYxjMy7YxhL3LKs4HR68Hh8E1wqYboSiT8t2+HnExXX8fKx15mbU83X71rCQ8/s5uFf7eUbdy+lQkyrPC4kScLr9eP1+i/q/aKLnyBcOtHUM8K6qhupyirn2UO/xbCG+ObdS3HaFX74qz0cOT4w2cUTBEEYFyLxj6DICvfNX48kwS/2P0t2lo1vrK/DblP43lO7eOQ3H9DUPjjZxRQEQbgkIvGfJteZw/raOzg22MITB57D65F48IsruW1NFfWt/Ty4cSdPvHDwokYYC4IgTAWijf8s6vIX0TP7Zn539EVah47zxflfYO2VJQQqennx8Da2tx8k/vskf7JuPoosrp2CIEwvIvGfwycqrqPKV8ET+5/lh7sexSpbiOoxvFYPlsIgu7u38LPfI5K/IAjTjkj85zHHX8VfXfE1nj/6IrppsLJwGXP8Vbxy7A2e5yX29L7MTzeZfHndQmRJDJYRBGF6EIn/AjxWN+tr7zhl2ycrb8Cm2PhN/fPs7XuZp1+2cs+NtcNzswuCIExloo3iIl1ftpo7qj+Fkt3J211/4LdvNkx2kQRBEEZF1PgvwXWlV9MWPME77ODFQ+8Si+ssnhOgrMBDlli8XRCEKUok/ksgSRKfVz/DiVAHx+bs49V9LrbsSk3vUFHg5aufXUhOlpgvRRCEqWVUiV9V1RrgSSAX6AE2aJpWf9o+3wLuAnQgAfy1pmmbx7e4U49VtvClhRv4/o5HcCzdw3Lfajzh2Wza1sJDz+zm63cvJd/vnOxiCoIgDBttG/9jwKOaptUAjwKPn2Wf94AVmqYtAr4I/JuqqhmR8Xz2LP586Zcp9ZbwZs8r7JL+g2tvjBDOfZ8H//ATHtn5BP0xMeWDIAhTwwUTv6qq+UAd8Fx603NAnaqqeSP30zRts6Zp4fTLD0hNBp47jmWd0grd+fz5ki9x/6J7AXiz8zUs+a1gjaD1HebBd37Ekd6mSS2jIAgCjK6ppww4rmmaDqBpmq6qalt6e9c53rMBOKppWuv4FHN6kCSJhYF5zM+tJZQI47G66R6I8ovX3qXJ+To/2v0Ydc6PUWGvJRrXsSgy19eVYLcqk110QRAyyLg/3FVV9Vrg74BPjPW96dXiL0pe3lSbNjk1F3p+fhY/rL6VbQcX8uiOX7BbfoUd3TtIts3GGAjwYWMv3/rjlXgvshfQ1It74mVizJCZcWdizDDxcUsXWgUp3dRzGMhN1/YVUg94qzVN6zpt31XAr4FPa5q2ewzlqAQae3qCFzX52XSZqzyuJ3i1YRtvdbzFQHyAXGsBnUcC5Jiz+PodK8bcA2i6xD2eMjFmyMy4MzFmGFvcsiydrDBXAU2j/YwLtvFrmtYJ7AXuTm+6G9hzlqS/Avg34I4xJv2MYVOs3FR9DQ9e9Zesr/0sVhsoFfsZKN/Ed157jCfeepOWziGxJKEgCBNqtE099wNPqqr6baCPVBs+qqq+AHxb07SdwE8AJ/C4qqon33ePpmkfjm+Rpz+LbOHq4pVcVXQFLUPH2dKwnd3mXnYmfs97O1/FPlTFdeUrWbusGodNDLUQBGF8XbCp5zKpJAOaes4nYSTZ2ryHN5q30ZVsxTQkpMEC6nKX8fH5CyjPzTljLqCZEPdYZWLMkJlxZ2LMcHmaekR1coqwyhauq1zBdZUr6Ah1sunwW+xhD7v1Tez+YBMYMg7ZzbWlq1hXfQOyJKZZEgTh4ojEPwUVuPP54tLPktA/xbbmD9l/vJXm3m769S428zI7T+znqyv+iDwys8eDIAiXRiT+KcyqWLmmqo5rquoA6OoL869vvUS7cwd/+87D1GkrWVW8hNr8cjEltCAIoyYS/zSSl+3iW7d+hjcOLOC3R59np7SVXX1bkfZ4yFfKmRuYzdWz5lHsy5gB04IgXASR+KcZSZK4fn4N1877X3SEB3l+71aOJA7RLh2io+8Ab+z6bzAsKIYdKw4KHcXcveRGSrMKJrvogiBMESLxT1OyJLGospQi983AzcQSCd5trGf38cP0JPqIGWFiZoRGeR//sONDSu3VXFGucjx4gpah48T1OLP9VVT7Z1HmLcEiW5AlGafFgdd28SOoBUGY+kTinyHsVivX1Mzjmpp5p2zf09jKrz54mRb9CK1H6rHhpCKrlHyXnf09h3i3fdcZx8p3BqjOnkWNfza1uTV4rO7LFYYgCJeBSPwz3NKqUpZU3sd7Whuvvd9EfWOUD5CoLfezdu4NlJWZDBp9GKaBYRoMxoc40t/I7s4P2Nr2HhISVb5yarLnICER1+PE9BjBRIiheJCoHmNh7lyuLbuaLJvoZSQI04FI/BlAkiRW1pawsraE7v4I7+xrZ9uBDp7afBhZkijNc5PQDSKxJLoh43PPpci7CJc/jDW7my79GC81vQqATbZiU2x4bB68Vjduq5vNx15nS8ubrCys44ayNRS6p9fzhMH4EA7Fjk0Ry2UKmUEk/gwT8Dv51Ooqbr26kpbOIO8e7KClM4jDZsFpU5BliYFgnP5gjIbjOpFYFlbLYqpLryY7y4XXYSPLaWNFbT65vtSkch3hLl5rfpPt7bvY2vYetdnVXFO6ClmSaR5spSV4nFAigm7qGIaOnr67MEyDEk8RN1V9nBJP0aT8Pur7Gnj0/Z8jAQsD86jLX0SJpxirYsEmW08ZKGdTbGLgnDAjiCkbprGJjjupG9S3DrC3vhutuY+hSIJQJEE8aSBLEstr87hxRTll+R6sFpmheJCtbe/x1vFtwyuOSUgUuPPx2bwokoIsySiygowEksTBnsNE9Sh1+YtYmr+IaDJGJBnBMA0cFjtOixO31UW23U+2w09pYS7tHf3E9BgGJm6La3gMg2EaHBts5USonUWB+Xhs53820TBwjH/Z+1P8dj/V2bPY2/khwUTonPvnOrK5pepGVhQuvewXgEw8xzMxZrg8UzaIxD+NTVbc3QMRXt3Vyh/2thGN6wDYrDJuhxWP04rbqSB5u8nL8rKgsJLq4lx8HvtZjxVKhHmt+U1eb32bmB6/4GdbZQsJIzn82m1xUejOx211c7S/kVAytQicQ7Hz8fJrub5sDXEjTn3fURoHm/HZsijzliBLEo9/8Eu8Njdfq7sfv92HbugcHWiiL9pPwkgQNxIYpgGAaZrs6nyflqHjFLoLWFW0HJtsQ5FkvDYP1dmzcFombqXRTDzHMzFmEIl/1MQJMjkisSQ7tU76g3FCkQThaJJgJMFQJM5gKE53f5ST32ZBjouFVTksnJ1LScCN027BblOQ07X1cCJMb7Qfp8WJy+pAlhQiyQjRZJSheIi+WD/90QFMm44RA7vFDqZJR7iL9nAng7EhqnwVzMtVCThzeOXYG+zt2odNthI3EkBqVtTkiItGriOb/1n3ANkO/6jiNUyDvV37+H3DZjrCpy4+J0sylVnl1GbPYY5/FlW+8uFnBjE9TsJInNE7Kq4nONLfQJWvAqfl/Gsx5OV5OdLaxr6eA0SSUeyKHbtio8JbSoE7f1Tln24m+/yeLCLxj5I4QaamaDxJc0eQxhODHGjq41BzH4mkcco+TruC22HF7bCS63NQWeilqiiLikIvHqf1jGOOJebGgWbeaXuPPFcuNdmzKfOUEElGaQ220RnuZmFg7qiT/kiGaRBORtANA8PU6Y70cKi3noO99TQPtWJiIksyAUcOQ4kgkWQUgMWB+Xyi4noqskrZ0b6H/27YTF8sdbG7rvRqri9bnW6uaqE12EbSSCJJMqZp0hRq4lDXUUzO/Puo8c9mTekqFgXmYZE/emwX1+NofUcYiA2S5wyQ7wrgs2eNazNVQk+w7cQOsh1+5uWoKPLFLyPaPNRK82ArVxQuw6ZYp/z5PVFE4h8lcYJMD/GEzuGWfnoGo0RiOtF4knAsSSiSJBRN0NEbpqMvMry/z22jOOCmONdNXraT/GwnVaXZdPUESSYNZFmiOOA+6wVisoQTERoGmjg60ERnuBuf3Yvf5iOiR3n7+HbCyQhZNi+D8SHKvSVcX7aGvV37eL9rH7IkDzctna7cV8KCnHkszVtItsNPTI8RSUb5oGs/b7e9S2+0D0VSKHDlUeQuIKJHqe87ekqzGIBDcVCZVcYsXwUlnqLhRG2YJqFEmFAiRDQZxW6x47I4cSh2kqZOwkigmwY1/tkUewoBaBlqY+OB52gPdQDgs3lZWbQcWZI5NthC82ArNsVGpa+cqqxyKrPKKfOWYFPO/L7e79rPE/ufIWEk8dt9rK38GJ9aeD3N7V10R3oJJUI4LI7hAYandx3ujvSg9R5hYd68S+pWnNATfNC9H6tsZUFg7qQ8zBeJf5SmWwIcLzMx7lA0QdOJIVo6gxzvDtLWHeJET3j4WcLZ5GTZKch2EU/ohKJJEkkDtdxPXU0e86typsxi9tFkjK1t73Kw9zBXFi6jrmDxcGJpC7az/cROsuxeKrxllHlLcFjsGKaBaZoUFvjP+V0bpsHB3sMc6W+kLdjOiVA7iqwwP6eWBYG55DkDdEd66Ix00Ro8QePAMdqC7We9e4DUA/lz/Qyg0F3ArKxy3m3fjcfqYn3tHeimwTtt77G/5xCSJFHsLqTcW0pUj9I02EJvtA9INYmVeIqY46tibq5KtX8W20/s4NeHf0d5VilrK27gleY3aBg4dt4LYaG7gAW5tRS7C9nZsZeDvYcxMXFZnNw+Zx1XFi0/ZeJC3dCp72/gQI9Gwkikm8rsuKxO3BYnTqsTre8I20/sJJRIPScq8RRxc+XHmeWvpL7vKIf7jhI3EiwKzGd+ropNsRHX4zQPHWcwPkSFt4xcZ/ZZv5+dHXvZ0b6H+YFariq64qwXv5NE4h+lmZgARyNT4jZNk6FIgq6+CFgUQsEYVkUioZsc7w7S0hGksz+Cw5ZqNjKBA429hGNJLIqEw2bBokhYFBmn3YLLbsFpT22TJAlJgiyXjYDfSZ7PgcdlxW5VsFkV/B7blFgFbby/62gySlekB9M0MTGRJAm3xY3H5sYmW0kYCcLpZywW2YJVtqKbOvu6D7Kr832O9jexJG8Bd9Xefsqzi2AiNDzWY6SB2BDHBptpHGymaaCZxsFjJIzk8HOXhYG53Df/C9gVG6Zpsr/nEK2xFuyGi4AzB4/VTUyPE0lG6Yn2crDnMPX9Deimjs+WxdXFV1CTPZvnGzbTMNBEtX8WxZ4i4nqcUCJMfX8DkWQEi2zBrtiIJWMkzVMrE7Ikszgwn9UlVzIYH+LFpi10hruHf+5Q7CiyQigRxiZbyXMFOBHqOOXilG33M8tXQZm3hFJvMRISzx99iWNDLXitHoYSQbw2D9eVribfFUBCwqbYmJtTPVwJEIl/lDIlAZ4uE+MebcxJ3UBr6edAYy/RuE5SN0jqBpGYTjiWJBxNYpgmpmmiGyYDoTixc9xV5PkdlOZ5yM1yYLelLgh2i4zNpmC3KjhtFjwuK16XFb/bjt02/ncYU+271g39ktrzTz7YPtCr4bF6+ET5tWcc70IxR5NROsJdlHqKRzRZGWxte5dNDa+QNJPDD8GrsipYlDeP2pwa7OmLUtJIEklGCSVChBIRAs5cfPaPmol0Q2dX5/sMxAapzp5FmacEIDWyvesDeiK9lHtLqcwqw2fPonGwmYb+JhoGjtEX6x8+js+Wxadn38SKwqUc7W/kpabXONRXf0osf7b4T5ibWzOquEcSiX8K/VFcLpkY90TFbJomwUiC7oEooWiCWNwgntDpHojQ0hWitTPIQChGLG5gnOdvRgLys52UFXgpST9/cDlSdxkn7yIcNoVcn2NMTVDiu55egokQrUNtDMaHWJy3YPhic1JPpI+YHsPERJEUCkf0zBJLLwrCZSJJEl6XDa/r/NM2nLxDiCV04onUxSEcSzIUTjAUjtMzGKWlI8ix9kF2Huo877H8Hhv5fidelw2Xw4LHaSUny0Ge30FulgPdMAlFEoSiSSwNvXT1hojFdfweOxWFXooDLhRZjCSeijxWN7U51ef8+dmeBVxOIvELwhhIkjT8vMB9/q73JHWDcDTVYykS04kndGIJnUgsSddAlM6+MF39Udp7wwSjqVHRSX30d7xWi4zbYUE3THTdxO20UBLwUBRw4XfbMU0Tw0zVCl12Cy5HaloOm03BYVVw2FIXG5tVFiu4ZRiR+AVhglgUmSy3jSz36CZ/M02TwVCcrv4o3YMRrIqcGgntsFJS7CM0FMVmkekeiHKsY4hj7UNEYkkURUaRJYbCcY53h/iwoQd9DE2mFkXG67KS5bLh89jwOK1YFAlZllEkCatVxmaRcdgslOS5qSrKOqMLrWGY9AxG6eyL4LRbKMp14bSL9DJViW9GEKYISZLweez4PHbm4DvlZ3nZLrqSqYfPxQE3xQE3q+YXnvU4Sd0gGteRJQlZBt0wiURTYyYisSSxhE4skZqNNRhJpEZbh+MMhhIMBOMc7wqlBqcZqWateNI4Y+BdwOfAYVMwTNB1g57BGEn91H2yvXZcdguGaWIYqZ5Dipz6z+20UpDtpCDHhdthJRJPEo0liScNdCO1f7bfSUGWg1klWWRdoAlOGBuR+AVhhrEoMh7nqW3/bselDXIzTJNI7KOR2M0dQyR1E0kCRZZYUh2gMMdFQbaLSCxJW0+Itu4w8YSOJEvI6Zakk81SQ+E4Ow51EoqeOsBMkSVkWUKWJBK6MdzZI9trT03zYZWxKjIJ3SCeSF0k/B4buT4HPredvqEoHX0RegajFPidzC7xUVWUhdvxUarzuGzkZqWOpxsmvUMxuvojWGSJPL8Tv9eOLEkYpkk0piNJ4LApM6o5TCR+QRAuSJYk3A4rcyuymVtx4QeTS8kb1XGDkQThWBKX3YLDpmBRPrpgeX1Odu07wdHjAxzvTj3YjiV0EkkDt9NKtldBlqAvGGN/Yy+DoQTZXhv52S4WVOZwojfMyztaztnsZbcpJBJn9tKyKKmmrUgsOTyMzWqRyXJZ081XqbEfVouMz23D77XjdlhIJFMXIxMoyHZSHHCT73eS0A2i6ZHq0bhOJJ4aZOh2WPG5beRk2cnPdo3q9zVeROIXBGHSeJzWc0654bBZqCnzU1M2uvmUTNM8o1aeSOq0doWIJ/T0PjAUSdA7GKVnMIrDppDnc5Lnd6IbJp39Ebr6IiSSRqobrsOCacJgODXxYCT20R1KPGnQ2R/hcEs/4WgSm1XBZpUxDPOMO5kL+cZdS5hbmTOm91wKkfgFQZgRztYUY7UoVBVljfoY8y/ys0+/6AyF45zoCdPVH8FmVXDaUr2oHLbUOA6rRSYYTTIYjBGN68wp9Z3n6ONPJH5BEIRLdPpF5+SYkPPdrfg8dkoC518saKKIcZLk4QAABa5JREFU0R+CIAgZRiR+QRCEDCMSvyAIQoYRiV8QBCHDiMQvCIKQYUbVq0dV1RrgSSAX6AE2aJpWf9o+CvAIsBYwgYc0TfvZ+BZXEARBuFSjrfE/BjyqaVoN8Cjw+Fn2+QIwB6gGVgHfVVW1cjwKKQiCIIyfCyZ+VVXzgTrgufSm54A6VVVPH5N9J/BTTdMMTdO6gP8CPjeehRUEQRAu3WiaesqA45qm6QCapumqqralt3eN2K8cODbidXN6n1FLryRzUfLyvBfeaQbKxLgzMWbIzLgzMWaY+Lin1MhdsfTi2GRi3JkYM2Rm3JkYM1z00otjMpo2/hagJP3w9uRD3OL09pGagYoRr8vPso8gCIIwyS6Y+DVN6wT2AnenN90N7Em344/078CXVFWV0+3/twG/Gc/CCoIgCJdutL167ge+qqrqYeCr6deoqvqCqqrL0/s8BTQA9cB24EFN0xrHubyCIAjCJRpVG7+maYeAlWfZfvOI/9eBB8avaIIgCMJEECN3BUEQMoxI/IIgCBlGJH5BEIQMIxK/IAhChpkqA7gUSA1GuFiX8t7pLBPjzsSYITPjzsSYYfRxj9hPGcvxJdMc+0jZCbAaeGuyCyEIgjBNrQHeHu3OUyXx24EVwAlAn+SyCIIgTBcKUATsAGKjfdNUSfyCIAjCZSIe7gqCIGQYkfgFQRAyjEj8giAIGUYkfkEQhAwjEr8gCEKGEYlfEAQhw4jELwiCkGGmypQNF01V1RrgSSAX6AE2aJpWP7mlGl+qquaSWuhmNhAntdjNVzRN61JV9UrgccAJNAF/lF41bcZQVfU7wHeBhZqm7ZvpMauq6gB+BHwciALbNE378kw+11VVXQf8HSCl//tbTdN+O5NiVlX1h8BngUrS53J6+zljnKj4Z0KN/zHgUU3TaoBHSSWEmcYEfqBpmqpp2kLgKPCQqqoy8DTwP9Lxvwk8NInlHHeqqtYBVwLH0q9nfMzAD0gl/Jr09/2t9PYZea6rqiqRqtjco2naEuAe4Mn0dz2TYv4v4BrS5/II54txQuKf1olfVdV8oA54Lr3pOaAuvebvjKFpWq+maW+M2LSd1ML2y4Copmkn5+h4DPj8ZS7ehFFV1U7qZB+5sttMj9kDbAC+pWmaCaBpWkcGnOsG4Ev/v5/U9C0BZlDMmqa9rWlay8ht5/teJ/I7n9aJHygDjqeXfTy5/GNbevuMlK4FPQA8D5QzovagaVo3IKuqmjNJxRtvDwJPa5rWNGLbTI95Nqlb+u+oqrpTVdU3VFVdzQw+19MXuM8Dv1NV9RipmvEGZnDMI5wvxgmLf7on/kz0z0AQ+JfJLshEUlV1FbAc+Mlkl+UyU4BZwB5N05YDfwn8FvBMaqkmkKqqFuCvgE9rmlYB3Ar8mhkc82Sb7om/BShRVVUBSP9bnN4+46QfDlUDd2qaZgDNpJp8Tv48ABiapvVOUhHH07XAXKBRVdUmoBTYDMxh5sYMqe80Sfr2XtO0d4FuIMLMPdeXAMWapm0FSP8bIvWcY6bGfNL5ctiE5bdpnfjTPTn2AnenN91NqqbUNXmlmhiqqn6PVPv2bZqmnZx+dRfgTDcFANwP/PtklG+8aZr2kKZpxZqmVWqaVgm0Ap8E/pEZGjMMN129DnwChnt15AOHmbnneitQqqqqCqCq6lyggFTvtZkaM3D+HDaR+W3aT8usqmotqe5O2UAfqe5O2uSWanypqjof2Efqjz+S3tyoadpnVFW9itSTfgcfdW3smJSCTqB0rX9dujvnjI5ZVdVZwC9IdeFLAP9H07QXZ/K5rqrqF4D/TeohL8B3NE37r5kUs6qqjwC3A4Wk7uJ6NE2bf74YJyr+aZ/4BUEQhLGZ1k09giAIwtiJxC8IgpBhROIXBEHIMCLxC4IgZBiR+AVBEDKMSPyCIAgZRiR+QRCEDCMSvyAIQob5/zAT/fKxIX75AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Epoch: 101 [0/60000 (0%)]\tLoss: 0.030194\n",
      "Train Epoch: 101 [8192/60000 (14%)]\tLoss: 0.039383\n",
      "Train Epoch: 101 [16384/60000 (27%)]\tLoss: 0.047205\n",
      "Train Epoch: 101 [24576/60000 (41%)]\tLoss: 0.052771\n",
      "Train Epoch: 101 [32768/60000 (54%)]\tLoss: 0.041035\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-18-17351e847ffc>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      7\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      8\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mepoch\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mepochs\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 9\u001b[0;31m     \u001b[0mevolnet_history\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtrain\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlog_interval\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mevol_net\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdevice\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtrain_loader\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moptimizer\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mepoch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mevolnet_history\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     10\u001b[0m     \u001b[0mevolnet_history\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtest\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mevol_net\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdevice\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtest_loader\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mevolnet_history\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     11\u001b[0m     \u001b[0mevolnet_history\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mplot\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-7-8324a1c401e7>\u001b[0m in \u001b[0;36mtrain\u001b[0;34m(log_interval, model, device, train_loader, optimizer, epoch, history)\u001b[0m\n\u001b[1;32m     36\u001b[0m         \u001b[0mdata\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtarget\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mdata\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mto\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdevice\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtarget\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mto\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdevice\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     37\u001b[0m         \u001b[0moptimizer\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mzero_grad\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 38\u001b[0;31m         \u001b[0moutput\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     39\u001b[0m         \u001b[0mpred\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0moutput\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmax\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mkeepdim\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;31m# get the index of the max log-probability\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     40\u001b[0m         \u001b[0mcorrect\u001b[0m \u001b[0;34m+=\u001b[0m \u001b[0mpred\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0meq\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtarget\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mview_as\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpred\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msum\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mitem\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/miniconda3/envs/learn/lib/python3.6/site-packages/torch/nn/modules/module.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m    487\u001b[0m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_slow_forward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    488\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 489\u001b[0;31m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    490\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mhook\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_forward_hooks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    491\u001b[0m             \u001b[0mhook_result\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mhook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-13-c2e08156f57e>\u001b[0m in \u001b[0;36mforward\u001b[0;34m(self, x)\u001b[0m\n\u001b[1;32m     40\u001b[0m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msample_evolutions\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     41\u001b[0m         \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mx\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mreshape\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m784\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 42\u001b[0;31m         \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mF\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrelu\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mset1\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     43\u001b[0m         \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mF\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrelu\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mset2\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     44\u001b[0m         \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfc1\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/miniconda3/envs/learn/lib/python3.6/site-packages/torch/nn/modules/module.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m    487\u001b[0m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_slow_forward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    488\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 489\u001b[0;31m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    490\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mhook\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_forward_hooks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    491\u001b[0m             \u001b[0mhook_result\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mhook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/synapses/synapses/SET_layer.py\u001b[0m in \u001b[0;36mforward\u001b[0;34m(self, x)\u001b[0m\n\u001b[1;32m    258\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    259\u001b[0m         \u001b[0mk\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mx\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0minds\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 260\u001b[0;31m         \u001b[0mk\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mk\u001b[0m \u001b[0;34m*\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mweight\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    261\u001b[0m         \u001b[0mz\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mscatter_add\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mk\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0minds_out\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    262\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mz\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbias\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "evol_net = EvolNet(evolution_proba=.2).to(device)\n",
    "print('number of params: ', count_params(sparse_net))\n",
    "\n",
    "optimizer = optim.SGD(evol_net.parameters(), lr=lr, momentum=momentum)\n",
    "\n",
    "evolnet_history = History()\n",
    "\n",
    "for epoch in range(1, epochs + 1):\n",
    "    evolnet_history = train(log_interval, evol_net, device, train_loader, optimizer, epoch, evolnet_history)\n",
    "    evolnet_history = test(evol_net, device, test_loader, evolnet_history)\n",
    "    evolnet_history.plot()\n",
    "    \n",
    "evolnet_history = test(evol_net, device, test_loader, evolnet_history)\n",
    "    \n",
    "show_MNIST_connections(evol_net)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "evolnet_history.val_loss.pop(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "epochs = 500\n",
    "\n",
    "for epoch in range(1, epochs + 1):\n",
    "    show_MNIST_connections(evol_net)\n",
    "    evolnet_history = train(log_interval, evol_net, device, train_loader, optimizer, epoch, evolnet_history)\n",
    "    evolnet_history = test(evol_net, device, test_loader, evolnet_history)\n",
    "    evolnet_history.plot()\n",
    "    \n",
    "evolnet_history = test(evol_net, device, test_loader, evolnet_history)\n",
    "    \n",
    "show_MNIST_connections(evol_net)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "keys = [key for key in optimizer.state.keys()]\n",
    "zeros = torch.zeros([2])\n",
    "inds = torch.tensor([0, 2]).long()\n",
    "optimizer.state[keys[0]]['momentum_buffer'][inds] = zeros\n",
    "optimizer.state[keys[0]]['momentum_buffer']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "another_optimizer = optim.SGD(evol_net.parameters(), lr=lr, momentum=momentum)\n",
    "another_optimizer.state_dict()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Experiment!\n",
    "This software is here to encourage experiments! I hope to see curious deep learners use the synapses library to create cool things; and I REALLY hope this inspires contributors to step in and optimize this promising algorithm!"
   ]
  }
 ],
 "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": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}