test_vae.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-06-15T03:43:14.835578Z",
     "start_time": "2019-06-15T03:43:13.022507Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /home/data/private/disk5/stevetod/anaconda3/envs/tf_py3/lib/python3.6/site-packages/tensorflow/python/framework/op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Colocations handled automatically by placer.\n"
     ]
    }
   ],
   "source": [
    "import sys\n",
    "import os\n",
    "os.environ['CUDA_VISIBLE_DEVICES'] = '1'\n",
    "import os.path as osp\n",
    "sys.path.append(os.path.abspath('./'))\n",
    "from data.pointcloud_dataset import load_one_class_under_folder\n",
    "from trainers.point_net_vae import PointNetVariationalAutoEncoder\n",
    "from trainers.autoencoder import default_train_params, Configuration\n",
    "from models.vae_models import mlp_architecture_ala_iclr_18\n",
    "from utils.dirs import create_dir\n",
    "from utils.tf import reset_tf_graph\n",
    "from utils.utils import plot_3d_point_cloud"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-06-15T03:43:14.874655Z",
     "start_time": "2019-06-15T03:43:14.838687Z"
    }
   },
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-06-15T03:43:15.021871Z",
     "start_time": "2019-06-15T03:43:14.876284Z"
    }
   },
   "outputs": [],
   "source": [
    "class_name = 'chair'\n",
    "data_root = '../latent_3d_points/data/shape_net_core_uniform_samples_2048/'\n",
    "train_root = './experiments'\n",
    "experiment_name = 'single_class_vae_test_summary'\n",
    "n_pc_points = 2048                # Number of points per model.\n",
    "bneck_size = 128                  # Bottleneck-AE size\n",
    "ae_loss = 'chamfer'                   # Loss to optimize: 'emd' or 'chamfer'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-06-15T03:43:16.232562Z",
     "start_time": "2019-06-15T03:43:15.024218Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6778 pclouds were loaded. They belong in 1 shape-classes.\n"
     ]
    }
   ],
   "source": [
    "pc_dataset = load_one_class_under_folder(data_root, class_name, verbose=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-06-15T03:43:16.267097Z",
     "start_time": "2019-06-15T03:43:16.234896Z"
    }
   },
   "outputs": [],
   "source": [
    "train_params = default_train_params()\n",
    "train_params['batch_size'] = 400"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-06-15T03:43:16.367401Z",
     "start_time": "2019-06-15T03:43:16.270569Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'batch_size': 400, 'training_epochs': 500, 'denoising': False, 'learning_rate': 0.0005, 'z_rotate': False, 'saver_step': 10, 'loss_display_step': 1}\n"
     ]
    }
   ],
   "source": [
    "print(train_params)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-06-15T03:43:16.493929Z",
     "start_time": "2019-06-15T03:43:16.370528Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "encoder, decoder, enc_args, dec_args = mlp_architecture_ala_iclr_18(n_pc_points, bneck_size)\n",
    "train_dir = osp.join(train_root, experiment_name)\n",
    "create_dir(osp.join(train_root, experiment_name))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-06-15T03:43:16.590855Z",
     "start_time": "2019-06-15T03:43:16.496484Z"
    }
   },
   "outputs": [],
   "source": [
    "conf = Configuration(n_input = [n_pc_points, 3],\n",
    "            loss = ae_loss,\n",
    "            training_epochs = train_params['training_epochs'],\n",
    "            batch_size = train_params['batch_size'],\n",
    "            denoising = train_params['denoising'],\n",
    "            learning_rate = train_params['learning_rate'],\n",
    "            train_dir = train_dir,\n",
    "            loss_display_step = train_params['loss_display_step'],\n",
    "            saver_step = train_params['saver_step'],\n",
    "            z_rotate = train_params['z_rotate'],\n",
    "            encoder = encoder,\n",
    "            decoder = decoder,\n",
    "            encoder_args = enc_args,\n",
    "            decoder_args = dec_args\n",
    "           )\n",
    "conf.experiment_name = experiment_name\n",
    "conf.held_out_step = 5   # How often to evaluate/print out loss on \n",
    "                         # held_out data (if they are provided in ae.train() ).\n",
    "conf.save(osp.join(train_dir, 'configuration'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-06-15T03:43:16.677666Z",
     "start_time": "2019-06-15T03:43:16.593989Z"
    }
   },
   "outputs": [],
   "source": [
    "load_pre_trained_ae = False\n",
    "restore_epoch = 500\n",
    "if load_pre_trained_ae:\n",
    "    conf = Configuration.load(train_dir + '/configuration')\n",
    "    reset_tf_graph()\n",
    "    vae = PointNetVariationalAutoEncoder(conf.experiment_name, conf)\n",
    "    vae.restore_model(conf.train_dir, epoch=restore_epoch)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-06-15T03:43:16.779875Z",
     "start_time": "2019-06-15T03:43:16.679722Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                    batch_size: 400\n",
      "                 consistent_io: None\n",
      "                         debug: False\n",
      "                       decoder: decoder_with_fc_only\n",
      "                  decoder_args: {'layer_sizes': [256, 256, 6144], 'b_norm': False, 'b_norm_finish': False, 'verbose': True}\n",
      "                       encoder: variational_encoder_with_convs_and_symmetry\n",
      "                  encoder_args: {'n_filters': [64, 128, 128, 256, 128], 'filter_sizes': [1], 'strides': [1], 'b_norm': True, 'verbose': True}\n",
      "               experiment_name: single_class_vae_test_summary\n",
      "                 gauss_augment: None\n",
      "                 held_out_step: 5\n",
      "                  is_denoising: False\n",
      "               latent_vs_recon: 0.01\n",
      "                 learning_rate: 0.0005\n",
      "                          loss: chamfer\n",
      "             loss_display_step: 1\n",
      "             loss_summary_step: 10\n",
      "                       n_input: [2048, 3]\n",
      "                      n_output: [2048, 3]\n",
      "                           n_z: None\n",
      "             saver_max_to_keep: None\n",
      "                    saver_step: 10\n",
      "                     train_dir: ./experiments/single_class_vae_test_summary\n",
      "               training_epochs: 500\n",
      "                      z_rotate: False\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(conf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-06-15T03:43:20.243906Z",
     "start_time": "2019-06-15T03:43:16.782565Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Building Encoder\n",
      "WARNING:tensorflow:From /home/data/private/disk5/stevetod/anaconda3/envs/tf_py3/lib/python3.6/site-packages/tflearn/initializations.py:119: UniformUnitScaling.__init__ (from tensorflow.python.ops.init_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use tf.initializers.variance_scaling instead with distribution=uniform to get equivalent behavior.\n",
      "encoder_conv_layer_0 conv params =  256 bnorm params =  128\n",
      "Tensor(\"single_class_vae_test_summary_2/Relu:0\", shape=(?, 2048, 64), dtype=float32)\n",
      "output size: 131072 \n",
      "\n",
      "encoder_conv_layer_1 conv params =  8320 bnorm params =  256\n",
      "Tensor(\"single_class_vae_test_summary_2/Relu_1:0\", shape=(?, 2048, 128), dtype=float32)\n",
      "output size: 262144 \n",
      "\n",
      "encoder_conv_layer_2 conv params =  16512 bnorm params =  256\n",
      "Tensor(\"single_class_vae_test_summary_2/Relu_2:0\", shape=(?, 2048, 128), dtype=float32)\n",
      "output size: 262144 \n",
      "\n",
      "encoder_conv_layer_3 conv params =  33024 bnorm params =  512\n",
      "Tensor(\"single_class_vae_test_summary_2/Relu_3:0\", shape=(?, 2048, 256), dtype=float32)\n",
      "output size: 524288 \n",
      "\n",
      "encoder_conv_layer_4 conv params =  32896 bnorm params =  256\n",
      "Tensor(\"single_class_vae_test_summary_2/Relu_4:0\", shape=(?, 2048, 128), dtype=float32)\n",
      "output size: 262144 \n",
      "\n",
      "encoder_conv_layer_4 conv params =  32896 bnorm params =  256\n",
      "Tensor(\"single_class_vae_test_summary_2/Relu_5:0\", shape=(?, 2048, 128), dtype=float32)\n",
      "output size: 262144 \n",
      "\n",
      "Tensor(\"single_class_vae_test_summary_2/z_mean:0\", shape=(?, 128), dtype=float32)\n",
      "Tensor(\"single_class_vae_test_summary_2/z_std:0\", shape=(?, 128), dtype=float32)\n",
      "Building Decoder\n",
      "decoder_fc_0 FC params =  33024 Tensor(\"single_class_vae_test_summary_2/Relu_6:0\", shape=(?, 256), dtype=float32)\n",
      "output size: 256 \n",
      "\n",
      "decoder_fc_1 FC params =  65792 Tensor(\"single_class_vae_test_summary_2/Relu_7:0\", shape=(?, 256), dtype=float32)\n",
      "output size: 256 \n",
      "\n",
      "decoder_fc_2 FC params =  1579008 Tensor(\"single_class_vae_test_summary_2/decoder_fc_2/BiasAdd:0\", shape=(?, 6144), dtype=float32)\n",
      "output size: 6144 \n",
      "\n",
      "WARNING:tensorflow:From /home/data/private/disk5/stevetod/anaconda3/envs/tf_py3/lib/python3.6/site-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use tf.cast instead.\n",
      "WARNING:tensorflow:From /home/data/private/disk5/stevetod/anaconda3/envs/tf_py3/lib/python3.6/site-packages/tensorflow/python/ops/math_grad.py:102: div (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Deprecated in favor of operator or tf.math.divide.\n",
      "allow growth\n"
     ]
    }
   ],
   "source": [
    "reset_tf_graph()\n",
    "vae = PointNetVariationalAutoEncoder(conf.experiment_name, conf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-06-15T04:56:48.698500Z",
     "start_time": "2019-06-15T03:43:20.247473Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 0001 training time (minutes)= 0.1916 loss= 7.837500811\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-1 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0002 training time (minutes)= 0.1466 loss= 3.091371328\n",
      "Epoch: 0003 training time (minutes)= 0.1467 loss= 1.978114963\n",
      "Epoch: 0004 training time (minutes)= 0.1475 loss= 1.454774797\n",
      "Epoch: 0005 training time (minutes)= 0.1485 loss= 1.151487745\n",
      "Epoch: 0006 training time (minutes)= 0.1479 loss= 0.953286793\n",
      "Epoch: 0007 training time (minutes)= 0.1477 loss= 0.811655033\n",
      "Epoch: 0008 training time (minutes)= 0.1482 loss= 0.705809403\n",
      "Epoch: 0009 training time (minutes)= 0.1496 loss= 0.622832164\n",
      "Epoch: 0010 training time (minutes)= 0.1475 loss= 0.556542229\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-10 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0011 training time (minutes)= 0.1470 loss= 0.502346892\n",
      "Epoch: 0012 training time (minutes)= 0.1474 loss= 0.456364544\n",
      "Epoch: 0013 training time (minutes)= 0.1465 loss= 0.418020727\n",
      "Epoch: 0014 training time (minutes)= 0.1460 loss= 0.384038579\n",
      "Epoch: 0015 training time (minutes)= 0.1471 loss= 0.354631765\n",
      "Epoch: 0016 training time (minutes)= 0.1462 loss= 0.329703527\n",
      "Epoch: 0017 training time (minutes)= 0.1461 loss= 0.306353066\n",
      "Epoch: 0018 training time (minutes)= 0.1463 loss= 0.285711015\n",
      "Epoch: 0019 training time (minutes)= 0.1457 loss= 0.267606938\n",
      "Epoch: 0020 training time (minutes)= 0.1465 loss= 0.251276424\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-20 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0021 training time (minutes)= 0.1476 loss= 0.236040549\n",
      "Epoch: 0022 training time (minutes)= 0.1465 loss= 0.222097676\n",
      "Epoch: 0023 training time (minutes)= 0.1462 loss= 0.209660576\n",
      "Epoch: 0024 training time (minutes)= 0.1457 loss= 0.198169697\n",
      "Epoch: 0025 training time (minutes)= 0.1474 loss= 0.187499585\n",
      "Epoch: 0026 training time (minutes)= 0.1469 loss= 0.177661458\n",
      "Epoch: 0027 training time (minutes)= 0.1458 loss= 0.168424533\n",
      "Epoch: 0028 training time (minutes)= 0.1459 loss= 0.159692759\n",
      "Epoch: 0029 training time (minutes)= 0.1465 loss= 0.151783944\n",
      "Epoch: 0030 training time (minutes)= 0.1469 loss= 0.144140172\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-30 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0031 training time (minutes)= 0.1479 loss= 0.137279689\n",
      "Epoch: 0032 training time (minutes)= 0.1471 loss= 0.130704365\n",
      "Epoch: 0033 training time (minutes)= 0.1475 loss= 0.124686384\n",
      "Epoch: 0034 training time (minutes)= 0.1469 loss= 0.119143721\n",
      "Epoch: 0035 training time (minutes)= 0.1471 loss= 0.113661379\n",
      "Epoch: 0036 training time (minutes)= 0.1469 loss= 0.108390825\n",
      "Epoch: 0037 training time (minutes)= 0.1458 loss= 0.103732280\n",
      "Epoch: 0038 training time (minutes)= 0.1462 loss= 0.099147583\n",
      "Epoch: 0039 training time (minutes)= 0.1472 loss= 0.094637196\n",
      "Epoch: 0040 training time (minutes)= 0.1470 loss= 0.090693678\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-40 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0041 training time (minutes)= 0.1462 loss= 0.086822988\n",
      "Epoch: 0042 training time (minutes)= 0.1469 loss= 0.083126449\n",
      "Epoch: 0043 training time (minutes)= 0.1470 loss= 0.079786293\n",
      "Epoch: 0044 training time (minutes)= 0.1471 loss= 0.076323615\n",
      "Epoch: 0045 training time (minutes)= 0.1466 loss= 0.073311739\n",
      "Epoch: 0046 training time (minutes)= 0.1472 loss= 0.070404375\n",
      "Epoch: 0047 training time (minutes)= 0.1481 loss= 0.067678257\n",
      "Epoch: 0048 training time (minutes)= 0.1465 loss= 0.064991764\n",
      "Epoch: 0049 training time (minutes)= 0.1460 loss= 0.062319839\n",
      "Epoch: 0050 training time (minutes)= 0.1467 loss= 0.059855451\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-50 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0051 training time (minutes)= 0.1458 loss= 0.057672868\n",
      "Epoch: 0052 training time (minutes)= 0.1468 loss= 0.055402887\n",
      "Epoch: 0053 training time (minutes)= 0.1467 loss= 0.053318333\n",
      "Epoch: 0054 training time (minutes)= 0.1466 loss= 0.051312425\n",
      "Epoch: 0055 training time (minutes)= 0.1464 loss= 0.049474316\n",
      "Epoch: 0056 training time (minutes)= 0.1471 loss= 0.047546929\n",
      "Epoch: 0057 training time (minutes)= 0.1463 loss= 0.045821222\n",
      "Epoch: 0058 training time (minutes)= 0.1462 loss= 0.044113037\n",
      "Epoch: 0059 training time (minutes)= 0.1469 loss= 0.042605306\n",
      "Epoch: 0060 training time (minutes)= 0.1465 loss= 0.041090579\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-60 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0061 training time (minutes)= 0.1463 loss= 0.039631466\n",
      "Epoch: 0062 training time (minutes)= 0.1464 loss= 0.038131532\n",
      "Epoch: 0063 training time (minutes)= 0.1463 loss= 0.036921018\n",
      "Epoch: 0064 training time (minutes)= 0.1462 loss= 0.035645968\n",
      "Epoch: 0065 training time (minutes)= 0.1463 loss= 0.034369327\n",
      "Epoch: 0066 training time (minutes)= 0.1472 loss= 0.033218745\n",
      "Epoch: 0067 training time (minutes)= 0.1481 loss= 0.032045223\n",
      "Epoch: 0068 training time (minutes)= 0.1467 loss= 0.030908859\n",
      "Epoch: 0069 training time (minutes)= 0.1467 loss= 0.029947860\n",
      "Epoch: 0070 training time (minutes)= 0.1458 loss= 0.028926015\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-70 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0071 training time (minutes)= 0.1460 loss= 0.028098349\n",
      "Epoch: 0072 training time (minutes)= 0.1463 loss= 0.027109528\n",
      "Epoch: 0073 training time (minutes)= 0.1476 loss= 0.026242042\n",
      "Epoch: 0074 training time (minutes)= 0.1458 loss= 0.025350900\n",
      "Epoch: 0075 training time (minutes)= 0.1467 loss= 0.024537199\n",
      "Epoch: 0076 training time (minutes)= 0.1468 loss= 0.023899287\n",
      "Epoch: 0077 training time (minutes)= 0.1468 loss= 0.023097636\n",
      "Epoch: 0078 training time (minutes)= 0.1466 loss= 0.022389184\n",
      "Epoch: 0079 training time (minutes)= 0.1462 loss= 0.021750108\n",
      "Epoch: 0080 training time (minutes)= 0.1461 loss= 0.021109854\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-80 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0081 training time (minutes)= 0.1466 loss= 0.020385054\n",
      "Epoch: 0082 training time (minutes)= 0.1462 loss= 0.019824850\n",
      "Epoch: 0083 training time (minutes)= 0.1464 loss= 0.019226025\n",
      "Epoch: 0084 training time (minutes)= 0.1475 loss= 0.018719878\n",
      "Epoch: 0085 training time (minutes)= 0.1464 loss= 0.018213859\n",
      "Epoch: 0086 training time (minutes)= 0.1466 loss= 0.017654024\n",
      "Epoch: 0087 training time (minutes)= 0.1472 loss= 0.017205323\n",
      "Epoch: 0088 training time (minutes)= 0.1468 loss= 0.016737666\n",
      "Epoch: 0089 training time (minutes)= 0.1459 loss= 0.016231033\n",
      "Epoch: 0090 training time (minutes)= 0.1468 loss= 0.015750424\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-90 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0091 training time (minutes)= 0.1466 loss= 0.015345540\n",
      "Epoch: 0092 training time (minutes)= 0.1475 loss= 0.015018355\n",
      "Epoch: 0093 training time (minutes)= 0.1464 loss= 0.014541283\n",
      "Epoch: 0094 training time (minutes)= 0.1466 loss= 0.014180468\n",
      "Epoch: 0095 training time (minutes)= 0.1466 loss= 0.013923568\n",
      "Epoch: 0096 training time (minutes)= 0.1462 loss= 0.013516890\n",
      "Epoch: 0097 training time (minutes)= 0.1471 loss= 0.013152969\n",
      "Epoch: 0098 training time (minutes)= 0.1468 loss= 0.012820316\n",
      "Epoch: 0099 training time (minutes)= 0.1464 loss= 0.012548381\n",
      "Epoch: 0100 training time (minutes)= 0.1463 loss= 0.012246312\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-100 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0101 training time (minutes)= 0.1462 loss= 0.011970727\n",
      "Epoch: 0102 training time (minutes)= 0.1461 loss= 0.011730791\n",
      "Epoch: 0103 training time (minutes)= 0.1469 loss= 0.011426881\n",
      "Epoch: 0104 training time (minutes)= 0.1459 loss= 0.011157245\n",
      "Epoch: 0105 training time (minutes)= 0.1471 loss= 0.010903923\n",
      "Epoch: 0106 training time (minutes)= 0.1472 loss= 0.010678241\n",
      "Epoch: 0107 training time (minutes)= 0.1464 loss= 0.010419530\n",
      "Epoch: 0108 training time (minutes)= 0.1473 loss= 0.010218151\n",
      "Epoch: 0109 training time (minutes)= 0.1465 loss= 0.009985776\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 0110 training time (minutes)= 0.1461 loss= 0.009792217\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-110 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0111 training time (minutes)= 0.1461 loss= 0.009612761\n",
      "Epoch: 0112 training time (minutes)= 0.1472 loss= 0.009417799\n",
      "Epoch: 0113 training time (minutes)= 0.1473 loss= 0.009236273\n",
      "Epoch: 0114 training time (minutes)= 0.1461 loss= 0.008990355\n",
      "Epoch: 0115 training time (minutes)= 0.1465 loss= 0.008852126\n",
      "Epoch: 0116 training time (minutes)= 0.1472 loss= 0.008716113\n",
      "Epoch: 0117 training time (minutes)= 0.1460 loss= 0.008525032\n",
      "Epoch: 0118 training time (minutes)= 0.1474 loss= 0.008356743\n",
      "Epoch: 0119 training time (minutes)= 0.1472 loss= 0.008235940\n",
      "Epoch: 0120 training time (minutes)= 0.1468 loss= 0.008102264\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-120 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0121 training time (minutes)= 0.1466 loss= 0.007933908\n",
      "Epoch: 0122 training time (minutes)= 0.1473 loss= 0.007831429\n",
      "Epoch: 0123 training time (minutes)= 0.1464 loss= 0.007676242\n",
      "Epoch: 0124 training time (minutes)= 0.1470 loss= 0.007565150\n",
      "Epoch: 0125 training time (minutes)= 0.1470 loss= 0.007429210\n",
      "Epoch: 0126 training time (minutes)= 0.1464 loss= 0.007321970\n",
      "Epoch: 0127 training time (minutes)= 0.1473 loss= 0.007201968\n",
      "Epoch: 0128 training time (minutes)= 0.1467 loss= 0.007077915\n",
      "Epoch: 0129 training time (minutes)= 0.1470 loss= 0.006994552\n",
      "Epoch: 0130 training time (minutes)= 0.1466 loss= 0.006869202\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-130 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0131 training time (minutes)= 0.1470 loss= 0.006771422\n",
      "Epoch: 0132 training time (minutes)= 0.1465 loss= 0.006667140\n",
      "Epoch: 0133 training time (minutes)= 0.1464 loss= 0.006568208\n",
      "Epoch: 0134 training time (minutes)= 0.1461 loss= 0.006489064\n",
      "Epoch: 0135 training time (minutes)= 0.1466 loss= 0.006403381\n",
      "Epoch: 0136 training time (minutes)= 0.1459 loss= 0.006294968\n",
      "Epoch: 0137 training time (minutes)= 0.1473 loss= 0.006233890\n",
      "Epoch: 0138 training time (minutes)= 0.1463 loss= 0.006158544\n",
      "Epoch: 0139 training time (minutes)= 0.1463 loss= 0.006089666\n",
      "Epoch: 0140 training time (minutes)= 0.1464 loss= 0.005984512\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-140 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0141 training time (minutes)= 0.1460 loss= 0.005923578\n",
      "Epoch: 0142 training time (minutes)= 0.1463 loss= 0.005831108\n",
      "Epoch: 0143 training time (minutes)= 0.1471 loss= 0.005770894\n",
      "Epoch: 0144 training time (minutes)= 0.1470 loss= 0.005723990\n",
      "Epoch: 0145 training time (minutes)= 0.1474 loss= 0.005628113\n",
      "Epoch: 0146 training time (minutes)= 0.1465 loss= 0.005577227\n",
      "Epoch: 0147 training time (minutes)= 0.1473 loss= 0.005521369\n",
      "Epoch: 0148 training time (minutes)= 0.1462 loss= 0.005481763\n",
      "Epoch: 0149 training time (minutes)= 0.1471 loss= 0.005400929\n",
      "Epoch: 0150 training time (minutes)= 0.1467 loss= 0.005339321\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-150 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0151 training time (minutes)= 0.1473 loss= 0.005276675\n",
      "Epoch: 0152 training time (minutes)= 0.1469 loss= 0.005230660\n",
      "Epoch: 0153 training time (minutes)= 0.1464 loss= 0.005191979\n",
      "Epoch: 0154 training time (minutes)= 0.1473 loss= 0.005145462\n",
      "Epoch: 0155 training time (minutes)= 0.1472 loss= 0.005094258\n",
      "Epoch: 0156 training time (minutes)= 0.1466 loss= 0.005053094\n",
      "Epoch: 0157 training time (minutes)= 0.1467 loss= 0.005008805\n",
      "Epoch: 0158 training time (minutes)= 0.1460 loss= 0.004958791\n",
      "Epoch: 0159 training time (minutes)= 0.1460 loss= 0.004913539\n",
      "Epoch: 0160 training time (minutes)= 0.1458 loss= 0.004873741\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-160 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0161 training time (minutes)= 0.1470 loss= 0.004805234\n",
      "Epoch: 0162 training time (minutes)= 0.1468 loss= 0.004787022\n",
      "Epoch: 0163 training time (minutes)= 0.1462 loss= 0.004754590\n",
      "Epoch: 0164 training time (minutes)= 0.1462 loss= 0.004706611\n",
      "Epoch: 0165 training time (minutes)= 0.1459 loss= 0.004691431\n",
      "Epoch: 0166 training time (minutes)= 0.1458 loss= 0.004637055\n",
      "Epoch: 0167 training time (minutes)= 0.1472 loss= 0.004596497\n",
      "Epoch: 0168 training time (minutes)= 0.1467 loss= 0.004586726\n",
      "Epoch: 0169 training time (minutes)= 0.1467 loss= 0.004539930\n",
      "Epoch: 0170 training time (minutes)= 0.1466 loss= 0.004529455\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-170 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0171 training time (minutes)= 0.1463 loss= 0.004500317\n",
      "Epoch: 0172 training time (minutes)= 0.1462 loss= 0.004443831\n",
      "Epoch: 0173 training time (minutes)= 0.1470 loss= 0.004429862\n",
      "Epoch: 0174 training time (minutes)= 0.1471 loss= 0.004413880\n",
      "Epoch: 0175 training time (minutes)= 0.1464 loss= 0.004382352\n",
      "Epoch: 0176 training time (minutes)= 0.1458 loss= 0.004334821\n",
      "Epoch: 0177 training time (minutes)= 0.1462 loss= 0.004334314\n",
      "Epoch: 0178 training time (minutes)= 0.1464 loss= 0.004317049\n",
      "Epoch: 0179 training time (minutes)= 0.1472 loss= 0.004295558\n",
      "Epoch: 0180 training time (minutes)= 0.1469 loss= 0.004258645\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-180 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0181 training time (minutes)= 0.1467 loss= 0.004223200\n",
      "Epoch: 0182 training time (minutes)= 0.1466 loss= 0.004224945\n",
      "Epoch: 0183 training time (minutes)= 0.1465 loss= 0.004223525\n",
      "Epoch: 0184 training time (minutes)= 0.1464 loss= 0.004176208\n",
      "Epoch: 0185 training time (minutes)= 0.1464 loss= 0.004172104\n",
      "Epoch: 0186 training time (minutes)= 0.1470 loss= 0.004142601\n",
      "Epoch: 0187 training time (minutes)= 0.1470 loss= 0.004118778\n",
      "Epoch: 0188 training time (minutes)= 0.1464 loss= 0.004105762\n",
      "Epoch: 0189 training time (minutes)= 0.1467 loss= 0.004093159\n",
      "Epoch: 0190 training time (minutes)= 0.1465 loss= 0.004061206\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-190 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0191 training time (minutes)= 0.1463 loss= 0.004058017\n",
      "Epoch: 0192 training time (minutes)= 0.1459 loss= 0.004041810\n",
      "Epoch: 0193 training time (minutes)= 0.1470 loss= 0.004027949\n",
      "Epoch: 0194 training time (minutes)= 0.1467 loss= 0.004000881\n",
      "Epoch: 0195 training time (minutes)= 0.1462 loss= 0.003999915\n",
      "Epoch: 0196 training time (minutes)= 0.1468 loss= 0.003991929\n",
      "Epoch: 0197 training time (minutes)= 0.1466 loss= 0.003987141\n",
      "Epoch: 0198 training time (minutes)= 0.1475 loss= 0.003968389\n",
      "Epoch: 0199 training time (minutes)= 0.1469 loss= 0.003956061\n",
      "Epoch: 0200 training time (minutes)= 0.1464 loss= 0.003950483\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-200 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0201 training time (minutes)= 0.1468 loss= 0.003939843\n",
      "Epoch: 0202 training time (minutes)= 0.1469 loss= 0.003916628\n",
      "Epoch: 0203 training time (minutes)= 0.1466 loss= 0.003914779\n",
      "Epoch: 0204 training time (minutes)= 0.1483 loss= 0.003903606\n",
      "Epoch: 0205 training time (minutes)= 0.1467 loss= 0.003894907\n",
      "Epoch: 0206 training time (minutes)= 0.1461 loss= 0.003881575\n",
      "Epoch: 0207 training time (minutes)= 0.1478 loss= 0.003853695\n",
      "Epoch: 0208 training time (minutes)= 0.1465 loss= 0.003859045\n",
      "Epoch: 0209 training time (minutes)= 0.1476 loss= 0.003868944\n",
      "Epoch: 0210 training time (minutes)= 0.1464 loss= 0.003855623\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-210 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0211 training time (minutes)= 0.1464 loss= 0.003824550\n",
      "Epoch: 0212 training time (minutes)= 0.1465 loss= 0.003848825\n",
      "Epoch: 0213 training time (minutes)= 0.1467 loss= 0.003828384\n",
      "Epoch: 0214 training time (minutes)= 0.1466 loss= 0.003805837\n",
      "Epoch: 0215 training time (minutes)= 0.1462 loss= 0.003782013\n",
      "Epoch: 0216 training time (minutes)= 0.1464 loss= 0.003812115\n",
      "Epoch: 0217 training time (minutes)= 0.1476 loss= 0.003807224\n",
      "Epoch: 0218 training time (minutes)= 0.1473 loss= 0.003781312\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 0219 training time (minutes)= 0.1471 loss= 0.003781892\n",
      "Epoch: 0220 training time (minutes)= 0.1462 loss= 0.003793763\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-220 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0221 training time (minutes)= 0.1472 loss= 0.003785137\n",
      "Epoch: 0222 training time (minutes)= 0.1464 loss= 0.003766306\n",
      "Epoch: 0223 training time (minutes)= 0.1467 loss= 0.003766520\n",
      "Epoch: 0224 training time (minutes)= 0.1465 loss= 0.003746874\n",
      "Epoch: 0225 training time (minutes)= 0.1462 loss= 0.003755748\n",
      "Epoch: 0226 training time (minutes)= 0.1461 loss= 0.003742191\n",
      "Epoch: 0227 training time (minutes)= 0.1462 loss= 0.003754515\n",
      "Epoch: 0228 training time (minutes)= 0.1467 loss= 0.003742621\n",
      "Epoch: 0229 training time (minutes)= 0.1461 loss= 0.003734220\n",
      "Epoch: 0230 training time (minutes)= 0.1468 loss= 0.003726120\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-230 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0231 training time (minutes)= 0.1467 loss= 0.003737113\n",
      "Epoch: 0232 training time (minutes)= 0.1467 loss= 0.003730791\n",
      "Epoch: 0233 training time (minutes)= 0.1468 loss= 0.003721167\n",
      "Epoch: 0234 training time (minutes)= 0.1461 loss= 0.003724661\n",
      "Epoch: 0235 training time (minutes)= 0.1469 loss= 0.003713615\n",
      "Epoch: 0236 training time (minutes)= 0.1463 loss= 0.003717070\n",
      "Epoch: 0237 training time (minutes)= 0.1461 loss= 0.003700975\n",
      "Epoch: 0238 training time (minutes)= 0.1462 loss= 0.003705442\n",
      "Epoch: 0239 training time (minutes)= 0.1462 loss= 0.003705402\n",
      "Epoch: 0240 training time (minutes)= 0.1475 loss= 0.003708351\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-240 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0241 training time (minutes)= 0.1466 loss= 0.003704907\n",
      "Epoch: 0242 training time (minutes)= 0.1467 loss= 0.003699790\n",
      "Epoch: 0243 training time (minutes)= 0.1466 loss= 0.003701372\n",
      "Epoch: 0244 training time (minutes)= 0.1466 loss= 0.003695587\n",
      "Epoch: 0245 training time (minutes)= 0.1467 loss= 0.003700059\n",
      "Epoch: 0246 training time (minutes)= 0.1461 loss= 0.003684892\n",
      "Epoch: 0247 training time (minutes)= 0.1481 loss= 0.003679504\n",
      "Epoch: 0248 training time (minutes)= 0.1477 loss= 0.003686592\n",
      "Epoch: 0249 training time (minutes)= 0.1460 loss= 0.003683635\n",
      "Epoch: 0250 training time (minutes)= 0.1461 loss= 0.003690291\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-250 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0251 training time (minutes)= 0.1466 loss= 0.003681074\n",
      "Epoch: 0252 training time (minutes)= 0.1466 loss= 0.003672549\n",
      "Epoch: 0253 training time (minutes)= 0.1468 loss= 0.003662694\n",
      "Epoch: 0254 training time (minutes)= 0.1478 loss= 0.003667881\n",
      "Epoch: 0255 training time (minutes)= 0.1457 loss= 0.003665638\n",
      "Epoch: 0256 training time (minutes)= 0.1468 loss= 0.003670399\n",
      "Epoch: 0257 training time (minutes)= 0.1463 loss= 0.003669060\n",
      "Epoch: 0258 training time (minutes)= 0.1464 loss= 0.003645610\n",
      "Epoch: 0259 training time (minutes)= 0.1463 loss= 0.003655638\n",
      "Epoch: 0260 training time (minutes)= 0.1458 loss= 0.003670913\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-260 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0261 training time (minutes)= 0.1466 loss= 0.003658244\n",
      "Epoch: 0262 training time (minutes)= 0.1460 loss= 0.003661439\n",
      "Epoch: 0263 training time (minutes)= 0.1460 loss= 0.003664350\n",
      "Epoch: 0264 training time (minutes)= 0.1466 loss= 0.003649177\n",
      "Epoch: 0265 training time (minutes)= 0.1477 loss= 0.003665061\n",
      "Epoch: 0266 training time (minutes)= 0.1463 loss= 0.003662240\n",
      "Epoch: 0267 training time (minutes)= 0.1462 loss= 0.003652568\n",
      "Epoch: 0268 training time (minutes)= 0.1462 loss= 0.003653409\n",
      "Epoch: 0269 training time (minutes)= 0.1459 loss= 0.003655611\n",
      "Epoch: 0270 training time (minutes)= 0.1468 loss= 0.003654243\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-270 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0271 training time (minutes)= 0.1462 loss= 0.003657504\n",
      "Epoch: 0272 training time (minutes)= 0.1464 loss= 0.003649931\n",
      "Epoch: 0273 training time (minutes)= 0.1464 loss= 0.003646270\n",
      "Epoch: 0274 training time (minutes)= 0.1461 loss= 0.003637995\n",
      "Epoch: 0275 training time (minutes)= 0.1459 loss= 0.003652601\n",
      "Epoch: 0276 training time (minutes)= 0.1463 loss= 0.003653298\n",
      "Epoch: 0277 training time (minutes)= 0.1470 loss= 0.003651635\n",
      "Epoch: 0278 training time (minutes)= 0.1459 loss= 0.003647132\n",
      "Epoch: 0279 training time (minutes)= 0.1463 loss= 0.003638485\n",
      "Epoch: 0280 training time (minutes)= 0.1460 loss= 0.003639631\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-280 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0281 training time (minutes)= 0.1471 loss= 0.003637360\n",
      "Epoch: 0282 training time (minutes)= 0.1462 loss= 0.003647467\n",
      "Epoch: 0283 training time (minutes)= 0.1459 loss= 0.003646484\n",
      "Epoch: 0284 training time (minutes)= 0.1465 loss= 0.003624521\n",
      "Epoch: 0285 training time (minutes)= 0.1457 loss= 0.003639481\n",
      "Epoch: 0286 training time (minutes)= 0.1465 loss= 0.003642142\n",
      "Epoch: 0287 training time (minutes)= 0.1468 loss= 0.003647394\n",
      "Epoch: 0288 training time (minutes)= 0.1462 loss= 0.003636476\n",
      "Epoch: 0289 training time (minutes)= 0.1463 loss= 0.003640932\n",
      "Epoch: 0290 training time (minutes)= 0.1466 loss= 0.003650288\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-290 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0291 training time (minutes)= 0.1458 loss= 0.003640490\n",
      "Epoch: 0292 training time (minutes)= 0.1472 loss= 0.003635810\n",
      "Epoch: 0293 training time (minutes)= 0.1462 loss= 0.003628801\n",
      "Epoch: 0294 training time (minutes)= 0.1459 loss= 0.003631254\n",
      "Epoch: 0295 training time (minutes)= 0.1459 loss= 0.003637818\n",
      "Epoch: 0296 training time (minutes)= 0.1457 loss= 0.003621975\n",
      "Epoch: 0297 training time (minutes)= 0.1462 loss= 0.003640444\n",
      "Epoch: 0298 training time (minutes)= 0.1470 loss= 0.003630837\n",
      "Epoch: 0299 training time (minutes)= 0.1461 loss= 0.003638428\n",
      "Epoch: 0300 training time (minutes)= 0.1466 loss= 0.003619997\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-300 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0301 training time (minutes)= 0.1459 loss= 0.003638291\n",
      "Epoch: 0302 training time (minutes)= 0.1463 loss= 0.003638818\n",
      "Epoch: 0303 training time (minutes)= 0.1465 loss= 0.003641810\n",
      "Epoch: 0304 training time (minutes)= 0.1467 loss= 0.003629992\n",
      "Epoch: 0305 training time (minutes)= 0.1464 loss= 0.003640060\n",
      "Epoch: 0306 training time (minutes)= 0.1474 loss= 0.003628513\n",
      "Epoch: 0307 training time (minutes)= 0.1461 loss= 0.003631918\n",
      "Epoch: 0308 training time (minutes)= 0.1467 loss= 0.003645520\n",
      "Epoch: 0309 training time (minutes)= 0.1461 loss= 0.003642449\n",
      "Epoch: 0310 training time (minutes)= 0.1461 loss= 0.003650425\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-310 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0311 training time (minutes)= 0.1471 loss= 0.003636650\n",
      "Epoch: 0312 training time (minutes)= 0.1464 loss= 0.003647843\n",
      "Epoch: 0313 training time (minutes)= 0.1467 loss= 0.003639824\n",
      "Epoch: 0314 training time (minutes)= 0.1464 loss= 0.003640790\n",
      "Epoch: 0315 training time (minutes)= 0.1468 loss= 0.003621319\n",
      "Epoch: 0316 training time (minutes)= 0.1466 loss= 0.003632297\n",
      "Epoch: 0317 training time (minutes)= 0.1466 loss= 0.003632776\n",
      "Epoch: 0318 training time (minutes)= 0.1464 loss= 0.003631218\n",
      "Epoch: 0319 training time (minutes)= 0.1465 loss= 0.003637140\n",
      "Epoch: 0320 training time (minutes)= 0.1461 loss= 0.003640551\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-320 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0321 training time (minutes)= 0.1462 loss= 0.003623578\n",
      "Epoch: 0322 training time (minutes)= 0.1459 loss= 0.003630728\n",
      "Epoch: 0323 training time (minutes)= 0.1466 loss= 0.003627818\n",
      "Epoch: 0324 training time (minutes)= 0.1458 loss= 0.003629053\n",
      "Epoch: 0325 training time (minutes)= 0.1461 loss= 0.003622341\n",
      "Epoch: 0326 training time (minutes)= 0.1471 loss= 0.003633448\n",
      "Epoch: 0327 training time (minutes)= 0.1475 loss= 0.003623989\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 0328 training time (minutes)= 0.1463 loss= 0.003640593\n",
      "Epoch: 0329 training time (minutes)= 0.1465 loss= 0.003633090\n",
      "Epoch: 0330 training time (minutes)= 0.1470 loss= 0.003619323\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-330 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0331 training time (minutes)= 0.1459 loss= 0.003635666\n",
      "Epoch: 0332 training time (minutes)= 0.1463 loss= 0.003635448\n",
      "Epoch: 0333 training time (minutes)= 0.1463 loss= 0.003637200\n",
      "Epoch: 0334 training time (minutes)= 0.1459 loss= 0.003631136\n",
      "Epoch: 0335 training time (minutes)= 0.1468 loss= 0.003625046\n",
      "Epoch: 0336 training time (minutes)= 0.1463 loss= 0.003629799\n",
      "Epoch: 0337 training time (minutes)= 0.1463 loss= 0.003629029\n",
      "Epoch: 0338 training time (minutes)= 0.1465 loss= 0.003620645\n",
      "Epoch: 0339 training time (minutes)= 0.1460 loss= 0.003633916\n",
      "Epoch: 0340 training time (minutes)= 0.1468 loss= 0.003641979\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-340 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0341 training time (minutes)= 0.1460 loss= 0.003611566\n",
      "Epoch: 0342 training time (minutes)= 0.1462 loss= 0.003638043\n",
      "Epoch: 0343 training time (minutes)= 0.1466 loss= 0.003638687\n",
      "Epoch: 0344 training time (minutes)= 0.1465 loss= 0.003625213\n",
      "Epoch: 0345 training time (minutes)= 0.1474 loss= 0.003625338\n",
      "Epoch: 0346 training time (minutes)= 0.1466 loss= 0.003625043\n",
      "Epoch: 0347 training time (minutes)= 0.1463 loss= 0.003629232\n",
      "Epoch: 0348 training time (minutes)= 0.1461 loss= 0.003627571\n",
      "Epoch: 0349 training time (minutes)= 0.1463 loss= 0.003616749\n",
      "Epoch: 0350 training time (minutes)= 0.1465 loss= 0.003626554\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-350 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0351 training time (minutes)= 0.1458 loss= 0.003630591\n",
      "Epoch: 0352 training time (minutes)= 0.1465 loss= 0.003627518\n",
      "Epoch: 0353 training time (minutes)= 0.1479 loss= 0.003619295\n",
      "Epoch: 0354 training time (minutes)= 0.1469 loss= 0.003631161\n",
      "Epoch: 0355 training time (minutes)= 0.1466 loss= 0.003641257\n",
      "Epoch: 0356 training time (minutes)= 0.1460 loss= 0.003634270\n",
      "Epoch: 0357 training time (minutes)= 0.1468 loss= 0.003625647\n",
      "Epoch: 0358 training time (minutes)= 0.1462 loss= 0.003630967\n",
      "Epoch: 0359 training time (minutes)= 0.1459 loss= 0.003635708\n",
      "Epoch: 0360 training time (minutes)= 0.1469 loss= 0.003634574\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-360 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0361 training time (minutes)= 0.1460 loss= 0.003632354\n",
      "Epoch: 0362 training time (minutes)= 0.1464 loss= 0.003638122\n",
      "Epoch: 0363 training time (minutes)= 0.1468 loss= 0.003638419\n",
      "Epoch: 0364 training time (minutes)= 0.1471 loss= 0.003612608\n",
      "Epoch: 0365 training time (minutes)= 0.1473 loss= 0.003631929\n",
      "Epoch: 0366 training time (minutes)= 0.1464 loss= 0.003634743\n",
      "Epoch: 0367 training time (minutes)= 0.1458 loss= 0.003635495\n",
      "Epoch: 0368 training time (minutes)= 0.1461 loss= 0.003631493\n",
      "Epoch: 0369 training time (minutes)= 0.1465 loss= 0.003623256\n",
      "Epoch: 0370 training time (minutes)= 0.1460 loss= 0.003625403\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-370 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0371 training time (minutes)= 0.1459 loss= 0.003632969\n",
      "Epoch: 0372 training time (minutes)= 0.1471 loss= 0.003608849\n",
      "Epoch: 0373 training time (minutes)= 0.1462 loss= 0.003622757\n",
      "Epoch: 0374 training time (minutes)= 0.1468 loss= 0.003628289\n",
      "Epoch: 0375 training time (minutes)= 0.1462 loss= 0.003609773\n",
      "Epoch: 0376 training time (minutes)= 0.1465 loss= 0.003622339\n",
      "Epoch: 0377 training time (minutes)= 0.1464 loss= 0.003621588\n",
      "Epoch: 0378 training time (minutes)= 0.1463 loss= 0.003638992\n",
      "Epoch: 0379 training time (minutes)= 0.1462 loss= 0.003639414\n",
      "Epoch: 0380 training time (minutes)= 0.1467 loss= 0.003631588\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-380 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0381 training time (minutes)= 0.1464 loss= 0.003640161\n",
      "Epoch: 0382 training time (minutes)= 0.1464 loss= 0.003628721\n",
      "Epoch: 0383 training time (minutes)= 0.1466 loss= 0.003624385\n",
      "Epoch: 0384 training time (minutes)= 0.1474 loss= 0.003605239\n",
      "Epoch: 0385 training time (minutes)= 0.1463 loss= 0.003630615\n",
      "Epoch: 0386 training time (minutes)= 0.1462 loss= 0.003613803\n",
      "Epoch: 0387 training time (minutes)= 0.1456 loss= 0.003625126\n",
      "Epoch: 0388 training time (minutes)= 0.1460 loss= 0.003636231\n",
      "Epoch: 0389 training time (minutes)= 0.1471 loss= 0.003629152\n",
      "Epoch: 0390 training time (minutes)= 0.1464 loss= 0.003619466\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-390 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0391 training time (minutes)= 0.1473 loss= 0.003617649\n",
      "Epoch: 0392 training time (minutes)= 0.1471 loss= 0.003609320\n",
      "Epoch: 0393 training time (minutes)= 0.1470 loss= 0.003636180\n",
      "Epoch: 0394 training time (minutes)= 0.1476 loss= 0.003629728\n",
      "Epoch: 0395 training time (minutes)= 0.1473 loss= 0.003616410\n",
      "Epoch: 0396 training time (minutes)= 0.1461 loss= 0.003628013\n",
      "Epoch: 0397 training time (minutes)= 0.1464 loss= 0.003618689\n",
      "Epoch: 0398 training time (minutes)= 0.1470 loss= 0.003623035\n",
      "Epoch: 0399 training time (minutes)= 0.1462 loss= 0.003638516\n",
      "Epoch: 0400 training time (minutes)= 0.1464 loss= 0.003608643\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-400 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0401 training time (minutes)= 0.1461 loss= 0.003628193\n",
      "Epoch: 0402 training time (minutes)= 0.1471 loss= 0.003620675\n",
      "Epoch: 0403 training time (minutes)= 0.1468 loss= 0.003633531\n",
      "Epoch: 0404 training time (minutes)= 0.1461 loss= 0.003619340\n",
      "Epoch: 0405 training time (minutes)= 0.1462 loss= 0.003620887\n",
      "Epoch: 0406 training time (minutes)= 0.1466 loss= 0.003627280\n",
      "Epoch: 0407 training time (minutes)= 0.1458 loss= 0.003628806\n",
      "Epoch: 0408 training time (minutes)= 0.1467 loss= 0.003644652\n",
      "Epoch: 0409 training time (minutes)= 0.1461 loss= 0.003628358\n",
      "Epoch: 0410 training time (minutes)= 0.1465 loss= 0.003611015\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-410 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0411 training time (minutes)= 0.1463 loss= 0.003618244\n",
      "Epoch: 0412 training time (minutes)= 0.1462 loss= 0.003631152\n",
      "Epoch: 0413 training time (minutes)= 0.1463 loss= 0.003625481\n",
      "Epoch: 0414 training time (minutes)= 0.1459 loss= 0.003623598\n",
      "Epoch: 0415 training time (minutes)= 0.1463 loss= 0.003635402\n",
      "Epoch: 0416 training time (minutes)= 0.1467 loss= 0.003628526\n",
      "Epoch: 0417 training time (minutes)= 0.1465 loss= 0.003626728\n",
      "Epoch: 0418 training time (minutes)= 0.1456 loss= 0.003624846\n",
      "Epoch: 0419 training time (minutes)= 0.1457 loss= 0.003620090\n",
      "Epoch: 0420 training time (minutes)= 0.1463 loss= 0.003631996\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-420 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0421 training time (minutes)= 0.1460 loss= 0.003626026\n",
      "Epoch: 0422 training time (minutes)= 0.1459 loss= 0.003631358\n",
      "Epoch: 0423 training time (minutes)= 0.1460 loss= 0.003616382\n",
      "Epoch: 0424 training time (minutes)= 0.1470 loss= 0.003616144\n",
      "Epoch: 0425 training time (minutes)= 0.1472 loss= 0.003607832\n",
      "Epoch: 0426 training time (minutes)= 0.1465 loss= 0.003634682\n",
      "Epoch: 0427 training time (minutes)= 0.1461 loss= 0.003634876\n",
      "Epoch: 0428 training time (minutes)= 0.1462 loss= 0.003620696\n",
      "Epoch: 0429 training time (minutes)= 0.1462 loss= 0.003625249\n",
      "Epoch: 0430 training time (minutes)= 0.1465 loss= 0.003630244\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-430 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0431 training time (minutes)= 0.1457 loss= 0.003626993\n",
      "Epoch: 0432 training time (minutes)= 0.1461 loss= 0.003635058\n",
      "Epoch: 0433 training time (minutes)= 0.1459 loss= 0.003625399\n",
      "Epoch: 0434 training time (minutes)= 0.1469 loss= 0.003632111\n",
      "Epoch: 0435 training time (minutes)= 0.1465 loss= 0.003628439\n",
      "Epoch: 0436 training time (minutes)= 0.1472 loss= 0.003629252\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 0437 training time (minutes)= 0.1472 loss= 0.003625530\n",
      "Epoch: 0438 training time (minutes)= 0.1462 loss= 0.003620772\n",
      "Epoch: 0439 training time (minutes)= 0.1474 loss= 0.003626516\n",
      "Epoch: 0440 training time (minutes)= 0.1464 loss= 0.003637195\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-440 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0441 training time (minutes)= 0.1460 loss= 0.003620645\n",
      "Epoch: 0442 training time (minutes)= 0.1460 loss= 0.003629882\n",
      "Epoch: 0443 training time (minutes)= 0.1461 loss= 0.003630272\n",
      "Epoch: 0444 training time (minutes)= 0.1473 loss= 0.003630423\n",
      "Epoch: 0445 training time (minutes)= 0.1465 loss= 0.003625551\n",
      "Epoch: 0446 training time (minutes)= 0.1462 loss= 0.003613739\n",
      "Epoch: 0447 training time (minutes)= 0.1462 loss= 0.003629072\n",
      "Epoch: 0448 training time (minutes)= 0.1465 loss= 0.003638783\n",
      "Epoch: 0449 training time (minutes)= 0.1467 loss= 0.003639270\n",
      "Epoch: 0450 training time (minutes)= 0.1470 loss= 0.003619494\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-450 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0451 training time (minutes)= 0.1461 loss= 0.003630472\n",
      "Epoch: 0452 training time (minutes)= 0.1460 loss= 0.003626220\n",
      "Epoch: 0453 training time (minutes)= 0.1467 loss= 0.003629853\n",
      "Epoch: 0454 training time (minutes)= 0.1461 loss= 0.003645380\n",
      "Epoch: 0455 training time (minutes)= 0.1473 loss= 0.003624235\n",
      "Epoch: 0456 training time (minutes)= 0.1458 loss= 0.003622099\n",
      "Epoch: 0457 training time (minutes)= 0.1477 loss= 0.003621669\n",
      "Epoch: 0458 training time (minutes)= 0.1461 loss= 0.003624879\n",
      "Epoch: 0459 training time (minutes)= 0.1469 loss= 0.003623893\n",
      "Epoch: 0460 training time (minutes)= 0.1462 loss= 0.003620655\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-460 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0461 training time (minutes)= 0.1462 loss= 0.003634166\n",
      "Epoch: 0462 training time (minutes)= 0.1462 loss= 0.003632716\n",
      "Epoch: 0463 training time (minutes)= 0.1465 loss= 0.003630775\n",
      "Epoch: 0464 training time (minutes)= 0.1463 loss= 0.003632854\n",
      "Epoch: 0465 training time (minutes)= 0.1460 loss= 0.003629725\n",
      "Epoch: 0466 training time (minutes)= 0.1466 loss= 0.003626185\n",
      "Epoch: 0467 training time (minutes)= 0.1473 loss= 0.003611984\n",
      "Epoch: 0468 training time (minutes)= 0.1460 loss= 0.003625143\n",
      "Epoch: 0469 training time (minutes)= 0.1466 loss= 0.003630864\n",
      "Epoch: 0470 training time (minutes)= 0.1458 loss= 0.003620630\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-470 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0471 training time (minutes)= 0.1471 loss= 0.003614969\n",
      "Epoch: 0472 training time (minutes)= 0.1463 loss= 0.003621258\n",
      "Epoch: 0473 training time (minutes)= 0.1472 loss= 0.003627453\n",
      "Epoch: 0474 training time (minutes)= 0.1462 loss= 0.003625986\n",
      "Epoch: 0475 training time (minutes)= 0.1460 loss= 0.003618288\n",
      "Epoch: 0476 training time (minutes)= 0.1470 loss= 0.003634828\n",
      "Epoch: 0477 training time (minutes)= 0.1460 loss= 0.003617136\n",
      "Epoch: 0478 training time (minutes)= 0.1472 loss= 0.003638499\n",
      "Epoch: 0479 training time (minutes)= 0.1469 loss= 0.003615318\n",
      "Epoch: 0480 training time (minutes)= 0.1463 loss= 0.003634426\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-480 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0481 training time (minutes)= 0.1460 loss= 0.003629110\n",
      "Epoch: 0482 training time (minutes)= 0.1462 loss= 0.003625749\n",
      "Epoch: 0483 training time (minutes)= 0.1460 loss= 0.003625124\n",
      "Epoch: 0484 training time (minutes)= 0.1473 loss= 0.003627329\n",
      "Epoch: 0485 training time (minutes)= 0.1469 loss= 0.003619269\n",
      "Epoch: 0486 training time (minutes)= 0.1464 loss= 0.003627426\n",
      "Epoch: 0487 training time (minutes)= 0.1462 loss= 0.003617908\n",
      "Epoch: 0488 training time (minutes)= 0.1465 loss= 0.003635421\n",
      "Epoch: 0489 training time (minutes)= 0.1465 loss= 0.003633216\n",
      "Epoch: 0490 training time (minutes)= 0.1463 loss= 0.003621336\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-490 is not in all_model_checkpoint_paths. Manually adding it.\n",
      "Epoch: 0491 training time (minutes)= 0.1457 loss= 0.003639923\n",
      "Epoch: 0492 training time (minutes)= 0.1473 loss= 0.003629468\n",
      "Epoch: 0493 training time (minutes)= 0.1462 loss= 0.003629496\n",
      "Epoch: 0494 training time (minutes)= 0.1459 loss= 0.003625159\n",
      "Epoch: 0495 training time (minutes)= 0.1457 loss= 0.003615960\n",
      "Epoch: 0496 training time (minutes)= 0.1464 loss= 0.003628085\n",
      "Epoch: 0497 training time (minutes)= 0.1470 loss= 0.003620029\n",
      "Epoch: 0498 training time (minutes)= 0.1479 loss= 0.003625943\n",
      "Epoch: 0499 training time (minutes)= 0.1459 loss= 0.003620781\n",
      "Epoch: 0500 training time (minutes)= 0.1458 loss= 0.003628491\n",
      "INFO:tensorflow:./experiments/single_class_vae_test_summary/models.ckpt-500 is not in all_model_checkpoint_paths. Manually adding it.\n"
     ]
    }
   ],
   "source": [
    "buf_size = 1 # Make 'training_stats' file to flush each output line regarding training.\n",
    "fout = open(osp.join(conf.train_dir, 'train_stats.txt'), 'a', buf_size)\n",
    "train_stats = vae.train(pc_dataset, conf, log_file=fout)\n",
    "fout.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-06-15T04:56:48.961876Z",
     "start_time": "2019-06-15T04:56:48.701773Z"
    }
   },
   "outputs": [],
   "source": [
    "feed_pc, feed_model_names, _ = pc_dataset.next_batch(10)\n",
    "reconstructions = vae.reconstruct(feed_pc)[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-06-15T04:56:49.452526Z",
     "start_time": "2019-06-15T04:56:48.964180Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAASUAAAEeCAYAAADM2gMZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjAsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+17YcXAAAgAElEQVR4nOy9eZwcdZ3//6zq6vuenjNzZjK575BAAnLITYQoKogiIsfCKiqrq4i6sv5w9evu8tVV+eriCohgCIJgUDErNxJCQiB3Mlfmvmd6enr6Pqrq90dnmplkJpnJXN2hno9HP2a6urqquqvq1e/3+/P+vN+CqqpoaGhoZAribB+AhoaGxnA0UdLQ0MgoNFHS0NDIKDRR0tDQyCg0UdLQ0MgoNFHS0NDIKKRTvK7lC2hoaEwHwlgvaJaShoZGRqGJkoaGRkahiZKGhkZGoYmShoZGRqGJkoaGRkahiZKGhkZGoYmShoZGRqGJkoaGRkahiZKGhkZGoYmShoZGRqGJkoaGRkahiZKGhkZGoYmShoZGRqGJkoaGRkahiZKGhkZGoYmShoZGRqGJkoaGRkahiZKGhkZGoYmShoZGRqGJkoaGRkahiZKGhkZGoYmShoZGRqGJkoaGRkahiZKGhkZGoYmShoZGRqGJkoaGRkahiZKGhkZGoYmShoZGRqGJkoaGRkahiZKGhkZGoYmShoZGRqGJkoaGRkahiZKGhkZGoYmShoZGRqGJkoaGRkahiZKGhkZGoYmShoZGRqGJkoaGRkahiZKGhkZGoYmShoZGRqGJkoaGRkahiZKGhkZGIc32AWhojBdVVdN/j39IkoQgCAiCMMtHqTFZNFHSmDGOF5Xh/w9/KIoy6rKxkGUZs9mcFiVRFNOPIZHSxCp70ERJY0KMJSZDy44XlOOfj0VtbS3FxcVYrVZgpIgMCc3xy4eQZXmElSTLMrIsj3h/IBDA5XKlhUqzqjIXTZQ+gJzMUgHSQjL0NxgMotPpMBgMYwqLqqon3OTDrZRTCYGiKOh0OnQ63Wl/rrGsIlVVOXjwIOvXrx+xfLg1NVysNGYXTZSykFO5QUOuznArZfiyU237eCulq6sLp9OJx+PJyht36HiHrC14//tKJpMnrKu5f7OLJkqzxFS5QUePHmXevHknbHu0m+lUbtBYnInuzmjfz9D3erz75/V6ycnJQa/Xa+7fDKCJ0iQYjxs0WtD2ZPGVibpBfr9/Ui7PeDlZPOhMYSyrqKmpCYfDccJyzf2bHj7QojSaoAx/frwbNPR3aNnJOHjwIPPnz8doNJ5grWi/tFPLTAjmcCtzaJ+a+zc9ZL0oTcQNUhQFv9+P3W4/5WjQ0PtP1w0aWmcmrJgzhWy6ccfr/qmqSnd3N3PmzBkhVNqP0thklSg1NTWxbds2brzxxnENMx/vCqmqSm1tLWeddda0XxjaBXfmMJpLPRqjCZWiKDQ3N1NUVHSCVaW5f6OTVdNMBgYG2L59e9p1Gm6NjPaQJGnE8+HWzUxcAB+EOMxUcaZ+V0OCNppbN+T+JRIJYrEY0WiUaDRKPB4nmUyOsPI/SGSVpaTT6VAUZYRvPxGGLoSZQPvFO3MYr6U0kfdOZPSvp6eHgoKC9A/rme7+ZZ0oDT9ZE2WmT+IH7RdOY3KMFRRvaGggPz//A+P+faBEaSY5Ey4OjfeZaktpohzvHZzJo39ZJUqSJJ1yKD6T0CylM4PJnsfpEISJuH+CIDA4OIjb7c4K9y+rAt06ne6EX4ZMJVNP+OlwJn2W02UyltJMcbKA+qFDh0gmk8Tj8XRQPRaLkUgkRgTVM4GsspSyyX2DM8tSOpM+y0TJREtpIvseciHHk/wpiiJGo3GmD3MEmihNE5p1MXmGEl4TiUT6MTSEfvz/TqcTl8uF0+nEYDDM9qGnyRQxP9nUpSEy5VizSpQkScoaUYLMOcmzzXjEZXBwkJqaGhRFSddHgtQvt16vTz8kSUKv12OxWNL/D7kf/f39NDY2kkgksFgsOJ1OnE4nDodj0sc/24HuyTCR5M/ZPlbIMlHSLKXZZSKWSyKRmJC4BAIB5s6di91uHzFqNB5isRhWq5WioqL0cYbDYfx+Pz09PdTX1xMMBtm3b98IofqgTAHKth9HTZSmkZm6GCb6azxRcYlEIoiiSFNT07gsl6HHRMSlvb0do9E4JUIhCAJWqxWr1cqcOXMA2L59O5WVlfj9fjo6OqiurkZVVRwOBw6HA6fTmRbEsbZ5OmSTpZQpaKI0TczERTAUrIxGo2khmSrLZfhrzc3N2O128vLypv0zTZQh4T/V9y0IAna7HbvdTklJCZCalxYIBPD7/bS0tBAIBBBFcYRQ2Wy2SR/jbAuCJkrTyJkaU5qMWxSJRKitrcVgMEyp5TKcTL6gxytKoyGKYtqdGyKZTKaFqqGhgVAoRCgUorq6Oi1UFotl3PvLBNdpIqKUCec6q0Qp0/OUjheXwcHBdC7IWOIyhE6nOy23aPfu3Sxfvvy05wOeCUzljSRJEm63G7fbnV62fft2cnNz8fv9dHd3Ew6HMRgMaUFzOp0n1M2aruM7HTRLaRqZKfdtKgK64XAYAJvNNuWWi8bMIggCubm55ObmppfFYjH8fj9+v5+2tjai0Sgmk2mEUJ2s0cJMoonSNCKK4oRO8mjikkgkaG9vn9KYy2jiUldXR15eHi6Xa0q/g9kgmy7omcJoNJKfn09+fj5AOrbn9/vp7++nqamJeDyOwWAgHo/j9XpxOBzo9foZP1ZNlKaBoSHe/v5+RFFk27ZtrFmz5rQCukNz52bCcpmpMq2Z8Gv8QUcQBMxmM2azmcLCQiB1/vv6+jh69Gg6NUGWZaxW64ynJmiiBGzbto27774bWZa5/fbbuffee0dd75lnnuG6667jnXfeYe3ataOuc+jQIb74xS+Sk5PDwMAAf/3rX1m0aNGolstQYbexToLP56O4uHjKPudYZNNFMB6yWfhm69gFQcBkMmGxWFi8eDGQGvELhUInpCbY7fa0UJ0sNeF0yLZzNy2iJMsyd911Fy+++CIlJSWsW7eOTZs2sWTJkhHrBQIBfvazn3HOOeecdHvLli3jjTfeAGDNmjX89Kc/nY7DnnKy7WLQmH5EURxXaoIgCGlLSpblrM8qnwjTIkq7du2iqqqKyspKAG644Qa2bt16gih997vf5Z577uGBBx6YjsOYVWbyItDE79TM1k05HkE4VWpCLBZjx44d6HS6dFrCRFITNFEilZ1bWlqafl5SUsLOnTtHrLNnzx5aW1u5+uqrZ1yUZuokzVRM6Uxhuj7LbIv26Xyu4akJHR0dnHvuuek0k9FSE4bEymQynbA/TZQY/SI4vsPDV7/6VX7zm99Mx+4zgky/CBRVZW/bIGVuM7m22Z9VP9vCMV1M5efS6/V4PB48Hk962VBqwuDgIO3t7aOmJmiiRMoyam1tTT9va2tLz0GCVCzp4MGDXHTRRQB0dXWxadMmnn/++TGD3VPF8Poy081s3WgJWeFgR4AF+VasxtFP8as1Xv7P3+qp8Fj49Y0rZvgIP1hM57U2ntSESCSCqqo0NDSkrarZSE0YL9MiSuvWraOuro7GxkaKi4vZsmULmzdvTr/udDrp6+tLP7/ooot44IEHpl2QYOaG0Gfql2m0z7Pl3Q4e3dHGBVU5fO8jC9LLh6yj8hwz+XYDBkmkPMd8wjYjCZntR32sKLaTb5/dgl9TwWxaCpO91ib6/tFSE3w+Hy0tLZhMpjFTE+x2O5KUGRlC03IUkiTx4IMPcsUVVyDLMrfeeitLly7lvvvuY+3atWzatGk6dptxzJalVGA3IokCRc6RgvJyTR///rejzMu18NBnVrD1zrVI4ok369PvdfLrt1pZW+bkx59YcsLrGuNnsoI4VYKq1+uZM2dO2mM5PjVhcHAQVVUpKytj3rx5k97fZJg2ady4cSMbN24csez+++8fdd3XXnttug7jBM4USykcl7n7mUMEA2EeWiiPMMcvX5zHh+blYNaPzHXJsxkxSCJlx6wjvW70XJjFhTY8Vj1nlTlHfV1jZpnstTSasI2VmpAJjTkyw16bQWYyA3o69+OPJGjzRYnHFfyRBA6LKf1apz+KIAhYDCMtpfIcM0/ftgaL4eQZxOvKXTx3x/uudDYFSUdjtt23yVpKM3UMQ7MfZpsP7tTyaWa6b4Iip4n/89FF/OMqMwXD4j49gRh3PnmAO588gD+SSC8/0D7IZx7dw7eer5nQsXUNRokmM6fTxQeNqRDUbBt9+8CJ0pliKQGsKnEw3z3S2JVEAb1OxCSJ6IbFi1LCAqHY+Eu/7G8f5JbH9/P913qm7JhPRjbdOONlKgThgyZKmvuWpfs50hXk31+sZ40rwfBE+RyrgcdvXgUwwk1bV+7iFzcsI29YTtLhzgChuMy68pNXMhA49QX9h72d9AzGue3cUgxS5v3WZbP7NhXHrolSBpNNJ+dkvNsyQF1PmGQU7jxO/MaKGVXmWtL/D0aTfP3ZIyiqyoOfWkZVnvWE9VcUO3j0ppX4eztOeizhuMx//70FWVHZUOliVUlmBciz2fWcqphSNpF1oiQIAoqiTGoW9ZlgKX10RSFmvQ5ruPOE1/50oJv63hB3fqh8TIEy60Xm51sZiCRGWE/HU+gwEvOd/Lu2GHR8fn0JXYMxFhVMvqb1mUQmWEqa+zbNDFWfPF1ROlPqD9lNEp9YXcS+fal4T013kHu3VnNepZuXqvuIJRXOqXBxbmXOiPc1ecMMRJIc7Q3R5A3z9cvm4TRPfsTlxnWTKwcznedktm9KLaY0MTLP+T8Fk20ecKbkKQ3fj6qqNPdH8EeS7G724zTrSSgqr9V6R6wbSyp85elDfGtrNa/VefFFkhzqCJx0+8FYklZ/gqSS/UI+G8x0RvdY28gmUcpaSykbmIoL6pk9nTyxq51vXFbJecdZPXvb/HQEZBYAFy/MxaTXsbvZx5PvdqIoKrua/SPW1+sEytxmugZj3HVBOW0DMeY4jdT1BDnaF+aCKs8Id6+5P8K3t1ZzuCtAidPAU7fnnTLHaSZQVZVEIkE8Hk83ZtDr9RiNmTklRnPfJkZWitJkOppkm6W0p9VPfzjOwY7ACFGq6Q7yra01yIk465fL5JgELqjKYUG+hYSi4rEaOH9eDg19YR56s5mPrShgQ2UOP79+KSogCgLhhMI/P3uEQCyJAHjPT6TdsHdb/HznT9UMRpKoaiqYHUsq0yJKQ626g8EgoVCIeDyeFp2hEsfxeHxEmePhZYwhNak7mUzicDhwuVy43e4JtUKaLqbCUtJEKcOZrKWUbXlKX7ukkg8vyOXcSveI5R6rAadZwmCQ0evev+AKHSa+eVlV+vn/bG/hjfp+wnGFDZU5qX7xx14zHhu6z7Xq0etEls2xp99n0AkICKwodrCpykC+w4zbMr7YkyzLJwjKaCIzvGdbJBKhs7MTk8mU7mFnNpvT/xsMhlFrWSuKQjKZxGazoSgKg4OD+Hw+ampqCIfDWK3WdPXGyQ6QnA6ZIAgTuQ5n+1ghC0VJkqSMmJ9zKqZK/DxWA5cuyj1h+f9sb0ESBW5eYiKakNl+uId15S481pEjadcsLyAQTTLXYyYQTbJ1fxcOk8Q1ywtYWmRn8y2rsRl1J8yDW17s4HefX4XVoKOjrQWdDvx+PzVdg+xuDXBBqQmjKI8Qm6HPO9TDbrigGAwGrFZreplerx8hMvv27WP+/PmTcsFEUcTlcuFyuZg7dy6qqhIKhejt7SUSibBjxw4MBkO6eJrL5ZqRov2TYapELZv6AmadKH3QLKWxtvt20wDBWJKukIH9u7t4dn8vH17g4Z8vqcQXTlDhSeUkmfUig9EEP321i7/X97O/I4AArC4yYZNSsZke34lWzHAXOZlMotfriUQi/PzNAeq8CUQxl0+tLjihG0wmIQhCuu+e1+tl7dq1RKNRfD4f3d3d1NbWIghC2t1zu90YDFNb8C5TUgKyiawUpQ9STGkoljPcdRIEgf/42CLaBqJ4ou3kYuQNq54luRL/+Lu99IUS/NN6N1UukZ++7WN3V5xoEnY1eLEaoMotMdDdRsRgYE93gr3dCW5em09hvjtt2UiSlP4MLS0tGAwGCgsL+VS8mz8d7OHS5aW43ScmXE6G6SyHO7Rtk8lEUVERRUVFACQSCQYGBvD5fDQ3N6fjUkMiZTafWG9qomgpARMjK0VpspbSTHE64ifLctpiCUWifPmPjQRjSb5zvgeP8f3Gmoqi4AAGIxHybDI/ON9KVEkSjieRVYVdXQlcrlxWzdVT6+/BIivEZZWBuIrF4Wb5slTLn+9t30NTf4Sz54ssqjh148yrlxdw9fKCCX+uTEWv15OXl0deXh6QilH5/X4GBgY4cuQIkUiEaDRKc3Mzbrcbu90+oWsoU1ICMs2KPRlZJ0qTzVOCmc3oHt4o8/iA73CXaShOJopiOu6CKJGQFVQErHY7xXk2JEni3j83UNcb5r8/vYxARwOVlZVYLBZ+uK2O/qjKogIHL9UHqeuX2XzLam4/fy7VXUHaBqLsaPTx8ZWF6eP84gXlbN3XTVVeyt1r9UUIx2Vequ7jaF+Y7141f9q/q0xCFMW0lTQUl3rzzTcRBIGmpiYCgQBGozG9jtPpPGVcKhPcN81SmkZmM6Y0XGTGGlEa3qlXVVV6e3tHBHcNBgNGoxG73T4iGDzWL9ljZRVEEjKFjlS9pKSSiifFkgpb93Vzcd77Iruq1MmuZj+XL87FatRxYVUqhUAUBJYU2VlSZGdNqZNn93ZhkEQWFthwW/Qc6Axwz3PVPH7zSr70+4MkZJXYsXIlNd1BikSVg10RsMQodLwfiN7Z5KMvGOeqpfmIo1z0fcE4RknEbjr1ZZapGd2CIKDT6SgrK6OsrAyASCSCz+ejq6uLmpqadIB9KHg+PC41FZ9LE6UMRxTFSY2+DRel4Ul4ownL0P/JZDL9HkmSRowqDXXpHS46kiTR29tLLBZLX8ini9OsHzENRBIFPr6qkNdqvVy80EOoO8idv68mlFB58Pql/PHOtTR6w5xflZMWskA0iSiA1Sjxx31dPPp2Kwc7Avzs+qXkWPSEY0mCsST//UYzHqsBfyTJF84v4y8He2joC9MlR/n521487/oosBuwmyS+edk87vtzLbKiUuY2s7zYMeK4W30RvrjlIBaDjt/evCqdfpBtjCYqQzWwh0rLDsWl+vv7aWxsRJZlnE4nbrebWCw2qRG+qRJrTZSmEUmSxgx0K4pySitmcHCQvr6+dGvv4aNHQyIzNGIzWtB3IkzXr/+9l1dx7+WpXKQ/Nyoc6Q6TkFWe3dPFNSsK+PJTh5B0AptvWU04nuTm3+7HrBf5zedWcuF8D/s7AmkXrshp4pb1Jfz8jWa2Vffxnx9bzNpyJ/vbA9T2hDnaF+GbH3JjkATy7QZqukMIAsiKyoXzPfQEYpR7LCc73KznVOf++LiULMvpfKmOjo60aA25fDabbdzXk+a+TSHbtm3j7rvvRpZlbr/9du69994Rr//4xz/m17/+NZIkkZeXxyOPPEJ5efmY23v99ddpaWnh6NGj/OQnP+GjH/0oZWVlxOPxEfGY4wVmKAlvaFlra+sJvbOmg5ka5atw6vj8ukIO9kQ4ryqVwa2XBGxGCUkUeOFgDy2+CAJw/wu1/PvHFvOz65aO2MbnN5TybusgR/vC2E1SOpHy2pUFzHGaWJib5H+uLWdOUSG/e6eDPJuBHKuBb19RNfpBAaVuM4/etBKDJM66lTTZltcTRafTpQVIr9ejKAo5OTn4fD4aGhoIBoOYTKYRcamx3HctJWCKkGWZu+66ixdffJGSkhLWrVvHpk2bRrTtXr16Nbt378ZisfDLX/6Se+65h6eeemrMbe7duxdZljGbzaxcuZLly5dTUFBwQhLeqcj0ZLmJ8lxNhPZoku9vWsy3tlbT4ovy9UsquXihh33tgzyyow1RAEWFV2v7aegLsXV/D5cvzktncIfiMl+4oJya7hALC1LD/EZJ5MsXzQWgtbUVnShwuCvEk7s7MOlFLl2cN2onlOFkQpPLTEAQhHSR/uPjUh0dHRw5cmSEkLlcrhG1sjVLaQrYtWsXVVVVVFZWAnDDDTewdevWEaL04Q9/OP3/+vXreeKJJ066zbvvvhuAmpoaNmzYcFKr6mScKZUnh9jeFiesJDjcGaTUbaZzMEaxy4ReJ9IfiiPpYHmRDZ1O5IJ5ObxS28/Tezpp6Avz4KeW8fiuNh5+qxXx2PHmWPRctGB0K9JpljBIIoUOI6fQozFp6Avz9/p+rl6ef0L2eSYyXcmPx8el4vF4Oi519OhRFEVJi9Nk8vJOdgyjkQniNS2i1N7eTmlpafp5SUkJO3fuHHP9hx9+mKuuumpc255s8mSm88hbrWxv8PGvG+enWyGNRTQhc+dqK2GDi/Pmubmgyk3nYJwipxF/JMFvd7aTZzPy408uTTeVbPKGafSGueZYrtHT73biCycodprItxuoyrfQNhDhd7s62Lg0j+XFDl6oHeTN5jDfu8bB729bg6QTRh1tGw//9Woj7zQPkJAVbj9vcoMA2cJ4bnSDwTCi060sy/j9ftrb2+nv72f79u3Y7XZcLhc5OTlYrdYZjUvNJNMiSqNZCGN9KU888QS7d+/m9ddfH9e2dTrdlI2+TSenu58Xq3tpH4hyuCswqijFkgr/e7iXpKzwq+0trMgV+O5HPOh1Ig+/1cL/bG/lysW53HpuGf3hBAKpRgKKqpKUVSo8Fn64aRGQOk+VeRYGogn+5aqqdEG4X77RxLN7O3mn2ccPNi3ijaYQDf1x9rQNMm+UsrkTYdOKAhKywvlVOadeeRh9wTiKqp5Wx97JxpRmo0WSTqcjJycHVVUxGAwsWLCAQCCAz+ejvr6eYDCI2WwmJycHl8s17XGpmWRaRKmkpITW1tb087a2trSZOpyXXnqJH/zgB7z++uvjnoiZLUXeTpfvfWQB1d1BLpo/ugv1Wp2XB15uwKwXkRWVntD7n+VgR6oRwGt1Xv7xgnJyrQaWzbHjMkt89Q+HaegLc99V87EadSzIt7G9wcfhziAVORYK7Ebqe0NU5Vm5amk+fznUQ11vmHueO8JnljsIxFUuXeghISuE4zJP7+lkw1w3S4vsox7nWFy6MJdLF544wfhkN44/kuAfNu9HUVT+58YVWdVKfKpETRAEHA4HDoeD8vJyVFVNx6Xa29tPiEu53e4Rbbg/8KK0bt066urqaGxspLi4mC1btrB58+YR6+zZs4c777yTbdu2pU3W8ZAtRd5O9yJYWGBj4UnqXC8ptLEg38o5FU7WlrlIeNvSr503L4e/H+3HbTVwsCNIQ1+Y+t4QF83PoX0gSjiucNdTB4gmVT65uohPrCpErxPJs+m5+5lDqKrAI59dQYXHQqHDRJsvSudglEfeTfDti4q448mDGHQCVy7J45G3Wnmrwccjn11J92CMJ3a1cdniPFYcl680FYiCgCQKyAgj2kYNMd033Gze0GOJmiAIWCwWLBYLxcWpGljxeByfz4fX6+Xo0aOoqorT6SQSieBwTP15mS6mRZQkSeLBBx/kiiuuQJZlbr31VpYuXcp9993H2rVr2bRpE9/4xjcIBoNcd911AJSVlfH888+fctvZMiEXpmcottRt5tc3rkBVVfrDCboD71+wC/Kt5NuMrC5xcH5VDqtK7Oxs8rP5nQ4evH4Z+9r9fOO5ahQVqruCLCmy8+wdZxFPyvzD5gOIgpAu4vbdq+bz5lEvLxzqJRyJASrhuExEgFUlTs4ud3HFklRezl8O9fDM3i7q+8L88oblE/o8CVk5ZXzKbpJ49KaVKCo4xpEdfjwznRIwVfue6PsNBgMFBQUUFKTihclkEr/fT21tLY2NjTQ0NGC329OW1ETiUjPJtOUpbdy4kY0bN45Ydv/996f/f+mll05ru9lSumS6T/aTuzt49O02Li+XuLMk9XlWFDt47s61qQJtgsDdH67ku3+uYWmRjQKHkYqoFYdRRzip8JWLKgjHUwXi3mnx4w3GuWxxLnaTRHN/hOb+CJ9eW8wNZxXT0tpKXBWp9JipyrOybI6d/xqW63TZolyO9oa4etn4LV4AbyjOPz55AJtR4guLT35ObMbZy/Od7Vn+p4skSXg8HpxOJ0VFRTidznRcqq6ujlAohMViSYtUplhTWZnRnQ2iBJP7lT3YEaBtIMpli3JHdVn80dTUl1Di/X0kZIVXavpwmiXOrcyhoS9Eqy/KE+90cNM5JRQ7jZxV5sJmksizG7nhkffItxvZMNdFKJ6kJxAH4BvPHcEbivOvGxdwQVUOkiiwvzPKgY4Adb1hvnRRxQjrptRt5gfHgucTIRBNpkuzJJTMzB+b7Vn+U5nRLYoiTqcTp9NJRUUFqqoSDofx+Xy0trYSCARYvXo1Ltepq0VMJ1knSpMdfZspJnMhKarKN/94hGhCwWWWWD93ZCnchKywrtzJhrkudIOpRpG/fKOJJ3d3oKhgNep48pbVPPp2G6FYEr1O4G9H+ti0oiBt4RzuDJCUVQYiCep7w8gKFBwLIK8qcbCn1U+Z20QsqVDfH2dZgYmbzimhKs8yqruVkBVeq/WiqHDxQs8JlSxHo8Jj4cefWIJZr8PbXH3a39epmM3Rp0wp8jZWXMpqtWK1WikpKQEYkbQ5W2SlKJ3pMSVRELhwvoe6ntCIrrZDPLqjlS3vdnLlkjw+WibS4Y/xyI5WokmVXKue8+flYDNJ6EQBu0ki12ZIZ2oPIekEPnVWEVctyeNwV4iGvjCLClPrDJ8+8qO/1fO/h7q5domLr1wxcnrKcJ7c3cFPX21EEOBbsSo+sbpoXJ91KKjvbR7X6lNKTyDGL15r4IIFuVy8MG/UdTJhOF3L6M5wPigxpXsumzfma4UOIzpRoMBuAJJYDSJGSUdcTnLlkjy+fNFcHn2rlQK7gX/68FzWlafM8Z2N/bxc4+WW9SV89jd7iScVTJLIeVU5DEYT/OrNVj40z4PNqOPlGi8VOWZMko5QQsEfff+H4ED7IC6LnlK3me7BGK/U9vHwW60IgFmvG1VIh2joC9PQF+bC+TnjsqaOR1VVtjf4sBklVpVMLgbyRl0ffznYzZGuwLnMYJcAACAASURBVJiiNFky2VLKVLJOlKaiyNtMMV3it2lFIZcszMVi0FFbW4vDJPH729ewq2mADy/w8Ny+Lh7a3oKiqJS6zKwrdxGOy9z9hyOEYjIvHunFKIkkZJU8u5FXqvvoHoyTbzcQjifZ2+rnX/5Ug1EncNeF5aCqvFAboO6xvdx1QTnfe6EOoyTy359exp1PHqA/lEAU4NpVhXznypMXhfvW1mp6AjFEYT4Xj5KvdCpqe0Lc/0IdOlHgqdvWYDOc/mTfDy/Io7YnyPnzxp6cPRU39GwkX2YzWSdKU2EpzURMaqp+mRRV5Q97OjFIIpuWF6S3ax02GrW90U9E1XPtqkJEQWBZkY0Cu5E8W2oZQCQhox7rchuIJfn+NQtZV+4i12bk/73ehMuiR1ZUbnl8P7edW4KqqsiqQKnLTInDQE8oSW8wjiCkmhEUOU0YJRG9KJJnM/CF88vYUHnqLO31c1280+w/7czwQoeRCk+q3VMqfeHUN+1YwuKxGfj2lQtP6zjGSyYFusdDJlhUH0hRyvSY0nAa+8I89GYLgiCwvsJNgWNkNnMkqfLjN9qRVYEKj5mzylw0eCMkZAWnWU84LtM2EOG9lkE+dVYRf9zXzWWL87hscX56lv8/nFfGeZVuvv18NR3+GK/UePn+NQtRVJU2f5TOYJJSp55vb1zIqhInT922Bp2Ymv92+7ml/NerjfSFEuMatv/qxZWT+j6cZj2/+syK9HNlBtqJZ2tKwBCa+zbNnOkZ3cdTlmNOteSWRPLsJ86qN0sCK4usvNk4yC//3sKvb3Rh1osoKuxu9vP1viNU5JjZ1x7gjg+V8cJdZ48Qj4a+MDkWPatKnXzlw3P5jxcbKHGb2bg0lXNU0x2kMsfAxZV2VpU4gVT1y6HP1zoQJZpUONIVPO3PmFRUqr1JKuIy7gmM/gxVDk0mk1itJ7e8ZjN5crLv12JKGY5erycWi532+7PNUtLrxJMWUxMEgY8uzeFwTzRtNVy5JJ/5eVa+90ItdqOeC+fn4A0lWFXiwGaUONobYnuDjyKHkf98qYE8u4H7rprPmlInf/iHs3CYJBRV5eG3WokmZH5wWREGKZVHdN+fa3i11su/blzApYty+ezZxSwssFHsNHLPc0dYP9fFx1edfOStJxDDKInpMr/P7u3kF3uj7A008u/XLiaZTBKPx0/6GN74UlVVVFXF5XLh8XjIyckZMbQ92fMwm4FqTZSygGxx32ZS/Ha3BVk2x8Y9l74/Yjcvz8o9l1Xx45cb6PRH8YUT/OVAN6FYkn965jBJWcFlMeALJ7AZdfzT04ewGiWevHU1OlGgezDG795pxx9JEgg4+PxZefgjCf56qIdoUuWJna1cuigXs17HBVU5vHCohx2NPhq94VFFaaiVd1NfgH/e2oBBBz+6NA9BSRL1BhFR0EX72bVrV7oO+vCHw+EY8XxoRryiKOnif0PzvhobGwHSFUazOdCsiVIWkC2iNFO81Rbht7v9WAw6Os6OkWc30tAX4q2Gfp7c3UmrL4r3WNmP/nCCFl8UvU7EYtDhMkuIAtz5oVJ+8fdWcqz6dPZ4vt3AqhIH24/209ifyvR2mCRWFjvY2eznQGeQmu4A5U498XicpW6Vaxc7mJ+TGhGMx+PEYrH0udLpdBgMBvwJHaqqIAoiNpsNu8XEJ0v1zHceZu3qlaedvCeK4ogyx4lEgv7+frq6uvB6vSSTSRobG/F4PBPu3TYZMkUQMuEYxktWilI2FHmbKfFrH5Qx6UXmeswsLbITSyp87rf7CESTuMwSTrPEneeXMddjoTzHjEESqcgxs7jQxr/8qYaBziB6nY6vXTKXSFxBVRTebR3g1Zo+PrfSxYpcHRWmKN3d3fT393N9RZzqThAFlYaaI4QdRn5fGwdB5I6z87GajSMsmuHlM4Z4YmGq9ZLVoOObf6ymoS/MHUtO/8Z5bFc7bzb4uf+axVTlp5Ix9Xp9enLqwMAAzc3N6PV6GhsbCQQC2Gw2cnNz8Xg8J+2CO9uiMhWNJGf7M0yUrBSlbCjyNhnaB6JEE/K4hs2vWWBleWkOFywuRhKFdEJkVCdy90UVCIJImdtMlcdINBbh0R2dxBJJCkUrYiICcpKjRxv43ZEYiqoS7DTwh3qZ+gGZv1X3Iooi/3y2nZioJ2p0c9HZbi5aryOaVNnbNohg1LG7uxpQuNPmIX8cnU2GyuDKikpNT4hYQqY/cvo33uv1Ppr7oxzqDKRF6XgkSaKkpISSklS6QzAYxOv1cvjwYaLR6Ih41PC+bZMlU2JK2UTWiVK2TMg93f0EY0m++NRBkrLCL25YTvkYJXGHGmPGEkkWugzEAv20elMu03c3WAhEEsR9LTzwbhydAGsLdLzTLZNQUk0BVhdZ+M7llURVkQKXlfr4Ud6o7+flHivziw3U+XoIJQUcRh06k4UH3u7DH/fjNLVR4jZx2aI8/u/LDZS5TXzxgnIURaV9IEp3IMY5FW4OdwbY3eLn2pWFYzaj1IkCP/3kEnoCcXTeoxP+roa+h3+5opLGgSSXLx5flYLhhfwrKipQFIWBgQG8Xi/Nzc3p7iO5ubno9fpZTwnQYkoZTrakBJwOqqoiqAp2g0gwphAZ9NEW9hKLxdKjTkOdd5sGFf7SmKTZL+MyD/CfVxZjNRuxWCz87kicna0RLl6QSzDRgQrs7Rfxx2V0AggC/GxnP7++sZg/7u4kmhzgM+uK2d8eoN0fZUOlG4MkklRUzip3EYorGHQCSTlJgzdOTyDGjeuKcZgkVhQ7uHZlIW0DEW57Yj+QmiLzo7/VE4zJWA26MefBJWSF+t4wCwus9PafeNN0+qP827Z6qvIs3LqhdERTzuFUeiwsLxvbqjzVTSmKIjk5OeTkpJI/k8kk/f399PT04PV6icfjNDQ04PF4cDgcE7rBZzulYGgbmihNI9kS6B5+EQyNPA09hovM0ENRlHSL6K+fpUfSWzERRxQNOJ1OjEbjiBbff32hlsP9PQgquEQ9JaWlmPSpYfvWwXZiSRWn2YDTJBFNpqoKHOoI0u6PEksqxJMK3YMxfv9eB6G4zEA4znevrKLIZaI8x0xcVnhydwfvNA+w7VAch0lk0/JCtrzbSZ7dSCSh8OwdZ6UrBnisBhYV2BBFgcd3tTEQTlLsMrG23MVfD/Ww5d0OvnZJJSuHVab825E+/uPFo1TmWvjyKHN9D3QE2NPq560GH8/v7+aHH13EhuMqJkwHkiSli/gHg0Hq6uowGo00NzczODiI1WrF4/GQm5uLxXJqdzUT3DdNlKaRTBElRVFOKjLRaJRIJMKuXbvSI0/DH3a7fYTQTLQf3WfPLsFi0FFpjrKiPDctSLGkwuWLU/PiNi7N5+IFHr6w5SCNfRH+3w3L+OmrTayYY+fKpfnYDCJOs0RPIM5z+3pYPsfJufNS1sJtG0pZXeLg3j8eAQEqXEY+s66EAoeJx3a2cf8Ltdx2bikXzvcw59iUk//65BLissrW/V0U2o3cdWEFpW4zP3+tibqeEG83+NKi5I8k2Nnkw2rUUeY28fghL3eXRanIfT8mdEFVDp9YXcQz73XgiyT41Zstpy1KCVklnlQwTLAxZiSeRKfTUVxcTHFxcaqGVSiE1+ulurqaSCSC0+lMj/wdH4+aimtNE6UMZzon5A5v+z2a0MTj8fTInyiKJwiNzWZL/y/LMk1NTSxbtmxajrUy18LXL53H0aNHsRpEfv5aI6G4zJv1/fQG45R7LFy1JA9/NIkkpmpbbzvUQ4c/yo3r5lDoMPLng920D8QQBHCZJdaUOegNxNjV7McoCfzy7y1ctbSAJS6FeR4jR7qD6HUCeVY93lCCR3a08cyeLn5943LueuoQJr1IjkXP/vYA376iilJ3Kh52/ZoiPBaJ69YUkVRUdjT4aOwL83KNlzK3CV84yY6OJAsO9fKFC98XpX3tg7x4pJdLFuVi0IlcuujkE3gVJVWy13ZcDCsUk/nOK30Y3nqHRz+3Zkw38Hgef7uVX7/ZyCcWGllxbGaLIAjYbDZsNhvl5eUoioLf78fr9dLS0pKOR3k8Htxu96x1Qzkebe7bNDLR0behqQhDIuPz+fD7/dTV1Z0gNIIgpNt7GwwGjMZUjMblcqWX6XS6cZ24SCQyY6Mejf0xtu7vJhSXURSVWFKhpT/Ml35/iLcbBxAFlYRi5ME3WgD4ycuNPHHLaoqdJkrdJpq84bQl8evtrbxa66Uqz0KrL8z+dh1mVc8PXutCUVMBqUgi9aOQY9ZTnmMmGE3iDcWRRAHXsRt+eLHMn7zaQPdgnHPmuukPJ/jhtjrsJonVJQ6uWZ5Prs2AGvFz1dKR5UO6BmPEZZWGvgi/+NQyXJaxxSQhK1z7y510BWI89JmVnFX+vkUVTcqEEyoJQSaakMctSm2+MElFpTc89vUmimK6nGxVVRXJZBKfz0dfXx91dXVEo9G0de50Oid802ujb1nAUJ7SwMAAkiSNOQ1hKCB8vNAoioJOpyM3N3dELs1U/0JM1faavGHMet0JE3GHM9dt5LrVRUg6AbtRx562AHvbBvFHUt9BXEnd4EMsPFbM7YGXG2jzRYgnVeLJJLc8vo9FBTbmuExcvjiX+r4w9X1hartlgnGVPJuBRYU2jnQGiCQUPrV2DrefW8q7rX4SskKJy8KPProIbyhOkdOU3l+Rw0RjX4RIQiESl5EV8IWTNPSFuWC+B7NeB8uMactqiKuXFfB24wB/P+rl56838d2rRi+LcrQvjC8Wprk/JSKNfeERopRj0fOdCzxUVlZiM0r89xuNLJvj4ENVY5csAfinS6rYUG7FHu8/6XrDkSSJvLw88vJSArt3715MJhOtra0cOnQIi8WSdvUsFsspr5Opcr0ywQIaL9MmStu2bePuu+9GlmVuv/127r333hGvx2IxPve5z/Huu+/i8Xh46qmnqKioGHN7mzdv5ic/+QkDAwPE43H8fj933HFHOi5zvPs01lCu3++nq6sLt3v6A6aT/YVq9Ib50lOHMEgCm29Znbp5j0MQBCSdwJ3nlxNPKjz8VivrK1zce/k87ti8n2KXiWBMpj+cSL9nfp6Vp97twGmSsBokIIkCRJMKe9oG+cy6OWzZ3cHiAhv1fSH6Iyp6HSwqTDUe+PyGUhYX2lhZ7KBrMMa//+0ooZiMpBOJJhV0osAf9nTyXqufr15cifOYO7W/fZCvXVJJJKHwUk0vC/JtGI/FeJKKyrN7O6nKtzE/34bFoEMnCly+KJf97YPU94R4fGcbN51TgqyovFjdxxynEZdJx9eeq0EnCCwvdnC0N0RV/siROFVVmWOXmJtr5W+He/jNjhZyrAZe+NKGk37/ZoOO1cV2WlsHTvsc6nQ68vPzsdvt6ZrYXq+X2tpawuEwDocjLVJj9T7MJkGZCqZFlGRZ5q677uLFF1+kpKSEdevWsWnTJpYsWZJe5+GHH8btdlNfX8+WLVv45je/yVNPPTXmNq+99lquv/56duzYwZYtW/jRj340HYc+ZUzFhWTW6zBIAnajlC4zcjIOdAR4ek8nkiiwttxJKC4TTSoUOYwjROkH/5vKCTLoBDYuzcNj1XP7eWV85feHqe8Ns69tkL5wgvVz3VyxOJf/fLGeaBLea/ED8Hp9P/dtnM9z+7p46t0OGrxhREHgYMcgNzz8LjpRREXFF07ySo2Xco+J284tZePSfEQBipxGrl1ZxCULPenRuwN9SX6xrxFVTVk2VyzJ457L5rG3fZBwXOZoJEzP7g5uPLuY3S0D/OeLRzHqRR797HJsBh1mg4RR0qHXiTR5w6wqHVn8fuh8nFXm4sL5HlaXzkxx/OE/TMNrYpeVlaGqKoODg/T19bFv3z6SySRut5vc3Nx0M8lsc72mgmkRpV27dlFVVUVlZap2zg033MDWrVtHiNLWrVv53ve+B8AnP/lJvvSlL53UVB2aCpApo2/jYbL7KXQY+d3nV6PXCaOWju0LxjnUE+NsWyrmsaTIxpVL8ih2mSh2mfmnD1fy/b/W0jUYHXX7CVnlj/t7MEkCa8rc6CWRuKzQ4o2gKip/3N9FmduC1aAjJssYJB2rS+y83eTn/77cQDShEIolsRl0hOIKwZiMRZ8qvHbB/Bz2tA5yuCvI4c4Qj3x2JRaDxEvVffxgWx2RuMLrdV6uWpJHOKFQYE41nFRUGIwmaO4PA1DfG0ZVVRbkW/n46kJiSYV5uVaq8izMy7Pisej5zWeXYbNa6QvGqekOnrTYnEkvsqLYyUAkMe7RuOlKfhQEId1dZN68eciynJ5UXF9fjyiK6bbdHo9n0tNNsoVpEaX29nZKS0vTz0tKSti5c+eY60iShNPpxOv1kpt78hGWbBGlqTK5h5pDjsa3tlZT1z3Il0UDn8jPx6zXjajtvbbMSUJWCcVlrlyci8em58ndKUtKFCDXqscfTVJgN2LRiwyEE0QSMtEEWIwSsViSJm+YYqeeYFwh12bgxrNL2NcewCAKqJLIkgInLb4oc5w6KvOsXL4olyd3d7DtcC8fqnTT0Bci327EYkhdagsLrMxxGmnwRtjTOsCORh8A850Cj920kofebOWtRh85ltTQ+v1XL+DxnW08f6CHX73Zwk9eaeSW9aU8dKzQmyzL6HUi9b0hXqru5bo1xWMKTSQuc+Mju6nrCeGy6Fk/N4dVpc6Tfv8zmfw4FOscugfi8TgHDhzA6/XS3t6OyWRK50dNRyPJTHETp0WURjsRx3/g8awzGtkiSjA9ox7xpMKP/laPpBMpzzHT3Bcgz3riaJKqquhEgfVz3bxc08fuFj9rSu2IArgsEusr3PzpQA8qEIpH+Pbz1eh1AgvyrNT2htCLcNXSPPSiyKs1Pciqis0g8tzeTgIxmb5Qyh3sPtYrrlDRc/M5pXz1D4cJxZLYTRJXLcvnCxdUkDPs+ErdZr52yTy++swhVDU1SpdUoHFQZY7TSFWehbebBpiXl0pK9FgNXLTAwwuHemkfiJJUVF6t83LN8vwRo2g/f7WBtxv7EYC7LhpZ3XLIApcVlaSs4jBKXLQglxKXiWA0eUIKwfHMVvKjwWDAarWSl5eHx+NJx6Pq6+sJBoMj4lEmk+nUG8wSpkWUSkpKaG1tTT9va2tjzpw5o65TUlKSbi88lOZ/MrKlccB0iV/bQJS/H01ZF7/93Aqun6/D7TQTSyoc6QpglHS81+qnJxDjr4d6ybPpMUsCC/KtVHeHUVVQlFSd7SFUFQySgNWg5/x5Lup6Q/gjSToGYty4bg41nQMc6Y3xXluAxr4wN58zh8d2prLGh1g+x4E/ksBh0rGowMq3rpjHv71QTzgh892r5pNvN2CQdNR0Bfnh/9YRTarMcRi5bk0hj+/qwGOQsRklekMJDJJAKPb+OV5V4uQ3N63ki08doGswxpHOAE+/18nt55Wl17lhbTGCAEvnOIglZL7xh4P0BOP8/FPLCRzrxGIzSTx68xoSsorFoOPm37wLAjx281ljphvMdvLjcFGzWCxYLBZKS0vT8Siv18uBAwdIJBK4XC5yc3PJyckZtTpDtjAtR75u3Trq6upobGykuLiYLVu2sHnz5hHrbNq0iccee4wNGzbwzDPPcPHFF59xltJ0MNdj5ksXlqPXiRQ6TMR8qbjDr95s5rl93ZgkkYFIgspcC7KqEk2oWI16DnUGCcVlFhfZ+dZllSwrdrCq1MEf9nRRnmPhtg0lXP/IHv7nrSCfWF3Aa7U+mrxhvvdCLbHE+3k6JoNEdXcYg04klnz/PFy9vID7/lyLP5JAUVS+8ORBGrwR7EYd//C7/anGBSoY9ToSskKOReLWc0u5fk0R51d5eObvB7jh0b18YlUhtd1BXqnpo8Jj5iPLCpAVlYfebKbUbeaGNXN44XAvZ1e48EcSfGtrNVaDjmK3hUUFNr735yOsLHbyYnUvqgqfffRdFFnh3vPcLATyjjXc9AbjJBQFSFlQx1PdFQCgYOxMjHExXeVwh8ejKisrkWU5Pam4oaEBQRDSSZzZdr1PiyhJksSDDz7IFVdcgSzL3HrrrSxdupT77ruPtWvXsmnTJm677TZuuukmqqqqyMnJYcuWLePadraI0nTtRxAEPrqi8ITlA5EksaTC+rku+oJxvnxRBd2DMcpzLEgibHm3g+f39+APx5mba+a2J/aj1wk89OkVGCSRUCxJPKmgqPDWUR+3nVvCozvaiEUVjJJIsUWHL6rij8R5pzmBUTfyRrn7mcNIYsoVq+8NMz/fitOsYyCcZGDY1+ASBRYW2PjFp5al4ldJhUKHkdp+ma5gjBcO9XKkO4SsqPx/L9RxVpkTk6Tj7aYBFBVuO7cUh0VPkcNI20CUup4QoXiSuOwl324EBGxGiYsX5tLljyKrEEjKDDPqGAgnMEoij37uLCDV1WQ43YMxvvJUanLxz66dN+uz/MeDTqcbUeQuHo/T399PZ2cnoVAonXrj8Xiw2WwZEz8ajWmz8TZu3MjGjRtHLLv//vvT/5tMJp5++ukJbzdbirzNJNGkwptH+zHoBD65uog1pU46/VG+vbUGURT47edWcu/lVRQ5TTy+q537/lzHe62p4f363hBzcy3U9AT56PICntvfRW8gzis1XuKyis2o43OrXBRadfx05wDeYJJIUkUnnhiATypg1Al8fHUhX76wgsfebueh7aksco9VosBu5L6N81lcaOdAe4BvbU3FsSIJmQqrwL2Xz6PEZebRt1vZfrSfRFLhtzvb+fqllXzlogr0OoEdDT427+5gdYmDT64p4uuXzuW9Fj9b3uuiJxDjkZvWsKLEmQ52+8Jxmjt6sCip0byOgSif+vUuzHodz/3jOSNaVQ1hNerwWA2ogFWv4/RbIsxe4wCDwUBhYSGFhYX4/X6WLFmStqICgQB2uz0tUicrcjcbZJ3jKUlSVhR5m0k30SDCpuUFtPujzD9WGM4gicceqXQCQRBwW/Qoikp1d5CqXAv5dgNlOSY2/fKdYy2SdJj0OvJsBj6ztpidTQM8s6eD/3qrl2KHRNtAkqHbIxKXEXi/65rHKhGOycxxmbl0YS52k57PrJvD240+6vvCLJ/j4OfXp+YBBmNJvvN8DZ2DMTxWPaCSUODKxbn86MUGAtEkX76wgife6aA8x8zXnz3MntZBzpvnJhpXyLXpOdgZpPqv9dz/kSo+taaIPx7oRVVV+kJxDJKIoqgIArgtBpJOIz5fmGA0yff/Uk1fMI5OEIglFKyjuGc2o8Tvbl0LgM/XP2lRmu0qAZBKqRle5C4QCOD1ejl06BDxeByXy0VeXh4lJSWT3tdkyTpRyhb3baYQhFS7o7surBix3GM18MTnVyEKQjpr+iPLCnh+fxe7mgcpsBv4wx1r8YbiRJMKiqJyXqWLYpeZz68vxW6SWDrHzvaGfnqDMXLNEm2DMoKqUuAwoigqvlAcQRCIySreUJJ8m55WX4Q7nzzIrz69nHUVLp64ZTU9gVh6pGwgnGDr/i70koDFIFLiMnHuXDf7G9roCcZ5q8FHLKlgMehYOseGP5LgjXofiqryel0/oiDwtUvm8l+vNBJOKFiMOp7b3829V8zHpNdx8cI8GvtCfGnLfhYX2njgk8vpDyX4/mu95Dij7Gnzo9eJXLuyCPcoo5ZDiMOSVWfb1ZmOoX+Hw4HD4WDu3LnpInfhcHhK93O6fCBFaSaYafHzhuK4zO8X/odURrisqHz/r3UMRBLc/5EFXFDlYW9bgKpjFpXHamBhgY3aniBnV7i5ckl+OjeqwG5gfr4VbyhGlcfI1atK+N077dz5oTIq86w89PcWcqwST73bhQr0BRMY9SKiqvLGUS8em4HKXAv5diNJRaWuJ8QrNX08trONNWVOInGZ5v4I1d1BInGFF6v7+I+PLaZzMMaPX66ndSDOO81+FuRZWFniYHGRjSOdIeY4TSQVlWBM5qndnbxa18+8vCC/ufksfrOjGVVNdQSu6wkRiSf5f9s7ONwTwR1S2bSikGVzHGxaefI2UMOZ7RZL081QkbuheNRs84ETJcjsWdOD0STP7e1iXbmTJUX2cb3nreYAv9zVxAXzcviXYZNWE3Iq1vS3I6m2SHvaBrl5fSlXLy/APWwIfFGBlVZfhN/ubOOBlxqYm2vhJ59Ywnf/XMvOJh+KAjFZockbpsEb4b6/1PHoTSv5t00Lue/PtdhNOnLMEp9eV8zHVhbx+M5WHtvVQV1PmA1zXfz1cC9Li2y8XtfP+fNyWFhgpdRl4mMrChiMJvnh/9ajkIrfLJ1jZ+kcO8/u6aR1II7LJLH51jXpY920PHX+1pY5ebtpAFlVuWShh0sWF/Dvf6vl97vbcVv0/OS65RS5zOxtG+SdtiCiIPKDjy5h/Thaiw8nk1ICPihkXd76ZAPdmZ7R/b+He3jozWb+86WGce+nuidCTyDG3jb/iNf+sKeL7/+1HlkB1FRDAkhZR0Nzzl480os/kuDB65ciCAKRhELnQJSjfSEa+sIICBh08FJ9kK0HepBEAaNO5KvPHObOzQdYU+pArxPxRZK0DcSOuV0ORMCsF3hkRyuHO4PsbR1EEAQMkogkivzpQDdb93fz8VVFFDpMWCXIGzYK9tPrl/Kz65aw+dbVo37muy6swGLQcagzyFc/XMHlSwroDybQiQIrSpycPTeHOU4Tr9X24rFIfHqF4wRBavNFePFID/Hk+zHKnkCMbz13iD/t7wLAF05Q542f9jUz26KWjWSdpZQtgW44vQvy7Ao3Z5X1c/mivFOvfIwci4TNKI0oFwKpia96ncC5lR48VsOoRdJ+s7ONRm+ENWUuHvr0cva1DaLXiZxT4eaBjy+mPxjn6XeaOdQTQSZVsG1liYOfv9ZEQ1+Yrfu7uWVDKU/saifHIqViVAkFUYC3G/0E4ymrtj8UY/NtZ/H9F+rY1z6IzajjnIrUpNhPrCrk4TcbGYi+P2l4MJrk2b1dvNsyyNkVLhwmHQfaqCi2hwAAIABJREFUA2xaUYDTrMdt0eMy65FEENTUPv716kVc01rIh+al3JCjPUGe3dOJJMDa4vcrB1R3Bdh2qJu3G3209Ie554oFfOyYO/d2Yz+v1PRS0x3kmhWFfPuFRroGY7jy+09Z6mQsMt19GyJTxC/rRClbGgec7gkuzzGnR6nGy5ULHCwqzWfpce7ehfM9bJjrHnMuWG8gxjXL8jnUGWDDXDf5diOXLX5fDFcWO1BUlQpLjDuebUJG5Kol+awocbCowMYXtxyk0x/DYZL4/kfm868v1PHM3i7+42OL+eKF5Tz09xaiCRlJJyCrAi9V93HT2cW4LHruOK+MCo+FSEKmNxhHBWq6Q+l913QFONgRYHfzAFv3daAXBcIJhe7uLi4uEWnwxVntlDmvxMDRmiM0iiJ5eXmcU5yH6Vi2+jvNPuJJGZdN//+3d97xbdXn/v8c7Wlb1h7eI3uQHaAkzWCEkjRAEiiQMEIo65beFpofNJ0ETEspt6RhlJFAGQ3cQuAyCgHCyLSTGMh0PBNb05JsSdaWzu8PcU40bckrdnLer5deHjo6S+d8zvN9vs+AVnrmUt+0sxl7mh0Yr5VCIeHTM5axc6ZA88wezCqNlbZRSXiweYIJqTK5cLZCAgZr+2eD806UzsXZNy6bhXlV6Z/iPA4LLXYv6tq6cPl4Fd3uqLnTi3u3HYbLH5vm57BZWDJBhSqlmE65ePrLVrx32Ip758gxVimAJ8KBXhazxkrlItw7vxQ1Hzfhua/awGYRMLtiAkUAuG66HiUyISzuII6a3fj0hB1cFoHZpQWYphcjEAjgg4Mt2PhZB0rz2bi+EpiW70JtbW0sbw8EFhez4Q5zUGsMYrpeBLufxOVTizFWV4AnXz2Mpk4vpEo5CggOVkzXw+tyoqWlBSa7C20BIcq1hRivzcMFWj7YcRn2P5lpAI/Nwp3zylCV1CcuX8jFfQsr6b83LC6CzdGF8do89IezHRIwGn1So1KURoNPaTiIkiSMrhD0Bb03HfjTx034xuhGMELihpl6AMAv/30UFncAfA4bPDYBc3cA6985jkl6KZ68NtZa5LilB75gBEZXCA/8QJVQ+QEAZpYUoEgmQDBCoqWzBzIhB7+/vBRafhBGoxEaIgCrx4NPjjgh4xEoi7TjwAEjXWq4yRqAJxjFcTuJpaVcTJs8Hnx+rPaToyeI3QeOgcMi8O+fzkKeINFSuWqSCp+dsGPnSQds7iA0MgmumqyDTqfDu+8fx3vfmvADRwj2Li/eswMVQh90Oh2EQiEurJDjworshmIsgoAgx2YDg81oiCgfTEadKI2WhNyhoLatC75QFD+okIEgCPzrgBHPfWXCggopNuh1KctTT8lLxykQiEQxo/hMmY5pRfngsgk8fNUYyEQ8nLR68PsPTqI0rvnlb5ZU4Zt2F9rNNrQ7vRCK7DhwqhtKAQkJO4JAIID7JkZwwh7C3zpJCNlR5Ied6O4WgM/nQyKRQKkQgM/zQlkoxOzZkxP2T1kUxOenvoGAy8HbTR7wlE5M1ufhZ28dRZ6AAw4rFl3NZRFw+8P4x642lCtE+PEULVZM02HFNB3eqGvH101O7Gm243BHN+6/tAqzy2SobevC4qmlsAY7YHR6wGMTdKCgXC6HSqVCQUFBVjfs2bZ0BgJjKQ0D5+vwzdETxIb3TiBKAk9fNxEVSjEdXZ3uknvnGzOe/foU7p1fiuVTtVg+9UxcjtsfRqlciGumqmGQshEIeGG0OVDAj1kUR44cQSAQQDgcxv7TYWxvCqM8n4UlY4J4/pALhnwenrq6ii5FrHWH8MrJbwCCgK608vsctBhKJbBFW4h8YeqltqvJCYs7BAk/CqeHxN92tmJGSQGi338/r6y5AP/zeQtWvXgIP54SqyYAAMWFIkz7vg7Sigu0mFNagHWvHwEA3Di7CJeOV+PS8WoAwIIxKrR1mMAK+1FRUYFwOEzXJzpy5AikUilUKhXdDTeZsz17drZF7Www6kSJxWKNmtm3wUQq4GCqIQ+eQIRuIrByug4V4iA00sSE0mg0iiPGbngDYdS3WjE5P9Yyiuro8nGzF/97MoziPBZ+c3Gs0eXORhdanSEc72ZhweRi8Pl8sNlsCPVu7LMdw5wyMUqKdPDVuuAJAQ9+dBoPLK5AqZANfQEbDy8dAw6LQJcvjP/7zoIfT9GgUBzbL02apgfBcBS7mhwoEHGxepYeB4+3YJ8N+Kbdhb9cMw4lhSJIBRyctPWgJxhBoYgLuZiLYISkwxko9Pl83PN9DSVDUvMBHocFCY+FnnCsMoBMxAGEBRg/XgWCAFwuF2w2G9ra2kAQBFQqFV474sXuVhcev2YiCllnf/aM8SmNcEbbCR4suGwWHl02FqFQCF0eD96u7URVIQfCkAs2awQOeyeCwVg8zbedEXx+IoxZWgF+XCUEi8Wi20Tx+Xzk6UNoCTbj8vFKTJoUqziwXuXHrmYHFo1RQCzmIUqSqDvVjSKZAJuXFSMcDuOgMwg2i4DRFYDVE0LdqW6UykVw9ATx9y/aoJTw4AmEcag9FpN0y9yijMfT0e3HwXYXAGBeVSHKIu24aFIxBFw2JunOOJVrlo1Fi92HOWUFuHy8Ei5/OMESA2LXxMJxSrx5oANHTW5M0KU6pV/7xonPWo2YWSJDbasTq2YacOclZXT5j8rKSgQCAdhsNtQ222Bzh/H1tw2YVyY9qz4dxtF9HjASvyCSJBGJRGhrJt6qoX6n2npzuVzsMkbxzyM9KC/kY8NFeeBwOCgqKqJbeh/8qg0hGCGU5mF8daw8bnOnF18etuOqyWqUK8XYvGoibJ4gomTM8tAXCLBy2hm/1NeNDvzxo0aopXz8ep4Cv9thhCJPjCsnqqCWcCHgcXDZ9+EDZlcAxm4/rO4A7vxBCQiCwMUViYGKta1OPP31Kdwyx4CLKwqhzeNjblnMUa6W8nGaIHDVJHXK96PNF9DxVwIum+4EnMz/fWvGc1+3YXeTA1tunp7yvicYBfl9/W8SQE8gdbKEz+fDYDBg82oFGixujCsk0H4q1qrb7XbTrZNyqfI4nOV0M31+JF7zvTEqRWk0nWRKbOIFJv5vaiaRw+GAz+fTL6FQiIKCAtpvE9/WW6LvQa29EZeNU0D6fWhSfHue1bMNqFaJab8LADy1sxV7W2PJrnf8oARvHTLjuV2nsHyKGnddUpqy3+o8PgQcFqqUInT2hGHrCcMV8iIQjuLrRj/+vHwcHV4wTiPBb5dUobatC8ctHvx5+biE2uL/d9iCJz9rQbc/jAffPYFpRfm4cqIKO086oJbycPtFJVmfz0w32cWVcuxtceLKieq0n7lthgK3zFOiQinG/x4yAiDh9oUgTdOYUlcggK4gJjxkJNZhRK1Ww2q14ttvv0UkEoFCoYBSqcyqwSRjKeXGqBSlkUA0Gk1o751s5YRCIXi9XtTX19PDJkpgJBIJ/Xe2HXfjqVSKseWmKQBiDRgS9osk0WTrwcySAlo0gNgUeiAcoeOZYpUgSfhC6f1zY9QSbP/pDBAAzGYz7purQEWxDo/vaAYJfF+1MQZBEJhZUoDfvX8S4WgUl1TKMev7aG1vMIKNHzXCE4igWimCKxBBc6cXVUoRxmkkmFWS2OooSpIgkP5G/sunzfi8wY6aZWMxUZcYKFqmEOPv10/JeM54bAJlWiksrgA2fd4MpzeItw4a8da6Wb2ef8pSEYvFKCsrQ1lZGUKhEN2m2+VyIT8/H0qlEgqFYkjK0DKiNAoYSkd1fJvvdMMoym9DEESC0PD5/ATLhsvloq6uDtOnpw4lhmKfKT48YsVfP2vBJH0e/nrNmZZWC8YosGBMLM0kSpK49gIt5pTJUCbPXOCLcigTBIEpWiGKtVL8beUEOL2hlG62XDYLd11SgtNOHybppfR2HD1BFMuEsLgD+P2VVQhEgAJhrDHks9dPSlhHty+Eu/51BEIuC5uvm0SXXKE4ZvbAG4ygvcuXIErHzB5s/Pg7/GiSBrf2YXXJRFxM0EpRe6qL7rDSG+luai6XSxdQI0kS3d3dsNlsaGlpSeiQKxaLM6w1e5jh2zkMSZIIh8MIBoMIh8MwmUy9+m3ixUYikdAdTCm/zUgheTZRLuaBzSISkluT+eW/j6HB2oMnrx2ftp9cb0j4sTy7dPx4SmKZ3t+934CPjtpw6xwDbrmwOEVkkun2hdHlC8HlJ+ALRlKWf2TpGDTazqSAAMDnDXY8/fUpmF1B7Gqy46JKOfgcFkrlooTPUjcmj8PCczddAJs7AAl/4O3aCYJAQUEBCgoKUFVVBb/fD5vNhuPHjyMQCMDv98Nut0Mmk/XruhlOURkp4jUqRSn55MU7idP5bpL9NuFwGOFwGCKRKKPfZrQyp0yGf98+HcJe+sUZu/0IR6JwxnXNHQoOnXbBH4qi7nQ3ftqLILn8YZxwRDDlAj7+/ONx4HFYabuLqKT8lJm3f39jgbErgPljFFgxXY973/gGLBaBbWtnIu97fxEl2t92dONwhwvLp+pQKObB7Q/3ep4ocrlZBQIBioqKUFRUhEgkgq+//hoWiwXHjx+n2yUplUrweJkfGvEwPqUhwuFwYNWqVWhtbUVpaSm2bdsGmUyWsEx9fT3uvPNOuFwusNlsPPTQQ1i1ahX9vsfjwZYtW2A0GtHV1YWFCxfi4YcfBpvNBpvNTvDbUNYN9XvyOL+2tjYlZeJcIBSJIhCOZrRkKP62YgIsrkCKXyYeTyCMB94+Bgmfg/+andri2huM4M87mqDNE+D2i4rSXvgPX1WNV/Z34K5Leh9SPfqfRnx9MgiO3ILrZqRGpvfGvfNK8GWjAz+ZXQogZskJeey0Vtlv3zsOU7cfBSIe9jY78FWjHX9cOq7X7P+BDJ/YbDa4XC7Gjx8PkiTh8Xhgs9lw6NAhkCRJC5RUOvCwg0wwopSBmpoaLFy4EOvXr0dNTQ1qamrw2GOPJSwjEonw8ssvo6qqCkajEdOnT8dll12GgoLYDUEQBEQiES655BJ88MEHePXVV6FSqUbUUOpsQZIkoiSJO17/Dp2eIJ5aORElhZl9Reksjs1ftuKIyYPfLamCUsqHxRVAU6cXBEGgJ5gH4fen2eUP45jZDTYBfHnSATaLwI2z9Gk7+U4x5GOKofcOtABQLBOADeC9wxZ8cMSKv62YgAIRF181OkAQSAkviKdaJUa1SgyhMGZ5/Ov2mWARREIFToplUzTY0+zAFEMe/nPUEqtV7R9aa5GCIAhIpVJIpVKUl5cjGAyis7MTLS0tcLvdKCgogEqlglwuT7DYGUtpiNi+fTt27twJAFizZg3mz5+fIkrV1dX07zqdDiqVCjabjRYlsViMW2+9FQDw4IMPQqNJbTN0PkJdcJEoCbc/jHCUhC+YWxoOSZL44IgNPYEwjlt6oJTyUa4QYcMVVRBy2cjn+RD8fp2PfdyIPS1OrJ5lwJo5BjRZe7Dxo5P45aKKhGqWufDTH5Sgmm3BX76J9YyzeYLwBMP444cnAQBbV0+BhM/B4zuaUCoX9RqUmc5HRt2YN88twc1zY1bbxmXj0e70YYxakrJ8PMbuAERDcJfweDzodLEEYqpGts1mQ2NjI3g8HlQqFZTKWBwYI0pDgMVigVYby73SarWwWq29Lr9//34Eg0FUVFSkfX+0neT+QJIkPmuwI1/IwYzi1OFTMlw2C09fNwkufxjlClGfy8dDEAQevqoard9HTlP/m11agCMmN3yhKKhnd5lCFIvkVogwv0qOKzfvh9sfxqKxSvywOv0w6IjRDacvhIvKZRkbK+bxWPjTj8fCF4qgSiVGIBzFZH1sWFMo5qG+vRtfNTmwt7ULN87S9+mg7/QEEAyTdLxRMhI+B2M1vZcb/q7DhQc+OI18ARtvV5UN2XVH1cimOkR7vV7YbDYcOXIEHo8HJ06cgEqlQn5+fs4jg/NalBYtWgSz2Zzy/40bN+a0HpPJhJtuuglbt249r4dmR80ePPqfRnBYLLx1+7Re/USU30Mh4UHRy6xbb0w15GNq0lDr9Tojtu5tx0yDEHfNjPkA115YjDWzDbQoaPL48AQi0OUnDgejJIlX9nUgGIni7W/MCEdJ/PWa8SmF6CiOOyJ4bf8xXDlBhZklBeBzWHj86jMhDRcY8rF6tgFFMmGCIKW76TyBMG59+RCC4Sj+cdMFaROWs4HLJkAQsZbmw4lIJEJJSQlKSkqwa9cuFBQUJCQQU76odAnE5wKDJko7duzI+J5arYbJZIJWq4XJZIJKpUq7nMvlwpVXXomHH34Yc+bMybi+0ZhQmyuGAgHGqCVQiLlp/TUUfT0FSZJEg7UH+gJBnw7wZFRSfiy8IGn8QokCSZJwekMQcGOtwuPp6PLjn7UdIEkS4zQS9AQiUEl52N3sRLlClJKka/VGEQiTaOzsQTp4HBZunpPd5ASbIMBlE4iSBHhsAsEcrIVQJIpHPmwACRIPXj4GT15VDAGHOGvWBkEQUKvVUKvVIEmSTiA+cOAACIKAUqmESqWCWCxOu4/ntaXUG0uXLsXWrVuxfv16bN26FcuWLUtZJhgMYvny5Vi9ejVWrFgxHLs1oskXcvH3VWfK4ppdfnDZLMjFuVlCnzXYUfNxE8ZrJPifFRNy+uzl45XQSHl450AbTK4gipPeJwgC/7NiAsyuQEKtJgDQ5fMh5bPh9IbwwOIKFMmE+PiYDY993IQSuRAv3pgYfX2hjoOLplahWjXwgEMhj41XbpmBcCSKPCEXp7v7/gyF2RXA5ydsAIBbLyxBgSD3iPuhgiCItAnEJ0+ehNfrhUwmg1KpRGFhIe0sZ0QpA+vXr8fKlSvxwgsvoLi4mG7XXVdXh2eeeQbPP/88tm3bhi+//BJ2ux1btmwBAGzZsgVTp04djl0c0XR0+fHT178Dj8PCK2um9mo5xdPlDUHEY4NFAPlCDswuP94/bMOl4xQpEdndvhB+9uYRiPmchKDK1w8YsbPBBQJRzE5TOrxIJkxZVwwCETJm4VjcARTJhDAUCCDisTE2jXOZwyIwsyw2RPQGIzjt9KFKJU4pU5ItsXMUO08nbH48/qUJSy8I46555b1+zlAgwH8vqgQJEkUyIU4PpD3uEEMlEBsMBkSjUTidTlitVjQ0NEAoFNJDPEaU0iCXy/Hpp5+m/H/GjBl4/vnnAQA33ngjbrzxxqzWR0Uxj7aT3V84rNgUN+XjSCbdcPaTYzb8aUczrhivxLbbYj6pJz5rxr/rzWhzePGHH41JWN7pDcHiDoLwBNETiKBAFBOl62foEAoEML8sNwuGzSLw5LXjYXEFMP37xODxWileuXkqdjU5YXMHoJSm6ZkN4I8fnkTdqW5cN12Lj4914sJyGX72w7Kcth9Pi8MPpy+CA21dCf8nSRIufzhWVzwupSa5UeVouM5YLBbkcjndULKnpwdWqxWnT5+G3++n01+ySSA+24zKiG6q+uRQJD+ORNR5fGxdPQVcNgvCpNIdyRfYjuM2vFprxCSdJJaX5QvRLbMXj1WipdOLJRNSfXqlchE2Lh0DflI09VRDPrTztfD7/Tnvd5lchLKkdI9X93fg9QNGzKssxMalY9N+Tvp98wGXPwyLO4DaJDHJlQUVeZDy2bhkUqKVtHXPKWzZcwp3/KAM188ypP3saPVfUgnEUqkUNpsNUql02BKIB8rI26MsON9ECQAtLOnoCUZQ29aFqYY8fHysEycsHkwvzsffVkxAWVx4wFRDHjZfNynjeuJLnWTilMOHp75oxRXjlXSCby5MK87HzpN2zC2TZVxm/aUV+OnFxZAKOJisz6NbjKcjFInC6gmhUiDIaAFw2QQuKpWmhAd09gQRJQF7TyDn4xgtkCQJNpudVQJxfn7f3/9wMCrvag6Hg3A4nFBD6HyFIAi8cNCJQ2Yrbpqlx3/9sBQzSvKxaIyCLkc7mHzRaMfXTQ50eUM5i5KjJwgBl4U31/ZeOYH1fWwSACzsYxt/+bQVXzQ68PNFlfjx1MwpKukE679+WIFLx6l6jVc6F9wE8fvfWwKxXq9HSUn2ta2GilEpSoPRkPJcuNgodBIODrPDMMgEMBQIsXJa5hQTAPjipB1vHjTh7nmlGKfpPaIZAIyuIH7zsRGTi/z474VlcPnCuKQyc+pHOqIkieXPHYg51H9YilvmJs/lDYA+vkaSJOEPR3HbKwcRjQKbrpsMMZ8DHoeFyVmkwYxm+hp+xicQj5SRx8jYixwZrI4mwyFKw7Gdaybk497LJmXshJvM9m8tOHC6G583dGYlShZPCN5QFEfNbhSKuLh7XmnO+xgr3Bb7PZrhPglHSXR6AlBL+RnP2Zcn7Xj/iBV3/aAEajELt88sxPIJ+ZhS0XvakTsQQZvdBxIkun1hiLOM2RrtD6/RuP+jVpQG0tHkXIMkyawFCQDumVeKCVoJlk5KLR2bjqk6Mf5rjhxTq0v6fYETBIF37piBlk4vphrOFPbf0+JEXVsXJvJIPL6jCZ+esOO+H5ZiUVUBXY8o/ufTX3ehqTsKaciJq6rF4PF4ICMR7NlzCoWFhdBoNAn93L462Yl9Jxy4eqIcj18zEVEyc+rJuQgjSsME5VPqL8PVZmk4LLL+rLtcIUK5IvvhE4sgMEUj6LXyQHOnF79+7wTmlslw7/zStMsUCLmYasijK3v6/X786T9NMLtDuKaMhNFrRSAQRkNjMwwREV16RiAQQC6XQyAQ4G6BF180OnHHxSVQSHgIh8N0FVC73Y729nYcOXIEhYWFUKpU+P37x+Hxh1AuF+Hqfg4ZR9tNHQ8jSsPE+dqQcqTSaOvBtgMmdHT5sbvZgVtmKFIsHL/fTz9IuFwuBIJYF91rJ8pQbw5gorIHd82eDrMnjDK5MO2NFAxH8bedx+ENRXDDTH1Cnh+LxaJnkahAQrPZjLnKMNrdgEEURjQa7VdC62iGEaVhYjAc3cPBYIrfZw2dcPSEcPVUTUqUcy7b8AYjePLzFhTLhLhxlr7P5Q8b3djwbgsuKRHiXkM47ZDqoc/sMHqimKtlY1FpFKdOnaItnPz8fFqAOJzU8rPl5cANiBXeEwl4KBdknjEkCIDLIcAKI229JAoqkDC/QIYb5cUIdJ5C0O/Dnj17IJVKodFoEOZK0Gz3YU6ZDJykigM2dwAnrR667O5ou6njYURpmDjfLCWXP4ya/zQhQpIYq5YkVIzMlIQZjJBpqy8eMbmx43gnOCwCK6dpaV8UVVI4WXA+buiG2RXAvtYA5n37bcKQispYv3KqEF81deHeJdV02ZQjRje6I1FMUwze7BaXzcI/fjIZwXA0bbncZJ7f1YbXa9uxsEyEtXN00Gq16O7uhsViwfqPDsPoJfFfF+uxam5lwszTQ9uP4pjJjfsvrcIEyei5TtIxmq5zilEpSqPNpzRQJHw2Lh+vhL0nmLFW0hOfNuOo2YONV43Blr2n8XmDHQ8vHYMZxQWIRqO0yKjZPlxWLkShgMSxI98hGAwiEo3igJWEOo+PqXop+Hw+xGIx5HI5bjOUQCk7DaWAxLRpY9Ju+7aLpLjtojN/O3qC+OXbx0CCxLPXT+7VF5UrIh4769w/qnEln3MmhYSK0ZnbSuCTY1YUcMLYv38/hEIh1Go1lEolxqglaLF7Yy3AQ95B2/ezBWMpDQMsFmtAs2+j7UtiEQT+e2H6RFKqJdSO413wBsLYdaQFx9u74Q0Esaf+GKJmNlgsFm3d8Pl83DJbm1DP/FC7C2/vOwYuO4TtF5UnzOSRJIkPG7phcQex43QEl09QYdH3AY07TnTiH1+fwl2XlND95IBYAbUyuQj+UASFIi5OO304dNqFBWPkOZdPGQhr5hThigkqeDqNKe89cFk17r+0ir4WPB4PzGYz6urqML+QhxVXF0OjFuH0aXu/r5eRYKUww7dhgsPhDErw5FAz0IuBJEkEg8G0TuNAIEC3jQKAe6bnwR4gMLskH9PLFDB6ophZJgcniw4tZXIRxqglKJIJwGWn7jOLAIJhEruanDjl8NGitLfZifYuP/a1diWIEo/DwubrzpQUWL/9OOrbXXAHwrhhZqof68MjVhxsC+GC6STanT789bMWXDFBicvHp6+71RvRKAmTyw9tngAsFgF1ngBee/p6SPH/k0gkqKysRGVlJXp6emCxWHDw4EEEAgEoFAq6602unG1ByEWUzva+UoxKURpNPqVM26EEh2oFlSw4kUgEBEGAx+PRFg6fz4dUKk3ormu329HV1YW5SaWDc0kWkIkSazfFQxAENl5WhBarC/s72Qkdbe+8pARTDHm9FvYHgPnVcnT7QgnxSRRufxh//awFvkAYS01ufGd0o7atCz3BSL9E6dmvW/FGbTtumVuCmy+MhQDk+l2LxWKUl5ejvLwcR48eRTQaRX19PYBYwUKNRgOBoO9Yp8FoJDlQSJIcdRVcR6UoDdRSGmpRooZUkUgEdrsd4XDirBUVWxM/NS4QCCAWi+nfR1IPOiGXhVIZD/MvKE34v1zMw1VZBGCuuECLFRdo074n4bOxfIoaR1s6UKkSo0gmhDcUwYW9JOz2RigcG9aHBim4lsvlIj8/HyqVCn6/H1arFd999x0ikQhdEVIkylwT/WwX/WeGb8MEm80ekKN7IFBDpnTWTSAQQCgUAkEQ4HA48Pv9cLvdEIvFKCgoSJgaH+x9GmrSbaPB2gNPIJxVdYFMEASBOy8pRa3QBiGXDSGXjbUX9j8v7q55ZfjRZA1KC88IBXVjNlg8EPHYMQd2jvsIxPLEiouLUVxcjGAwCKvViqNHjyIUCkGlUkGtVkMiOZO2w/iU+seoFaWhspQyCY7f76cFh81mJ1g4mWJxvvnmG5SUlGTdDbW/x5ItoUgUR0xujFFLUuoy5bqNnkAY//2/RxCJAk9eO77PVkXx2HuC4LCIXsux9BcOm4VyRWqpkzZnABs+bgKXzcKb62ZlPYOX6Trh8Xh01cdQKERXfPT7/VAqlVCr1RA1OnPJAAAgAElEQVQK0weBZstoFJTBYNSKUn9m36hYHOoiIkkyob03VXsmXnCoWBw+n9+v0qJDPUz8oqkb0UAAlZV9L791bzteqzNi0VgFHrwsiw/0Ap/LRplchM6eUE4dVKzuAG5/7VtwWCy6n9twIOaxIOCykSfggNNL4GV/4HK50Ov10Ov1CIfDsNlsaG5uhsfjQTgcRnd3N/Ly8vp17TDDt1FCOkspGo1mHFJRM1VsNht8Pp/26+Tl5UEul4PP54PH4w36lzfUF0ODtQebdptBRqNYOC3UZ0ChJi/WnUSbl34WKUqSaOn0orhQ2GdfNQ6LwFMrJ/brovcEIvAGg+jo8udkYQHASWsPTjl9mF8lp6O6s9m+UsLDm7fPBItF9Hls8eR6fBwOB1qtFlptrFpnXV0d2tra4Ha70yYM9wUjSkOEw+HAqlWr0NraitLSUmzbtg0yWXpHpsvlwrhx47B8+XJs2rQp4b36+nocP34c33zzDZqamtDW1obq6mr6xMdHGwuFQshkMlpw4mcgjh49Co1GkzD+HyqG0lIyFAgwTiUEhwxBKuj7q/zRJDV+WC3POHT55/4OvLyvHUsmqDLGRSWT6wWvkvKhyxeg0daD+nZXzqL0wDvH0O0Lg89h9TnrR0F9B/y4IWtdWxf+8P5xrJyux42zs2vdlCvUQ3Dy5MmIRqOw2+10/7bCwkKo1WrIZLKMs2ODNfvGiFIaampqsHDhQqxfvx41NTWoqalJadtNsWHDBsybNy/te7t374bb7YZUKsWcOXPwwx/+EGq1Oucpz9EW0Z0JMZ+DP15egs7Ozl5zwYBYt5IvGx2YWybLWEuIz2GBIAgIuLmdz1MOH9yBcMZGk8k8sKgcu5qduHScMqftAMDcMhm+7XChWCZANBpFNBrt1013uKMbVncA+1qdfYrSYARPZkoYPnbsGAoKCqBWqyGXyxOuZWb4NoRs374dO3fuBACsWbMG8+fPTytKBw4cgMViweWXX466urqU9++66y4AwIMPPoixY8fSrcD7w0iYGRlOXtxzGm8dMuPScQr8dkl12mVWTtNiXlUhVN93GYlESby09zQCXi8uL0/vN/KFIrj3zcMIhsmsHd6T9HmYpE+NWYqH8hmSJEm/AODn88/MzFGxXCRJwu/3g8VigcPhpH1IJd+YK2cYoM4X9DlzONDrJJ0gxHceIUkSXV1dMJvNOHHiBPLy8qBWq6FQKBhRGkosFgstIFqtFlarNWWZaDSKX/ziF3jllVfStmOKZzBm34aD4bLIstnGzJIC7G3pwpzSgozLEAQBTd6ZoMBTTh/eOGBCNBLBdLUSFWk+w2WzoJHyYe8JQZZFkixFNBql95skSUQiEbjd7rRBiSxWzIJLflHiQwkYNZFBvUcJVLrzI+KxccWEvmOsBnJTZ/NZgiAgk8kgk8noov4WiwWNjY0QCoUIBoMIh8P9DiMZjQ/fQROlRYsWwWw2p/x/48aNWX1+8+bNWLJkCYqK+h7fj/TgyeEk2xvm4orCrH0wFMUyIa6ZqkHA54VSnN4PxWERePr6SYiSoGe1kq0c6vdM+08QBMrLy9HY2IhoNAq1Wg2VSgWhMLt4IkqcWCwWIpEIQpEo8P3EBwCEQqGz8n3nus34hOHq6mrY7XYcPXoU+/fvh0AggEajoRtM5rre0cSgidKOHTsyvqdWq2EymaDVamEymaBSpaYP7NmzB1999RU2b94Mj8eDYDAIiUSCmpqalGVHS5rJaBc/NovAnT8ogc1mg8dzplVssuhQfp1Qmq8k3qqJt3IAJAyzqBmrQCAAi8WCw4cP062BersRI5EI/H4//ar5vB1HrQHcPZUPnSh27nk8HtRqNQKBANjsWIJyLn7IgdzUAykfLBaLIZFIMG3aNHg8HlgsFtTV1YHL5UKj0UClUvUZA8fkvmVg6dKl2Lp1K9avX4+tW7di2bJlKcu8+uqr9O9btmxBXV1dWkECBq+bCUMq8YJD/aSGVl6vN0EcqIuYzWYnCA+Afudb8fl8Omra6/XCaDSitrYWHA4HYrEYHA6HDvmgKkkKBAJ6xtXcQyJKsCDTlWLO2DPDM8opHg6H6XywbARqINfJYOS+UedTIpFAIpGgoqICXq8XZrMZBw8eBJvNptNd0iUMMz6lDKxfvx4rV67ECy+8gOLiYrz55psAgLq6OjzzzDN06+5sYXxK/d9GJgdypnUTBAG5XI5AIICjR4+Cz+fTT2kWi4UoScIbjEDCy/1SooJX4y0dn89Hx5gBsaF6Xl4eWCwW/H4/HA4HpFIpysrKUmarAGDT9SqcdvowrTjRgR0vPv0VqP4c31BcayKRiE4Y9vl8sFgsCQnDVDT5UO7DUDIsoiSXy9M6r2fMmJFWkG6++WbcfPPNGdc3Woq8DTfJQhM/tEomk9M4/vd4ysvLUVZWhk5nN5ydVrS0tKCgoAAvHg6i3uTFo8vGYUZJohOdKi4XLzbUKxQKAQAdV0a94nMEM1XV7O7uhtlsRmNjI2QyGTQaDfLz80EQBNR5fKgzBIdS9CZQbDY7wfKjzsnZIBtBEQqFKC0tRWlpacLQNxKJQKVS0alRo4lzJqI7F0arTymdAzkSicDj8cDj8SREpVNiQ83aDMbQ6n8+b8b/fWfFb6+sxsVzqmCz2dBhPwGfL4xDR05A1CNFOBxOGFrFB7NKpVKoVCoIBIK09bqzId4ZHI1G4XA40N7ejmPHjkGhUECr1eYUFJtOoCjHOJXO1N/vcKBWSq6fjx/6UgnDTqcTbrcbWq02JWF4pDJqRelc7PuWyYGcDupGys/Ph1arRUNDA1gsFn3xDbT0SXyBOcrSOXbKCl8ggN31x8Dt5IDNZuNns/LhDLFhEJNwuVwAYk5rnU6X8yxRrrBYLCgUCigUCkQiEXR2dqKpqQl+vx8qlQoajabPGbx4Rzllzfl8Pvh8PjoOSq/XIxgMplhQIxkqYdjpdEKv18Pv96ckDEul0hFpRY1aURoNw7f4LzydA7m3aXIAWVs5VNtlyjG8f/9+FBQUQKfTZUwEpXIF44dV1O/U0IoqMCcUCiEQCPDry6vQ7olieqkcPE560QsEAjCbzTh06BB4PB60Wi2USuWQ38jxDt9QKASbzYZjx44hHA5DJpNBIpHQVhx1nPGOcuoY8/Pz6SJu1PmPt6AA0D6ovgRquC2lTOvgcrkoLCyETqdDOBxGZ2cnmpub0dPTA4VCAbVajfz8kdO+fFSKEpfLpR2h/WEoRCmdlUMNL5L9I9lOk+eKSCRCZWUlKioq6Ez1np4eSCQSCAQChEIh+P1+2gKI9+WIxWIoFAoIBIJeqyH0FUXG5/NRUlKCkpISeDwemEwmNDc3Iz8/Hzqdjvb9DBbUMcVbOZTIUje13W6HyWQCh8OBQqFAaWkpxGJx1tZkpiEe9V5vAnW2RSkZDocDjUYDjUZDW5enTp2C2+3GxRdfPKRldrLex7O9A/1huIu85WrlUDWXSktL0dHRgfr6esjlchgMhl6rFGYLVdky+Sb0+XwJFREEAgGUSiWCwSAcDgddYkOj0QzLEEQikaCqqgqVlZUJvh+VSgWtVtvnuaAK6iULDvUTiN1k8ZaOUqmkhTb5GL1eLywWC44ePQqhUAiNRgOFQpHTUJcSKOqaiEajCAaDKVHk1P4PlKFMM4m3LqPR6JAPt7Nl1IrSYDq6+zNNHh9FnMnK4fF4GDt2LKLRKKxWK44dOwYA0Ov1UCqVGW+G+DIsyUOrYDAIIGYtUjciVfdJKBSCz+dnFBzKctm3bx9kMhl0Ot2w+BWosAK5XI5IJAKr1Yrjx48jHA5DqVRCKpUiEokkiA5lCVMlg6ljzcvLo2fnchVWkUiEsrIylJaW0t1LmpubkZeXB41G02vGfjLx33+8QMWnufQ3WZhiOKsEjCQ/2agVpVwc3cl5VtFoFG63O21dm77yrPoDi8WiTWav14v29nY0NTVBLBZDKpXSCaXxztX4qXKRSITCwkIIBIIB1X2iLJeKigrY7Xa0tLTA7/dDo9FAq9UOuukeH4eUbOVQRfWMRiPC4TDt91CpVBCJRBlDAgYDgiAglUohlUpRWVlJhxg0NDSkhBhkA4vFoi06r9cLv98Pr9eLnp4eOn8tU6JwbzAJuaOI+Ny3/uRZabVatLW1ob6+nr4hs+lOkQ3JQ450fg4WiwWpVIpoNAqLxQIgFvRWVVXVrzY+uRJfRiMYDMJsNqO+vh48Hg86nQ4KhSKrGyjeoksWneQ4JKFQCKFQmFFcKSvuxIkT9IxiLsXQ+ktvIQZKpZKuu5XsJKdm6KjvlM1m09Yc1diSquUVb0FRcVDZnN/zVZSIPkzEERlh+Nhjj2HHjh149tlnUVBwJmAvnYXTmwM5HA7DZDLBaDRCJBLBYDD0eSOke/pTv8cPreJTH+Idyun2w+/3w2g0wmKxID8/H3q9vl/lUweK2+2G0WiEw+FAYWEhtFot3QAhWXSojizxohP/sz+lg4HY+XU6nTCZTHC5XFAqldBqtRCLU+tuDybU8JE6Pq/Xi+7ubvT09CAajYLH40EqlUIikaQca7YCTlnp2UaRU+dh/Pjx/T6uAwcOYMKECVk9dIfjgRhHxotjVIpSa2srtm7diu3bt2PKlClYt24dpkyZ0u/1UVHC7e3t8Hg8tJ8j2ZlMOdf5fH6K2AiFwgGX1CVJEg6HAx0dHfD5fNDpdNBoNEPmgOwtRicYDNLWjlQqhVwuh1gsThCdoYbyP5nNZoTDYWg0GqjV6n4NM6ljjRee5NAAypqLt+wEAkHCfkSj0QHtB3BmBo+y7jMJlMPhgMViwbhx4/q1HYARpWGHJEl8+umn2LRpE+x2O9auXYtly5b1erFQQpMu9SF+KBgIBCASiWgH6HDdiBTBYBAmkwkmkwkSiQR6vT7n4Uzy07+vGJ3436kYHWo/zGYzhEIhdDodCgsLh90xSsU/WSwWOv4pfuYsWWDjjzv5WOMFJ9cee1Qqh8VioasYqFSqftc76k2gnE4nbDYbxo4d2691A7H80kmTJmUlOIwoDTJtbW14+umn8c4772DOnDkYO3YsKisr6aTF+ATPTEOr+IuTslra29sRDAah1+sHJVI6V5KtOI1GA51OBx6P12eMDhUWkDzc6E+zS5Ik6eGd0+mEQqGATqcb8mEVBZVH5/P50NXVBbvdjp6eHvompmYj01k6Q/WdUdn6Vqu13yEG8VACRfmgHA4HAoEAxowZ0++HQG1tLaZMmZKVVceI0hDw61//GnV1dQgGgwgGgygtLcUNN9yAWbNmQSAQ9Hto5ff70dHRAavVisLCQhgMhiG/GdPFIvX09KC7uxs+n48OfqRCAZJFZygtmWg0CpvNRs+cUaktA7Ek40UneZhFzUimO06qa+1w+p+SIUmSDjHo7OzsM8SA+m6TjzXeL0lF06tUKuTl5aVNFM6G/fv344ILLsjqu2FEaRiora3Fpk2bcOTIEaxZswarVq0aUEIidTO2t7cDAAwGQ79TKOJzy5ItnUwxOtRPPp8Pj8eDjo4OdHV1Qa1WQ6fTDdoMYi4EAgF6eCcWi+nhXfIDID6tJdmvEx9hns7SyWZoFIlEYLPZYDKZEAqF6BCM4Y5Qpizbjo4OOJ1OiMViiEQiOuyDCvrMZNml80smD/FyEah9+/Zh+vTpWZ1DRpSGkc7OTjz//PN47bXXcMkll+D2229HVVXVgNbZ09ODjo4O2O12qFQq6PX6BFHoK0YHSM0tixedbK26SCQCi8WCjo4OcDgc6PX6rKf0B5NIJEK3EHK73fQNFgqF6Jm6+GFz/DEPdhtzyu9jNpvB5XLpMIfBHMbFW3bpfFjU0BkA7T5QKBQwGAyQSrPr+pJpu/Fxd32luezduxczZ87s89gJghhuAT+/RYkiEongvffew+bNmwEA69atw2WXXdavi5V68vf09MBqtcJutwNAgpmcabp8KBpfAqCtJ4fDAYVCAb1ePyhpLQASnvTJlg5Vs4c6RqrhZ3d3NwiCgE6ng1qtHnTxyQYq/qmzszOn+CfqoZIsOuksu2TnebrricozM5vNOVUx6I1sBGrPnj2YPXt2nw8pRpRGACdOnMDf//53fPHFF1i5ciVWr14NuVxOv5+uQFlfMTokSdL1rKnyHWcjwZFKa+no6ADQd1oLkHoTJlt26SydbMIDfD4fTCYTrFYrpFIptFotZDLZsMdgJcc/KRQKKJVKEASRIjrxgZ/JojMYlh3VNn6wQgySk8Dj8/D27t2LOXPmMKI0mmhubsaGDRvw2WefQaFQQCgUoqamJuUmTA4M7I34oEyxWAyDwTDo2fHZQpUzsVqtyM/PpwUhXngo0YmPv0q+CQcjsrirqwtGoxEulwsqlQo6nW5AlkJfUJH1ya9AIIBQKERbPHl5eXRsmlAo7HfgZ38YzBCDaDRKH6PX66VnBy+++GJwudxehYkRpRFEe3s73nvvPZSUlMBsNuODDz6A0WjEbbfdhquvvnpAzj/qRmxvb4fX64VOp6OjpAcbynGebOnEP/kB0L8rFApoNBqIxeJhvQmBM34wk8kEANDpdFCpVDkPo5MDIjOlfiS/4n12VJoN5X+i6j8Nd+gHkBpioNVqIZfL6X2Jn7VLZ90lDympvEkej0enm2Rq2MmI0ginvb0dzz77LN555x1cccUVWLt2LQwGw4DWGQwGYTQaYTKZUFBQkLPDM3m2Lv4GjHecp5vRSRadkZLWAsRuRGp4l1xzKT5ROd2MXbqAyFxSP5Lp6emByWSCzWZDXl7eWRlqUgGvDocDNpsNbrebDqZMjscSiUS08PRlzSanuSTn4TGiNEoIBoN466238Mwzz0Amk2HdunWYP3/+gFNJ7HY72tvbEQqFYDAY6M4g8U/B5JQPIHG2Lv4G7K/jPD6txe/3Q6vVDmlaS7rtx1t3DocDTqeTLj0bL7LJxzyUTvOhzL/ry4FOCa1IJKKPNRQKweFwoLu7GzKZDFqtdkAPkWSBoizX0tLSQZsYyQJGlAbKoUOHsGnTJhw6dAg33XQTrr/+euTl5WX9+XQBcx6PB11dXXSTRLFYnJBf1lvsymAzGGkt6UgebiTXS0oWHkpkqWqRbDYbOp3urA2p+hP/1JsvC+i/A52qYmA2m+F2uxOqGPQGScbqp7e0tNCv1tZWtLa2wmazgc1mQ6/X469//SsqKytzP0n9gxGlwcLhcODFF1/EK6+8grlz52LdunUYO3ZsQsmSTDdgfGG25Nmrzs5OtLe3gyAIGAyGsxJrBPSe1pKO5Py6ZL8Oh8NJa+lkG4sVP6Tqq+74UEP5nyixLCgooMMfqDpKVIxS/PFSw6zBrBFFhRiYTCYEAgGcPn0aSqUSJEnSotPc3IxTp04hGAxCKpXSveKoV2VlJW2lnwUYURpsDhw4gN///vfYt28f2Gw2SkpKUFNTk2B2p3Os9kVPTw/a29vhcDjSBmUOJ6FQiJ5FZLPZCU0hfT5fn9n1g93YkRpqer1eug7WUEQhUw+Y+IJtyQ5lqveg3++HRCKBVqsdUGJuNvvkdrvTWjsulwsSiQQnTpwASZJYvXo1LrzwQlRUVKC0tBRCoXAk1lRiRGmwOX36NFpaWlBaWgqv14vnnnsOO3bswNVXX42bb74ZKpVqQOuPj9Sm2uWkS98YKMmR5/Gv+EhsSoyoFj1FRUXDnmNGEQqF6BkzDodDD+9yEcH46fN0M3fxDuX4V/KkATXDajKZ0N3d3a/ec9R6IpEI3WghXnja2toQCAQgkUjopqAVFRW0taNWq+ljN5lMdN2nEc7oEqWPPvoIP/vZzxCJRLB27VqsX78+7XJvvfUWVqxYgdraWsyYMWOY9zIVr9eLV199Ff/4xz9QWVmJdevWYebMmQMWEpfLhfb2drhcLmi12pxK16YbViY70DPFJ6VzeI+UtBYKj8cDo9EIu91OtxGiSgzHO9Hj43eGYuYOODOkMhqNtP+JqkAJnLF2WltbacGhflosFrBYLOj1epSVlaG8vJwWnrKyMohEopFo7QyE0SNKkUgE1dXV+OSTT2AwGDBz5ky8/vrrKdX33G43rrzySgSDQWzatGlEiBIFSZLYvXs3nnrqKbS1tWHNmjVYsWLFgAMF44dTEomEDsrs7akPJCb2xr8G6kAfyrSWbKD8WVSgoNPphMvlomt+i0QieuIgflg9VLOL1IyWyWRCU1MTrFYrPv30U+zduxfRaBRCoZD27cQLT2Vl5bB1mBlBjB5R2rNnD373u9/hP//5DwDg0UcfBQD8v//3/xKWu++++7Bo0SI8/vjjePzxx0eUKMVjMpnw3HPP4a233sLixYuxdu1alJaW5rSOdMmf3d3d8Hg8iEQi4PP5yMvLo2+8eF/WcFzo6dJaBsOBmi5OiXqlcyhTL4Ig6IRcPp8PnU4HuVw+KOeCKlNCWTktLS1oa2tDS0sLzGYznetHOZMrKipAEAR2796NxYsXY+HChQPeh3OEjKI04hoHdHR0oKjoTMtDg8GAffv2JSxz6NAhnD59Gj/60Y/w+OOPD/cu5oRWq8Vvf/tbPPjgg3j77bdx9913QyQSYd26dVi4cCFYLFbOyZ9UeySqVGtHRwcdkZzNFPFgk9ytxWg0oqWlBYWFhdDr9b3uT6YIZWpoGX/cVPdaoVDYZ3hAcXExiouL6cJ0jY2NkMvl0Ol0ve4PZe1Q7ZeSncp+vx9isRhlZWW0tTN37lxUVVVBq9VmFL6lS5dmcSYZgBEoSukst/ghRjQaxc9//nNs2bJlGPdq4HC5XKxcuRLLly/H/fffj/Xr18Nut0MsFuNXv/oVJk2aRN98UqkUKpUqq9gVNptN9zLr7OzEyZMnEYlE6EqZwz0koLr0lpeXw263o6GhAcFgEDKZDEKhMGH6PF3IALXcYKW+SKVSjBkzBtFoFJ2dnWhsbEQwGMQ333yDcePGwel0plg7QOxhQjmU586dixtuuAFlZWWQSCTnmm9nxDHiRMlgMOD06dP03+3t7dDpdPTfbrcbhw8fxvz58wEAZrMZS5cuxbvvvjtih3DxcDgczJ07F9dffz1kMhnef/99PPvss5gxYwZuv/12lJSU9Gu9BEHQbZN8Ph86Ojqwb9++Iff1UEGh8dPmybN3fD4fLpcLFouFDswc7PpGyVAOecraiXcqEwQBLpeLY8eOQa1W42c/+xnmzJmDyspKaLXasxKkyXCGEedTCofDqK6uxqeffgq9Xo+ZM2fitddew4QJE9IuP3/+fDz++OPo7OzsdcbuiSeewPPPPw8OhwOlUokXX3yx3wIw2ESjUezYsQNPPfUUXC4X1q5di6VLlw7YIRvv66GCMqlyHbkQHyCZzpGeKR0kef8HM62FJEn4fD5abJqbmxOsHZIkodFo6CEWZb2Vl5fT1g5JkmhsbBxwwT+GfjF6HN0A8MEHH+C+++5DJBLBrbfeioceegi/+c1vMGPGjJSx+fz58/HYY4/hJz/5Sa8zdp9//jlmz54NkUiEp59+Gjt37sS//vWv4T60PmlpacHTTz+NDz/8EMuWLcMtt9wCrVY74PV6PB60t7fD6XRCrVZDr9cnTFWn82l5vd6UAMnk6fP+DmWySWuhmnXGWzuUCHm9XgiFQpSWliY4lcvLy2EwGBLaqTOMSEaXKOVKtjN2FIcOHcI999yDXbt2Dds+5orP58Mbb7yBZ599FkVFRbjjjjswd+7cft9oVBMCj8cDq9WKrq4uADGfFIfDoWOVkmfwhnIoQ5IkvF4vGhoaYLVa8eSTT9LZ6w6HAyRJQq1WJ0yfU8IjlUoZ0RndjJ7Zt/6QzYxdPC+88AKuuOKK4di1fiMUCnHLLbfg5ptvxv79+/HUU0/hwQcfxJo1a7By5cqUaOpMNaN9Pl+KQ1mlUqGkpAShUAgWiwUej4cOPBzsGB5qCElZO9QQq7W1FR6PBwKBACUlJSgvL8c111yD5uZmHDx4EHv27BnUXDGG0cM5IUp9zdjF889//hN1dXX44osvhnq3BgWCIDB79mzMnj0bp0+fxiOPPILHHnsMFRUVEIlEWLt2LV1/KF3YgFAo7HUGTqlU0kGZBw4cgFQqhcFgyDrplYolih9aUT+NRiOi0ShUKhUqKipQVlaGKVOm4Oqrr0ZFRcVZS6xlGNmcE6LU14wdxY4dO7Bx40Z88cUXw91OZlDgcrmIRqO48847Ybfb8d133+HFF1/ELbfcgsWLF/d7qMXlclFcXIyioiI4nU60tbXB7/dDJpNBrVZDIpHAZrOlOJQpa4fP56O4uJj27Sxfvhzl5eUoKiqiC9kzMGTLOeFTymbG7tChQ7j22mvx0UcfJcy29JVnFwgEsHr1ahw4cAByuRz/+te/co7IHkqOHj2Kv//979i1axdWrVqFm266CYWFhTmtg3J0x1s7VCfcDz74gD6/EydOTHEqD0bNJYbzknPb0Q30PWO3aNEifPfdd/RMVnFxMd5+++0+8+w2b96Mb7/9Fs888wzeeOMNvP322yNy1s7lcuHll1/GSy+9hEmTJmHdunWYOnUq/T4VPBgfoUz9dLlc4PP5KCkpSXEq63Q67NixAzKZDBdddNFZPEKGbBhFyeznvij1h2xm7S677DL87ne/w9y5cxEOh6HRaGCz2UasdUCSJD7//HNs2rQJx44dg1arRXd3NyKRCORyeULpC8raORstjxgGn1GWzH5uz771l2xm7eKX4XA4yM/Ph91uh0KhGNZ9zRaCILBgwQIsWLAAe/bsgUgkwvjx4welTRLDyGb//v10kCgAXHfdddi+fXuKKG3YsAEPPPDAiM0bPa9qJSSTzaxdLjN7I425c+diypQpw95CieHskO4hS1VuoIhPZh+pnNeilM2sXfwy4XAY3d3dOTuSGRg++ugjjBkzBpWVlaipqUl5/4knnsD48eMxefJkLFy4EG1tbTlvI9tk9r/85S85r3s4Oa9FaebMmTh58iRaWloQDAbxxhtvpKSxLJwmdM4AAAb5SURBVF26FFu3bgUQcw4uWLCAsToYciISieDuu+/Ghx9+iKNHj+L111/H0aNHE5a54IILUFdXh2+//RbXXnstHnjggZy3k0sye2lpKfbu3YulS5eirq6u/wc3FFB9yDO8znnef/99sqqqiiwvLycffvhhkiRJcsOGDeT27dtJkiRJn89HXnvttWRFRQU5c+ZMsqmpKeHzdrudXLRoEVlZWUkuWrSIdDgcKds4dOgQOWfOHHL8+PHkpEmTyDfeeGPoD4xhxLB7927y0ksvpf9+5JFHyEceeSTj8gcPHiQvvPDCnLcTCoXIsrIysrm5mQwEAuTkyZPJw4cPZ1x+3rx5ZG1tbc7bGSQy6s55L0oD5f777ycfffRRkiRJ8tFHHyUfeOCBlGVOnDhBNjQ0kCRJkh0dHaRGoyGdTuew7ifD2ePNN98kb7vtNvrvl19+mbz77rszLn/33XeTf/zjH/u1rb4esvEwonSOUl1dTRqNRpIkSdJoNJLV1dV9fmby5Mm0SDGcXT788EOyurqarKiooB8u8fj9fnLlypVkRUUFOWvWLLKlpSXnbWzbti1FlO655560y77yyivk7NmzSb/fn/N2RhkZdee89ikNBhaLhQ7I1Gq1sFqtvS6/f/9+BINBVFRUDMfuMfRCNr6eF154ATKZDI2Njfj5z3+OX/3qVzlvJ9c0qHfffXdUpkENGr0p1lnRzxHIwoULyQkTJqS83nnnHTI/Pz9h2YKCgozroSypPXv2DPUuM2RBNr6eSy+9lNy9ezdJkjGfjVwuJ6PRaE7bycbXc/DgQbK8vPx8sqAz6s55HTyZLTt27Mj4nlqthslkglarhclkytiE0uVy4corr8TDDz+MOXPmDNWunlM4HA6sWrUKra2tKC0txbZt2yCTyRKWqa+vx5133gmXywU2m42HHnoIq1atymr9wxU8y+FwsGnTJlx22WV0GtSECRMS0qDuv/9+eDwerFixAkAsDerdd9/NehvnEszwbYDEhwxs3boVy5YtS1kmGAxi+fLlWL16NX3RAbGbbvHixaiqqsLixYvhdDozbsflckGv1+Oee+4Z/IMYodTU1GDhwoU4efIkFi5cmDa+RyQS4eWXX8aRI0fw0Ucf4b777qML2PUFOYzBs0uWLEFDQwOamprw0EMPAQD+8Ic/0CEoO3bsgMViQX19Perr689bQQIYURow69evxyeffIKqqip88skndAJkXV0d1q5dCwDYtm0bvvzyS2zZsgVTp07F1KlTUV9fn9VNR7FhwwbMmzdvWI5ppLB9+3asWbMGALBmzRq88847KctUV1fTVR90Oh1UKhVsNltW62eCZ0covY3tzsI487wi25m7uro6ctWqVeRLL73U61TyuUYu/jqSJMl9+/aRY8eOJSORSFbrz8bXs2nTJvKOO+4gSZIkX3/9dXLFihU5HAFDLzCzbyORbGbuotEofvGLX+DPf/7zcO9eCkMx3Fy0aBEmTpyY8tq+fXtO+2YymXDTTTfhpZdeyrrXXbyvZ9y4cVi5ciXt66GGT7fddhvsdjsqKyvxxBNP9GrNMgwOjKN7iFm0aBHd4DCejRs3ZvX5zZs3Y8mSJQkO2bMFNdxcv349ampqUFNTg8ceeyztstkON8/2JMKSJUuwZMmShP/94Q9/oH8XCAR48803c1onw8BgRGmIGehNt2fPHnz11VfYvHkzPB4PgsEgJBLJWXlib9++HTt37gQQ8/FQ7a2SOXDgACwWCy6//PIB5VVRkwjr16/PeRKBYRTT29jurIw0zyN++ctfJqSo3H///b0uf7Z9Stn4eCKRCDlv3jzy1KlTA97fzs5OcsGCBWRlZSW5YMEC0m63kyRJkrW1tXSE9CuvvEJyOBxyypQp9OvQoUP93ibDsMHEKY1E1q9fj5UrV+KFF15AcXExPUyoq6vDM888g+eff37Y92kkDTflcjk+/fTTlP/PmDGDPjc33ngjbrzxxgFvi2HkcF6Xw2XIjTFjxmDnzp30cHP+/Pk4ceJEwjI33HADvvrqK7BYLHq4eddddzEOYoZkmBrdDAPn/vvvh1wupx3dDocDf/rTnzIuv2XLFtTV1WHTpk3DuJcMo4SMosSEBDBkTTaBogwMA4WxlBgYGM4GjKXEwMAwOmBEiYGBYUTBiBIDA8OIghElBgaGEQUjSgwMDCMKRpQYGBhGFIwoMTAwjCgYUWJgYBhRMKLEwMAwomBEiYGBYUTBiBIDA8OIghElBgaGEQUjSgwMDCMKRpQYGBhGFIwoMTAwjCgYUWJgYBhRMKLEwMAwomBEiYGBYUTRV4uljCUrGRgYGIYCxlJiYGAYUTCixMDAMKJgRImBgWFEwYgSAwPDiIIRJQYGhhEFI0oMDAwjiv8PJkF6d0QhMFQAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "i = 3\n",
    "plot_3d_point_cloud(reconstructions[i][:, 0], \n",
    "                    reconstructions[i][:, 1], \n",
    "                    reconstructions[i][:, 2], in_u_sphere=True);\n",
    "\n",
    "plot_3d_point_cloud(feed_pc[i][:, 0], \n",
    "                    feed_pc[i][:, 1], \n",
    "                    feed_pc[i][:, 2], in_u_sphere=True);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python3.6 tf",
   "language": "python",
   "name": "tf_py3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}