pretrained-resnet.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Pretrained ResNet model visualization"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this notebook we will show how to load a pretrained ResNet-18 model, to visualize it's structure, weights and\n",
    "propagate a sample image to show it's contents\n",
    "\n",
    "The notebook doesn't need CUDA support and needs the following rocks:\n",
    "```\n",
    "luarocks install optnet\n",
    "```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "nn = require 'nn'\n",
    "graphgen = require 'optnet.graphgen'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "-- download ResNet-18 model (47MB)\n",
    "local name = 'resnet-18-cleared.t7'\n",
    "local url = 'https://www.dropbox.com/s/76m5w4dmutgmou3/'..name\n",
    "\n",
    "if not paths.filep(name) then os.execute('wget '..url) end\n",
    "\n",
    "net = torch.load(name)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Show text representation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "nn.Sequential {\n",
       "  [input -> (1) -> (2) -> (3) -> (4) -> (5) -> (6) -> (7) -> (8) -> (9) -> (10) -> (11) -> output]\n",
       "  (1): nn.SpatialConvolution(3 -> 64, 7x7, 2,2, 3,3) without bias\n",
       "  (2): nn.SpatialBatchNormalization\n",
       "  (3): nn.ReLU\n",
       "  (4): nn.SpatialMaxPooling(3x3, 2,2, 1,1)\n",
       "  (5): nn.Sequential {\n",
       "    [input -> (1) -> (2) -> output]\n",
       "    (1): nn.Sequential {\n",
       "      [input -> (1) -> (2) -> (3) -> output]\n",
       "      (1): nn.ConcatTable {\n",
       "        input\n",
       "          |`-> (1): nn.Sequential {\n",
       "          |      [input -> (1) -> (2) -> (3) -> (4) -> (5) -> output]\n",
       "          |      (1): nn.SpatialConvolution(64 -> 64, 3x3, 1,1, 1,1) without bias\n",
       "          |      (2): nn.SpatialBatchNormalization\n",
       "          |      (3): nn.ReLU\n",
       "          |      (4): nn.SpatialConvolution(64 -> 64, 3x3, 1,1, 1,1) without bias\n",
       "          |      (5): nn.SpatialBatchNormalization\n",
       "          |    }\n",
       "          |`-> (2): nn.Identity\n",
       "           ... -> output\n",
       "      }\n",
       "      (2): nn.CAddTable\n",
       "      (3): nn.ReLU\n",
       "    }\n",
       "    (2): nn.Sequential {\n",
       "      [input -> (1) -> (2) -> (3) -> output]\n",
       "      (1): nn.ConcatTable {\n",
       "        input\n",
       "          |`-> (1): nn.Sequential {\n",
       "          |      [input -> (1) -> (2) -> (3) -> (4) -> (5) -> output]\n",
       "          |      (1): nn.SpatialConvolution(64 -> 64, 3x3, 1,1, 1,1) without bias\n",
       "          |      (2): nn.SpatialBatchNormalization\n",
       "          |      (3): nn.ReLU\n",
       "          |      (4): nn.SpatialConvolution(64 -> 64, 3x3, 1,1, 1,1) without bias\n",
       "          |      (5): nn.SpatialBatchNormalization\n",
       "          |    }\n",
       "          |`-> (2): nn.Identity\n",
       "           ... -> output\n",
       "      }\n",
       "      (2): nn.CAddTable\n",
       "      (3): nn.ReLU\n",
       "    }\n",
       "  }\n",
       "  (6): nn.Sequential {\n",
       "    [input -> (1) -> (2) -> output]\n",
       "    (1): nn.Sequential {\n",
       "      [input -> (1) -> (2) -> (3) -> output]\n",
       "      (1): nn.ConcatTable {\n",
       "        input\n",
       "          |`-> (1): nn.Sequential {\n",
       "          |      [input -> (1) -> (2) -> (3) -> (4) -> (5) -> output]\n",
       "          |      (1): nn.SpatialConvolution(64 -> 128, 3x3, 2,2, 1,1) without bias\n",
       "          |      (2): nn.SpatialBatchNormalization\n",
       "          |      (3): nn.ReLU\n",
       "          |      (4): nn.SpatialConvolution(128 -> 128, 3x3, 1,1, 1,1) without bias\n",
       "          |      (5): nn.SpatialBatchNormalization\n",
       "          |    }\n",
       "          |`-> (2): nn.SpatialConvolution(64 -> 128, 1x1, 2,2) without bias\n",
       "           ... -> output\n",
       "      }\n",
       "      (2): nn.CAddTable\n",
       "      (3): nn.ReLU\n",
       "    }\n",
       "    (2): nn.Sequential {\n",
       "      [input -> (1) -> (2) -> (3) -> output]\n",
       "      (1): nn.ConcatTable {\n",
       "        input\n",
       "          |`-> (1): nn.Sequential {\n",
       "          |      [input -> (1) -> (2) -> (3) -> (4) -> (5) -> output]\n",
       "          |      (1): nn.SpatialConvolution(128 -> 128, 3x3, 1,1, 1,1) without bias\n",
       "          |      (2): nn.SpatialBatchNormalization\n",
       "          |      (3): nn.ReLU\n",
       "          |      (4): nn.SpatialConvolution(128 -> 128, 3x3, 1,1, 1,1) without bias\n",
       "          |      (5): nn.SpatialBatchNormalization\n",
       "          |    }\n",
       "          |`-> (2): nn.Identity\n",
       "           ... -> output\n",
       "      }\n",
       "      (2): nn.CAddTable\n",
       "      (3): nn.ReLU\n",
       "    }\n",
       "  }\n",
       "  (7): nn.Sequential {\n",
       "    [input -> (1) -> (2) -> output]\n",
       "    (1): nn.Sequential {\n",
       "      [input -> (1) -> (2) -> (3) -> output]\n",
       "      (1): nn.ConcatTable {\n",
       "        input\n",
       "          |`-> (1): nn.Sequential {\n",
       "          |      [input -> (1) -> (2) -> (3) -> (4) -> (5) -> output]\n",
       "          |      (1): nn.SpatialConvolution(128 -> 256, 3x3, 2,2, 1,1) without bias\n",
       "          |      (2): nn.SpatialBatchNormalization\n",
       "          |      (3): nn.ReLU\n",
       "          |      (4): nn.SpatialConvolution(256 -> 256, 3x3, 1,1, 1,1) without bias\n",
       "          |      (5): nn.SpatialBatchNormalization\n",
       "          |    }\n",
       "          |`-> (2): nn.SpatialConvolution(128 -> 256, 1x1, 2,2) without bias\n",
       "           ... -> output\n",
       "      }\n",
       "      (2): nn.CAddTable\n",
       "      (3): nn.ReLU\n",
       "    }\n",
       "    (2): nn.Sequential {\n",
       "      [input -> (1) -> (2) -> (3) -> output]\n",
       "      (1): nn.ConcatTable {\n",
       "        input\n",
       "          |`-> (1): nn.Sequential {\n",
       "          |      [input -> (1) -> (2) -> (3) -> (4) -> (5) -> output]\n",
       "          |      (1): nn.SpatialConvolution(256 -> 256, 3x3, 1,1, 1,1) without bias\n",
       "         "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       " |      (2): nn.SpatialBatchNormalization\n",
       "          |      (3): nn.ReLU\n",
       "          |      (4): nn.SpatialConvolution(256 -> 256, 3x3, 1,1, 1,1) without bias\n",
       "          |      (5): nn.SpatialBatchNormalization\n",
       "          |    }\n",
       "          |`-> (2): nn.Identity\n",
       "           ... -> output\n",
       "      }\n",
       "      (2): nn.CAddTable\n",
       "      (3): nn.ReLU\n",
       "    }\n",
       "  }\n",
       "  (8): nn.Sequential {\n",
       "    [input -> (1) -> (2) -> output]\n",
       "    (1): nn.Sequential {\n",
       "      [input -> (1) -> (2) -> (3) -> output]\n",
       "      (1): nn.ConcatTable {\n",
       "        input\n",
       "          |`-> (1): nn.Sequential {\n",
       "          |      [input -> (1) -> (2) -> (3) -> (4) -> (5) -> output]\n",
       "          |      (1): nn.SpatialConvolution(256 -> 512, 3x3, 2,2, 1,1) without bias\n",
       "          |      (2): nn.SpatialBatchNormalization\n",
       "          |      (3): nn.ReLU\n",
       "          |      (4): nn.SpatialConvolution(512 -> 512, 3x3, 1,1, 1,1) without bias\n",
       "          |      (5): nn.SpatialBatchNormalization\n",
       "          |    }\n",
       "          |`-> (2): nn.SpatialConvolution(256 -> 512, 1x1, 2,2) without bias\n",
       "           ... -> output\n",
       "      }\n",
       "      (2): nn.CAddTable\n",
       "      (3): nn.ReLU\n",
       "    }\n",
       "    (2): nn.Sequential {\n",
       "      [input -> (1) -> (2) -> (3) -> output]\n",
       "      (1): nn.ConcatTable {\n",
       "        input\n",
       "          |`-> (1): nn.Sequential {\n",
       "          |      [input -> (1) -> (2) -> (3) -> (4) -> (5) -> output]\n",
       "          |      (1): nn.SpatialConvolution(512 -> 512, 3x3, 1,1, 1,1) without bias\n",
       "          |      (2): nn.SpatialBatchNormalization\n",
       "          |      (3): nn.ReLU\n",
       "          |      (4): nn.SpatialConvolution(512 -> 512, 3x3, 1,1, 1,1) without bias\n",
       "          |      (5): nn.SpatialBatchNormalization\n",
       "          |    }\n",
       "          |`-> (2): nn.Identity\n",
       "           ... -> output\n",
       "      }\n",
       "      (2): nn.CAddTable\n",
       "      (3): nn.ReLU\n",
       "    }\n",
       "  }\n",
       "  (9): nn.SpatialAveragePooling(7x7, 1,1)\n",
       "  (10): nn.View(512)\n",
       "  (11): nn.Linear(512 -> 1000)\n",
       "}\t\n"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(tostring(net))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Show network graph"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<script type=\"text/javascript\">\n",
       "  $(function() {\n",
       "    $(\"#b87a236e-39de-4871-c8d4-3339d14ca9fe\").html('<?xml version=\"1.0\" encoding=\"UTF-8\" standalone=\"no\"?>\\n<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\\n \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\\n<!-- Generated by graphviz version 2.36.0 (20140111.2315)\\n -->\\n<!-- Title: G Pages: 1 -->\\n<svg width=\"617pt\" height=\"4580pt\"\\n viewBox=\"0.00 0.00 617.00 4580.00\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\\n<g id=\"graph0\" class=\"graph\" transform=\"scale(1 1) rotate(0) translate(4 4576)\">\\n<title>G</title>\\n<polygon fill=\"white\" stroke=\"none\" points=\"-4,4 -4,-4576 613,-4576 613,4 -4,4\"/>\\n<!-- n1 -->\\n<g id=\"node1\" class=\"node\"><title>n1</title>\\n<polygon fill=\"#00ffff\" stroke=\"#00ffff\" points=\"242,-4572 174,-4572 174,-4536 242,-4536 242,-4572\"/>\\n<text text-anchor=\"middle\" x=\"208\" y=\"-4557\" font-family=\"Times,serif\" font-size=\"10.00\">Input</text>\\n<text text-anchor=\"middle\" x=\"208\" y=\"-4546\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 1</text>\\n</g>\\n<!-- n2 -->\\n<g id=\"node2\" class=\"node\"><title>n2</title>\\n<polygon fill=\"#ffb90f\" stroke=\"#ffb90f\" points=\"335,-4500 81,-4500 81,-4464 335,-4464 335,-4500\"/>\\n<text text-anchor=\"middle\" x=\"208\" y=\"-4485\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(3 &#45;&gt; 64, 7x7, 2,2, 3,3) without bias</text>\\n<text text-anchor=\"middle\" x=\"208\" y=\"-4474\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 2</text>\\n</g>\\n<!-- n1&#45;&gt;n2 -->\\n<g id=\"edge1\" class=\"edge\"><title>n1&#45;&gt;n2</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M208,-4535.7C208,-4527.98 208,-4518.71 208,-4510.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"211.5,-4510.1 208,-4500.1 204.5,-4510.1 211.5,-4510.1\"/>\\n</g>\\n<!-- n3 -->\\n<g id=\"node3\" class=\"node\"><title>n3</title>\\n<polygon fill=\"#00eeee\" stroke=\"#00eeee\" points=\"278.25,-4428 137.75,-4428 137.75,-4392 278.25,-4392 278.25,-4428\"/>\\n<text text-anchor=\"middle\" x=\"208\" y=\"-4413\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialBatchNormalization</text>\\n<text text-anchor=\"middle\" x=\"208\" y=\"-4402\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 3</text>\\n</g>\\n<!-- n2&#45;&gt;n3 -->\\n<g id=\"edge2\" class=\"edge\"><title>n2&#45;&gt;n3</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M208,-4463.7C208,-4455.98 208,-4446.71 208,-4438.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"211.5,-4438.1 208,-4428.1 204.5,-4438.1 211.5,-4438.1\"/>\\n</g>\\n<!-- n4 -->\\n<g id=\"node4\" class=\"node\"><title>n4</title>\\n<polygon fill=\"#00eeee\" stroke=\"#00eeee\" points=\"242,-4356 174,-4356 174,-4320 242,-4320 242,-4356\"/>\\n<text text-anchor=\"middle\" x=\"208\" y=\"-4341\" font-family=\"Times,serif\" font-size=\"10.00\">nn.ReLU</text>\\n<text text-anchor=\"middle\" x=\"208\" y=\"-4330\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 3</text>\\n</g>\\n<!-- n3&#45;&gt;n4 -->\\n<g id=\"edge3\" class=\"edge\"><title>n3&#45;&gt;n4</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M208,-4391.7C208,-4383.98 208,-4374.71 208,-4366.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"211.5,-4366.1 208,-4356.1 204.5,-4366.1 211.5,-4366.1\"/>\\n</g>\\n<!-- n5 -->\\n<g id=\"node5\" class=\"node\"><title>n5</title>\\n<polygon fill=\"#cd3333\" stroke=\"#cd3333\" points=\"290.25,-4284 125.75,-4284 125.75,-4248 290.25,-4248 290.25,-4284\"/>\\n<text text-anchor=\"middle\" x=\"208\" y=\"-4269\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialMaxPooling(3x3, 2,2, 1,1)</text>\\n<text text-anchor=\"middle\" x=\"208\" y=\"-4258\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 4</text>\\n</g>\\n<!-- n4&#45;&gt;n5 -->\\n<g id=\"edge4\" class=\"edge\"><title>n4&#45;&gt;n5</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M208,-4319.7C208,-4311.98 208,-4302.71 208,-4294.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"211.5,-4294.1 208,-4284.1 204.5,-4294.1 211.5,-4294.1\"/>\\n</g>\\n<!-- n6 -->\\n<g id=\"node6\" class=\"node\"><title>n6</title>\\n<polygon fill=\"#7fffd4\" stroke=\"#7fffd4\" points=\"258.25,-4212 -0.25,-4212 -0.25,-4176 258.25,-4176 258.25,-4212\"/>\\n<text text-anchor=\"middle\" x=\"129\" y=\"-4197\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(64 &#45;&gt; 64, 3x3, 1,1, 1,1) without bias</text>\\n<text text-anchor=\"middle\" x=\"129\" y=\"-4186\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 5</text>\\n</g>\\n<!-- n5&#45;&gt;n6 -->\\n<g id=\"edge5\" class=\"edge\"><title>n5&#45;&gt;n6</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M188.472,-4247.7C178.618,-4238.97 166.514,-4228.24 155.804,-4218.75\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"158.109,-4216.12 148.303,-4212.1 153.467,-4221.36 158.109,-4216.12\"/>\\n</g>\\n<!-- n11 -->\\n<g id=\"node11\" class=\"node\"><title>n11</title>\\n<polygon fill=\"#8b8878\" stroke=\"#8b8878\" points=\"261.25,-3852 184.75,-3852 184.75,-3816 261.25,-3816 261.25,-3852\"/>\\n<text text-anchor=\"middle\" x=\"223\" y=\"-3837\" font-family=\"Times,serif\" font-size=\"10.00\">nn.CAddTable</text>\\n<text text-anchor=\"middle\" x=\"223\" y=\"-3826\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 9</text>\\n</g>\\n<!-- n5&#45;&gt;n11 -->\\n<g id=\"edge11\" class=\"edge\"><title>n5&#45;&gt;n11</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M233.49,-4247.86C245.386,-4238.58 258.717,-4226.14 267,-4212 287.495,-4177.02 287,-4163.54 287,-4123 287,-4123 287,-4123 287,-3977 287,-3933.08 261.237,-3887.73 242.403,-3860.44\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"245.22,-3858.37 236.573,-3852.24 239.515,-3862.42 245.22,-3858.37\"/>\\n</g>\\n<!-- n7 -->\\n<g id=\"node7\" class=\"node\"><title>n7</title>\\n<polygon fill=\"goldenrod\" stroke=\"goldenrod\" points=\"199.25,-4140 58.75,-4140 58.75,-4104 199.25,-4104 199.25,-4140\"/>\\n<text text-anchor=\"middle\" x=\"129\" y=\"-4125\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialBatchNormalization</text>\\n<text text-anchor=\"middle\" x=\"129\" y=\"-4114\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 6</text>\\n</g>\\n<!-- n6&#45;&gt;n7 -->\\n<g id=\"edge6\" class=\"edge\"><title>n6&#45;&gt;n7</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M129,-4175.7C129,-4167.98 129,-4158.71 129,-4150.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"132.5,-4150.1 129,-4140.1 125.5,-4150.1 132.5,-4150.1\"/>\\n</g>\\n<!-- n8 -->\\n<g id=\"node8\" class=\"node\"><title>n8</title>\\n<polygon fill=\"goldenrod\" stroke=\"goldenrod\" points=\"163,-4068 95,-4068 95,-4032 163,-4032 163,-4068\"/>\\n<text text-anchor=\"middle\" x=\"129\" y=\"-4053\" font-family=\"Times,serif\" font-size=\"10.00\">nn.ReLU</text>\\n<text text-anchor=\"middle\" x=\"129\" y=\"-4042\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 6</text>\\n</g>\\n<!-- n7&#45;&gt;n8 -->\\n<g id=\"edge7\" class=\"edge\"><title>n7&#45;&gt;n8</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M129,-4103.7C129,-4095.98 129,-4086.71 129,-4078.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"132.5,-4078.1 129,-4068.1 125.5,-4078.1 132.5,-4078.1\"/>\\n</g>\\n<!-- n9 -->\\n<g id=\"node9\" class=\"node\"><title>n9</title>\\n<polygon fill=\"#ffd39b\" stroke=\"#ffd39b\" points=\"258.25,-3996 -0.25,-3996 -0.25,-3960 258.25,-3960 258.25,-3996\"/>\\n<text text-anchor=\"middle\" x=\"129\" y=\"-3981\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(64 &#45;&gt; 64, 3x3, 1,1, 1,1) without bias</text>\\n<text text-anchor=\"middle\" x=\"129\" y=\"-3970\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 7</text>\\n</g>\\n<!-- n8&#45;&gt;n9 -->\\n<g id=\"edge8\" class=\"edge\"><title>n8&#45;&gt;n9</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M129,-4031.7C129,-4023.98 129,-4014.71 129,-4006.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"132.5,-4006.1 129,-3996.1 125.5,-4006.1 132.5,-4006.1\"/>\\n</g>\\n<!-- n10 -->\\n<g id=\"node10\" class=\"node\"><title>n10</title>\\n<polygon fill=\"#cd3333\" stroke=\"#cd3333\" points=\"229.25,-3924 88.75,-3924 88.75,-3888 229.25,-3888 229.25,-3924\"/>\\n<text text-anchor=\"middle\" x=\"159\" y=\"-3909\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialBatchNormalization</text>\\n<text text-anchor=\"middle\" x=\"159\" y=\"-3898\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 8</text>\\n</g>\\n<!-- n9&#45;&gt;n10 -->\\n<g id=\"edge9\" class=\"edge\"><title>n9&#45;&gt;n10</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M136.416,-3959.7C139.794,-3951.81 143.87,-3942.3 147.623,-3933.55\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"150.947,-3934.67 151.67,-3924.1 144.513,-3931.92 150.947,-3934.67\"/>\\n</g>\\n<!-- n10&#45;&gt;n11 -->\\n<g id=\"edge10\" class=\"edge\"><title>n10&#45;&gt;n11</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M174.82,-3887.7C182.57,-3879.22 192.039,-3868.86 200.524,-3859.58\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"203.197,-3861.85 207.362,-3852.1 198.031,-3857.12 203.197,-3861.85\"/>\\n</g>\\n<!-- n12 -->\\n<g id=\"node12\" class=\"node\"><title>n12</title>\\n<polygon fill=\"#8b8878\" stroke=\"#8b8878\" points=\"257,-3780 189,-3780 189,-3744 257,-3744 257,-3780\"/>\\n<text text-anchor=\"middle\" x=\"223\" y=\"-3765\" font-family=\"Times,serif\" font-size=\"10.00\">nn.ReLU</text>\\n<text text-anchor=\"middle\" x=\"223\" y=\"-3754\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 9</text>\\n</g>\\n<!-- n11&#45;&gt;n12 -->\\n<g id=\"edge12\" class=\"edge\"><title>n11&#45;&gt;n12</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M223,-3815.7C223,-3807.98 223,-3798.71 223,-3790.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"226.5,-3790.1 223,-3780.1 219.5,-3790.1 226.5,-3790.1\"/>\\n</g>\\n<!-- n13 -->\\n<g id=\"node13\" class=\"node\"><title>n13</title>\\n<polygon fill=\"gold\" stroke=\"gold\" points=\"273.25,-3708 14.75,-3708 14.75,-3672 273.25,-3672 273.25,-3708\"/>\\n<text text-anchor=\"middle\" x=\"144\" y=\"-3693\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(64 &#45;&gt; 64, 3x3, 1,1, 1,1) without bias</text>\\n<text text-anchor=\"middle\" x=\"144\" y=\"-3682\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 10</text>\\n</g>\\n<!-- n12&#45;&gt;n13 -->\\n<g id=\"edge13\" class=\"edge\"><title>n12&#45;&gt;n13</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M203.472,-3743.7C193.618,-3734.97 181.514,-3724.24 170.804,-3714.75\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"173.109,-3712.12 163.303,-3708.1 168.467,-3717.36 173.109,-3712.12\"/>\\n</g>\\n<!-- n18 -->\\n<g id=\"node18\" class=\"node\"><title>n18</title>\\n<polygon fill=\"#8b6508\" stroke=\"#8b6508\" points=\"276.25,-3348 199.75,-3348 199.75,-3312 276.25,-3312 276.25,-3348\"/>\\n<text text-anchor=\"middle\" x=\"238\" y=\"-3333\" font-family=\"Times,serif\" font-size=\"10.00\">nn.CAddTable</text>\\n<text text-anchor=\"middle\" x=\"238\" y=\"-3322\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 14</text>\\n</g>\\n<!-- n12&#45;&gt;n18 -->\\n<g id=\"edge19\" class=\"edge\"><title>n12&#45;&gt;n18</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M248.49,-3743.86C260.386,-3734.58 273.717,-3722.14 282,-3708 302.495,-3673.02 302,-3659.54 302,-3619 302,-3619 302,-3619 302,-3473 302,-3429.08 276.237,-3383.73 257.403,-3356.44\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"260.22,-3354.37 251.573,-3348.24 254.515,-3358.42 260.22,-3354.37\"/>\\n</g>\\n<!-- n14 -->\\n<g id=\"node14\" class=\"node\"><title>n14</title>\\n<polygon fill=\"darkorchid\" stroke=\"darkorchid\" points=\"214.25,-3636 73.75,-3636 73.75,-3600 214.25,-3600 214.25,-3636\"/>\\n<text text-anchor=\"middle\" x=\"144\" y=\"-3621\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialBatchNormalization</text>\\n<text text-anchor=\"middle\" x=\"144\" y=\"-3610\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 11</text>\\n</g>\\n<!-- n13&#45;&gt;n14 -->\\n<g id=\"edge14\" class=\"edge\"><title>n13&#45;&gt;n14</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M144,-3671.7C144,-3663.98 144,-3654.71 144,-3646.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"147.5,-3646.1 144,-3636.1 140.5,-3646.1 147.5,-3646.1\"/>\\n</g>\\n<!-- n15 -->\\n<g id=\"node15\" class=\"node\"><title>n15</title>\\n<polygon fill=\"darkorchid\" stroke=\"darkorchid\" points=\"181,-3564 107,-3564 107,-3528 181,-3528 181,-3564\"/>\\n<text text-anchor=\"middle\" x=\"144\" y=\"-3549\" font-family=\"Times,serif\" font-size=\"10.00\">nn.ReLU</text>\\n<text text-anchor=\"middle\" x=\"144\" y=\"-3538\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 11</text>\\n</g>\\n<!-- n14&#45;&gt;n15 -->\\n<g id=\"edge15\" class=\"edge\"><title>n14&#45;&gt;n15</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M144,-3599.7C144,-3591.98 144,-3582.71 144,-3574.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"147.5,-3574.1 144,-3564.1 140.5,-3574.1 147.5,-3574.1\"/>\\n</g>\\n<!-- n16 -->\\n<g id=\"node16\" class=\"node\"><title>n16</title>\\n<polygon fill=\"chartreuse\" stroke=\"chartreuse\" points=\"273.25,-3492 14.75,-3492 14.75,-3456 273.25,-3456 273.25,-3492\"/>\\n<text text-anchor=\"middle\" x=\"144\" y=\"-3477\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(64 &#45;&gt; 64, 3x3, 1,1, 1,1) without bias</text>\\n<text text-anchor=\"middle\" x=\"144\" y=\"-3466\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 12</text>\\n</g>\\n<!-- n15&#45;&gt;n16 -->\\n<g id=\"edge16\" class=\"edge\"><title>n15&#45;&gt;n16</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M144,-3527.7C144,-3519.98 144,-3510.71 144,-3502.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"147.5,-3502.1 144,-3492.1 140.5,-3502.1 147.5,-3502.1\"/>\\n</g>\\n<!-- n17 -->\\n<g id=\"node17\" class=\"node\"><title>n17</title>\\n<polygon fill=\"aliceblue\" stroke=\"aliceblue\" points=\"244.25,-3420 103.75,-3420 103.75,-3384 244.25,-3384 244.25,-3420\"/>\\n<text text-anchor=\"middle\" x=\"174\" y=\"-3405\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialBatchNormalization</text>\\n<text text-anchor=\"middle\" x=\"174\" y=\"-3394\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 13</text>\\n</g>\\n<!-- n16&#45;&gt;n17 -->\\n<g id=\"edge17\" class=\"edge\"><title>n16&#45;&gt;n17</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M151.416,-3455.7C154.794,-3447.81 158.87,-3438.3 162.623,-3429.55\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"165.947,-3430.67 166.67,-3420.1 159.513,-3427.92 165.947,-3430.67\"/>\\n</g>\\n<!-- n17&#45;&gt;n18 -->\\n<g id=\"edge18\" class=\"edge\"><title>n17&#45;&gt;n18</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M189.82,-3383.7C197.57,-3375.22 207.039,-3364.86 215.524,-3355.58\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"218.197,-3357.85 222.362,-3348.1 213.031,-3353.12 218.197,-3357.85\"/>\\n</g>\\n<!-- n19 -->\\n<g id=\"node19\" class=\"node\"><title>n19</title>\\n<polygon fill=\"#8b6508\" stroke=\"#8b6508\" points=\"275,-3276 201,-3276 201,-3240 275,-3240 275,-3276\"/>\\n<text text-anchor=\"middle\" x=\"238\" y=\"-3261\" font-family=\"Times,serif\" font-size=\"10.00\">nn.ReLU</text>\\n<text text-anchor=\"middle\" x=\"238\" y=\"-3250\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 14</text>\\n</g>\\n<!-- n18&#45;&gt;n19 -->\\n<g id=\"edge20\" class=\"edge\"><title>n18&#45;&gt;n19</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M238,-3311.7C238,-3303.98 238,-3294.71 238,-3286.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"241.5,-3286.1 238,-3276.1 234.5,-3286.1 241.5,-3286.1\"/>\\n</g>\\n<!-- n20 -->\\n<g id=\"node20\" class=\"node\"><title>n20</title>\\n<polygon fill=\"#ff3030\" stroke=\"#ff3030\" points=\"290,-3204 26,-3204 26,-3168 290,-3168 290,-3204\"/>\\n<text text-anchor=\"middle\" x=\"158\" y=\"-3189\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(64 &#45;&gt; 128, 3x3, 2,2, 1,1) without bias</text>\\n<text text-anchor=\"middle\" x=\"158\" y=\"-3178\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 15</text>\\n</g>\\n<!-- n19&#45;&gt;n20 -->\\n<g id=\"edge21\" class=\"edge\"><title>n19&#45;&gt;n20</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M218.225,-3239.7C208.246,-3230.97 195.989,-3220.24 185.143,-3210.75\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"187.378,-3208.06 177.548,-3204.1 182.769,-3213.32 187.378,-3208.06\"/>\\n</g>\\n<!-- n25 -->\\n<g id=\"node25\" class=\"node\"><title>n25</title>\\n<polygon fill=\"chartreuse\" stroke=\"chartreuse\" points=\"480.25,-3132 233.75,-3132 233.75,-3096 480.25,-3096 480.25,-3132\"/>\\n<text text-anchor=\"middle\" x=\"357\" y=\"-3117\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(64 &#45;&gt; 128, 1x1, 2,2) without bias</text>\\n<text text-anchor=\"middle\" x=\"357\" y=\"-3106\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 19</text>\\n</g>\\n<!-- n19&#45;&gt;n25 -->\\n<g id=\"edge26\" class=\"edge\"><title>n19&#45;&gt;n25</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M260.524,-3239.96C272.669,-3230.17 287.483,-3217.22 299,-3204 316.097,-3184.37 331.973,-3159.33 342.777,-3140.81\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"345.866,-3142.46 347.8,-3132.04 339.792,-3138.98 345.866,-3142.46\"/>\\n</g>\\n<!-- n21 -->\\n<g id=\"node21\" class=\"node\"><title>n21</title>\\n<polygon fill=\"#e0eeee\" stroke=\"#e0eeee\" points=\"215.25,-3132 74.75,-3132 74.75,-3096 215.25,-3096 215.25,-3132\"/>\\n<text text-anchor=\"middle\" x=\"145\" y=\"-3117\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialBatchNormalization</text>\\n<text text-anchor=\"middle\" x=\"145\" y=\"-3106\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 16</text>\\n</g>\\n<!-- n20&#45;&gt;n21 -->\\n<g id=\"edge22\" class=\"edge\"><title>n20&#45;&gt;n21</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M154.787,-3167.7C153.354,-3159.98 151.632,-3150.71 150.035,-3142.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"153.444,-3141.3 148.177,-3132.1 146.561,-3142.58 153.444,-3141.3\"/>\\n</g>\\n<!-- n22 -->\\n<g id=\"node22\" class=\"node\"><title>n22</title>\\n<polygon fill=\"#e0eeee\" stroke=\"#e0eeee\" points=\"206,-3060 132,-3060 132,-3024 206,-3024 206,-3060\"/>\\n<text text-anchor=\"middle\" x=\"169\" y=\"-3045\" font-family=\"Times,serif\" font-size=\"10.00\">nn.ReLU</text>\\n<text text-anchor=\"middle\" x=\"169\" y=\"-3034\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 16</text>\\n</g>\\n<!-- n21&#45;&gt;n22 -->\\n<g id=\"edge23\" class=\"edge\"><title>n21&#45;&gt;n22</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M150.933,-3095.7C153.606,-3087.9 156.826,-3078.51 159.801,-3069.83\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"163.203,-3070.7 163.136,-3060.1 156.582,-3068.43 163.203,-3070.7\"/>\\n</g>\\n<!-- n23 -->\\n<g id=\"node23\" class=\"node\"><title>n23</title>\\n<polygon fill=\"aliceblue\" stroke=\"aliceblue\" points=\"315.25,-2988 46.75,-2988 46.75,-2952 315.25,-2952 315.25,-2988\"/>\\n<text text-anchor=\"middle\" x=\"181\" y=\"-2973\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(128 &#45;&gt; 128, 3x3, 1,1, 1,1) without bias</text>\\n<text text-anchor=\"middle\" x=\"181\" y=\"-2962\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 17</text>\\n</g>\\n<!-- n22&#45;&gt;n23 -->\\n<g id=\"edge24\" class=\"edge\"><title>n22&#45;&gt;n23</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M171.966,-3023.7C173.289,-3015.98 174.878,-3006.71 176.352,-2998.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"179.828,-2998.55 178.068,-2988.1 172.928,-2997.37 179.828,-2998.55\"/>\\n</g>\\n<!-- n24 -->\\n<g id=\"node24\" class=\"node\"><title>n24</title>\\n<polygon fill=\"#458b00\" stroke=\"#458b00\" points=\"286.25,-2916 145.75,-2916 145.75,-2880 286.25,-2880 286.25,-2916\"/>\\n<text text-anchor=\"middle\" x=\"216\" y=\"-2901\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialBatchNormalization</text>\\n<text text-anchor=\"middle\" x=\"216\" y=\"-2890\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 18</text>\\n</g>\\n<!-- n23&#45;&gt;n24 -->\\n<g id=\"edge25\" class=\"edge\"><title>n23&#45;&gt;n24</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M189.652,-2951.7C193.636,-2943.73 198.45,-2934.1 202.868,-2925.26\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"206.106,-2926.61 207.448,-2916.1 199.845,-2923.48 206.106,-2926.61\"/>\\n</g>\\n<!-- n26 -->\\n<g id=\"node26\" class=\"node\"><title>n26</title>\\n<polygon fill=\"darkorchid\" stroke=\"darkorchid\" points=\"321.25,-2844 244.75,-2844 244.75,-2808 321.25,-2808 321.25,-2844\"/>\\n<text text-anchor=\"middle\" x=\"283\" y=\"-2829\" font-family=\"Times,serif\" font-size=\"10.00\">nn.CAddTable</text>\\n<text text-anchor=\"middle\" x=\"283\" y=\"-2818\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 20</text>\\n</g>\\n<!-- n24&#45;&gt;n26 -->\\n<g id=\"edge27\" class=\"edge\"><title>n24&#45;&gt;n26</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M232.562,-2879.7C240.675,-2871.22 250.588,-2860.86 259.471,-2851.58\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"262.243,-2853.75 266.629,-2844.1 257.186,-2848.91 262.243,-2853.75\"/>\\n</g>\\n<!-- n25&#45;&gt;n26 -->\\n<g id=\"edge28\" class=\"edge\"><title>n25&#45;&gt;n26</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M354.166,-3095.93C349.111,-3066.37 337.767,-3003.84 324,-2952 314.881,-2917.66 301.705,-2878.9 292.769,-2853.77\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"296.01,-2852.43 289.339,-2844.2 289.421,-2854.8 296.01,-2852.43\"/>\\n</g>\\n<!-- n27 -->\\n<g id=\"node27\" class=\"node\"><title>n27</title>\\n<polygon fill=\"darkorchid\" stroke=\"darkorchid\" points=\"320,-2772 246,-2772 246,-2736 320,-2736 320,-2772\"/>\\n<text text-anchor=\"middle\" x=\"283\" y=\"-2757\" font-family=\"Times,serif\" font-size=\"10.00\">nn.ReLU</text>\\n<text text-anchor=\"middle\" x=\"283\" y=\"-2746\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 20</text>\\n</g>\\n<!-- n26&#45;&gt;n27 -->\\n<g id=\"edge29\" class=\"edge\"><title>n26&#45;&gt;n27</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M283,-2807.7C283,-2799.98 283,-2790.71 283,-2782.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"286.5,-2782.1 283,-2772.1 279.5,-2782.1 286.5,-2782.1\"/>\\n</g>\\n<!-- n28 -->\\n<g id=\"node28\" class=\"node\"><title>n28</title>\\n<polygon fill=\"darkviolet\" stroke=\"darkviolet\" points=\"336.25,-2700 67.75,-2700 67.75,-2664 336.25,-2664 336.25,-2700\"/>\\n<text text-anchor=\"middle\" x=\"202\" y=\"-2685\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(128 &#45;&gt; 128, 3x3, 1,1, 1,1) without bias</text>\\n<text text-anchor=\"middle\" x=\"202\" y=\"-2674\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 21</text>\\n</g>\\n<!-- n27&#45;&gt;n28 -->\\n<g id=\"edge30\" class=\"edge\"><title>n27&#45;&gt;n28</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M262.978,-2735.7C252.874,-2726.97 240.464,-2716.24 229.482,-2706.75\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"231.647,-2703.99 221.792,-2700.1 227.07,-2709.29 231.647,-2703.99\"/>\\n</g>\\n<!-- n33 -->\\n<g id=\"node33\" class=\"node\"><title>n33</title>\\n<polygon fill=\"dimgrey\" stroke=\"dimgrey\" points=\"337.25,-2340 260.75,-2340 260.75,-2304 337.25,-2304 337.25,-2340\"/>\\n<text text-anchor=\"middle\" x=\"299\" y=\"-2325\" font-family=\"Times,serif\" font-size=\"10.00\">nn.CAddTable</text>\\n<text text-anchor=\"middle\" x=\"299\" y=\"-2314\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 25</text>\\n</g>\\n<!-- n27&#45;&gt;n33 -->\\n<g id=\"edge36\" class=\"edge\"><title>n27&#45;&gt;n33</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M310.108,-2735.88C322.53,-2726.69 336.376,-2714.31 345,-2700 365.923,-2665.27 365,-2651.54 365,-2611 365,-2611 365,-2611 365,-2465 365,-2420.84 338.432,-2375.56 319.009,-2348.35\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"321.742,-2346.16 312.997,-2340.18 316.103,-2350.31 321.742,-2346.16\"/>\\n</g>\\n<!-- n29 -->\\n<g id=\"node29\" class=\"node\"><title>n29</title>\\n<polygon fill=\"darkgreen\" stroke=\"darkgreen\" points=\"272.25,-2628 131.75,-2628 131.75,-2592 272.25,-2592 272.25,-2628\"/>\\n<text text-anchor=\"middle\" x=\"202\" y=\"-2613\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialBatchNormalization</text>\\n<text text-anchor=\"middle\" x=\"202\" y=\"-2602\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 22</text>\\n</g>\\n<!-- n28&#45;&gt;n29 -->\\n<g id=\"edge31\" class=\"edge\"><title>n28&#45;&gt;n29</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M202,-2663.7C202,-2655.98 202,-2646.71 202,-2638.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"205.5,-2638.1 202,-2628.1 198.5,-2638.1 205.5,-2638.1\"/>\\n</g>\\n<!-- n30 -->\\n<g id=\"node30\" class=\"node\"><title>n30</title>\\n<polygon fill=\"darkgreen\" stroke=\"darkgreen\" points=\"239,-2556 165,-2556 165,-2520 239,-2520 239,-2556\"/>\\n<text text-anchor=\"middle\" x=\"202\" y=\"-2541\" font-family=\"Times,serif\" font-size=\"10.00\">nn.ReLU</text>\\n<text text-anchor=\"middle\" x=\"202\" y=\"-2530\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 22</text>\\n</g>\\n<!-- n29&#45;&gt;n30 -->\\n<g id=\"edge32\" class=\"edge\"><title>n29&#45;&gt;n30</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M202,-2591.7C202,-2583.98 202,-2574.71 202,-2566.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"205.5,-2566.1 202,-2556.1 198.5,-2566.1 205.5,-2566.1\"/>\\n</g>\\n<!-- n31 -->\\n<g id=\"node31\" class=\"node\"><title>n31</title>\\n<polygon fill=\"#9a32cd\" stroke=\"#9a32cd\" points=\"336.25,-2484 67.75,-2484 67.75,-2448 336.25,-2448 336.25,-2484\"/>\\n<text text-anchor=\"middle\" x=\"202\" y=\"-2469\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(128 &#45;&gt; 128, 3x3, 1,1, 1,1) without bias</text>\\n<text text-anchor=\"middle\" x=\"202\" y=\"-2458\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 23</text>\\n</g>\\n<!-- n30&#45;&gt;n31 -->\\n<g id=\"edge33\" class=\"edge\"><title>n30&#45;&gt;n31</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M202,-2519.7C202,-2511.98 202,-2502.71 202,-2494.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"205.5,-2494.1 202,-2484.1 198.5,-2494.1 205.5,-2494.1\"/>\\n</g>\\n<!-- n32 -->\\n<g id=\"node32\" class=\"node\"><title>n32</title>\\n<polygon fill=\"cornsilk\" stroke=\"cornsilk\" points=\"304.25,-2412 163.75,-2412 163.75,-2376 304.25,-2376 304.25,-2412\"/>\\n<text text-anchor=\"middle\" x=\"234\" y=\"-2397\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialBatchNormalization</text>\\n<text text-anchor=\"middle\" x=\"234\" y=\"-2386\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 24</text>\\n</g>\\n<!-- n31&#45;&gt;n32 -->\\n<g id=\"edge34\" class=\"edge\"><title>n31&#45;&gt;n32</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M209.91,-2447.7C213.553,-2439.73 217.954,-2430.1 221.994,-2421.26\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"225.206,-2422.65 226.181,-2412.1 218.84,-2419.74 225.206,-2422.65\"/>\\n</g>\\n<!-- n32&#45;&gt;n33 -->\\n<g id=\"edge35\" class=\"edge\"><title>n32&#45;&gt;n33</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M250.067,-2375.7C257.939,-2367.22 267.555,-2356.86 276.173,-2347.58\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"278.878,-2349.81 283.117,-2340.1 273.748,-2345.05 278.878,-2349.81\"/>\\n</g>\\n<!-- n34 -->\\n<g id=\"node34\" class=\"node\"><title>n34</title>\\n<polygon fill=\"dimgrey\" stroke=\"dimgrey\" points=\"336,-2268 262,-2268 262,-2232 336,-2232 336,-2268\"/>\\n<text text-anchor=\"middle\" x=\"299\" y=\"-2253\" font-family=\"Times,serif\" font-size=\"10.00\">nn.ReLU</text>\\n<text text-anchor=\"middle\" x=\"299\" y=\"-2242\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 25</text>\\n</g>\\n<!-- n33&#45;&gt;n34 -->\\n<g id=\"edge37\" class=\"edge\"><title>n33&#45;&gt;n34</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M299,-2303.7C299,-2295.98 299,-2286.71 299,-2278.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"302.5,-2278.1 299,-2268.1 295.5,-2278.1 302.5,-2278.1\"/>\\n</g>\\n<!-- n35 -->\\n<g id=\"node35\" class=\"node\"><title>n35</title>\\n<polygon fill=\"#00cdcd\" stroke=\"#00cdcd\" points=\"352.25,-2196 83.75,-2196 83.75,-2160 352.25,-2160 352.25,-2196\"/>\\n<text text-anchor=\"middle\" x=\"218\" y=\"-2181\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(128 &#45;&gt; 256, 3x3, 2,2, 1,1) without bias</text>\\n<text text-anchor=\"middle\" x=\"218\" y=\"-2170\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 26</text>\\n</g>\\n<!-- n34&#45;&gt;n35 -->\\n<g id=\"edge38\" class=\"edge\"><title>n34&#45;&gt;n35</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M278.978,-2231.7C268.874,-2222.97 256.464,-2212.24 245.482,-2202.75\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"247.647,-2199.99 237.792,-2196.1 243.07,-2205.29 247.647,-2199.99\"/>\\n</g>\\n<!-- n40 -->\\n<g id=\"node40\" class=\"node\"><title>n40</title>\\n<polygon fill=\"#8b4513\" stroke=\"#8b4513\" points=\"546,-2124 294,-2124 294,-2088 546,-2088 546,-2124\"/>\\n<text text-anchor=\"middle\" x=\"420\" y=\"-2109\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(128 &#45;&gt; 256, 1x1, 2,2) without bias</text>\\n<text text-anchor=\"middle\" x=\"420\" y=\"-2098\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 30</text>\\n</g>\\n<!-- n34&#45;&gt;n40 -->\\n<g id=\"edge43\" class=\"edge\"><title>n34&#45;&gt;n40</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M321.874,-2231.98C334.212,-2222.2 349.269,-2209.26 361,-2196 378.342,-2176.4 394.497,-2151.36 405.501,-2132.83\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"408.604,-2134.46 410.619,-2124.06 402.558,-2130.93 408.604,-2134.46\"/>\\n</g>\\n<!-- n36 -->\\n<g id=\"node36\" class=\"node\"><title>n36</title>\\n<polygon fill=\"blanchedalmond\" stroke=\"blanchedalmond\" points=\"275.25,-2124 134.75,-2124 134.75,-2088 275.25,-2088 275.25,-2124\"/>\\n<text text-anchor=\"middle\" x=\"205\" y=\"-2109\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialBatchNormalization</text>\\n<text text-anchor=\"middle\" x=\"205\" y=\"-2098\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 27</text>\\n</g>\\n<!-- n35&#45;&gt;n36 -->\\n<g id=\"edge39\" class=\"edge\"><title>n35&#45;&gt;n36</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M214.787,-2159.7C213.354,-2151.98 211.632,-2142.71 210.035,-2134.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"213.444,-2133.3 208.177,-2124.1 206.561,-2134.58 213.444,-2133.3\"/>\\n</g>\\n<!-- n37 -->\\n<g id=\"node37\" class=\"node\"><title>n37</title>\\n<polygon fill=\"blanchedalmond\" stroke=\"blanchedalmond\" points=\"268,-2052 194,-2052 194,-2016 268,-2016 268,-2052\"/>\\n<text text-anchor=\"middle\" x=\"231\" y=\"-2037\" font-family=\"Times,serif\" font-size=\"10.00\">nn.ReLU</text>\\n<text text-anchor=\"middle\" x=\"231\" y=\"-2026\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 27</text>\\n</g>\\n<!-- n36&#45;&gt;n37 -->\\n<g id=\"edge40\" class=\"edge\"><title>n36&#45;&gt;n37</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M211.427,-2087.7C214.355,-2079.81 217.887,-2070.3 221.14,-2061.55\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"224.446,-2062.7 224.647,-2052.1 217.884,-2060.26 224.446,-2062.7\"/>\\n</g>\\n<!-- n38 -->\\n<g id=\"node38\" class=\"node\"><title>n38</title>\\n<polygon fill=\"#6e8b3d\" stroke=\"#6e8b3d\" points=\"378.25,-1980 109.75,-1980 109.75,-1944 378.25,-1944 378.25,-1980\"/>\\n<text text-anchor=\"middle\" x=\"244\" y=\"-1965\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(256 &#45;&gt; 256, 3x3, 1,1, 1,1) without bias</text>\\n<text text-anchor=\"middle\" x=\"244\" y=\"-1954\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 28</text>\\n</g>\\n<!-- n37&#45;&gt;n38 -->\\n<g id=\"edge41\" class=\"edge\"><title>n37&#45;&gt;n38</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M234.213,-2015.7C235.646,-2007.98 237.368,-1998.71 238.965,-1990.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"242.439,-1990.58 240.823,-1980.1 235.556,-1989.3 242.439,-1990.58\"/>\\n</g>\\n<!-- n39 -->\\n<g id=\"node39\" class=\"node\"><title>n39</title>\\n<polygon fill=\"#79cdcd\" stroke=\"#79cdcd\" points=\"349.25,-1908 208.75,-1908 208.75,-1872 349.25,-1872 349.25,-1908\"/>\\n<text text-anchor=\"middle\" x=\"279\" y=\"-1893\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialBatchNormalization</text>\\n<text text-anchor=\"middle\" x=\"279\" y=\"-1882\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 29</text>\\n</g>\\n<!-- n38&#45;&gt;n39 -->\\n<g id=\"edge42\" class=\"edge\"><title>n38&#45;&gt;n39</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M252.652,-1943.7C256.636,-1935.73 261.45,-1926.1 265.868,-1917.26\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"269.106,-1918.61 270.448,-1908.1 262.845,-1915.48 269.106,-1918.61\"/>\\n</g>\\n<!-- n41 -->\\n<g id=\"node41\" class=\"node\"><title>n41</title>\\n<polygon fill=\"#ee2c2c\" stroke=\"#ee2c2c\" points=\"384.25,-1836 307.75,-1836 307.75,-1800 384.25,-1800 384.25,-1836\"/>\\n<text text-anchor=\"middle\" x=\"346\" y=\"-1821\" font-family=\"Times,serif\" font-size=\"10.00\">nn.CAddTable</text>\\n<text text-anchor=\"middle\" x=\"346\" y=\"-1810\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 31</text>\\n</g>\\n<!-- n39&#45;&gt;n41 -->\\n<g id=\"edge44\" class=\"edge\"><title>n39&#45;&gt;n41</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M295.562,-1871.7C303.675,-1863.22 313.588,-1852.86 322.471,-1843.58\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"325.243,-1845.75 329.629,-1836.1 320.186,-1840.91 325.243,-1845.75\"/>\\n</g>\\n<!-- n40&#45;&gt;n41 -->\\n<g id=\"edge45\" class=\"edge\"><title>n40&#45;&gt;n41</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M417.166,-2087.93C412.111,-2058.37 400.767,-1995.84 387,-1944 377.881,-1909.66 364.705,-1870.9 355.769,-1845.77\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"359.01,-1844.43 352.339,-1836.2 352.421,-1846.8 359.01,-1844.43\"/>\\n</g>\\n<!-- n42 -->\\n<g id=\"node42\" class=\"node\"><title>n42</title>\\n<polygon fill=\"#ee2c2c\" stroke=\"#ee2c2c\" points=\"383,-1764 309,-1764 309,-1728 383,-1728 383,-1764\"/>\\n<text text-anchor=\"middle\" x=\"346\" y=\"-1749\" font-family=\"Times,serif\" font-size=\"10.00\">nn.ReLU</text>\\n<text text-anchor=\"middle\" x=\"346\" y=\"-1738\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 31</text>\\n</g>\\n<!-- n41&#45;&gt;n42 -->\\n<g id=\"edge46\" class=\"edge\"><title>n41&#45;&gt;n42</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M346,-1799.7C346,-1791.98 346,-1782.71 346,-1774.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"349.5,-1774.1 346,-1764.1 342.5,-1774.1 349.5,-1774.1\"/>\\n</g>\\n<!-- n43 -->\\n<g id=\"node43\" class=\"node\"><title>n43</title>\\n<polygon fill=\"#0000cd\" stroke=\"#0000cd\" points=\"399.25,-1692 130.75,-1692 130.75,-1656 399.25,-1656 399.25,-1692\"/>\\n<text text-anchor=\"middle\" x=\"265\" y=\"-1677\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(256 &#45;&gt; 256, 3x3, 1,1, 1,1) without bias</text>\\n<text text-anchor=\"middle\" x=\"265\" y=\"-1666\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 32</text>\\n</g>\\n<!-- n42&#45;&gt;n43 -->\\n<g id=\"edge47\" class=\"edge\"><title>n42&#45;&gt;n43</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M325.978,-1727.7C315.874,-1718.97 303.464,-1708.24 292.482,-1698.75\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"294.647,-1695.99 284.792,-1692.1 290.07,-1701.29 294.647,-1695.99\"/>\\n</g>\\n<!-- n48 -->\\n<g id=\"node48\" class=\"node\"><title>n48</title>\\n<polygon fill=\"gainsboro\" stroke=\"gainsboro\" points=\"400.25,-1332 323.75,-1332 323.75,-1296 400.25,-1296 400.25,-1332\"/>\\n<text text-anchor=\"middle\" x=\"362\" y=\"-1317\" font-family=\"Times,serif\" font-size=\"10.00\">nn.CAddTable</text>\\n<text text-anchor=\"middle\" x=\"362\" y=\"-1306\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 36</text>\\n</g>\\n<!-- n42&#45;&gt;n48 -->\\n<g id=\"edge53\" class=\"edge\"><title>n42&#45;&gt;n48</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M373.108,-1727.88C385.53,-1718.69 399.376,-1706.31 408,-1692 428.923,-1657.27 428,-1643.54 428,-1603 428,-1603 428,-1603 428,-1457 428,-1412.84 401.432,-1367.56 382.009,-1340.35\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"384.742,-1338.16 375.997,-1332.18 379.103,-1342.31 384.742,-1338.16\"/>\\n</g>\\n<!-- n44 -->\\n<g id=\"node44\" class=\"node\"><title>n44</title>\\n<polygon fill=\"burlywood\" stroke=\"burlywood\" points=\"335.25,-1620 194.75,-1620 194.75,-1584 335.25,-1584 335.25,-1620\"/>\\n<text text-anchor=\"middle\" x=\"265\" y=\"-1605\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialBatchNormalization</text>\\n<text text-anchor=\"middle\" x=\"265\" y=\"-1594\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 33</text>\\n</g>\\n<!-- n43&#45;&gt;n44 -->\\n<g id=\"edge48\" class=\"edge\"><title>n43&#45;&gt;n44</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M265,-1655.7C265,-1647.98 265,-1638.71 265,-1630.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"268.5,-1630.1 265,-1620.1 261.5,-1630.1 268.5,-1630.1\"/>\\n</g>\\n<!-- n45 -->\\n<g id=\"node45\" class=\"node\"><title>n45</title>\\n<polygon fill=\"burlywood\" stroke=\"burlywood\" points=\"302,-1548 228,-1548 228,-1512 302,-1512 302,-1548\"/>\\n<text text-anchor=\"middle\" x=\"265\" y=\"-1533\" font-family=\"Times,serif\" font-size=\"10.00\">nn.ReLU</text>\\n<text text-anchor=\"middle\" x=\"265\" y=\"-1522\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 33</text>\\n</g>\\n<!-- n44&#45;&gt;n45 -->\\n<g id=\"edge49\" class=\"edge\"><title>n44&#45;&gt;n45</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M265,-1583.7C265,-1575.98 265,-1566.71 265,-1558.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"268.5,-1558.1 265,-1548.1 261.5,-1558.1 268.5,-1558.1\"/>\\n</g>\\n<!-- n46 -->\\n<g id=\"node46\" class=\"node\"><title>n46</title>\\n<polygon fill=\"#8b7d6b\" stroke=\"#8b7d6b\" points=\"399.25,-1476 130.75,-1476 130.75,-1440 399.25,-1440 399.25,-1476\"/>\\n<text text-anchor=\"middle\" x=\"265\" y=\"-1461\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(256 &#45;&gt; 256, 3x3, 1,1, 1,1) without bias</text>\\n<text text-anchor=\"middle\" x=\"265\" y=\"-1450\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 34</text>\\n</g>\\n<!-- n45&#45;&gt;n46 -->\\n<g id=\"edge50\" class=\"edge\"><title>n45&#45;&gt;n46</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M265,-1511.7C265,-1503.98 265,-1494.71 265,-1486.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"268.5,-1486.1 265,-1476.1 261.5,-1486.1 268.5,-1486.1\"/>\\n</g>\\n<!-- n47 -->\\n<g id=\"node47\" class=\"node\"><title>n47</title>\\n<polygon fill=\"#bcee68\" stroke=\"#bcee68\" points=\"367.25,-1404 226.75,-1404 226.75,-1368 367.25,-1368 367.25,-1404\"/>\\n<text text-anchor=\"middle\" x=\"297\" y=\"-1389\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialBatchNormalization</text>\\n<text text-anchor=\"middle\" x=\"297\" y=\"-1378\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 35</text>\\n</g>\\n<!-- n46&#45;&gt;n47 -->\\n<g id=\"edge51\" class=\"edge\"><title>n46&#45;&gt;n47</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M272.91,-1439.7C276.553,-1431.73 280.954,-1422.1 284.994,-1413.26\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"288.206,-1414.65 289.181,-1404.1 281.84,-1411.74 288.206,-1414.65\"/>\\n</g>\\n<!-- n47&#45;&gt;n48 -->\\n<g id=\"edge52\" class=\"edge\"><title>n47&#45;&gt;n48</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M313.067,-1367.7C320.939,-1359.22 330.555,-1348.86 339.173,-1339.58\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"341.878,-1341.81 346.117,-1332.1 336.748,-1337.05 341.878,-1341.81\"/>\\n</g>\\n<!-- n49 -->\\n<g id=\"node49\" class=\"node\"><title>n49</title>\\n<polygon fill=\"gainsboro\" stroke=\"gainsboro\" points=\"399,-1260 325,-1260 325,-1224 399,-1224 399,-1260\"/>\\n<text text-anchor=\"middle\" x=\"362\" y=\"-1245\" font-family=\"Times,serif\" font-size=\"10.00\">nn.ReLU</text>\\n<text text-anchor=\"middle\" x=\"362\" y=\"-1234\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 36</text>\\n</g>\\n<!-- n48&#45;&gt;n49 -->\\n<g id=\"edge54\" class=\"edge\"><title>n48&#45;&gt;n49</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M362,-1295.7C362,-1287.98 362,-1278.71 362,-1270.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"365.5,-1270.1 362,-1260.1 358.5,-1270.1 365.5,-1270.1\"/>\\n</g>\\n<!-- n50 -->\\n<g id=\"node50\" class=\"node\"><title>n50</title>\\n<polygon fill=\"#ff7256\" stroke=\"#ff7256\" points=\"415.25,-1188 146.75,-1188 146.75,-1152 415.25,-1152 415.25,-1188\"/>\\n<text text-anchor=\"middle\" x=\"281\" y=\"-1173\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(256 &#45;&gt; 512, 3x3, 2,2, 1,1) without bias</text>\\n<text text-anchor=\"middle\" x=\"281\" y=\"-1162\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 37</text>\\n</g>\\n<!-- n49&#45;&gt;n50 -->\\n<g id=\"edge55\" class=\"edge\"><title>n49&#45;&gt;n50</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M341.978,-1223.7C331.874,-1214.97 319.464,-1204.24 308.482,-1194.75\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"310.647,-1191.99 300.792,-1188.1 306.07,-1197.29 310.647,-1191.99\"/>\\n</g>\\n<!-- n55 -->\\n<g id=\"node55\" class=\"node\"><title>n55</title>\\n<polygon fill=\"dimgrey\" stroke=\"dimgrey\" points=\"609,-1116 357,-1116 357,-1080 609,-1080 609,-1116\"/>\\n<text text-anchor=\"middle\" x=\"483\" y=\"-1101\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(256 &#45;&gt; 512, 1x1, 2,2) without bias</text>\\n<text text-anchor=\"middle\" x=\"483\" y=\"-1090\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 41</text>\\n</g>\\n<!-- n49&#45;&gt;n55 -->\\n<g id=\"edge60\" class=\"edge\"><title>n49&#45;&gt;n55</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M384.874,-1223.98C397.212,-1214.2 412.269,-1201.26 424,-1188 441.342,-1168.4 457.497,-1143.36 468.501,-1124.83\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"471.604,-1126.46 473.619,-1116.06 465.558,-1122.93 471.604,-1126.46\"/>\\n</g>\\n<!-- n51 -->\\n<g id=\"node51\" class=\"node\"><title>n51</title>\\n<polygon fill=\"darksalmon\" stroke=\"darksalmon\" points=\"338.25,-1116 197.75,-1116 197.75,-1080 338.25,-1080 338.25,-1116\"/>\\n<text text-anchor=\"middle\" x=\"268\" y=\"-1101\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialBatchNormalization</text>\\n<text text-anchor=\"middle\" x=\"268\" y=\"-1090\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 38</text>\\n</g>\\n<!-- n50&#45;&gt;n51 -->\\n<g id=\"edge56\" class=\"edge\"><title>n50&#45;&gt;n51</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M277.787,-1151.7C276.354,-1143.98 274.632,-1134.71 273.035,-1126.11\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"276.444,-1125.3 271.177,-1116.1 269.561,-1126.58 276.444,-1125.3\"/>\\n</g>\\n<!-- n52 -->\\n<g id=\"node52\" class=\"node\"><title>n52</title>\\n<polygon fill=\"darksalmon\" stroke=\"darksalmon\" points=\"331,-1044 257,-1044 257,-1008 331,-1008 331,-1044\"/>\\n<text text-anchor=\"middle\" x=\"294\" y=\"-1029\" font-family=\"Times,serif\" font-size=\"10.00\">nn.ReLU</text>\\n<text text-anchor=\"middle\" x=\"294\" y=\"-1018\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 38</text>\\n</g>\\n<!-- n51&#45;&gt;n52 -->\\n<g id=\"edge57\" class=\"edge\"><title>n51&#45;&gt;n52</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M274.427,-1079.7C277.355,-1071.81 280.887,-1062.3 284.14,-1053.55\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"287.446,-1054.7 287.647,-1044.1 280.884,-1052.26 287.446,-1054.7\"/>\\n</g>\\n<!-- n53 -->\\n<g id=\"node53\" class=\"node\"><title>n53</title>\\n<polygon fill=\"#ff7256\" stroke=\"#ff7256\" points=\"441.25,-972 172.75,-972 172.75,-936 441.25,-936 441.25,-972\"/>\\n<text text-anchor=\"middle\" x=\"307\" y=\"-957\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(512 &#45;&gt; 512, 3x3, 1,1, 1,1) without bias</text>\\n<text text-anchor=\"middle\" x=\"307\" y=\"-946\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 39</text>\\n</g>\\n<!-- n52&#45;&gt;n53 -->\\n<g id=\"edge58\" class=\"edge\"><title>n52&#45;&gt;n53</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M297.213,-1007.7C298.646,-999.983 300.368,-990.712 301.965,-982.112\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"305.439,-982.575 303.823,-972.104 298.556,-981.297 305.439,-982.575\"/>\\n</g>\\n<!-- n54 -->\\n<g id=\"node54\" class=\"node\"><title>n54</title>\\n<polygon fill=\"darksalmon\" stroke=\"darksalmon\" points=\"412.25,-900 271.75,-900 271.75,-864 412.25,-864 412.25,-900\"/>\\n<text text-anchor=\"middle\" x=\"342\" y=\"-885\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialBatchNormalization</text>\\n<text text-anchor=\"middle\" x=\"342\" y=\"-874\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 40</text>\\n</g>\\n<!-- n53&#45;&gt;n54 -->\\n<g id=\"edge59\" class=\"edge\"><title>n53&#45;&gt;n54</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M315.652,-935.697C319.636,-927.728 324.45,-918.1 328.868,-909.264\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"332.106,-910.614 333.448,-900.104 325.845,-907.483 332.106,-910.614\"/>\\n</g>\\n<!-- n56 -->\\n<g id=\"node56\" class=\"node\"><title>n56</title>\\n<polygon fill=\"#caff70\" stroke=\"#caff70\" points=\"447.25,-828 370.75,-828 370.75,-792 447.25,-792 447.25,-828\"/>\\n<text text-anchor=\"middle\" x=\"409\" y=\"-813\" font-family=\"Times,serif\" font-size=\"10.00\">nn.CAddTable</text>\\n<text text-anchor=\"middle\" x=\"409\" y=\"-802\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 42</text>\\n</g>\\n<!-- n54&#45;&gt;n56 -->\\n<g id=\"edge61\" class=\"edge\"><title>n54&#45;&gt;n56</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M358.562,-863.697C366.675,-855.22 376.588,-844.864 385.471,-835.583\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"388.243,-837.749 392.629,-828.104 383.186,-832.908 388.243,-837.749\"/>\\n</g>\\n<!-- n55&#45;&gt;n56 -->\\n<g id=\"edge62\" class=\"edge\"><title>n55&#45;&gt;n56</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M480.166,-1079.93C475.111,-1050.37 463.767,-987.844 450,-936 440.881,-901.66 427.705,-862.896 418.769,-837.766\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"422.01,-836.434 415.339,-828.202 415.421,-838.797 422.01,-836.434\"/>\\n</g>\\n<!-- n57 -->\\n<g id=\"node57\" class=\"node\"><title>n57</title>\\n<polygon fill=\"#caff70\" stroke=\"#caff70\" points=\"446,-756 372,-756 372,-720 446,-720 446,-756\"/>\\n<text text-anchor=\"middle\" x=\"409\" y=\"-741\" font-family=\"Times,serif\" font-size=\"10.00\">nn.ReLU</text>\\n<text text-anchor=\"middle\" x=\"409\" y=\"-730\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 42</text>\\n</g>\\n<!-- n56&#45;&gt;n57 -->\\n<g id=\"edge63\" class=\"edge\"><title>n56&#45;&gt;n57</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M409,-791.697C409,-783.983 409,-774.712 409,-766.112\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"412.5,-766.104 409,-756.104 405.5,-766.104 412.5,-766.104\"/>\\n</g>\\n<!-- n58 -->\\n<g id=\"node58\" class=\"node\"><title>n58</title>\\n<polygon fill=\"#caff70\" stroke=\"#caff70\" points=\"462.25,-684 193.75,-684 193.75,-648 462.25,-648 462.25,-684\"/>\\n<text text-anchor=\"middle\" x=\"328\" y=\"-669\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(512 &#45;&gt; 512, 3x3, 1,1, 1,1) without bias</text>\\n<text text-anchor=\"middle\" x=\"328\" y=\"-658\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 43</text>\\n</g>\\n<!-- n57&#45;&gt;n58 -->\\n<g id=\"edge64\" class=\"edge\"><title>n57&#45;&gt;n58</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M388.978,-719.697C378.874,-710.965 366.464,-700.24 355.482,-690.75\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"357.647,-687.995 347.792,-684.104 353.07,-693.291 357.647,-687.995\"/>\\n</g>\\n<!-- n63 -->\\n<g id=\"node63\" class=\"node\"><title>n63</title>\\n<polygon fill=\"#8b2323\" stroke=\"#8b2323\" points=\"463.25,-324 386.75,-324 386.75,-288 463.25,-288 463.25,-324\"/>\\n<text text-anchor=\"middle\" x=\"425\" y=\"-309\" font-family=\"Times,serif\" font-size=\"10.00\">nn.CAddTable</text>\\n<text text-anchor=\"middle\" x=\"425\" y=\"-298\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 47</text>\\n</g>\\n<!-- n57&#45;&gt;n63 -->\\n<g id=\"edge70\" class=\"edge\"><title>n57&#45;&gt;n63</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M436.108,-719.878C448.53,-710.691 462.376,-698.312 471,-684 491.923,-649.274 491,-635.542 491,-595 491,-595 491,-595 491,-449 491,-404.845 464.432,-359.558 445.009,-332.35\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"447.742,-330.158 438.997,-324.176 442.103,-334.306 447.742,-330.158\"/>\\n</g>\\n<!-- n59 -->\\n<g id=\"node59\" class=\"node\"><title>n59</title>\\n<polygon fill=\"#9a32cd\" stroke=\"#9a32cd\" points=\"398.25,-612 257.75,-612 257.75,-576 398.25,-576 398.25,-612\"/>\\n<text text-anchor=\"middle\" x=\"328\" y=\"-597\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialBatchNormalization</text>\\n<text text-anchor=\"middle\" x=\"328\" y=\"-586\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 44</text>\\n</g>\\n<!-- n58&#45;&gt;n59 -->\\n<g id=\"edge65\" class=\"edge\"><title>n58&#45;&gt;n59</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M328,-647.697C328,-639.983 328,-630.712 328,-622.112\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"331.5,-622.104 328,-612.104 324.5,-622.104 331.5,-622.104\"/>\\n</g>\\n<!-- n60 -->\\n<g id=\"node60\" class=\"node\"><title>n60</title>\\n<polygon fill=\"#9a32cd\" stroke=\"#9a32cd\" points=\"365,-540 291,-540 291,-504 365,-504 365,-540\"/>\\n<text text-anchor=\"middle\" x=\"328\" y=\"-525\" font-family=\"Times,serif\" font-size=\"10.00\">nn.ReLU</text>\\n<text text-anchor=\"middle\" x=\"328\" y=\"-514\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 44</text>\\n</g>\\n<!-- n59&#45;&gt;n60 -->\\n<g id=\"edge66\" class=\"edge\"><title>n59&#45;&gt;n60</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M328,-575.697C328,-567.983 328,-558.712 328,-550.112\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"331.5,-550.104 328,-540.104 324.5,-550.104 331.5,-550.104\"/>\\n</g>\\n<!-- n61 -->\\n<g id=\"node61\" class=\"node\"><title>n61</title>\\n<polygon fill=\"#bcee68\" stroke=\"#bcee68\" points=\"462.25,-468 193.75,-468 193.75,-432 462.25,-432 462.25,-468\"/>\\n<text text-anchor=\"middle\" x=\"328\" y=\"-453\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialConvolution(512 &#45;&gt; 512, 3x3, 1,1, 1,1) without bias</text>\\n<text text-anchor=\"middle\" x=\"328\" y=\"-442\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 45</text>\\n</g>\\n<!-- n60&#45;&gt;n61 -->\\n<g id=\"edge67\" class=\"edge\"><title>n60&#45;&gt;n61</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M328,-503.697C328,-495.983 328,-486.712 328,-478.112\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"331.5,-478.104 328,-468.104 324.5,-478.104 331.5,-478.104\"/>\\n</g>\\n<!-- n62 -->\\n<g id=\"node62\" class=\"node\"><title>n62</title>\\n<polygon fill=\"cadetblue\" stroke=\"cadetblue\" points=\"430.25,-396 289.75,-396 289.75,-360 430.25,-360 430.25,-396\"/>\\n<text text-anchor=\"middle\" x=\"360\" y=\"-381\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialBatchNormalization</text>\\n<text text-anchor=\"middle\" x=\"360\" y=\"-370\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 46</text>\\n</g>\\n<!-- n61&#45;&gt;n62 -->\\n<g id=\"edge68\" class=\"edge\"><title>n61&#45;&gt;n62</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M335.91,-431.697C339.553,-423.728 343.954,-414.1 347.994,-405.264\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"351.206,-406.654 352.181,-396.104 344.84,-403.744 351.206,-406.654\"/>\\n</g>\\n<!-- n62&#45;&gt;n63 -->\\n<g id=\"edge69\" class=\"edge\"><title>n62&#45;&gt;n63</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M376.067,-359.697C383.939,-351.22 393.555,-340.864 402.173,-331.583\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"404.878,-333.814 409.117,-324.104 399.748,-329.051 404.878,-333.814\"/>\\n</g>\\n<!-- n64 -->\\n<g id=\"node64\" class=\"node\"><title>n64</title>\\n<polygon fill=\"#8b2323\" stroke=\"#8b2323\" points=\"462,-252 388,-252 388,-216 462,-216 462,-252\"/>\\n<text text-anchor=\"middle\" x=\"425\" y=\"-237\" font-family=\"Times,serif\" font-size=\"10.00\">nn.ReLU</text>\\n<text text-anchor=\"middle\" x=\"425\" y=\"-226\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 47</text>\\n</g>\\n<!-- n63&#45;&gt;n64 -->\\n<g id=\"edge71\" class=\"edge\"><title>n63&#45;&gt;n64</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M425,-287.697C425,-279.983 425,-270.712 425,-262.112\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"428.5,-262.104 425,-252.104 421.5,-262.104 428.5,-262.104\"/>\\n</g>\\n<!-- n65 -->\\n<g id=\"node65\" class=\"node\"><title>n65</title>\\n<polygon fill=\"#caff70\" stroke=\"#caff70\" points=\"507,-180 343,-180 343,-144 507,-144 507,-180\"/>\\n<text text-anchor=\"middle\" x=\"425\" y=\"-165\" font-family=\"Times,serif\" font-size=\"10.00\">nn.SpatialAveragePooling(7x7, 1,1)</text>\\n<text text-anchor=\"middle\" x=\"425\" y=\"-154\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 48</text>\\n</g>\\n<!-- n64&#45;&gt;n65 -->\\n<g id=\"edge72\" class=\"edge\"><title>n64&#45;&gt;n65</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M425,-215.697C425,-207.983 425,-198.712 425,-190.112\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"428.5,-190.104 425,-180.104 421.5,-190.104 428.5,-190.104\"/>\\n</g>\\n<!-- n66 -->\\n<g id=\"node66\" class=\"node\"><title>n66</title>\\n<polygon fill=\"#caff70\" stroke=\"#caff70\" points=\"462,-108 388,-108 388,-72 462,-72 462,-108\"/>\\n<text text-anchor=\"middle\" x=\"425\" y=\"-93\" font-family=\"Times,serif\" font-size=\"10.00\">nn.View(512)</text>\\n<text text-anchor=\"middle\" x=\"425\" y=\"-82\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 48</text>\\n</g>\\n<!-- n65&#45;&gt;n66 -->\\n<g id=\"edge73\" class=\"edge\"><title>n65&#45;&gt;n66</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M425,-143.697C425,-135.983 425,-126.712 425,-118.112\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"428.5,-118.104 425,-108.104 421.5,-118.104 428.5,-118.104\"/>\\n</g>\\n<!-- n67 -->\\n<g id=\"node67\" class=\"node\"><title>n67</title>\\n<polygon fill=\"#00bfff\" stroke=\"#00bfff\" points=\"480.5,-36 369.5,-36 369.5,-0 480.5,-0 480.5,-36\"/>\\n<text text-anchor=\"middle\" x=\"425\" y=\"-21\" font-family=\"Times,serif\" font-size=\"10.00\">nn.Linear(512 &#45;&gt; 1000)</text>\\n<text text-anchor=\"middle\" x=\"425\" y=\"-10\" font-family=\"Times,serif\" font-size=\"10.00\">Storage id: 49</text>\\n</g>\\n<!-- n66&#45;&gt;n67 -->\\n<g id=\"edge74\" class=\"edge\"><title>n66&#45;&gt;n67</title>\\n<path fill=\"none\" stroke=\"black\" d=\"M425,-71.6966C425,-63.9827 425,-54.7125 425,-46.1124\"/>\\n<polygon fill=\"black\" stroke=\"black\" points=\"428.5,-46.1043 425,-36.1043 421.5,-46.1044 428.5,-46.1043\"/>\\n</g>\\n</g>\\n</svg>\\n'); // clear any previous plot in window_id     \n",
       "  });\n",
       "</script>\n",
       "<div id=\"b87a236e-39de-4871-c8d4-3339d14ca9fe\"></div>\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "-- optnet needs to propagate through the network to generate the graph\n",
    "-- so we create a dummy input for it\n",
    "local input = torch.randn(1,3,224,224):float()\n",
    "\n",
    "-- we will generate svg image from our graph\n",
    "local svgname = paths.tmpname() .. '.svg'\n",
    "graph.graphvizFile(graphgen(net,input), 'dot', svgname)\n",
    "itorch.svg(svgname)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualize first layer filters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZAAAAGQCAIAAAAP3aGbAAAgAElEQVR4nO2daXSkZ3WtT1V9NWuWSrNa6pZaPbvdk9s2tvGADbYJxoQwEwIEboINl4TLDWEOjoGwgrnkgkNyIUAwhBBmbDMYT9jGU7u73ZPds4bWrJJUqlLNw/2dfc5axU+/a+3nn/eqevXW9311upb3u8/x/e37rxBFrCkDynOHx0FpH9iOysaL9FKn507iUk+9AMr4oxdA+fzeT+mlFtqroARDeVAK4SwokVhZL+VPNqJUDoNw56/vBOVdt71dL/Xc8+dBWZmYwV2tql019uqlehIeKPHudaD89v7/AuW+X31LL1Vbq4ASDOPVq1bxIzd1JfVSjRu7QUkuroJyzej/AOX2Oz+ql9q/YyMo6YU0KMVVvF/zM/jnRGRhCsWarwbK1+/5Eii33vQqvVSpHAClsubDXeVRkSjeLBFpacE9BCIlUH7ww/tA+e6KXklemMLvYHZmAZRg5gwoc9PP6KWaY/2gbL70laD85Wa8y7fdfpteanUpCEp6aRqUwa1toIxuxcVFZPwsPmwzk/iIfueer4Li1wsRQshLExYsQogzsGARQpyBBYsQ4gwsWIQQZ2DBIoQ4g1erFrR65a6rQNnQg9brI4fwoMPE5Gm9VFDQ47zuZlxck2xBg1NE5ruLoFQ60RnNr0czOBAxTPHSWAMo0aW+ursK1tD/FpHtOwZAmQ3h5udncFfVUKuxq/IiKNm0cXeAhRl1REMkWEXTPeQPgbI2nwKlnEdXXkT6+vHWe0HjkgIjQ4aTPbK+A5SJKh5iWEzhPV2ZwMsiImMvoMe/kjXOrwDj55a1mJzHwzGZZbwO/hBevUir8Y99b18UlKbelrq7+uw7PqJFT51oyZXx+X/9y/Eg0RXrX6uXysTwVE1zGc8iiOD98teMcxsL4/gkVyo5UDbuwI88sj2ulzp2CM9kzJ4xvvW4q7qvIISQlwgsWIQQZ2DBIoQ4AwsWIcQZWLAIIc7gHX7wuFavv2wTKG96BWakA2tomX3ndwf0Ur7wLCjRGtpqBnE0I0Qk34eW2cqGJVBSWzFrHU9gvFZEfIkeUPwvYKpTk80bnl0ogkW/HMPX5Hzo77SgSyki0hjuAiVVxOisJp83LJiFFfyAqdPohbX6lTFantdLlS7DG+2F1SdUtHYZu+ruawYlvYx3MH0Br2c6hT6UiKyt4MdJZ1HRbLvEePYa+vHj+OL4kWvKPm1qMSzj9gi6hEsLeOs1LV3GIzox/zwonofX4dlzaOStLE7qpQpFTG73bVIm9fbdIKSTxrO3PK8Mx068NL1b8avUOWw8MMk5XH9mrP7Tzl9YhBBnYMEihDgDCxYhxBlYsAghzsCCRQhxBm86g//bX0SeeOARUFpV09iObgwJ7hoyQm0nx7BfcLiEKTBNrsnIhS0pM2VuCM3E1E4VfIujTSkiDTU0Mmr5zrq7Ovb0b7T4zg//OSjVAjqV2WW0Fytlw83pWL8FlANPYP9lTUvauFb3fXMKlEAV2xOPn8Uw19v+1LBKO6ro9zU1GCkzoKEposXGNnSLEuopmjuNQcVqGds9i0huDd23vOoKreneYXiXO1+B7aq7N+GuIjG0/44+/KJeaukIbn5xyvh+AV7ceM22PnQhI3l0fhtqa6CEMkYaNJbH+9XkRz9aky9Ynp2yS70mfMnIXtxnMGA8DMuzeAcX5+qnQfkLixDiDCxYhBBnYMEihDgDCxYhxBlYsAghzuCNXGRMP53Loft26hw6ShddeSUoV28d0ksFStjWMthSv0pW1FBMESn5cPBn2Yemz1oae1r6q8rGEKnl2nFXxfqdIZcKGHwTkeZe9E/337QflIGL8SIcf+aUXmpu4Sgo7e2JurtK9GIzTBGJqYu8msKo3UW7hkHJVwwDN1/GC9iaw2azoqJphYIRAEwto4db9ePVC8XRj5OAkdorFNFRWs0bfxHILxk+1MGfToDiBXCyb3IBr97qjBESLKjkYCZfP+HY0mt4domODaBMnUYbNNqGvvxayLgIwTU0f3v66z/tjXHDC25qx+eqrVM7y3hPU0vGIzo3oyxOf/08L39hEUKcgQWLEOIMLFiEEGdgwSKEOAMLFiHEGViwCCHO4Em7Efo9qxzb5XF09CNt2BD5lldfp5fq6MVzBs+ePFx3Wx3KiBWRjin0QX1BbLnrX+sHJRjGwxAiUhvD8xAd80ZyG2jZsU+LrcM7QIkVMOq8WsTzCk09Rqp5YQYnho4Mb627q2wcz0yIyPar8DzEL793AhRfDVskxzPGJMtT4xiCrfbjJdVDaHMZo510agmPyyzN4UGHSgkfvnCjYYp39OBhi2K4vil+7qBxMGX8GO4hs4inatQMU5Egmvci0t2Dra8b2ur/Jnj0gDomIpJOqWfGNwTK5iHsa7wuYRy2SHSoOb4XnsIX7cGzOE3WpNhED35NGuP4eCycwqMVJ48Y82szaXxZ70j9BgT8hUUIcQYWLEKIM7BgEUKcgQWLEOIMLFiEEGfwpubntLpxNzpfBx/6NSjlpPK5rOq3ax+2/d2xY7DutoKzRvCysxE9weYaxkEbVjAoWzBi1NK8hOHPFjUXVuP5jPGneQ8tst5hjALXymiINDaqiK/Iq3bfgIs3bqu7q2rXWS1e+Wa8gGs1bBV9+EE0p1YLxkTPsSR+wJE9eCM04+NJLXZH8QIuz2JP4dVVdJ2Cqj2xiDQqP661wWh/DJRLhrUXblCGo25ZXMM72JQwdjU0ir2Vy/76LZIv2fkKLT7w4E9AWQvhX1zJrgPlqt0YaBeRdsHbupg5WHdXNcsGbe/HWx/w8H6dPIT3dOy0MRe5qw+/JuuHjD4FAH9hEUKcgQWLEOIMLFiEEGdgwSKEOAMLFiHEGXw1NRmREEJemvAXFiHEGViwCCHOwIJFCHEGFixCiDOwYBFCnMF79JHfaHVZtS1sbcRelJFiDJQXfj+jl4p0YcKrdzumwK665GpQ3v7Xb9ZLSRnLq1fBAGB4DbtT5v24TxFZjqqsYgg/88++9BVQPvzB2/VS+UWclDmzgP0Ve7ZiHC8UMbpophawd2i8ETf/5S99GZQPvP8v9FLVPE4M9dRHLlUxm7awiLkwESkG8Na3N+Mbv3H3F0H557u/ZuzKh21IW5uxTWh2Ff/c5Bkjida/pRuUWhU/8nvf8+eg/Pz++/RSXlQN/jyPS82cx2f7yltG9VK5Gl7AmWl8PN54y2tAueFO41ql1azfRB477u4uHgGls4rdU0UkGcAAYDaAPWM/d9fnQfmPv32dXmo+jkHgbAAf0ZUMXs8NcaPrb38ew8htxWlQLvv8r0DhLyxCiDOwYBFCnIEFixDiDCxYhBBnYMEihDgDCxYhxBm8gvK/RaRYxkR0JIHGZEsBRyoefeBxvdSlb74YlNZe45wBbitr+KAtOWyIHMri5vuXsZVtVVARkZPNuPlsS/3aXVwyhoMef+YcKM1bcHbs0O5eUMaOo4MrIoEQ7qFcMSabAl7YOCERCOOVCQXQX19eQpt8dcHoipur4Pq5TP3AvM96Sc2Hrn9bL/YUPncceyuvLBkXYWsPHpdJzSzU3VUgaGyroRNPaWTH8UaPHcTjAje+wbrscbzs2XGj6zQQXTN2tSGFByn2rz0Pyssr+I1LhYwuwwcC2PF8JlK/yfXJBD60IjIVvQiUyTW8EXMBPEDTEza+8iPBAVAGq+pYg3oXf2ERQpyBBYsQ4gwsWIQQZ2DBIoQ4AwsWIcQZvKoYxlDFh2ZHex9OHg0tokty6gAO5hSRq/7sElC0KaMpqXCmiJRq6HMN5DCxvGUCraKIz1iqWoqAcs7KSAOzE4bjk46hxfOOv1VzMSvomzz+g+N6qa52NFz8USONDNTEGAHb1IZTKkc34x6aevAj//x7L+qljjyDczFn83j1rF0Z1l6sEffQEMGI7+9/fhiUq9+wWS/VNYip6dnzhh0MlANoNItIpB2f5NZuvBETB3HesFc1Lnu0BT+gP1D/N8GGnOFv7sxhNvjK4jP4xgZ8zYPRfXqpGT9+nEy4/tjgpVXjLk/50NU9W0NPPN2JufTlnPEwjE2hclk73lMNf2ERQpyBBYsQ4gwsWIQQZ2DBIoQ4AwsWIcQZvJoYvkk4hkZeMIS2QlDZQIW0YWnF1FKeWkqTVy6eiFTV8u1p7Cm8ex6NvLbUvF7Kl1fprQCaHZpsxDA7bv7g1aBsuxRdrTtuwR64a2eMVrbruzEIlsrUd76mXzRecyaHYmEVvbD3f/4qUHbu69JLffXvfg/KM8/U9y6r1nPV0ooRzjPPYfvj08fQMvvMvX+il1pZPglKbjFbd1e+IhrNItLcjm5poAP/Ic/NYu5y7hw+eyIyslOZ6fH6zle3b1GLnQHMEhZa8fH7aRzDdvd4r9VLzXmYCuz3ztbdVbgJjWYRmS2j514awFbRy1l8jMsF44dRcxPmGQ/nT9XdFX9hEUKcgQWLEOIMLFiEEGdgwSKEOAMLFiHEGTy/6nIpImXVL3JtFS2Y4iraH6N7sYWgiBSVmZhOKjcH82RSthJYywF0H6YFTZ+VAGa+4gHDPApXUYwWDP8IGNhk+CYD3WgMfetDPwTl2M+PgXLzn1yhl/LH0D/1++s7qgUrtTc1iW98+iG0YBYW8S6/5yNb9VIvuwZnZ1b9mC7UhExzrIy7evaxM6C86p07QekaNBq9Pv3wLChBr34XzZx1k/MFNDS9ZnRUaxG8Vktzxq6qRTTR/H4jq4vUDEd1MtgJyvFYDyj3RTC1+qhcqZfqDoyB0uGbrLspnxqgKyK1Et563yr68l2NCVAmx40Ry34ffsczJaP7Mb6r7isIIeQlAgsWIcQZWLAIIc7AgkUIcQYWLEKIM7BgEUKcwSvnDfOy6kefdWUKw7TBJWxPPKjGhYpIMY9vXJvDNrL6WIOKT4qIrKoBjQer/aDES7jz4ayRLJ2NYyR7XnXF1QQCxuzM88/h+mefQc/4susxDt0ybHRknk2jQ+yzTwf8N7rWG//qBGLo8ZeqQ6Dc+5/Y0jqXxiiyiOy/DBPRwbLh6AM1n3GtlhcxNV0J4/3afz3a+WMHcVStiCQncanmsHHoBAgGjFB9fgUf0dYIbn7jVX2g5LL4/ItIIYkfJ+avfwdXfcaztxTHJ2SyAb8n5zw8QNBSxmM9IrLN/wIoQ3/AsYZY2uh4viGIPQKyWUyhZ1fxxEmn33hEWz08Y5KIGHlygL+wCCHOwIJFCHEGFixCiDOwYBFCnIEFixDiDJ7PcjF0ILqwgmZiPo1KY79hwVQCmBpNrdSPGdd8hg/lhTFZmmvBMO14Ag2X6ho2txWRtUbcVb5BNU1WREOGmzM/kQSlXe2hNYFXZiVn+HHVikrhRowpsEBC9R0WkUgcHdVgGEPaySQ6iWW1ARE5dRIjqV5940sqVWOpTBr9uN7hBlAa4ui+JS8Y+eFwCN8YVh9ZY0bJkxMY6K0oC7d7Dw4HTZeMUH1yHH3eat4wE4F81LigeR9+v6phNEZ7lScY9U3opXb7D4HSUzXMRKC7C+1aEZldxA+YTo/hHjL40CasCx+t4AXcsr6+z8tfWIQQZ2DBIoQ4AwsWIcQZWLAIIc7AgkUIcQZfrVbfHSOEkJcC/IVFCHEGFixCiDOwYBFCnIEFixDiDCxYhBBn8P7pa/+i1VMHzoKSaMEWnes2YvCncxN2ZRSRmSUMao2fxoaEd/zNx0D534/M66VOn8Pehi8cOQ9K13oM1i2vLuml8ksYMNyyCed3/uwvXg7Kv9/zLb1UuYaRKD2btlDGYF2paAbfUIxHcKl3vvmdoHzmzjv1QqWcCmP6MJGX6MSE4/K0kY+riUrWhXFXn/7034DyiU98Wi9VVdNqKzE1srSAtnWb1X52KYnBOi+KS915x2dA+cLff14vFVrCT+0l8ZkJVTEfV2s0AoDe7ktAyfUPg3L7q/BJ+8aCXknue+xBUIpz06DMj0/hBuJG3LVcwihfqIqhzsf//kOgvOban+mlBlvaQIl7eL9GL7oId9WMOVYRaQjiG8slVN7wfnzS+AuLEOIMLFiEEGdgwSKEOAMLFiHEGViwCCHOwIJFCHEGb3nCOEBw/jAea/Cvx5GNI9twpGLPeqOnassGHMPZoCZEGhs4gbariJydx/MQS0s46/HkWTyv0NuOOxcRXw3XT45hp2NNrWzsqjWBbnpTHx4B8XvY7ja3hn2HRSSziqcfKn9Ag12pGP/qhNRFzi5jU+amfjTmq2Icazh/EF335raBupvyK7dbRDz1oZcyuPlYg+oTHTKOgBSz+HGizfXbSUdDRq/esurVW57GAwS54zh5NDiDrxGRwMiToLRffxm+SB1rOP70I3qpx+7/FSiFOWx/XMrjtYr4jfm12SKeaCnqJuiK2cUZLXYG8Ram1AEafzM+e0GrB3RHC56LChjNzNfh4volhBDy0oQFixDiDCxYhBBnYMEihDgDCxYhxBm8bNbwoVZm0TdJhtGUKZbx//w3JNA3FBG/smDibfVdwoZ1g1pc34RDHOfOoFPT24aOUnPcmlKZQmujI45TRTWri9YgzjL6Mp09eB1612EkO9pkOF8rKXQhV+fxI2tqReMOVgtomY0fxRuxaf9GUG58/5Be6j8+hynczISRJwcKecNw3NCP8fhHvnsAlPd8+jpQjvz6sF7K78Pkds1vjkn971j5fG8UjezgdrRBwwfHQCk9iIagiCyexzRy/scP4YvuRmFl0piqW1QjS+NRjBCHImj/JdOLeqlQCL8Cja2GdQ6sZo0vTtWPz23Fhzda34h4BCPTIhJRc5ELufqeOH9hEUKcgQWLEOIMLFiEEGdgwSKEOAMLFiHEGbxA1bCrlqcwYJiIoxdWyWOPWn8Vk00i0tiIwbp4B1pmmlOnDmmxuIBZwnAQd1VcwqxTIZPSS2Uy2Ks3uhMNF83kqWUtjh1By6y0iptv68aP3DmIl0VEOjc24BtVLPESdPakVlHdkEUaY2iiZfDiycdfjz1wfzz+Pr3U1X+8C5RHfoC9qjXFqhGW7ExgfOz4k+dA2XvDB3BXX/iJXmrPfszorZYMrw1fE8VQp4hITw8Iwc1oJvZctxuUyFX9eqXAL58CZfGZU3V3la0YuyqtovsW6cHHIxzEh7av1Yh5VufxaZdo/dzl9NkLWlxqwz0k+vHqVfxYB/x+w/7THzqfw3bqIniR+QuLEOIMLFiEEGdgwSKEOAMLFiHEGViwCCHO4EVVe0ARCakOgcUyGg0XzoyBklvep5fy1dCFDPrqOxTxouF8pdLYX7Qzjot7QRwkuTpvpKtiCT1vUjkpir3XbNJiazs6qicfx4Tj84+iUrVMtP6LMHK1YU83KO94B77LsyZ6xjvRqXndbdtAOfOXeGU++kc/1Eu9+7N7QWloQatIE7R6e2bU/NpYB9qgNXVlclnjWkUa0CBO5QyTGvByhjkbG9wOylIA26WOKZd80OpkmxjeCkrwYSMICUSajevpb0Rn2ad+Xnh+/PIONKE7LCLzZTTkwk0YnzQ20GJUhuYB9E9bWnHObf8G7DxcWrFGxZbwGxfw6v9+4i8sQogzsGARQpyBBYsQ4gwsWIQQZ2DBIoQ4AwsWIcQZvKYuY/rp4LYRUFr0QYcCusgr08YsUl8WjeR83nCpga1D7VpcmsY9lAIYPI54eNChvd8weiWHczFD1d66u/JFjEGqg1fg+huv3gHKTSWcnZmbMYZGrixg3jVXrH/YolIz/tWZT2ESeP0gXr13fQzPKzz2mzN6qReexNuqjo4YxKLGAYKx8zib89Lr8RzAc7/CZHXvkOHBrxXw4Ev98LpI8r4HDHEe4/GB9Ti/M9eKHbSLq/gIiUgsiQ9kIGI0FwBmV40vjj+OX5N4C1734hqGimfTxpdrUTXRbvfVb0bc14dHakQkEsQTEispPBzTlsQ9BMrGiZNsHJ/bdM44oAPwFxYhxBlYsAghzsCCRQhxBhYsQogzsGARQpzBi8SN4GXPJpxjWk1j99JwCDPMKeW2iEiwhJbBckb1RkUPTXoThtkxPIjWhk+ZaIEwLt4aMYryygwGO7taMd2qKYvRyrbmR9HfjNHZ5kZMAic24FBMEenMYAfh5WndNBbx/MbOy1U0zWbncKlqE5oy219mWMa5HBqaUdV/WeOrGI7PcgZ7TA9twgjxqcM4i7RnveESZtWgVi9c3ydct2qMgJ144llQSpMYaK+uwz34o8afq6i+v7lc/fGuuawxdDartjrnqaXURS434rMnIhXB73hosP4g1ZDfeI1vcQWlKn4r8xHcQ3XFakDQhMZrLaC/qhfBf/MXFiHEGViwCCHOwIJFCHEGFixCiDOwYBFCnMFXqxm2AiGEvAThLyxCiDOwYBFCnIEFixDiDCxYhBBnYMEihDiD76Ybb9XqDW+4BJTEEOaYymksdovjRgAwm8IunRU1LvFTf/NJUL72ma/ppSoNqv2mD6NbBb/ap89ofenlMLUnxTQIf/2Jd4Hygds/oJcq+dBm9ZdRiTdi7mxpyTBn443YnbKWw6DWP33tblD+62Yj1NaqOsRuGFwPSjGMr8kWjTtohOEa+kF42Sd+C8pdT39QL+XPDIHiS2KosyWl/hHNGv0qy10qZdmJyruvvQOUOz79Gb2UT/3BQqEESsCHFzkUM1qJFtXI0kAJb/Qn7vg7UN5359/rpWazmDadKuPI0sEg5g13BXHYsIjsC2IT10gOQ51X3PEkKF/84l16qZAfP3XYU09RAR9as3GuTyVzg1Fc/EN/9WFQ+AuLEOIMLFiEEGdgwSKEOAMLFiHEGViwCCHOwIJFCHEG76a3Xq7VV71lFF/nQ8fx0IM4SHImbXR6rea1x2+cM8B3odktIlKIqxMSUTS8yyGlWOHudBr7O0dyRvtj3JUY/nogjP2Cs0lsI9vd3wbKzILRTjpSwyvj+esPUr1o+8VaLGfQ428Z2ADKUhZPcnQHjH7ZKxm8reFoY91dldeMvsbFC9h1NzaOV6ZyAc37ppzxNGQS+AGX1+OUVrlW7apk3OW8suFrVbTqfRU83FFYMx4GPc01Eqs/HNS8msk4PqLnp7GB9QkPx9A+6eEQXxF5eSN+nbc3vAjKFfptAeObc2EaL/LCMk6BbenAx3hogzGlOKrm8eaMOcUIf2ERQpyBBYsQ4gwsWIQQZ2DBIoQ4AwsWIcQZvPWbjYmeAVXIThxAL+Der58CJTWLvoaIJLrQJQm3WvbKf6ccMippoaEASqgFvbZoywIoOSv8nJrcgn8x2l53VzWrvkfVGNOVjBotmUH/Y3BHn17qmXtfAOXy/QN1d5VLq9mWImHBbZVWMqBUUjgh1Z8w/Lhcag6UsrIXNQ0+wx3LrOD6DfPoG647iV5Yx5RhlS6OoJKP4mU3dpUwRsA2x/AJSWXwA64so1KtWI9xDc3Eqq/+gN7uZTTcRaStHYPoF8I4efRothtfUzHM2efVZNOamjesiTcb1+pVV+wHZXQvhurzKqz+7AP4YIvIuWNoOPqk/tBZ/sIihDgDCxYhxBlYsAghzsCCRQhxBhYsQogzeJVVtIpE5OTv0Ri6/9vnQXn6R1OgJPqMxFCb6kUcDmH/WU2lhoagiITi6HO19mLv1871B0FJV40OwqG4soEmttfdlU9HxURKVfw4XR3o1Dzz0HOgfOqX2BVaRL76oe+DcsuNRjQMyGWMzFcwhDZQOoP/OGmXq5AxzKOaapvsb8S4nyZQNEy0BuVntiXRROs9jW/cOmPYfy/68LIn19X3mNZS1q48dLd37tmEe7gUk5gBwxKXM4cvgHLh2HzdXa0T1e5ZpGVNNTuO4Ve1lEL/9GAJ7W8RuaD7htd66u5qdcaI9v3nb58C5ZkH/h3fuIC72n0VhhlFZNc1aC/G6j9W/IVFCHEHFixCiDOwYBFCnIEFixDiDCxYhBBn8M4cMnJhq7NoEJx4FCcvVksYFuvoM1J7XZsxkZQYrt9xtFQyPJiqSgUGPTQToxH0N8VIREl0aRaUQtQI9wEByyVcy6EnMtqHabjJH2HCsbHN+IDBZozalbL183HreowUZDCI1mEggh5ryI/Xs1w2PuBSFa3DWtZol4pLlQwv2B/GvrVl1dZyLYpvnG8z4niVduyDWgvUz6hWs0YsceYCum+PfP15UE49j554a4fRnXVwNxrEfeuU9fU2FFrjxl2+thUDhntzaEFe2XsGlJ+ujemlpmIY2Ezp4bGKXMZw6oe34bN985t3g9I9iPHkpSQOfBWRM8fwA+ZS9c8P8BcWIcQZWLAIIc7AgkUIcQYWLEKIM7BgEUKcgQWLEOIMXqBqREZDysnu70cbPtGJ8x8vfmWHXqpvC64favgDBqlaYzhX53EP5Zpyqcuq7W+zCu+KFGb68S+uYP9ZTcVnuP61ChrzVeWv9wygr3zg3iN6qX3X4VzMmZnFurtaKhgfsJZCIznaje10W9pwV82txmWvZfE8xKadO+vuqhY3dpVrUnnaPswGn9uP50vmVq3pp71465cGjTm++C7LNw814T/b+/8Iw8/Xvh6n1eZShuu/vKY+oGc8M8AvljBZLSJSxoNEW4J4tGJzAUP1N/qMHPXZ2hAqFeOrCnSPGsdlvCA+22cunAblwPOqn4IaQysijU144CPYVj/9zF9YhBBnYMEihDgDCxYhxBlYsAghzsCCRQhxBi9n/Q/8XBUtnkgXvmzd8CAog3uNFsk1D22L5eX6IVV/yYgsl1cxG1z0Y5fVRdU5NxIy5ozmUyrqnKk/SNUvVjPiAAZ6kyn0d4Z34rV64YkJvdSGTWjkrc7iUppU2YjORhvR9JzL4JXpa8SdryQxXisi/iCmfENe/Qa7pQbjLleaMTVdG0TnN6ni66mKYbRlmvC5ynXXj87GmoylgjG8Dj7l85bUYxxvM744gRL+AsijxWrwRHCfFr0ARr6LseOg7AqMgRJZNPoatxTwsg9HrAbP/53UnGE46tblfg9N/+Ygnh/wRysAlsMAACAASURBVI3LXi7hc1sqG84yLlX3FYQQ8hKBBYsQ4gwsWIQQZ2DBIoQ4AwsWIcQZfLWa4XwRQshLEP7CIoQ4AwsWIcQZWLAIIc7AgkUIcQYWLEKIM3hf+sI/GbIabxmNYWmr4RxVCXhG+SuVUSxmsYHkX33gNlDu/tYX9VJrZ06AcmnfKChXXnoDKN974pRe6sAc5uPat+4C5WNvxmjkvZ+8RS+14GGbxMkANpB83r8ZlIr1T0VHCZODfeUkKH/3mY+B8s//+nW9VDiMSbfMIkbDGhswQ+fFjXxcsYK5s1wZ33j7u98Oyt3ff4deqjeGSbd4FVN7q2Uc1fmjgyf1Ut03vQeUQh6brH71FfhcvfjQAb2Ur4QP5Mwq3oi4YGo1v2rMZPV8OKA32oA3+uI3vAKUd733L/VS4SC+0V/DqF0lhVOQSz4jjleNYbC0GsJY7re/8mVQ3vvj+4yllmdAeeph7HraEMHrefaMEYmNDAyDsjBzFpT8b38ACn9hEUKcgQWLEOIMLFiEEGdgwSKEOAMLFiHEGViwCCHO4EVb0KIWES+KYqRBzUON47mGcNBYai2D7m922WjpCywtL2jx/nu+BcpsI7YeXl/AhshXDGzTSzUF8WX5kO5ZjMcaLviMHtBnQ0OgjMd2gHKish2UloIxITVRfhKUrprRphaoFC0nW/1LVCvj/cpn8UWbNxnTZM9P4qzTcrX+KNxNVex0LCIb43hC4umDaMx/8z8fA2WqBZsFi8jIZbh+Z+F83V3NzRgzWQtZ7K1cax7APcxgz+6hBB6pEZFUAR/I+VU8bIETWUUqVgOCUgW36vnxDhYzeOvbW4xbU23Gzc+sGp2UgY4wNuwWEYnhMZQrr34lKA1qXG0x96Jeab6Cm+9YjwcdNPyFRQhxBhYsQogzsGARQpyBBYsQ4gwsWIQQZ/DSyaxWCzk1WjKPnkU+i9ZGyBqXGI2jUirXH6Q6Mqq9FOkbuQiUIw/iaMnjvz0CSsuwYTgu+LtAadqA0x8105GEFue8daCMFzrwNX6cPNpomWgj1XFQNlQwDmrgGZe9JiiWfXhPjz08Dcrw5n691O4rMCj7u4dO192Uv9KmxZk0Pg3f/+kZUJ54Di2tbJPxz+rb862gNBUn6+4qHjdsUF8Qjexw20ZQDj55DpQXX8RnT0SiTbj5PfuNZwYIhowP6FPx9XALXr0LR3Hw7eJJtHRF5JJb94BSMy4Dki/iRxaRWgbtRXXxpFDAj7Nnz1a91PksOvXTs0YtAvgLixDiDCxYhBBnYMEihDgDCxYhxBlYsAghzuB5PvQ1RCSdw7jf9Av4P/BPPYvuW3reaITaPYwdhAcu+gN8k4SR2nvbZz4MyucufA6UpTY0p1qtprHJs+hzhVqwB65mJoDde0VkptoCyqIPLUh/KQXKrvIzeqlrCxija8xhLEsTtfoa+/34qXdchk7lr799DJS73nu/Xuof770VlG0b0aHTnDha0OKZ89hBeOI8Ph5DUYywVcJG+LS7iEaXb8640cDKsvGIFgPYL/vMYYwl/uAb2Kt30zZ0EkXkwhyanuuG3qBehR/Zr8J3IpJew5TlhisxvRhOtIPy/fd9Qy8Vuf8wKMN78WHQzE8ZvnmXynWuJNH0Pz+J7vOp2Rf0UtkcfikCVhgZ4C8sQogzsGARQpyBBYsQ4gwsWIQQZ2DBIoQ4g9fej0kxEWkbxtTS6KVo7V36JuzKWMwYIcF8Bi0enxF9Qx5+FiOBIvL6d74clJ5XXwPKb6fQW2nvw/6fItLVgEbG5LIxbxUoV62wZA3/4lAFB74OlvHP/VnJmH66r4Ruzhk/zmTV+K0sYb6I96JzBPtV3vHTd4LyptF/0Ev92+dx+Ojr/hLnwmqSBfTCROTp59CcbV3Bl/X0oEsY7lMze0XCAfzUZ85jElOzvIz5NRFZvw4ts1wNE3mRNAbr8mNG8K0XU6SS8K/V3VWsbPibqQsZUE4+iVfvDV/AO5heNnb1wvcfB6Vpvv6uwmHDC/YX8Oucy2JL1UIBn72uHiNLGA1iGPnJ5/CLY2yg7isIIeQlAgsWIcQZWLAIIc7AgkUIcQYWLEKIM7BgEUKcwSsVjOBlLI6FrGMQR2AmRtD1bGgyhjguz6DPujRe31LNBIzRkj9/+AlQhl/+ClCe+/VDoHzvjOGVxgvY2LU1pno5K9rKeIJBRBqC2DQ2XsVLujt0CJRtVeMURSaCHv/Z4BZQRtS7qmnjWhXyeAcf+81RUG59BwbFP/ZvmHMWkd/9BA9bPP/sFChXXIXvSqeNQwbb+jEW7iXwuapF0DgPJdDgFxHPjzlqv1c/KN66E/+ciBRWccznnn141ucL/3ojvqto/LnECPbLbmusv6tyzUh3t8UwZnzoh0+D8gv1jdt0iXGAoJrCR7Q4g+dsNB0J41hDTwzzyW1t+JErcWwHMDaPD4yIDA7huaihy3fV3RV/YRFCnIEFixDiDCxYhBBnYMEihDgDCxYhxBm8/JoahCiyvITi1DlsiBx6BotdUxv6GiISDSlfpmK4WsDlu43E8mIWA6jZMvpHL3vtXlCCy9iJVUTGHkfDZWIMZ1JqWqrGtSr78OO0BbAPbzyHV2++ihFfEZnzMIV7IoJ9eF+p3uW3orMRD03PVWXX/v5e9C6Htxg9oFeWcLpqsYJTdTWvvslohD357BgosQheh3OzOGJ283bDfX71VWjknbLG3AIRZWCJSDG7CMp0Aa9VtRX3MDqwSS81PY3h6px6PDTVmNGmvK0BO19vy+IHnLzvIC61sKqX8oXwYQi31J+kunU/dmQWkXMH0V5cWMT7VfbwIrd0GAcDQnEsF1MX9BTYbfDf/IVFCHEGFixCiDOwYBFCnIEFixDiDCxYhBBn8NVq9V0MQgh5KcBfWIQQZ2DBIoQ4AwsWIcQZWLAIIc7AgkUIcQZv5gtGQ85UCdNVhRoOszz5Iib7kumkXqq5HxNDxVwjKH/2b29F5Zvf0EsVMtjwsyGMIawgBrBksYx9NUUkHVSxrArmrY695xJQbnvfbXqpSBR7MFZLOEjSU90jA358l4iU8phVzBfRw/3qV/4vKB//+Uf1UvEURjgrNVSqq/jPVbVmzGStrsMgmBfHCOcnX/GPoHzjG1/TS/k8vDJzCxi+K+fxFsYajCyh58fN+wU3//4P3A7KRz73Tb2UfwIHlKaXMB/XHsW+ml39xozb2uAgKP2XYjPW16iY7F133WUsVcKIaEB95KqHz5U/ZJj+eqlqCdOgH/rwh0F595+9Sy/VEG8AJZnC5GxHJ2YJ56Zw2KqI9PZhjHQuiVnCe779HVD4C4sQ4gwsWIQQZ2DBIoQ4AwsWIcQZWLAIIc7AgkUIcQZv/OhZrV5665Wg/PbRJ0EZU4cYGitocIpI0IeHGAqz9Yc4Jss4cVNEqiE8oLCWwymVRUGl1oB9h0VkuYwtYlfK7XV31dxqDFuNqf62NdWyuFZEqz63ZszXLBXQk/aC9ZsRB9IdWozO4wGCYB4t8GoOJ2Xm9KkQkWwA+zvX8JYaTJ42zrj0deFfvPmV+0FJ5bCl9ZFDY3qpom66G8STN5qOPmNkadiP12r6OJ71eeKxH4PijxiXvXMHdui+QQ9J3YHTfytV4yxCJILPVQHHocryEn7j2nqxc7SIeCG8rbWacaoGiDUYt3lxFQ8oxOJ4XCZYwZ03d+BhCBE5fwYrz6atODZYw19YhBBnYMEihDgDCxYhxBlYsAghzsCCRQhxBi/ZghaJiJxaQ6Nrroq2gm8AQ5xPPPEjvVRnGic7tnfU9wKWK2r8qogvgF5DNIIWTCaL7ls+a4x3XS7hxymG6vtxC5PGTNbsIjo1YyfQSZkZx9fE2o1ddW9sAmXdqDH4U+3AcL5iy+jUxGbRP6qu4vUstRgTPavKdwr/Ad5lpNvInP/ipy+A8rm//jkoN79xHyhX3ILTNEUkE8Z7kc0YY26BeLdhGUc9vIB9oxhsDqTVVcgbQfFcDb9NxT/AE/dVjeuZU9/Bc+M463f0Mhx8O7rJGNA7fQK7BqysGiY1sLRiJJZb29A63DA8DMqdH8cs9x1f/Jhe6nwTZs6PHcHHQ8NfWIQQZ2DBIoQ4AwsWIcQZWLAIIc7AgkUIcQbv0DnsdCwij82Pg9LUiOG+WASDdV1t6/VSK3NobaSmj9bdVr5qVFKfH/9ioYjxsTU1F9YfxPyaiNR8yvSpGqYPULHSarUoRrz69uJfvOp2NEY3XoxhRhHxVXDzqYX6bk6gYHQQ7phFcfSsyhLOoTl1ot+Ij2U70fkKBAxnGdh4kZG7/JMPvg2UNWyKK/fc8WtQfvP9Z/RSoxejHRZtrJ8lLKbRLxORRBjd0rUo2rWLpRdBCacNa0/7tQd/+whKn/xTVMLGztcq2Id676uw//I1t+7CP/czw5R89l78OncNGR4u0NdnhCWXLuCuvvpj7Dp9PP0IKFu8y/VSr3vr60DZe/n2urviLyxCiDOwYBFCnIEFixDiDCxYhBBnYMEihDiDV17EmZEikggOgDKXPAZKoAmdFN/483qp8+emQNm6ZU/dbYXE6IiYVB0Xc3mMHJar2FyxGjC6aBaVJ5gPGF4b0LvRMBy7BjHPNXgxNi/t3oKRwOVxdFtE5NiDeK0WxrN1d+WrGh5TPIOfet1ZtPaaJtGCzJUNU3JxBP9VS4vqqao4/HO01UTk0R89C8rGy/tBec2H8fFIjqkbLzJ2fBaUSrG+oxr2jIacm7ahOTWvvMSaavdZmTVs5bUArr+0YviSQCRodBxt6MKlutrwKfrNl4+D8pUP3q+Xuvq1O0HZsMN4koFz52a0ODiK9+vPd70FlO0+nB37qa9+XC+1qloWP3b/0/gilUHkLyxCiDOwYBFCnIEFixDiDCxYhBBnYMEihDgDCxYhxBm8F6asgZc+NMUzJcx11vJ4HmI5NaaX8kcXQWmO6BmYSKFiHDKoqJmsgQi+zIth5jaZNNzunGAv4KCah6rxV4zQb3YRxYM/Ow/K/JfQ2x4/jp2jRaRaxqW6hvDsiCbTUNBiugH/KZpvxrBuKYvGfCC+opeqiDpV0GycMwAGR7BzroiUCvgBFw5hH977Dz4BSmuPEcmONaLHX43U/9c35jMehmoVn+3OBF726YFeUAIdxsmbDh8+fr5xPKqiqZSsltOq4fPRh3Gi7bHfTYDymve+TK901Vs2gjIxhS0JNEPrhrR4/KnDoFx6LR5DueFPbgZlLW88Vz//zn2gvO5Nf1x3V/yFRQhxBhYsQogzsGARQpyBBYsQ4gwsWIQQZ/Bm5g2zo6kJ/YjOOBoZZ0/h1MPUIr5LREYHMMP5smFjhihQDRsdhP0BFAMVtPYyNVWCo4bHVErhx6kqw1GTnDCsvYkl9JhSM5hYrlUx3drVaUxIbe7CPYTi9f9FqYWM6OxCOzqAJy7CoHhrJ169pGcYuIFWdCGrcSNPDmSrRrrbp7aa2IB+XLWE96umzFMRKa/hRfb7/4Am1xVj5w1NeC/2bcQvRVxN9l2eN4y2lRLer8FN6+ruqlgw7rIerhpQGelLbkT7r3vYSHefU23Q06v1m1zPLBr+5u5L94Ly1O8OgXLlK9A3fPCHj+mlXvfW14Ny+MjBurviLyxCiDOwYBFCnIEFixDiDCxYhBBnYMEihDiDr6YmjxJCyEsT/sIihDgDCxYhxBlYsAghzsCCRQhxBhYsQogzeP94x2u0OrkOU2bl8Cgo1128G5TfPWk0oqxFsTHp2157NSj7IrjUXR/5F71ULJLCfYZxZOlitA0Ur2IkHG+t4MzXwWVs3rjxc78A5St7jLTajtFNoCz70Hh9bg7brqYbcOcikit3gZJvxmjYt7/3a1D+9n3/Uy+1s/duUJ7Z/0bcw/fOgrLuykv0Up9t+i4o28LXgHLgj/4LlKd+gP0kRWQthWHMagcmMTfEMLU3sHW9XuqBZ8dA+epdXwTl/sefwdfcbTyi9/8MN7/ltX8KipdX4buVM3qpmbPYQLXsx7Dkd/99Gyif/frn9VJbL94ASnsHTj/NF7AtaS5ttJ+dODcPyuQFHEP7D3/9GVBC+4yn/dXvfjsobW340C5M4NPua+vTSzUOjeBSIRwJ/OUr3goKf2ERQpyBBYsQ4gwsWIQQZ2DBIoQ4AwsWIcQZWLAIIc7gdW3botXS9Bgov5lEqz4awdGSv3jgiF5q7x5c/4Wz2Px1H/q8Eg0b7mwwgAcUKq04t/KFEDbYzYUMS7UthQ12r/Ch447dZ0X6R9DBFZFwK548KCjrfCKMg1QnTlqmeAbFrTvwzITm/BfxxImIvP3QF0C50PEgKNsvx/MQRy7C0x4i8vWHvwfKrbu76+5qLIlnUERkrVYGpb+MXaGLPmyEvZQ2ZtyeH8P7lakazb6BU6ePavHE0adBGZvBJ7nkS4Ay0I/HL0RkcXIMlFgU3yiCj/tIL36VRGSoEw/o9PTjZQ/48XqmFvGYiIi0BPBr0hqpPzZ486VXazEQwKMVrc24+SkPD1v4SrhPEalksV1165b9dXfFX1iEEGdgwSKEOAMLFiHEGViwCCHOwIJFCHEGb3gj+mUikowMgjKwiAHU44/jdMZs8eV6qfHaECj/6x8eAeVP/x3fNZDHuKaI9DagibYiOJM1EEdzbzGKvoaIvJjD7PEmZQNpBts6tdgyiDbo8SO4z9g8RmcrM4ZN41XRXqlOGvYKsO3ov2rxV2feA8orFzDl/vENaJld2mpMP/2PY78C5blr0KG7S74OyroIhn5FZErQa17Xiq5uYwgd1e42NLlE5OS0evzK9Qf03v/oI1qshXFXoWa8Xxt68TWLS+f1UgtL46D0JtRAVEUubVz2hQnsGuB56KhGfJhPTqeMUbhLCyqSnTXMRKC0gk+jiCxMoFP/0C+exH024I3oaDec+kwWXd1ocwe+qAcD0vyFRQhxBhYsQogzsGARQpyBBYsQ4gwsWIQQZ/DGnjdq1kAf/n/+8g702h6dxCxVd7vhx42dQAdk4Wj9HFNXZkWLIxHsYnzehzm+UhTjSJPKWxGRvmb0BM+soV2lCcWNobPxBhR7E3hJG05i79dYo7GrtgrGsnJGe2ek/bcYGxSRh7ZgUCvUcAyUT0avB+Unv3pOL/XGi9DN8e+q711GQ4YP1dmCEdFiBJdq7kF7cc1DU1JEzh5+AZT+IcOKAq794xu0eO733wTFl7sASjW9DErqgpGc7VZ22OoCPrSaomESysxJbDScQa9PpFwCoZA3TP/VFbzsXri+o7o0Z7STjsfQw020o4fb09IASsUzHpjOGH552324uIa/sAghzsCCRQhxBhYsQogzsGARQpyBBYsQ4gxez9WGZ5dtwv/zXzs6CUryLPoR4VFjOGhXBBtINnXrEBaGEKsZw7NriqHX0F5EM7HRj9ZGReXXRMQfx10tFup7l/k1w7uMhrC15oZmtNWSoz2g+MpG5iuTRJckmTGGWQL/9NRjWrxt9y5Qzl13AJTKFzEsdi59UC91YR06X5OPo233VpUiPVszPKblSczH+WbRKkquoTtWLhmOaiaPD8PmDiPpCVTnDMNxbAnFHhVzW1k7CcrkWfxGiMgNr74OlIYevPXGBl7EjywiqYkpUNJpfDz86gdHWI2hFZGGZvw2hZrwsmu6+43vYJvycIeUC9/Zjb1S54tGoLJvAA8V9PXU72TLX1iEEGdgwSKEOAMLFiHEGViwCCHOwIJFCHEGFixCiDN4vb0btDp/AZOlUkDPeNsNaFH3thrm5Ym5F0E5OYz9ZzXLaWNkabER/XUviaHivqZZUDY3YZBVRFrK06AEy8bpB6BrxUiWBs5iQLlQxjBrNIBmcFPMyAbXVNx0pLel7q4i++7UYmUN+xqPPv8GUDpfhQNEH5vFqyci2w7huY1092V1d5Up92vx/Coe5uiI4QGaxMAOUM7N69SvBAfwCSk11b9WI1v1eFz54b24/oZOPByQaMU/Vw2jeS8ix0/gWYQr12ESWHPmjBH6nTg0A0p2EW9EOIinXsItxumcrvV4LzrX199V24hxyGD7rr2gVFaxDrS04ZmhQiqplxoexgYEvcP14+v8hUUIcQYWLEKIM7BgEUKcgQWLEOIMLFiEEGfwevzYZVVELjShmbKrB/OZbUGMOg/vMVxC/wpmRNuOo28o8m7471LYyLtOlgZAWc2gSziwjB9nQ9VwKBI+zP1Gs/Vdwv4W/HMikl3GK1MpoMfU5cM3pivGB1w/ipZZk2BzW82r59+qxc9+/5egXH8zNvntrV4FSnvacAmfuwFtu9w96mU3oXBh1vCCF5dQrKm8/IOPnwblod89q5d66BEcpHrp9fWbXF8/avi8J16Oz9WRx/DqtXZuAmWwDd8lIuUqBno7G+p7l3nLoV5O4yO6uobPdtinBvQWjQem2oF+XIPUDz8PDAxpcf0W9TCs4PTTShZ3HikaM1mPH8Qe09GWUXyROizAX1iEEGdgwSKEOAMLFiHEGViwCCHOwIJFCHEGX61W3x0jhJCXAvyFRQhxBhYsQogzsGARQpyBBYsQ4gwsWIQQZ/AefuInWu3swnzQgUdx+um5o9gRMZGwwlx+1RSxAZX3vP8vQHmTb71e6T1/dQMoy+ewVWO+B+drdgYv0kvNdV0LytoIbv4v3oixqW/+29f1UoUy/sVIDINa+TL2Fy3ljEGqjar9ZqWK/6K8+53vAuVDH/mUsas89kG96bp1oJw4gHfw0YNGEm10B/arrFawM+qX/gG7nr7ld3hZRGRj/tegtKmwWDCGtnVLxUiilSrYM3NyAVN7H7/6UlDCe/5cL1Ws9oLS78M06OVvw36tD/3Lm/RSNyd2ghIZwsv+tXs+DcqPfvCwXurp3+Ho1plJTIMuJ/Ep6u8xQoKJnhC+bBO2S/0ft+Fzdd9P/1kvVc3jN667A69eoAtv6vOHMDYoIo3qO3jr7s24VAIjt/yFRQhxBhYsQogzsGARQpyBBYsQ4gwsWIQQZ2DBIoQ4g9ffZ0xeHD+JbX+f/OlZUAoptMAT1xjdYKtxnCrqD6JnrGmIGaMlr2s9A8rpAQ8U3zqc/jgVNPoaNwsuFd64te6uylZQ3IuhZ7y4hn14AxE81jC02ZgZWVEDQ6fUlFaDGh4TEZHlFbTARy7C4x059Zpf3We0ky6qiahNrUZ/Z2CrHzsdi8jGOF72RtUDupTBcaEd5YBeKnkIR5Y2qb7eInisYSB7iV5qoBmN+YZt+BS9/ko8ZzPc9VG9VKgJR8z++jvP6ZcB/+fz/6nFk8/jgFJ/FVuQt3biN669f1gvFevEU0oSwy+O5uia8R1MTOOX9/AhPFoxefaHoLx4XA1mFsk34mmVu333gvLQ058Ahb+wCCHOwIJFCHEGFixCiDOwYBFCnIEFixDiDN7MBcNj+sU3cNbpk/eiv3PN63aD0rsFE5Uikq6gtRFpru8xNRVx9KOIzCUwjTx8HS71zf93AZTgfsP+a9qM6xePduKL8POJeIbhmFdJ4HAM7Y9X3Yzm0dmHDPvvvm8/DUr7cJN+GZCaM+7gmRfRY+oYxLTw4LYeUMbPHdJL9Y/ieMvGDsO2A9bGMC0vIn1RfIr6mvDxqAXwH9HxBw7rpYpPHgdltWMLvuhNHwGhbXOzXqoQxDs4cy26uj/68f8FpdPK52+5DJ21W1rrX6vBIRy/KiIdbXjrE13o0G3ahcnqXS8zdhXy8JIW/oAfKtv6VTBdpLsZw9Xby/gXH1rBjzy9akxrLlXQu9zzmovr7oq/sAghzsCCRQhxBhYsQogzsGARQpyBBYsQ4gze7x807KonfonJwcQIWgbXvhlNmb6LDUNkbh49rEIRjSFN+8VG3rBUQuPpzAz2VJ3Now8Vmf+9XmqxCYN1a3m0F0XQgolGjARWUHUxHh3dBMpT/4Ku69c++gu91I4b8ZJe/vr6CcfknNFtuZxRF7CGPYtbe7Ar9MLyql6qUMA3lqv1k2gtc0Zf48D4U6Cs37wBlOcPPg9KJG89MBMYhCwvTtTd1dbrMDYoInLyVyCM7MRdrex8DSiV2IBe6eB3vwHK7degQay56iadgpSOdgz5Dm1rB6WlE+9gTYzLnpzCwGalUv+XyshmlUAUyRzFOO3EEWzlfODAL0HpbzMaN58+9hAoU4dVmFb2w3/zFxYhxBlYsAghzsCCRQhxBhYsQogzsGARQpzBK61ie0wR2bwbU2avfjdOiLzi9eiSzCXH9VL5GQxqZdfqp6vOlYyJntmTGARrTmDma294DyjJQcNoO1fAzben0V7U5Ffws4hIaQW3+uwh9Fgf+S8cJLnnLdv1Un/1XZzWOX4el9IEre6R+Rr2QV2dxX2uzGBvz5ZuI9QWjqFdFfgDPKaBNiMx2j6OecblIxgJHOpChy7XYnRnbRq5DpSjKXwYNIXv/FSL63ZharT8AFrbj1VmQbl1aVovlT42BMoPO/DW7xRMzI2OGJe9uwv3sG4UPcFSFTN6xbLRFLfUgpZxeaWiXwbE1Y0QkdwYmulzU5habWjG+3X0uFEZ+i/bBcrGXcZ1APgLixDiDCxYhBBnYMEihDgDCxYhxBlYsAghzsCCRQhxBs/v4aBTERnahz5o1zCarFMTaFVOzaJNLiJz5zFP619Dx13ja1bTO0UqginchTUsuEfj6MH3ncbgsYhs7cZzG6fCRkNYILdiHAHxQvhxAgk8Z3DLJ68CZb+Vah47i5f06R8dBeXq/40p3MFNRkh1ZgXbVZ9+Uc3FLKEFvnWv1RW3VzW5DtU/mJIKGRb1RAGzx888h4cDvHV49So7jZjxdAg9/vTm+g12O/cZYuv1eK2OpLFV9M4ObFgcf84I9G7cgN+mazBBbzA9ZcyvxWddpLkd/2IgjmdHcjnjyzU9jkHxC5O4z8u34bsWjxmHeAqTeB2a/fhVbarNg9LeYXSm3rAB39hdqN8Wgb+wCCHOwIJFCHEGFixCiDOwYBFCnIEFixDirp1JPgAAAUNJREFUDF641WhG3NONFky8EV+2soBZ07Fzhom2OItWVEtIGyCKGPZdFZEXlZ/ZfhTNPn8B9zAdM1K4gdMHQTncpCPEN8J/h1sNY0i7hMGIUoK4q6d+iTNTRWR2TKWRw8ZsWiAQMlokd3fhp546i3Zt3EOzr7/f+IDBBpWaTqPrpDmfNMa7joxiZ+qu3mFQJn14TxeG8TUiEhhBWytdwGS1JnjTa7V44El8GPa9GltvB5MY/s/tN6zSrQkc0Nu/bHhtQLjB8Hlb1RezVMG7MzOFX6XpGcP0Hz+LBnE6aTQXABaLRpi8RfmSQwM4XXjHRgz2r9aMqP+mfXiRyyun6u6Kv7AIIc7AgkUIcQYWLEKIM7BgEUKcgQWLEOIMvlrN6KlKCCEvQfgLixDiDCxYhBBnYMEihDgDCxYhxBlYsAghzvD/ATL5hDDKAaPrAAAAAElFTkSuQmCC",
      "text/plain": [
       "Console does not support images"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 400,
       "width": 400
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "local w = net:get(1).weight\n",
    "local im = image.toDisplayTensor{input=w, padding=1, nrow=8, scaleeach=true}\n",
    "itorch.image(image.scale(im, 400, 400, 'simple'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Test the network with a random image from wikipedia"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "local image_url = 'http://upload.wikimedia.org/wikipedia/commons/e/e9/Goldfish3.jpg'\n",
    "image_name = paths.basename(image_url)\n",
    "if not paths.filep(image_name) then os.execute('wget '..image_url)   end"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "Console does not support images"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 224,
       "width": 224
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "-- Rescales and normalizes the image\n",
    "function preprocess(im, img_mean)\n",
    "  -- rescale the image\n",
    "  local im3 = image.scale(im,224,224,'bilinear')\n",
    "  -- subtract imagenet mean and divide by std\n",
    "  for i=1,3 do im3[i]:add(-img_mean.mean[i]):div(img_mean.std[i]) end\n",
    "  return im3\n",
    "end\n",
    "\n",
    "local transform = {\n",
    "   mean = {\n",
    "      0.48462227599918,\n",
    "      0.45624044862054,\n",
    "      0.40588363755159,\n",
    "   },\n",
    "   std = {\n",
    "      0.22889466674951,\n",
    "      0.22446679341259,\n",
    "      0.22495548344775,\n",
    "   }\n",
    "}\n",
    "\n",
    "local im = image.load(image_name)\n",
    "I = preprocess(im, transform):float()\n",
    "itorch.image(I)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "predicted class 1: \tgoldfish, Carassius auratus\t\n",
       "predicted class 2: \ttench, Tinca tinca\t\n",
       "predicted class 3: \tgar, garfish, garpike, billfish, Lepisosteus osseus\t\n",
       "predicted class 4: \teft\t\n",
       "predicted class 5: \taxolotl, mud puppy, Ambystoma mexicanum\t\n"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "os.execute('wget https://raw.githubusercontent.com/szagoruyko/torch-opencv-demos/master/imagenet_classification/synset.t7')\n",
    "local synset_words = torch.load('synset.t7', 'ascii')\n",
    "\n",
    "local _,classes = net:forward(I:view(1,3,224,224)):view(-1):sort(true)\n",
    "for i=1,5 do\n",
    "  print('predicted class '..tostring(i)..': ', synset_words[classes[i] ])\n",
    "end"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# More examples\n",
    "* https://github.com/facebook/fb.resnet.torch the original package used to trained ResNet networks, has also feature extraction and fine-tuning on a custom dataset examples\n",
    "* https://github.com/szagoruyko/torch-opencv-demos/blob/master/imagenet_classification/demo.lua run a network in realtime and visualize predicted classes\n",
    "* https://github.com/Aysegul/torch-visbox visualize network outputs in real time using webcam\n",
    "* https://github.com/szagoruyko/loadcaffe pre-trained networks from caffe"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "iTorch",
   "language": "lua",
   "name": "itorch"
  },
  "language_info": {
   "name": "lua",
   "version": "5.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}