mxnet/mnistLenet.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#  Introduction"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This notebook is aimed at helping people learn about creating Convolutional Neural Networks (ConvNets) in [Julia](http://julialang.org/) using [MXNet](http://mxnet.io/), specifically the Julia interface [MXNet.jl](http://github.com/dmlc/MXNet.jl). \n",
    "\n",
    "It assumes a basic knowledge of Julia as well as supervised machine learning principles such as training vs eval and test sets.  It follows a [previous notebook](https://github.com/ultradian/julia_notebooks/blob/master/mnistMLP.ipynb) on using Multi Layer Perceptron to identify MNIST characters with MXNet.  There is some overlap, but you should know everything in the previous notebook before exploring this one.  In particular, we will be using the same dataset, but I will not be reviewing any of the information on the background of the data or details of its processing besides around its use in ConvNets.\n",
    "\n",
    "Data is obtained from https://www.kaggle.com/c/digit-recognizer/data and assumed to be in a `data` folder. Files are named `train.csv` and `test.csv`.  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Check data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "using DataFrames"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " 12.023967 seconds (35.65 M allocations: 1.177 GB, 2.61% gc time)\n"
     ]
    }
   ],
   "source": [
    "# read in data\n",
    "@time train = readtable(\"data/train.csv\");"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Plots.PyPlotBackend()"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "using Plots\n",
    "pyplot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<img src=\"data:image/png;base64,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\" />"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# take a quick look with Plots\n",
    "plot([heatmap(rotl90(reshape(Array(train[i,2:end]), 28, 28)), aspect_ratio=:equal, legend=false) for i=1:16]...)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<img src=\"data:image/png;base64,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\" />"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "histogram(train[:label], ticks=collect(0:9))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Prepare data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "42000-element Array{Int64,1}:\n",
       " 1\n",
       " 0\n",
       " 1\n",
       " 4\n",
       " 0\n",
       " 0\n",
       " 7\n",
       " 3\n",
       " 5\n",
       " 3\n",
       " 8\n",
       " 9\n",
       " 1\n",
       " ⋮\n",
       " 0\n",
       " 5\n",
       " 3\n",
       " 1\n",
       " 9\n",
       " 6\n",
       " 4\n",
       " 0\n",
       " 1\n",
       " 7\n",
       " 6\n",
       " 9"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# separate out labels\n",
    "X = transpose(Array(train[:,2:end]))\n",
    "y = Array(train[:,1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "42000"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "N = size(X)[2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0,255)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "extrema(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "33.408911169825075"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mean(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "6190.186921854022"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "var(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "# scale X to get variance around one\n",
    "X = X./80;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9672167065396925"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "var(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.41761138962281336"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mean(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "# shift X to get mean close to zero\n",
    "X = X.-0.4176;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Split data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "# mx.Convolution() method requires data in a 4D tensor\n",
    "X = reshape(X, 28,28,1,N);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "# split the train data into a training set (cv_X) and an eval set (eval_X)\n",
    "split = 0.8\n",
    "cv_X = X[:,:,:,1:floor(Int,split*N)]\n",
    "eval_X = X[:,:,:,floor(Int,split*N)+1:N];"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "8400-element Array{Int64,1}:\n",
       " 0\n",
       " 7\n",
       " 7\n",
       " 2\n",
       " 2\n",
       " 6\n",
       " 5\n",
       " 7\n",
       " 8\n",
       " 5\n",
       " 3\n",
       " 0\n",
       " 2\n",
       " ⋮\n",
       " 0\n",
       " 5\n",
       " 3\n",
       " 1\n",
       " 9\n",
       " 6\n",
       " 4\n",
       " 0\n",
       " 1\n",
       " 7\n",
       " 6\n",
       " 9"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cv_y = y[1:floor(Int,split*N)]\n",
    "eval_y = y[floor(Int,split*N)+1:N]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Setup providers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "using MXNet"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1000"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "batch_size = 1000"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MXNet.mx.ArrayDataProvider(Array{Float32,N}[\n",
       "Float32[-0.4176 -0.4176 … -0.4176 -0.4176; -0.4176 -0.4176 … -0.4176 -0.4176; … ; -0.4176 -0.4176 … -0.4176 -0.4176; -0.4176 -0.4176 … -0.4176 -0.4176]],Symbol[:data],Array{Float32,N}[\n",
       "Float32[1.0 0.0 … 2.0 2.0]],Symbol[:softmax_label],1000,33600,true,0.0f0,0.0f0,MXNet.mx.NDArray[mx.NDArray{Float32}(28,28,1,1000)],MXNet.mx.NDArray[mx.NDArray{Float32}(1000,)])"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_provider = mx.ArrayDataProvider(cv_X, cv_y, batch_size=batch_size, shuffle=true)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    },
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MXNet.mx.ArrayDataProvider(Array{Float32,N}[\n",
       "Float32[-0.4176 -0.4176 … -0.4176 -0.4176; -0.4176 -0.4176 … -0.4176 -0.4176; … ; -0.4176 -0.4176 … -0.4176 -0.4176; -0.4176 -0.4176 … -0.4176 -0.4176]],Symbol[:data],Array{Float32,N}[\n",
       "Float32[0.0 7.0 … 6.0 9.0]],Symbol[:softmax_label],1000,8400,false,0.0f0,0.0f0,MXNet.mx.NDArray[mx.NDArray{Float32}(28,28,1,1000)],MXNet.mx.NDArray[mx.NDArray{Float32}(1000,)])"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "eval_provider = mx.ArrayDataProvider(eval_X, eval_y, batch_size=batch_size, shuffle=false)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Setup model "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Like the [previous MLP model](mnistMLP.ipynb#MLP), we will set up a `SymbolicNode` with data input with the `mx.Variable()` method and an output layer with `mx.SoftmaxOutput()` that is fed from a `mx.FullyConnected()` layer with 10 nodes to represent the 10 digits.  What differs will be the addition of `Convolution` and `Pooling` layers.  \n",
    "\n",
    "This model is loosely based on the famous LeNet5 network described in LeCun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998). Gradient-based learning applied to document recognition. Proceedings of the IEEE, 86(11), 2278-2324.  The article is available on ResearchGate at https://www.researchgate.net/profile/Yann_Lecun/publication/2985446_Gradient-based_learning_applied_to_document_recognition/links/0deec519dfa1983fc2000000/Gradient-based-learning-applied-to-document-recognition.pdf.  \n",
    "\n",
    "Each convolutional layer is composed of a set of *feature maps* which each of which is extracted from the input layer by a corresponding set of *filter banks* which store the corresponding set of trainable weights and bias.  The filters will be convolved over the input layer during feedforward, creating the activation map for the next layer.  In order to reduce the effect of small variations in input to the feature detector, the next layer is usually a *pooling layer* which 'coarsens' and reduces the size of the activation map using some particular function such as averaging values in a sub-region (average pooling), or selecting the maximum value in the sub-region (max pooling), or even functions like L2-norm pooling.\n",
    "\n",
    "In the original model (see the image below from the LeCun paper), the network uses a 5x5 kernel which makes 25 trainable weights plus one bias.  There are 6 filters of this type to create 6 feature maps.  Note that the size of the feature map is smaller than te input because running a 5x5 kernel over a rectangular input will produce 4 less results in each direction.  Also note that the 'subsampling' layers reduce the size by half as they take a 2x2 neighborhood and reduce it to a single value.\n",
    "\n",
    "![](https://www.researchgate.net/profile/Haohan_Wang/publication/282997080/figure/fig10/AS:305939199610894@1449952997905/Figure-10-Architecture-of-LeNet-5-one-of-the-first-initial-architectures-of-CNN.png)\n",
    "\n",
    "There are a lot of differences between our model and the original LeNet5.  These include:\n",
    "* the original increased the size of images to 32x32 to make sure there was better centering for the feature maps\n",
    "* the original 'subsampled' by adding the four values, then applying the 'squashing' activation\n",
    "* input to the second convolution layer were specifically tailored to select a fixed distribution of patterns from the previous layer; we are just letting backprop do the work\n",
    "* the original used an Euclidean Radial Basis Function for output that was calculated from a 84 node layer which in turn corresponded to vectors derived from ASCII characters drawn on a 7x12 bitmap. We are using just making a single layer MLP with softmax\n",
    "\n",
    "The original LeNet5 reported an error rate on the test set of 0.95%.  An earlier version LeNet4 got an error rate of 1.1% using 4 feature maps in the first convolutional layer, creating pairs to get 8 subsampling maps, then 16 feature maps in the second convolutional layer. \n",
    "\n",
    "We are going to use the same kernel size on a 28x28 image, halving the result by using a max pooling layer.  When I tried this with 6 filters in the first convolutional layer and 16 in the second, I got a maximum of 75% accuracy, so we are going to increase the number of filters although this increases the memory demand and computation time.\n",
    "\n",
    "Clearly the fine tuning done by LeCun et al. helped make their network more efficient."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MXNet.mx.SymbolicNode(MXNet.mx.MX_SymbolHandle(Ptr{Void} @0x0000000019f888d0))"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# set up net\n",
    "data = mx.Variable(:data)\n",
    "# first conv, takes 28x28 -> 24x24 (with convolution) -> 12x12 (with max pooling)\n",
    "conv1 = @mx.chain mx.Convolution(data, kernel=(5,5), num_filter=12, name=:conv1)  =>\n",
    " mx.Activation(act_type=:tanh) =>\n",
    " mx.Pooling(pool_type=:max, kernel=(2,2), stride=(2,2))\n",
    "# second conv, takes 12x12 -> 8x8 -> 4x4\n",
    "conv2 = @mx.chain mx.Convolution(conv1, kernel=(5,5), num_filter=25, name=:conv2) =>\n",
    " mx.Activation(act_type=:tanh) =>\n",
    " mx.Pooling(pool_type=:max, kernel=(2,2), stride=(2,2))\n",
    "# fully-connected\n",
    "fc   = @mx.chain mx.Flatten(conv2) =>\n",
    " mx.FullyConnected(name=:fc1, num_hidden=100) =>\n",
    " mx.Activation(name=:relu1, act_type=:tanh) =>\n",
    " mx.FullyConnected(name=:fc2, num_hidden=10) \n",
    "# softmax loss\n",
    "lenet = mx.SoftmaxOutput(fc, name=:softmax)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    },
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MXNet.mx.FeedForward(MXNet.mx.SymbolicNode(MXNet.mx.MX_SymbolHandle(Ptr{Void} @0x0000000019f888d0)),MXNet.mx.Context[CPU0],#undef,#undef,#undef)"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# change context to gpu(number) if you have a gpu\n",
    "model = mx.FeedForward(lenet, context=mx.cpu())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MXNet.mx.SGD(MXNet.mx.SGDOptions(0.05,0.9,0,1.0e-5,MXNet.mx.LearningRate.Fixed(0.05),MXNet.mx.Momentum.Fixed(0.9)),#undef)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "optimizer = mx.SGD(lr=0.05, momentum=0.9, weight_decay=0.00001)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[1m\u001b[34mINFO: Start training on MXNet.mx.Context[CPU0]\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: Initializing parameters...\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: Creating KVStore...\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: TempSpace: Total 69 MB allocated on CPU0\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: Start training...\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: == Epoch 001/001 ==========\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Training summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.1068\n",
      "\u001b[0m"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " 47.598640 seconds (8.54 M allocations: 583.137 MB, 1.23% gc time)\n"
     ]
    }
   ],
   "source": [
    "# run one epoch and then check initial states\n",
    "@time mx.fit(model, optimizer, train_provider, eval_data=eval_provider, n_epoch=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    },
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Dict{Symbol,MXNet.mx.NDArray} with 8 entries:\n",
       "  :fc1_weight   => mx.NDArray{Float32}(400,100)\n",
       "  :fc1_bias     => mx.NDArray{Float32}(100,)\n",
       "  :conv1_bias   => mx.NDArray{Float32}(12,)\n",
       "  :fc2_weight   => mx.NDArray{Float32}(100,10)\n",
       "  :conv2_bias   => mx.NDArray{Float32}(25,)\n",
       "  :conv1_weight => mx.NDArray{Float32}(5,5,1,12)\n",
       "  :conv2_weight => mx.NDArray{Float32}(5,5,12,25)\n",
       "  :fc2_bias     => mx.NDArray{Float32}(10,)"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# see parameters of model\n",
    "model.arg_params"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "# let's examine the weights conv1\n",
    "c1 = copy(model.arg_params[:conv1_weight]);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "12×3 Array{Any,2}:\n",
       " (-0.00921913,0.00957952)   0.000911697  0.00556884\n",
       " (-0.0100759,0.00889772)   -0.00103914   0.00497863\n",
       " (-0.00936757,0.00924012)   0.0012334    0.00584902\n",
       " (-0.00839963,0.00972352)   0.00249005   0.00634543\n",
       " (-0.00975356,0.00962675)   7.92435f-5   0.0065193 \n",
       " (-0.0101899,0.00847369)   -0.00219759   0.00557271\n",
       " (-0.0102178,0.00908929)   -0.00104016   0.0063156 \n",
       " (-0.00952072,0.00978716)   0.000580064  0.00687523\n",
       " (-0.00787795,0.00973415)   0.00123118   0.00603861\n",
       " (-0.00955902,0.00924187)   0.000818616  0.00604525\n",
       " (-0.00985298,0.0100028)   -0.00113303   0.00614327\n",
       " (-0.00896623,0.00936259)   0.00200388   0.00641203"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# let's get the extrema mean and standard deviation for each of the 6 groups\n",
    "[ [extrema(c1[:,:,1,i]) for i in 1:12] [mean(c1[:,:,1,i]) for i in 1:12] [std(c1[:,:,1,i]) for i in 1:12] ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1×3 Array{Any,2}:\n",
       " (-0.0102178,0.0100028)  0.000328184  0.00611495"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# let's also get the extrema mean and standard deviation for all\n",
    "[ extrema(c1[:,:,1,:]) mean(c1[:,:,1,:]) std(c1[:,:,1,:]) ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<img src=\"data:image/png;base64,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\" />"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# plot mean and std\n",
    "plot([ [mean(c1[:,i]) for i in 1:12] [std(c1[:,i]) for i in 1:12] ], legend=true, label=[\"mean\" \"std\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<img src=\"data:image/png;base64,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\" />"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# plot them as 5x5 arrays\n",
    "plot([heatmap(c1[:,:,1,i], aspect_ratio=:equal, legend=false, clims=(-0.8,0.8)) for i=1:12]...)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "# let's examine the weights conv2 as well\n",
    "c2 = copy(model.arg_params[:conv2_weight]);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "25×3 Array{Any,2}:\n",
       " (-0.0101052,0.00994338)    0.00140964   0.00625768\n",
       " (-0.00934925,0.00969299)  -0.0004845    0.00631823\n",
       " (-0.00705089,0.00984438)   0.00140995   0.00537533\n",
       " (-0.00973907,0.00930746)   0.00175598   0.00575009\n",
       " (-0.00924765,0.00964172)   5.39319f-5   0.00640249\n",
       " (-0.0098497,0.00725531)   -0.00150254   0.00522301\n",
       " (-0.00951758,0.00843546)   0.000251659  0.00600261\n",
       " (-0.00931521,0.00854179)  -0.00145474   0.0052768 \n",
       " (-0.00933483,0.00841678)  -0.000575758  0.00531668\n",
       " (-0.00977549,0.00963705)  -0.000173146  0.00659104\n",
       " (-0.00960067,0.00881964)   0.00119637   0.00622745\n",
       " (-0.00961633,0.00908362)   0.000417128  0.00483483\n",
       " (-0.00911649,0.00978981)   0.00203165   0.00554724\n",
       " (-0.00931348,0.00806423)  -0.00124122   0.0054029 \n",
       " (-0.009475,0.00971875)     0.00151365   0.00593198\n",
       " (-0.00991442,0.00878294)  -0.000495969  0.00571382\n",
       " (-0.00942977,0.00875635)  -0.000427015  0.00547125\n",
       " (-0.00977247,0.00736236)  -0.00190297   0.00556499\n",
       " (-0.00956265,0.00969471)  -0.00130624   0.00561866\n",
       " (-0.00916946,0.00824434)   0.000631118  0.00544019\n",
       " (-0.00948183,0.00967124)   2.50861f-6   0.0060356 \n",
       " (-0.0096112,0.00952886)    0.00118258   0.00666542\n",
       " (-0.00920097,0.0099337)   -0.00117764   0.00604961\n",
       " (-0.00984673,0.00856263)  -0.00329432   0.00451825\n",
       " (-0.00862562,0.00947686)   0.000445087  0.0057599 "
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# let's get the extrema mean and standard deviation for each of the 25 groups\n",
    "[ [extrema(c2[:,:,1,i]) for i in 1:25] [mean(c2[:,:,1,i]) for i in 1:25] [std(c2[:,:,1,i]) for i in 1:25] ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1×3 Array{Any,2}:\n",
       " (-0.0101052,0.00994338)  -6.93917f-5  0.00578866"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# let's also get the extrema mean and standard deviation for all\n",
    "[ extrema(c2[:,:,1,:]) mean(c2[:,:,1,:]) std(c2[:,:,1,:]) ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<img src=\"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAlgAAAGQCAYAAAByNR6YAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAAPYQAAD2EBqD+naQAAIABJREFUeJzs3Xl8FPX9P/DXzO7M3jnIQRKSkIQACeFMuAW55BAUUYt4oIhVsVWrbRVrbX+Vav1WW631aLUVq1ZFvEDx4igQQOTKxZGDKyTcOcix9+zuzO+PJZFAjt3s7Mzs7uf5ePhAyMzn817ebPLe+VyUIAgCCIIgCIIgCNHQcgdAEARBEAQRbkiBRRAEQRAEITK13AH4o6GhAevXr0dGRgZ0Op3c4RAEQRAEEUHsdjtOnDiB2bNnIz4+vttrQ6rAWr9+PRYvXix3GARBEARBRLD3338fd9xxR7fXhFSBlZGRAcD7wnJzc+UNRmYemxkN//wtKLUaqpgE9LnzN6BocUd8H330Ubz88suitklIj+Qx9JEchgeSx9BXUVGBxYsXt9cj3QmpAqttWDA3Nxf5+fkyRyMvyw/fojkxGnH3/B6NK1cgFi0w5M8QtY+YmJiI/3sOBySPoY/kMDyQPIYPX6YpkUnuIcpeuh2aAUOhGzYB2uET0frNexDcnKh9cJy47RHyIHkMfSSH4YHkMbKE1BMswstjaYHzSClibvoZACB63lKc//MyWL7/BqYpC0Trp7S0VLS2CPmQPIY+ksPwoNQ82mw2VFZWyh2GIuXk5ECv1/fqXlJghSDHgZ2AAOiGXwUAYPqmwTBuFswbVsEwbiZorUGUfgYPHixKO4S8SB5DH8lheFBqHisrK1FQUCB3GIpUVFTU62FdUmCFINvF4UFVVJ/2P4uasxjWos0wb/4M0XPvEqWfBx54QJR2CHmRPIY+ksPwoPQ8kgVkP2qbzB4IUmCFGI+1tcPwYBtVTDxMVy+AZevnME66rkPx1Vu33XZbwG0Q8iN5DH0kh+FB6XkkC8jERSa5hxjv8KDQPjx4KdOMhYBajdYNH8oQGUEQBEEQbUiBFWJspdvBZg3t9AkVrTch6ppFsO78Fu76MwH3tXLlyoDbIORH8hj6SA7DA8ljZCEFVgjhrWY4D5dCP3Jyl9cYJ82HKioWLd+8G3B/xcXFAbdByI/kMfSRHIYHksfIQgqsEGI/sBMQeOhGTOryGorVIGrOnbCXFII7eSSg/l5//fWA7ieUgeQx9JEchgeSx8hCCqwQ0t3w4KX0Y66BOikdLeveligygiAIghAHTdN47rnnMH78eGRkZGDNmjV47rnnMGbMGAwaNAiFhYXt165fvx6TJ0/G6NGjMW7cOGzduhUAcO7cOUyfPh2jR4/G0KFD8fDDD0MQBADAO++8g2uuuQa33347hg8fjjFjxqC6ulr010FWEYYI7/BgCWJuXNbjtZRKheh5S9G4cgUcVcXQDiarQgiCIIjeOd4qoDmATehjWCArivLrHpPJhF27dmHz5s2YP38+/vGPf2Dv3r349NNP8fjjj2PPnj04fvw4VqxYgfXr18NkMuHo0aO4+uqrUVNTg9jYWKxbtw4GgwEejwcLFizAxx9/jEWLFgEA9u3bh7KyMvTv3x9PPvkknn/+ebzxxhu9f5GdIAVWiLAf/ME7PNjJ6sHOaIeOB5s5BC3r3oZm4EjRD4ImCIIgwl+DQ8DAj93ghd63oaKAc4vViNf6XmS1FUIFBQWw2Wy49dZbAQBjxozBkSPe6S/fffdde1HV3pdKhdraWiQnJ2P58uX4/vvvIQgC6urqMGzYsPZ2J0yYgP79+7f//6uvvtr7F9gFUmCFCHvpdrCZeVBFx/l0PUVRiL7+HtS/8hjspdugz5/qd5/z58/Hl19+6fd9hLKQPAaP4PHAefwg7GU74KwsQvSNy6DLGyd6PySH4SEU8xivpXDkFnXAT7D8Ka4AQKvVAvAWTADAsmz7791ud/t1M2fOxAcffHDF/c8++yzq6+uxZ88esCyLX//613A4HFe0D3iHJC9tUyykwAoBvM0Mx+ESxNxwn1/3abKGQps3Di3fvAvd8KtAqRm/7n/ooYf8up5QJpJHcQkeN5yHS2Hf/z3s+3eCt7ZAFZMAimHRsm4ltLmjQdEqUfskOQwPoZpHf4f3pDJr1iysWLECBw4cwLBhwwAAe/bswdixY9Hc3IykpCSwLItz587h448/xsKFCyWNjxRYIcB+4AeA93S7erAr0dctxfkXfg7rD9/COHm+X/fOmjXL7/4I5SF5DJzg5uCoKoG9dDvsh3ZBsFmgik+Gftws6EdOApM2CFxNFepffhT20u29emLcHZLD8EDy6BuKonz6fXZ2Nj788EMsW7YMNpsNHMchPz8f77//Ph555BH85Cc/wdChQ5GSktLh756iqA5tXv57sZACKwTYy7aDzRzi8/DgpZjkDOjHzEDr+g+hHzsTtEYXhAgJIvzwnAPOin2w7f8ejoO7IThtUCemwTjpeuiGTwLTL6vDN2VNRg60Q8ag9bsPoBs5WfSnWAQRKTweT/v/G43GDr9PTU1Fa2tr++9nzJiBGTNmXNFGWloadu/e3Wn7S5YswZIlS9p/P2/ePMybN0+M0DsgBZbC8TYLHFUliJ5/b6/biLr2TtiKt8Ky5XNEzblDxOgIIrzwDhsc5XtgL9sBR8VeCJwTTEomTNNuhm7kJDBJ/bu9P2rOYtS99AjsxYXQj54uUdQEQSgRWVqmcPaDPwAeN/S9GB5so45NhHHyfJi3fAqPpdnn+9auXdvrPgnlIHnsHm+zwLp3ExreehpnfrcIF977M9wXzsM063b0fWol+i7/J6Lm3NFjcQUAbPpgaPPGoXX9BxAu+dQdKJLD8EDyGFnIEyyF864eHAJVTHxA7ZiuWQTrD9/BvGEVYm76mU/3rFq1CgsWLAioX0J+oZZHj7kJ9tLtgJoBzWhAsVpQ7JW/0hd/hYrxe/6Ex9ICx8EfYC/7Ho7DJYDHDTYjF9Hz7oZu+FVQxyX1Ov6oOYtR9+LDsBVtgWHsNb1u51KhlkOicySPkYUUWArmHR4sDmh4sI3KEAXTjIVo/e59GKfc6NMPkNWrVwfcLyG/UMqjIAhofOc5cMcPAQLv2000DYrRgtJoQDFa0O1FWCeFGaOB69QROI/uBwQBbNZQxNxwH7TDJ0IdkyDKa2DTBkI7dALMGz6EvmAaKFXgc7FCKYdE10geIwspsBTMfnAX4HFDN8K3zUV7YpyyAJbtX6L1m3fR584nRGmTIMTkOLAT3LEDiLv/GWhzR0NwOSFwTgic45JfHeAv/zOXE4LT+yt/8Zq2r/GWlg73qeOSEHPzg9ANnwiVKTYoryNqzmLU/fVB2Pb9D4ZxZOUYQUQiUmApWNvqQbE+WdOsFlFzFqP541dgnPYTsKkDRGmXIMQguF1o/vItaAbne/eSoihQrBZgtQCi5Q7PL2zqAGiHT0TrhlXQj54OSkW+1RJEpCGT3BWKt1vhqCyGbsRkUds1jJsNdUI/tHz1H1HbJYhAWXasg6fxPGIW3B+UPWmkFjVnMTyNZ2Hbu0nuUAgi5NXU1ODNN9/s9prRo0d3OAhabqTAUijv6kEXdCN7v3qwM5RKhajrlsJZuQ+OI2XdXrt06VJR+ybkEQp59Fhb0br+QxgmzAGTnCF3OKJgU7KgGzEJrRtWQXC7AmorFHJI9Izksfeqq6t7PIw5WBuG9hYpsBTKXrodbEauaMODl9INvwpM+mC0rFsJQej6BE+y63B4CIU8mtd/APA8oq69U+5QRBU1ZzE8TXWwBvgUKxRySPSM5NE3drsdixYtQl5eHkaOHInZs2fjZz/7GaqqqjBq1Kj2lZg7d+7EyJEjMWzYMNxzzz1BOU8wEKTAUqD24cGR4g4Ptmk7CNpVexj2sh1dXnfbbbcFpX9CWkrPo+v8SVh2fAXTzFuDNulcLkxyBnQjJsMc4FMspeeQ8A3Jo2++++47tLS04NChQygtLcXq1avxxhtvYPDgwSgpKcHatWvBcRwWLVqEl156CQcOHMBtt92GsrLuR2WkRmZeKpD90C7v8KDI868upR04AtrcMWj9+h3ohk0gk3AJ2bR8+RZU0XEwTQnP/YGi5tyB888/AOvuDTBeJf5xHAQRbO6Gs+Dtll7fT+uMUMcn+3z9yJEjUVFRgQcffBBTpkzB3LlzrxhtqaysBMMwmD7de2LCzJkzkZWV1esYg4H8VFUge+l2sP1zoI4Vf3jwUlHXLUXdXx+Eddd68o2fkIXjcAkch3ajz11PgmJYucMJCiapP3SjpsC8cRUM42aCUofn6yTCk8fSgnN/+qnv+9J1hqaR/MdVUBl9Ww2cmZmJiooK/O9//8OmTZuwfPlyvPzyyz3ep6T5VwApsBSHd1jhqCxC9Ly7g94X2y8L+oJpaF3/AfSjZ4DWaDt8fceOHZg0SdxJ9oT0lJpHgfegZe2/wWbkQjfqarnDCaqo2Xfg/J+XeT/MTLre7/uVmkPCP6GYR5UxGklPrQz4CZavxRUAnD59GjExMbj++usxe/ZsrF27FnFxcWhpaWm/JicnB263G1u3bsXUqVOxadMmHDt2rNcxBgMpsBTGcXA34HYFbf7V5aKuvQu2ku2wbFuLqJm3dvjaCy+8EHLfDIgrKTWPtj0b4TpzHAmP/k1xnzzFxvRNgz5/Clo3fgTDuNl+P61Tag4J/4RqHv0Z3hPDgQMH8OSTT0IQBLjdbtx1112YOHEi8vLyMGzYMAwYMABr167F6tWr8fOf/xwejwdjxozByJEjJY2zJ6TAUhhb+/BgoiT9qeOSYLxqHsz/+xiGiXOhMkS1f+2jjz6SJAYiuJSYR95hQ8s370KXPxWajFy5w5GEafYdsBUXwvrDtzBefYNf9yoxh4T/SB59M2fOHMyZM+eKP1+3bl2H30+YMAElJSVSheU3sopQQbzDg/ske3rVxjTrVkAAzBs7vvn1er2kcRDBocQ8mjd/At5mQfR1kbMvEJOYCv3o6WjdtBoC5/TrXiXmkPAfyWNkIQWWgjgO7fEOD46Q9hGyyhgD0/SbYdm+Du4L5yXtm4g87qZ6mLd8BtPUm6Du01fucCQVNet28JZmWH74Vu5QCIIIMlJgKYitdDuY9MGy/NAxTr0JtN6I1m//K3nfRGRp/ept0Bo9TNcskjsUyakTUqAfPQPmTavBcw65wyEIIohIgaUQvMMGR8Ve6CUeHmxDa3SImn07bPv+B9eZagDA448/LksshLiUlEeupgq2oi2ImrcEtDYyh0uiZt0O3toC6/df+3yPknJI9B7JY2QhBZZCOA5Ju3qwM4YJ10Idl4yWr98BAKSnp8sWCyEepeRREAQ0r30TTHIGDOMi98gQdXwy9GNnwvy/T8A7fXuKpZQcEoEheYwsZBWhQtjK5BsebEOp1IiauwQX3vs/OI8dxMMPPyxbLIR4lJJHe9kOcNXliP/Zc6BoldzhyCpq5m2w7dkE6/frYJq+sMfrlZJDIjBKz2NFRYXcISiGGH8XpMBSAN5ph6NiH6KvvUvuUKAbORnMlk/Rsm4lEh55Kez3JyKkIbg4tKxbCe2QsdAOzpc7HNmp45JgGDcL5v99CsNV14HW6CSPQRAEWLZ8BnfDWcTeouwf/IQ0Fi9eLHcIYYUUWArgOLQbcHGSrx7sDEXTiL5uKRr++Vs4DvwA3fCJcodE9JK74Qwa/v009GOugWnGQlmLZcv2L+BpqkP8fX+ULQalMc28FdY9G2HZvg5R19wiad+Cx4PmT1+D9YdvAYpC9HV3g9abJI2BUI6cnBwUFRXJHYYi5eTk9PpeUmApgL10O5j0QVDHJckdCgBAOzgfmkEjUffl20gfNoE8xQpBHnMz6t94CoLditav3gZvbkL0DfeBoqWfdumxNKN1wyoYJs4Dk0TmoLRR9+kLw7jZsGz5FMZJ13U76b+ysjKgb/SX4p12XHj3OTgqi2GafQfM6z+A80QFdEPGitI+0TUx8ygmvV6P/HzyZFlsZJK7zHinHXYZVw92xTR9IVQNp8AdOyB3KISfeIcNDf/6PQTOgcRfvYKPGlSwbFuLpg9fhOBxSx5P63fvA6AQNYcMP1zONPNW8A47LNu/7Pa65cuXi9Kfp/UC6l99HM5jhxC/7BlEzVkM2hQL7vghUdonuidWHonQQAosmTkO7bk4PKisAkszOB9CXArMWz+XOxTCD4Lbhcb/PAt33WnE3/8s1HFJuOW5f6LPnU/AVlKIxpV/lHT/Jde5Glh3foOo2bf7ddhrpFDHJsAwYQ7MWz4F77B2ed1rr70WcF+uc7Wo+9uj8JibkPCLv0I7OB8URUGTlQdndXnA7RM9EyOPROggBZbM7GXbwaQNVMzwYBuKotBn5iI4Du2Gq+6U3OEQPhB4Hk2rXoLz6AHE3fsHsKkDAHiXhuvzpyL+3qfhPFqGhjeeAm+zSBJTyxdvQRXbF8bJ10vSXyiKumYRBM4Jy7Yvurwm0OX9zmMHUPfKr0Bp9Uh89GWw/bLav8ZmDgFXWwXB7QqoD6JnZJuGyEIKLBnxTgcc5XuhH3m13KF0Sl8wDbQhGpZta+UOhfBBy7qVsBVvRZ/Fj0M7cMQVX9fmjkb8z/8M97la1L/6GDwtjUGNx1GxD46KvYie/1NQajaofYUyVUw8jBOuhXnL50EpfG3FW1H/j9+C7ZeNxIf/CnVsQoeva7LyABcH7tRR0fsmiEhGCiwZOcr3QHA5Zd1ctDsUw8I46XrYdm8EbzXLHQ7RDfOWz2DZ8hlibnwA+lFdF+yajFwk/OKv4G0W1L3ya7gbzgQlHsHjQfMX/wabNRS64VcFpY9wYrpmEQQ3J+qHGUEQYN78CS6892foR12N+GXPgNYbr7iO6TcAFKsh87AIQmSkwJKRvXSbIocH2zz//PMwTJoHQeBh2en7sR6EtGz7NqPli3/DdM0iGK++4YqvP//88x1+zyT1R8IjL4KiVaj7+6/BnTomekzWXd/Bfa4GMQvuJ6tQfaCKjoNx4jyYt67p9CnW5TnsicB70PzZP9Dy5UqYZt6K2DseA6VmOr2WUqnB9s+BkxRYQedvHonQRgosmbQNDyptcvulbDYbVMYYGMZcA8v2L8kcDQVyVBbhwocvQj92JqLm3d3pNTab7Yo/U/fpi4RHXoQqJh71rz0Op4irRXmHFa3fvgf96Blg0weJ1m64M81YCHjcMBeuueJrneWwKzznQOPbz8C682vE3PILRM+7u8cil83MA1ddDkEQ/I6b8J0/eSRCHymwZNI2PKi07RkutWLFCgCAccoC8K0XYCsplDki4lJc7WE0vv0MtDn5iF30SJc/RNvyeDmVMQYJDz4PNm0Q6t94CvaDP4gSl3njaghOB6Kuu1uU9iKFKqoPDFfNg2XrmiuG5LvK4eU85mbUv/YEnIdLEXfv0zBOnOvTfZrMIeCtLXCTBS1B5WseifBACiyZ2Eu3gUnNhjo+We5QesQk9Yc2dwwsW9eQT7gK4a4/g4Z//R5Mcgb6LHkKlKp3ewbTWj3il/0RuiFjvU89dm8ILK7GczAXroFx+k+gjkno+QaiA9OMhQDvgXnrZ37f66o7hbqXfwlPUx0SHv6LXxuHspm5AEWDqybDhAQhFlJgyaB9eFDBT68uZ5x6I1ynj8F5dL/coUQ8T+sF1L/xFGi9CXH3rQCt0QbUHqVm0WfJkzCMm42mVS/BvPnTXrfV8tV/QOtNME3/SUAxRSqVKRaGydfDsu0LeKytPt/nrC5H/d9/BUqlRuKjfwObNtCvfmmtAUxKJpzHyX5YBCEWUmDJwFGh/OFBAGhoaGj/f82gUWCSM2AhG4/Kqn2XdpcT8Q/8yafNOy/NY1coWoWYW34B08xb0fLlW2j+cqXfTyud1eWwlxQiet7dshxeHC5M038CXDyIuU13ObSX7UD9P34DdVJ/JD76Uq8XzbCZQ8gTrCDz5b1IhI+QLLDcrRfkDiEg9tLtYFIHQB2fInco3brnnnva/5+iKBin3uTdePT8SRmjilyC24XGt5+Bu+Es4h94Fuo+fX2679I8doeiKETPuxvRC5bBsvkTNK1+GYLH41tsgoCWtf8CkzoA+jHX+HQP0TmVMQbGyfO9T7EszQC6zqG5cC0a3/kTdEMnIOGBPwV0YLMmKw/u+tPwmJt73QbRPV/fi0R4CMkCiztRIXcIvcZzDjjK9yh69WCbp59+usPv9QVTQZtiYSkkG49KTeB5XPjwRTiPHUTcT/8ANiWr55suujyPPTFNvRGxdzwG256NaHznTxBcXI/32EsKwdVUIvqG+2U5UDrcGKfdDFAUzJu9T7Euz6HA82he8yZa1rwB47Sb0efOJ0AxgW3mymblAQB5ihVE/r4XidAWkt8JQ7nAcpTvhcA5Fbt7+6UuP12dUrMwTr4etr2b4LG0yBSVPAQ3B3dzvTx9CwJavvg37CWF6HPn8k53ae/O5Xn0hWHMNYj76R/grCxCw5u/6/acPIFzomXd29AOneB3bETnVMZoGK++AdYdX8Jjbu6QQ4Fz4sK7z8Gy7QvE3PxzxMy/V5SiVh2TAFVsItkPK4h6814kQleIFliVPg9dKI29dDuYfgOgTlD28GBXDBPnQYAA685v5A5FUq0bV+Pc03ei/h+/gb1sh6T//ixbPoWlcA1ibv65pPP2dHnjEP+z58CdPo7615bDY27q9Dpz4Rp4WhoRPf+nksUWCUxTbwYoFcybP2n/M4+lBfX/fBKO8r2Iu+f3ME6eL2qfbJZ3P6xI5ssTW4LwRUgWWALnAFdbJXcYfvMOD+4OqdWDl1MZo70bj+74EoI7cr4ROY+UgUkdAMHFofE/z+LcM0vQuuHDLosOsVj3bmrfjds4SfoDkzVZeUh8+C/wtDah/u+/hrvxXIeve8xNMG9aDePk+WASUyWPL5zRBhOMU26AdcdX8LRegLvhDOr//iu4608j/sHnoRs2QfQ+NZlDwJ08Ap5ziN52KHAeO4jTyxegYeUfQ/JnDKEsIVlg0Vo9HJVFcofhN0fFPgiccs8evNzKlSs7/XPjlBvBtzbBVhwZG48Kbhe4k4ehHz0DiY+8hMTHXoc2ZzTMG1fj7NN3ovG9P8N5/JDoe4Q5KvahadXfoB8/G1Fzl/S6na7y6CsmJROJj7wEAQLq/v4ruM6eaP9a6zfvASoVombdHlAfROdMU28CVCqU/eUx1L38SwBA4qN/gyYjJyj9sVl5AO8BVxOZxYV170bQpmi4z9Wg7qVHUP/P38JxpEy093ag70UitIRkgcWkD4ajYp/cYfjNXrodTEoWmIR+cofik+Li4k7/nOmbBu2QsbBs/TwiNh51nT4OuDiwGUMAAGzqAMTe+iiSV7yP6Ovvgav2MOpf+TXq/vIgLD98C94Z+Kd/rrYKjf95Ftrc0Yhd+IuAzvPrKo/+UMclIfGRl6AyxaLulcfgrC6H60w1rLvWI2r2YtCG3q9eI7pG600wTrkRieYzUMenIOGRl4K6+phJ6g9Ka4jIYULB7YJ9//cwjJ2Fvk/+C33u/i14SwsaXn8C9X//FeyHdgf8/U6M9yIROnq3/bPMNJm5cJV+DY+1FSpDlNzh+ETgnHAc2g3TNYvkDsVnr7/+epdfM069CQ3/+A2cR0qhHTRKwqik56wuBxgWbOqADn9O600wTb0JxqsXwFlVDMv3X6H541fR8sVbMIybBcNV83o1bOaqO4WGN/8fmJRM9FnyJCiVKqD4u8ujP1SmWCQ89AIa3noaDf94Eur4ZKjjk2GcdJ0o7ROdM01fCFVMPAwF00GxmqD2RdEqsBm5ETnR3XG4BILNAt2oKaBoFfQjr4ZuxGQ4KvbBvHEVGv/9BzApmTBdswi6kZNB0f6/L8V6LxKhITSfYGXkAoIAZ1WJ3KH4zFGxFwLnCJnhwZ5oBo4Ak5IFy5bw33iUqz4ENm0QKDXT6dcpmoY2dzTi730aSb97G8ar5sG2bzPOP3cv6v/5W9gP/ACB921SvKf1AhreeAq0Mdq7Szsb2C7tYqN1BiQ88CdoBufDdfYEouff2+tjegjf0BotjBOuDXpx1UaTlQfuRIXP/2bDhb1kG9SJqWBSMtv/jKIo6IaMQeIjLyHh4b+CjuqDC+/9Geefuw/WH76NqHmohP9CssBSm2KhTs4IqXlYttLtYFIyw2YiMEVRME67CY6KvXCdq5E7nKARBAHOExXQZA7x6Xp1XBKir78HyU//F7F3PAbBYUPjyhU498xStG78qH3jyM7wDisa3vw94PEg/oFnFft0lmJYxC39HRIfey0oE60JeWmy8iA4rGH9vr6c4OJgP7DT+/Sqi+F4zYChSHjgT0j89atg+mWh6eNXcO6Ze2DeukaUaQFE+AnJAgsAtDkFcFQVhcQcoLbhwXB5etVGP2oK6Kg+Yb3xqKepDnxLo/cwXD9QDAvDmGuQ+MuXkfjrV6EZOBKtGz7E2T/ciQvvvwDniYoO/3YFN4fGlc/A3XjOu0t7bKLYL0VUlEoFNjVb7jCIIGDSBwG0ClwEDRM6KosgOGzQj+p5f0I2bSDilv4OfZ94E5pBI9Hy5b9x7o93oXXDh+BtZgmiJUJFSBdYfEsj3CHwKctRuc87PBgCu7dfav787vfYodQMjJPnw7p3U7dPZkJZ22RfNsO/AutSbNpA9Ln9V0h++n1Ez1sCZ3UF6l/+JepefBjWXevBOx248P5f4aw+hPj7ngaTnCFS9F495ZFQPilzSLNasGkDvXMPI4StpBDq5AwwSf19vodJSkefOx5D0lNvQzdqClo3foSzK5agZd3bXW7fQt6LkSVkCyxN1lBQjCYkVhPaSreDSc4A0zdN7lD88tBDD/V4jXHiPFA0DeuOryWISHrO6nKoE/pBZYwJuC2VIQqm6T9B0lMrEXf/M1BF9UHT6pdx5qmFsJftQNxdv4FmwDARou7IlzwSyiZ1DtnMIRHzBIvnHHAc3OXT06vOqOOSEPuTB5H8/96FcdJ1sOz4Cmf/uARNn74O94XzHa4l78XQ5zpX6/O1ITs7lWJYsAOGwVFZ5D19XqHaVw8qOMauzJo1q8draIMJ+rEzYdmxDqYZCwM+D01puBPlYH2cf+UriqaXMZS/AAAgAElEQVShGzIGuiFj4G44C+uu78AkZ0A3/CpR+2njSx4JZZM6h5qsPFi2fg53Uz3UsQmS9i01R/ke7wjDqCkBtaMyxSL6+ntgmnELLDvWwVK4Btad30BfMA2maxaB6ZtG3oshTvB40LzmDZ+vD9knWACgzR0N5/GDit512FFZBMFphy4Ezh7sLePVC8BbW2Ar2iJ3KKLinXa4TleLXmBdSh2fjOjrlkJfMC1ofRCEv9jMyDn42V6yDUzqANH2J6T1RkTNug1J/+89RM+/F87DpTj/5/vR+J9nO2zSS4QeR+U+8NZWn68P7QIrpwBwu+A8ekDuULpkKyn0bi4aYsOD/mASU6HNGwdzmG08ytVUAgIPTQDzrwgiFKlMMVAn9Av7/bB4hw328j3QjQzs6VVnaI0Wpqk3Iun3byP2ll+AO3UUDW/+Lqy+R0Ya6+71fp0jHNIFljox1Xv6u0K3a+CdDjgO7YIuX/w3rxTWrvV9daBp6k1wn6uBsyp8dirmqstB6YxQ902XO5SA+JNHQpnkyCGbGf4HPzsO7QZcXK/nX/mCUrMwTLgWMTfcD09zA/iWxqD1RQSPx9wMx8Hd0A71fWuakC6wKIrybtdQqcyJ7o5DuyBwTugDHNuXy6pVq3y+lh0wDExqNsxbw2fjUWd1BTQZuaDokH6b+JVHQpnkyKEmKw+uM9Xg7VbJ+5aKraQQbP8cqOOSgt4XkzYQAMCdOhr0vgjx2fb9D6BoaIeM8fme0P7JAUCTUwB33akrVmsoga1YujdvMKxevdrnaymKgmnqTXBWFoXFPAOB58GdqPB7/ysl8iePhDLJkUM2cwggeN8H4Yi3WeCoKIIuiE+vLqWKiQdtiAZ38ogk/RHiEQQB1t3roRs2ASqd0ef7Qr7A0g4cCdC04nZ197559wW8MiWU6EZOhio6HubCNXKHEjD3+VoIDmv7Ac8EEWnUiamgDdFhux+W/eAPgMcl2QbQFEWBSR0AF3mCFXK4miq4z9XCMH62X/eFfIFF641g++cobh6W/cBOgHdDH2a7t3eHUjMwTJ4P277NXW60Fyqc1eUATYPtP1juUAhCFhRFeffDCtOVhPaSbWCz8qCOkW4bCjY1G65TxyTrjxCHbfd6qGISoBk00q/7Qr7AAi4em3O4BILHLXco7WwlhWCzhkIVEy93KJIyTrwWFE3DsuMruUMJCFddDiYlC7RGJ3coBCEbTVYeuJpKRX1vFYPH2gpHVbHk82OZtGx4WhpC/gNoJOGdDtiKC6EfNwsUrfLr3jApsEZDcNi8y+oVwGNpgfNwCfT5U+UOJSBLly71+x5ab4J+3GxYv/8KAucMQlTS8M6/Co/hwd7kkVAWuXLIZuVB4JxwnQ6vpy6O/d8DggDdiEmS9vu7V/4NAOQpVgixl+2A4LTBMHam3/eGRYHFpGWD1psUMw/LXrYDAKAbEZyduaXS212HTVcvAG9tha1os8gRScNjboa7/jQ0YVJgkd2jQ59cOWRTswGGDbv9sGwl26DJHgZVVB9J+x1zzbWgtAYy0T2EWHevh2bgyF4tVguLAouiVdAMzldOgVVSCM3AkaKcXyen2267rVf3qRNSoB06Aeata0JyU722VVPh8gSrt3kklEOuHFJqBmz64LDaD8tjbobzSJksp2vcdvvtYNOyQ2qiu3nrGpj/9wns+3fCda4Wgtsld0iScdWfBnfsgN+T29uE7FmEl9PmFKDpo7/BY2mByhgtWxyelkY4jx1A7K2/lC0GJTBNvQn1rz4GZ2URtLmj5Q7HL87qcqii46GScPIrQSiVJisP1l3rIQgCKIqSO5yA2ct2ABQkHx5sw6Rmt49yKJ2nuQEta98EVAzguVhY0TTUfZKgTkxt/4+5+Cttig2LfyNtbLs3gNIaoBs2sVf3h1WBBUGQfe6TrXQ7QKt6nZBwwWblgUkbCPPWz0OuwOKqD4HNzA2rbxQE0Vts5hCYN34ET8NZv44JUSpb+wiDPB/E2dRsWLZ8Bt5qBm0wyRKDr7jaKgBA8u//A1AUXHWnvPtO1p2Cu/4U7Ad/gKfxPCDwAABKq/cWXAn9LhZfad5fE1JAs1o5X4rfBI8H1r2boB89DRSr6VUbYVNgqaLjwCRnwFGxT9YCy168Fdrc0aD1vm9GplQ7duzApEm9+5TXtvHohf8+D9eZajApmSJHFxyCmwN38giiw+hw7kDySCiDnDnUZOQCFAVn9aGQL7A8LY3gjh9E7KJHZel/x44dGDcwGwDAnT4K7aBRssThK66m6uLTfO9qeFV0HDBwRIdrBDcHd8NZuOtO/1iA1Z+Co7Kow8HIqtjEK554MckZks+D85Wjch/4lkYYxvVueBAIowIL8O7qbivaItujbHfjOXA1lehz5xOS9x0ML7zwQkDf1HUjJ0O1biXMW9egz+2/EjGy4OFOHQPcLmjCYAf3NoHmkZCfnDmk9Saok/qDO36oVyuplKR9hGG4PCMML7zwAr5YuxaURgfXqVAosCp73AuQUrNgkvqDSeqPyze18Vhbf3ziVXcKrrpTcB4phXXnt94hRzWDxEdfBps6IHgvopdsuzeASckEk5rd6zbCqsDS5hTAsuUzuM5Wg03Jkrx/W8k2UIwG2qHjJe87GD766KOA7qdUahivvgEtX7+H6OvuVuwnlUtx1eWgGA2Yfsp7w/dWoHkk5Cd3DjWZeXAeOyBrDGKwl26DdnA+aL08Q3MfffQRKJoG0y8L3EllT3QXeA+4k0cQNev2XrehMkRBlTnkihXZgscDT9N5NLz5e7R++x7i71sRaLii8liaYT+4C9E33BfQw5qwWEXYRpM1FBSrkW1Xd3tJIbR548Jmc0q9Xh9wG4bx14JSqUJm41GuuhxM+iBQqvD57CFGHgl5yZ1DTVYe3Odr4bG0yBpHINxN9eCqyyU7e7AzbXn07uiu7ALLff4kBKc9KKdZUCoV1PEpiJqzGI5Du+FU2HmXtr3eg531o6cH1E5YFVgUw0KTPVyW7Rpc50/CdfoY9PmRc/agL2i9EYbx3o1Hec4hdzjdEgQBzhPlYbP/FUGIpW3LEu5E6G7XYC/dBqgZ6IZNkDsUMKnZcNefBu+wyh1Kl7iaSoCiwaQNDFofulFToE7OQOvX7watD391ONjZEBVQW2FVYAGAZnABnMcOgXdK+8PcXrINlEYPbe4YSfsNBcarF4C3mb2fChTMc+E8+NamsNn/iiDEourTF6roeDiPh26BZSsphDZ3DGitQe5QvBu4AnCdPi5zJF3jaqrAJKUHdUSGomlEX3sXnEdK4ThcGrR+/NF+sPO4wDf3DbsCS5tTAHhccB4tk6xPQRBgK94K3bAJoBhWsn6D7fHHHxelHXV8MnTDJsJSuAYCz4vSZjC0babIZoTPBHdAvDwS8pE7hxRFgc3KC9mDn90NZ+GqPQy9jMODwI95VPdNBxhW0fOwvBPcc4Lej3bYBDDpg9D6zbuK2Ji6/WDnwYEvQAi7AkudmApVbKKk87BcZ6vhrjsJXZgND6anp4vWlnHqTXDXnYKjYq9obYrNWV0OdWJawI+FlUbMPBLyUEIONZlDwNUegeDi5A7Fb7bSiwuQ8sbJGkdbHimVCmxKFlynlHlkDu+0w3W2Jijzry5HURSi5y4Bd6ICjvI9Qe+vOzx38WDnsTP9Pti5M2FXYFEUBW1OgaTzsOzFhaD1JsUvufXXww8/LFpbbOYQMOmDYdm6RrQ2xcZVl4MNo+0Z2oiZR0IeSsghm5UHeFzgTh6WOxS/2Uu2QZs3VvYFSJfmkUnNBqfQie6uU0cBgZfkCRYAaAbngx0wDK3fvCfrKIe9tPcHO3fGpwLryJEjmDhxIgYPHoyxY8eivLzzcfivvvoKubm5GDRoEG6++WaYzeb2r+3evRsjRozA4MGDMWPGDJw5c6b9axkZGcjJycGoUaMwatQofPLJJwG9KG3uaLjrT8PdeC6gdnwhCAJsJYXQjZgESs0Evb9Q1bbxqPNIqXevKYXhHVa4zp4gE9wJogtMciYojS7kDn521Z3yLkAapawRBjZ1ANznT0k+X9gXXE0lKFYLdZI0T07bnmK5Th+Dfb98xwi1H+wcnyxKez4VWMuWLcMDDzyAqqoqPPHEE7j77ruvuMZiseDee+/FF198gcOHDyMlJQXPPPMMAIDnedxxxx145ZVXUFVVhblz5+LRR3/cSZeiKHz88ccoKSlBSUkJFi5cGNCL0gwcCdC0JE+xXLWH4Wk8B53C3rxKpBsxCarYRFgKlfcUi6up8n5iyyAFFkF0hlKpwGbkhtzBz/aSQlAaneIWIDGp2YDAw3VGeRPduZoqMGkDRRkm85VmwFBocgrQ+s1/IfAeyfpt464/4z3YWYTJ7W16LLDq6upQVFSExYsXAwBuuukmnDx5EsePd/xH8e233yI/Px+DBg0CAPz85z/HqlWrAABFRUVgGAZTpniLkPvvvx/r1q0Dx/04li/m5DZaZwCbkStJgWUr3graFAtN9rCg9yW1yspKUdujVCoYr74BtuKt8LQ0itp2oLjqcu+O1YmpcociOrHzSEhPKTnUZObBWX1I0YtVLmcrKYRu6PhenycnpkvzyCRnACq1IvfDkmqC++Wi5y6Bu+4kbPs2S963dc/Fg52HXyVamz0WWCdPnkRycjJo2nspRVFIT09HbW1th+tqa2s7TMTs378/zp49C57nUVtbi/79+7d/zWQyISoqqsMw4Z133onhw4fj3nvvRUNDQ8AvTJtTAOfhUgged8BtdUXgedhLt0M/crKklb5Uli9fLnqbhvFzQKnUsO5eL3rbgXBWl4PNyAVFh920xKDkkZCWUnLIZg2BYLPAfb6254sVwHX2BNznahUzwnBpHik1AyY5Q3ErCT0tjfA0N0AjwQT3y7Hpg6AdPhGt6z+A4HZJ1q/g8cC6ZyP0Bb0/2Lkzov00CWQ7+W3btmH//v0oLi5GfHw8lixZEnA82pwCCE4buBPB++THVZfD09IAnYyHSwfTa6+9JnqbtM4A3bAJsO3booglucDFIyFqKsN2/6tg5JGQllJyyPbPAWgazhAZJrSVFILSGqDNyZc7FABX5lGJO7pzNVUAAEaGAgsAoq+9C54L52HdvUGyPh1VRd6Dncf3/mDnzvRYYKWlpbU/iQK8Q3mXP60CvMtPa2pq2n9/4sSJ9idfl3/NbDajpaUFKSkp7X0AgFqtxiOPPILt27d3G9PcuXMxf/78Dv9NmDABa9eubb+GSR0IN6vDe3+88pPfgw8+iJUrV3b4s+LiYsyfP/+Kp2d/+MMf8Pzzz3f4s9raWsyfPx+nN6+FKjax/VHqq6++esV+NTabDfPnz8eOHR0n7q1atQpLly69IrZFixZ1eB0AsGHDBsyfPz9or+Py4Ye213FpjsV8HbqC6XDXnWz/xhLs13Gpzl6H61wNBIcNf1v9pV+v43Jyvw6g83ykp6eHxesAwiMfvXkdGzduVMTroDU6MP0GgLs40V3Z+bgeF3auh274RFBqtsPruJxUr6Nt9KftdTCp2d7vPy5OMe+PD198FjDFQh2T0OXrAIL3/mCSM1BkU6Pui5UQOGevX4c//642vvQHeOJSOhzs/Oqrr+K6667rUGfMmzcPM2f6scJQ8MHUqVOFd955RxAEQfjkk0+EMWPGXHFNa2urkJiYKFRWVgqCIAgPPvig8PjjjwuCIAgej0cYMGCAsGXLFkEQBOEvf/mLsHDhQkEQBMFqtQpNTU3t7bz44ovClClTOo2jqKhIACAUFRX5ErbQ8O7/Cef++pBP1/qLd7uF00/dIjSt/VdQ2g9n7X93a96UOxRBEATBvOMr4eQvrxU8TrvcoRCE4jV99k/hzB+XyB1Gj5wnjwgnH5kt2Mv3yh1KlxzVFcLJR2YLzpoquUNpV/facqF+5QpZY3DVnRZO/vJaoXXzp0Hvy21uEk7+ap7QunWNT9f7U4f4dKLtm2++ibvvvhvPPfccoqOj8Z///Ke9GkxJScGyZctgMpnw1ltvYcGCBXC73Rg2bBjefdd7vhBN03j//fexbNkyOBwO9OvXD//9738BAOfPn8fNN98Mj8cDQRAwYMAAvPfee75XiN3Q5hSgqaQQHkszVMYYUdps4zxaBt7SAn2YDg8GE6VSQTdqCmwlhYie/1PZ569x1eVg+g0AzWpljYMgQgGblQfLtrXwtDRCFR0ndzhdspdsA22IgmbQSLlD6RKbkgnQNLhTR8GmD5I7HO90idojMM28VdY41AkpMIydBfOmj2GYcC1obfAOO/dOqKcCPti5Mz7NwRo0aBB27tyJqqoq7NmzB3l5eQCAFStWYNmyZe3XXX/99aioqMCRI0fw+eefw2QytX9t/PjxKCsrQ1VVFTZv3ox+/foBADIzM1FcXIyysjLs378fa9asEW3XYu3gfEAQ4KwqEaW9S9mKt0Id3/GRYri5/FGqmPQF08C3NMJ59EDQ+vBVuB/wHMw8EtJQUg7b3itOBR+bI7TtTzj8KlAqn54jSOLyPFKsBkzfdLhOKmNHd/f5kxCcNkl2cO+Jafbt4B02WLat7fniXhIEAdZd66EbNj4oJ3iE35KpS6ii48CkZIq+XYPg5mDfvxO6/CkBTe5XOpvNFrS22f45UMUlw1Yk/XLcS3nMTfA0nA3r/a+CmUdCGkrKoSo6Dqq4ZHAKPvjZVXsYngvnoZP57MHLdZZHJe3oztVUARQFNm2g3KFAHZsI41VzYd78GXibuecbesFVexjuczUwjBN3cnubsC6wAECTMxqOqiJRV6w5Kosh2C2K2xlYbCtWrAha2xRFQV8wFfayHbKebdZ+wHNW+BZYwcwjIQ2l5VCTlafoJ1i2kkLQxhhoBgyXO5QOOssjk5oN15kTQd1SyFdcTSXUSf2DOiTnD9PMWwHeDfOWz4LSvlXEg507E/YFljanAHxrE1xnqkVr01ZSCHVyhnejOKLX9AXTIThssMt4wKezuhyqmIQOK2YIgugemzkErtPHwDvtcodyBe/+hNugGzkJlEr5+xOyadmAxwXXOfn3FuNqqhQxPNhGZYqFcfINsBSuhcfcLGrb3oOdt4p2sHNnwr7A0mQNAcVq4BRpmJDnHHAc+CHsn15JgembBiY1W9ZhQu5ERdjuf0UQwaLJygN4HlyNMnaYvxR3ogKe5oaQ+R7N9BsAUJTs+2HxTgdc506ATVdOgQUApukLAZqGedNqUdu1l+2A4BDvYOfOhH2BRalZaLJHwFG5T5T2HOV7IXCOkHnzBkKMHfV7oi+YBsehvUEbY++O4OLA1R4J6wnugDR5JIJLaTlUJ6aB1pva98NSEntJIejoOLCZeXKHcoXO8khrdFAnpMo+D8t16ijA87IckdMd2mCCaerNsHz/FdzN9aK1a921HpqBI0Q72LkzYV9gARePzTl+SJTH2faSQjBpA6FOSBEhMmW75557gt6HPn8qwLthL5P+BHXu1FHA4wr7J1hS5JEILqXlkKJpsJlD4FRYgSXwHtjKtkM/YrIij73qKo9MWrbsKwm5mkrvqsak/j1fLDHj1AWgNTqY138oSns/HuwcnMntbZT3LzAINDkFgMcN59H9AbXDO6ywl++JiKdXAPD0008HvQ9VdBw0A0fAVrQl6H1djqsu935DScmUvG8pSZFHIriUmEM2Mw9cTSUEj0fuUNo5jx0E39qkuNWDbbrKI5uaDdfp4xB4+f4uuZpKMGmDFDlvjdYaYJpxC6y7N8DdcKbnG3oQjIOdOxMRBZY6oR9UcUkBb9dgP7gLcHGKffOKLT9fmvO79AXT4Tx2QNTHv75wnigHmz5YUfvkBINUeSSCR4k51GQNgeC0w3XmuNyhtLOXbPMeX5aRK3coneoqj0xqNgSXE+66UxJH9COu9rCiJrhfzjjpetDGaLR+935A7Qh8cA527kxEFFgURUE7uCDgie724kKwmUOgjk0UKTICgPdThEoNe3GhZH0KggDueHnYDw8SRLCwaYMAFdO+1YncBI8H9v07oBs5OeT2J2T7DQAA2eZheVovwNNUp7gJ7peiWA2iZt0GW9EWuM7V9HxDFxyVxd6DncfNEjG6zkVEgQV452G560/D3XC2V/fzVjMclUXkaJwgoHUG6PLGSTpM6Gk8C97STAosguglimHBpg9UzDys9uPLQnAKB603QhWfDNdJeQosrqYKABQ3wf1yhvFzoIpNROu3/+11G7bd68GkZIKRYDPViCmwNINGALSq18OE9v07AEGAbsQkkSNTrstPKA8mfcF0uE4fC+iTiT+cFz91a/orcyhBTFLmkQgOpeaQzfRuOCrmRs69ZSsphCo+WZIfnL3VXR5ZGXd052oqQUf1gSomXpb+fUWpGUTNvgP2sh3gerEowGNphv3gLujHzZbkKWfEFFi01gA2I7fXBZatpBCagcOhiuojcmTKVVxcLFlf2iGjQemMkj3F4qrLoe6bDtpg6vniECdlHongUGoONVl54Fsa4blwXtY4BLcL9v3fQz9K2ceXdZdHJjUbrlPHIPC8hBF5cbVVYPvnKPrvro1+9AyoE1PR+s17ft/rPdjZuz2QFCKmwAIAbe5oOI+U+X0kgcfcBOeR/SH56DkQr7/+umR9UWoW+hGTYCvaIsmnYa46vA94vpSUeSSCQ6k5ZBVy8LPjcAkEm0XxC5C6yyObmg3BaYO7sXfTWHpL4HnvBPf0QZL221uUSoWoOXfCUbHXr+HpHw92ngCVMTqIEf4osgqsnAIIThu4ExV+3Wcv3Q5QFHTDI2d4UA66gmnwXDgf9EmzvN0K17kaMv+KIAKkMkRB3Tdd9oOf7SXboE5MA5MculuuMKnZACD5PCx33UkIDpvi519dSjdyMpiULLR8/Y7PH8hdJ4N7sHNnIqrAYvoNAG2MhqPCv13dbSWF0OYURMRwkpw0A4ZBFRMf9GFCrqYSEATFLuUmiFAi98HPgouD/cBO6EZdHRJDXF1RGaOhik2UfB4WV1MFUBTYdOXOXbscRdOImnsXuGMH4Dxc4tM91l3roYqJD9rBzp2JqAKLomloBuf7NQ/L3VQP7vghxT96DgcUTUOXPxX20m1BPVneWV0O2hAFdWJq0PogiEjBZg6B++wJWY67AgBHZREEhy0spnB452FJu6M7V1PpnY+qNUjab6C0eePA9s9By9fv9vgUS4qDnTsTUQUWAGgHF8B16qjPJ3PbS7cBaga6YROCHJnyzJ8/X/I+9QXTwVtbA94UtjtcdTnYjNyQ/rTrDznySIhLyTnUZHnP/HP6OfVCLLaSQjDJGWCS0mXp3x895ZFNHQDu1FFJV2VytVWK3v+qKxRFIWreErhqq+A4tKvba3882Dn4e19dKvIKrBzvTrrOKt9W5diKt0I3ZGzIVfdieOihhyTvk0nJhDqpf9CGCQXeA66mMqLmX8mRR0JcSs6hKi4ZdFSsLAc/85wDjoO7oAuRp1c95ZFJzYZgs0i2KpPnHHCdqVb0Du7d0QwcCU32cLR+8163qy9tuzcE/WDnzkRcgaWK6gOm3wCfnpC468/AdfIIdPmh8eYV26xZ0lb7gPdTib5gGhwHfxDlcO7Luc7WQHDaI2YFISBPHglxKTmHFEVBk5kny4ajjvK9EDhHyAwP9pRHtm2iu0TzsFynjgI8H1IT3C/lfYp1N1xnqr2jTZ1wN5yB8+h+6CXYuf1yEVdgAd7VhI7Koh73G7GVFIJitdAOGStRZATg3aNE4JywH9gpettc9SGAVnmP+SAIQhRsZh64k4chuDlJ+7WXFIJJzYY6IUXSfoNFFR0HOqqPZBPduZoq74H3yRmS9BcMmswh0A4Zg9Zv/9vpwePW3d6DnfUy7AIQkQWWJqcAvKW5x0NKbSWF0A6bAJrVShQZAQDqPn3BZuXBtk/8YUKuuhxManbQD/kkiEiiycoDXBw4CbcY4B022Mv3hMzTK1+xqdmS/T1yNVVgUgeCUkk38TsYouYugbv+NGx7N3X4c4H3wLZnE/T5U2X5nh+ZBVbmEFCsttthQtfZE3CfPRF2b15/rF27Vra+9fnT4Dxc7PNiBF85T1RE1PAgIG8eCXEoPYdMvyxQrEbSg58dh3YDLg66kZMl6zNQvuSxbSWhJBsu11SG7PyrS7Gp2dCNmITW9R90eIrqqCyGp6UBhvHS7X11qYgssCg1A83AEXB2U2DZSraB0hnbJ8VHolWrVsnWt/ebJgV7SaFobXpaGuFpPAc2M7L2v5Izj4Q4lJ5DSqUG2z9X0nlYtpJCsP1zoI5LkqzPQPmSRzY1G7ylBXxLY1Bj8Zib4GmqC4sCCwCirr0LnuYGWH/4rv3PbLvXe1eYynQ+ZUQWWIB3HpazurzTidSCIMBeUgjd8Img1KwM0SnD6tWrZetbZYyGNne0qKsJ25aRazLzRGszFMiZR0IcoZBDNnMIuOpySZ688DYLHBVFIbc/oS95ZNK8E92DPQ+Lq6kEgJCd4H45Jikd+tHT0bpxFXjOIfnBzp2J4AJrNOBxw3mk7IqvuU4dhbv+NPT5U6UPjGinL5gGrqYS7vozorTHVZdDFZsIVXScKO0RBPEjTVYeeGsL3HWngt6X/eAPgMcVUsODvlLFJIA2RAd9JSFXUwU6KhaqmISg9iOlqNl3gLe0wrpjXfscXv3o6bLFE7EFljohBaq4ZDgqrzw2x1ZSCNoYDU32CBkiI9poh44HpdGJ9hSLqy6PqP2vCEJKbEYOQNHelbpBZi/ZBjYrD+owKg7aUBQF5uKGo8HE1Xg3GA2nDZfV8ckwjJ8N86aPYd35DXRDx0t2sHNnIrbAAn7cruFSAs/DXrINuhGTQ35lRaijWS10wybCVrQ54GEHgXOCO3U04ia4E4RUaK0BTEpmUOdheZobcOGDv8JRsRf6AvmeTAQbm5od1EOfBZ737uAeJsODlzLNug0854C77iQM4+fIGkvEF1iehrNwN/w4BMXVVMLTVAd9hF07k24AACAASURBVG4ueqmlS5fKHQL0BdPgrj8d8ONy7tQRwOOOyCdYSsgjEZhQySGblReUlYQC50Trhg9x7rmfwlGxFzG3/AKGCdeK3k+w+ZpHJi0bnpYGeMxNQYnDXXcKgsMWNhPcL6WOSYBpyo1QxSdLerBzp7HI2rvMNANHALQKjsoiGCd5N6qzF28FHR0HNsImQndGCbtHawaNAm2MgW3fZrABrAThqitAsVowyZkiRhcalJBHIjChkkNN5hBYt38Jj7kJKlNswO15FxxtQ8u6lfC0XoDx6hsQNet20LrQPLrM1zz+uKP7MahyR4seB1dTCVBU2G64HDXvbkRdu1jSg507E9FPsGitHmzmkPZhQoH3wFa6HfqRV4OiI/qvBgBw2223yR0CKJUK+vwpsJUUQuCv3KXXV87qcrD9cyJy2FcJeSQCEyo5ZC8e/CzGUyzu5BHUv/oYLrz3f2D6ZaHvb95EzA33hWxxBfieR1VcMiitIWjzsLjaw1AnpoX032V3KJpWxA4AEV9FaHMK4DxSBsHtgvPoAfDmJrJ6UGH0BdPBt17odMWnLwRBAHeCTHAniGBTxyRAFZsY0DwsT0sjLnz4Iupe+gV4uxXxP3sO8fc+DSahn3iBKhxFUWDTsoO2kjBcNhhVuogeIgS8BVbr1+/AWV0Oe0khVHFJYNLD87FpqGLSB0EVnwxb0RZoB/u/8au74Qx4SwuZ4E4QEmCzenfws+DiYN76OcybVoNSM4i5+UEYJlwbkU+dAe+O7vayHaK3K3BOuM5Uh+QctlAT8U+wmH4DQBuj4SjfA3vZDuhHTQmZZauljQL+VdH9gdWB2LFD/Dd3b1AUBX3BdNjLvofAOf2+v224IhxXzPhCKXkkei+UcqjJzIPr1FHwnMOn6wVBgK10O879331o/fa/MIyfg6Sn3oZx0nVhV1z5k0c2NRueC+fBW82ixsCdOgrwHvIESwIRX2BRNA3t4HxYd3wF3maGLoRWDy7f7cHPv/eg2RmcnZNfeOGFoLTbG/qCaRCcNtjLd/t9L1ddDnVSf9B6YxAiUz4l5ZHonVDKoSYrD+A94GqqeryWO3UU9a8tx4V3/gQmqT/6PvEGYm5cFrbvVX/yyFyc6M6dFneYkKutAhgWTHKGqO0SV4r4AgsANLmjIbicUCemhcwqs6MtAjaeFuARgI2ng1NgffTRR0FptzeYxFQwaQPbd+f1h7O6XNHDg3a3AM4TvONFlJRHondCKYfqpP7eCdrdbDjqMTfhwkd/Q92LD4O3tiB+2bOIv/+PYPqmSRip9PzJozqhHyiNTvR5WFxNFdjUbFCqiJ8hFHTkbxiAdnABQFHQ54fO8OC/KnnEaoAELfB1LY+FWeLXynq9XvQ2A6EvmI6WdSvBW82gDSaf7uFtFrjP1cA0/SdBjq735n3nwaBoCm9MDs5wiNLySPgvlHJI0TQ0mUPgPH7lSkLBzcFS+AVaN6wCpVIh5qafwTBxXtgNBXbFnzxSNA2mXxY4kTcc5WoqoRt+lahtEp0jBRYAlSkGCQ8+DyZE9gRxegS8XcXj7oE0GBp45wgPXhBAh0hx2Fv6/Clo+eLfsJVth3HiXJ/uaT/QVKFPsGotAracFdDgFABExg8ZIvyxmUNg3vwJBN4DilZBEAQ4DuxE8xdvwdN0HsZJ1yNq9mKfPyhFKjY1G46KK49z6y2PuRmeC+cjdj6q1MgQ4UWa7OGgNVq5w/DJZ9UCGp3Aslwac9Mp1NmBoobgn2AvN1VUH2gGjfTrbEJn9SHQxmio41OCGFnvfXrcu0ihoslbOBNEONBk5UFw2OA6VwPu9HE0/OM3aHz7GagT+qHv8jcQc9PPSHHlAyY1G+760+AdVlHaa//ASSa4S4IUWCHojQoe05IpDI6hMLEvhWgW+KZW/B/Ojz/+uOhtBkpfMA3csQNwN9X7dD1XXQE2Y4hih34/qRaQagDcAlDRHJw+lJhHwj+hlkMmfRCgUqPpwxdR99eH4Gm5gLj7n0HCA8+CSUqXOzzZ+JvH9h3dTx8XpX+utgq0MQaq2ERR2iO6RwqsEHPogoDt5wQ8kOtNHUNTmNWPwtcnxS+w0tOV941QN3wiwLCwF2/t8VrB4wFXU6nYCe41ZgG76gT8bpQ3l/sbg/MES4l5JPwTajmkWS3YjFy4G88jesH96PvEP6EbMkbusGTnbx7VfdMBhhVtHhZXUwW2/2DFfuAMN6TACjFvVvJI1AELMn58g8xLp7G3XsB5m7g/oB9++GFR2xMDrTVAlzcetqLNPV7rOlsNgXOAzcyVIDL/fVrNQ6MCbh9AY0AUUHYhOAWWEvNI+CcUcxj/0/+H5D+8C9OUBWTF2kX+5pFSqcCmZImyklDg+YsFFpl/JRVSYIUQq0vAe0d4/HQwDVb1Y4E1J9X7/9+diow5PPrR0+A6Uw3X2RPdXsdVlwMqNdjU3h8SHUyfVAu4NpWCiaUwvA+FsiA9wSIIOdB6E2hteJ51JyUmNRvcqSMBt+OuPw3BYSXzryRECqwQsvq4gFYOuG9wx7T11VMYk0Dh69rg7equJNqc0aD0Rtj2df8Uy1ld7t3vhdVIFJnvTpgF7K4TcMvF7TVG9KFQdkGAIJAiiyCIH7GpA+A+fwq807ed8bvStvErGyKr5cMBKbBCyBsVPOakUciMunL8fF4ahQ2nBbh48X5AV1ZWitaWmCg1A/2IybAVb4XAd11UctXKPeC5bXjwunRvLkfEUWhwAOfs4vel1DwSviM5DA+9ySOTmg0IPFxnApvoztVWQp2YFra75CsRKbBCRFG9gL31P05uv9zcdAotHLDzvHgF1vLly0VrS2z60dPhaaprP2fwcp7mBnia6hRbYH1yXMDcNO/wIAAM7+P9NRjDhErOI+EbksPw0Js8Msn9AZU64HlYbRPcCemQAitEvFn5/9l78/C4qitf+92nSrMsq0qSR0meR1kSGMtJSEIgBENsMCRkIkPfmC9NOgkkuekPEm53bsPtTieQvp3ujvkaukPSU2KM0wk22IwJIZABA8a25EE2LlmSZ8uSXCVLVqnq7O+Po7JlWSXVcKrqDPt9Hh6jUtWprfM7wzp7r/VbUapLYHXN2NUfV1UKphSZa9ewfv1607ZlNvlz6vCUV8VNdh88bAReBbOtl+B+OCTZfvri8iDA7EkwKS8zAZaVdVQkhtLQGaSio/AafQPDaQRYMjzI0NGASnDPMirAsgHBsORn70i+sEjDq40dYGlC8OFqwdZO8/KwrFwaLjSN4quuY2Dnq8jI0GW/D7ftw1MxDc/kihyMbnx+3qZTOGJ5EAz9GvyC3RmoJLSyjorEUBo6g1R1zK+ez9CRQyl/b/joIdCjagYry6gAywb89B2d81H4wuLx5VpTq7Gnx/BXcgPFV12H3h/i/P63LvvdYNteS85eATw5vDxYmndpsNwwnOiuUCgUI8mrns/Q8cPIoXBKnw+3t0JePnkz5pg7MMW4qADL4kgp+ed9OrfUCmaWjG8Od8NMgUfANhNnsaxM3ow55E2ffVnrHBkeZOjIO5bMv2oLGrl0YzXnbqyA/b1wPqKCLIVCcZG86vmgRye0polHuH2/UVGt/MiyigqwLM4fT0mau4mb3D6S8gLB+6YJtpqUh/XQQw+Zsp1MUnTVBznf8kf08/0XXgt3HjCmw+fU5XBkYzPW8mCMRr8gmoGWOXbQUTE+SkNnkKqO+TPmgKalnIcV7mglv1YtD2YbFWBZnEf36cyZBDdUJ9baYE2N4NfHJAMmzIL09/dP/KYcU3zVtcihQQZ2//7Ca4NtexEFRUb1jcV4MiBZU3v58iDAMr9AYL6jux10VIyP0tAZpKqjyC8gb2ptSpWE0b5eomdOqAT3HKACLAvTfV6yMSD54mINLcHeUatrNAai8Jvj6d+kH3zwwbS3kWm8vinkz6u/pJow3LaX/FmLEZonhyO7nEBQ8maX5ONzxj7tSvOE0TLH5EpCO+ioGB+loTNIR0fD0T35AOuCwahKcM86KsCyMP9+UEeXsG5R4jIt9cGsUnPtGqxO8fJrGTywk2iwGykl4cP7LJl/9fM2nSIPrBljeTBGY4VqmaNQKC4nr3o+Q8fakNFIUp8Lt7eilU7G45+aoZEp4mHLAMsN7USklDy6T+f2OYIpRYl3PhdCsLpGY2un7or9BFB8xTWgafS//Vsip46gnwtSYMEAa7zlwRiNw1YNbtFOoVAkRn7NfIgMMXSiI6nPhdv3GzP6Ca6CKMzDlgHWqQy0E7EavzkuOXA2seT20aypFbSFoPVsemPo6upKbwNZQiuZROGSFfS/9WvCh/eBEJbLNwgEJW+NszwYo8EvODMIx0xMubGLjor4KA2dQTo65s2cB0IklYcldZ1wxwG1PJgjbBlg7e11/tP9o/t0FpfDNdOSf+q4boag0EPazZ/vvPPOtD6fTYqv+iBDHQc4t/1F8qbNQisqyfWQLmFTYOLlQTCWCMHcPCw76agYG6WhM0hHR62gCG9VdVJ5WJGuY8iBPlVBmCPsGWD1ODvAOtkv+UWb0XcwlWndYq/guhnp2zU88MADaX0+mxTVvQtRUEz4ULMl7RmebNO5uVZQMs7yIBj5c2V5mOrobicdFWOjNHQG6eqYVzM/qRmscLvRXFoFWLlBBVgW5McHdLwa/MmC1OVZUyN49YQkGE59Xy1fvjzlz2YbkV9AUcPVAOTPsZaD+6GgZEcXY5qLjkYMt8wx06rBTjoqxkZp6AzS1TG/ej5DRw8h9WhC7w+3t+KdUo1WXJrW9ypSw54BVq9zE92juuRf9ul8ap7AV5B6UuLqGo2IhBePOnM/jUXJu2+CvHwK5jfkeiiXsCmgU+yN36h7NKqSUKFQjEVe9XxkeJDIqSMJvT+W4K7IDbYMsIJhaAvlehSZ4YWjksN9qSW3j2ROmWBJOWxLMw/LThTMW8bM72zC65uS66FcwpMBnTU1Ey8PxmisELSeVS1zFArFpeTPnAeQUB6WHAozdKyN/NqFmR6WIg62DLAA3jztzJvPo/t0rqiAlVXpl9SurtHY1inRU5zte/zxx9MeQ7YR+QW5HsIlvHNW8vYZ+EQCy4MxGv2gS9jTY84Y7Kij4lKUhs4gXR214lI8ldMZ6pw4wAofPQTRiJrByiG2DLCmFcMbXc4LsDr7JM90GM7tZniWrKkVnBiAnWdS+/yOHTvSHoPb2dQ2vDw4QfXgSOp85rbMUTraH6WhMzBDx/wEHd3D7a3gzSNvxpy0v1ORGrYMsJb6nDmD9aNW42b8mfnmyPK+aYJJeanbNTzyyCOmjMPNPBkwqgeLvYkHWCV5ggWTzbNqUDraH6WhMzBDx7wLie7jX9fD7fvJr56P8Oal/Z2K1LBngFUueKsr9aUvKzKkS/51v85n5mlMyjfHcTdPE6yqFmztdM5+shMHz0p2Jrk8GCPm6K5QKBQjya+ejzzfT+TM8XHfF25vVfYMOcaWAVadTxAaggNpOpVbiWfaJcf74YtpJrePZk2NxvZTktMD6madbTYFdEq88OEEqwdHErNqcGq1rEKhSI28aiPRfbw8rGjfWaJnjisH9xxjywBrSblxw3LSMuGj+3TeNUVwZaW5/aI+XCOQwHNHnLOv7EIqy4MxGisEPYNw5FwGBqZQKGyLp7QcT3nVuHlY4Y5WAJXgnmNsGWBNyhcsKIM3HBJgHQpKXjgq07ZmGItpxYKrKgXbOpPPw1q7dq3p43ELB3olu7pTWx6Eiy1zzFgmVDraH6WhMzBLx7yaBeM6uofbW9FKJuOpmGbK9ylSw5YBFsCKKsGbDqkk/Nf9OuX58Im5mel2vrpG8FynJKInt7/uvvvujIzHDWxqS315EKCmBMrzzUl0VzraH6WhMzBLx/zqeQwdeSduCoFhMLrIlGp0RerYNsBqqhK83ZV80GA1BqOSH7fq/I+FWkpLSYmwplbQG4Y/nkpuX61atSoj43EDTwZ0bpklKEpRUzNb5igd7Y/S0BmYpWNe9Xz0/hDR7pOX/U5KSbhDJbhbAdsGWCsqBQNR2GuSGWOu+OVhyenz8MXFmZOiqUpQVUjazZ8VidHaK9ndDZ+Yk56mqmWOQqEYi/zq+QBjLhNGuo4h+/tUgrsFsG2AdWWlQBPYfpnw0X06H5guWOLL3FSuJgQ3VQu2ppCHpUieTQGd0jy4KcXlwRiNfsHBIPSrljkKhWIEnskVaGW+MRPdw+37AVSAZQFsG2CV5hm99uyc6L6vR/LK8cwkt49mTa1Gc7fhFp8oTz31VAZH5FyebNO5pTb15cEYDRWxljnpHeNKR/ujNHQGZuqYX72AoSOHLns93N6Kt2omWvEk075LkRq2DbDAWCa0s1XDY/t1qgrhI7Mzn4i4qlrgESRVTbhhw4YMjsiZ7O+VNKdRPTiSZT5jlnZXiq2OYigd7Y/S0BmYqWNe9XzCnQcvS3SPJbgrco+tA6ymKiMJeDBqvyCrPyL59wM6dy7SKPBkPsDyFQiunirYloSr+8aNGzM4ImcSWx68sTp9TYu8goWT07dqUDraH6WhMzBTx/zq+eh9vehnLz6ByUiYoaMB5X9lEWwdYK2oEgzp0GzDliJPBiS9Ybgrg8nto1lTI3jpqOS8yunJGJvadNaasDwYo8GvEt0VCsXl5NUYie4j87DCRwIQjagZLItg6wCr0S/wCns6uj+6T+fGasHcsuz5lKyu1eiPwG9P2G9/2YF9PeYtD8ZoVC1zFArFGHjKq9BKyi6pJAy37wdPHnkz5uZwZIoYtg6wCr2Cer/9Et3f7pK8fio7ye0jWeYzDCyVXUNm2NSmM8mk5cEYjRWCs2Ho6DNtkwqFwgEIIYw8rJEBVkcr+dXzEN68HI5MEcPWARbY09H9sX06M4rh5trsuuwKIVhdo7G1U09oRmTdunVZGJVz2BTQWTtLUGiiYWyDP/2WOUpH+6M0dAZm65hfPf+SGayh9la1PGghbB9gNVVp7Omxj1dQKCz56SGdLyzW8GrZb2OwplZwKAgHz078XuUenTh7eyQtPeYuDwJUl4CvIL2WOUpH+6M0dAZm65hXM59obxfRUC/Rc0EiXcdUgruFsH2AtaJSEJWw0yaJwD99R6c/Al9YlJtd/8EZggIPCZmO3nHHHVkYkTPYFDCWB1fNNDdoFkJcyMNKFaWj/VEaOgOzdRzp6B5ubzVeUzNYlsH2AdYyPxR47JGHJaXk0X06N9cKakpz04SzJE9w7fTk7BoUE7OpTedWk5cHYzRWiLStGhQKhfPwVExHFJYQPvIO4fb9aCVleCqm53pYimFsH2DlaYIrKuxhOPr6KcmubrKe3D6aNTWCV45LQmHr7zM7sKdbsicDy4MxGvyCg2fh3JDSS6FQXEQIQX7N/AszWPm1ixAiNw/visuxfYAF0FRpj0T3R/fpzC41fxkpWVbXagzp8Ktj4++z1157LUsjsjeb2nTK8gy3/EzQ6BdIoCXFljlKR/ujNHQGmdAx5ug+1KES3K2GIwKsFVWC1l4IWnhGpvu8ZGNActcSDU8OkttHMq9MsGgybO0YPw/r4YcfztKI7M2mgLE8mClH/jofaAJ2d6f2eaWj/VEaOoNM6JhfPZ9o90n0/pBKcLcYzgiwKo0n/B0WnsX6j4M6ER3uXGiNXb66RmNb5/gGlk888UQWR2RP9nRL9vbCxzO0PAiG39uiyalXEiod7Y/S0BlkQse84UR3gPxaNYNlJaxxt0+TxeVQ4rVuonssuf2jcwRTi62xPr6mVnCsH3aNMytSXFycvQHZlCczvDwYo7Ei9ZY5Skf7ozR0BpnQ0Vs1A5FfiLdyBlrJJNO3r0gdRwRYHk2w3MJ5WL89IWk9m/vk9pG8f5qgNG/iZUJFfKSUbAro3DY7c8uDMRr9RiWhapmjUChGIjQP+XPrKFjQmOuhKEZhnTt+mjRVCcvOYD26T2fhZLh2ujVmrwDyPYIbZiq7hnTY0wP7euHjczJ/GjX4BcEhaFctcxQKxSgq7/zflN/+pVwPQzEKxwRYKyoFbSE4c95aAcOpAcl/txl9B61WPrumRuOPp2TcfXbvvfdmeUT2YlObzuR8uCHDy4NgLBFCanlYSkf7ozR0BpnSUeQXILz5Gdm2InWcE2BVGTegtyy2TPjLwzq6hM8tsN6u/nCNQJfw/JGx91ltbW2WR2QfpJQ8GdC5LYPVgyOZUQwVBaTk6K50tD9Kw+zQH5H8sCWaMc85paO7sN5dP0Xml8HkfOslum9ul7x/mqCy0FqzVwAzSgRXVsTPw7rnnnuyPCL70NID+zNcPTgSIQQN/tQc3ZWO9kdpmB1+9o7kq3/QueX5aEb629pRx3/Zp3Oy31r3VbvgmABLCMGKSms5uvcNSX51VLJ2lvWCqxirazSeOyKJ6tbZb3ZgU2B4eTCLprHpVBIqFIqJ2dqhM3cSbD8tufm5zARZduL0gOSLr0X5j4OqGCoVHBNgwXCiu4WWCJ8/IgnrsHaWdXfzmlpB9yC8bqHA1OrElgc/MluQn4XlwRiNFYJDQSNwVygU5jIYlbx0TPKFxRrP3uRRQRbQPDxj3qx6oaaEde/8KbCiSnD0HBy3yHTmlnadOp/hnG5VVlYJKgpga8fl+2z//v05GJH1ae6G1rPZqR4cSaxlTrIXO6WjvZFS8t9vBnI9DMfz6glJ35Axq//+6RrbhoMsM5cL7XYuxtpzpdqmy+04KsBqGk50t8IyYUSXbO2Q3Grh2SswPMRuqhFs67x8Cvi+++7LwYisz6Y2nfJ8+FCWe0ouKQePIOk8LKWjvXn5mORjb9fw0lG1TJNJtnVIZpZAg9/4+ZrhIOv1U5K1JgVZdjsXYw9ze3uNe5oiOax990+SmhKoKsQShqO/Pyk5M4il869irKnR2HkGjp67dL+tX78+RyOyLheqB7O8PAhGy5zF5bDrTHKfUzram9azxnn5N2+rACuTbO3UWV0jLrHTiQVZfzApyLLbudjSY9xXB6PwTjDXo7EfjgqwhBCsqLJGovuWdsm0oouzalbmxmqBJuDZUaajqqT4cnZ3w4Gz8IksVQ+OptEvkrZqUDram0DImLl85bjk1eMqyMoE75yVHDhrPGyOZmSQdesLUQbSCLLsdC7qUtLSI/nUPGOfqDys5HFUgAUXHd1z2VJESsnmdp1bZgk0i5mLjoW/UPCeKUK1zUmATQEdXwFcPyM3ujb4Bc3dEl21zHENgaDk2umCej98Z6c6RzPBtk6dfA2uj7Ps/4HpGltv9PD7k5K1aQZZdqGjD/qG4LoZgqlFKg8rFRwXYK2oFJw+D53ncjeG/b3GdOraWvvs3tU1gpeOSQaj6iSKx0hz0WwvD8ZorBCEhuBwKCdfr8gBgZBkfpngL67w8PwRyRunVZBlNts6JR+YLijNi39eXzvjYpCV7kyWHYjNWC3zCeqHH+wUyWGfCCBBYo7uuTQc3dyuU+yN/zRkRdbUavQNGZU0MR566KEcjsh67O6Gg8HcLQ9Cai1zlI72RUpJIAgdO37Dx+YIFk6G76hcLFM5NyT5zXHJ6pqJr9exIOu1E6kFWXY6F1t6JJPzoboE6n0qwEoFxwVY04sFM0tyW0m4pV2yaqagyGufAKvBDzNLLrVr6O/vz+GIrMeTseXBHAbO04qgsjC5ljlKR/vSPQjBISgdOIVHE9x/hYfN7VLd7Ezk18ckg1HjITMRrp2hsfUmI8i6Lckgy07nYnO3ZJnPSPqv9xsefJlqIeRUHBdgATRVipxVEp7sl/zxlOTW2fbatUIIVo+ya3jwwQdzOCJrccFcdJYgT8tdgCWEoDHJljlKR/sSCBk63/+FTwHwmfmCWaXwtzujuRyWo9jaIZlfBgsmJ35eXzccZL16QvKRF6OcTzDIstO52NIjqfcb+2SZHySwt1cFWMlgryggQWKVhLlIdH+mQyIErElgutlqrKnROHDWqKhRXMqubiOvLpfLgzFUyxz3EBgujZ9bZvybpwm+1aix8ZCkVd3s0kZKybZOPeHZq5FcN0PjmRs9/Pa45LYkgiw7MKRL9vfCMp/xc51PIICW7pwOy3bk/m6RAVZUCnrDcCgHvh1bOnSuniKoKrJfgHX9TEG+xpimo27n6Xaj9+AHLZBX1+gXBEIQDDvngq4Ym0BI4i+AyfkXj7vPL9SYXgzf26VmsdKlpccoiEok/2osPjhT4+kbPbzisCDrQC8M6VyYwSr2CuaVKauGZHFmgJWjRPf+iOTFI9Zu7jwepXmCD0wXF/Kwurq6cjwi63DgrKTOl9vlwRgNw4nuiZZNKx3tSyAomTtJXKJhoVdwb4PGfx6UHA6pG146bOswCpI+MD318/r6mcZM1ivHJ14utMu52NxzsYIwRr1fXHhdkRiODLAqCgVzJmXf0f2lo5KBqLWbO0/EmlrBb45Lzg1J7rzzzlwPxzK0hWDOpFyPwmBJOXhF4pWESkf7cihoLA+O1vBPF2v4C+ChXWq2OR22dko+NFNQkKbtSizI+s1xyUdfih9k2eVcbOmWzCg2PBJjKKuG5LFvJDABMcPRbLL5sM6iybCoPPezHKmyukYjrMOvjkkeeOCBXA/HMgRCxkyCFSjwCJb4Em+Zo3S0L7HjbrSGJXmCb9Rr/LhVv6zFlSIxegYlvz8px3RvT4Xrh5cLXz4muT1OkGWXc7GlR14yewXGbNbJATg9oI63RHFsgLWiUrCjSxLNUoPKqC55ukPaevYKjEqaBWVGZc3y5ctzPRxLMBCRHO+HORYJsMBwdE+0klDpaE/CUUnnOZhbJsbU8MtLNYq98He71SxWKrxwRBKV8GETC5I+NFNjyyoPvx4OskYbN9vlXGzuvlhBGCP2s3J0Txx7RwPj0FQlOBeB1rPZ+b7tpyWnz8OtNs2/GsnqWo1tnXpO2w1ZifY+499YJZcViFk1qJY5zqWjD3QJc+MsTZflC766TOOxfbqaVUiBbZ06DX6oKTX3mn1Dtcbm4SDre0UUhwAAIABJREFUoy9eHmRZnXNDkkAIlo0KsOaXQYFHJbong2MDrOWV2U1039wuqSqEd0+xf4C1pkZw5Bw0q5JcwEg0BmvNYDVWGA8QAdXh3rHEPLDGW5r+ap2GR4MftKhZrGTQpeTZTslqk5YHR7NqOMj61THJ7TYLsvZcSHC/9HWvJlhargKsZHBsgFWWL1g0OXuO7lvadW6uFXgsUGWWLtdMF5R44cFfbs/1UCxBW0iSp8HM4lyP5CINw0+XiTi6P/7445kejiIDBIISr4Ca0vgaVhQKvrxEY/0enZ5BdeNLlDeHVxzW1Gbuer2q2lgufOmY5GPDy4V2OBdbekAAS32X75tlPkFzT/bHZFccG2DBcKJ7FioJD56V7Ou1d/XgSAo8gvdPE7x5tjDXQ7EEgRDMKsVSwfO0YsGUIhLKw9qxY0cWRqQwm0AIZk0yZg7G0/Ab9RpDOvxwj5rFSpStnRJfQeZXHFZVa2y+wcOLR40g6423d2X0+8yguVsyr8zwvhpNvV+wp0elJiSKMyKCOKyoEuw8IxnKcKL7lnadQg/cYAETSrO4okKgT1+a62FYgjYLVRCOpNGfmKP7I488koXRKMxmZOXqeBpOLRbctVjjH1p0Qsp8NiG2dUhurBZ4s/DQdGONxlPDQVbwo3+f8e9Ll5EtckZT7xf0DUF7KMuDsimODrCaqgSD0czb+29uN7xUSvKsdxNOlXq/kYellh2MpRor5V/FUC1znE3MZDQR7m3Q6BuCf96nZrEm4kS/5M2uzOVfjcVNNRp/u0LjyYBMqjl0Log1eR6LWOClDEcTw9EB1hUVAo/IrOFo13nJ707a355hNLEcH7cnNEopaQtZq4IwRoNfcLgPzqpZC8chpbxgMpoI1aWCdQs1/m+zbvkbeK557ohEADdVZ/eh6ZrpgqiEnRZ+KDo9IDk5QNwZrBnF4CtQ94VEcVZUMIpir6DOl9lE920dEl3CLRlMlswFi8ohT1MnUvcgBIesVUEYo1EFwY6lZ/i4S2Zp+puNGmfOw4/2q1ms8djaobMyB/1iG/xGr9dsG2AnQ8sYLXJGIoRgmU8oL6wEcXSABYbh6BunM3fB2dyu864pgmnF1rsBp0OeJijsbWe3y60a2hIolc8Vi4eD4ImWCdeuXZulESnM4oJFQ5lx3CWi4dwywafnCR7erdvKFiCbDOmSF45I1phoLpoo+R5B8ZkDbLdygNUtyddgweT471EtcxLH+QFWlaC5m4x0OT8fkTx/RDrCXHQsmqYXuP5EivlMWaUP4UjyPYIl5RNbNdx9991ZGpHCLA4NH3cxk9FENbz/Cg9Hz8F/HHT3eRuP352QBIfIav7VSN5fW2zpGazmHmn0Oh0n+b/eB629RqcBxfg4PsBqqhJEZGJ+Qcny62OScxFYW+vM3bhq6QyaXV6S2xaSTM438g6siOHoPv57Vq1alZ3BKEwjEDJsBMoLjBtdohou8Qk+Nkfw3Z1RIllqE2YntnVKphXBlZW5+f6PLZ/FgbPQa9HioZbu+PlXMer9xj11f2+WBmVjnBkZjKDeL8jTMpOHtaVdMncSLPVN/F470lChSnIDIWP2SghrzlI2VhjT9dnquanIDslUEI7mL6700BaCDYfUMTGarR06H64RaDk6n5uqjO/NZOFVqkgpx2zyPJo6n6okTBTHB1gFHkGjX5g+LatLyZYOnVtnaZa9+aZL/fCJlGhTYSfSFrKmRUOMBr+gP3JxSUnhDAKh+D0IJ6KxQnBLreBvd0ZV4D2CwyHJ3l5Yk8MVh0XlMCnPmonuHX0QGpp4Bqu8QFBTYuRrKcbH8QEWGHlYZj8xvNUlOd4Pax2afwXwxoubXV+SG7CoyWiMxoqJg+CnnnoqW8NRmEQgKC8kuEPyGv7FlRr7e+EXh9177o5mW6eOV8CHcmgIvWXzZlZUCrafsp4usev86CbPY6ES3RMjoQDr4MGDXH311SxatIiVK1eyd+/eMd/3zDPPsGTJEhYuXMjtt99OKHRxben111+nsbGRRYsWcf3113Ps2LGkt58qKyoF+3qhb8i8A2JLu5Ej8b5p1r35pssTT2yg3idcO4MV1SXtIWsmuMeYUiSYVjR+juGGDRuyOCJFugzpko5zl1auJqvhu6ZofGim4DtvR5EuzqEcybYOyfunCSbn5+6avWHDhqy1cEuWlh5JWR7UlEz83nq/UEuECZBQgPXFL36RP/uzP6O1tZVvfvObfP7zn7/sPX19fXzhC19g8+bNHDhwgBkzZvDXf/3XAOi6zmc+8xn+6Z/+idbWVlavXs3Xv/71pLafDk1VAl3C2yYe1JvbddbUZKfVQq7YuHEjDS4+kY6cg4jkkpkEKzKRo/vGjRuzOBpFunT0gS5h3giT0VQ0/MsrNXZ1w9YOd56/IxmISH59TLI6x36FGzduZOUUwdFzcOyctXRp6ZEs84uEUl6W+QQdyuR4QiYMsE6dOsVbb73FZz/7WQA++tGP0tnZSSAQuOR9zz77LMuXL2fhwoUAfPnLX77w1PXWW2+Rl5fHBz7wAQDuuusunn76acLhcMLbT4elPijymJdY2BaUNHfDrQ5zbx+Ler/gwFlc6Q4d88Cycg4WGHlYbp1ldCKBoDnea9dME7xvquBvduqun8X6zXHJQBTW5MieYSSxRHer5WE1d8sLebcTEcvT2uPSh+9EmfBo6+zsZPr06Wia8VYhBLW1tXR0dFzyvo6ODmpray/8PGvWLI4fP46u63R0dDBr1qwLv5s0aRJlZWUcO3Ys4e2ng1cTXFlpXqL7lg6dfA1uzHKrhVzQ4Deepve5sCQ3MLzCPbs0t+OYiMYKQXufdUu/FckRCEk8AmrSPO6EEPzllRqvn5L86pi7j42tHZLZpYY5b66pKYEpRdYKsIZ0yf5eWOZP7P2Ly8Ej3J2fmwimhfNWr6RbUSlMs2rY0i754AzBpByu5WeLOhdXEraFJDOKodBrbZ1jLXPcqJETCQRhVun4Zo+JsqpasKJS8Ddvu7d9jpSSrR06a2qtUfEthGBllfmV7elw8CyE9fgtckZT4BEsmgzNLu/0MRETBlg1NTUXZqLAOFhHz1YB1NbW0t7efuHnw4cPX5iZGv27UCjE2bNnmTFjRsLbH8nq1atZu3btJf+95z3vuazS5oUXXrjQYqKpSnAwaDzlf+UrX+Hxxx+/5L07duxg7dq1dHV1XfL6X/3VX/HQQw9d+LlnUPLKcZ2j2x5j//79l7z3hz/8Iffee+8lr/X397N27Vpee+21S17fsGED69atu+xv++QnPznu3zGSdP4OMGYd165dG/fvWLduHZPyBXMnwY6TYdv+HSNJRo9AUBI+2mr5v2NROeQPt8wZ6+9Yt26dI/QAZxxXE/0dgdClFYRf+cpXeO9735vS3yGE4Is1Z3jluOS/Xj+c1b9jNLnSY38vHO6D1TUi53/HunXr6Ojo4OALG3j9ZOSSpdtc6hGzXIgeaU5Yj3mFA2x4tdnR5/nNN998SZyxZs0abrjhhsvGEBeZANdee638t3/7NymllJs2bZJNTU2XvScYDMopU6bI/fv3Syml/MpXviLvvfdeKaWU0WhUzps3T7788stSSim///3vy49//ONJbV9KKd966y0JyLfeeiuRYV/Cvh5d8i9h+dKRaNKfHclPD0Yl/xKWnSE9re3YgZ/97GdSSilvfX5I3rB1KMejyT7veWpIfu7X9vi7r/jvsPzCK2OPNaajQsojfbrsGrD2ubv8F2F5128jl7yWjoZRXZf1Pw/LG7fZ41g2m7/bFZGFj4dl/1DudY/p+GyHcR852Jv7MUkp5V++EZHT/jOc1Gf++q2I9P17WOq6Nf6GbJFMHJLQEuFjjz3GY489xqJFi3j44Yf5yU9+ciEafOyxxwAjr+pHP/oRt912GwsWLODYsWN8+9vfBkDTNP7rv/6Lr33tayxatIht27bxgx/8YMLtm8nCyYbBW7rLhFvada6qFFSX5n6qOdPccccdgHuTqNtGzSRYmUa/YFec6fqYjm7nUFByxS8ifP0P0VwPZVwCwctNRtPRUBOC/3WFh+ePyIw2vrcqWzuMlI4iCyz1x3RcMZzobpXGzy3dckKD0dHU+wU9g3C8P0ODcgDeRN60cOFCfv/731/2+oMPPnjJz7fccgu33HLLmNt497vfza5du5LavploQnBVZXr+I+Go5NlOyZ835L4SJZvU+wUnB+DUgGRKUe4vUtmgPyI5MWD9CsIYDX7BkwGdqC7xONg6JFW6zks+/GyErvOwcxxLi1zTfV7SGzbfGuTjcwT/uwz+9m2dX65yz/UrGJa8ekLyj1db62+uLBTMmWQkun96fq5HY7S9Sbanbiwga+6WzChR15yxsNZRl2FWVKWX6P7KcaMTuxvsGUbSMOJEcguHhysIU21Xkm0aKwQDUXhHtcy5jIGI5NYXovSG4Rv1Gq1njaopKxIImWPRMBqPJvhfV3p4ql266jx+8agkImG1BewZRmOVRPdzQ5JAcOIWOaOZPQlKvKon4XhY76jLIE2VRjn76YHUDojN7ZJZpYZ1gRuIJSTOL4NCj7sCrJgXkV1msGItc8YyHB2dWOomdCn53G+ivN0leeZGD7fUCoZ0eOdsrkc2NoHhAHlu2aWvm6HhZ+YLZpXC3+609hKpmWzr0FlaDrMtch6P1LGpSrCjSxLJcbC/t1ciSdyiIYYmBHU+1TJnPFwVYK1Io5O5lJIt7TprHdzceTQPP/wwYDz91rmsZU5bSJKvwYwE2kZYgcpCwYzisVvmxHR0I/e+rvOLNsmGD3pYOUW7YDtiVYPEQEhSng++gkuvMWZomKcJvtWosfGQpLXXmn+/mehSsq1TsjqHzZ1HM1LHpipj1nlPTw4HBLR0gwCWlid/X6v3Gw7wirGxzpGXBeZMAn9BaonuO89A5zlnN3cezRNPPHHh/+v97vI8CYSMKXDNRsF0g3/sljkjdXQTP2yJ8vfNOv90tcats41LXVWRoKrQ2gHW6NkrME/Dzy/UmF4M39vl/FmsnWfgxACsqbHOOTxSx+WVAk3k3nC0uds45kryUgmwBHt7jL6tistxVYAlhGG6l8oBvaVdpyzPaD/hFoqLiy/8f4Nf0NIjXXMitYWk6XkwmaaxYuxZxpE6uoXNh3W+9gedP6/XuLvOc8nv6nzCugFWcOz8K7M0LPQK7m3Q+M+DksMha+4Ds9jWYVyz32uha/ZIHUvzBEvLYXuOKztbehJvkTOaep/gvMr9jIurAiwYTnRPYYlwS4fO6lpBvsc6J2s2qfcbJ9Ihl5xIgZC0Tf5VjEa/oPOcUYnmZl4/pXPHr6PcPkfw8Lsuv8TV+QR7LbpEFghJ5mXYGuRPF2v4CuChXc62bNjaKVlVLcizcFVtkwUS3Zu7jSbPqVDvwgKoZHBdgNVUJTjeD0eT6GTe2SfZ0UXSZaxOosFF7ViklLSFLk80tjpu0igeh4KSW56PsrxS8J/XesZc4q3zwQELVhIO6ZKOPvMrCEdTkif4Rr3Gj1t1jiVxHbQTXeclr5+SlqweHMnKKkFzt2ELkwu6zht2NKnOYFUVCaYUqUrCeFj76MsAKyqHE92TeGp4ukPHK+DDFlrLzwYj2x1McdGJ1HUe+obsU0EYY1E5FHguD7BGt61wKjGvq/J82LzKE7eH5FKfUUl40GKVhJ19EJVjB/Zma/iVpRrFXvi73c6cxXqu06iMs9o1e7SOTVUaUZk7b7ZYi5xUZ7DACM5aXPxQNx6uC7BmlsC0JDuZbz4s+cB0QXmBtU7WTDO6H2SDX7DbwiaNZtGWIS+iTOPVBHW+y60axuvr6RTORyS3DXtdPfthLxWF8bWzaiXheB5YZmtYli/46jKNR/fpKdvWWJltnUbHjWnF1jqHR+tY7zceinK1TNjSY1RLL5ic+jbq/cqqIR6uC7CEEEnlYQXDkpePS251UfVgjHvuueeSn+t9whUzWIFhk9E5NjEZHUnDGC1zRuvoNHQp+ZPfRNnRJXn6Rs+EOUyVhcZsrOUCrCB4BNSUXv67TGj41ToNjwY/aHHWLFZElzx3RLKm1nrX7NE65nsEV1QItp/K1QwWLC4nrTy1er/gnWDuljmtjOsCLDDysN48LS/pZB6P5zolQzrc4jL39rFoqBAcCkLfkLNPpLaQxFeALWcsG4erPXNtXphNvrld5+dtkp990MO7piR2ntaVW6+SMBCS1Jamd7NLhopCwZeWaKzfo9M7aK19kQ6vn5L0DMJqiy0PxqMpxcp2M2juSb4H4WiW+UACey12PlkBV0YNKyoFZwYvtkMZjy0dOg1+6zgB55J6iy6tmE0gKG05ewWGVcNg1Hr5RZli/Z4of7db5x/fo3Hb7MQvZ1a0ajgUzL41yDfqNcI6/H97nTOLtbVTUlVoPEjbgZVTBAeD0JPlIFdKSUu3ZFmKCe4x6nwCAbTk2DDVirgzwErQ0X1Il2ztkK7rPRhj//79l/y81AeagN0ONxxtC9kv/ypGrJJwpKP7aB2dQszr6hv1Gvcs80z8gRHU+YwgNBy1TpAVz2QUMqfhtGLBuoUa/9CiO2aJZ1uHzk3VwpImwWPpGAsE0+mTmwqd5yA4RNoBVkmeYG6ZsmoYC1dGDlOKBLWlEycWvnbC6GzvJvf2kdx3332X/FzkFSxwwYlkRw+sGBWFgpkllya6j9bRCWwf9rr6yGzB98fwupqIOp8gIq010xfPZBQyq+G9DRpnBuHHrfafxTrSJ9nVDWssaqkzlo4LJ0NZXvYT3WOVf+kuEcJwfq7D7wupYM2jMAusqBQTPjFsaZfMKIarKu15s02X9evXX/Zag9/ZPQkjF7yIcj2S1GkcpdFYOtqZQFBy8/NRrhzH62oillpsubtn0HiYixdgZVLDuWWCT80VfH+3bjlvsGR59ojEI2BVtTWv2WPpqA0XXmU7wGrulkzKg9oxiiqSRVUSjo1rA6ymKsFbXRI9TqK7lJLNLmvuPJqxSsNjJ1IiBQJ2JOZFZNcZLDDysEYuETrJpuHMecnq5y56XRXF8bqaiIpCwdQi2GMRR/fAcIeEeEuEmdbwW1d46OiDDe9YY3+kytYOnaunisuaZVuFeDo2VQm2Z3sGq8fIvzLj/rbMJzgxYHjRKS7i2gBrRZUgOBR/iWBPj5GL40Z7hvFo8Au6B+FYf65HkhliHlh2DrAa/IKj54xgJFMMRCQvHc1u9VnM6+rMIGy7yUvlOF5XiWClRPfxPLCyQb1fcHOt4Hu7onEfOq3OYFTy0lFpm+rBkTRVCY4l2WEkXYwWOeZsK7bMqAxHL8W1AVZs2S9eovvmdp3SPLhuhv1O1kzi9N5TgRAIYJbNlwghMy1zwlHJP++NMm9jhBu2Ran6zwg3bIvwyJ4onX2ZOyZ0Kfkfr0R5s0vy9CoP8yenf15aKsAKSsrzwZ9m0JgO91+hsa/XSI2wI789LjkXsW7+1XisHE50z9YyYUSX7OtNvUXOaBZMNgxTnXpfSBX7HYkm4SsQzC+Lf0BvaZfcWC0ocGlzZ4CHHnrostdmT4ISr3NPpLaQZGYJttZ9wWQo9FxMdB9Lx2SJ6pL/OKCzeFOEr/xO5/oZgj/e6uEfr9YQwNf/oFO7IcKKX0b4mx1R05eRv7VdZ1NA8rPrPLx7qjmXrVgl4aAFKgkDE/S+NEPDibh6qsY10wTf3anbMgVgW6ekusTwZbIq8XSsLoGpSXYYSYeDZyGsp9ciZyReTbCk3B2t1JLBm+sB5JIVVWMnuh87J9l+WnJ3XXKl306jv//ydUBNCOodnOgeCGXfi8hsjJY5F/OwxtIxUaSU/PKw5NtvRtnbCx+ZLdiyynPhwvyuKfDlpdA7KHm2U/JUu85Du3W+/ZbO3Elw22yNW2cJ3jtV4EnRQPORPVG+P+x19ZE55j0T1vkEUWk0fq43aakkVSY67tLRMBnuv0Ljw89FefmY5IMz7XUebO3QWVNj7ZzZeDoKIViZxUT3luFAKF2LhpEs8xmNqxUXce0MFhgOuju6Lne9fqbDqESx41q+mTz44INjvl7vd/IMlj1b5Iym0X9xBiuejuMhpeT5Tp2mp6Lc/lKUmSWC7bd5+MUN3jGfessLBHfM19h4vZeuz3l59iYPN8zU2HBI5wPPRJn20wh3vhJh8+Hk/Jaebtf56h90/ucyja8m6XU1EVbqSRiYwGQ0FQ1T4cZqwZUV8N1d9rJsOHhWcjAIqy3YHmck4+nYNBxgZWP2sLlbMrUIqorM21/1w10k7Dj7mSlcHWCtqBIMRGFf76Wvb2nXed80MW7DWDfT4Bfs7cX2Jd1j0RaSzJ2gl50daKwQ7OkhpZY5r53QufaZKDc9F6XAAy+v8fDCai9NVYldLgo8gptqNB59v4cjn/byx1s9fGGRxh9PSW57MUrlf0T4yAsR/u2APm7V0RundT417HX1d+82/1LlKxBML859i4+ILmnvG3+JMFsIIfhWo4eXjkrePG2fIGtbp06+BtfbOGe2qUrQG4Z3gpn/rhYTWuSMpt4v6BuC9j5TN2trXB1gLa80LP5HLhP2DUleOiZZa/EnoVxS7xcM6XCgd+L32om+IcmpAXtXEMZo8AvCOrQmodGOLsMC4f1PRwkOSZ650cNrt3i4dkbqlwlNCN41ReO7Kz3s/XgerR/38uBVGqfOw52vRJn6XxGufSbCD5qjBIIXz8O2Ya+rRn/qXleJYIVE95g1iFWWpm+fYxgKf3enjQKsDsm10wUledbYh6nQlMVE92YTWuSMJpYw79TVjVRwdYBVmmck5o08oF88IhmMwq1J9DVzKl1dXWO+HjuRnJaHFetNaWeT0RiNFRdb5sTTMcb+XsknXopw1S8jHApKNn7Qw1sf8bKm1vx8loXlgnsbPfxurZfjn/Hy2Ps8lHrh/jd05m2M0PDfQ/zvN6N8+LkIZXnpeV0lwlILNH0+FLNoGGfmdCINzcSjCe5r9PDLw5L9FvEJG4++IclvjkvW2OCheDwdKwoFcyeRcT+s/ojkUNAcB/eRzCyByfkqwBqJ66OIFVXiEquGze06S8thngOWidLlzjvvHPN1/3A7FqdVjAQc4IEVw1cgqCkxguB4Oh4OSda9EqHu5xFePy358TUe9nzMyyfmaVnp4za1WPCFxRrP3GTkbf38Qx4a/YL1e3XODMKzN3lNzREZizqf4J1gbisJA0Gjx+d4jtrxNMwUn1tgLJ8+tCua1e9NhV8fk4R1WF1j/dvZRDo2ZSHRfW+PRGJ+taUQwmiZ47D7Qjq4uooQjAP6iUM64aiR2P5Mh+RPF1v/RM0GDzzwQNzfNfgFu88460RqC0kKPTCtONcjMYfGCsGuM5LvjNLxRL/kO2/rPLZfx1cA//BujbuWaDm1pijNE9w+R3D7HI0hXXI+ApPyMz+eOp+xPNfaCw0VGf+6MQmEJLUlkDdOleV452ImKPAI/rxe45vbdR68SlJbat2Hjq0dkoWTMcUbLdNMpOPKKsHmdqNl0XjHQzq09Bj/1pm8RAjGrNirJ+yztJxpXB9JrKg0clWau+H3JyVnBpV7e4zly5fH/Z0Tn1QCQcPnKxuzN9mgwW9YNcR07D4v+db2KHOfiPBf7+j8n6s0Ap/0cs8yj6V8v/I0kZXgCqxRSRgITlxYMd65mCnuWqJRlg//d7d1b5hSSrZ16raYvYKJdWyqEpyPGp1EMkVLt2TuJDKSr1bvh/29hiGxQgVYNFYIvMKoWNrSYZSurpxinZuNVWmoEHT0kdVWKZmmzQEeWCNp9AuO9xs38L/eEWXOExHW79H5Rr1G26e8fOsKj62Tgs2gvEAwozjHAVbImnl/pXmCe+o0/nW/zukBa57nzd1w5By2yL9KhOWVAk1kNtG9udv8CsIY9X5BREJrnBZ0bsP1AVaRV7DMb7TM2dKuc0utcMwMRiaJJbq3OGgWKxCSjsi/ihFLdK/7eYS/eVvnzkUagU95+ZsmD+UWbYabC3JdSRiwsDXIPXUamoB/2mPNWaxtnTolXnj/NGvuv2QpyRPU+WD7qczt71iT50xQpyoJL8H1ARYY07JPHZYcOAtrZ6ldEuPxxx+P+7vF5eAVzqkklFLSZtGZhFSZXwZXTxWsiB7k4Ce9/OA9HqZkOGncjtT5BHtzVC3XMyjpGYR5EwT2452LmaSiUHDXYo31e3SCYeud61s7JDfMtE9Ls0R0bKrMXKJ793nJsX7zWuSMxlcgqC5RAVYMFU0AKyo1zgxCkQc+ZLP2EJlkx44dcX+X7xEsLscxrRFODUB/xBkVhDE8muB3a7007HzE0knKuSZWSXg+CYd5s2iLWYNMYDI63rmYab5Rr3EuAo/ts9YsVs+g5PenJKtt1Nw5ER1XThG09JBUx4NEia04mNXkeSxiju4KFWABhlUDwKpqkVHPHbvxyCOPjPv7Bgf1JGxLwIvIrkyko9tZ6gM9R3kjMXPViXL/cqlhdangTxYI/r5Zz0kQGo/nj0h0aa+WZono2FSlEZXwdpf5+7q5W5KnwcJy0zd9gXqfUDNYw6gAC6PyYXoxfHa+2h3JUO83TiQn9J4KDM8kOKEPoSI5lpbnrpIwEJJMzgdfQda/Oinua/BwcgD+/aB1ZrG2deg0+mFmiX0CrESo90OBJzOJ7i09RnpHpiwgwFh+bO/DkkvK2UZFFBgH29FPe/nYXLU7kqHBLwg5pPdUW0hSUQBlWbIHUFiH8gLDODcnAVbQyPsz2zHfbBaWCz42R/DwLj2l/pZmE9Ulzx6RrLHR8mCi5GmCKysyk4eViRY5o4lVKKplQhVgXcDqFzgrEjuRnDAdHAg6q4JQkRx1OWqZY+UKwtHcf4WHQAg2BXJ/vr/ZJek6b6/lwWRoqhKmt8yRUmakyfNolpSDR0CLQ/Jz00EFWIq4rF27dtzfV5dAeb4zKgnbQhMnGtuViXRU5M6q4VAwMe81K2h4ZaXgxmrB93ZFc54WsLVD4i+LzDioAAAgAElEQVSAd9vMszBRHZuqjMKL7vPm7ecj5+BsmIzPYBV4BAsnO+PBO11UgKWIy9133z3u74UQF/Kw7I7TPLBGMpGOCiPAOhSEgSwmcUd0SXtfYtYgVtHw/is0dnfDts4cB1idOjdWCzwZzCXKBInquHK48OpNExPdL1QQZngGK/YdTuv0kQoqwFLEZdWqVRO+xwmVhEO6pPPcxJVcdiURHd1OnQ8k2a0k7Owz+iAmskRoFQ2vmSZ4zxTBd3fmJtldSqPd044u+IQNc2YT1XHBZCjLMzfRvblbUpo3flNxs1jmc04BVDrY7whVWIp6Pxw4C4M27j3V0WeU6asKQveyJAc9CQMxaxAbBfZCCO6/QuN3JyWvHs9ukBXVJV96TeehXTo/eLfGbbOde/vShKCpytxE95bhBPdsdCqp9wu6B+F4f8a/ytI49whVZIUGvyAqYV9vrkeSOk72wFIkxuR8w4E6uwEWaCI7MwpmsqZWsMwH392VvQArHJV85uUo/9qq8+NrPHy93pO1784VZie6N/dIlvlM29y4qEpCAxVgKeLy1FNPTfieWMLk7jP2PZECQXve6BIlER0V2U90DwQlNSVGV4SJsJKGmhB86woPz3ZKdmbhvO+PSG57McovD0s2Xe9h3SL73raS0bGpymjWfvRc+vs4okv29WYn/wqM1YBir0p0t++Rqsg4GzZsmPA9k/IFcyZh64TGtpCkuiSz5nu5JBEdFTkIsEKSeQnOmlpNw0/ONc777+2MZvR7egclN26L8tvjkq03efjoHHvfspLRMZbobsYy4TtBGIxmvoIwhiYEdcrRXQVYivhs3LgxoffV+4W9Z7BCiZXK25VEdXQ7dT5BIJiZHnBjETMZTQSraejVBPc2aGxqk7xzNjP769SA5LqtEfb0Sl5a7eFDM+1/u0pGx5klMK0IU5YJW7qzV0EYo95n7wdvM7D/EavIOQ02L8ltC6kEd8XFSsL9WcontJPJ6FisW6hRVQjf321+LlZHn+T9T0c40Q+v3Ozl3VPdd6sSJia6t/RIphRBVVEWAyy/YG+PUZzgVtx31CpMp95n5Ap0mWiKl03aHD6DpUiMbPYk7B2UdA/aq4JwNIVewf9cpvFvB3SOmZAnFKO1V/K+LRGGdHhtrTersy5WY2WV4M3TEj1Nu4Pmbkl9lpYHY9T7BeejcCiY1a+1FCrAUqRNQ4V9W+aEwkbLjTk2nklQmMOkfEFtKezNQoDVNtxc3O7dA760VKPICz9oMWcWa0eX5H1PR5iUB6/e4k04R82pNFUJesPwTpr+bC09kmVZDlQvtFKz8epGuqgASxGXdevWJfS++WVG93c7Go5euNE5eIkwUR0VxixWNmawkvXAsqqGZfmCLy/VeHSfTs9gevvtt8d1rnsmwpxJgt/e4mVmifOCq2R1XBFLdE/D0X0gInknmL0E9xhTigRVhfZ88DYLFWAp4pKo67BXE9T57HkixW50Tm2TA9ZxAbcD2aokPBSUlOWBvyCx91tZw68v04josH5P6rNY2zp0bnw2yooqwa9We6godOb5mKyOFYWCeWWw/VTqx+S+XsNIud6f8iZSximt1FJFBViKuNxxxx0Jv7feJ9htw+7pbSFJkQemFuV6JJkjGR3dTp1P0BbKfCVhIGgsD4oEXbWtrOGUIsH/s0jjH1t0zg0lv9+eOKRz6wtRbqwWbL3Rw6R8ZwZXkJqO6Sa6xwKcWI5hNqn3C1ebjaoAS2EKDX7jyd9uFSOBoFFBmOiNTuFsYpWE+3oy+z1Oswb5fxs0esPwo9bkZrEe3Rvl07+O8un5gp9/yEOh1zn7xCxWVgnePiMZSvHa2tIjmTOJnASu9T7BO1luom4lVIClMIV6v6A/YrT/sBNtNi+VV5jL0iz1JLS7RcNoZk8SfHqe4O9264QT7Ev6vZ1RvvQ7nbvrNH7yAQ9ehxr9pktTlVGN15LiCkHzcA/CXLDMbyxP7s3wA4tVUQGWIi6vvfZawu9t8NuzkjAQko7Ov4LkdHQ7pXmCWaWZDbAiuqQ9lFxhhR00/GajhyPn4KfvjL/vpJR88/Uo97+h81fLNf7xPVpWGhBbgVR0vLJCoInUHd1bemTOrC7qfO6uJFQBliIuDz/8cMLvnVpsVIzYqZJQSklbkjc6O5KMjorMJ7ofOQcRmZwHlh00rPMLbp0leGhXNG6qQFSXfPG1KA/v1vmH92g8cJXHVcvzqehYkmc0137jdPJFBD2DkqPnsl9BGKM0TzB30kUnebehAixFXJ544omk3m+3ipETA3A+6uwKQkheR7dT5xPs7c3ccRwIGttOxuPJLhref4VG61l4qv3y/ReOSj79cpTHWyU/+YCHry3z5GCEuSVVHZuqREotc3LRImc0drsvmIkKsBRxKS4uTur9DX5hqxmstpgXkYNyYcYiWR3dztLhSsJUKuISIRACTUBtaeKfsYuG75qicd10wXd36sgR7uP9EcmtL0R56rDk59d7+PxCd956UtWxqUqwpyf5Y7K5R+IVsHBySl9rCst89m6llg7uPMoVGaHeLzgUzNyNyWwCwy0cVB9CxUjqfMa/+zI0ixUISWpKIN/jzMD+/is03uqSvHTU2H+9g5JV26K8ekKy7SYPH5mjbjvJsrJKIyrh7TPJHZMt3bC4PLfHWr3faKV2xqat1NJBHekK02jwGyXu2TBqNIO2kKSq0MgTUChiLLnQkzAz2w8EnWXRMJoPzRRcVWnMYp3sl1y3NcLeXsmv1ni4fqa65aTCMj8UepJPdG/uzn6LnNHU27QAygzU0a6Iy7333pvU+5f6BAJotonhqBsqCCF5Hd1OaZ5gdgYrCQOh5HsQ2klDIQT3X6Hx8nHJVU9FODkAv73Zy7umqNtNqjrmaYIrK5IzHJVSGhWEOUpwj7FgMuRruNJwVB3xirjU1tYm9f5ir2DBZPtUEralcKOzI8nqqMhsJWEqM1h20/AjswVLy40b62u3eHM+i2IV0tEx2UT3Y/3QGybn+z5PEywpt8+Dt5moAEsRl3vuuSfpz9ipYiQQdMcMVio6up1MBVhnw5Izg8kXVthNQ00IXr3Fy+7bvY4vIkmGdHRsqjJyXLsTzGWKXYdzPYMFRpBnl/uCmagAS2EqsUrCkRVEViQclRw5l5wXkcI91PkE7X3QZ3LBRqywwuneawD+QqHyG01kZZWxL9/sSuyYbOmRlHhhlgWOtXqf0ZPQ6veFeEgpCYYlB89Kdp1J3I/Mm8ExKVxIvU9wZtDwmJpu4cry9j4jIV9VECrGIuZAva9X0lRlXpAQcIk1iMJ85k+Gyfmw/ZRkVfXE748luFvBJb/eLwgNQUefNQI+MIKm7kE4OQCnBiQnB+DkiH9PDXDJa+ejwx9sVwGWwgT279/P4sWLk/pMQ8XFipHpxbk/seNxwQPLBTNYqejodhaXG//u6YGmKvO2GwhKJuVBRUFyn1MaOoN0dNSEYEWl4I0kZrCurLDG9W1kJeGsLFxzz0ckr52UnOi/NGgaGUydGjA6KoykwANTi2BqkWBqEdT74UNFGlNGvHbmgIePfyexcagASxGX++67jy1btiT1mTmToMRrJLon8pSVKwJBiUdATRJmj3YlFR3dTkmeYM4k8ysJA8OtmZJtD6M0dAbp6rhyiuAnrYaJ63jHUFSX7O2Bz823RoBVXWLMvjX3SG6eldnvCkclq5+L8vJx49ydlHcxaJpSZJjhGj9ffC0WPE3Km/jc3HEi8X2qAixFXNavX5/0ZzQhbJHQ2BYynLS9mjUuQJkkFR0VmUl0DwRlSsuDSkNnkK6OTZWC7+6Eo+egepyHw0NBY0krly1yRiKEMBzdM3xfkFLypdei/O6k5LmbPFwzXVDkzd0+UEnuirikWlJc77O+VYNbPLDAfiX+ViEjAVZIJtWDMIbS0Bmkq2MsH3AiP6xYa5pcNXkei2xUmD+8S+fHBySPX+Phxhotp8EVqABLkQEa/IK9PRDRrRtktYXcUcmlSJ06n6CjD0Jhc47jqC45rI47RRrMLDGKhybyw2rpNrpUTLVQHmy9H/b3Gkt4meDnAZ1vvaHz7Ss1PrvAGqGNNUahcBT1fkFYhwNncz2S+LS5aAZLkRqxSsK9JvUkPHLOSKp1Q2GFIjMIIWiqmtjRvblHWmr2CowK84jMzH1h+ymdz/0myqfmCR68yjphjXVGorAcDz30UEqfs3rvqbNhozzXLQFWqjq6ncXlIDCvJ2E6Fg1KQ2dgho4rqwRvdkn0cTylWrqlZfKvYizL0H2ho0+y9oUoV1YIfnKNJ+kCkkyiAixFXPr7+1P6XEWhYEaxdfOw2kLGv25okwOp6+h2ir2CuWWw16Q8rEDQCNhmpVC5qjR0Bmbo2FQlOBuGg3FmggYikoPB3LfIGY2vQDCz5GJ+mBkEw5Kbn49Q5IGnVnkozHHO1WhUgKWIy4MPPpjyZxssXEkYCBrjcssMVjo6up2l5eYlugdCkppSyPckf9wpDZ2BGTquqBw/0X1/L+jSKDayGvU+QYtJ94WILvnUr6O0h2DrTV6mFFnveq4CLEVGqB9umWNF2kJGC4mqwlyPRGF1zKwkPJRCk2eFYjT+QsH8svgBVuzBts5iOVhgbiXhN/6o88IRyc8/5GGpBf9WUAGWIkM0+I1ebmdNqsAyk0DIMES10lq9wprU+QSd54yliHQJqApChUk0VYm4lYQtPZLZpTAp33rXt2U+wWETKnPX74nywz06j7xX44Zq64Yx1h2ZIud0dXWl/NlYgqVZ08Fm0hZKzezRrqSjo9u5UElowixWqiajoDR0Cmbp2FQlePuMZGgMK5xYD0IrcuG+kMb5tK1D52t/0PlGvcYXl3jMGlpGUAGWIi533nlnyp9dXA5eYc1KwkDQXRYN6ejodhaXgybSryQ8G5acGUzdokFp6AzM0nFllWAwCs3dl/+upcd6FYQxlpSDR4w97kTYfUbyyV9HublW8PBK64cv1h+hImc88MADKX+2wCNYVA67UzyRMoUuJYf73LVUk46ObqfIK5hrQk/CdCtXlYbOwCwdr6wUeAS8cVq/5PXeQcmRc9ZycB9JoVewYHJqM1gn+o2Kwfll8NPrPHhs0OZMBViKuCxfvjytz1uxkvB4PwxG3VNBCOnr6HbMSHSPVa6mOoOlNHQGZulY7BXU+S5PdI8FLladwQJS6knYHzG8rqISnl7lpTTPun/fSFSApcgY9X5Bc49EjmOIl23aQund6BTuo84n0nZzD4Qkk/KgUlWuKkxi5RiJ7i3dEq+ARZNzNKgEiFUSJnpf0KXkT34TZU+P5OkbvVSX2ufarQIsRcZo8BuGeJ3ncj2SiwSCxr+zXbREqEiPpT7BkXPpVcQGgsaytKpcVZhFU5XGnh44N3TxuGzugUXlqXmtZYt6n+DMIJwYSOz9f/mmzi/aJD+7zsPySuv+XWOhAixFXB5//PG0Ph+bpt59xlozWFOLoMQmU8xmkK6ObseMSsJAmpWrSkNnYKaOK6cIdAlvj7i+tnRbrwfhaJKpMP9Jq853d+p8/10at862X7hivxErssaOHTvS+nxNCUzON7c1QroEXNjkOV0d3c7iyelXEgbSNBlVGjoDM3Ws80GhB7afMq6vUkpLNnkezdwyKPJMXGH+8jGdu16NctdijW/U2zNUseeoFVnhkUceSevzQgjqfdZydG9zodljujq6nUKvYF4alYRRfbhyNY3el0pDZ2Cmjnma4MoKwRtdxnF5vB96Bq2d4A6gCUGdT4z74N3aK7n9pSjXzhCsf69m26V1FWApMkpDhbUqCd04g6VIn3QqCY/2w5CuCisU5rNyirhQSRi7zlrVZHQk9f74XlhnzkvWPB9hWhFsut5Dng3sGOKhAixFRqn3QWsvDEZzH2Sdj0iOncNVLu4Kc0gnwDqUpkWDQhGPpirBoaARlLT0SIq9Rhswq1PvN86n6Cgn+sGo5CMvRgmGYeuNXsoL7H3OqABLkVEaKgQRaXR4zzXtfSCxxwVIYS3qfIJj/YaRY7IEgiCAWeq4U5hM03BV3ZunJc3dkjqfQLPBclq9X3A+avTnjCGl5K5Xo2w/LXlqlYc5DngQVgGWIi5r165NexuxhEsrLBO61QPLDB3dzoVKwhT8sAIhSXWJ0d0gVZSGzsBsHedPhvJ82H5a0tJjrBjYgbHuC3+7U+c/Dkp+co2Hq6c6IzRxxl+hyAh333132tsoyxfMKsUSie6BkGHCV12S65FkFzN0dDsL06gkTNeiAZSGTsFsHTUhWFEl+OMpyZ4e6zZ5Hs3UIsN0NxZgPXlI5y/f1HnwKo075jsnLHHOX6IwnVWrVpmyHau0zGkLGcs0duhhZSZm6ehmCr2C+WWpVRLGTEbTQWnoDDKh48oqwUtHJeej1q8gjBGrMG/ukfzxpM6fvBLls/MF377SWSGJs/4ahSWp91vDqqFNVRAq0iDVRPdAKD0PLIViPJqqBOHhns9W98AaSb1f8IeTkltfjLKiUvCjazy2tWOIhwqwFBmnwW8kCJ85n9sgK12zR4W7SSXACoYlXedhngMSdhXWpKnKOLYqC42lN7tQP3xfKPXCL2/wpJWjaFVUgKWIy1NPPWXKdmLT1rleJmwLubOC0Cwd3U6dTwybOSZ+HLcNV0mlYzIKSkOnkAkdZ5YIZhQbs1d2mgG6boagqUqw9SYvVUX2GXcyqABLEZcNGzaYsp2FkyFfy22A1TMo6Q27r4IQzNPR7cQqCZOZxQqY5IGlNHQGmdLx3gaNuxbb63Y+r0yw/TYvi8ude0325noACuuyceNGU7bj1QRLfbmtJIzNJMxJcybBjpilo9tZOBk8wgiw3jctsc8EQpLSPGP5Jh2Uhs4gUzp+vd6Tke0q0sNeIa/CtjT4Bc1pNMtNl9hMgkpyV6RKgceoJNybxHEcGO59aaelG4VCYQ4qwFJkhXq/oKVbosvczGK1hSST8qCiICdfr3AIySa6q8IKhcK9qABLkRUa/IJzkYtLddkmMJzgrmYSFOmQbIB1KJi+yahCobAnKsBSxGXdunWmbStWSbj7TO5msNw6k2Cmjm6nzic4MQDdCViORHXJ4b70TUZBaegUlI7uQgVYiriY6To8LdYaIQWjRjMIBN1rMqpcwM0jmUrCo/0wpGPKDJbS0BkoHd2FCrAUcbnjjjtM21asNUIuKgmjuqS9L30vIrtipo5uZ+Fk8IrEAiyzLBpAaegUlI7uQgVYiqzRUJGbnoTH+iGsqwpCRfrkewQLJifW9DkQAgHMKs34sBQKhQVRAZYia9T7BAfPQn8ku0FWW8i8mQSFItFE90BQMrPEaBStUCjchwqwFHF57bXXTN1eQwVIYG+W87ACw5WLs13YJgfM19HtLPUJ9vYmEGCFpGk9CJWGzkDp6C5UgKWIy8MPP2zq9up8AgHs7jZ1sxPSFpJML4Yil84kmK2j26krF5wcmLh5ecxk1AyUhs5A6eguVICliMsTTzxh6vaKvYYTdrbzsNxcQQjm6+h2Eq0kNNNkVGnoDJSO7kIFWIq4FBcXm77Nen/2E93bTJxJsCOZ0NHNLEigkjAUlpw+b45FAygNnYLS0V2oAEuRVRr82bdqCITcPYOlMJd8j2DhBJWEsY4Fbg7sFQq3owIsRVap9wtOn4eT/dkJsgYikuP95s0kKBQwcSVhIFa5qo47hcK1qABLEZd7773X9G02xFrmZGkW6/DwTMIcF88kZEJHtzNRgHUoKCnxQlWhOd+nNHQGSkd3oQIsRVxqa2tN3+bcMpheDP+6Xzd922OhPLAyo6PbqfMZM7GnB8YOsgIh41g3q7m40tAZKB3dhQqwFHG55557TN+mJgQPr/SwqU3yq6OZD7ICIUmeBjNcnFuaCR3dztLhSsJ4flhmVhCC0tApKB3dhQqwFFnnM/MF75squOf3UYb0zC4VtoVgdil4NPfOYCnMZ8FkyNPiVxIGQuYGWAqFwn6oAEuRdYQQrH+vh9az8MOWzM5itakKQkUGyNPiVxJGdcnhkHubiysUCgMVYCnisn///oxtu7FC8OUlGg/s0DmewYrCQFC6vpIrkzq6mXiJ7rHm4mbOYCkNnYHS0V2oAEsRl/vuuy+j2/8/KzQKPPDN16MZ2b6UkraQuysIIfM6upV4AVbMosGsPoSgNHQKSkd3oQIsRVzWr1+f0e37CgTfa/Lwn+9IXjth/lJh9yAEh9xdQQiZ19Gt1PkEXefh1KhKwkAQBDCr1LzvUho6A6Wju1ABliIu2SgpXrdI0FQluPt3USImJ7zHLBrcnoOlSsMzQ7yehIGQZGYJFJrYXFxp6AyUju5CBViKnKIJwSPv1djdDY/t+//bu/+YJu88DuDvp61DneAvMIqIVRbAOVCqzultB8ky9ZzpEQlziTJ1kmGGU2PiXC67qfGCMf7Y4VxuW+apQ8MyzpzGnURzntnYzk3WgiynomeLhWlgeChVfrff+6NSLaWlYMvTp32/EkP68NR+nrz5ysfn+T7fx79nsUzNjq+cbEyB8ExU73cS+nuJBiJSJjZYJLs5MSqsSZLw/k92jws3DoTZKjDyKcelSCJ/06gkJI0ELve4k9DEOwiJCGywyItdu3YN2mcVzFEDAP5Q7r8J7yYrH7YLDG6O4aa3ie6BWAOLGYYG5hhe2GCRRy0tLYP2WTHDJPxptgoHqwUuNvjnUiHXwHIYzBzDTXeDJYSjybJ2CDS0+v8hz8wwNDDH8MIGizzavn37oH7e2mkqzBgLrPu3HXbx5JcKTWywAAx+juFk+mgJd9qBhlbHa/PDh4v7+8wpMwwNzDG89NlgXb9+HfPnz0dSUhKef/55XL582eO+X3/9NaZNm4bExERkZWXBarU6v/fjjz9ixowZSEpKwssvv4xbt245v6fVapGcnIy0tDSkpaWhpKTkCQ+LlEitknBgvhrlvwr8tfrJGiybXeAm58JQgPW8k9DEh4sT0UN9Nlh5eXlYu3YtqqursWXLFqxatarX/e7fv4/c3FycPHkS165dQ2xsLHbs2AEAsNvtWL58Ofbv34/q6mosXrwYGzdudL5XkiR89dVXqKioQEVFBbKzs/1zdKQ4vxmvQs4zEt67aMP/2gbeZNU9ALoEl2igwEqIAp567E5CU7PAcA0wbpjMhRGR7Lw2WA0NDTAYDFixYgUAYOnSpaitrYXJZHLbt7S0FDqdDomJiQCAt99+G8XFxQAAg8GAIUOGID09HQDw1ltv4dSpU+jo6HC+X/jhkhD5V2Njoyyfu2uuGh124APDwOdimXkmwUmuHMOBRiUhedSjZxJ231ghSf79uWOGoYE5hhevDVZtbS0mTJgAlcqxmyRJiI+Ph8VicdvXYrG4LKI2efJk3L59G3a7HRaLBZMnT3Z+LzIyElFRUS6XCXNycpCamorc3Fz+EAaJN998U5bPnTBcwjadCn+5YkflnYE13qaHV6f9uZq2UsmVY7h4/E5CkzUwz75khqGBOYYXv01yf5L/sX377beoqqqC0WhEdHQ0Vq5c6a+y6Als27ZNts9+5zkVkkYC6763DejspjkAq2krlZw5hoNnR0n4z13HnYSmZoGEAJw1ZYahgTmGF7cGq6ioyDnZ/Ny5c7h9+zZsNsfaREIItzNV3eLj43Hz5k3n65qaGufZr57fs1qtuHfvHmJjYwEAkyZNAgBoNBps2LABZWVlXotevHgx9Hq9y5958+bhxIkTLvudPXsWer3e7f35+fk4ePCgyzaj0Qi9Xu929mzr1q1ua5dYLBbo9Xq3J6N/9NFH2Lx5s8u2lpYW6PV6fPfddy7bi4uLsXr1arfali1bFjTHodPpZDuOnysrMOr0VnxfL3Dsv48aLF+Pw9QsEGG9HVJ5DPQ4dDpdSBwHEJx5TB8toakdWPjaKpia7S43VvjrOCoqKgJ+HEBo5BHMx6HT6ULiOIDQyKOv41iyZIlLn/Hqq6/ilVdecavBI9GHjIwMcfjwYSGEECUlJWLOnDm97tfc3CzGjRsnrl69KoQQIj8/X2zevFkIIYTNZhMJCQni/PnzQgghdu/eLbKzs4UQQjx48EA0NTU5/569e/eK9PT0Xj/DYDAIAMJgMPRVNoWI1/7ZKcYXdYh77fZ+vW/eiU7xxvnOAFVF9Eh1k13gsw5xuNom8FmH+MdNm9wlEVGA9KcP0fTVgH366adYtWoVCgoKMHLkSBw6dMilG4yNjUVeXh4iIyPx+eefIzMzE11dXUhJScGRI0cAACqVCkePHkVeXh7a2towceJEFBUVAQDq6+uRlZUFm81xKSghIQFffPGF7x0ihbQ9c9VILunCdqMde19Q+/w+k1VgQRyXeaPAS4gCItTAKYvjpoxAzMEiIuWRhFDO7XtGoxGzZs2CwWBwuXxFgXHw4EGsWbNG7jKws9KGP/5kx6WlGkwf0/cvr5YugacPdeFIuhpvJLLJCpYcQ9nM4524YQXudwKtqzV+n/vHDEMDc1S+/vQh/O1DHhmNRrlLAABsSlFhSiSw/oJvE97NzY6vU/gcQgDBk2Momz5awv1OBOzGCmYYGphjeGGDRR59/PHHcpcAAIhQS9g/X41/3RL4m9mHBqt7DSxeqgEQPDmGsu4V3QO17hozDA3MMbywwSJF+N0kFfSTJWz6wYYHnd6bLJNVIEINTBg+SMVR2HvUYMlcCBEFDTZYpBgfvqDGr21AQaX3Fd7NVkA7AlD5eTVtIk+e7W6weNaUiB5ig0WKMTVKwnszVNhTZcf1e57PYpmtgs8gpEE1NRKYEyPht+P5c0dEDmywyKPeFnWT25YZKsQOBzZ4mfAeqMeVKFUw5hhq1CoJFzM1yIgNzD+pzDA0MMfwwgaLPFq3bp3cJbgZppHw4Tw1SmsFTlncGywhBMxW3kH4uGDMkfqHGYYG5hhe2GCRRwsWLJC7hF79frKEhXESNl6wobXLtclqbHOsRRSou7mUKFhzJN8xw9DAHAqPOfoAAAS6SURBVMMLGyxSHEmSUDhPjboHwO4q1wnv3Us0cA4WERHJiQ0WKVLSKAmbUlTYWWlHjfXRWSyT1fH18QfuEhERDTY2WORRz6eZB5v301QYEwFs+sHm3Ga2CoyOAEY+xTNY3YI9R+obMwwNzDG8sMEij3bt2iV3CV6NGCJh7wtq/L1G4Eyt41KhqVlw/lUPwZ4j9Y0ZhgbmGF7YYJFHMTExcpfQp2VTJWRMkLD+gg3tNt5B2Bsl5EjeMcPQwBzDCxssUjRJkvDRfDVuNAN//tnuWAOLZ7CIiEhmGrkLIHpSz42R8M50FXZU2NFm4xksIiKSH89gUUjYNkuFp4cANsElGoiISH6KOoPV2toKALhy5YrMlYSHixcvwmg0yl2Gz9aNsOODq3Z01ahh/JVNVjel5UjumGFoYI7K191/dPcj3kjC0wPdgtCxY8ewYsUKucsgIiKiMHb06FEsX77c6z6KarAaGxtx5swZaLVaDBs2TO5yiIiIKIy0traipqYGCxcuRHR0tNd9FdVgERERESkBJ7kTERER+RkbLCIiIiI/Y4NFbrRaLZKTk5GWloa0tDSUlJTIXRL5YP369ZgyZQpUKhWqqqqc2xsaGrBo0SIkJiYiJSUFZWVlMlZJ3jye4aVLl5zbMzIyMHXqVOeYLCwslLFK6kt7ezsyMzORlJSEmTNnYsGCBbhx4wYAjkcl8ZajT2NSEPWg1WrFpUuX5C6D+qmsrEzU1dW55bd69Wqxfft2IYQQ5eXlIi4uTnR2dspVJnnhKcOMjAxx8uRJGSuj/mhraxOlpaXO1wcOHBAZGRlCCI5HJfGWoy9jkmewqFeC9z4ozosvvoiJEye6bS8pKcHatWsBALNnz0ZsbCy++eabwS6PfOApQ4BjUkkiIiKwaNEi5+u5c+eipqYGAMejknjLEeh7TLLBol7l5OQgNTUVubm5aGxslLscGqA7d+6gs7MT48aNc27TarWwWCwyVkUD8e677yI1NRWvv/46zGaz3OVQPxQWFiIzM5PjUeG6c+zW15hkg0VuysrKUFVVBaPRiOjoaKxcuVLuksjPJIkr3StJUVERqqurUVVVhZdeeglLliyRuyTyUUFBAUwmE3bu3OlxH47H4NczR1/GJBsschMXFwcA0Gg02LBhAydhKtjYsWOh0WhQX1/v3FZTU4P4+HgZq6L+6h6TAJCfnw+TyYSmpiYZKyJf7NmzBydOnEBpaSmGDh3K8ahQPXMEfBuTbLDIRUtLC+7evet8XVxcDJ1OJ2NFNBCPzw3Izs7GJ598AgAoLy/HL7/8gvT0dLlKIx91Z2iz2Vx+IR8/fhzjx4/H6NGj5SqNfLBv3z58+eWXOHv2LKKiopzbOR6VpbccfR2TXMmdXJjNZmRlZcFms0EIgYSEBBQWFvJ/WAqQl5eH06dPo76+HmPGjEFUVBSuXbuGhoYG5OTkwGw2IyIiAgcOHOA/6EGqtwwrKyuRnp6O9vZ2qFQqxMTEYN++fUhJSZG7XPKgrq4O8fHxSEhIwIgRIwAAQ4cOxYULFzgeFcRTjufOnfNpTLLBIiIiIvIzXiIkIiIi8jM2WERERER+xgaLiIiIyM/YYBERERH5GRssIiIiIj/7P34PIao1ddmDAAAAAElFTkSuQmCC\" />"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# plot mean and std\n",
    "plot([ [mean(c2[:,i]) for i in 1:25] [std(c2[:,i]) for i in 1:25] ], legend=true, label=[\"mean\" \"std\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<img src=\"data:image/png;base64,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\" />"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# plot them as 5x5 arrays\n",
    "plot([heatmap(c2[:,:,1,i], aspect_ratio=:equal, legend=false, clims=(-0.4,0.4)) for i=1:25]...)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[1m\u001b[34mINFO: Start training on MXNet.mx.Context[CPU0]\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: Initializing parameters...\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: Creating KVStore...\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: TempSpace: Total 69 MB allocated on CPU0\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: Start training...\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: == Epoch 001/018 ==========\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Training summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.1098\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:               time = 38.4253 seconds\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Validation summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.1056\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: == Epoch 002/018 ==========\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Training summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.1116\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:               time = 35.0255 seconds\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Validation summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.1056\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: == Epoch 003/018 ==========\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Training summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.1028\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:               time = 34.9135 seconds\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Validation summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.1056\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: == Epoch 004/018 ==========\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Training summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.1425\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:               time = 35.1945 seconds\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Validation summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.3078\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: == Epoch 005/018 ==========\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Training summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.5372\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:               time = 35.8175 seconds\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Validation summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.6329\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: == Epoch 006/018 ==========\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Training summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.7383\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:               time = 35.7084 seconds\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Validation summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.6476\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: == Epoch 007/018 ==========\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Training summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.8100\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:               time = 35.9042 seconds\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Validation summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9003\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: == Epoch 008/018 ==========\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Training summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9109\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:               time = 42.1679 seconds\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Validation summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9330\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: == Epoch 009/018 ==========\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Training summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9375\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:               time = 37.0213 seconds\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Validation summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9453\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: == Epoch 010/018 ==========\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Training summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9493\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:               time = 38.6728 seconds\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Validation summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9509\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: == Epoch 011/018 ==========\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Training summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9572\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:               time = 37.6517 seconds\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Validation summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9594\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: == Epoch 012/018 ==========\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Training summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9630\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:               time = 36.5996 seconds\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Validation summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9644\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: == Epoch 013/018 ==========\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Training summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9679\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:               time = 36.9148 seconds\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Validation summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9662\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: == Epoch 014/018 ==========\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Training summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9711\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:               time = 36.5323 seconds\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Validation summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9706\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: == Epoch 015/018 ==========\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Training summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9740\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:               time = 40.0272 seconds\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Validation summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9712\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: == Epoch 016/018 ==========\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Training summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9766\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:               time = 37.4118 seconds\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Validation summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9756\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: == Epoch 017/018 ==========\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Training summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9783\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:               time = 39.4677 seconds\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Validation summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9750\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: == Epoch 018/018 ==========\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Training summary\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:           accuracy = 0.9795\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO:               time = 38.4914 seconds\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: ## Validation summary\n",
      "\u001b[0m"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "756.474779 seconds (8.40 M allocations: 4.244 GB, 0.90% gc time)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[1m\u001b[34mINFO:           accuracy = 0.9759\n",
      "\u001b[0m\u001b[1m\u001b[34mINFO: Finish training on MXNet.mx.Context[CPU0]\n",
      "\u001b[0m"
     ]
    }
   ],
   "source": [
    "# run 18 more epochs\n",
    "@time mx.fit(model, optimizer, train_provider, eval_data=eval_provider, n_epoch=18)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "# let's reexamine the weights conv1\n",
    "c119 = copy(model.arg_params[:conv1_weight]);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "12×3 Array{Any,2}:\n",
       " (0.140932,0.571721)     0.441203   0.102162 \n",
       " (-0.452934,0.326453)   -0.159804   0.218757 \n",
       " (0.0845838,0.539062)    0.391558   0.120643 \n",
       " (0.179966,0.395499)     0.325503   0.0561188\n",
       " (-0.172542,0.430873)    0.126497   0.181701 \n",
       " (-0.381267,-0.202552)  -0.330923   0.0479099\n",
       " (-0.43614,0.343286)    -0.116384   0.223075 \n",
       " (-0.432596,0.344055)    0.0364184  0.213315 \n",
       " (-0.22719,0.486746)     0.216132   0.194413 \n",
       " (0.221292,0.479855)     0.394016   0.0648994\n",
       " (-0.138333,0.266782)    0.0448276  0.138303 \n",
       " (0.214362,0.443765)     0.359387   0.0599086"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# get the extrema mean and standard deviation for each of the groups\n",
    "[ [extrema(c119[:,:,1,i]) for i in 1:12] [mean(c119[:,:,1,i]) for i in 1:12] [std(c119[:,:,1,i]) for i in 1:12] ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1×3 Array{Any,2}:\n",
       " (-0.452934,0.571721)  0.144036  0.283974"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# let's also get the extrema mean and standard deviation for all\n",
    "[ extrema(c119[:,:,1,:]) mean(c119[:,:,1,:]) std(c119[:,:,1,:]) ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<img src=\"data:image/png;base64,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\" />"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# plot mean and std\n",
    "plot([ [mean(c1[:,i]) for i in 1:12] [std(c1[:,i]) for i in 1:12]  [mean(c119[:,i]) for i in 1:12] [std(c119[:,i]) for i in 1:12] ],\n",
    "legend=true, label=[\"mean1\" \"std1\" \"mean19\" \"std19\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<img src=\"data:image/png;base64,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\" />"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# plot them as 5x5 arrays\n",
    "plot([heatmap(c119[:,:,1,i], aspect_ratio=:equal, legend=false, clims=(-0.8,0.8)) for i=1:12]...)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "# let's examine the weights conv2 as well\n",
    "c219 = copy(model.arg_params[:conv2_weight]);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "25×3 Array{Any,2}:\n",
       " (-0.222406,0.0387741)   -0.0753508   0.0718755\n",
       " (-0.115673,0.124138)     0.01178     0.0593914\n",
       " (-0.391768,-0.0860749)  -0.203252    0.0826154\n",
       " (-0.117941,0.127839)    -0.00398249  0.0582859\n",
       " (-0.166522,0.0927904)   -0.0315113   0.0712754\n",
       " (-0.170387,-0.0130706)  -0.0784524   0.0475003\n",
       " (0.0374652,0.234771)     0.166074    0.0529682\n",
       " (-0.169853,0.0991063)    0.0113204   0.071699 \n",
       " (-0.13934,0.0476891)    -0.0339089   0.051801 \n",
       " (-0.201406,0.100754)    -0.00687966  0.08301  \n",
       " (0.130506,0.301634)      0.237295    0.049125 \n",
       " (-0.136162,0.114778)    -0.0263614   0.0747324\n",
       " (-0.123029,0.115635)     0.0170031   0.0615863\n",
       " (-0.0662929,0.198753)    0.0684825   0.0657384\n",
       " (-0.0730168,0.108024)    0.00616589  0.0573075\n",
       " (-0.128949,0.0872836)   -0.0216839   0.067694 \n",
       " (0.0599301,0.30402)      0.184179    0.0702485\n",
       " (-0.111923,0.159116)     0.0582421   0.058886 \n",
       " (-0.0914282,0.06807)     0.00311806  0.0496163\n",
       " (-0.149503,-0.0118382)  -0.0748741   0.0360288\n",
       " (-0.17349,0.112259)     -0.0282332   0.0718451\n",
       " (-0.0418719,0.131363)    0.053017    0.0506313\n",
       " (-0.00146018,0.234283)   0.0980172   0.0674581\n",
       " (-0.120674,0.11386)      0.0335362   0.0544203\n",
       " (-0.449376,-0.229012)   -0.31009     0.0567545"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# let's get the extrema mean and standard deviation for each of the 25 groups\n",
    "[ [extrema(c219[:,:,1,i]) for i in 1:25] [mean(c219[:,:,1,i]) for i in 1:25] [std(c219[:,:,1,i]) for i in 1:25] ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1×3 Array{Any,2}:\n",
       " (-0.449376,0.30402)  0.00214602  0.12555"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# get total extrema, mean and standard deviation\n",
    "[ extrema(c219[:,:,1,:]) mean(c219[:,:,1,:]) std(c219[:,:,1,:]) ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<img src=\"data:image/png;base64,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\" />"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# plot mean and std\n",
    "plot([ [mean(c2[:,i]) for i in 1:25] [std(c2[:,i]) for i in 1:25] [mean(c219[:,i]) for i in 1:25] [std(c219[:,i]) for i in 1:25] ], \n",
    "legend=true, label=[\"mean\" \"std\" \"mean19\" \"std19\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<img src=\"data:image/png;base64,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\" />"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# plot them as 5x5 arrays\n",
    "plot([heatmap(c219[:,:,1,i], aspect_ratio=:equal, legend=false, clims=(-0.4,0.4)) for i=1:25]...)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<img src=\"data:image/png;base64,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\" />"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# plot them as 5x5 arrays without scaling to see relationships within feature maps\n",
    "plot([heatmap(c219[:,:,1,i], aspect_ratio=:equal, legend=false) for i=1:25]...)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Run on test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  "
     ]
    }
   ],
   "source": [
    "@time test = readtable(\"data/test.csv\");"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "# use same transform as used on training set\n",
    "test_X = transpose(Array(test))\n",
    "test_X = (test_X./80).-0.4176\n",
    "test_X = reshape(test_X, 28,28,1,size(test_X)[2]);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MXNet.mx.ArrayDataProvider(Array{Float32,N}[\n",
       "Float32[-0.4176 -0.4176 … -0.4176 -0.4176; -0.4176 -0.4176 … -0.4176 -0.4176; … ; -0.4176 -0.4176 … -0.4176 -0.4176; -0.4176 -0.4176 … -0.4176 -0.4176]],Symbol[:data],Array{Float32,N}[],Symbol[],1000,28000,false,0.0f0,0.0f0,MXNet.mx.NDArray[mx.NDArray{Float32}(28,28,1,1000)],MXNet.mx.NDArray[])"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_provider = mx.ArrayDataProvider(test_X, batch_size=batch_size, shuffle=false)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[1m\u001b[34mINFO: TempSpace: Total 33 MB allocated on CPU0\n",
      "\u001b[0m"
     ]
    }
   ],
   "source": [
    "tpreds = mx.predict(model, test_provider)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "# create submission\n",
    "open(\"LeNetsubmission.csv\", \"w\") do f\n",
    "    write(f, \"ImageId,Label\\n\")\n",
    "    for i = 1:size(tpreds)[2]\n",
    "        write(f, string(i),\",\",string(indmax(tpreds[:,i])-1),\"\\n\")\n",
    "    end\n",
    "end"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "When I submitted the resulting file to https://kaggle.com/c/digit-recognizer/submit, I got a score of 0.97257 on the test set.  Not as good as the 1.1% error LeCun et al. got with their smaller network in the 1998 paper, but done with a lot less manual tuning.  Proper adjustments are still important, as this network is susceptible to overfitting."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "[<img style=\"float: left;\" src=\"https://i.creativecommons.org/l/by-sa/4.0/88x31.png\">](http://creativecommons.org/licenses/by-sa/4.0/)  \n",
    "\n",
    "Licensed under a [Creative Commons Attribution-ShareAlike 4.0 International License](http://creativecommons.org/licenses/by-sa/4.0/)."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Julia 1.1.1",
   "language": "julia",
   "name": "julia-1.1"
  },
  "language_info": {
   "file_extension": ".jl",
   "mimetype": "application/julia",
   "name": "julia",
   "version": "1.1.1"
  },
  "toc": {
   "nav_menu": {
    "height": "150px",
    "width": "252px"
   },
   "navigate_menu": true,
   "number_sections": true,
   "sideBar": true,
   "threshold": 4,
   "toc_cell": false,
   "toc_section_display": "block",
   "toc_window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}