demo_process_results.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from chebai.result.base import ResultFactory, ResultProcessor\n",
    "from chebai.preprocessing.datasets.chebi import ChEBIOver100\n",
    "from chebai.preprocessing.datasets.base import XYBaseDataModule\n",
    "from chebai.models.electra import Electra\n",
    "from chebai.models.base import ChebaiBaseNet\n",
    "import torch\n",
    "import os\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import tqdm\n",
    "import matplotlib.pyplot as plt\n",
    "from torchmetrics.classification import MulticlassF1Score, MultilabelF1Score\n",
    "import numpy as np\n",
    "from functools import partial"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "outputs": [],
   "source": [
    "# not used\n",
    "class CustomResultsProcessor(ResultProcessor):\n",
    "\n",
    "    @classmethod\n",
    "    def _identifier(cls) -> str:\n",
    "        return 'custom_results_processor'\n",
    "\n",
    "    def process_prediction(self, proc_id, features, labels, pred, ident):\n",
    "        print(f'id: {proc_id}')\n",
    "        print(f'features: {features}')\n",
    "        print(f'labels: {labels}')\n",
    "        print(f'pred: {pred}')"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "outputs": [],
   "source": [
    "model_path_v148 = os.path.join('logs', 'chebi100_bce_unweighted', 'version_6', 'checkpoints',\n",
    "                          'per_epoch=99_val_loss=0.0252_val_micro-f1=0.89.ckpt')\n",
    "model_path_v200 = 'electra_c100_bce_unweighted.ckpt'\n",
    "model_v148 = Electra.load_from_checkpoint(model_path_v148).to('cpu')\n",
    "model_v200 = Electra.load_from_checkpoint(model_path_v200).to('cpu')\n",
    "data_module_v200 = ChEBIOver100()\n",
    "data_module_v148 = ChEBIOver100(chebi_version_train=148)\n",
    "#dataset = torch.load(data_path)\n",
    "#processors = [CustomResultsProcessor()]\n",
    "#factory = ResultFactory(model, data_module, processors)\n",
    "#factory.execute(data_path)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "outputs": [],
   "source": [
    "# construct mapping (not used)\n",
    "filename_200 = 'classes.txt'\n",
    "filename_148 = f'classes_v148.txt'\n",
    "with open(os.path.join(data_module_v200.raw_dir, filename_200), \"r\") as file:\n",
    "    v200_classes = file.readlines()\n",
    "with open(os.path.join(data_module_v200.raw_dir, filename_148), \"r\") as file:\n",
    "    v148_classes = file.readlines()\n",
    "# find index in orig_classes for every class in new_classes\n",
    "mapping = []\n",
    "for ind, v200_class in enumerate(v200_classes):\n",
    "    if v200_class in v148_classes:\n",
    "        mapping.append(v148_classes.index(v200_class))\n",
    "    else:\n",
    "        mapping.append(None)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "outputs": [],
   "source": [
    "# get list of classes that appear in v200 and v148\n",
    "common_classes = []\n",
    "for v200_class in v200_classes:\n",
    "    if v200_class in v148_classes:\n",
    "        common_classes.append(v200_class)\n",
    "# get filter if a class in v200/v148 is a common class\n",
    "common_classes_mask_v200 = torch.tensor([[c in common_classes for c in v200_classes]])\n",
    "common_classes_mask_v148 = torch.tensor([[c in common_classes for c in v148_classes]])"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "701\n"
     ]
    }
   ],
   "source": [
    "print(len(common_classes))"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "outputs": [],
   "source": [
    "# (not used)\n",
    "#mapping = [-1 if new_class not in orig_classes else orig_classes.index(new_class) for new_class in\n",
    "#                   new_classes]\n",
    "#input = torch.tensor(np.random.random([1, 854]))\n",
    "def _apply_mapping(input, index):\n",
    "    orig_ind = mapping[index]\n",
    "    if orig_ind is not None:\n",
    "        return input[0, orig_ind].item()\n",
    "    return None\n",
    "# mapping between model outputs / labels for chebi v200 (with 854 classes) and chebi v148 (with 709 classes)\n",
    "def apply_mapping(mapping: [], input: torch.Tensor):\n",
    "        input = input.detach().numpy()\n",
    "        output = np.array(np.zeros((1, len(mapping))))\n",
    "        for ind, value in enumerate(input[0]):\n",
    "            if mapping[ind] is not None:\n",
    "                output[0, mapping[ind]] = value\n",
    "        return torch.tensor(output)\n"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([  3.4000,   5.8000,   6.2000, 111.0000, 405.0000])\n",
      "0\n",
      "None\n",
      "1\n",
      "None\n",
      "2\n",
      "0\n",
      "3\n",
      "None\n",
      "4\n",
      "1\n",
      "tensor([[  6.2000, 405.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,\n",
      "           0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000,   0.0000]],\n",
      "       dtype=torch.float64)\n"
     ]
    }
   ],
   "source": [
    "input = torch.Tensor([[3.4, 5.8, 6.2, 111, 405]])\n",
    "print(input[0])\n",
    "for ind, value in enumerate(input[0]):\n",
    "    print(ind)\n",
    "    print(mapping[ind])\n",
    "print(apply_mapping(mapping, input))"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "outputs": [],
   "source": [
    "# get predictions from model\n",
    "def evaluate_model(model: ChebaiBaseNet, data_module: XYBaseDataModule, common_classes_mask = None, test_file=None):\n",
    "    collate = data_module.reader.COLLATER()\n",
    "    if test_file is None:\n",
    "        test_file = data_module.processed_file_names_dict['test']\n",
    "    data_path = os.path.join(data_module.processed_dir, test_file)\n",
    "    data_list = torch.load(data_path)\n",
    "    preds_list = []\n",
    "    labels_list = []\n",
    "\n",
    "    for row in tqdm.tqdm(data_list):\n",
    "        processable_data = model._process_batch(collate([row]), 0)\n",
    "        model_output = model(processable_data)\n",
    "        preds, labels = model._get_prediction_and_labels(processable_data, processable_data[\"labels\"], model_output)\n",
    "        if common_classes_mask is not None:\n",
    "            preds = preds[common_classes_mask]\n",
    "            labels = labels[common_classes_mask]\n",
    "            preds_list.append(preds.unsqueeze(0))\n",
    "            labels_list.append(labels.unsqueeze(0))\n",
    "        else:\n",
    "            preds_list.append(preds)\n",
    "            labels_list.append(labels)\n",
    "\n",
    "    test_preds = torch.cat(preds_list)\n",
    "    test_labels = torch.cat(labels_list)\n",
    "    print(test_preds.shape)\n",
    "    print(test_labels.shape)\n",
    "    f1_macro = MultilabelF1Score(test_preds.shape[1], average='macro')\n",
    "    f1_micro = MultilabelF1Score(test_preds.shape[1], average='micro')\n",
    "    print(f'Macro-F1 on test set with {test_preds.shape[1]} classes: {f1_macro(test_preds, test_labels):3f}')\n",
    "    print(f'Micro-F1 on test set with {test_preds.shape[1]} classes: {f1_micro(test_preds, test_labels):3f}')"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 16486/16486 [07:24<00:00, 37.09it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([16486, 854])\n",
      "torch.Size([16486, 854])\n",
      "Macro-F1 on test set with 854 classes: 0.603181\n",
      "Micro-F1 on test set with 854 classes: 0.902437\n"
     ]
    }
   ],
   "source": [
    "evaluate_model(model_v200, data_module_v200, test_file='test_martin_server.pt')"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "Results:\n",
    "- model_v200, test set from Martin:\n",
    "  - Macro-F1 on test set with 854 classes: 0.607030\n",
    "  - Micro-F1 on test set with 854 classes: 0.903165\n",
    "- model_v200, test set from me (?) (i dont know where it is from, but it was there and lead to bad results)\n",
    "- Macro-F1 on test set with 854 classes: 0.003233\n",
    "- Micro-F1 on test set with 854 classes: 0.155634\n",
    "- model_v200, test set from Martin (only using classes also present in chebi_v148):\n",
    "  - Macro-F1 on test set with 701 classes: 0.623063\n",
    "  - Micro-F1 on test set with 701 classes: 0.905059\n",
    "- model_v148, test set from ?\n",
    "  - Macro-F1 on test set: 0.510064\n",
    "  - Micro-F1 on test set: 0.854736\n",
    "- model_v148, test set from Martin (modified for training with 709 classes)\n",
    "  - Macro-F1 on test set with 709 classes: 0.513283\n",
    "  - Micro-F1 on test set with 709 classes: 0.854591\n",
    "- model_v148, test set from Martin (modified for training with 709 classes, only using classes also present in chebi_v200)\n",
    "  - Macro-F1 on test set with 701 classes: 0.519968\n",
    "  - Micro-F1 on test set with 701 classes: 0.855442"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 16486/16486 [07:21<00:00, 37.36it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([16486, 701])\n",
      "torch.Size([16486, 701])\n",
      "Macro-F1 on test set with 701 classes: 0.623063\n",
      "Micro-F1 on test set with 701 classes: 0.905059\n"
     ]
    }
   ],
   "source": [
    "evaluate_model(model_v200, data_module_v200, common_classes_mask_v200)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 16486/16486 [05:31<00:00, 49.69it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([16486, 709])\n",
      "torch.Size([16486, 709])\n",
      "Macro-F1 on test set with 709 classes: 0.513283\n",
      "Micro-F1 on test set with 709 classes: 0.854591\n"
     ]
    }
   ],
   "source": [
    "evaluate_model(model_v148, data_module_v148)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 16486/16486 [05:16<00:00, 52.07it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([16486, 701])\n",
      "torch.Size([16486, 701])\n",
      "Macro-F1 on test set with 701 classes: 0.519968\n",
      "Micro-F1 on test set with 701 classes: 0.855442\n"
     ]
    }
   ],
   "source": [
    "evaluate_model(model_v148, data_module_v148, common_classes_mask_v148)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([1, 854])\n"
     ]
    }
   ],
   "source": [
    "print(common_classes_mask_v200.shape)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "       epoch          variable     value\n",
      "0        0.0    val_loss_epoch       NaN\n",
      "1        0.0    val_loss_epoch       NaN\n",
      "2        0.0    val_loss_epoch       NaN\n",
      "3        0.0    val_loss_epoch       NaN\n",
      "4        0.0    val_loss_epoch       NaN\n",
      "...      ...               ...       ...\n",
      "17227    NaN  train_loss_epoch       NaN\n",
      "17228    NaN  train_loss_epoch       NaN\n",
      "17229    NaN  train_loss_epoch       NaN\n",
      "17230   99.0  train_loss_epoch       NaN\n",
      "17231   99.0  train_loss_epoch  0.013129\n",
      "\n",
      "[17232 rows x 3 columns]\n"
     ]
    }
   ],
   "source": [
    "# visualize results from csv\n",
    "df = pd.read_csv(os.path.join('server-results', 'version_6', 'metrics.csv'))\n",
    "df_loss = df.melt(id_vars='epoch', value_vars=['val_loss_epoch', 'train_loss_epoch'])\n",
    "df_macro = df.melt(id_vars='epoch', value_vars=['train_macro-f1', 'val_macro-f1', ])\n",
    "df_micro = df.melt(id_vars='epoch', value_vars=['train_micro-f1', 'val_micro-f1', ])\n",
    "print(df_loss)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 640x480 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "lineplt = sns.lineplot(df_loss, x='epoch', y='value', hue='variable')\n",
    "plt.show()"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 640x480 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.lineplot(df_micro, x='epoch', y='value', hue='variable')\n",
    "plt.show()"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 640x480 with 1 Axes>",
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAkAAAAGwCAYAAABB4NqyAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/SrBM8AAAACXBIWXMAAA9hAAAPYQGoP6dpAABp5klEQVR4nO3dd3hUZcLG4d9Meu+FQBJa6CV0ARVUFBQRcFVEVERW188uwiqoYFlFXXtFXF0XV8TuoiCKCNjoRXovCSWNkN5nzvfHCQORAAkkMynPfV1zZebMe868c0Dy+FaLYRgGIiIiIo2I1dUVEBEREXE2BSARERFpdBSAREREpNFRABIREZFGRwFIREREGh0FIBEREWl0FIBERESk0XF3dQXqIrvdzqFDhwgICMBisbi6OiIiIlIFhmGQm5tLTEwMVuvp23gUgCpx6NAhYmNjXV0NEREROQvJyck0a9bstGUUgCoREBAAmDcwMDDQxbURERGRqsjJySE2Ntbxe/x0FIAqcazbKzAwUAFIRESknqnK8BUNghYREZFGRwFIREREGh0FIBEREWl0NAboHNhsNkpLS11dDWkgPDw8cHNzc3U1REQaBQWgs2AYBikpKWRlZbm6KtLABAcHEx0drfWnRERqmQLQWTgWfiIjI/H19dUvKzlnhmFQUFBAWloaAE2aNHFxjUREGjYFoGqy2WyO8BMWFubq6kgD4uPjA0BaWhqRkZHqDhMRqUUaBF1Nx8b8+Pr6urgm0hAd+3ulsWUiIrVLAegsqdtLaoP+XomIOIcCkIiIiDQ6CkAiIiLS6CgASY3Zt28fFouF9evXV/mcW265hREjRpy2zMCBA7n//vvPqW4iIiIn0iwwqTGxsbEcPnyY8PBwV1dFRETqiJyiUnKLyrBawM1iwWKx4Ga14OPhho+n62a7KgBJjSgpKcHT05Po6GhXV0VERFzAZjfIKSzlYFYh65OzWJ+cxbqko+xOz6+0/P8NbMVDQ9o5uZbHqQusEZo5cyYxMTHY7fYKx4cPH86tt97K7t27GT58OFFRUfj7+9OrVy9+/PHHCmWbN2/OU089xc0330xgYCC33377SV1gNpuN8ePH06JFC3x8fGjbti2vvvpqpXV64okniIiIIDAwkDvuuIOSkpJT1r+4uJiJEyfStGlT/Pz86NOnD0uWLDmneyIiImdmGAaHsgqZv/Ewz8zfyvUzlzHwn4vp+sQPtH5kPt2eWsiVr//Ko19v4vM1Bxzhx9PdioebBesJE12tLp70qhagRujaa6/lnnvuYfHixVxyySUAZGZmsmDBAubPn09eXh5XXHEFTz/9NF5eXsyaNYthw4axfft24uLiHNd54YUXmDp1KtOmTav0c+x2O82aNeOzzz4jLCyM33//ndtvv50mTZpw3XXXOcotWrQIb29vlixZwr59+xg3bhxhYWE8/fTTlV737rvvZsuWLcyZM4eYmBi++uorhgwZwsaNG0lISKjBOyUi0viU2uyk5hRxOLuIg0cLSc4s4MDRQpKPFrArLY+03OLTnh/k40HnpkF0iwsmMdZ8hPl7VShjGAaGUZvf4swshuHqKtQ9OTk5BAUFkZ2dTWBgYIX3ioqK2Lt3Ly1atMDb29tFNTx3I0aMICwsjPfeew8wW4WeeOIJkpOTsVpPbhjs1KkTd9xxB3fffTdgtgB169aNr776ylFm3759tGjRgnXr1pGYmFjp5959992kpKTw+eefA+Yg6G+++Ybk5GTHIoAzZsxg0qRJZGdnY7VaGThwIImJibzyyiskJSXRsmVLkpKSiImJcVx30KBB9O7dm2eeeaZG7o+rNJS/XyJSPxSUlPH7riMs3p7GpkM5HM4qJD2v+LThxM1qoV10AImxwXSNDaZ5mB8hvh4E+3oS5OOBp7vrOpdO9/v7z9QC1EiNGTOG2267jbfeegsvLy8++ugjrr/+eqxWK3l5eTz++OPMmzePw4cPU1ZWRmFhIUlJSRWu0bNnzzN+zptvvsn7779PUlIShYWFlJSUnBSOunbtWmFl7b59+5KXl0dycjLx8fEVym7cuBGbzUabNm0qHC8uLtbWJCIiJyiz2fl99xHm/nGIlOwign09CPXzJMTXE093Kyv2ZrJ8zxFKyuwnnevhZiEq0JtmIT7EhvjSLMSX2FAf4sN86dAkyKWDl2uKAlAjNWzYMAzDYN68efTq1YtffvmFl19+GYCJEyeycOFCXnjhBVq3bo2Pjw/XXHPNSeNy/Pz8TvsZc+bMYeLEibz44ov07duXgIAA/vnPf7JixYqzrndeXh5ubm6sWbPmpL2y/P39z/q6IiINQZnNzuZDOfxv/SHm/nGIjLzTd1cBNA324eJ2kfRtFUazEB+ig7wJ9/PC6upBOrVMAaiR8vb25uqrr+ajjz5i165dtG3blu7duwPw22+/ccsttzBy5EjADB379u2r9mf89ttv9OvXjzvvvNNxbPfu3SeV++OPPygsLHRsBrp8+XL8/f2JjY09qWy3bt2w2WykpaVxwQUXVLtOIiL1mc1ucCS/mLScYtJzi0nLLSIps4DdafnsTs9j35F8Sm3H+69CfD0Y2qUJibEhZBeWcjS/hMyCEvKKyujUNJCL2kbSOtK/UW7DowDUiI0ZM4Yrr7ySzZs3c+ONNzqOJyQk8OWXXzJs2DAsFguPPfbYSTPGqiIhIYFZs2bx/fff06JFCz788ENWrVpFixYtKpQrKSlh/PjxPProo+zbt49p06Zx9913VzoWqU2bNowZM4abb76ZF198kW7dupGens6iRYvo0qULQ4cOrf6NEBGpQwzD4Eh+Cfsy8tmTbgYb85FPUmYBNvvph+76erpxcbtIRiQ25cI2ES4dk1OXKQA1YhdffDGhoaFs376dG264wXH8pZde4tZbb6Vfv36Eh4fz0EMPkZOTU+3r/+1vf2PdunWMGjUKi8XC6NGjufPOO/nuu+8qlLvkkktISEjgwgsvpLi4mNGjR/P444+f8rr//ve/+cc//sGDDz7IwYMHCQ8P57zzzuPKK6+sdh1FRFyppMzOxoNZrNx7lE0Hs9l3JJ/9RwrIKy475TkWC4T5eRIR4E1kgBdNQ3xoHeFPq0h/WkX4ERPk0+C7r2qCZoFVojHMApO6SX+/RBoewzA4WlBKSnYRKTmFpGQXczCrgLX7s1iXfJSi0pNb2C0WaBLoTcsIM9S0jvSnVYQ/LSL8iPD3wt1NrTqV0SwwERERF8jIK2b5niPsTM1jb0a+43G6Fp1QP096Nw+le3wwrSL8iQ8zZ115e9T/mVZ1mQKQiIjIWbLZDdYnZ7F0expLdqSz4UD2KcuG+XkSHeRNdKA3UUHedIwJpE+LUFpFNM5ByK6mACQiIlIFxWU2th7OZcuhHDYfymbL4Ry2Hc6lsNRWoVyHJoF0bhpEiwg/Wob70TLCTy06dZACkIiIyAkMwyC3uIz03GK2Hc5lbdJR1iYdZfPBHEpsJ4/XCfB258KECAa2jWBAmwgiAzV+rz5QABIRkUbHMAxScorYnpLreOzOyCcjt5j0vOJKV0cGc12dTk2D6BATSMeYIDo0CaRFuB9umnVV7ygAiYhIg2QYBpsP5bB0RzqbD2WTVVBa/ijhaEHpSV1Xfxbg5U58uC/dYkPoHh9Mt9gQ4sN8NV6ngVAAEhGRBqGkzM7ejHy2HM7m151H+HlnOumn2bnczWqhRbgfbaMDaBcVQEKUP1GB3oT7exER4KUxOw2cApCIiNQ7BSVlbDiQzbqkLDYdymZnai570vMp+9Mqyb6ebvRrFcZ5LcOICPAiyOf4ruVNgrwVchoxBSAREanzDMNg9f6jfPvHIVbvP8q2lNxKt4Tw93InIcqfXs1DGdgmgh7NQ/ByV8iRkykAyVlp3rw5999/P/fff7+rq1JnfP3110ycOJG9e/dyzz338Morr7i6SiL13uHsQr5ce5DP1xxgb0Z+hfeiAr3oHhdC19hg2kYH0CYqgJggb43RkSpRAGpEBg4cSGJiYo38Yl61ahV+fn7nXqkG5G9/+xvjxo3j3nvvJSAggKKiIu644w7WrFnD1q1bufLKK/n6669dXU0Rl8vIK+Z/6w+RnFlAmd2OzW5QajMotdkpKLGRX1xGfvnPPel5HGvo8fV04/JOTbioXQTd40KICfZx7ReRek0BSBwMw8Bms+Hufua/FhEREU6okXOVlJTg6el5Vufm5eWRlpbG4MGDiYmJASA/Px8fHx/uvfdevvjii5qsqkiddqxr6sSp4Ta7wc870pmzKolFW9NOGqtzOr1bhHJtj2Zc0bkJfl76tSU1Q7up1QDDMCgoKXP6ozr72N5yyy0sXbqUV199FYvFgsVi4YMPPsBisfDdd9/Ro0cPvLy8+PXXX9m9ezfDhw8nKioKf39/evXqxY8//ljhes2bN6/QkmSxWPjXv/7FyJEj8fX1JSEhgblz51apbkuWLMFisfD999/TrVs3fHx8uPjii0lLS+O7776jffv2BAYGcsMNN1BQUOA4b8GCBZx//vkEBwcTFhbGlVdeye7duytc+8CBA4wePZrQ0FD8/Pzo2bMnK1asAODxxx8nMTGRf/3rXxU2H01KSmL48OH4+/sTGBjIddddR2pq6mnrHxAQAMDFF1+MxWJhyZIl+Pn58fbbb3PbbbcRHR1dpXshUp9tT8ll6v82kfjED7SaMp+OUxfQ55kfGfTSUs6bvohxH6zi+82plNkNEmOD+b+BrbjvkgQevLQNDw1px6ND2/PMyM68en0i/7q5J7Nv68OvD13Ep3/ry7U9YxV+pEbpb1MNKCy10WHq907/3C1PDsbXs2p/hK+++io7duygU6dOPPnkkwBs3rwZgIcffpgXXniBli1bEhISQnJyMldccQVPP/00Xl5ezJo1i2HDhrF9+3bi4uJO+RlPPPEEzz//PP/85z95/fXXGTNmDPv37yc0NLRKdXz88cd544038PX15brrruO6667Dy8uL2bNnk5eXx8iRI3n99dd56KGHALOFZcKECXTp0oW8vDymTp3KyJEjWb9+PVarlby8PAYMGEDTpk2ZO3cu0dHRrF27Frv9+AJnu3bt4osvvuDLL7/Ezc0Nu93uCD9Lly6lrKyMu+66i1GjRrFkyZJK692vXz+2b99O27Zt+eKLL+jXr1+Vv7NIfVdUauP7zSl8tDyJlfsyK7yXX2Ijv8RGao45FT3E14OR3ZoxqlcsbaMDXFFdEQcFoEYiKCgIT09PfH19Ha0R27ZtA+DJJ5/k0ksvdZQNDQ2la9eujtdPPfUUX331FXPnzuXuu+8+5WfccsstjB49GoBnnnmG1157jZUrVzJkyJAq1fEf//gH/fv3B2D8+PFMnjyZ3bt307JlSwCuueYaFi9e7AhAf/nLXyqc//777xMREcGWLVvo1KkTs2fPJj09nVWrVjkCSevWrSucU1JSwqxZsxxdegsXLmTjxo3s3buX2NhYAGbNmkXHjh1ZtWoVvXr1Oqnenp6eREZGOu6dWnukoTtwtIDF29NZsi2N33ZnUFRq/k+Fm9XCpe2juPG8eNo3CSCvuIzcojJyikrBQDOypE5RAKoBPh5ubHlysEs+tyb07Nmzwuu8vDwef/xx5s2bx+HDhykrK6OwsJCkpKTTXqdLly6O535+fgQGBpKWllblepx4flRUFL6+vo7wc+zYypUrHa937tzJ1KlTWbFiBRkZGY6WnaSkJDp16sT69evp1q3baVtj4uPjK4xn2rp1K7GxsY7wA9ChQweCg4PZunUrvXr1omPHjuzfvx+ACy64gO+++67K31GkPrHZDfYdyWdnai7bU/LYkZbL1sM57EmvOBsrJsib63rFcn2vOKKDju+DFebv5ewqi1SZAlANsFgsVe6Kqov+PJtr4sSJLFy4kBdeeIHWrVvj4+PDNddcQ0lJyWmv4+HhUeG1xWKp0N10Jieeb7FYzni9YcOGER8fz7vvvktMTAx2u51OnTo56unjc+YZImczk23+/PmUlpZW+TNE6gvDMNh/pIBfd2Xw684Mft+dQU5R2UnlrBboER/CwLaRXNwuknbRAZp6LvVO/f2tLdXm6emJzXb6vW8AfvvtN2655RZGjhwJmC1C+/btq+XaVc+RI0fYvn077777LhdccAEAv/76a4UyXbp04V//+heZmZlVHpPTvn17kpOTSU5OdrQCbdmyhaysLDp06ACYrUYi9V1RqY2dqXnsSM1lZ1oeO1PN1p1D2UUVyvl4uNE60p+EKH/aRAXQJsqf7nEhBPue3YxJEQCK88DL36VVUABqRJo3b86KFSvYt28f/v7+p2ydSUhI4Msvv2TYsGFYLBYee+yxarXkOENISAhhYWHMnDmTJk2akJSUxMMPP1yhzOjRo3nmmWcYMWIE06dPp0mTJqxbt46YmBj69u1b6XUHDRpE586dGTNmDK+88gplZWXceeedDBgw4KSuwqrYsmULJSUlZGZmkpuby/r16wFITEys9rVEzkV2YSlr9x9l5b5MVu7NZOOBbEpsJ/937eFmoUd8COe3Dqd/63A6Nw3C3U0ThqUGbf8Ovr4Trp8N8ZX/W+wMCkCNyMSJExk7diwdOnSgsLCQf//735WWe+mll7j11lvp168f4eHhPPTQQ+Tk5Di5tqdntVqZM2cO9957L506daJt27a89tprDBw40FHG09OTH374gQcffJArrriCsrIyOnTowJtvvnnK61osFv73v/9xzz33cOGFF2K1WhkyZAivv/76WdXziiuucIwXAujWrRtAtZYwEKmOMpudQ1lFbC9v0Tn22J9ZwJ//2oX6eZIQebxlp3VkAF1jg+p1l77UYXY7/PxPWPKM+XrFDJcGIIuhf4lPkpOTQ1BQENnZ2QQGBlZ4r6ioiL1791ZYN0akpujvl1TH0fwSftudwep9R9l3JJ/9RwrKV1eu/J/15mG+9GoeSu8W5iMu1Fdjd+q7Y7/Ca/LPMSsJNn0BO36AhEuh//1graQV0DBg789QlA0BTSAgGvyjwL2S7tGiHPj6/2Dbt+br3rfD4GfAzePksufgdL+//0wxX0Sknigps7Mu6Si/7Mzgl53pbDiYfVKrDoCXu5UW4X50aBJIe8cjQLOy6qrSQljzAaRugoGTIahZ1c7b/RN8cx94BcGIN6FJ1zOfU5niXMg+CPt+MYNP0rLj7yX9Dskr4ep3wDvo+PG8NJh7L+yoZBasXwREtIPI9hDZAQKbwg+PQMYOcPOEK1+GbjeeXV1rkMsD0Jtvvsk///lPUlJS6Nq1K6+//jq9e/eutOzmzZuZOnUqa9asYf/+/bz88suVbsZZnWtK7bvjjjv473//W+l7N954IzNmzHByjUTqh+IyG3sz8vl1Zwa/7cpgxd5MCkoqTmRoGxVAv9ZhtIkKID7MlxbhfkQFeGO1qmWnzisrgXWz4OcXIPeweWz7Arj239DiwlOfV1oIPz5udiEd8+4lcNEU6H8fWE9YIqUgEzZ/Cek7oKwQSovMnyUFkJsC2QegOPtPH2CB5udDs56w7C0z5My8CK7/yAw1W78xg1fBETPQRHcxA1FeCthKID/dfOz7peJlA2Jg1H+hWY9zuWs1xqUB6JNPPmHChAnMmDGDPn368MorrzB48GC2b9/uWFjuRAUFBbRs2ZJrr72WBx54oEauKbXvySefZOLEiZW+d6YmSpHG4Ng+WV+sPcCBo4Vk5pdwNL+E3OKTp6CH+XnSr3U4FyaEc0FCRIV1d6SesJXCH3Ng6fOQXb6+WlAseAVA2haYNQIufRL63nVy19ah9fDl7ZCx3Xzd66+Ql2qGkkVPwM6FMPwNs7Vl/WxzwLG99Mx18g6C8DbQcaT5CDT3NKT9VfDJTZC52wxZLQfA9vnme1GdzZahqI7ma8OAwqOQtR/StkHaZkjbCunbIaoTDHsVAqLO9e7VGJeOAerTpw+9evXijTfeAMButxMbG8s999xz0oyeP2vevDn333//SS1A53LNYzQGSFxFf78al4y8Yj5ZlczHK5M4cLSw0jI+Hm70bhHqmJXVLjpArTt1VcZO2PSl2fIR3Rk6XwMx3Y+HmJICWPdf+P01yE42j/lHwYWToPvNYLfBtw/Ahjnmex2vhl7jzTCTvt187PsF7GXmecPfgoRBZvBYPxu+ewhKck+uV5Ou0Opi8PQDdx/w8DZ/BkRBYDMIamqGr1PJz4DPb4W9S8sPWMyWpoumgHvd6latF2OASkpKWLNmDZMnT3Ycs1qtDBo0iGXLlp3mzJq/ZnFxMcXFxY7XdW3Gk4jUb4UlNlbuyyTpSD4Hsgo5lFXEwaMFbDyYTanN/H/QIB8PrunRjN4tQgnz8yTUz5MwPy8Cfdw1ULmuspWZ4WTn9+bYmZSNx9/b9wssfwtCW0Hna8HNHZbPgIIM832/COh3D/S6DTx9j583cgY07QHfTza7rjZ/efLnth8GV74KfmHma4sFuo2B5v3hy79B8nLz+l1GQdfREN3p3L6nXzjc+CUsfQ72/wYXP+bS2Vs1xWUBKCMjA5vNRlRUxeawqKgoxx5Vzrrm9OnTeeKJJ87qM0VEKpNVUMKirWl8vzmFn3emO/bL+rOuscHc2CeOYV1j8K6h7W3qLVuZGRSc7dhspuSVEN8P4vpWPuspcy/sXgSHN0DKBrN7p+yEhSOt7tDyInPmVPIK2Dbf7Dpa+uzxMsFx0O9ecxCwRyUryVss0Od2swVp3gRzwcCItuWPdmaYaZJY+ayvkOYwbj6kbzO7s2pyhpWbO1z8SM1drw5w+SDoumDy5MlMmDDB8TonJ6fCXlAiIqez/0g+21Ny2ZWex660PHan5bHpUA62E6ajNw32oWNMIE1DfGga7ENMsA8Jkf4kRGlXdGxl8OM0WPku9LwVLnuq8l/eW7+BJc+ZAWPA3ysPECc6uh/2LDZnS+UcNsevtLvS7BKyWMx1aXZ8B7+8CAfXHD8vsCl0utpsuTHsZpDZNs8c0/Jnnv7mYOGOI83xMr7lq873+ZsZXrbPh42fQ0kedB9rXrcqwSS+L9x5Fr0hVrfjY3LktFwWgMLDw3FzcyM1NbXC8dTU1LPeTftsr+nl5YWXV93qxxSRus1uN1i8PY23l+xm9f6jlZZpFx3AZR2jGdwxig5NAtWVVZnCLHN8ye5F5usVb5vTwa/9z/EuHlupOetpmTm2k9SNsPkrGPYKtBx4/Fp2u9nysvkr83pHdlX8rAMrzYX4guLMELX/d0jfar7n7g0tBphTwHMOwu+vm48TWdzMFqLYPmYLTXRnCGlReWsRmFs9dLnOfEid47IA5OnpSY8ePVi0aBEjRowAzAHLixYt4u67764z1xQROVGpzc63Gw4xY8ketqeaA0493Cy0iQqgdaQ/rSP8aR3pT8eYIOLCfM9wtUaiJN8cIBzWquJg2yO7YfYoOLLTHJR73v/Bypnm+JmZA81p134RZkBK+t08p+to2LMUju6FWcOh6w3QY6zZ0rLpy+ODi8EMLM16mQOAA2NgxwLYtcicebX6PbOMV6A50Pi8O8E/0pwmvvMH2PS5OSXd6gatLzFbjhIuO97CI/WeS7vAJkyYwNixY+nZsye9e/fmlVdeIT8/n3HjxgFw880307RpU6ZPnw6Yg5y3bNnieH7w4EHWr1+Pv78/rVu3rtI15eydauZdQ7Vt2zZuueUW1q9fT7t27Rz7eEnjUFRqY11SFhsOZHHgaCEHswo5cLSAA0cLHWvx+Hu5M+a8OMb3b0FkYCOdtZeXZs5OCowxF/A7NiuopMAMEpu/gh3fm2vPYDHHpsR0g7DWsOx1cxXhwKbmvlAxiWZryZwbIHMPvHeZ2YqSnw6eATDiLehwlbmq8KInYdW/4I/Z5uMYzwBofyW0G2qupXPi4n3dbzLrtfsns2ssKBZ63AI+wcfLeHibn9HhKigrNutc2crGUu+5NACNGjWK9PR0pk6dSkpKComJiSxYsMAxiDkpKQnrCU2Lhw4dcuylBPDCCy/wwgsvMGDAAJYsWVKla4pU1bRp0/Dz82P79u34+5u7Fj/99NPMmzeP9evX4+npSVZWlmsrKTUmv7iM1fuPsnLvEVbsyeSPA1mOGVp/Fu7vybj+LbjxvHiCfGp2KX+XSdtmhhWruznrxz/SbH3xizCnXJ84U6kg0xyPs+kLs7XGODbA22JuiRAYY65nU1pw/BzPAHOKdsb242vYgNlCM+qj4+vDRLaH236Cz8eb3VhlhRDZEUZ9aLYgAXgHwtAXzLA0bwJk7DKng3e+1mylOd3YIE9fMyC1v/LM96SOTfGWmqW9wCqhdYAqV99agAzDwGaz4e5+djm/Z8+eDB06tMIMwWnTphEcHMyBAwd47733ajwANea/X85WVGpjxd5Mlu85wvI9R9h4IPukPbQiA7zo1TyU5uG+NA32pWmID81CfIgL9cXDFTuk221mq0dQM7OFo6psZXBondkaEtri+CBcux12LYTlb5stIqfjGWCGFK9AcwaU/YRFGoPjzLViTgw8x44fW1ivSaLZWnR4vVmXwxvMmU0DHjJbXSr7rsveMMcIXTipYgD7M8Oo2b2wpN6qF+sANSiGcfJ/+M7g4Vvl/+hnzpzJ448/zoEDByq0qg0fPpywsDAeeeQRJkyYwPLly8nPz6d9+/ZMnz6dQYMGnVXVLBYLM2bM4JtvvuGnn34iPj6e999/n4iICP7617+yatUqunbtyocffkirVub/1e3evfuMdSguLmbq1KnMnj2btLQ0YmNjmTx5MuPHj2fJkiVcdNFFzJ8/n0cffZSNGzfyww8/0LdvXyZNmsScOXPIycmhZ8+evPzyy/Tq1eu09QdYs2YNTz75JNOmTePxxx93hKEPPvjgrO6LuN6utDw+XpnEF2sPkFVQcYXcZiE+9G4RynktwujdIpT4sDq0WajdDt/cay6kB9DlerMV5HQL2AHkHILPbjEHB4PZwhPa0uyKSttqTtMGsFihzRDwDTPDTH465KdBXnr51gm5cOSERfaiO5sL9XUcaYYqwzC3Rsjab26vENSs4iKAYAaogMHQZvCZv6/VzVxsryrqyp+R1CsKQDWhtACeiXH+5045ZK7sWQXXXnst99xzD4sXL+aSSy4BIDMzkwULFjB//nzy8vK44oorePrpp/Hy8mLWrFkMGzaM7du3ExcXd1bVe+qpp3jppZd46aWXeOihh7jhhhto2bIlkydPJi4ujltvvZW7776b774zN9OrSh1uvvlmli1bxmuvvUbXrl3Zu3cvGRkZFT734Ycf5oUXXqBly5aEhITw97//nS+++IL//Oc/xMfH8/zzzzN48GB27dpFaGjlAxoPHz7MoEGDGDJkCBMnTnR0gUn9VFBSxg+bU5m9MomVezMdx6MDvTk/IZzzWobRp0UosaF1dNCyYZibSa77rxlUwFwt+MBK+Mt70LR75eftWQpfjDfDjLuPGRRKC8zF+zJ2mGW8gsyxMb1vh5D4yj+7OLd8r6dUcyG/iPYQ0aZiOYvF7DrzCzcX8hOp4xSAGomQkBAuv/xyZs+e7QhAn3/+OeHh4Vx00UVYrVa6dj2+k/BTTz3FV199xdy5c896Bt24ceO47jpz+udDDz1E3759eeyxxxg82Py/v/vuu6/C4PSuXbuetg47duzg008/ZeHChY5WoZYtW570uU8++SSXXnopAPn5+bz99tt88MEHXH755QC8++67LFy4kPfee49JkyZVWvfo6Gjc3d3x9/c/62UZxLVKyuz8vCOduX8cYuGWVApLzYHLVgtc3C6SG/rEMaBNJG51ZVuJ4lxI3WIuZvfn/ZKWPmeuKgww/E1z6vUXfz0+UPiiKdD2iuNbGtjt8Nsr8NNT5vicqE5w3SzzvJyDxwOQp7/ZguN1moBvsZhjbrwDIbx1bX17EadTAKoJHr5ma4wrPrcaxowZw2233cZbb72Fl5cXH330Eddffz1Wq5W8vDwef/xx5s2bx+HDhykrK6OwsJCkpKSzrl6XLl0cz48NQu/cuXOFY0VFReTk5BAYGHjGOqxfvx43NzcGDBhw2s/t2bOn4/nu3bspLS2lf//+jmMeHh707t2brVvN9T/+vFt9Xl7eWX9ncT673WBrSg77Mgocs7QOHC1gbVIW2YXHu7jiQn0Z2a0po3rFEhN8hgX0nCHnsDnuJnklHFhlDho27Dh24u44EjoMhw2fwhJzJixDnoPEG8zn//crzL0Xts41N8FcVD5WzSvQHOtzbDp41xtg6IvHx9AEx5qP1pc49euK1DUKQDXBYqlyV5QrDRs2DMMwmDdvHr169eKXX37h5ZdfBmDixIksXLiQF154gdatW+Pj48M111xDSUnJWX+eh8fx2THHxlFUdsxut1epDj4+Vful5edXvT+L0+1WL3WTYRhsPJjNtxsO8+0fhziUXVRpucgAL67sEsNViTF0bRZU/fE8eenmTKQWFx7fHftclBWbu3Ov+695XcfsqXJ+EWZ31b5fzMf8SWCYLVdc9Aicd8fxsj4hZqvO2lnmCsrZSeaU8uIc8+HmCVf801x9WGNkRE6iANSIeHt7c/XVV/PRRx+xa9cu2rZtS/fu5tiB3377jVtuuYWRI0cCZivIvn37nFq/M9Whc+fO2O12li5dWuXB2a1atcLT05PffvuN+HhzfENpaSmrVq1yzGaLjIwkMjKyRr+L1IyCkjJ+3pFOel4J2QUlZBWUcrSglNX7M9l/5PjEAz9PN9o3CaRZiA/NQnxpFuJDq0h/useFnH0XV/Iq+PQmyD1sjrtpORASx5izr860BcOJcg6ZLTz7fjW3RCg8PgaJpj3MlYWb9YJmvSGwCWQlmdPRN30Bh/8wy/W925wJ9WcWi7kIYI+x5uviPPPzcg+Z6+wENTu77y7SCCgANTJjxozhyiuvZPPmzdx4442O4wkJCXz55ZcMGzYMi8XCY4895miZcZYz1aF58+aMHTuWW2+91TEIev/+/aSlpTnGGv2Zn58f//d//8ekSZMIDQ0lLi6O559/noKCAsaPH1/tOiYlJZGZmUlSUhI2m82xOGLr1q01ULoGFZSU8dHyJN75eTcZeZW3Qnp7WLmkfRTDusQwsG1EzW4kuuY/MH8i2ErMlpbCo+biebt/MruYojubm2CWFUNpoblVg3egWfbYo/AoHFhthpETBTQxVzNOHFP5mJrgOHP2U//7zJWSsw+YLVBVacXx8jcHJ/95gLKInEQBqJG5+OKLCQ0NZfv27dxwww2O4y+99BK33nor/fr1Izw8nIceeoicnByn1q0qdXj77beZMmUKd955J0eOHCEuLo4pU6ac9rrPPvssdrudm266idzcXHr27Mn3339PSEhItes4depU/vOf/zheH1uYc/HixQwcOLDa15OK8ovLmL2iYvBpGuxD56ZBBPt6EOTrQbCPJ3GhvgxsG4GfVw3/E1ZWAgsePr5NQrsrYeQMs1vqjzmw/mOzq2n/byefm32Ka1qs5kJ+zXqY12t1sTnFuyrCWh1f/E9EapQWQqyEFkIUV2lMf79W7DnC1+sPcTi7kJTsItJyi8nMP97aExfqy90Xt2Zkt6a1s+hgWbE5rTtjp/k4stPcHDNtC2Axx9xc8GDFjS6PbbaZe8icVu7hbW6iafUwx90UHj3+cPeCpj3N7R3qwRhBkYZACyGKSJ21LSWH5xds56dtaZW+Hx/my10X1XDwObQOFk+Ho/ugKMscLFxW+cBpvALhL/+qfLE+qxXi+9ZMnUTEpRSApNo++ugj/va3v1X6Xnx8PJs3b3ZyjaSuMwyDfUcKeP2nnXy17iCGAW5WC9d0b0aP+BAiA72ICvQmOtCbYF+PymdrZR+EvUth789ml5RvGPiEmj/9wiCuL0S0qzhWpjgPFj8NK2acPOMKzJab0JYQnlD+aAMtLzIHI4tIg6YAJNV21VVX0adPn0rfO3GauzReO1NzWbg1lV1peexOy2N3ej55xcf3jhrauQkPXtaGlhFnGDh+dD/8/rq5Xs6RXWf+4LAEcxfv9sMgNwXmTYScA+Z7na+FbjeZA5S9g8wdwD0DKnZxiUijoQAk1RYQEEBAwBn2H5JGxzAMVu8/yowlu1lUSfeW1QL9W4cz8bK2dI0NPtPFYMMnZoApKd9/ymI1N9RsOQBCW5WPtck095/KPmBOMz+yE3550XwcExwPV74Erc9uXzsRaZgUgM6Ss6eIS+NQH/9e2ewGi7amMmPpbtYmZQEQYsnlqhYQ0aobrSIDaR3pT1yYL17uJ8x+OvwH7F8GUR3MNXCO7QheeBS+nQCbvzRfx/Yxp4TH9zdbbU6lKAd2/gBb/ge7fjQHOfe7GwY8fPqdxEWkUVIAqiZPT0+sViuHDh0iIiICT0/PurNbtNRbhmFQUlJCeno6VqsVT09PV1fpjPKLy/h8zQHe/20v+48UEEEWN3us4aagP2idvw7LIRvkx5pbNzS7AdwDwFYG276FFe9A0u/HL+bmBbG9zQUBN3xqdltZ3GDgZDj/AXCrwj9V3oHQ+RrzUVpojvnR7CsROQVNg6/EmabRlZSUcPjwYQoKCio5W+Ts+fr60qRJkzoZgAzDICOvhKTMfH7YlMLiVetoVbKdROsuervvIpEdWDjhnxN374ozreL7m2N6jo3Jsbqbx9K3Q15KxQ8LbQlX/8tcO0dEpIo0Db6WeXp6EhcXR1lZGTabzdXVkQbCzc0Nd3f3OtOiWFxm48ctaczfeJjd6XkkZRZQWlLM/e6fM97tZyZbsuDPOS2mu7mBZ/th5t5Z2+aZ+17tWXJ88UDfMOgxDnqNN8sYhrni8b6fzS6x4Fg4f8LpdygXETlHagGqRHUSpEhDs+lgNp+vOcDX6w+SVXB8N/VYSypveLxOV+seAOwWNyxRHbE062nuadVigBleKpOVbO5a7hsGHUYcH+8jIlKD1AIkIlVmtxusP5DFj1tSWbgllZ1peY73ogO9+UuPplxhXU77VVOxluSCdzBc+RLWNpdXfXBxcCz0vat2voCIyFlQABJppLYezuE/v+/jx61pZOQVO457ulm5tGMU13WP4Xz/w7itfQvWlu9/FnueuUryqVp6RETqCQUgkUZmW0oOry3ayfyNxwceB3i5M6BtBCPiCulfthyfg+/DV8vN/a0AsMAFE2DglKrNyBIRqeP0L5lIQ3RoHexaBJEdoHl/8A5iy6Ec3ly8i3kbDwPmjhFXdG7C6F5x9G7ihuevz8Oid8A4YWC/V6C5xUTfu8wFCEVEGggFIJGGJHMP/PQP2PSF45Dd4sYOt9b8WNSePHtbmlliSOzUiXsHtaVNhB/88TG8Pc3cXwug5UBIuMycoh7dGaxulX+WiEg9pgAk0hDkZ8DS52H1+2AvxcDCZp+e+BUk04IU2pVtp5379uPl9/rAl60BA1I3mcfCEuDy56D1JS75CiIizqQAJFKflRbB8rcwfnkJS/meWb8aXXmmZBRbipoDcHF0MeOb7qenfQNeGVvMNXfKCiF1o3kNT38Y8Hfo83/gXvcWYBQRqQ0KQCJ1VV46rPsQNn9lrp/T9nJoMwRC4s3FAzd9QdnCabjnHMACbLC34Nmy0fxu70RMkDd/S4zhL92b0SbqTxvX2sogaz9k7IS8VLO7K7CJS76iiIiraCHESmghRHEZw4Ck5bD6PXNTT1vJyUUiO5BXZiUg0+y6OmSE8nzp9Sz1GsDlXZoyvGsMvZqHYrXWjRWlRUScRQshitQXhgFH98K+X48/cg4efz+mO/S4BYpzKNk8D/eDK7GmbSEAyDe8eKtsOBtib+Davm15rmNUxd3WRUTklBSARFwlZSN8erM5c+tE7j7mjua9xmOPTmTx9jQ+3pTET3taEmDkMdC6nliPHOhyHSPO786kP3dxiYjIGSkAibhCYRZ8ciMc3QdWD2jWE5qfbz6a9cbu7sN3m1J4/dNf2JaS6zitTfNYLuzVnys6N8HHU609IiJnSwFIxNkMA/53lxl+guLg9iXgFwaAzW4wf+NhXv9pFTtSzT25Arzcub53LKN6xdE6Ujuki4jUBAUgEWdb9iZs+xbcPOG6D8AvjPziMr5Ye4B//7aPvRn5AAR4uzOufwvG929BkK+Ha+ssItLAKACJ1DTDMMf1ZO6B6C4QEHX8vaQV8OM08/ngZzjk14H/zN/KxyuTyCkqAyDQ253x57fklv7NCfJR8BERqQ0KQCLnyjDM2Vu7f4JDa819uIqyy9+0QLNe0P5KiD8fPh8H9jLS44fy2LZuLPx6MTa7uRJF8zBfxvVvwTU9muHnpf80RURqk/6VFTkXSStg0ZOw/9eKx928IKgZZO6GAyvNR7lkSwxDtl9NPqkA9G0ZxvjzW3Bxu0it3SMi4iQKQCJnI2UT/PQU7Fhgvnbzgk5/gdjeENPN3IXd3ROyD2LfNp/UlV8QfmQlRYYnfy2+Fzz9ual7M27qG3/ySs0iIlLrFIBETqWsBHb9aO6snrUfinKgOBeKc6DEnKGFxQqJY2Dgw2aLz5+szfZl2sr2bDx4DwEU0CrcmxvP78LIbk3xVzeXiIjL6F9gkRMZBhxcCxvmwMbPoTDz1GU7joSLHoHwhJPe2pWWxztLd/PZmgOAOZV9wmU9uem8eNzdrLVVexERqSIFIJFjDAO+vR/WfHD8mF8kdL4W4vuCdxB4BYJXgLk5qU9whdOP5pfwzYZDfLHmAH8cyHYcv6ZHMx4a0o6IAC+nfA0RETkzBSCRY9bOMsOPxWqO5+lyPbQcCG6n/88kv7iMaXM387/1Bym1mTO63KwWBraJ4M6LWtMjPqT26y4iItWiACQCcHgDzJ9kPr/4MbhgQpVOO5hVyF//s5qth3MA6BgTyNXdmzE8MYZwf7X4iIjUVQpAIkXZ5qaktmJoMwT631+l09YmHeX2WWvIyCsm3N+TN2/oTp+WYbVbVxERqREKQNK4Ofbl2mvuyzXibbCeeZDy/9YfZNLnGygps9MuOoB/je1JsxBfJ1RYRERqggKQNG7L34at35g7sl/7AfiGnrb45kPZvLN0D3P/OATAoPaRvHJ9N01pFxGpZ/SvtjQudru5VcWuheYaPwdWm8cHPw3NelR6imEY/L77CDOW7uaXnRmO47df2JKHhrTDTas3i4jUOwpA0vAVZJr7dO38wQw9BUcqvt99LPS+vdJTl+85wjPzt7KhfFq71QJXdonh9gtb0qlpUG3XXEREaokCkDQcRTmQe9h85Bw2V2/e/RMcWAWG/Xg5r0BoOQBaXwqtB0FQ05MutS8jn+nfbeX7zeZ+Xd4eVq7vFcf481sQG6qxPiIi9Z0CkNR/OYfhi7+evCHpiSI7QMJlkHApxPYBN49Ki2UXlvL6op38Z9k+Sm0GblYLN/SO4/5BCYRpWruISIOhACT124E1MOcGyEsxX3sFQWATCIiGgBiI7WW29ATHnvYyNrvBJ6uSeeGH7WTmlwAwsG0Ej1zRngRtVioi0uAoAEn99cccmHuvuX5PRHu4/iMIa1Xty6zal8njczez+ZC5mGHrSH8eu7IDA9pE1HSNRUSkjlAAkvrHboMfp8Hvr5uv214BV8809+iqhvTcYp76dotjSnuAtzsPDGrDTX3j8dCGpSIiDZoCkNQveenwxa2w92fz9YWTYOCUKi1eeIxhGMz94xDT5m4mq6AUiwVG947jwUvbaJyPiEgjoQAk9cf+ZfD5OHOWl4cfDH8DOl1drUuk5RQx5atN/LjVnN3VoUkgz1/TRVPaRUQaGQUgqfsMA5a9AQungWGD8LYw6kOIaFvlS9jsBp+tTuaZ+VvJKSrDw83CvRcncMfAVuruEhFphBSApG47vAF+espcxBCg0zUw7FXw8q/yJX7dmcE/5m1hW0ouAF2aBfH8NV1oFx1YGzUWEZF6QAFI6qbklfDzC7Dze/O11QOGTIdefwVL1bae2JWWyzPzt/HTtjQAAr3dufeSBG7p1xx3tfqIiDRqCkBStxzZDd/ef3yQs8UKHa+GCydCZPsqXcJuN5jx825e/GEHNruBu9XCTX3juffiBEL8PGuv7iIiUm8oAEndUZwHs6+DI7vA6g5dr4fzJ1RrbZ/M/BIe/HQ9i7enA+Zu7VOuaE/LiKp3mYmISMOnACR1x/yJZvgJiIFbv4OQ5tU6ffW+TO75eB2Hs4vwcrfyxFUdGdUrFksVu8xERKTxUACSumH9bPjjY7PL65r3qhV+bHaDd07o8moR7sdbY7rTvokGOYuISOUUgMT10nfAvAfN5wMnQ3y/Kp+6Ky2XiZ9tYH1yFgDDusYw/erO+Hvpr7aIiJyafkuIa5UWmosblhZAiwvhggerdFqZzc7MX/bwyo87KSmzE+DlzmNXduDans3U5SUiImekACSu9f0jkLoJfMPh6nfB6nbGU/Zl5HPvnHVsOJANmLu2T7+6M02CfGq7tiIi0kAoAInrrJgJq98zn1/9DgREn/GU5XuOcMd/15BVUEqAtzvThnXkL92bqtVHRESqRQFIXGPj5/Dd383nFz0KrQed8ZRPVyfzyFcbKbUZdG0WxDs39SQ6yLuWKyoiIg2RApA4365F8NUdgAG9bzcXOTwNu93g+e+3M2PpbgCGdm7Ci9d1xdvjzN1lIiIilVEAEuc6sBo+uQnspdDpLzDkudNubVFQUsaET/5gweYUAO65uDUPDGqD1aouLxEROXsu3xDpzTffpHnz5nh7e9OnTx9Wrlx52vKfffYZ7dq1w9vbm86dOzN//vwK7+fl5XH33XfTrFkzfHx86NChAzNmzKjNryBVlb4DProWSvOh1cUwYgZYT/1XMDWniFHvLGfB5hQ83ay8PKorD17WVuFHRETOmUsD0CeffMKECROYNm0aa9eupWvXrgwePJi0tLRKy//++++MHj2a8ePHs27dOkaMGMGIESPYtGmTo8yECRNYsGAB//3vf9m6dSv3338/d999N3PnznXW15LKFGTC7GuhMBOa9oDrPgT3U+/LtelgNsPf+I2NB7MJ9fPko9v6MLJbMydWWEREGjKLYRiGqz68T58+9OrVizfeeAMAu91ObGws99xzDw8//PBJ5UeNGkV+fj7ffvut49h5551HYmKio5WnU6dOjBo1iscee8xRpkePHlx++eX84x//qLQexcXFFBcXO17n5OQQGxtLdnY2gYFaTfic2crgo2tgz2IIjoPbFoNf+CmLL9ySyn1z1lFQYqNVhB//vqU3cWG+TqywiIjURzk5OQQFBVXp97fLWoBKSkpYs2YNgwYdn/1jtVoZNGgQy5Ytq/ScZcuWVSgPMHjw4Arl+/Xrx9y5czl48CCGYbB48WJ27NjBZZdddsq6TJ8+naCgIMcjNjb2HL9dI1SUY3ZxVZanf5xmhh8PX7j+49OGn1nL9nH7h6spKLFxfutwvryzv8KPiIjUOJcFoIyMDGw2G1FRURWOR0VFkZKSUuk5KSkpZyz/+uuv06FDB5o1a4anpydDhgzhzTff5MILLzxlXSZPnkx2drbjkZycfA7frBEyDJg9Ct7sBbOugkPrjr+34VNYZrbwMeItiO50ysu8uXgXU/+3GcOAG/rE8e9xvQjy8ajlyouISGPU4GaBvf766yxfvpy5c+cSHx/Pzz//zF133UVMTMxJrUfHeHl54eXl5eSaNiDJKyDpd/P53p9h5kDoeDV0HAlz7zGPX/Cg+boShmFOc397iTnN/d5LEnhgUIIWNxQRkVrjsgAUHh6Om5sbqampFY6npqYSHV35isDR0dGnLV9YWMiUKVP46quvGDp0KABdunRh/fr1vPDCC6cMQHKOlr9l/mw/zOzm2vApbP7SfAAkDDYXO6yE3W7wxDeb+c+y/QBMuaIdt1/Yyhm1FhGRRsxlXWCenp706NGDRYsWOY7Z7XYWLVpE3759Kz2nb9++FcoDLFy40FG+tLSU0tJSrH+aWu3m5obdbq/hbyAAZCXD1vJB6QMnw9Uz4W8/Q6tLzGPhbeAv71Y63d1mN3joiw38Z9l+LBb4x4hOCj8iIuIULu0CmzBhAmPHjqVnz5707t2bV155hfz8fMaNGwfAzTffTNOmTZk+fToA9913HwMGDODFF19k6NChzJkzh9WrVzNz5kwAAgMDGTBgAJMmTcLHx4f4+HiWLl3KrFmzeOmll1z2PRu0Ve+CYTN3co/qaB5r0gVu+hLStkFQU/AKOOm0MpudBz/7g/+tP4Sb1cIL13bRNHcREXEalwagUaNGkZ6eztSpU0lJSSExMZEFCxY4BjonJSVVaM3p168fs2fP5tFHH2XKlCkkJCTw9ddf06nT8YG1c+bMYfLkyYwZM4bMzEzi4+N5+umnueOOO5z+/Rq8knxY84H5/Lw7T34/sl2lp5Xa7Nz/yXrmbTiMu9XC66O7cXnnJrVXTxERkT9x6TpAdVV11hFo1Fa9B/MmQEgLuGftaVd1PqakzM69H69jweYUPNwsvHlDdy7reOZd4EVERM6kOr+/G9wsMHESux1WlG8x0ueOKoWf4jIbd320jh+3puLpZuXtG7tzSfuoM54nIiJS0xSA5Ozs+QkydoBnACTecMbixWU2/u+/a/lpWxqe7lZm3tSDgW0jnVBRERGRkykAydlZXt760+1G8D59M2NRqY07/ruGJdvT8XK38q+xPbkgIcIJlRQREamcApBUX+pm2LUQsECf209btKjUxu0fruHnHel4e1h5f2wv+rU+9VYYIiIizqAAJNVTeBQ+vdl83m4ohLY8ZdGiUhu3zVrNLzsz8PFw4/1betG3VZiTKioiInJqCkBSdbYy+PxWOLILApvBlS+fsuiJ4cfX0ww/57VU+BERkbpBAUiq7odHYPdP5nYXoz8G/8oHMdvtBg9+9ocj/Hwwrje9W4Q6ubIiIiKn5rKtMKSeWf3v49PeR75jrvZ8Cs8t2Ma8DYfxcLPwr7E9FX5ERKTOUQCSM9v7C8yfaD6/+FHocNUpi85ato93ft4DwD+v6Uq/VhrwLCIidY8CkJyaYcC6/8JH14K9DDpdAxdMPGXxhVtSeXzuZgAmDW7LiG5NnVVTERGRatEYIKlccR7MexA2zDFftx4Ew98Ai6XS4n8kZ3HPx2uxGzC6dyx3DtSu7iIiUncpAMnJUjfDZ7eYKz1brGa3V/8HTrndxf4j+dz6wSqKSu0MaBPBU8M7YTlFUBIREakLFICkoqTlMGs4lBVBQAxc8x7E9ztl8SN5xYx9fyVH8kvoGBPIm2O64+6mnlUREanbFICkosXPmOGnxYVwzb/B79SDmAtKyrj1P6vZd6SAZiE+/HtcL/y99FdKRETqPv2vuhyXthX2LjW7vYa/edrwU2azc8/sdfyRnEWwrwf/ubU3kQHeTqysiIjI2VMAkuNWzjR/tr0CguNOWcwwDB773yYWbUvDy93Ke2N70irC30mVFBEROXcKQGIqzII/ymd89fnbaYt+tCKJj1cmY7XAa6O70SNeCx2KiEj9ogAkpnX/hdICiOwAzS84ZbEDRwuYPn8rAA9f3o7BHaOdVUMREZEaowAkYLfBqnfN571vP+VaP4Zh8PAXG8kvsdEzPoS/nn/qneBFRETqMgUggZ0/wNF94B0MXa47ZbE5q5L5dVcGXu5Wnr+mC1ar1voREZH6SQFIYMU75s/uN4GnX6VFDmYV8vQ8s+tr4mVtaalBzyIiUo8pADV26dthz2Jz6nuv2yotYhgGk7/cSF5xGd3igrn1/BZOrqSIiEjNUgBq7I5NfW9zOYTEV1rks9UH+HlHOp7uVv55TVfc1PUlIiL1nAJQY5a6BdbPNp+fYur7waxCnvp2CwATLm1D60h1fYmISP2nANRY5R+Bj0eZU99bDDC3vvgTu93g75//QW5xGYmxwfxVXV8iItJAKAA1RmUl8OnNkJUEIS3g2g8qnfr+4fL9/LbrCN4eVl66rqs2ORURkQZDv9EaG8OA+RNh/6/gGQCj54DvySs570nPY/p35QseDmmnWV8iItKgKAA1Nitnwtr/ABa45n2IbHdSkTKbnQc/+4OiUjv9W4dxc9/mTq+miIhIbVIAakz2/gwLJpvPL30C2lxWabF3ft7DuqQsArzc+ec1XbXgoYiINDgKQI1FaSHMvQcMG3S5HvrdW2mxrYdzeOXHHQBMu6ojMcE+zqyliIiIUygANRa/vmxudxEQA0NfrHTQs91uLnhYajO4tEMUf+ne1Pn1FBERcQIFoMbgyG749RXz+ZDp4FX5gOaPVyWxPjkLfy93nhreCcspNkUVERGp7846AO3atYvvv/+ewsJCwNwuQeogw4D5k8BWDK0uhg7DKy2WkVfMc99tA+DBy9oQHeTtzFqKiIg4VbUD0JEjRxg0aBBt2rThiiuu4PDhwwCMHz+eBx98sMYrKOdoy/9g9yJw84QrXqi06wvgmXlbySkqo2NMIDedV/mWGCIiIg1FtQPQAw88gLu7O0lJSfj6+jqOjxo1igULFtRo5eQcFecen/V1/gMQ1qrSYr/vzuDLdQexWODpkZ214KGIiDR47tU94YcffuD777+nWbNmFY4nJCSwf//+GquY1IClz0HuIQiONwNQJYrLbDz69SYAxvSJIzE22IkVFBERcY1q/69+fn5+hZafYzIzM/Hy8qqRSkkNyNgFy94yn1/xAnhUPp393Z/3sCc9n3B/TyYNPnlRRBERkYao2gHoggsuYNasWY7XFosFu93O888/z0UXXVSjlZNzsOQZc82fhMtOueBhak4Rr/+0C4BHh3YgyMfDmTUUERFxmWp3gT3//PNccsklrF69mpKSEv7+97+zefNmMjMz+e2332qjjlJdKRth0xfm80umnrLYGz/torjMTo/4EIYnxjipciIiIq5X7RagTp06sWPHDs4//3yGDx9Ofn4+V199NevWraNVq8oH2YqT/fQP82env0B050qLJGcWMGdVEgATL2urNX9ERKRRqXYLEEBQUBCPPPJITddFakLyStixACxuMHDKKYu9tmgnpTaD/q3D6NsqzIkVFBERcb1qB6Cff/75tO9feOGFZ10ZOUeGAYueNJ8n3gDhrSsttic9jy/WHgDgwcvaOqt2IiIidUa1A9DAgQNPOnZi94nNZjunCsk52LME9v1iLno44KFTFnvlx53YDbikXSTd40KcVz8REZE6otpjgI4ePVrhkZaWxoIFC+jVqxc//PBDbdRRquLE1p+e4yE4ttJi21Jy+GbDIQAmXNbGWbUTERGpU6rdAhQUFHTSsUsvvRRPT08mTJjAmjVraqRiUk3b5sGhteDhBxdMOGWxl37YgWHA0M5N6Bhz8p+liIhIY1Bjex5ERUWxffv2mrqcVNeqd82f590B/pGVFtlwIIsftqRitcADlyY4sXIiIiJ1S7VbgDZs2FDhtWEYHD58mGeffZbExMSaqpdUR2kRJC03n3cZdcpixxY9HJHYlNaRAc6omYiISJ1U7QCUmJiIxWLBMIwKx8877zzef//9GquYVMOBVVBWBP5REF75uJ496Xn8uDUVgDsvqnx2mIiISGNR7QC0d+/eCq+tVisRERF4e3vXWKWkmvaWL03Q4kI4xYKG7/26F8OAQe0jaR3p78TKiYiI1D3VDkDx8fG1UQ85FycGoEocySvm8zXmuj9/vaCls2olIiJSZ1UpAL322mtVvuC999571pWRs1CcBwdXm89PEYA+XL6f4jI7XZoF0adFqBMrJyIiUjdVKQC9/PLLVbqYxWJRAHK2pOVgL4PgOAhpftLbRaU2Ply2H4DbLmipPb9ERESoYgD687gfqUP2LjV/nqL158u1BzmSX0LTYB8u7xTtxIqJiIjUXTW2DpC4iGP8z4CT3rLbDf71yx4Abj2/Be5u+uMWERGBs9wN/sCBA8ydO5ekpCRKSkoqvPfSSy/VSMWkCgqPQkr5ukzNLzjp7Z+2pbEnI58Ab3dG9ap8awwREZHGqNoBaNGiRVx11VW0bNmSbdu20alTJ/bt24dhGHTv3r026iinsv93MOwQlgCBTU56e2Z5688NfeLw9zqrrCsiItIgVbtPZPLkyUycOJGNGzfi7e3NF198QXJyMgMGDODaa6+tjTrKqZxm+vuWQzms3JuJu9XCLf2aO7deIiIidVy1A9DWrVu5+eabAXB3d6ewsBB/f3+efPJJnnvuuRqvoJzGaQLQJ6uSABjcKZomQT7OrJWIiEidV+0A5Ofn5xj306RJE3bv3u14LyMjo+ZqJqeXlwZpW8znfxr/U1Rq46t1BwEY1VNjf0RERP6s2gNDzjvvPH799Vfat2/PFVdcwYMPPsjGjRv58ssvOe+882qjjlKZfb+YP6M6g19YhbcWbEohp6iMpsE+nN863AWVExERqduqHYBeeukl8vLyAHjiiSfIy8vjk08+ISEhQTPAnOk03V9zyru/rusZi9WqhQ9FRET+rNoB6JlnnuHGG28EzO6wGTNm1HilpApOEYD2ZuSzfE8mFgtc27OZCyomIiJS91V7DFB6ejpDhgwhNjaWSZMm8ccff9RGveR0sg9A5h6wuEF8vwpvfbo6GYABbSKICdbgZxERkcpUOwD973//4/Dhwzz22GOsWrWK7t2707FjR5555hn27dtXC1WUk+z+yfwZ0w28Ax2Hy2x2x67v12vhQxERkVM6q70RQkJCuP3221myZAn79+/nlltu4cMPP6R169Y1XT+pzK4fzZ+tB1U4vHh7Oum5xYT5eXJxuygXVExERKR+OKfNoUpLS1m9ejUrVqxg3759REXpl26ts5XB7iXm84RLK7x1bO2fv/Rohqe79v0SERE5lbP6Lbl48WJuu+02oqKiuOWWWwgMDOTbb7/lwIED1b7Wm2++SfPmzfH29qZPnz6sXLnytOU/++wz2rVrh7e3N507d2b+/Pknldm6dStXXXUVQUFB+Pn50atXL5KSkqpdtzrp4GoozgafELMLrFxKdhE/bUsDzNlfIiIicmrVDkBNmzbliiuuICMjg5kzZ5Kamsr777/PJZdcgsVSvSnXn3zyCRMmTGDatGmsXbuWrl27MnjwYNLS0iot//vvvzN69GjGjx/PunXrGDFiBCNGjGDTpk2OMrt37+b888+nXbt2LFmyhA0bNvDYY4/h7e1d3a9aNx3r/mp1MVjdHIe/WHsAuwG9mofQOtLfRZUTERGpHyyGYRjVOeHdd9/l2muvJTg4+Jw/vE+fPvTq1Ys33ngDALvdTmxsLPfccw8PP/zwSeVHjRpFfn4+3377rePYeeedR2JiomM6/vXXX4+HhwcffvhhletRXFxMcXGx43VOTg6xsbFkZ2cTGBh4mjNdYOZAOLQORrwNiTc4Dl/+6i9sPZzD83/pwnUaAC0iIo1QTk4OQUFBVfr9Xe0WoNtuu61Gwk9JSQlr1qxh0KDjA3mtViuDBg1i2bJllZ6zbNmyCuUBBg8e7Chvt9uZN28ebdq0YfDgwURGRtKnTx++/vrr09Zl+vTpBAUFOR6xsXU0QOSlm+EHzBagcoeyCtl6OAeLBQZ10DgsERGRM3HZSNmMjAxsNttJA6ejoqJISUmp9JyUlJTTlk9LSyMvL49nn32WIUOG8MMPPzBy5Eiuvvpqli5desq6TJ48mezsbMcjOTn5HL9dLdmz2PwZ3RkCoh2Hj4396R4XQqifpytqJiIiUq9UeyXousxutwMwfPhwHnjgAQASExP5/fffmTFjBgMGDKj0PC8vL7y8vJxWz7N2qunv5QHo4naRzq6RiIhIveSyFqDw8HDc3NxITU2tcDw1NZXo6OhKz4mOjj5t+fDwcNzd3enQoUOFMu3bt6//s8Dsdti1yHx+QgAqLLHx664MAC5prwAkIiJSFS4LQJ6envTo0YNFixY5jtntdhYtWkTfvn0rPadv374VygMsXLjQUd7T05NevXqxffv2CmV27NhBfHx8DX8DJ0v5AwoywDMAmvV2HF62J4PiMjtNg31oGxXgwgqKiIjUHy7tApswYQJjx46lZ8+e9O7dm1deeYX8/HzGjRsHwM0330zTpk2ZPn06APfddx8DBgzgxRdfZOjQocyZM4fVq1czc+ZMxzUnTZrEqFGjuPDCC7noootYsGAB33zzDUuWLHHFV6w5x7q/Wg4A9+PjfBZtPd79Vd1lCERERBorlwagUaNGkZ6eztSpU0lJSSExMZEFCxY4BjonJSVhtR5vpOrXrx+zZ8/m0UcfZcqUKSQkJPD111/TqVMnR5mRI0cyY8YMpk+fzr333kvbtm354osvOP/8853+/WqUo/vrEschwzAcA6AvVveXiIhIlVV7HaDGoDrrCDhFYRY83xIMG9y3AULM7rwth3K44rVf8Pawsn7qZXh7uJ3+OiIiIg1Yra4DJC6wZ4kZfsLbOMIPwE/bzAHh57cOV/gRERGpBgWg+uAU098XOaa/a/FDERGR6lAAqg+Slps/W17kOHQkr5j1yVmA1v8RERGpLgWguq6sBDL3mM+jOjoOL9mejmFAx5hAooMayEavIiIiTqIAVNdl7jbH/3gGQGCM4/Cx2V+XqPVHRESk2hSA6rr08kUdwxOgfJ2fkjI7P+9IB+Di9hr/IyIiUl0KQHVdxk7zZ0Rbx6F1SUfJLS4j3N+TLk2DXFQxERGR+ksBqK7LONYC1MZxaG1SFgC9W4RitWr1ZxERkepSAKrrjnWB/akFCKBbbIgraiQiIlLvKQDVZXb78S6w8hYgwzBYVz79vVtcsGvqJSIiUs8pANVl2clQVghWDwhpAcDBrELSc4txt1ropPE/IiIiZ0UBqC7L2GH+DGsFbua+tevKx/90iAnU9hciIiJnSQGoLks/eQD0sQDULTbY+fURERFpIBSA6rKMSgZAJ5cPgI7TAGgREZGzpQBUlzkGQJsBqLjMxuaDOYAGQIuIiJwLBaC6zDEF3uwC23IohxKbnVA/T+JCfV1YMRERkfpNAaiuys+AwkzAAmEJQMXxPxaLFkAUERE5WwpAddWx1p/gWPA0W3u0/o+IiEjNUACqqxxbYJy8AnSiVoAWERE5JwpAdVV6+RpA5TPA0nKLOHC0EIsFusRqAUQREZFzoQBUVzlagMzxP+vLx/8kRPoT6O3hokqJiIg0DApAddWxFqDyLjDH+B91f4mIiJwzBaC6qDgPcg6Yz8u7wBw7wGsAtIiIyDlTAKqLjpQvgOgbDr6hlNnsbDiQDWgFaBERkZqgAFQX/WkA9I7UPApKbPh7udM60t+FFRMREWkYFIDqooyKm6Ae2/+ra2wQblYtgCgiInKuFIDqovSKm6Cud6wAre4vERGRmqAAVBdlHJsBZrYArS+fAZYYG+ya+oiIiDQwCkB1ja0UMveYzyPaUlhiY3d6HgCdm2kBRBERkZqgAFTXZO4Bexl4+EFgU7am5GA3INzfk8gAL1fXTkREpEFQAKprHN1fCWCxsPlQDgAdYoK0A7yIiEgNUQCqa451f4W1AmDLIXP9n04xga6qkYiISIOjAFTXHN1v/gxpAeBoAeoYo/E/IiIiNUUBqK45us/8GdKcUpudbSm5AHRUC5CIiEiNUQCqa04IQLvT8ygps+Pv5U5cqK9LqyUiItKQKADVJXYbZCWZz0Oas+lg+QDoJoFYtQK0iIhIjVEAqktyDoG9FKweEBjD5vIB0B2bqvtLRESkJikA1SXHur+CY8HqpgHQIiIitUQBqC45YfyP3W6w1RGA1AIkIiJSkxSA6pITAlDy0QJyi8vwdLfSOtLfpdUSERFpaBSA6pKsY2sAHR8A3TYqAA83/TGJiIjUJP1mrUtOaAE6NgC6kwZAi4iI1DgFoLqkQgA6vgeYiIiI1CwFoLqiOA/y083nJwQgDYAWERGpeQpAdcWx8T8+IaSVeJGRV4zVAu2jFYBERERqmgJQXVFJ91erCH98PN1cVycREZEGSgGornAsghjPpoPlK0Cr+0tERKRWKADVFZW0AGkFaBERkdqhAFRXnBiADqsFSEREpDYpANUVR81B0Hl+sSRnFgLQQQFIRESkVigA1QV2u2MW2I7iMACahfgQ7OvpylqJiIg0WApAdUFeKpQVgcWNLQUBALSLDnBxpURERBouBaC64Nj4n6BmHMguA6BZiK/r6iMiItLAKQDVBScMgD6YZY7/aRrs47r6iIiINHAKQHWBIwDFc/BoAQBNQxSAREREaosCUF2gFiARERGnUgCqC8oDUGlgHKk5xYA5C0xERERqhwJQXVA+BT7dPQYAbw8roX6aAi8iIlJbFIBcrbQQcg8DkEwkYHZ/WSwWV9ZKRESkQVMAcrWsJPOnVyD7870AaKop8CIiIrVKAcjVTpgBdiC7CNAAaBERkdqmAORqJ84AO2rOANMAaBERkdqlAORqxwJQcDwHs8rXAFILkIiISK1SAHK1StYAUguQiIhI7VIAcrXyAGQLbs7hrPIxQApAIiIitUoByNWyDwJwxD2SMruBu9VCZIC3iyslIiLSsNWJAPTmm2/SvHlzvL296dOnDytXrjxt+c8++4x27drh7e1N586dmT9//inL3nHHHVgsFl555ZUarnUNsJVCcTYAB0v8AGgS7I2bVWsAiYiI1CaXB6BPPvmECRMmMG3aNNauXUvXrl0ZPHgwaWlplZb//fffGT16NOPHj2fdunWMGDGCESNGsGnTppPKfvXVVyxfvpyYmJja/hpnpyCz/ImF/QXmys8aAC0iIlL7XB6AXnrpJW677TbGjRtHhw4dmDFjBr6+vrz//vuVln/11VcZMmQIkyZNon379jz11FN0796dN954o0K5gwcPcs899/DRRx/h4eHhjK9SfYXlAcgnhIPZJQA0DdYiiCIiIrXNpQGopKSENWvWMGjQIMcxq9XKoEGDWLZsWaXnLFu2rEJ5gMGDB1cob7fbuemmm5g0aRIdO3Y8Yz2Ki4vJycmp8HCKgiPmT98wDpSvAaQB0CIiIrXPpQEoIyMDm81GVFRUheNRUVGkpKRUek5KSsoZyz/33HO4u7tz7733Vqke06dPJygoyPGIjY2t5jc5S44AFKop8CIiIk7k8i6wmrZmzRpeffVVPvjggypvKDp58mSys7Mdj+Tk5FquZbkTWoAOHjUXQWymMUAiIiK1zqUBKDw8HDc3N1JTUyscT01NJTo6utJzoqOjT1v+l19+IS0tjbi4ONzd3XF3d2f//v08+OCDNG/evNJrenl5ERgYWOHhFOWDoA2f4y1A6gITERGpfS4NQJ6envTo0YNFixY5jtntdhYtWkTfvn0rPadv374VygMsXLjQUf6mm25iw4YNrF+/3vGIiYlh0qRJfP/997X3Zc5GeQAq9AimqNSOxQJNghSAREREapu7qyswYcIExo4dS8+ePenduzevvPIK+fn5jBs3DoCbb76Zpk2bMn36dADuu+8+BgwYwIsvvsjQoUOZM2cOq1evZubMmQCEhYURFhZW4TM8PDyIjo6mbdu2zv1yZ1LeBZZFAACRAV54uje4XkkREZE6x+UBaNSoUaSnpzN16lRSUlJITExkwYIFjoHOSUlJWK3HQ0G/fv2YPXs2jz76KFOmTCEhIYGvv/6aTp06ueornL3yAJRh9we0BpCIiIizWAzDMFxdibomJyeHoKAgsrOza3c80LsXw8E1fN/5Zf62KophXWN4fXS32vs8ERGRBqw6v7/V3+JK5S1AB4rNxQ81BV5ERMQ5FIBcqXwQ9J58L0BdYCIiIs6iAOQqZSVQbK44vSPX3KpDU+BFREScQwHIVQqPmj8tVnbkuAFaBFFERMRZFIBcpXz8j907hOwiO6AWIBEREWdRAHKV8gBU6hUMQIivB76eLl+VQEREpFFQAHKV8gBU6B4EqPVHRETEmRSAXKXQnAGWYykPQBr/IyIi4jQKQK5S3gKUibkKdLMQX1fWRkREpFFRAHKV8jWA0sq0DYaIiIizKQC5SnkL0KFiM/hoDJCIiIjzKAC5SnkA2l9UHoDUAiQiIuI0CkCuUt4FllzsDUBkoJcrayMiItKoKAC5yrFB0PYALBYI9fV0cYVEREQaDwUgVylvATpKAKG+nri76Y9CRETEWfRb1xXKSqAkF4BMI4CIAHV/iYiIOJMCkCuUL4JoYCUHX8L9FYBEREScSQHIFcrH/xR5BGFgJdxf439EREScSQHIFcoDUL5bIIBagERERJxMAcgVygNQjqU8AGkMkIiIiFMpALlC+QywTAIAtQCJiIg4mwKQK5QHoHSbuQ+YZoGJiIg4lwKQK5R3gaWWmjvAaxC0iIiIcykAucKxjVDLA1CEusBEREScSgHIFcrXAco0yrfB8FMLkIiIiDMpALnCsX3AjABCtA2GiIiI0+k3ryuUB6CjRoC6v0RERFxAAcgVHBuh+hMeoO4vERERZ1MAcrayYijJA8wuMK0BJCIi4nwKQM5W3vpjx41cbYQqIiLiEgpAznbCPmDmRqgKQCIiIs6mAORsjn3AzG0wtAq0iIiI8ykAOduxKfCOfcA0CFpERMTZFICcrXwRxIzyfcDUBSYiIuJ8CkDOVj4IOrXMD1AXmIiIiCsoADmbYxVoswVI22CIiIg4nwKQs1XYBsMDD22DISIi4nT67etsx1aBNgLU/SUiIuIiCkDOdsIsMA2AFhERcQ0FIGcrbwHKMvwVgERERFxEAcjZ1AIkIiLicgpAzlRaBKX5gDkGSDvBi4iIuIYCkDOVL4Jow0oOvkSoBUhERMQlFICcybEPWCBgIVyzwERERFxCAciZygPQ0fJFENUCJCIi4hoKQM5UPgMs3a59wERERFxJAciZTlgFGiBMO8GLiIi4hAKQM52wCnSwtsEQERFxGXdXV6BR6XwNW+3N+GRhhsb/iIiIuJACkDOFtWJHqDd/GOvpqwAkIiLiMuqDcbL03GIATYEXERFxIQUgJ8vIKwEgXAOgRUREXEYByMky8spbgNQFJiIi4jIKQE52rAssQl1gIiIiLqMA5GTHWoA0C0xERMR1FICcTF1gIiIirqcA5ER2u8GRY4OgAzQIWkRExFUUgJwou7CUMrsBQJifWoBERERcRQHIidLLu7+CfDzwdNetFxERcRX9FnaiDM0AExERqRMUgJwo3TEAWuN/REREXEkByImOrwKtFiARERFXUgByopIyO94eVgUgERERF7MYhmG4uhJ1TU5ODkFBQWRnZxMYGFij1zYMgzK7gYebsqeIiEhNqs7vb3cn1UnKWSwWPNwsrq6GiIhIo6ZmCBEREWl06kQAevPNN2nevDne3t706dOHlStXnrb8Z599Rrt27fD29qZz587Mnz/f8V5paSkPPfQQnTt3xs/Pj5iYGG6++WYOHTpU219DRERE6gmXB6BPPvmECRMmMG3aNNauXUvXrl0ZPHgwaWlplZb//fffGT16NOPHj2fdunWMGDGCESNGsGnTJgAKCgpYu3Ytjz32GGvXruXLL79k+/btXHXVVc78WiIiIlKHuXwQdJ8+fejVqxdvvPEGAHa7ndjYWO655x4efvjhk8qPGjWK/Px8vv32W8ex8847j8TERGbMmFHpZ6xatYrevXuzf/9+4uLizlin2hwELSIiIrWjOr+/XdoCVFJSwpo1axg0aJDjmNVqZdCgQSxbtqzSc5YtW1ahPMDgwYNPWR4gOzsbi8VCcHBwpe8XFxeTk5NT4SEiIiINl0sDUEZGBjabjaioqArHo6KiSElJqfSclJSUapUvKirioYceYvTo0adMg9OnTycoKMjxiI2NPYtvIyIiIvWFy8cA1abS0lKuu+46DMPg7bffPmW5yZMnk52d7XgkJyc7sZYiIiLibC5dByg8PBw3NzdSU1MrHE9NTSU6OrrSc6Kjo6tU/lj42b9/Pz/99NNp+wK9vLzw8tLqzCIiIo2FS1uAPD096dGjB4sWLXIcs9vtLFq0iL59+1Z6Tt++fSuUB1i4cGGF8sfCz86dO/nxxx8JCwurnS8gIiIi9ZLLV4KeMGECY8eOpWfPnvTu3ZtXXnmF/Px8xo0bB8DNN99M06ZNmT59OgD33XcfAwYM4MUXX2To0KHMmTOH1atXM3PmTMAMP9dccw1r167l22+/xWazOcYHhYaG4umpndhFREQaO5cHoFGjRpGens7UqVNJSUkhMTGRBQsWOAY6JyUlYbUeb6jq168fs2fP5tFHH2XKlCkkJCTw9ddf06lTJwAOHjzI3LlzAUhMTKzwWYsXL2bgwIFO+V4iIiJSd7l8HaC6SOsAiYiI1D/1Zh0gEREREVdweRdYXXSsUUwLIoqIiNQfx35vV6VzSwGoErm5uQBaEFFERKQeys3NJSgo6LRlNAaoEna7nUOHDhEQEIDFYqnRa+fk5BAbG0tycrLGF9Uy3Wvn0b12Ht1r59G9dp6auteGYZCbm0tMTEyFCVSVUQtQJaxWK82aNavVzwgMDNR/UE6ie+08utfOo3vtPLrXzlMT9/pMLT/HaBC0iIiINDoKQCIiItLoKAA5mZeXF9OmTdPeY06ge+08utfOo3vtPLrXzuOKe61B0CIiItLoqAVIREREGh0FIBEREWl0FIBERESk0VEAEhERkUZHAciJ3nzzTZo3b463tzd9+vRh5cqVrq5SvTd9+nR69epFQEAAkZGRjBgxgu3bt1coU1RUxF133UVYWBj+/v785S9/ITU11UU1bjieffZZLBYL999/v+OY7nXNOXjwIDfeeCNhYWH4+PjQuXNnVq9e7XjfMAymTp1KkyZN8PHxYdCgQezcudOFNa6fbDYbjz32GC1atMDHx4dWrVrx1FNPVdhLSvf67Pz8888MGzaMmJgYLBYLX3/9dYX3q3JfMzMzGTNmDIGBgQQHBzN+/Hjy8vJqpH4KQE7yySefMGHCBKZNm8batWvp2rUrgwcPJi0tzdVVq9eWLl3KXXfdxfLly1m4cCGlpaVcdtll5OfnO8o88MADfPPNN3z22WcsXbqUQ4cOcfXVV7uw1vXfqlWreOedd+jSpUuF47rXNePo0aP0798fDw8PvvvuO7Zs2cKLL75ISEiIo8zzzz/Pa6+9xowZM1ixYgV+fn4MHjyYoqIiF9a8/nnuued4++23eeONN9i6dSvPPfcczz//PK+//rqjjO712cnPz6dr1668+eablb5flfs6ZswYNm/ezMKFC/n222/5+eefuf3222umgoY4Re/evY277rrL8dpmsxkxMTHG9OnTXVirhictLc0AjKVLlxqGYRhZWVmGh4eH8dlnnznKbN261QCMZcuWuaqa9Vpubq6RkJBgLFy40BgwYIBx3333GYahe12THnroIeP8888/5ft2u92Ijo42/vnPfzqOZWVlGV5eXsbHH3/sjCo2GEOHDjVuvfXWCseuvvpqY8yYMYZh6F7XFMD46quvHK+rcl+3bNliAMaqVascZb777jvDYrEYBw8ePOc6qQXICUpKSlizZg2DBg1yHLNarQwaNIhly5a5sGYNT3Z2NgChoaEArFmzhtLS0gr3vl27dsTFxenen6W77rqLoUOHVrinoHtdk+bOnUvPnj259tpriYyMpFu3brz77ruO9/fu3UtKSkqFex0UFESfPn10r6upX79+LFq0iB07dgDwxx9/8Ouvv3L55ZcDute1pSr3ddmyZQQHB9OzZ09HmUGDBmG1WlmxYsU510GboTpBRkYGNpuNqKioCsejoqLYtm2bi2rV8Njtdu6//3769+9Pp06dAEhJScHT05Pg4OAKZaOiokhJSXFBLeu3OXPmsHbtWlatWnXSe7rXNWfPnj28/fbbTJgwgSlTprBq1SruvfdePD09GTt2rON+VvZviu519Tz88MPk5OTQrl073NzcsNlsPP3004wZMwZA97qWVOW+pqSkEBkZWeF9d3d3QkNDa+TeKwBJg3HXXXexadMmfv31V1dXpUFKTk7mvvvuY+HChXh7e7u6Og2a3W6nZ8+ePPPMMwB069aNTZs2MWPGDMaOHevi2jUsn376KR999BGzZ8+mY8eOrF+/nvvvv5+YmBjd6wZOXWBOEB4ejpub20mzYVJTU4mOjnZRrRqWu+++m2+//ZbFixfTrFkzx/Ho6GhKSkrIysqqUF73vvrWrFlDWloa3bt3x93dHXd3d5YuXcprr72Gu7s7UVFRutc1pEmTJnTo0KHCsfbt25OUlATguJ/6N+XcTZo0iYcffpjrr7+ezp07c9NNN/HAAw8wffp0QPe6tlTlvkZHR580UaisrIzMzMwaufcKQE7g6elJjx49WLRokeOY3W5n0aJF9O3b14U1q/8Mw+Duu+/mq6++4qeffqJFixYV3u/RowceHh4V7v327dtJSkrSva+mSy65hI0bN7J+/XrHo2fPnowZM8bxXPe6ZvTv3/+k5Rx27NhBfHw8AC1atCA6OrrCvc7JyWHFihW619VUUFCA1VrxV6Gbmxt2ux3Qva4tVbmvffv2JSsrizVr1jjK/PTTT9jtdvr06XPulTjnYdRSJXPmzDG8vLyMDz74wNiyZYtx++23G8HBwUZKSoqrq1av/d///Z8RFBRkLFmyxDh8+LDjUVBQ4Chzxx13GHFxccZPP/1krF692ujbt6/Rt29fF9a64ThxFphh6F7XlJUrVxru7u7G008/bezcudP46KOPDF9fX+O///2vo8yzzz5rBAcHG//73/+MDRs2GMOHDzdatGhhFBYWurDm9c/YsWONpk2bGt9++62xd+9e48svvzTCw8ONv//9744yutdnJzc311i3bp2xbt06AzBeeuklY926dcb+/fsNw6jafR0yZIjRrVs3Y8WKFcavv/5qJCQkGKNHj66R+ikAOdHrr79uxMXFGZ6enkbv3r2N5cuXu7pK9R5Q6ePf//63o0xhYaFx5513GiEhIYavr68xcuRI4/Dhw66rdAPy5wCke11zvvnmG6NTp06Gl5eX0a5dO2PmzJkV3rfb7cZjjz1mREVFGV5eXsYll1xibN++3UW1rb9ycnKM++67z4iLizO8vb2Nli1bGo888ohRXFzsKKN7fXYWL15c6b/PY8eONQyjavf1yJEjxujRow1/f38jMDDQGDdunJGbm1sj9bMYxgnLXYqIiIg0AhoDJCIiIo2OApCIiIg0OgpAIiIi0ugoAImIiEijowAkIiIijY4CkIiIiDQ6CkAiIiLS6CgAiYiISKOjACQiUgVLlizBYrGctNmriNRPCkAiIiLS6CgAiYiISKOjACQi9YLdbmf69Om0aNECHx8funbtyueffw4c756aN28eXbp0wdvbm/POO49NmzZVuMYXX3xBx44d8fLyonnz5rz44osV3i8uLuahhx4iNjYWLy8vWrduzXvvvVehzJo1a+jZsye+vr7069eP7du31+4XF5FaoQAkIvXC9OnTmTVrFjNmzGDz5s088MAD3HjjjSxdutRRZtKkSbz44ousWrWKiIgIhg0bRmlpKWAGl+uuu47rr7+ejRs38vjjj/PYY4/xwQcfOM6/+eab+fjjj3nttdfYunUr77zzDv7+/hXq8cgjj/Diiy+yevVq3N3dufXWW53y/UWkZmk3eBGp84qLiwkNDeXHH3+kb9++juN//etfKSgo4Pbbb+eiiy5izpw5jBo1CoDMzEyaNWvGBx98wHXXXceYMWNIT0/nhx9+cJz/97//nXnz5rF582Z27NhB27ZtWbhwIYMGDTqpDkuWLOGiiy7ixx9/5JJLLgFg/vz5DB06lMLCQry9vWv5LohITVILkIjUebt27aKgoIBLL70Uf39/x2PWrFns3r3bUe7EcBQaGkrbtm3ZunUrAFu3bqV///4Vrtu/f3927tyJzWZj/fr1uLm5MWDAgNPWpUuXLo7nTZo0ASAtLe2cv6OIOJe7qysgInImeXl5AMybN4+mTZtWeM/Ly6tCCDpbPj4+VSrn4eHheG6xWABzfJKI1C9qARKROq9Dhw54eXmRlJRE69atKzxiY2Md5ZYvX+54fvToUXbs2EH79u0BaN++Pb/99luF6/7222+0adMGNzc3OnfujN1urzCmSEQaLrUAiUidFxAQwMSJE3nggQew2+2cf/75ZGdn89tvvxEYGEh8fDwATz75JGFhYURFRfHII48QHh7OiBEjAHjwwQfp1asXTz31FKNGjWLZsmW88cYbvPXWWwA0b96csWPHcuutt/Laa6/RtWtX9u/fT1paGtddd52rvrqI1BIFIBGpF5566ikiIiKYPn06e/bsITg4mO7duzNlyhRHF9Szzz7Lfffdx86dO0lMTOSbb77B09MTgO7du/Ppp58ydepUnnrqKZo0acKTTz7JLbfc4viMt99+mylTpnDnnXdy5MgR4uLimDJliiu+rojUMs0CE5F679gMraNHjxIcHOzq6ohIPaAxQCIiItLoKACJiIhIo6MuMBEREWl01AIkIiIijY4CkIiIiDQ6CkAiIiLS6CgAiYiISKOjACQiIiKNjgKQiIiINDoKQCIiItLoKACJiIhIo/P/vWzQft1Du9IAAAAASUVORK5CYII=\n"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# values are not correct\n",
    "sns.lineplot(df_macro, x='epoch', y='value', hue='variable')\n",
    "plt.show()"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}