DEMO/CNN-Binary-Example-DAVIS.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "os.chdir('../')\n",
    "\n",
    "import DeepPurpose.DTI as models\n",
    "from DeepPurpose.utils import *\n",
    "from DeepPurpose.dataset import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Beginning Processing...\n",
      "Beginning to extract zip file...\n",
      "Default binary threshold for the binding affinity scores are 30, you can adjust it by using the \"threshold\" parameter\n",
      "Done!\n",
      "in total: 30056 drug-target pairs\n",
      "encoding drug...\n",
      "unique drugs: 68\n",
      "drug encoding finished...\n",
      "encoding protein...\n",
      "unique target sequence: 379\n",
      "protein encoding finished...\n",
      "splitting dataset...\n",
      "Done.\n"
     ]
    }
   ],
   "source": [
    "X_drug, X_target, y = load_process_DAVIS('./data/', binary=True)\n",
    "\n",
    "drug_encoding = 'CNN'\n",
    "target_encoding = 'CNN'\n",
    "train, val, test = data_process(X_drug, X_target, y, \n",
    "                                drug_encoding, target_encoding, \n",
    "                                split_method='random',frac=[0.7,0.1,0.2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# use the parameters setting provided in the paper: https://arxiv.org/abs/1801.10193\n",
    "config = generate_config(drug_encoding = drug_encoding, \n",
    "                         target_encoding = target_encoding, \n",
    "                         cls_hidden_dims = [1024,1024,512], \n",
    "                         train_epoch = 100, \n",
    "                         LR = 0.001, \n",
    "                         batch_size = 256,\n",
    "                         cnn_drug_filters = [32,64,96],\n",
    "                         cnn_target_filters = [32,64,96],\n",
    "                         cnn_drug_kernels = [4,6,8],\n",
    "                         cnn_target_kernels = [4,8,12]\n",
    "                        )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Let's use 1 GPU/s!\n",
      "--- Data Preparation ---\n",
      "--- Go for Training ---\n",
      "Training at Epoch 1 iteration 0 with loss 0.69403857\n",
      "Validation at Epoch 1 , AUROC: 0.7636400901824144 , AUPRC: 0.16826621955981202 , F1: 0.0\n",
      "Training at Epoch 2 iteration 0 with loss 0.12913653\n",
      "Validation at Epoch 2 , AUROC: 0.8419476327116213 , AUPRC: 0.30344635054864355 , F1: 0.0\n",
      "Training at Epoch 3 iteration 0 with loss 0.21572796\n",
      "Validation at Epoch 3 , AUROC: 0.8387220741955318 , AUPRC: 0.30598321657052374 , F1: 0.0\n",
      "Training at Epoch 4 iteration 0 with loss 0.14563164\n",
      "Validation at Epoch 4 , AUROC: 0.8509146341463415 , AUPRC: 0.31637410466513377 , F1: 0.0547945205479452\n",
      "Training at Epoch 5 iteration 0 with loss 0.1452313\n",
      "Validation at Epoch 5 , AUROC: 0.8490648698503792 , AUPRC: 0.33100461183944246 , F1: 0.2962962962962963\n",
      "Training at Epoch 6 iteration 0 with loss 0.13702923\n",
      "Validation at Epoch 6 , AUROC: 0.85576706292273 , AUPRC: 0.3491745417930331 , F1: 0.2967032967032967\n",
      "Training at Epoch 7 iteration 0 with loss 0.079116836\n",
      "Validation at Epoch 7 , AUROC: 0.8826783152285305 , AUPRC: 0.3794391385685332 , F1: 0.3163265306122449\n",
      "Training at Epoch 8 iteration 0 with loss 0.17364198\n",
      "Validation at Epoch 8 , AUROC: 0.8666350686616111 , AUPRC: 0.37262355345134507 , F1: 0.31182795698924726\n",
      "Training at Epoch 9 iteration 0 with loss 0.16354859\n",
      "Validation at Epoch 9 , AUROC: 0.9049267267882763 , AUPRC: 0.4460224533672988 , F1: 0.47698744769874474\n",
      "Training at Epoch 10 iteration 0 with loss 0.09402539\n",
      "Validation at Epoch 10 , AUROC: 0.8950143472022956 , AUPRC: 0.43166481822117664 , F1: 0.44067796610169496\n",
      "Training at Epoch 11 iteration 0 with loss 0.10640566\n",
      "Validation at Epoch 11 , AUROC: 0.9071607911457267 , AUPRC: 0.49207847162612867 , F1: 0.38775510204081637\n",
      "Training at Epoch 12 iteration 0 with loss 0.09256137\n",
      "Validation at Epoch 12 , AUROC: 0.9132045501127279 , AUPRC: 0.48881340507971016 , F1: 0.33862433862433866\n",
      "Training at Epoch 13 iteration 0 with loss 0.07567432\n",
      "Validation at Epoch 13 , AUROC: 0.9162456445993032 , AUPRC: 0.5140960138404388 , F1: 0.4748858447488584\n",
      "Training at Epoch 14 iteration 0 with loss 0.06538307\n",
      "Validation at Epoch 14 , AUROC: 0.916312256609961 , AUPRC: 0.4987124076180094 , F1: 0.3979591836734694\n",
      "Training at Epoch 15 iteration 0 with loss 0.076862425\n",
      "Validation at Epoch 15 , AUROC: 0.9238368518138964 , AUPRC: 0.5282982579558476 , F1: 0.4672897196261683\n",
      "Training at Epoch 16 iteration 0 with loss 0.106477425\n",
      "Validation at Epoch 16 , AUROC: 0.9178520188563231 , AUPRC: 0.5317619377395247 , F1: 0.39999999999999997\n",
      "Training at Epoch 17 iteration 0 with loss 0.080990925\n",
      "Validation at Epoch 17 , AUROC: 0.9288096946095511 , AUPRC: 0.5338708848946752 , F1: 0.45933014354066987\n",
      "Training at Epoch 18 iteration 0 with loss 0.13484664\n",
      "Validation at Epoch 18 , AUROC: 0.9236447017831523 , AUPRC: 0.5216281810075476 , F1: 0.5263157894736842\n",
      "Training at Epoch 19 iteration 0 with loss 0.07682976\n",
      "Validation at Epoch 19 , AUROC: 0.9364623898339823 , AUPRC: 0.5711966451919173 , F1: 0.44791666666666663\n",
      "Training at Epoch 20 iteration 0 with loss 0.06302365\n",
      "Validation at Epoch 20 , AUROC: 0.9359602377536381 , AUPRC: 0.5596483573586336 , F1: 0.5087719298245614\n",
      "Training at Epoch 21 iteration 0 with loss 0.06367779\n",
      "Validation at Epoch 21 , AUROC: 0.9241007378561181 , AUPRC: 0.5572449781575782 , F1: 0.48076923076923084\n",
      "Training at Epoch 22 iteration 0 with loss 0.069616705\n",
      "Validation at Epoch 22 , AUROC: 0.9230964336954295 , AUPRC: 0.5303964386954687 , F1: 0.48888888888888893\n",
      "Training at Epoch 23 iteration 0 with loss 0.09091424\n",
      "Validation at Epoch 23 , AUROC: 0.9273262963722075 , AUPRC: 0.5361992178978408 , F1: 0.4867256637168142\n",
      "Training at Epoch 24 iteration 0 with loss 0.048069615\n",
      "Validation at Epoch 24 , AUROC: 0.9199887271981965 , AUPRC: 0.5360963003830524 , F1: 0.46363636363636357\n",
      "Training at Epoch 25 iteration 0 with loss 0.08967747\n",
      "Validation at Epoch 25 , AUROC: 0.9315202910432466 , AUPRC: 0.55585434098351 , F1: 0.4745762711864407\n",
      "Training at Epoch 26 iteration 0 with loss 0.045414694\n",
      "Validation at Epoch 26 , AUROC: 0.9277823324451733 , AUPRC: 0.5380648335374518 , F1: 0.49315068493150677\n",
      "Training at Epoch 27 iteration 0 with loss 0.059319347\n",
      "Validation at Epoch 27 , AUROC: 0.9327756712441075 , AUPRC: 0.525848213214875 , F1: 0.4147465437788019\n",
      "Training at Epoch 28 iteration 0 with loss 0.054195795\n",
      "Validation at Epoch 28 , AUROC: 0.9356579217052674 , AUPRC: 0.5401656688644135 , F1: 0.5294117647058822\n",
      "Training at Epoch 29 iteration 0 with loss 0.096503206\n",
      "Validation at Epoch 29 , AUROC: 0.9320583111293297 , AUPRC: 0.5206459171023787 , F1: 0.4933920704845815\n",
      "Training at Epoch 30 iteration 0 with loss 0.04534679\n",
      "Validation at Epoch 30 , AUROC: 0.9204857552777208 , AUPRC: 0.4990984922615908 , F1: 0.5\n",
      "Training at Epoch 31 iteration 0 with loss 0.04636871\n",
      "Validation at Epoch 31 , AUROC: 0.9302187948350071 , AUPRC: 0.5151041265370552 , F1: 0.4793388429752066\n",
      "Training at Epoch 32 iteration 0 with loss 0.046937253\n",
      "Validation at Epoch 32 , AUROC: 0.9305185488829678 , AUPRC: 0.5315189702185577 , F1: 0.4978902953586498\n",
      "Training at Epoch 33 iteration 0 with loss 0.06999016\n",
      "Validation at Epoch 33 , AUROC: 0.9331164172986267 , AUPRC: 0.5205500565069556 , F1: 0.47058823529411764\n",
      "Training at Epoch 34 iteration 0 with loss 0.032049872\n",
      "Validation at Epoch 34 , AUROC: 0.9327346792375487 , AUPRC: 0.5281469460884153 , F1: 0.5224489795918367\n",
      "Training at Epoch 35 iteration 0 with loss 0.049913596\n",
      "Validation at Epoch 35 , AUROC: 0.924715617954499 , AUPRC: 0.5469737284097115 , F1: 0.49773755656108604\n",
      "Training at Epoch 36 iteration 0 with loss 0.06840132\n",
      "Validation at Epoch 36 , AUROC: 0.9276491084238574 , AUPRC: 0.5206383568887594 , F1: 0.4813278008298755\n",
      "Training at Epoch 37 iteration 0 with loss 0.03628567\n",
      "Validation at Epoch 37 , AUROC: 0.9170962287353966 , AUPRC: 0.5521375721464997 , F1: 0.5158371040723982\n",
      "Training at Epoch 38 iteration 0 with loss 0.029465776\n",
      "Validation at Epoch 38 , AUROC: 0.92115699938512 , AUPRC: 0.5227006853495088 , F1: 0.5344827586206896\n",
      "Training at Epoch 39 iteration 0 with loss 0.042498406\n",
      "Validation at Epoch 39 , AUROC: 0.929588542734167 , AUPRC: 0.5105382258569237 , F1: 0.47104247104247104\n",
      "Training at Epoch 40 iteration 0 with loss 0.032960795\n",
      "Validation at Epoch 40 , AUROC: 0.9327628612420579 , AUPRC: 0.5268282351996987 , F1: 0.496\n",
      "Training at Epoch 41 iteration 0 with loss 0.06273056\n",
      "Validation at Epoch 41 , AUROC: 0.933813281410125 , AUPRC: 0.523471612202944 , F1: 0.475\n",
      "Training at Epoch 42 iteration 0 with loss 0.038919788\n",
      "Validation at Epoch 42 , AUROC: 0.9385145521623284 , AUPRC: 0.5631597066801645 , F1: 0.551440329218107\n",
      "Training at Epoch 43 iteration 0 with loss 0.032266103\n",
      "Validation at Epoch 43 , AUROC: 0.9306927649108423 , AUPRC: 0.5323701821678755 , F1: 0.5271317829457364\n",
      "Training at Epoch 44 iteration 0 with loss 0.04825941\n",
      "Validation at Epoch 44 , AUROC: 0.9363086698093872 , AUPRC: 0.5332425469389527 , F1: 0.47698744769874474\n",
      "Training at Epoch 45 iteration 0 with loss 0.056063652\n",
      "Validation at Epoch 45 , AUROC: 0.9367057798729248 , AUPRC: 0.48889719217828603 , F1: 0.44628099173553715\n",
      "Training at Epoch 46 iteration 0 with loss 0.037880063\n",
      "Validation at Epoch 46 , AUROC: 0.9302854068456652 , AUPRC: 0.5412396623748551 , F1: 0.5238095238095238\n",
      "Training at Epoch 47 iteration 0 with loss 0.037543602\n",
      "Validation at Epoch 47 , AUROC: 0.9247771059643369 , AUPRC: 0.5211528037106868 , F1: 0.4869565217391305\n",
      "Training at Epoch 48 iteration 0 with loss 0.027733017\n",
      "Validation at Epoch 48 , AUROC: 0.9335212133633943 , AUPRC: 0.5415796039312415 , F1: 0.5271317829457364\n",
      "Training at Epoch 49 iteration 0 with loss 0.031213483\n",
      "Validation at Epoch 49 , AUROC: 0.9399134043861448 , AUPRC: 0.5250739999291815 , F1: 0.5118110236220473\n",
      "Training at Epoch 50 iteration 0 with loss 0.025122339\n",
      "Validation at Epoch 50 , AUROC: 0.929227300676368 , AUPRC: 0.5093319224525819 , F1: 0.4505928853754941\n",
      "Training at Epoch 51 iteration 0 with loss 0.02059749\n",
      "Validation at Epoch 51 , AUROC: 0.9397340643574503 , AUPRC: 0.5248697987226367 , F1: 0.5227272727272727\n",
      "Training at Epoch 52 iteration 0 with loss 0.033719\n",
      "Validation at Epoch 52 , AUROC: 0.9380738880918221 , AUPRC: 0.545761885352334 , F1: 0.5413533834586466\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training at Epoch 53 iteration 0 with loss 0.03376828\n",
      "Validation at Epoch 53 , AUROC: 0.937210493953679 , AUPRC: 0.5418339983477827 , F1: 0.5217391304347825\n",
      "Training at Epoch 54 iteration 0 with loss 0.024219895\n",
      "Validation at Epoch 54 , AUROC: 0.9386298421807746 , AUPRC: 0.5222720002941696 , F1: 0.4836065573770492\n",
      "Training at Epoch 55 iteration 0 with loss 0.0301464\n",
      "Validation at Epoch 55 , AUROC: 0.9332803853248616 , AUPRC: 0.5173871664260096 , F1: 0.5037037037037037\n",
      "Training at Epoch 56 iteration 0 with loss 0.021212561\n",
      "Validation at Epoch 56 , AUROC: 0.9371669399467104 , AUPRC: 0.5239589621462077 , F1: 0.43946188340807174\n",
      "Training at Epoch 57 iteration 0 with loss 0.039023615\n",
      "Validation at Epoch 57 , AUROC: 0.9352608116417298 , AUPRC: 0.5261320356960077 , F1: 0.49137931034482757\n",
      "Training at Epoch 58 iteration 0 with loss 0.029012958\n",
      "Validation at Epoch 58 , AUROC: 0.9394189383070302 , AUPRC: 0.5570794763794581 , F1: 0.49811320754716976\n",
      "Training at Epoch 59 iteration 0 with loss 0.026636513\n",
      "Validation at Epoch 59 , AUROC: 0.9454114572658332 , AUPRC: 0.574369266581412 , F1: 0.5306122448979591\n",
      "Training at Epoch 60 iteration 0 with loss 0.011317954\n",
      "Validation at Epoch 60 , AUROC: 0.9385606681697068 , AUPRC: 0.5182290693284501 , F1: 0.5084745762711865\n",
      "Training at Epoch 61 iteration 0 with loss 0.017281646\n",
      "Validation at Epoch 61 , AUROC: 0.93493799959008 , AUPRC: 0.527837023340734 , F1: 0.50199203187251\n",
      "Training at Epoch 62 iteration 0 with loss 0.024695056\n",
      "Validation at Epoch 62 , AUROC: 0.917083418733347 , AUPRC: 0.46747837263743863 , F1: 0.4684014869888476\n",
      "Training at Epoch 63 iteration 0 with loss 0.03964399\n",
      "Validation at Epoch 63 , AUROC: 0.9341514654642344 , AUPRC: 0.5140048850708893 , F1: 0.4649446494464945\n",
      "Training at Epoch 64 iteration 0 with loss 0.018066056\n",
      "Validation at Epoch 64 , AUROC: 0.9425548268087725 , AUPRC: 0.5532193445140754 , F1: 0.5267489711934156\n",
      "Training at Epoch 65 iteration 0 with loss 0.026209088\n",
      "Validation at Epoch 65 , AUROC: 0.9416734986677597 , AUPRC: 0.545691035097604 , F1: 0.5296442687747036\n",
      "Training at Epoch 66 iteration 0 with loss 0.012367565\n",
      "Validation at Epoch 66 , AUROC: 0.9366852838696454 , AUPRC: 0.5524494451787633 , F1: 0.5188284518828452\n",
      "Training at Epoch 67 iteration 0 with loss 0.03219722\n",
      "Validation at Epoch 67 , AUROC: 0.938014962082394 , AUPRC: 0.536142676539815 , F1: 0.5220883534136546\n",
      "Training at Epoch 68 iteration 0 with loss 0.029184632\n",
      "Validation at Epoch 68 , AUROC: 0.9381251281000205 , AUPRC: 0.5105273519357686 , F1: 0.4615384615384615\n",
      "Training at Epoch 69 iteration 0 with loss 0.024376426\n",
      "Validation at Epoch 69 , AUROC: 0.9397904283664686 , AUPRC: 0.5697391962756181 , F1: 0.5081967213114753\n",
      "Training at Epoch 70 iteration 0 with loss 0.026943563\n",
      "Validation at Epoch 70 , AUROC: 0.9405846484935438 , AUPRC: 0.5390892483251566 , F1: 0.48936170212765956\n",
      "Training at Epoch 71 iteration 0 with loss 0.032199685\n",
      "Validation at Epoch 71 , AUROC: 0.9448068251690921 , AUPRC: 0.5588207588163694 , F1: 0.5468164794007491\n",
      "Training at Epoch 72 iteration 0 with loss 0.007292864\n",
      "Validation at Epoch 72 , AUROC: 0.9441202090592334 , AUPRC: 0.5583111987026674 , F1: 0.5344129554655871\n",
      "Training at Epoch 73 iteration 0 with loss 0.019239172\n",
      "Validation at Epoch 73 , AUROC: 0.9368646238983399 , AUPRC: 0.5471383028573246 , F1: 0.5110132158590308\n",
      "Training at Epoch 74 iteration 0 with loss 0.0063166097\n",
      "Validation at Epoch 74 , AUROC: 0.943331112932978 , AUPRC: 0.5406544632797723 , F1: 0.5276595744680851\n",
      "Training at Epoch 75 iteration 0 with loss 0.019895848\n",
      "Validation at Epoch 75 , AUROC: 0.9383813281410126 , AUPRC: 0.5438196559861319 , F1: 0.5317460317460317\n",
      "Training at Epoch 76 iteration 0 with loss 0.018401688\n",
      "Validation at Epoch 76 , AUROC: 0.9404232424677187 , AUPRC: 0.4840329611153865 , F1: 0.4773662551440329\n",
      "Training at Epoch 77 iteration 0 with loss 0.021034708\n",
      "Validation at Epoch 77 , AUROC: 0.9346075015372002 , AUPRC: 0.558989820846276 , F1: 0.5299145299145299\n",
      "Training at Epoch 78 iteration 0 with loss 0.02179912\n",
      "Validation at Epoch 78 , AUROC: 0.9404155564664889 , AUPRC: 0.5422541628311739 , F1: 0.5179282868525896\n",
      "Training at Epoch 79 iteration 0 with loss 0.020792892\n",
      "Validation at Epoch 79 , AUROC: 0.941537712646034 , AUPRC: 0.5295737820108503 , F1: 0.4919354838709677\n",
      "Training at Epoch 80 iteration 0 with loss 0.037752084\n",
      "Validation at Epoch 80 , AUROC: 0.9405897724943636 , AUPRC: 0.5579116978469638 , F1: 0.5333333333333333\n",
      "Training at Epoch 81 iteration 0 with loss 0.006811868\n",
      "Validation at Epoch 81 , AUROC: 0.9444686411149826 , AUPRC: 0.5574673842372466 , F1: 0.5121951219512194\n",
      "Training at Epoch 82 iteration 0 with loss 0.027747478\n",
      "Validation at Epoch 82 , AUROC: 0.9396674523467924 , AUPRC: 0.5503000405290495 , F1: 0.4999999999999999\n",
      "Training at Epoch 83 iteration 0 with loss 0.012403611\n",
      "Validation at Epoch 83 , AUROC: 0.9418425906948146 , AUPRC: 0.578131480608976 , F1: 0.5682656826568265\n",
      "Training at Epoch 84 iteration 0 with loss 0.013104705\n",
      "Validation at Epoch 84 , AUROC: 0.9398980323836852 , AUPRC: 0.5741387123402509 , F1: 0.5176470588235293\n",
      "Training at Epoch 85 iteration 0 with loss 0.016158246\n",
      "Validation at Epoch 85 , AUROC: 0.9398032383685181 , AUPRC: 0.5539465381354913 , F1: 0.5296442687747036\n",
      "Training at Epoch 86 iteration 0 with loss 0.0110019045\n",
      "Validation at Epoch 86 , AUROC: 0.9279834494773519 , AUPRC: 0.5162382073756486 , F1: 0.5352112676056339\n",
      "Training at Epoch 87 iteration 0 with loss 0.024818772\n",
      "Validation at Epoch 87 , AUROC: 0.9278463824554212 , AUPRC: 0.5408340926911602 , F1: 0.5263157894736842\n",
      "Training at Epoch 88 iteration 0 with loss 0.045672685\n",
      "Validation at Epoch 88 , AUROC: 0.9368287558926008 , AUPRC: 0.5577093944539343 , F1: 0.5263157894736842\n",
      "Training at Epoch 89 iteration 0 with loss 0.015461825\n",
      "Validation at Epoch 89 , AUROC: 0.9356425497028079 , AUPRC: 0.5526410238601228 , F1: 0.5354330708661418\n",
      "Training at Epoch 90 iteration 0 with loss 0.012365005\n",
      "Validation at Epoch 90 , AUROC: 0.934264193482271 , AUPRC: 0.582690359573256 , F1: 0.5654008438818565\n",
      "Training at Epoch 91 iteration 0 with loss 0.014866872\n",
      "Validation at Epoch 91 , AUROC: 0.9439280590284895 , AUPRC: 0.571619657918198 , F1: 0.5657370517928287\n",
      "Training at Epoch 92 iteration 0 with loss 0.049499005\n",
      "Validation at Epoch 92 , AUROC: 0.9448862471817996 , AUPRC: 0.5747947224062507 , F1: 0.5355648535564853\n",
      "Training at Epoch 93 iteration 0 with loss 0.012299232\n",
      "Validation at Epoch 93 , AUROC: 0.9380636400901825 , AUPRC: 0.5621987326629349 , F1: 0.5439999999999999\n",
      "Training at Epoch 94 iteration 0 with loss 0.018092467\n",
      "Validation at Epoch 94 , AUROC: 0.944843974175036 , AUPRC: 0.5501493882348499 , F1: 0.5166666666666666\n",
      "Training at Epoch 95 iteration 0 with loss 0.014761863\n",
      "Validation at Epoch 95 , AUROC: 0.9472817175650747 , AUPRC: 0.5785907761359472 , F1: 0.5344129554655871\n",
      "Training at Epoch 96 iteration 0 with loss 0.047205735\n",
      "Validation at Epoch 96 , AUROC: 0.9354273416683746 , AUPRC: 0.601732041427251 , F1: 0.5537190082644627\n",
      "Training at Epoch 97 iteration 0 with loss 0.009167598\n",
      "Validation at Epoch 97 , AUROC: 0.9438409510145522 , AUPRC: 0.6043662457184974 , F1: 0.549800796812749\n",
      "Training at Epoch 98 iteration 0 with loss 0.008935517\n",
      "Validation at Epoch 98 , AUROC: 0.9436103709776593 , AUPRC: 0.5530776450016752 , F1: 0.5058365758754864\n",
      "Training at Epoch 99 iteration 0 with loss 0.011557452\n",
      "Validation at Epoch 99 , AUROC: 0.94342974994876 , AUPRC: 0.5197582702967057 , F1: 0.5247148288973384\n",
      "Training at Epoch 100 iteration 0 with loss 0.00818947\n",
      "Validation at Epoch 100 , AUROC: 0.9287366775978685 , AUPRC: 0.5251459289228797 , F1: 0.5403225806451614\n",
      "--- Go for Testing ---\n",
      "Testing AUROC: 0.9196785488678763 , AUPRC: 0.5605287830157307 , F1: 0.5884413309982488\n",
      "--- Training Finished ---\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAEaCAYAAAAVJPDdAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nO3dd5xU1fnH8c9DE5ClowJLB0E6uFLsDQUVRAUDxoKNRGKJmmJiiWj8aSwxYoiKJYgNkUTBrghIREAWBKWIIFKWIk16XTi/P87d3ZntA1N2dr/v12tfc+vc5+7CPHPOueccc84hIiKSpVyiAxARkZJFiUFERMIoMYiISBglBhERCaPEICIiYZQYREQkjBKDJDUzW2hmZxZxTGMz22lm5eMUVkyZ2RAz+yJk3ZlZy0TGJKWLEoPEhJmtMLM9wQfyT2b2bzOrFu3rOOfaOeemFnHMKudcNefcwWhfP/hQ3hXc5xoz+3tJS0Bmdr6ZTTOzHWa20cw+N7N+iY5LSi4lBomlvs65akBX4CTgntwHmJfs/w47Bfd5BvAL4LoEx5PNzAYAbwFjgFTgWOA+oO9hvFdp+FtJMeiPLDHnnFsDfAi0BzCzqWb2kJlNB3YDzc2shpm9aGbrgm/efw395m1mN5rZ4uBb7yIz6xpsX2Fm5wbL3cws3cy2B6WUvwfbmwbf7CsE6w3MbKKZbTGzZWZ2Y8h17jezcWY2JrjWQjNLK+Z9LgOmA51D3u9w7+suM/shZPslkf7ezcyAvwMPOudecM5tc84dcs597py7MeR+Xw05J/fvKvff6s9mlp7rOreb2cRg+Sgze9zMVgV/g2fNrEqksUtiKTFIzJlZI+AC4OuQzVcBQ4EUYCXwMpAJtAS6AOcBNwTnDwTuB64GqgP9gM35XOop4CnnXHWgBTCugJDeADKABsAA4P/M7JyQ/f2AsUBNYCLwz2LeZxvgNGBZyObDva8fgveqAQwHXjWz+sWJI0RroBEwPsLzcgv9Wz0NtDazViH7rwBeD5b/BhyPT44tgYb4EookE+ecfvQT9R9gBbAT2Ir/4P8XUCXYNxV4IOTYY4F9WfuDbYOBKcHyx8BthVzn3GB5Gv5DtG6uY5oCDqiA/6A8CKSE7H8YGB0s3w9MCtnXFthTyH06YDuwK1h+AzjqSO8rn+vMAy4OlocAX+SKoWU+55wS7KtcyPveD7ya3+8qv79VsO1V4L5guRWwA6gKWPB7aBFybE/gx0T/e9RPZD8qMUgs9XfO1XTONXHODXPO7QnZtzpkuQlQEVhnZlvNbCvwHHBMsL8R/ht0Ua7Hf1v9zsxmm9lF+RzTANjinNsRsm0l/pttlvUhy7uByllVKwXoClTDty90B44+0vsys6vNbF7Iee2BuoXEkJ+s0kekJY3cVudafx2f4MCXFt5xzu0G6uETxJyQuD8KtksSUWKQRAkd1nc1/pt13SCR1HTOVXfOtQvZ36LIN3RuqXNuMP6D92/AeDM7Otdha4HaZpYSsq0xsOZwbyS4tnPOjQNmkFN1clj3ZWZNgOeBm4E6zrmawAL8N/JILAmucVkhx+zCf5hnOS6fY3IPwfwJUNfMOuMTRFY10iZgD9Au5H5rON8wL0lEiUESzjm3Dv9h84SZVTezcmbWwszOCA55AfidmZ0YPBnTMvjwDGNmV5pZPefcIXwVFvhqo9BrrQa+BB42s8pm1hFf0ngtSrfzCDDUzI47gvs6Gv9hvDG4r2sJGu4j4ZxzwB3AvWZ2bUgMp5rZqOCwecDp5vt61AD+VIz3zcS3WzwG1AY+DbYfwie0J83smCD2hmZ2fqSxS2IpMUhJcTVQCVgE/Iz/4KkP4Jx7C3gI/810B/AO/gMpt97AQjPbiW+IHuSc25vPcYPxdelrgbeBvzjnPo3GTTjnvgU+B35/uPflnFsEPIEvffwEdMA/7XQ48Ywn5xHatcH7/RWYEOz/FHgT+AaYA7xXzLd+HTgXeCtIFFn+iG98n2lm24FJ+EZwSSLmv1SIiIh4KjGIiEgYJQYREQmjxCAiImGUGEREJExhnXaSQt26dV3Tpk0THYaISFKZM2fOJudcvp0Pkz4xNG3alPT09KIPFBGRbGa2sqB9qkoSEZEwSgwiIhJGiUFERMIkfRuDSGly4MABMjIy2Ls3v5E8RCJXuXJlUlNTqVixYrHPUWIQKUEyMjJISUmhadOm+AnYRA6fc47NmzeTkZFBs2bNin1e3KqSzOwlM9tgZgsK2G9mNiKYavGbrCkORcqSvXv3UqdOHSUFiQozo06dOhGXQOPZxjAaP/plQfrgZ4NqhZ9G8Jk4xCRS4igpSDQdzr+nuCUG59w0YEshh1wMjAkmPJkJ1DyMOW6L74u74cWWsG1FzC4hIpKMStJTSQ0Jn0Iwg/DpFrOZ2VAzSzez9I0bNx7e1bZ8B1t/gBUfHd75IiKlVElKDPmVd/KdLMI5N8o5l+acS6tX7zCnk23ax7/++OHhnS9SSpUvX57OnTvTvn17Bg4cyO7du/Ns79u3L1u3bi3ineLPOcfZZ5/N9u3bEx1KgebMmUOHDh1o2bIlt956K/nNiTN16lRq1KhB586d6dy5Mw888ED2vq1btzJgwADatGnDCSecwIwZMwD43e9+x+TJk6MSY0lKDBn4ydGzpOJnnIqNZkFzx6rPIHNfzC4jkmyqVKnCvHnzWLBgAZUqVeLZZ5/Ns7127dqMHDkyqtc9ePBg0QcV4YMPPqBTp05Ur149rteNxE033cSoUaNYunQpS5cu5aOP8q+1OO2005g3bx7z5s3jvvvuy95+22230bt3b7777jvmz5/PCSecAMAtt9zCI488EpUYS9LjqhOBm81sLNAd2BbMmRsbKalQtwNs+hbWfAFNzonZpUQOyxMxaoS+s/izNp522ml88803ebb37Nkz3+0AY8aM4fHHH8fM6NixI6+88gpDhgzhoosuYsCAAQBUq1aNnTt3MnXqVIYPH079+vWZN28effv2pUmTJgwbNgyA+++/n5SUFO68804ee+wxxo0bx759+7jkkksYPnx4nmu/9tprDB06NHu9f//+rF69mr1793Lbbbdl76tWrRp33HEHH3/8MU888QRVqlThjjvuYOfOndStW5fRo0dTv359nn/+eUaNGsX+/ftp2bIlr7zyClWrVi327y+3devWsX37dnr27AnA1VdfzTvvvEOfPn2Kdf727duZNm0ao0ePBqBSpUpUqlQJgCZNmrB582bWr1/Pcccdd9gxQnwfV30DP4dtazPLMLPrzezXZvbr4JAPgOX4+WKfB4bFPKhmqk4SKUhmZiYffvghHTp0CNt+8OBBPvvsM/r165fnnIULF/LQQw8xefJk5s+fz1NPPVXkdb766iseeughFi1axKBBg3jzzTez940bN46BAwfyySefsHTpUr766ivmzZvHnDlzmDZtWp73mj59OieeeGL2+ksvvcScOXNIT09nxIgRbN68GYBdu3bRvn17Zs2aRffu3bnlllsYP348c+bM4brrruPuu+8G4NJLL2X27NnZ38xffPHFPNecMmVKdpVP6M/JJ5+c59g1a9aQmpqavZ6amsqaNWvy/b3MmDGDTp060adPHxYuXAjA8uXLqVevHtdeey1dunThhhtuYNeuXdnndO3alenTD2t68DBxKzE45wYXsd8Bv4lTOF6zPjD7UVjxIfB4XC8tUqQIvtlH0549e+jcuTPgSwzXX3992PYVK1Zw4okn0qtXrzznTp48mQEDBlC3bl0AateuXeT1unXrlt35qkuXLmzYsIG1a9eyceNGatWqRePGjRkxYgSffPIJXbp0AWDnzp0sXbqU008/Pey9tmzZQkpKSvb6iBEjePvttwFYvXo1S5cupU6dOpQvX57LLrsMgCVLlrBgwYLs+zl48CD16/sHIhcsWMA999zD1q1b2blzJ+eff36e+M866yzmzZtX5H0C+bYn5Pc4adeuXVm5ciXVqlXjgw8+oH///ixdupTMzEzmzp3L008/Tffu3bntttt45JFHePDBBwE45phjWLv2yGvgS1JVUvw1OAUqpcDmRbB9FVRvnOiIRBIuqy2hoO3btm3joosuYuTIkdx6661hxzjn8v2gq1ChAocOHco+Zv/+/dn7jj766LBjBwwYwPjx41m/fj2DBg3KPudPf/oTv/rVrwqNPes65cqVY+rUqUyaNIkZM2ZQtWpVzjzzzOyOXpUrV6Z8+fLZ792uXbvsRtxQQ4YM4Z133qFTp06MHj2aqVOn5jlmypQp3H777Xm2V61alS+//DJsW2pqKhkZGdnrGRkZNGjQIM+5oW0kF1xwAcOGDWPTpk2kpqaSmppK9+7ds39Xoe0Ke/fupUqVKoX9ioqlJDU+x1/5itD4XL+s6iSRYqlRowYjRozg8ccf58CBA2H7zjnnHMaNG5ddZbNli++61LRpU+bMmQPAhAkT8pwXatCgQYwdO5bx48dnt0mcf/75vPTSS+zcuRPwVTIbNmzIc27r1q1Zvnw5ANu2baNWrVpUrVqV7777jpkzZ+Z7vdatW7Nx48bsxHDgwIHsqpsdO3ZQv359Dhw4wGuvvZbv+Vklhtw/uZMCQP369UlJSWHmzJk45xgzZgwXX3xxnuPWr1+fXbr46quvOHToEHXq1OG4446jUaNGLFmyBIDPPvuMtm3bZp/3/fff0759+3zjjETZTgygdgaRw9ClSxc6derE2LFjw7a3a9eOu+++mzPOOINOnTpxxx13AHDjjTfy+eef061bN2bNmpWnlJD7PXbs2EHDhg2zq3TOO+88rrjiCnr27EmHDh0YMGAAO3bsyHPuhRdemP2tvnfv3mRmZtKxY0fuvfdeevToke/1KlWqxPjx4/njH/9Ip06d6Ny5c/aH+oMPPkj37t3p1asXbdq0ifj3lJ9nnnmGG264gZYtW9KiRYvshudnn302+wmw8ePH0759ezp16sStt97K2LFjs0tiTz/9NL/85S/p2LEj8+bN489//jPgE9qyZctIS0s74hgtvzqvZJKWluaOaAa3HRkwqhFUrAa/2QzlK0UvOJEILV68OPvxQ4ncunXruPrqq/n0008THUrcvf3228ydOze7vSFUfv+uzGyOcy7fLKISQ0oq1G0PB3b6x1ZFJGnVr1+fG2+8sUR3cIuVzMxM7rzzzqi8lxIDqBe0lCjJXopPtMsvvzyiDm6lxcCBA6lZs2ae7Yfz70mJAdTOICVG5cqV2bx5s5KDREXWfAyVK1eO6Lyy/bhqloan+DaGzQth+2qo3qjoc0RiIOtxxsMeHFIkl6wZ3CKhxAC+wbnJubDsHd/ZrePQos8RiYGKFStGNNOWSCyoKimLqpNERAAlhhxNg9FWV06Cg/sLP1ZEpBRTYshSvTHUaRs8tnrkg1CJiCQrJYZQBT22uuJTWDYh/vGIiCSAEkOorHaGFSGJYcsS+M95MKE/ZO5NTFwiInGkxBCq4alQ8WjYtMA/tgow7Q85+w9lJiau3JyDbT+CO5ToSESkFFJiCFXhKGgczOS24iPYtBB+mJjYmEIdyoTvxsJrJ8ELzWHmQ4mOSERKIfVjyK1ZH58MfvzQT/tZEuzfCQtehDlPwvaVOdu3r0hYSCJSeikx5JbVzrBqEliCC1T7d8Kcv8Pcf8Den/22Wq2gdlv4QY3hIhIbqkrKrXoTqH0C7N8B+7ZB/Z6+3SGanIMFo2FMZ1j5Wd79Bw/A/GfhxZbw5V98UqjfE/r9F4YshhZ9oxuPiEgIlRjy06wPbFnslzv9Gj4bFr333rcNPv0VLAkmPF/+HjQ5J2d/xhcw6Vd+ulGA47rB6X+DRmdGLwYRkUKoxJCfZhf418q14fiB0XvfLUvgtW45SSHU3p/hk6Hw5mk+KdRsAReNgytmKimISFypxJCfxmfDqQ/DsV2g4pFPrA34TnLvDfQlhrodoOFpMP9fOfs+uhp2rYdyFaHbXdD9z1ChiKFyN3wNY0+Htldq4D8RiRolhvyYQfe7ovd+S8bBB1fCoQPQ6lLo/TJ8+7zft/Q/MPcpwEGDU+C8UX5ojuLY8LV/dYeUGEQkapQYYu3bl+CTGwAHJ94JZzwa/rTTjtV+vedwX0ooV77o90wJ5ouoWM2P7SQiEkVKDLG09L85SeHUh6Dbn3xpBOCoYAq+ag3hwtch9fTiv2/T8+C672H7Khh/btTDFpGyTYkhEjvXQu3ji3dsxhfw/hWAg1P+6ksDodpcAUfVgNQzoEqdyGOp1Qp2/RT5eSIiRdBTScWRNSbRv9vA3q1FH795EUzoBwf3Qaeb8iYF8MNvtLr08JJC3gBh1ZScR1yLcvBAFK4pIqWVEkNxZO4JFhxs+a7wY/dsgf/08Y+ftuwPZz+dU30UK2tnwFtnwzsX+/XMffD1P2HqHeED/235Hj4aAk9VgVkPxzYmEUlaqkqK1K61he+fchvsWOU7pl3wevEakw9XdsJx/mXvz36QvS/+7EdfBV9lVbEqzPyr7z+RVfr5aU7s4hKRpKbEEKkdqwvet2wiLH4VKlSBC16NXh+IghzTBVpeAkcf64fQ2LsZ3h8cfsyES2DnGsD5PhLHngjrZ8c2LhFJaqpKKo6r5kGLoJpmR0b+x+zZ4oeyADj1/3zjcKxVrAoX/9dfL8vR9aHX8zl9IXZmQPmK0GkYXL8MTvpD/u8lIhJQiaE4jukExw/wI5oWlBim3Op7Ljc8FbreGt/4KteC0x4BKw+db/KD/q2bCT8vhQ43QI97oFoDf+z6r+Ibm4gkHSWG4kpJ9a87VsGBPeHVRMsmwOLXfBXS+f9OzHDd3f4Yvn7eKDj7qaJHhl3zJcx+FGo0g7OejF18IpI04poYzKw38BRQHnjBOfdIrv2NgZeBmsExdznnPohnjAXK6m289ksYWROuWwrVG/sqpE+DKqTTHoZaLRMXYygrV3hS2PYjjD8fVn7i18tV9Pd4cG/+j9eKSJkRt8RgZuWBkUAvIAOYbWYTnXOhD9/fA4xzzj1jZm2BD4Cm8YqxUNUa5iwf3O+rZKo3hpkPwu6f/KB4XW5JXHyR2jDXv1Y8Gg7s8uM4fX6n39b2GkhpWPC5IlKqxbPOoxuwzDm33Dm3HxgLXJzrGAdUD5ZrAEU8GxpHuUc6rVTd94Se/4xfP/vpxM/4VhxV6vnXitWg+91ww495SxaH1AFOpCyLZ1VSQyD0Wc8MoHuuY+4HPjGzW4CjgXwHAjKzocBQgMaNG0c90AKlNAp/XHXWw753c6vLfAN1Mkg9Ha6Y5ed7yOp1/YtpvhT03i98G4qIlGnx/IqbX/dfl2t9MDDaOZcKXAC8Ypb3a7hzbpRzLs05l1avXr0YhFqAyz7JWd6xGr4dBRicfH/8YjhSZlC/W/hQHMd2hQY9Yt9DW0SSQjwTQwbQKGQ9lbxVRdcD4wCcczOAykDduERXHHXaQJNefnn2o/5bduvLoW77xMYVaxnTYNIw2Lw40ZGISBzEMzHMBlqZWTMzqwQMAibmOmYVcA6AmZ2ATwwb4xhj8f38vX8tzU/wrP4cxp0Fb57h21K+GVXwsbvWw+YixpESkaQQtzYG51ymmd0MfIx/FPUl59xCM3sASHfOTQTuBJ43s9vx1UxDnHO5q5tKjsZnQ72OiY4i+lZPhY9f9q+h8muU3rwIZj/uhwI5lAl9Xoa2V8UjShGJkbj2Ywj6JHyQa9t9IcuLgFPiGVPEQvNUlzj3cI61vVv868fX+tejasKJd/inrabfk3Occ7DmC1+dtvy9YKMBDj661j/x1OqSeEYuIlGUBM9XljBrv8hZbn5h4uKIhaPr5yyf/ADcuAJ63usnFAI/MuvSt+GNk+HN031SqFDZzzlx3ffQ415wB/3TTT8vS8gtiMiR05AYkcrc61/r94RypezXd+aTsOlb6DjUj7+U2zejcvptVK4NnW+GLr+Bqsf4bScPh7XTYdVk2PB1yekFLiIRKWWfbHHQaxR8+wL0G5/oSKKv+QX+J7esJ4bdQaiWCif9Hjpcn7djnJlPGABfPw0zhsP5L/nHY0UkaSgxRKrjjf6nLGl+EayeAo3OhvbX+WlJi7Lmf/515SdKDCJJRm0MUrTqjaHvW35I76KSwnHdoXwlSAl6pC8ZB883hcWv5xxzcL8fjfaNU+CNU/0ER1+P9FOSikjCWUl+GrQ40tLSXHp6eqLDkFDuEEy/D2Y9lLOt7VVw+qMw/zn45lnf7yG3+j3g8qlFJ59DB/2jtIcOQLPe0YxcpMwwsznOubT89qkqSaLPykGDk/3TTNUa+r4OP37o56PO6gtRtz1sWuCXq6X67etmwqrPoEZTn0CWv+cnIGo90B+35XtY9DIsHONnpgNo2d8/QbX1B2jRN7ZzbIuUESoxSGwtegU+vDpYMWh5se//0ehM2LYcdq6DhifDhEv9DHkpjfMO5NfjHv+k09ovC79WxaNhyCJf9SUihVKJQRIn9cygh3gX/2hrjWY5+2q28D+QM4DfjlW+g1zVen4yIYCZf/WvFav5sanaXeMTyvuDwq91YJdvy+h+V0xvSaS0U2KQ2KreCAZ+VvRx7a/3fURaXQptBkOlarA+3feH2LYc6rT1+0IfkW14Kmz6BirVgLFBh/l5T0PbK3OmYhWRiKkqSUqH/TtgfC9YNwtqtYZB03I63olIHoVVJelxVSkdKqXAJR/4QQ1/XgJz/pHoiESSlhKDlB5VavtxmwD2bk5sLCJJTIlBShnNQidypJQYpHT6ZhQsfDnRUYgkJSUGKV1Cn1r64d3ExSGSxJQYpHQ5fgA0Oc8v71oP370Jc5+GDfMTG5dIElE/BildKlT280ms/MTPDbF2ut9eozlcu9gP8CcihVKJQUqf1DN8b+sGJ0OzPn7btuXwek/Y+zPs+imx8YmUcCoxSOlTtW5Ob+v9O+Hp6oCDDXNhZDCR0G+25D9LnYioxCClXKVqvgoptxdbwL5tOev7tscvJpESTiUGKf1qt4Y+Y2DPJihXCSbf7KuUVk7yw3UvftXPdX3Rm36QPpEyTolByoa2V+UsLxoD67+CdweEH7NpIbSOb1giJZGqkqTs6Xkf1Gnnn2A6fgA075voiERKlMMuMZhZTXIlFufcliOOSCTWml/of7J8ORyWqzOcSJaIEoOZNQGeBc4CKobuAhygeRUlef2UDocyoZxqWKVsi/R/wL+BmsB1wFp8MhApHX78AJ6sCL1H+1niRMqoSBNDN6CHc25BLIIRSYg6bcPXf/xIiUHKtEgbn38EjopFICIJ03og3PQTNO1dvON3/QTrZ8c2JpEEijQx3AY8bGYtYxGMSMJUPQbaXu2Xl4yFZxvAqik5+/dth4Vj4D+94dnj4LVuMPV3cOhgYuIViaFIq5Im4EsMS8xsH5AZutM5Vz1agYnEXYUqOcu71sFbZ8PxQYe35RMhc2/48XOe8P0hLnkXjqoRvzhFYizSxHBzTKIQKQma9YFT/grT78nZ9v24nOXU06HNFXDoAEy+xW9b8z/4Z024binUUkFaSgdzLrkfLEpLS3Pp6emJDkNKm33b4KXWcHR9aDMY2gyC6o1z9q+dCW/0zFlv2hvqnOATR9VjoUpdqFgl7/uKlBBmNsc5l5bvvkgTg5kdBfwSaIt/XHUh8IZzbt+RBno4lBgkYXZkwKhG+e9r1gcu/cC3QaydDlYBGvT0w38vfRuWve3HZ7rgNWihntcSf4Ulhkg7uLUFPgKqA98Gm28EhptZb+dcPsNYhp3fG3gK3xHuBefcI/kcczlwPz7pzHfOXRFJjCJxk5IKg7+Et86FzN3h+9bPhknDYOl/YXch8z9kTFNikBIn0jaGp4Cvgaucc9sBzKw68CrwD+D8gk40s/LASKAXkAHMNrOJzrlFIce0Av4EnOKc+9nMjokwPpH4atATbt3pl78bC/u3w6Rf+5Fc5z+T9/ijavrSROZeX2oAPwXpD+/Cqs+gSS9oeCrs2ex7YNdtFz6PtUgcRFSVZGa7gZOccwtzbe8AzHTOFfgv2Mx6Avc7584P1v8E4Jx7OOSYR4HvnXMvFDcmVSVJiXLwAIw91bdRHH8ZtBoAx3T2H/5H1cxpd5j9GEz7Q3BS1ogy+ah6DNywXMlBoi5qVUnAXvyQGLnVCPYVpiGwOmQ9A+ie65jjAcxsOr666X7n3Ee538jMhgJDARo3bpx7t0jilK8Iv5yVd3u1+uHr9Tr7EsGhTD8P9XEnwZov/L4azaFqPVg3C3ZvgNmPQ9qdfv6I5e/BuhlwVC1ffbVnky9h1GzhX5sWWGgXKbZISwwvAyfh2xVmBpt7As8BXznnri3k3IHA+c65G4L1q4BuzrlbQo55DzgAXA6kAv8D2jvnthb0vioxSNLas8U3Rtdu42ea27MZ3CGfFMA/FfXz95G9Z4t+0H9C9GOVUieaJYbbgJfxH9hZXT7LAROB3xZxbgYQ+ghHKn4gvtzHzHTOHQB+NLMlQCtA4w9I6VOltv/JXq8Tvv+kP8AnN+SsNzgZml8Ejc+GzD2+tDHhEl/dtG25P+aHifDqSbDtB+g0DFZ+6jvhnfY3OPG3vnQiUoTD6scQNBK3wVeOLnLOLSvGORWA74FzgDX4D/srQtsrgqeWBjvnrjGzuviG7s7Ouc0Fva9KDFKq7VgDP83xSaFq3YKP27kOnmtQ+Hu1vARaXepLFUdpkIKyLpolBgCcc0uBpRGek2lmNwMf49sPXnLOLTSzB4B059zEYN95ZrYIXyL5fWFJQaTUS2nof4pSrT78aq1vf0h/wrdNNO0NZvD10/6YZUH/iR73wSnDYxu3JLUiSwxmNgL4k3NuV7BcIOfcrdEMrjhUYhApwk9z4Yu7Ycdq2LwQKlSF9tfC8QOh0RmJjk4S5EhLDB3Ima2tQyHHJffYGiKl1bFd4bIP4ccP4b8X+KeZ5o2E78fD9ct8w7dIiCITg3PurPyWRSTJNOsDg2fAd2/A1yN8j+z0J+DkvyQ6Milhjnhy22BuhgznXFH9GEQk0Rr08D9NzoV3+sGM+2Htl74t4swn/UCAUuZFNFGPmf2fmV0TLJuZfYp/0midmfWIRYAiEgMNToYqQX+JlZ/Aio99j+1DmYWfJ2VCpDO4/RJYEiz3AToDPYAxwMMFnSQiJUyVOuRUOoAAABPXSURBVL59ofdoKBc0Ie7dAvOfS2hYUjJEmhiOxXdCA7gAGOec+wp4GugSzcBEJMaOqg7troHb90O14JHYyTfD7k2JjUsSLtLEsBloEiyfB0wOlivgO7uJSDIaMCln+Zl6fn7rpf/1Q3RImRNpYvgP8HrQtlAbPzcD+CqlIns/i0gJVacNnP5ozvpH18DEy2DVlMTFJAkTaWK4AxgBLAJ6Oed2BdvrA/kMPi8iSeOk30Ov56B+Tz89KcD4c/1wG1KmRJQYnHOZzrknnHO3Oee+Dtn+ZCRzKIhICdVxKFzxpR9PKcsPExMXjyREkf0YzKwrMM85dyhYLpBzbm7UIhORxOn5F/j2BcCpnaEMKk4Ht3TgOGBDsOzIv6HZ4QfHE5Fkl9IQOv0K5j8Lnw2D5hdA9SZFnyelQnESQzNgY8iyiJQFFUPGUHq+KZx4O3S9TQmiDCjOWEkr81sWkVKu629h3r/8oHsAc570EwSdq+dMSrtIh8S42cyuzGf7lWY2LHphiUjCpTSEa76FJuflbNu/M3HxSNxE+rjqb4HV+WxfAdx+xNGISMlSszkM+Bj6jEl0JBJHkSaGVCC/6qSMYJ+IiCS5SBPDenwv59y6AhpgRaS0y/gc3h0IKz5JdCQSQ5HOx/A6MMLMdgFTg21nAf8AXotiXCJSklQ9xr/uWO1/vh/v+zq0vASO6ZTY2CTqIi0x/AWYDnwM7A5+PgS+BO6NbmgiUmI06QUDPoVTH8rZNmM4TLwkcTFJzERUYnDOHQAGm9l9+ColA+Y65zSAnkhpZuX8rG+pZ8BPc/zIqwC7fkpsXBITkZYYAHDOLQWmAeOVFETKkPIVod9/4Dc/+/XM3TlJQkqNSPsxVDSzR81sB7AGaBps/5v6MYiUIeVCpmB5/wo4dDCh4Uh0HU4bQ1/gSmBfyPavgCFRiklESrpK1eCcf/rlg/vwQ6VJaRFpYhgM/No5NwEIHXJxAXB81KISkZKv8zDf9gCwa31iY5GoijQxNCD/Dm4ViPzRVxFJdllDco9uBwf2JDYWiZpIE8NC4PR8tl8OzDnycEQkqRw/wL/u3+77N0ipEGliGA48bWZ34+deGGhm/wbuAh6MdnAiUsL1fQsq1/HL/24Ne7cmNh6Jikin9nwXXzo4D9/G8BegFdDXOTcp+uGJSInXpFfO8shafqhuzfqW1IrdLmBmFfAJYZZz7ozYhSQiSeWiN2BnBqz5wq9/9huo3QYan53YuOSwFbvE4JzLBP4LpMQuHBFJSn3GQLshOev7tiUsFDlykbYxzAdaxiIQEUliNZpB739Dy/6JjkSiINLEcD/whJn1N7NGZlY79CcG8YmISJxFmhjeBzrgq5RWABuDn03Ba6HMrLeZLTGzZWZ2VyHHDTAzZ2ZpEcYnIiJHKNJOaWcd7oXMrDwwEuiFn/FttplNdM4tynVcCnArMOtwryUiIoevWInBzKoCjwH9gYrAJOBW51wks7Z1A5Y555YH7zkWuBhYlOu4B4FHgd9F8N4iUpIsextqtoC67XOGzZCkUdy/2HD8IHnvA2/gv/U/E+G1GgKhXSMzgm3ZzKwL0Mg5915hb2RmQ80s3czSN24ssgZLROJt0SswphP8vTy80AKWvpPoiCQCxU0MlwLXO+eGOuduAy4E+gfVQ8Vl+WzLHpLRzMoBTwJ3FvVGzrlRzrk051xavXr1IghBRGKq9aC827Ythx8mxj8WOWzFTQyNgP9lrTjnvgIy8YPqFVdG8D5ZUoG1IespQHtgqpmtAHoAE9UALZJE2vwC7nRw0wY47W/Q7lq/3R2E3RvBaXjuZFDcxFAe2J9rWyaRNV7PBlqZWTMzqwQMArK/Rjjntjnn6jrnmjrnmgIzgX7OufQIriEiJUHVetDtD9DwVL++aAw8cwxM+nVi45JiKe4HuwGvmlno5DyVgefNbHfWBudcv4LewDmXaWY3Ax/jE81LzrmFZvYAkO6cU1lTpLSpc0L4+jej4OThcPRxiYlHisVcMYp2wQiqRXLOXXvEEUUoLS3NpaerUCFSYu3ZDD/Nhf+cl7Ot06+hzRWw7UeoUgeaX5i4+MooM5vjnMu3qr5YJYZEfOCLSClRpY6vUqp9AmxZ7LfNf9b/ZKmWCn3HQYOeiYlRwugBYxGJvYpV4NpFcPEEqFgNjukCDU7J2b8zA944GTKmJS5GyabpOEUkflr2g1t35KxvXgwfDYH1X/n1N8+AK9Ph2BMTEp54KjGISOLUOQF+OQtOfShn26YFiYtHACUGESkJuv8Z2gz2yx8Nge0rExpOWafEICIlQ+gjrEv/m7g4RIlBREqI7ndD1WP9suaMTiglBhEpGarUgRN+megoBCUGERHJRYlBRETCKDGISMmz7UeNxJpASgwiUvLMGwkLRyc6ijJLiUFESo7GZ+cs/7w0cXGUcUoMIlJyNL8wvBe0JIQSg4iUTD++D/t3JjqKMkmJQURKmGB6+I3fwOxHExtKGaXEICIlS8v+OcszH9TwGAmgxCAiJUudE+CS93LWF7+WuFjKKCUGESl5mvWBtN/75aX/hecawvR7YUdGYuMqI5QYRKTksXLh80DvXAsz/wqv94CtP6hROsaUGESkZEo9Hfq/C80vytm2cw282BKeTvG9oyUmlBhEpGQygxYXwSXvwu0H/DzRoV5oDvu2Jya2Uk6JQURKvnIV4Kq5cMchOPH2nO1v9ISD+xMXVymlxCAiycMMzvw7NDjFr29eBD9+mNiYSiElBhFJPhe/nbM8oT+81Ab2bE5cPKWMEoOIJJ+q9eC0R3LWf14Cb5wC+7YlLqZSRIlBRJJTtz/CsE1QpZ5f/3kJfDc2sTGVEkoMIpK8qtSBwdNz1if9Wm0OUaDEICLJrVYr6D06Z33tzISFUlooMYhI8mt3DaT9zi9vWQzuUGLjSXJKDCJSOlSs5l+/fwtG1oatyxMbTxJTYhCR0qHp+TnL+7bBLM0Ed7iUGESkdGjQA27eCjVb+vUDuxIbTxKLa2Iws95mtsTMlpnZXfnsv8PMFpnZN2b2mZk1iWd8IpLkjqoBpzzol5e8Cc83gx/eTWxMSShuicHMygMjgT5AW2CwmbXNddjXQJpzriMwHtC8fiISmaNq5ixvXwHv9IMPr4YDuxMWUrKJZ4mhG7DMObfcObcfGAtcHHqAc26Kcy7rrzcTSI1jfCJSGjTpBX3GwPEDcrYtegVGHA1zn0pcXEkknomhIbA6ZD0j2FaQ64F8e6qY2VAzSzez9I0bN0YxRBFJeuXKQ9uroO9bcMNyqFI3Z9+U38LOdYmLLUnEMzFYPttcvgeaXQmkAY/lt985N8o5l+acS6tXr14UQxSRUqVGMxi2EQZ/mbPtxeaqVipCPBNDBtAoZD0VWJv7IDM7F7gb6Oec2xen2ESkNGvQEzrd5Jcz98JzDTQaayHimRhmA63MrJmZVQIGARNDDzCzLsBz+KSwIY6xiUhpd+6/oPHZfnnfNvhXXVgyzg+hoZ7SYeKWGJxzmcDNwMfAYmCcc26hmT1gZv2Cwx4DqgFvmdk8M5tYwNuJiETu0g+h9S9y1t/7hZ8F7u/lYdzZ8PPSxMVWgphz+VbzJ420tDSXnp6e6DBEJJl8+yJ89TBs/SHvvkveh+YXxD+mODOzOc65tPz2qeeziJQ9Ha6H65fBLdvh/H9DnXY5+757I3FxlRBKDCJSdlVKgfZD4JpvoVswGMPiV2HuiISGlWhKDCIiZnD85TnrU26DH95LXDwJpsQgIgJwbBcYODlnfdb/JS6WBFNiEBHJ0uhMOPkBv7xuBiwck9BwEkWJQUQkixm0vTJn/aNr4AmD9bMhyZ/gjIQSg4hIqBrN4PIp4dte6wZ/LwdbliQmpjhTYhARya3RmfDbfTlzO2T5dxtfgvjsFsgsvSP2KDGIiOSnfCXocQ/ccQhOfTh837x/wurJ+Z9XCigxiIgUxgy63+UTxJUhoyx89UiprVqqkOgARESSghkce6Lv7/D9OMiY5quWAGqfANUaQMv+0PFXUL5iYmM9QioxiIhE4qTfQ0rj8G1bFsOqz2DyLfCPSr4dYuWkpH2SSSUGEZFIHJcGQ1fCwf3+w3/bcihfGT69Mfy48b1ylqvUg8w98MvZUKdNfOM9DBpdVUQkWg7sgW+eg6m3F3xMxWowZBFUb1TwMXFQ2OiqSgwiItHmHLiDfn7pfVth2Tvw5X05+49Ng2Z9oNbx/tHYlNS4h6jEICKSaO4QvHkmrPlfwcccf7lvrzh5OLS6JKbhFJYY1MYgIhIPVg7Oex6+/qdvl1jxsS9VhPp+nH+deCnU6wi128KFr/lz4xmqSgwiIgl0YBcsm+BfNy+EuU/lPabf23DcSZDSMGqXVVWSiEiy2L3Bd5x7fxDsXBu+r9FZ0O4aP8FQ0/Oh4tGHfRlN7SkikiyqHgOpp8GNq+DUXHNCrJ4CHw2BiZfB7MdjFoISg4hISVSuPHT/E9zp4Nrv/FNMDU6Bep39/t0bYnfpmL2ziIhER+3WcOkHMPgL6HCD37Z9hX8cNgaUGEREktGPH8C038fkrfW4qohIMml8NtRuAwd2w1G1YnIJJQYRkWRS5wS4dnFML6GqJBERCaPEICIiYZQYREQkjBKDiIiEUWIQEZEwSgwiIhJGiUFERMIoMYiISJikH3bbzDYCKw/z9LrApiiGkwx0z2WD7rlsOJJ7buKcq5ffjqRPDEfCzNILGo+8tNI9lw2657IhVvesqiQREQmjxCAiImHKemIYlegAEkD3XDbonsuGmNxzmW5jEBGRvMp6iUFERHJRYhARkTBlIjGYWW8zW2Jmy8zsrnz2H2Vmbwb7Z5lZ0/hHGV3FuOc7zGyRmX1jZp+ZWZNExBlNRd1zyHEDzMyZWdI/2licezazy4O/9UIzez3eMUZbMf5tNzazKWb2dfDv+4JExBktZvaSmW0wswUF7DczGxH8Pr4xs65HfFHnXKn+AcoDPwDNgUrAfKBtrmOGAc8Gy4OANxMddxzu+SygarB8U1m45+C4FGAaMBNIS3Tccfg7twK+BmoF68ckOu443PMo4KZguS2wItFxH+E9nw50BRYUsP8C4EPAgB7ArCO9ZlkoMXQDljnnljvn9gNjgYtzHXMx8HKwPB44x8wsjjFGW5H37Jyb4pzbHazOBFLjHGO0FefvDPAg8CiwN57BxUhx7vlGYKRz7mcA59yGOMcYbcW5ZwdUD5ZrAGvjGF/UOeemAVsKOeRiYIzzZgI1zaz+kVyzLCSGhsDqkPWMYFu+xzjnMoFtQJ24RBcbxbnnUNfjv3EksyLv2cy6AI2cc+/FM7AYKs7f+XjgeDObbmYzzax33KKLjeLc8/3AlWaWAXwA3BKf0BIm0v/vRapwROEkh/y++ed+Rrc4xySTYt+PmV0JpAFnxDSi2Cv0ns2sHPAkMCReAcVBcf7OFfDVSWfiS4X/M7P2zrmtMY4tVopzz4OB0c65J8ysJ/BKcM+HYh9eQkT986sslBgygEYh66nkLVpmH2NmFfDFz8KKbiVdce4ZMzsXuBvo55zbF6fYYqWoe04B2gNTzWwFvi52YpI3QBf33/YE59wB59yPwBJ8okhWxbnn64FxAM65GUBl/GBzpVWx/r9HoiwkhtlAKzNrZmaV8I3LE3MdMxG4JlgeAEx2QatOkirynoNqlefwSSHZ652hiHt2zm1zztV1zjV1zjXFt6v0c86lJybcqCjOv+138A8aYGZ18VVLy+MaZXQV555XAecAmNkJ+MSwMa5RxtdE4Org6aQewDbn3LojecNSX5XknMs0s5uBj/FPNLzknFtoZg8A6c65icCL+OLmMnxJYVDiIj5yxbznx4BqwFtBO/sq51y/hAV9hIp5z6VKMe/5Y+A8M1sEHAR+75zbnLioj0wx7/lO4Hkzux1fpTIkmb/omdkb+KrAukG7yV+AigDOuWfx7SgXAMuA3cC1R3zNJP59iYhIDJSFqiQREYmAEoOIiIRRYhARkTBKDCIiEkaJQUREwigxiJRAweivAwpaF4klJQaREGY2OvgQdmaWaWarzOwZM6uV6NhE4kWJQSSvSUB9oClwA9AX+FciAxKJJyUGkbz2OefWO+cynHOfAG8C52XtNLMaZjYqmDxlh5l9nnvMJTPrYWaTzWyXmW0LJkNqEOzrbWb/M7OfzWyLmX0cDN0gUiIoMYgUwsyaA72BA8G6Ae/jhzW+COiCn/hnctYY+GbWCZiCH6LgFPyAfePIGYLmaOAf+LkFzsQP8/5uMPaPSMKV+rGSRA5DbzPbiR+Lp3Kw7Y7g9SygM1DPObcn2HavmfUFrsJPAvQHYL5zbmjIey7OWnDO/Sf0YmZ2LbAdnyi+iPK9iERMiUEkr2nAUKAKfga0FsCIYN+JQFVgY65J/ioHx4EvRbxd0JubWQv8THLdgXr4kns5oHHU7kDkCCgxiOS12zm3LFi+1cymAPfiZwYrB/wEnJbPeduD16KmhX0XWAP8KnjNBBbh5zAWSTglBpGiDQc+NLNRwFzgWOCQc66geQ3mAmfnt8PM6gAnAL9xzk0JtnVF/xelBFHjs0gRnHNTgYXAPfhHWacDE8ysTzBhTE8zG25mWaWIx4AuwZNLncystZndYGaNgZ+BTcCNZtbSzM4AnsWXGkRKBCUGkeL5O37KyMb4SVEmA8/jp8ocB7QmmE7ROTcPOBdog58pbhZ+8qcDwbzDvwA6AguAkfhqqmSfWlVKEU3UIyIiYVRiEBGRMEoMIiISRolBRETCKDGIiEgYJQYREQmjxCAiImGUGEREJIwSg4iIhPl/60vkDUHZLjAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "model = models.model_initialize(**config)\n",
    "model.train(train, val, test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}