examples/basic_example.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "446b4a64",
   "metadata": {},
   "source": [
    "# Basic Example of Using the RevelioNN Library"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "ccb3bcca",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import sys\n",
    "import torch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "ebdfc914",
   "metadata": {},
   "outputs": [],
   "source": [
    "cur_path = os.path.dirname(os.path.abspath(\"__file__\"))\n",
    "root_path = os.path.dirname(cur_path)\n",
    "sys.path.append(root_path)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "30f7ef70",
   "metadata": {},
   "source": [
    "## Preparation of the convolution neural network model"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d474af90",
   "metadata": {},
   "source": [
    "### Initializing the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "8df0dd70",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using cache found in C:\\Users\\aaagafonov/.cache\\torch\\hub\\pytorch_vision_v0.10.0\n",
      "C:\\Users\\aaagafonov\\anaconda3\\envs\\onto-xai-final\\lib\\site-packages\\torchvision\\models\\_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.\n",
      "  warnings.warn(\n",
      "C:\\Users\\aaagafonov\\anaconda3\\envs\\onto-xai-final\\lib\\site-packages\\torchvision\\models\\_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=ResNet18_Weights.IMAGENET1K_V1`. You can also use `weights=ResNet18_Weights.DEFAULT` to get the most up-to-date weights.\n",
      "  warnings.warn(msg)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<All keys matched successfully>"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from main_net_classes.resnet18_scdb import ResNet18, NUM_CHANNELS, IMG_SIDE_SIZE, transformation\n",
    "\n",
    "main_net = ResNet18()\n",
    "main_net.load_state_dict(torch.load('SCDB_ResNet18_C1.pt'))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0c4c0d2f",
   "metadata": {},
   "source": [
    "### Converting the model to RevelioNN format"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "3aaf4b44",
   "metadata": {},
   "outputs": [],
   "source": [
    "from revelionn.utils.model import convert_to_rvl_format\n",
    "\n",
    "convert_to_rvl_format(main_net, 'SCDB_ResNet18_C1', 'C1', 'resnet18_scdb', 'ResNet18', 'transformation', IMG_SIDE_SIZE, NUM_CHANNELS)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e7492d35",
   "metadata": {},
   "source": [
    "## Training mapping networks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "fa54cf70",
   "metadata": {},
   "outputs": [],
   "source": [
    "device = torch.device('cuda')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "520d1606",
   "metadata": {},
   "source": [
    "### Initializing the MappingTrainer class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "1bf3b4ba",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using cache found in C:\\Users\\aaagafonov/.cache\\torch\\hub\\pytorch_vision_v0.10.0\n",
      "C:\\Users\\aaagafonov\\anaconda3\\envs\\onto-xai-final\\lib\\site-packages\\torchvision\\models\\_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.\n",
      "  warnings.warn(\n",
      "C:\\Users\\aaagafonov\\anaconda3\\envs\\onto-xai-final\\lib\\site-packages\\torchvision\\models\\_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=ResNet18_Weights.IMAGENET1K_V1`. You can also use `weights=ResNet18_Weights.DEFAULT` to get the most up-to-date weights.\n",
      "  warnings.warn(msg)\n"
     ]
    }
   ],
   "source": [
    "from revelionn.mapping_trainer import MappingTrainer\n",
    "\n",
    "trainer = MappingTrainer('SCDB_ResNet18_C1.rvl', os.path.join(root_path, 'main_net_classes'), ['bn'], 20, 100, os.path.join(root_path, 'trained_models', 'mapping_models'),\n",
    "                         device, 'D:\\\\Datasets\\\\scdb_custom\\\\images',\n",
    "                         'D:\\\\Datasets\\\\scdb_custom\\\\scdb_partition_corrected\\\\C1_mapping_train.csv', \n",
    "                         'D:\\\\Datasets\\\\scdb_custom\\\\scdb_partition_corrected\\\\C1_mapping_val.csv', \n",
    "                         'name', 100, 6, None)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "95e99f95",
   "metadata": {},
   "source": [
    "### Training for simultaneous mapping network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "ff1c398a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100.. Train loss: 0.713.. Valid loss: 0.693.. Train acc: 0.037.. Valid acc: 0.033.. Train AUC: 0.739.. Valid AUC: 0.747.. \n",
      "Train AUC HexStar: 0.782.. Train AUC EllStar: 0.925.. Train AUC TEStarmarker: 0.777.. Train AUC Hexagon: 0.722.. Train AUC Star: 0.691.. Train AUC Ellipse: 0.910.. Train AUC Triangle: 0.553.. Train AUC Starmarker: 0.550..  \n",
      "Val AUC HexStar: 0.806.. Val AUC EllStar: 0.908.. Val AUC TEStarmarker: 0.795.. Val AUC Hexagon: 0.784.. Val AUC Star: 0.668.. Val AUC Ellipse: 0.906.. Val AUC Triangle: 0.563.. Val AUC Starmarker: 0.543.. \n",
      "Validation loss decreased (inf --> 0.693029).  Saving model ...\n",
      "Epoch 2/100.. Train loss: 0.629.. Valid loss: 0.562.. Train acc: 0.075.. Valid acc: 0.042.. Train AUC: 0.755.. Valid AUC: 0.748.. \n",
      "Train AUC HexStar: 0.804.. Train AUC EllStar: 0.931.. Train AUC TEStarmarker: 0.812.. Train AUC Hexagon: 0.773.. Train AUC Star: 0.653.. Train AUC Ellipse: 0.937.. Train AUC Triangle: 0.574.. Train AUC Starmarker: 0.553..  \n",
      "Val AUC HexStar: 0.809.. Val AUC EllStar: 0.911.. Val AUC TEStarmarker: 0.800.. Val AUC Hexagon: 0.793.. Val AUC Star: 0.655.. Val AUC Ellipse: 0.917.. Val AUC Triangle: 0.561.. Val AUC Starmarker: 0.539.. \n",
      "Validation loss decreased (0.693029 --> 0.561806).  Saving model ...\n",
      "Epoch 3/100.. Train loss: 0.524.. Valid loss: 0.514.. Train acc: 0.151.. Valid acc: 0.194.. Train AUC: 0.749.. Valid AUC: 0.719.. \n",
      "Train AUC HexStar: 0.805.. Train AUC EllStar: 0.936.. Train AUC TEStarmarker: 0.812.. Train AUC Hexagon: 0.776.. Train AUC Star: 0.622.. Train AUC Ellipse: 0.941.. Train AUC Triangle: 0.563.. Train AUC Starmarker: 0.538..  \n",
      "Val AUC HexStar: 0.803.. Val AUC EllStar: 0.915.. Val AUC TEStarmarker: 0.802.. Val AUC Hexagon: 0.795.. Val AUC Star: 0.416.. Val AUC Ellipse: 0.923.. Val AUC Triangle: 0.569.. Val AUC Starmarker: 0.528.. \n",
      "Validation loss decreased (0.561806 --> 0.514383).  Saving model ...\n",
      "Epoch 4/100.. Train loss: 0.494.. Valid loss: 0.498.. Train acc: 0.190.. Valid acc: 0.169.. Train AUC: 0.740.. Valid AUC: 0.746.. \n",
      "Train AUC HexStar: 0.812.. Train AUC EllStar: 0.932.. Train AUC TEStarmarker: 0.804.. Train AUC Hexagon: 0.792.. Train AUC Star: 0.539.. Train AUC Ellipse: 0.935.. Train AUC Triangle: 0.575.. Train AUC Starmarker: 0.532..  \n",
      "Val AUC HexStar: 0.809.. Val AUC EllStar: 0.901.. Val AUC TEStarmarker: 0.800.. Val AUC Hexagon: 0.805.. Val AUC Star: 0.634.. Val AUC Ellipse: 0.913.. Val AUC Triangle: 0.581.. Val AUC Starmarker: 0.526.. \n",
      "Validation loss decreased (0.514383 --> 0.498054).  Saving model ...\n",
      "Epoch 5/100.. Train loss: 0.476.. Valid loss: 0.480.. Train acc: 0.186.. Valid acc: 0.168.. Train AUC: 0.762.. Valid AUC: 0.767.. \n",
      "Train AUC HexStar: 0.805.. Train AUC EllStar: 0.926.. Train AUC TEStarmarker: 0.819.. Train AUC Hexagon: 0.793.. Train AUC Star: 0.698.. Train AUC Ellipse: 0.940.. Train AUC Triangle: 0.592.. Train AUC Starmarker: 0.523..  \n",
      "Val AUC HexStar: 0.803.. Val AUC EllStar: 0.906.. Val AUC TEStarmarker: 0.821.. Val AUC Hexagon: 0.796.. Val AUC Star: 0.728.. Val AUC Ellipse: 0.919.. Val AUC Triangle: 0.615.. Val AUC Starmarker: 0.550.. \n",
      "Validation loss decreased (0.498054 --> 0.480095).  Saving model ...\n",
      "Epoch 6/100.. Train loss: 0.458.. Valid loss: 0.473.. Train acc: 0.247.. Valid acc: 0.224.. Train AUC: 0.777.. Valid AUC: 0.771.. \n",
      "Train AUC HexStar: 0.804.. Train AUC EllStar: 0.930.. Train AUC TEStarmarker: 0.836.. Train AUC Hexagon: 0.793.. Train AUC Star: 0.745.. Train AUC Ellipse: 0.944.. Train AUC Triangle: 0.613.. Train AUC Starmarker: 0.554..  \n",
      "Val AUC HexStar: 0.806.. Val AUC EllStar: 0.893.. Val AUC TEStarmarker: 0.823.. Val AUC Hexagon: 0.809.. Val AUC Star: 0.774.. Val AUC Ellipse: 0.915.. Val AUC Triangle: 0.575.. Val AUC Starmarker: 0.572.. \n",
      "Validation loss decreased (0.480095 --> 0.473090).  Saving model ...\n",
      "Epoch 7/100.. Train loss: 0.450.. Valid loss: 0.467.. Train acc: 0.247.. Valid acc: 0.227.. Train AUC: 0.783.. Valid AUC: 0.774.. \n",
      "Train AUC HexStar: 0.808.. Train AUC EllStar: 0.926.. Train AUC TEStarmarker: 0.831.. Train AUC Hexagon: 0.799.. Train AUC Star: 0.794.. Train AUC Ellipse: 0.941.. Train AUC Triangle: 0.610.. Train AUC Starmarker: 0.552..  \n",
      "Val AUC HexStar: 0.806.. Val AUC EllStar: 0.896.. Val AUC TEStarmarker: 0.826.. Val AUC Hexagon: 0.803.. Val AUC Star: 0.815.. Val AUC Ellipse: 0.915.. Val AUC Triangle: 0.573.. Val AUC Starmarker: 0.554.. \n",
      "Validation loss decreased (0.473090 --> 0.466946).  Saving model ...\n",
      "Epoch 8/100.. Train loss: 0.443.. Valid loss: 0.459.. Train acc: 0.286.. Valid acc: 0.267.. Train AUC: 0.785.. Valid AUC: 0.775.. \n",
      "Train AUC HexStar: 0.805.. Train AUC EllStar: 0.928.. Train AUC TEStarmarker: 0.828.. Train AUC Hexagon: 0.797.. Train AUC Star: 0.821.. Train AUC Ellipse: 0.941.. Train AUC Triangle: 0.588.. Train AUC Starmarker: 0.568..  \n",
      "Val AUC HexStar: 0.797.. Val AUC EllStar: 0.896.. Val AUC TEStarmarker: 0.832.. Val AUC Hexagon: 0.801.. Val AUC Star: 0.785.. Val AUC Ellipse: 0.922.. Val AUC Triangle: 0.586.. Val AUC Starmarker: 0.578.. \n",
      "Validation loss decreased (0.466946 --> 0.459363).  Saving model ...\n",
      "Epoch 9/100.. Train loss: 0.436.. Valid loss: 0.453.. Train acc: 0.263.. Valid acc: 0.225.. Train AUC: 0.799.. Valid AUC: 0.798.. \n",
      "Train AUC HexStar: 0.808.. Train AUC EllStar: 0.927.. Train AUC TEStarmarker: 0.842.. Train AUC Hexagon: 0.800.. Train AUC Star: 0.882.. Train AUC Ellipse: 0.941.. Train AUC Triangle: 0.610.. Train AUC Starmarker: 0.582..  \n",
      "Val AUC HexStar: 0.797.. Val AUC EllStar: 0.891.. Val AUC TEStarmarker: 0.845.. Val AUC Hexagon: 0.806.. Val AUC Star: 0.888.. Val AUC Ellipse: 0.922.. Val AUC Triangle: 0.619.. Val AUC Starmarker: 0.614.. \n",
      "Validation loss decreased (0.459363 --> 0.453370).  Saving model ...\n",
      "Epoch 10/100.. Train loss: 0.428.. Valid loss: 0.445.. Train acc: 0.279.. Valid acc: 0.313.. Train AUC: 0.805.. Valid AUC: 0.798.. \n",
      "Train AUC HexStar: 0.805.. Train AUC EllStar: 0.930.. Train AUC TEStarmarker: 0.844.. Train AUC Hexagon: 0.796.. Train AUC Star: 0.900.. Train AUC Ellipse: 0.945.. Train AUC Triangle: 0.627.. Train AUC Starmarker: 0.591..  \n",
      "Val AUC HexStar: 0.799.. Val AUC EllStar: 0.896.. Val AUC TEStarmarker: 0.846.. Val AUC Hexagon: 0.798.. Val AUC Star: 0.910.. Val AUC Ellipse: 0.920.. Val AUC Triangle: 0.616.. Val AUC Starmarker: 0.599.. \n",
      "Validation loss decreased (0.453370 --> 0.444586).  Saving model ...\n",
      "Epoch 11/100.. Train loss: 0.421.. Valid loss: 0.441.. Train acc: 0.316.. Valid acc: 0.324.. Train AUC: 0.808.. Valid AUC: 0.807.. \n",
      "Train AUC HexStar: 0.810.. Train AUC EllStar: 0.923.. Train AUC TEStarmarker: 0.855.. Train AUC Hexagon: 0.798.. Train AUC Star: 0.930.. Train AUC Ellipse: 0.941.. Train AUC Triangle: 0.620.. Train AUC Starmarker: 0.586..  \n",
      "Val AUC HexStar: 0.799.. Val AUC EllStar: 0.897.. Val AUC TEStarmarker: 0.870.. Val AUC Hexagon: 0.791.. Val AUC Star: 0.942.. Val AUC Ellipse: 0.929.. Val AUC Triangle: 0.617.. Val AUC Starmarker: 0.612.. \n",
      "Validation loss decreased (0.444586 --> 0.441011).  Saving model ...\n",
      "Epoch 12/100.. Train loss: 0.411.. Valid loss: 0.435.. Train acc: 0.331.. Valid acc: 0.339.. Train AUC: 0.817.. Valid AUC: 0.813.. \n",
      "Train AUC HexStar: 0.810.. Train AUC EllStar: 0.928.. Train AUC TEStarmarker: 0.874.. Train AUC Hexagon: 0.799.. Train AUC Star: 0.936.. Train AUC Ellipse: 0.949.. Train AUC Triangle: 0.641.. Train AUC Starmarker: 0.602..  \n",
      "Val AUC HexStar: 0.799.. Val AUC EllStar: 0.891.. Val AUC TEStarmarker: 0.871.. Val AUC Hexagon: 0.809.. Val AUC Star: 0.941.. Val AUC Ellipse: 0.918.. Val AUC Triangle: 0.660.. Val AUC Starmarker: 0.612.. \n",
      "Validation loss decreased (0.441011 --> 0.434932).  Saving model ...\n",
      "Epoch 13/100.. Train loss: 0.408.. Valid loss: 0.438.. Train acc: 0.356.. Valid acc: 0.299.. Train AUC: 0.827.. Valid AUC: 0.824.. \n",
      "Train AUC HexStar: 0.829.. Train AUC EllStar: 0.923.. Train AUC TEStarmarker: 0.870.. Train AUC Hexagon: 0.810.. Train AUC Star: 0.967.. Train AUC Ellipse: 0.943.. Train AUC Triangle: 0.676.. Train AUC Starmarker: 0.599..  \n",
      "Val AUC HexStar: 0.829.. Val AUC EllStar: 0.899.. Val AUC TEStarmarker: 0.879.. Val AUC Hexagon: 0.806.. Val AUC Star: 0.956.. Val AUC Ellipse: 0.913.. Val AUC Triangle: 0.687.. Val AUC Starmarker: 0.626.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 14/100.. Train loss: 0.404.. Valid loss: 0.426.. Train acc: 0.336.. Valid acc: 0.307.. Train AUC: 0.832.. Valid AUC: 0.832.. \n",
      "Train AUC HexStar: 0.843.. Train AUC EllStar: 0.926.. Train AUC TEStarmarker: 0.879.. Train AUC Hexagon: 0.813.. Train AUC Star: 0.967.. Train AUC Ellipse: 0.944.. Train AUC Triangle: 0.686.. Train AUC Starmarker: 0.600..  \n",
      "Val AUC HexStar: 0.846.. Val AUC EllStar: 0.900.. Val AUC TEStarmarker: 0.880.. Val AUC Hexagon: 0.806.. Val AUC Star: 0.980.. Val AUC Ellipse: 0.915.. Val AUC Triangle: 0.698.. Val AUC Starmarker: 0.634.. \n",
      "Validation loss decreased (0.434932 --> 0.425616).  Saving model ...\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 15/100.. Train loss: 0.396.. Valid loss: 0.419.. Train acc: 0.343.. Valid acc: 0.310.. Train AUC: 0.842.. Valid AUC: 0.839.. \n",
      "Train AUC HexStar: 0.861.. Train AUC EllStar: 0.928.. Train AUC TEStarmarker: 0.882.. Train AUC Hexagon: 0.823.. Train AUC Star: 0.976.. Train AUC Ellipse: 0.940.. Train AUC Triangle: 0.703.. Train AUC Starmarker: 0.623..  \n",
      "Val AUC HexStar: 0.849.. Val AUC EllStar: 0.913.. Val AUC TEStarmarker: 0.881.. Val AUC Hexagon: 0.806.. Val AUC Star: 0.975.. Val AUC Ellipse: 0.919.. Val AUC Triangle: 0.702.. Val AUC Starmarker: 0.665.. \n",
      "Validation loss decreased (0.425616 --> 0.419496).  Saving model ...\n",
      "Epoch 16/100.. Train loss: 0.393.. Valid loss: 0.416.. Train acc: 0.344.. Valid acc: 0.339.. Train AUC: 0.845.. Valid AUC: 0.844.. \n",
      "Train AUC HexStar: 0.859.. Train AUC EllStar: 0.935.. Train AUC TEStarmarker: 0.881.. Train AUC Hexagon: 0.818.. Train AUC Star: 0.974.. Train AUC Ellipse: 0.943.. Train AUC Triangle: 0.707.. Train AUC Starmarker: 0.646..  \n",
      "Val AUC HexStar: 0.854.. Val AUC EllStar: 0.908.. Val AUC TEStarmarker: 0.880.. Val AUC Hexagon: 0.828.. Val AUC Star: 0.984.. Val AUC Ellipse: 0.920.. Val AUC Triangle: 0.719.. Val AUC Starmarker: 0.659.. \n",
      "Validation loss decreased (0.419496 --> 0.416041).  Saving model ...\n",
      "Epoch 17/100.. Train loss: 0.394.. Valid loss: 0.410.. Train acc: 0.331.. Valid acc: 0.319.. Train AUC: 0.850.. Valid AUC: 0.843.. \n",
      "Train AUC HexStar: 0.877.. Train AUC EllStar: 0.929.. Train AUC TEStarmarker: 0.885.. Train AUC Hexagon: 0.844.. Train AUC Star: 0.978.. Train AUC Ellipse: 0.940.. Train AUC Triangle: 0.712.. Train AUC Starmarker: 0.636..  \n",
      "Val AUC HexStar: 0.841.. Val AUC EllStar: 0.907.. Val AUC TEStarmarker: 0.888.. Val AUC Hexagon: 0.823.. Val AUC Star: 0.980.. Val AUC Ellipse: 0.917.. Val AUC Triangle: 0.724.. Val AUC Starmarker: 0.664.. \n",
      "Validation loss decreased (0.416041 --> 0.410123).  Saving model ...\n",
      "Epoch 18/100.. Train loss: 0.390.. Valid loss: 0.412.. Train acc: 0.329.. Valid acc: 0.322.. Train AUC: 0.853.. Valid AUC: 0.849.. \n",
      "Train AUC HexStar: 0.874.. Train AUC EllStar: 0.936.. Train AUC TEStarmarker: 0.886.. Train AUC Hexagon: 0.840.. Train AUC Star: 0.980.. Train AUC Ellipse: 0.938.. Train AUC Triangle: 0.723.. Train AUC Starmarker: 0.645..  \n",
      "Val AUC HexStar: 0.875.. Val AUC EllStar: 0.914.. Val AUC TEStarmarker: 0.900.. Val AUC Hexagon: 0.855.. Val AUC Star: 0.976.. Val AUC Ellipse: 0.922.. Val AUC Triangle: 0.702.. Val AUC Starmarker: 0.647.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 19/100.. Train loss: 0.385.. Valid loss: 0.403.. Train acc: 0.326.. Valid acc: 0.339.. Train AUC: 0.853.. Valid AUC: 0.850.. \n",
      "Train AUC HexStar: 0.868.. Train AUC EllStar: 0.935.. Train AUC TEStarmarker: 0.890.. Train AUC Hexagon: 0.839.. Train AUC Star: 0.979.. Train AUC Ellipse: 0.940.. Train AUC Triangle: 0.721.. Train AUC Starmarker: 0.649..  \n",
      "Val AUC HexStar: 0.873.. Val AUC EllStar: 0.911.. Val AUC TEStarmarker: 0.892.. Val AUC Hexagon: 0.830.. Val AUC Star: 0.985.. Val AUC Ellipse: 0.924.. Val AUC Triangle: 0.720.. Val AUC Starmarker: 0.666.. \n",
      "Validation loss decreased (0.410123 --> 0.402930).  Saving model ...\n",
      "Epoch 20/100.. Train loss: 0.378.. Valid loss: 0.391.. Train acc: 0.320.. Valid acc: 0.336.. Train AUC: 0.858.. Valid AUC: 0.862.. \n",
      "Train AUC HexStar: 0.877.. Train AUC EllStar: 0.937.. Train AUC TEStarmarker: 0.899.. Train AUC Hexagon: 0.840.. Train AUC Star: 0.982.. Train AUC Ellipse: 0.948.. Train AUC Triangle: 0.733.. Train AUC Starmarker: 0.650..  \n",
      "Val AUC HexStar: 0.891.. Val AUC EllStar: 0.919.. Val AUC TEStarmarker: 0.902.. Val AUC Hexagon: 0.831.. Val AUC Star: 0.986.. Val AUC Ellipse: 0.928.. Val AUC Triangle: 0.750.. Val AUC Starmarker: 0.687.. \n",
      "Validation loss decreased (0.402930 --> 0.391010).  Saving model ...\n",
      "Epoch 21/100.. Train loss: 0.369.. Valid loss: 0.380.. Train acc: 0.340.. Valid acc: 0.357.. Train AUC: 0.864.. Valid AUC: 0.864.. \n",
      "Train AUC HexStar: 0.891.. Train AUC EllStar: 0.944.. Train AUC TEStarmarker: 0.898.. Train AUC Hexagon: 0.851.. Train AUC Star: 0.983.. Train AUC Ellipse: 0.952.. Train AUC Triangle: 0.732.. Train AUC Starmarker: 0.665..  \n",
      "Val AUC HexStar: 0.880.. Val AUC EllStar: 0.925.. Val AUC TEStarmarker: 0.907.. Val AUC Hexagon: 0.833.. Val AUC Star: 0.982.. Val AUC Ellipse: 0.937.. Val AUC Triangle: 0.743.. Val AUC Starmarker: 0.706.. \n",
      "Validation loss decreased (0.391010 --> 0.380082).  Saving model ...\n",
      "Epoch 22/100.. Train loss: 0.366.. Valid loss: 0.380.. Train acc: 0.341.. Valid acc: 0.365.. Train AUC: 0.863.. Valid AUC: 0.867.. \n",
      "Train AUC HexStar: 0.887.. Train AUC EllStar: 0.945.. Train AUC TEStarmarker: 0.897.. Train AUC Hexagon: 0.848.. Train AUC Star: 0.983.. Train AUC Ellipse: 0.954.. Train AUC Triangle: 0.724.. Train AUC Starmarker: 0.663..  \n",
      "Val AUC HexStar: 0.898.. Val AUC EllStar: 0.931.. Val AUC TEStarmarker: 0.902.. Val AUC Hexagon: 0.851.. Val AUC Star: 0.986.. Val AUC Ellipse: 0.938.. Val AUC Triangle: 0.737.. Val AUC Starmarker: 0.692.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 23/100.. Train loss: 0.358.. Valid loss: 0.373.. Train acc: 0.365.. Valid acc: 0.395.. Train AUC: 0.870.. Valid AUC: 0.867.. \n",
      "Train AUC HexStar: 0.889.. Train AUC EllStar: 0.943.. Train AUC TEStarmarker: 0.913.. Train AUC Hexagon: 0.848.. Train AUC Star: 0.985.. Train AUC Ellipse: 0.957.. Train AUC Triangle: 0.739.. Train AUC Starmarker: 0.683..  \n",
      "Val AUC HexStar: 0.890.. Val AUC EllStar: 0.929.. Val AUC TEStarmarker: 0.913.. Val AUC Hexagon: 0.845.. Val AUC Star: 0.987.. Val AUC Ellipse: 0.939.. Val AUC Triangle: 0.734.. Val AUC Starmarker: 0.702.. \n",
      "Validation loss decreased (0.380082 --> 0.372972).  Saving model ...\n",
      "Epoch 24/100.. Train loss: 0.352.. Valid loss: 0.364.. Train acc: 0.384.. Valid acc: 0.395.. Train AUC: 0.874.. Valid AUC: 0.876.. \n",
      "Train AUC HexStar: 0.899.. Train AUC EllStar: 0.951.. Train AUC TEStarmarker: 0.908.. Train AUC Hexagon: 0.862.. Train AUC Star: 0.987.. Train AUC Ellipse: 0.964.. Train AUC Triangle: 0.736.. Train AUC Starmarker: 0.685..  \n",
      "Val AUC HexStar: 0.910.. Val AUC EllStar: 0.939.. Val AUC TEStarmarker: 0.917.. Val AUC Hexagon: 0.850.. Val AUC Star: 0.990.. Val AUC Ellipse: 0.951.. Val AUC Triangle: 0.747.. Val AUC Starmarker: 0.704.. \n",
      "Validation loss decreased (0.372972 --> 0.364071).  Saving model ...\n",
      "Epoch 25/100.. Train loss: 0.349.. Valid loss: 0.362.. Train acc: 0.391.. Valid acc: 0.414.. Train AUC: 0.878.. Valid AUC: 0.874.. \n",
      "Train AUC HexStar: 0.906.. Train AUC EllStar: 0.950.. Train AUC TEStarmarker: 0.915.. Train AUC Hexagon: 0.868.. Train AUC Star: 0.985.. Train AUC Ellipse: 0.961.. Train AUC Triangle: 0.743.. Train AUC Starmarker: 0.699..  \n",
      "Val AUC HexStar: 0.905.. Val AUC EllStar: 0.930.. Val AUC TEStarmarker: 0.914.. Val AUC Hexagon: 0.859.. Val AUC Star: 0.992.. Val AUC Ellipse: 0.949.. Val AUC Triangle: 0.734.. Val AUC Starmarker: 0.709.. \n",
      "Validation loss decreased (0.364071 --> 0.361589).  Saving model ...\n",
      "Epoch 26/100.. Train loss: 0.348.. Valid loss: 0.362.. Train acc: 0.391.. Valid acc: 0.399.. Train AUC: 0.882.. Valid AUC: 0.879.. \n",
      "Train AUC HexStar: 0.905.. Train AUC EllStar: 0.953.. Train AUC TEStarmarker: 0.916.. Train AUC Hexagon: 0.883.. Train AUC Star: 0.986.. Train AUC Ellipse: 0.964.. Train AUC Triangle: 0.746.. Train AUC Starmarker: 0.699..  \n",
      "Val AUC HexStar: 0.899.. Val AUC EllStar: 0.940.. Val AUC TEStarmarker: 0.922.. Val AUC Hexagon: 0.854.. Val AUC Star: 0.988.. Val AUC Ellipse: 0.950.. Val AUC Triangle: 0.761.. Val AUC Starmarker: 0.718.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 27/100.. Train loss: 0.353.. Valid loss: 0.364.. Train acc: 0.373.. Valid acc: 0.381.. Train AUC: 0.875.. Valid AUC: 0.877.. \n",
      "Train AUC HexStar: 0.899.. Train AUC EllStar: 0.948.. Train AUC TEStarmarker: 0.920.. Train AUC Hexagon: 0.868.. Train AUC Star: 0.985.. Train AUC Ellipse: 0.962.. Train AUC Triangle: 0.747.. Train AUC Starmarker: 0.675..  \n",
      "Val AUC HexStar: 0.902.. Val AUC EllStar: 0.942.. Val AUC TEStarmarker: 0.922.. Val AUC Hexagon: 0.856.. Val AUC Star: 0.987.. Val AUC Ellipse: 0.953.. Val AUC Triangle: 0.752.. Val AUC Starmarker: 0.699.. \n",
      "EarlyStopping counter: 2 out of 20\n",
      "Epoch 28/100.. Train loss: 0.342.. Valid loss: 0.350.. Train acc: 0.394.. Valid acc: 0.398.. Train AUC: 0.883.. Valid AUC: 0.881.. \n",
      "Train AUC HexStar: 0.914.. Train AUC EllStar: 0.954.. Train AUC TEStarmarker: 0.918.. Train AUC Hexagon: 0.876.. Train AUC Star: 0.987.. Train AUC Ellipse: 0.966.. Train AUC Triangle: 0.751.. Train AUC Starmarker: 0.700..  \n",
      "Val AUC HexStar: 0.909.. Val AUC EllStar: 0.936.. Val AUC TEStarmarker: 0.926.. Val AUC Hexagon: 0.859.. Val AUC Star: 0.988.. Val AUC Ellipse: 0.952.. Val AUC Triangle: 0.758.. Val AUC Starmarker: 0.718.. \n",
      "Validation loss decreased (0.361589 --> 0.349915).  Saving model ...\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 29/100.. Train loss: 0.336.. Valid loss: 0.356.. Train acc: 0.412.. Valid acc: 0.408.. Train AUC: 0.885.. Valid AUC: 0.883.. \n",
      "Train AUC HexStar: 0.914.. Train AUC EllStar: 0.957.. Train AUC TEStarmarker: 0.918.. Train AUC Hexagon: 0.882.. Train AUC Star: 0.989.. Train AUC Ellipse: 0.969.. Train AUC Triangle: 0.753.. Train AUC Starmarker: 0.697..  \n",
      "Val AUC HexStar: 0.910.. Val AUC EllStar: 0.941.. Val AUC TEStarmarker: 0.928.. Val AUC Hexagon: 0.864.. Val AUC Star: 0.994.. Val AUC Ellipse: 0.954.. Val AUC Triangle: 0.761.. Val AUC Starmarker: 0.712.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 30/100.. Train loss: 0.336.. Valid loss: 0.338.. Train acc: 0.402.. Valid acc: 0.417.. Train AUC: 0.886.. Valid AUC: 0.887.. \n",
      "Train AUC HexStar: 0.910.. Train AUC EllStar: 0.957.. Train AUC TEStarmarker: 0.927.. Train AUC Hexagon: 0.879.. Train AUC Star: 0.988.. Train AUC Ellipse: 0.970.. Train AUC Triangle: 0.757.. Train AUC Starmarker: 0.699..  \n",
      "Val AUC HexStar: 0.906.. Val AUC EllStar: 0.948.. Val AUC TEStarmarker: 0.931.. Val AUC Hexagon: 0.866.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.958.. Val AUC Triangle: 0.769.. Val AUC Starmarker: 0.722.. \n",
      "Validation loss decreased (0.349915 --> 0.338452).  Saving model ...\n",
      "Epoch 31/100.. Train loss: 0.330.. Valid loss: 0.338.. Train acc: 0.414.. Valid acc: 0.406.. Train AUC: 0.886.. Valid AUC: 0.891.. \n",
      "Train AUC HexStar: 0.910.. Train AUC EllStar: 0.958.. Train AUC TEStarmarker: 0.922.. Train AUC Hexagon: 0.883.. Train AUC Star: 0.989.. Train AUC Ellipse: 0.968.. Train AUC Triangle: 0.762.. Train AUC Starmarker: 0.697..  \n",
      "Val AUC HexStar: 0.925.. Val AUC EllStar: 0.949.. Val AUC TEStarmarker: 0.932.. Val AUC Hexagon: 0.881.. Val AUC Star: 0.992.. Val AUC Ellipse: 0.963.. Val AUC Triangle: 0.772.. Val AUC Starmarker: 0.716.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 32/100.. Train loss: 0.331.. Valid loss: 0.333.. Train acc: 0.419.. Valid acc: 0.440.. Train AUC: 0.890.. Valid AUC: 0.893.. \n",
      "Train AUC HexStar: 0.914.. Train AUC EllStar: 0.959.. Train AUC TEStarmarker: 0.934.. Train AUC Hexagon: 0.893.. Train AUC Star: 0.989.. Train AUC Ellipse: 0.972.. Train AUC Triangle: 0.764.. Train AUC Starmarker: 0.698..  \n",
      "Val AUC HexStar: 0.917.. Val AUC EllStar: 0.954.. Val AUC TEStarmarker: 0.931.. Val AUC Hexagon: 0.884.. Val AUC Star: 0.991.. Val AUC Ellipse: 0.963.. Val AUC Triangle: 0.777.. Val AUC Starmarker: 0.730.. \n",
      "Validation loss decreased (0.338452 --> 0.332710).  Saving model ...\n",
      "Epoch 33/100.. Train loss: 0.325.. Valid loss: 0.334.. Train acc: 0.418.. Valid acc: 0.433.. Train AUC: 0.888.. Valid AUC: 0.890.. \n",
      "Train AUC HexStar: 0.914.. Train AUC EllStar: 0.963.. Train AUC TEStarmarker: 0.923.. Train AUC Hexagon: 0.898.. Train AUC Star: 0.991.. Train AUC Ellipse: 0.972.. Train AUC Triangle: 0.759.. Train AUC Starmarker: 0.688..  \n",
      "Val AUC HexStar: 0.917.. Val AUC EllStar: 0.954.. Val AUC TEStarmarker: 0.931.. Val AUC Hexagon: 0.887.. Val AUC Star: 0.991.. Val AUC Ellipse: 0.962.. Val AUC Triangle: 0.763.. Val AUC Starmarker: 0.712.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 34/100.. Train loss: 0.325.. Valid loss: 0.339.. Train acc: 0.414.. Valid acc: 0.421.. Train AUC: 0.890.. Valid AUC: 0.897.. \n",
      "Train AUC HexStar: 0.924.. Train AUC EllStar: 0.967.. Train AUC TEStarmarker: 0.924.. Train AUC Hexagon: 0.904.. Train AUC Star: 0.988.. Train AUC Ellipse: 0.974.. Train AUC Triangle: 0.767.. Train AUC Starmarker: 0.674..  \n",
      "Val AUC HexStar: 0.920.. Val AUC EllStar: 0.954.. Val AUC TEStarmarker: 0.940.. Val AUC Hexagon: 0.889.. Val AUC Star: 0.992.. Val AUC Ellipse: 0.965.. Val AUC Triangle: 0.783.. Val AUC Starmarker: 0.730.. \n",
      "EarlyStopping counter: 2 out of 20\n",
      "Epoch 35/100.. Train loss: 0.314.. Valid loss: 0.327.. Train acc: 0.435.. Valid acc: 0.431.. Train AUC: 0.899.. Valid AUC: 0.897.. \n",
      "Train AUC HexStar: 0.927.. Train AUC EllStar: 0.967.. Train AUC TEStarmarker: 0.933.. Train AUC Hexagon: 0.912.. Train AUC Star: 0.989.. Train AUC Ellipse: 0.978.. Train AUC Triangle: 0.776.. Train AUC Starmarker: 0.710..  \n",
      "Val AUC HexStar: 0.921.. Val AUC EllStar: 0.953.. Val AUC TEStarmarker: 0.939.. Val AUC Hexagon: 0.895.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.963.. Val AUC Triangle: 0.787.. Val AUC Starmarker: 0.723.. \n",
      "Validation loss decreased (0.332710 --> 0.327500).  Saving model ...\n",
      "Epoch 36/100.. Train loss: 0.311.. Valid loss: 0.321.. Train acc: 0.434.. Valid acc: 0.419.. Train AUC: 0.898.. Valid AUC: 0.897.. \n",
      "Train AUC HexStar: 0.928.. Train AUC EllStar: 0.968.. Train AUC TEStarmarker: 0.923.. Train AUC Hexagon: 0.916.. Train AUC Star: 0.988.. Train AUC Ellipse: 0.978.. Train AUC Triangle: 0.775.. Train AUC Starmarker: 0.704..  \n",
      "Val AUC HexStar: 0.924.. Val AUC EllStar: 0.960.. Val AUC TEStarmarker: 0.940.. Val AUC Hexagon: 0.898.. Val AUC Star: 0.992.. Val AUC Ellipse: 0.970.. Val AUC Triangle: 0.778.. Val AUC Starmarker: 0.716.. \n",
      "Validation loss decreased (0.327500 --> 0.320514).  Saving model ...\n",
      "Epoch 37/100.. Train loss: 0.307.. Valid loss: 0.313.. Train acc: 0.447.. Valid acc: 0.443.. Train AUC: 0.900.. Valid AUC: 0.907.. \n",
      "Train AUC HexStar: 0.927.. Train AUC EllStar: 0.971.. Train AUC TEStarmarker: 0.936.. Train AUC Hexagon: 0.916.. Train AUC Star: 0.990.. Train AUC Ellipse: 0.981.. Train AUC Triangle: 0.777.. Train AUC Starmarker: 0.705..  \n",
      "Val AUC HexStar: 0.938.. Val AUC EllStar: 0.963.. Val AUC TEStarmarker: 0.947.. Val AUC Hexagon: 0.913.. Val AUC Star: 0.995.. Val AUC Ellipse: 0.971.. Val AUC Triangle: 0.793.. Val AUC Starmarker: 0.733.. \n",
      "Validation loss decreased (0.320514 --> 0.313238).  Saving model ...\n",
      "Epoch 38/100.. Train loss: 0.301.. Valid loss: 0.316.. Train acc: 0.445.. Valid acc: 0.455.. Train AUC: 0.903.. Valid AUC: 0.906.. \n",
      "Train AUC HexStar: 0.928.. Train AUC EllStar: 0.972.. Train AUC TEStarmarker: 0.942.. Train AUC Hexagon: 0.922.. Train AUC Star: 0.989.. Train AUC Ellipse: 0.981.. Train AUC Triangle: 0.782.. Train AUC Starmarker: 0.707..  \n",
      "Val AUC HexStar: 0.936.. Val AUC EllStar: 0.960.. Val AUC TEStarmarker: 0.948.. Val AUC Hexagon: 0.916.. Val AUC Star: 0.990.. Val AUC Ellipse: 0.975.. Val AUC Triangle: 0.793.. Val AUC Starmarker: 0.727.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 39/100.. Train loss: 0.306.. Valid loss: 0.320.. Train acc: 0.456.. Valid acc: 0.440.. Train AUC: 0.905.. Valid AUC: 0.903.. \n",
      "Train AUC HexStar: 0.934.. Train AUC EllStar: 0.972.. Train AUC TEStarmarker: 0.938.. Train AUC Hexagon: 0.923.. Train AUC Star: 0.989.. Train AUC Ellipse: 0.980.. Train AUC Triangle: 0.791.. Train AUC Starmarker: 0.716..  \n",
      "Val AUC HexStar: 0.936.. Val AUC EllStar: 0.957.. Val AUC TEStarmarker: 0.941.. Val AUC Hexagon: 0.918.. Val AUC Star: 0.991.. Val AUC Ellipse: 0.970.. Val AUC Triangle: 0.790.. Val AUC Starmarker: 0.718.. \n",
      "EarlyStopping counter: 2 out of 20\n",
      "Epoch 40/100.. Train loss: 0.295.. Valid loss: 0.307.. Train acc: 0.467.. Valid acc: 0.450.. Train AUC: 0.908.. Valid AUC: 0.910.. \n",
      "Train AUC HexStar: 0.938.. Train AUC EllStar: 0.975.. Train AUC TEStarmarker: 0.941.. Train AUC Hexagon: 0.933.. Train AUC Star: 0.992.. Train AUC Ellipse: 0.983.. Train AUC Triangle: 0.792.. Train AUC Starmarker: 0.710..  \n",
      "Val AUC HexStar: 0.943.. Val AUC EllStar: 0.965.. Val AUC TEStarmarker: 0.944.. Val AUC Hexagon: 0.926.. Val AUC Star: 0.991.. Val AUC Ellipse: 0.972.. Val AUC Triangle: 0.799.. Val AUC Starmarker: 0.738.. \n",
      "Validation loss decreased (0.313238 --> 0.307317).  Saving model ...\n",
      "Epoch 41/100.. Train loss: 0.292.. Valid loss: 0.305.. Train acc: 0.456.. Valid acc: 0.446.. Train AUC: 0.909.. Valid AUC: 0.911.. \n",
      "Train AUC HexStar: 0.942.. Train AUC EllStar: 0.976.. Train AUC TEStarmarker: 0.940.. Train AUC Hexagon: 0.939.. Train AUC Star: 0.989.. Train AUC Ellipse: 0.982.. Train AUC Triangle: 0.787.. Train AUC Starmarker: 0.717..  \n",
      "Val AUC HexStar: 0.943.. Val AUC EllStar: 0.965.. Val AUC TEStarmarker: 0.951.. Val AUC Hexagon: 0.932.. Val AUC Star: 0.995.. Val AUC Ellipse: 0.973.. Val AUC Triangle: 0.806.. Val AUC Starmarker: 0.723.. \n",
      "Validation loss decreased (0.307317 --> 0.305402).  Saving model ...\n",
      "Epoch 42/100.. Train loss: 0.291.. Valid loss: 0.299.. Train acc: 0.473.. Valid acc: 0.453.. Train AUC: 0.910.. Valid AUC: 0.912.. \n",
      "Train AUC HexStar: 0.945.. Train AUC EllStar: 0.977.. Train AUC TEStarmarker: 0.939.. Train AUC Hexagon: 0.944.. Train AUC Star: 0.988.. Train AUC Ellipse: 0.984.. Train AUC Triangle: 0.791.. Train AUC Starmarker: 0.713..  \n",
      "Val AUC HexStar: 0.947.. Val AUC EllStar: 0.960.. Val AUC TEStarmarker: 0.950.. Val AUC Hexagon: 0.937.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.971.. Val AUC Triangle: 0.801.. Val AUC Starmarker: 0.736.. \n",
      "Validation loss decreased (0.305402 --> 0.299055).  Saving model ...\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 43/100.. Train loss: 0.290.. Valid loss: 0.301.. Train acc: 0.469.. Valid acc: 0.448.. Train AUC: 0.912.. Valid AUC: 0.917.. \n",
      "Train AUC HexStar: 0.944.. Train AUC EllStar: 0.978.. Train AUC TEStarmarker: 0.939.. Train AUC Hexagon: 0.946.. Train AUC Star: 0.990.. Train AUC Ellipse: 0.982.. Train AUC Triangle: 0.798.. Train AUC Starmarker: 0.723..  \n",
      "Val AUC HexStar: 0.945.. Val AUC EllStar: 0.966.. Val AUC TEStarmarker: 0.958.. Val AUC Hexagon: 0.935.. Val AUC Star: 0.992.. Val AUC Ellipse: 0.977.. Val AUC Triangle: 0.823.. Val AUC Starmarker: 0.740.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 44/100.. Train loss: 0.294.. Valid loss: 0.304.. Train acc: 0.479.. Valid acc: 0.478.. Train AUC: 0.911.. Valid AUC: 0.916.. \n",
      "Train AUC HexStar: 0.946.. Train AUC EllStar: 0.973.. Train AUC TEStarmarker: 0.938.. Train AUC Hexagon: 0.944.. Train AUC Star: 0.991.. Train AUC Ellipse: 0.983.. Train AUC Triangle: 0.802.. Train AUC Starmarker: 0.712..  \n",
      "Val AUC HexStar: 0.952.. Val AUC EllStar: 0.965.. Val AUC TEStarmarker: 0.950.. Val AUC Hexagon: 0.948.. Val AUC Star: 0.990.. Val AUC Ellipse: 0.977.. Val AUC Triangle: 0.809.. Val AUC Starmarker: 0.735.. \n",
      "EarlyStopping counter: 2 out of 20\n",
      "Epoch 45/100.. Train loss: 0.294.. Valid loss: 0.297.. Train acc: 0.464.. Valid acc: 0.443.. Train AUC: 0.913.. Valid AUC: 0.917.. \n",
      "Train AUC HexStar: 0.947.. Train AUC EllStar: 0.976.. Train AUC TEStarmarker: 0.946.. Train AUC Hexagon: 0.942.. Train AUC Star: 0.990.. Train AUC Ellipse: 0.984.. Train AUC Triangle: 0.805.. Train AUC Starmarker: 0.717..  \n",
      "Val AUC HexStar: 0.948.. Val AUC EllStar: 0.968.. Val AUC TEStarmarker: 0.949.. Val AUC Hexagon: 0.945.. Val AUC Star: 0.990.. Val AUC Ellipse: 0.976.. Val AUC Triangle: 0.816.. Val AUC Starmarker: 0.749.. \n",
      "Validation loss decreased (0.299055 --> 0.296782).  Saving model ...\n",
      "Epoch 46/100.. Train loss: 0.285.. Valid loss: 0.285.. Train acc: 0.475.. Valid acc: 0.473.. Train AUC: 0.914.. Valid AUC: 0.922.. \n",
      "Train AUC HexStar: 0.953.. Train AUC EllStar: 0.976.. Train AUC TEStarmarker: 0.938.. Train AUC Hexagon: 0.950.. Train AUC Star: 0.989.. Train AUC Ellipse: 0.984.. Train AUC Triangle: 0.801.. Train AUC Starmarker: 0.721..  \n",
      "Val AUC HexStar: 0.955.. Val AUC EllStar: 0.967.. Val AUC TEStarmarker: 0.954.. Val AUC Hexagon: 0.946.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.978.. Val AUC Triangle: 0.822.. Val AUC Starmarker: 0.760.. \n",
      "Validation loss decreased (0.296782 --> 0.285247).  Saving model ...\n",
      "Epoch 47/100.. Train loss: 0.275.. Valid loss: 0.292.. Train acc: 0.494.. Valid acc: 0.482.. Train AUC: 0.920.. Valid AUC: 0.921.. \n",
      "Train AUC HexStar: 0.962.. Train AUC EllStar: 0.979.. Train AUC TEStarmarker: 0.945.. Train AUC Hexagon: 0.960.. Train AUC Star: 0.990.. Train AUC Ellipse: 0.985.. Train AUC Triangle: 0.812.. Train AUC Starmarker: 0.729..  \n",
      "Val AUC HexStar: 0.960.. Val AUC EllStar: 0.966.. Val AUC TEStarmarker: 0.953.. Val AUC Hexagon: 0.955.. Val AUC Star: 0.991.. Val AUC Ellipse: 0.975.. Val AUC Triangle: 0.816.. Val AUC Starmarker: 0.753.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 48/100.. Train loss: 0.276.. Valid loss: 0.298.. Train acc: 0.501.. Valid acc: 0.438.. Train AUC: 0.918.. Valid AUC: 0.923.. \n",
      "Train AUC HexStar: 0.960.. Train AUC EllStar: 0.977.. Train AUC TEStarmarker: 0.942.. Train AUC Hexagon: 0.959.. Train AUC Star: 0.991.. Train AUC Ellipse: 0.985.. Train AUC Triangle: 0.805.. Train AUC Starmarker: 0.725..  \n",
      "Val AUC HexStar: 0.957.. Val AUC EllStar: 0.970.. Val AUC TEStarmarker: 0.958.. Val AUC Hexagon: 0.943.. Val AUC Star: 0.992.. Val AUC Ellipse: 0.979.. Val AUC Triangle: 0.817.. Val AUC Starmarker: 0.769.. \n",
      "EarlyStopping counter: 2 out of 20\n",
      "Epoch 49/100.. Train loss: 0.271.. Valid loss: 0.286.. Train acc: 0.505.. Valid acc: 0.470.. Train AUC: 0.920.. Valid AUC: 0.923.. \n",
      "Train AUC HexStar: 0.960.. Train AUC EllStar: 0.976.. Train AUC TEStarmarker: 0.945.. Train AUC Hexagon: 0.961.. Train AUC Star: 0.990.. Train AUC Ellipse: 0.984.. Train AUC Triangle: 0.818.. Train AUC Starmarker: 0.729..  \n",
      "Val AUC HexStar: 0.955.. Val AUC EllStar: 0.966.. Val AUC TEStarmarker: 0.957.. Val AUC Hexagon: 0.954.. Val AUC Star: 0.989.. Val AUC Ellipse: 0.977.. Val AUC Triangle: 0.834.. Val AUC Starmarker: 0.753.. \n",
      "EarlyStopping counter: 3 out of 20\n",
      "Epoch 50/100.. Train loss: 0.268.. Valid loss: 0.298.. Train acc: 0.489.. Valid acc: 0.472.. Train AUC: 0.922.. Valid AUC: 0.921.. \n",
      "Train AUC HexStar: 0.962.. Train AUC EllStar: 0.981.. Train AUC TEStarmarker: 0.943.. Train AUC Hexagon: 0.961.. Train AUC Star: 0.992.. Train AUC Ellipse: 0.985.. Train AUC Triangle: 0.816.. Train AUC Starmarker: 0.738..  \n",
      "Val AUC HexStar: 0.958.. Val AUC EllStar: 0.965.. Val AUC TEStarmarker: 0.946.. Val AUC Hexagon: 0.956.. Val AUC Star: 0.989.. Val AUC Ellipse: 0.976.. Val AUC Triangle: 0.815.. Val AUC Starmarker: 0.761.. \n",
      "EarlyStopping counter: 4 out of 20\n",
      "Epoch 51/100.. Train loss: 0.271.. Valid loss: 0.296.. Train acc: 0.514.. Valid acc: 0.469.. Train AUC: 0.923.. Valid AUC: 0.922.. \n",
      "Train AUC HexStar: 0.958.. Train AUC EllStar: 0.979.. Train AUC TEStarmarker: 0.948.. Train AUC Hexagon: 0.962.. Train AUC Star: 0.992.. Train AUC Ellipse: 0.986.. Train AUC Triangle: 0.816.. Train AUC Starmarker: 0.740..  \n",
      "Val AUC HexStar: 0.955.. Val AUC EllStar: 0.965.. Val AUC TEStarmarker: 0.954.. Val AUC Hexagon: 0.951.. Val AUC Star: 0.988.. Val AUC Ellipse: 0.978.. Val AUC Triangle: 0.822.. Val AUC Starmarker: 0.766.. \n",
      "EarlyStopping counter: 5 out of 20\n",
      "Epoch 52/100.. Train loss: 0.266.. Valid loss: 0.276.. Train acc: 0.509.. Valid acc: 0.489.. Train AUC: 0.926.. Valid AUC: 0.928.. \n",
      "Train AUC HexStar: 0.965.. Train AUC EllStar: 0.978.. Train AUC TEStarmarker: 0.942.. Train AUC Hexagon: 0.963.. Train AUC Star: 0.991.. Train AUC Ellipse: 0.984.. Train AUC Triangle: 0.825.. Train AUC Starmarker: 0.759..  \n",
      "Val AUC HexStar: 0.962.. Val AUC EllStar: 0.968.. Val AUC TEStarmarker: 0.957.. Val AUC Hexagon: 0.959.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.978.. Val AUC Triangle: 0.829.. Val AUC Starmarker: 0.782.. \n",
      "Validation loss decreased (0.285247 --> 0.275903).  Saving model ...\n",
      "Epoch 53/100.. Train loss: 0.255.. Valid loss: 0.287.. Train acc: 0.513.. Valid acc: 0.475.. Train AUC: 0.929.. Valid AUC: 0.931.. \n",
      "Train AUC HexStar: 0.968.. Train AUC EllStar: 0.983.. Train AUC TEStarmarker: 0.950.. Train AUC Hexagon: 0.966.. Train AUC Star: 0.992.. Train AUC Ellipse: 0.988.. Train AUC Triangle: 0.818.. Train AUC Starmarker: 0.769..  \n",
      "Val AUC HexStar: 0.963.. Val AUC EllStar: 0.963.. Val AUC TEStarmarker: 0.962.. Val AUC Hexagon: 0.967.. Val AUC Star: 0.988.. Val AUC Ellipse: 0.975.. Val AUC Triangle: 0.836.. Val AUC Starmarker: 0.799.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 54/100.. Train loss: 0.254.. Valid loss: 0.282.. Train acc: 0.525.. Valid acc: 0.494.. Train AUC: 0.929.. Valid AUC: 0.930.. \n",
      "Train AUC HexStar: 0.968.. Train AUC EllStar: 0.982.. Train AUC TEStarmarker: 0.947.. Train AUC Hexagon: 0.970.. Train AUC Star: 0.992.. Train AUC Ellipse: 0.987.. Train AUC Triangle: 0.823.. Train AUC Starmarker: 0.766..  \n",
      "Val AUC HexStar: 0.956.. Val AUC EllStar: 0.973.. Val AUC TEStarmarker: 0.957.. Val AUC Hexagon: 0.959.. Val AUC Star: 0.990.. Val AUC Ellipse: 0.981.. Val AUC Triangle: 0.838.. Val AUC Starmarker: 0.786.. \n",
      "EarlyStopping counter: 2 out of 20\n",
      "Epoch 55/100.. Train loss: 0.255.. Valid loss: 0.275.. Train acc: 0.527.. Valid acc: 0.494.. Train AUC: 0.929.. Valid AUC: 0.930.. \n",
      "Train AUC HexStar: 0.970.. Train AUC EllStar: 0.981.. Train AUC TEStarmarker: 0.942.. Train AUC Hexagon: 0.970.. Train AUC Star: 0.993.. Train AUC Ellipse: 0.987.. Train AUC Triangle: 0.822.. Train AUC Starmarker: 0.767..  \n",
      "Val AUC HexStar: 0.961.. Val AUC EllStar: 0.969.. Val AUC TEStarmarker: 0.955.. Val AUC Hexagon: 0.955.. Val AUC Star: 0.992.. Val AUC Ellipse: 0.978.. Val AUC Triangle: 0.835.. Val AUC Starmarker: 0.794.. \n",
      "Validation loss decreased (0.275903 --> 0.274623).  Saving model ...\n",
      "Epoch 56/100.. Train loss: 0.251.. Valid loss: 0.264.. Train acc: 0.529.. Valid acc: 0.507.. Train AUC: 0.931.. Valid AUC: 0.932.. \n",
      "Train AUC HexStar: 0.969.. Train AUC EllStar: 0.987.. Train AUC TEStarmarker: 0.947.. Train AUC Hexagon: 0.969.. Train AUC Star: 0.994.. Train AUC Ellipse: 0.990.. Train AUC Triangle: 0.824.. Train AUC Starmarker: 0.772..  \n",
      "Val AUC HexStar: 0.968.. Val AUC EllStar: 0.969.. Val AUC TEStarmarker: 0.957.. Val AUC Hexagon: 0.963.. Val AUC Star: 0.994.. Val AUC Ellipse: 0.980.. Val AUC Triangle: 0.830.. Val AUC Starmarker: 0.791.. \n",
      "Validation loss decreased (0.274623 --> 0.263743).  Saving model ...\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 57/100.. Train loss: 0.258.. Valid loss: 0.280.. Train acc: 0.529.. Valid acc: 0.499.. Train AUC: 0.930.. Valid AUC: 0.930.. \n",
      "Train AUC HexStar: 0.966.. Train AUC EllStar: 0.980.. Train AUC TEStarmarker: 0.947.. Train AUC Hexagon: 0.970.. Train AUC Star: 0.993.. Train AUC Ellipse: 0.986.. Train AUC Triangle: 0.827.. Train AUC Starmarker: 0.767..  \n",
      "Val AUC HexStar: 0.964.. Val AUC EllStar: 0.966.. Val AUC TEStarmarker: 0.955.. Val AUC Hexagon: 0.962.. Val AUC Star: 0.991.. Val AUC Ellipse: 0.979.. Val AUC Triangle: 0.828.. Val AUC Starmarker: 0.799.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 58/100.. Train loss: 0.251.. Valid loss: 0.268.. Train acc: 0.540.. Valid acc: 0.500.. Train AUC: 0.934.. Valid AUC: 0.933.. \n",
      "Train AUC HexStar: 0.972.. Train AUC EllStar: 0.983.. Train AUC TEStarmarker: 0.950.. Train AUC Hexagon: 0.975.. Train AUC Star: 0.992.. Train AUC Ellipse: 0.989.. Train AUC Triangle: 0.824.. Train AUC Starmarker: 0.784..  \n",
      "Val AUC HexStar: 0.967.. Val AUC EllStar: 0.973.. Val AUC TEStarmarker: 0.962.. Val AUC Hexagon: 0.960.. Val AUC Star: 0.994.. Val AUC Ellipse: 0.981.. Val AUC Triangle: 0.829.. Val AUC Starmarker: 0.798.. \n",
      "EarlyStopping counter: 2 out of 20\n",
      "Epoch 59/100.. Train loss: 0.247.. Valid loss: 0.264.. Train acc: 0.536.. Valid acc: 0.526.. Train AUC: 0.934.. Valid AUC: 0.932.. \n",
      "Train AUC HexStar: 0.974.. Train AUC EllStar: 0.983.. Train AUC TEStarmarker: 0.947.. Train AUC Hexagon: 0.976.. Train AUC Star: 0.993.. Train AUC Ellipse: 0.987.. Train AUC Triangle: 0.829.. Train AUC Starmarker: 0.786..  \n",
      "Val AUC HexStar: 0.968.. Val AUC EllStar: 0.971.. Val AUC TEStarmarker: 0.956.. Val AUC Hexagon: 0.966.. Val AUC Star: 0.994.. Val AUC Ellipse: 0.981.. Val AUC Triangle: 0.832.. Val AUC Starmarker: 0.788.. \n",
      "EarlyStopping counter: 3 out of 20\n",
      "Epoch 60/100.. Train loss: 0.248.. Valid loss: 0.278.. Train acc: 0.540.. Valid acc: 0.501.. Train AUC: 0.934.. Valid AUC: 0.931.. \n",
      "Train AUC HexStar: 0.969.. Train AUC EllStar: 0.981.. Train AUC TEStarmarker: 0.949.. Train AUC Hexagon: 0.969.. Train AUC Star: 0.992.. Train AUC Ellipse: 0.987.. Train AUC Triangle: 0.834.. Train AUC Starmarker: 0.791..  \n",
      "Val AUC HexStar: 0.967.. Val AUC EllStar: 0.963.. Val AUC TEStarmarker: 0.956.. Val AUC Hexagon: 0.971.. Val AUC Star: 0.990.. Val AUC Ellipse: 0.978.. Val AUC Triangle: 0.826.. Val AUC Starmarker: 0.800.. \n",
      "EarlyStopping counter: 4 out of 20\n",
      "Epoch 61/100.. Train loss: 0.249.. Valid loss: 0.271.. Train acc: 0.549.. Valid acc: 0.502.. Train AUC: 0.933.. Valid AUC: 0.933.. \n",
      "Train AUC HexStar: 0.972.. Train AUC EllStar: 0.982.. Train AUC TEStarmarker: 0.945.. Train AUC Hexagon: 0.973.. Train AUC Star: 0.991.. Train AUC Ellipse: 0.987.. Train AUC Triangle: 0.825.. Train AUC Starmarker: 0.789..  \n",
      "Val AUC HexStar: 0.964.. Val AUC EllStar: 0.972.. Val AUC TEStarmarker: 0.957.. Val AUC Hexagon: 0.964.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.980.. Val AUC Triangle: 0.834.. Val AUC Starmarker: 0.805.. \n",
      "EarlyStopping counter: 5 out of 20\n",
      "Epoch 62/100.. Train loss: 0.242.. Valid loss: 0.267.. Train acc: 0.543.. Valid acc: 0.529.. Train AUC: 0.938.. Valid AUC: 0.936.. \n",
      "Train AUC HexStar: 0.975.. Train AUC EllStar: 0.986.. Train AUC TEStarmarker: 0.950.. Train AUC Hexagon: 0.973.. Train AUC Star: 0.994.. Train AUC Ellipse: 0.990.. Train AUC Triangle: 0.832.. Train AUC Starmarker: 0.804..  \n",
      "Val AUC HexStar: 0.970.. Val AUC EllStar: 0.972.. Val AUC TEStarmarker: 0.958.. Val AUC Hexagon: 0.964.. Val AUC Star: 0.994.. Val AUC Ellipse: 0.980.. Val AUC Triangle: 0.837.. Val AUC Starmarker: 0.811.. \n",
      "EarlyStopping counter: 6 out of 20\n",
      "Epoch 63/100.. Train loss: 0.242.. Valid loss: 0.266.. Train acc: 0.529.. Valid acc: 0.519.. Train AUC: 0.939.. Valid AUC: 0.938.. \n",
      "Train AUC HexStar: 0.973.. Train AUC EllStar: 0.985.. Train AUC TEStarmarker: 0.953.. Train AUC Hexagon: 0.972.. Train AUC Star: 0.993.. Train AUC Ellipse: 0.989.. Train AUC Triangle: 0.840.. Train AUC Starmarker: 0.806..  \n",
      "Val AUC HexStar: 0.964.. Val AUC EllStar: 0.971.. Val AUC TEStarmarker: 0.958.. Val AUC Hexagon: 0.962.. Val AUC Star: 0.992.. Val AUC Ellipse: 0.981.. Val AUC Triangle: 0.838.. Val AUC Starmarker: 0.836.. \n",
      "EarlyStopping counter: 7 out of 20\n",
      "Epoch 64/100.. Train loss: 0.243.. Valid loss: 0.261.. Train acc: 0.530.. Valid acc: 0.518.. Train AUC: 0.938.. Valid AUC: 0.936.. \n",
      "Train AUC HexStar: 0.974.. Train AUC EllStar: 0.984.. Train AUC TEStarmarker: 0.951.. Train AUC Hexagon: 0.975.. Train AUC Star: 0.993.. Train AUC Ellipse: 0.988.. Train AUC Triangle: 0.831.. Train AUC Starmarker: 0.807..  \n",
      "Val AUC HexStar: 0.966.. Val AUC EllStar: 0.971.. Val AUC TEStarmarker: 0.953.. Val AUC Hexagon: 0.970.. Val AUC Star: 0.991.. Val AUC Ellipse: 0.979.. Val AUC Triangle: 0.837.. Val AUC Starmarker: 0.821.. \n",
      "Validation loss decreased (0.263743 --> 0.260799).  Saving model ...\n",
      "Epoch 65/100.. Train loss: 0.239.. Valid loss: 0.264.. Train acc: 0.554.. Valid acc: 0.524.. Train AUC: 0.940.. Valid AUC: 0.936.. \n",
      "Train AUC HexStar: 0.972.. Train AUC EllStar: 0.985.. Train AUC TEStarmarker: 0.954.. Train AUC Hexagon: 0.974.. Train AUC Star: 0.993.. Train AUC Ellipse: 0.989.. Train AUC Triangle: 0.841.. Train AUC Starmarker: 0.816..  \n",
      "Val AUC HexStar: 0.962.. Val AUC EllStar: 0.974.. Val AUC TEStarmarker: 0.960.. Val AUC Hexagon: 0.966.. Val AUC Star: 0.990.. Val AUC Ellipse: 0.982.. Val AUC Triangle: 0.838.. Val AUC Starmarker: 0.819.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 66/100.. Train loss: 0.228.. Valid loss: 0.256.. Train acc: 0.563.. Valid acc: 0.520.. Train AUC: 0.943.. Valid AUC: 0.939.. \n",
      "Train AUC HexStar: 0.979.. Train AUC EllStar: 0.989.. Train AUC TEStarmarker: 0.952.. Train AUC Hexagon: 0.978.. Train AUC Star: 0.994.. Train AUC Ellipse: 0.989.. Train AUC Triangle: 0.846.. Train AUC Starmarker: 0.820..  \n",
      "Val AUC HexStar: 0.967.. Val AUC EllStar: 0.974.. Val AUC TEStarmarker: 0.958.. Val AUC Hexagon: 0.968.. Val AUC Star: 0.992.. Val AUC Ellipse: 0.981.. Val AUC Triangle: 0.842.. Val AUC Starmarker: 0.828.. \n",
      "Validation loss decreased (0.260799 --> 0.255920).  Saving model ...\n",
      "Epoch 67/100.. Train loss: 0.235.. Valid loss: 0.249.. Train acc: 0.561.. Valid acc: 0.531.. Train AUC: 0.941.. Valid AUC: 0.940.. \n",
      "Train AUC HexStar: 0.974.. Train AUC EllStar: 0.985.. Train AUC TEStarmarker: 0.951.. Train AUC Hexagon: 0.975.. Train AUC Star: 0.995.. Train AUC Ellipse: 0.987.. Train AUC Triangle: 0.840.. Train AUC Starmarker: 0.819..  \n",
      "Val AUC HexStar: 0.967.. Val AUC EllStar: 0.973.. Val AUC TEStarmarker: 0.957.. Val AUC Hexagon: 0.971.. Val AUC Star: 0.991.. Val AUC Ellipse: 0.980.. Val AUC Triangle: 0.846.. Val AUC Starmarker: 0.833.. \n",
      "Validation loss decreased (0.255920 --> 0.249180).  Saving model ...\n",
      "Epoch 68/100.. Train loss: 0.227.. Valid loss: 0.261.. Train acc: 0.557.. Valid acc: 0.528.. Train AUC: 0.943.. Valid AUC: 0.940.. \n",
      "Train AUC HexStar: 0.980.. Train AUC EllStar: 0.988.. Train AUC TEStarmarker: 0.955.. Train AUC Hexagon: 0.981.. Train AUC Star: 0.993.. Train AUC Ellipse: 0.992.. Train AUC Triangle: 0.836.. Train AUC Starmarker: 0.820..  \n",
      "Val AUC HexStar: 0.968.. Val AUC EllStar: 0.976.. Val AUC TEStarmarker: 0.961.. Val AUC Hexagon: 0.968.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.982.. Val AUC Triangle: 0.838.. Val AUC Starmarker: 0.831.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 69/100.. Train loss: 0.233.. Valid loss: 0.262.. Train acc: 0.564.. Valid acc: 0.495.. Train AUC: 0.942.. Valid AUC: 0.939.. \n",
      "Train AUC HexStar: 0.974.. Train AUC EllStar: 0.984.. Train AUC TEStarmarker: 0.954.. Train AUC Hexagon: 0.973.. Train AUC Star: 0.994.. Train AUC Ellipse: 0.989.. Train AUC Triangle: 0.843.. Train AUC Starmarker: 0.825..  \n",
      "Val AUC HexStar: 0.966.. Val AUC EllStar: 0.976.. Val AUC TEStarmarker: 0.959.. Val AUC Hexagon: 0.966.. Val AUC Star: 0.994.. Val AUC Ellipse: 0.981.. Val AUC Triangle: 0.839.. Val AUC Starmarker: 0.829.. \n",
      "EarlyStopping counter: 2 out of 20\n",
      "Epoch 70/100.. Train loss: 0.233.. Valid loss: 0.245.. Train acc: 0.569.. Valid acc: 0.546.. Train AUC: 0.943.. Valid AUC: 0.943.. \n",
      "Train AUC HexStar: 0.980.. Train AUC EllStar: 0.983.. Train AUC TEStarmarker: 0.952.. Train AUC Hexagon: 0.980.. Train AUC Star: 0.994.. Train AUC Ellipse: 0.989.. Train AUC Triangle: 0.840.. Train AUC Starmarker: 0.825..  \n",
      "Val AUC HexStar: 0.973.. Val AUC EllStar: 0.973.. Val AUC TEStarmarker: 0.961.. Val AUC Hexagon: 0.973.. Val AUC Star: 0.992.. Val AUC Ellipse: 0.983.. Val AUC Triangle: 0.841.. Val AUC Starmarker: 0.844.. \n",
      "Validation loss decreased (0.249180 --> 0.244911).  Saving model ...\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 71/100.. Train loss: 0.222.. Valid loss: 0.246.. Train acc: 0.571.. Valid acc: 0.552.. Train AUC: 0.943.. Valid AUC: 0.940.. \n",
      "Train AUC HexStar: 0.982.. Train AUC EllStar: 0.986.. Train AUC TEStarmarker: 0.953.. Train AUC Hexagon: 0.980.. Train AUC Star: 0.996.. Train AUC Ellipse: 0.989.. Train AUC Triangle: 0.838.. Train AUC Starmarker: 0.824..  \n",
      "Val AUC HexStar: 0.969.. Val AUC EllStar: 0.976.. Val AUC TEStarmarker: 0.959.. Val AUC Hexagon: 0.973.. Val AUC Star: 0.991.. Val AUC Ellipse: 0.984.. Val AUC Triangle: 0.830.. Val AUC Starmarker: 0.835.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 72/100.. Train loss: 0.227.. Valid loss: 0.240.. Train acc: 0.579.. Valid acc: 0.555.. Train AUC: 0.943.. Valid AUC: 0.942.. \n",
      "Train AUC HexStar: 0.978.. Train AUC EllStar: 0.986.. Train AUC TEStarmarker: 0.954.. Train AUC Hexagon: 0.979.. Train AUC Star: 0.993.. Train AUC Ellipse: 0.991.. Train AUC Triangle: 0.836.. Train AUC Starmarker: 0.823..  \n",
      "Val AUC HexStar: 0.972.. Val AUC EllStar: 0.980.. Val AUC TEStarmarker: 0.966.. Val AUC Hexagon: 0.973.. Val AUC Star: 0.994.. Val AUC Ellipse: 0.987.. Val AUC Triangle: 0.839.. Val AUC Starmarker: 0.829.. \n",
      "Validation loss decreased (0.244911 --> 0.240111).  Saving model ...\n",
      "Epoch 73/100.. Train loss: 0.227.. Valid loss: 0.244.. Train acc: 0.539.. Valid acc: 0.520.. Train AUC: 0.941.. Valid AUC: 0.941.. \n",
      "Train AUC HexStar: 0.981.. Train AUC EllStar: 0.988.. Train AUC TEStarmarker: 0.953.. Train AUC Hexagon: 0.978.. Train AUC Star: 0.994.. Train AUC Ellipse: 0.991.. Train AUC Triangle: 0.841.. Train AUC Starmarker: 0.802..  \n",
      "Val AUC HexStar: 0.975.. Val AUC EllStar: 0.976.. Val AUC TEStarmarker: 0.958.. Val AUC Hexagon: 0.975.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.983.. Val AUC Triangle: 0.843.. Val AUC Starmarker: 0.828.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 74/100.. Train loss: 0.222.. Valid loss: 0.239.. Train acc: 0.573.. Valid acc: 0.546.. Train AUC: 0.946.. Valid AUC: 0.942.. \n",
      "Train AUC HexStar: 0.978.. Train AUC EllStar: 0.989.. Train AUC TEStarmarker: 0.957.. Train AUC Hexagon: 0.979.. Train AUC Star: 0.995.. Train AUC Ellipse: 0.991.. Train AUC Triangle: 0.850.. Train AUC Starmarker: 0.832..  \n",
      "Val AUC HexStar: 0.971.. Val AUC EllStar: 0.976.. Val AUC TEStarmarker: 0.965.. Val AUC Hexagon: 0.970.. Val AUC Star: 0.996.. Val AUC Ellipse: 0.984.. Val AUC Triangle: 0.839.. Val AUC Starmarker: 0.835.. \n",
      "Validation loss decreased (0.240111 --> 0.238847).  Saving model ...\n",
      "Epoch 75/100.. Train loss: 0.219.. Valid loss: 0.253.. Train acc: 0.577.. Valid acc: 0.546.. Train AUC: 0.946.. Valid AUC: 0.943.. \n",
      "Train AUC HexStar: 0.979.. Train AUC EllStar: 0.988.. Train AUC TEStarmarker: 0.954.. Train AUC Hexagon: 0.978.. Train AUC Star: 0.995.. Train AUC Ellipse: 0.990.. Train AUC Triangle: 0.845.. Train AUC Starmarker: 0.838..  \n",
      "Val AUC HexStar: 0.969.. Val AUC EllStar: 0.975.. Val AUC TEStarmarker: 0.961.. Val AUC Hexagon: 0.970.. Val AUC Star: 0.991.. Val AUC Ellipse: 0.984.. Val AUC Triangle: 0.844.. Val AUC Starmarker: 0.850.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 76/100.. Train loss: 0.220.. Valid loss: 0.236.. Train acc: 0.593.. Valid acc: 0.547.. Train AUC: 0.948.. Valid AUC: 0.947.. \n",
      "Train AUC HexStar: 0.981.. Train AUC EllStar: 0.989.. Train AUC TEStarmarker: 0.956.. Train AUC Hexagon: 0.982.. Train AUC Star: 0.996.. Train AUC Ellipse: 0.992.. Train AUC Triangle: 0.847.. Train AUC Starmarker: 0.839..  \n",
      "Val AUC HexStar: 0.973.. Val AUC EllStar: 0.979.. Val AUC TEStarmarker: 0.964.. Val AUC Hexagon: 0.973.. Val AUC Star: 0.995.. Val AUC Ellipse: 0.984.. Val AUC Triangle: 0.856.. Val AUC Starmarker: 0.850.. \n",
      "Validation loss decreased (0.238847 --> 0.236413).  Saving model ...\n",
      "Epoch 77/100.. Train loss: 0.215.. Valid loss: 0.239.. Train acc: 0.593.. Valid acc: 0.560.. Train AUC: 0.948.. Valid AUC: 0.946.. \n",
      "Train AUC HexStar: 0.982.. Train AUC EllStar: 0.988.. Train AUC TEStarmarker: 0.953.. Train AUC Hexagon: 0.982.. Train AUC Star: 0.996.. Train AUC Ellipse: 0.990.. Train AUC Triangle: 0.852.. Train AUC Starmarker: 0.840..  \n",
      "Val AUC HexStar: 0.973.. Val AUC EllStar: 0.978.. Val AUC TEStarmarker: 0.961.. Val AUC Hexagon: 0.973.. Val AUC Star: 0.994.. Val AUC Ellipse: 0.983.. Val AUC Triangle: 0.850.. Val AUC Starmarker: 0.858.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 78/100.. Train loss: 0.212.. Valid loss: 0.246.. Train acc: 0.584.. Valid acc: 0.543.. Train AUC: 0.950.. Valid AUC: 0.945.. \n",
      "Train AUC HexStar: 0.982.. Train AUC EllStar: 0.988.. Train AUC TEStarmarker: 0.960.. Train AUC Hexagon: 0.984.. Train AUC Star: 0.997.. Train AUC Ellipse: 0.991.. Train AUC Triangle: 0.853.. Train AUC Starmarker: 0.845..  \n",
      "Val AUC HexStar: 0.972.. Val AUC EllStar: 0.979.. Val AUC TEStarmarker: 0.962.. Val AUC Hexagon: 0.976.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.983.. Val AUC Triangle: 0.842.. Val AUC Starmarker: 0.855.. \n",
      "EarlyStopping counter: 2 out of 20\n",
      "Epoch 79/100.. Train loss: 0.221.. Valid loss: 0.253.. Train acc: 0.574.. Valid acc: 0.539.. Train AUC: 0.949.. Valid AUC: 0.945.. \n",
      "Train AUC HexStar: 0.980.. Train AUC EllStar: 0.988.. Train AUC TEStarmarker: 0.954.. Train AUC Hexagon: 0.982.. Train AUC Star: 0.994.. Train AUC Ellipse: 0.991.. Train AUC Triangle: 0.855.. Train AUC Starmarker: 0.845..  \n",
      "Val AUC HexStar: 0.974.. Val AUC EllStar: 0.970.. Val AUC TEStarmarker: 0.967.. Val AUC Hexagon: 0.975.. Val AUC Star: 0.995.. Val AUC Ellipse: 0.984.. Val AUC Triangle: 0.848.. Val AUC Starmarker: 0.845.. \n",
      "EarlyStopping counter: 3 out of 20\n",
      "Epoch 80/100.. Train loss: 0.215.. Valid loss: 0.237.. Train acc: 0.591.. Valid acc: 0.554.. Train AUC: 0.950.. Valid AUC: 0.947.. \n",
      "Train AUC HexStar: 0.982.. Train AUC EllStar: 0.989.. Train AUC TEStarmarker: 0.957.. Train AUC Hexagon: 0.981.. Train AUC Star: 0.994.. Train AUC Ellipse: 0.992.. Train AUC Triangle: 0.853.. Train AUC Starmarker: 0.849..  \n",
      "Val AUC HexStar: 0.975.. Val AUC EllStar: 0.977.. Val AUC TEStarmarker: 0.964.. Val AUC Hexagon: 0.974.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.983.. Val AUC Triangle: 0.848.. Val AUC Starmarker: 0.858.. \n",
      "EarlyStopping counter: 4 out of 20\n",
      "Epoch 81/100.. Train loss: 0.209.. Valid loss: 0.241.. Train acc: 0.603.. Valid acc: 0.534.. Train AUC: 0.951.. Valid AUC: 0.944.. \n",
      "Train AUC HexStar: 0.983.. Train AUC EllStar: 0.988.. Train AUC TEStarmarker: 0.959.. Train AUC Hexagon: 0.984.. Train AUC Star: 0.995.. Train AUC Ellipse: 0.992.. Train AUC Triangle: 0.855.. Train AUC Starmarker: 0.849..  \n",
      "Val AUC HexStar: 0.976.. Val AUC EllStar: 0.978.. Val AUC TEStarmarker: 0.959.. Val AUC Hexagon: 0.977.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.983.. Val AUC Triangle: 0.839.. Val AUC Starmarker: 0.847.. \n",
      "EarlyStopping counter: 5 out of 20\n",
      "Epoch 82/100.. Train loss: 0.215.. Valid loss: 0.235.. Train acc: 0.594.. Valid acc: 0.574.. Train AUC: 0.949.. Valid AUC: 0.947.. \n",
      "Train AUC HexStar: 0.980.. Train AUC EllStar: 0.987.. Train AUC TEStarmarker: 0.957.. Train AUC Hexagon: 0.983.. Train AUC Star: 0.996.. Train AUC Ellipse: 0.989.. Train AUC Triangle: 0.849.. Train AUC Starmarker: 0.848..  \n",
      "Val AUC HexStar: 0.973.. Val AUC EllStar: 0.978.. Val AUC TEStarmarker: 0.964.. Val AUC Hexagon: 0.974.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.985.. Val AUC Triangle: 0.843.. Val AUC Starmarker: 0.867.. \n",
      "Validation loss decreased (0.236413 --> 0.235322).  Saving model ...\n",
      "Epoch 83/100.. Train loss: 0.211.. Valid loss: 0.238.. Train acc: 0.594.. Valid acc: 0.566.. Train AUC: 0.950.. Valid AUC: 0.945.. \n",
      "Train AUC HexStar: 0.986.. Train AUC EllStar: 0.987.. Train AUC TEStarmarker: 0.960.. Train AUC Hexagon: 0.984.. Train AUC Star: 0.995.. Train AUC Ellipse: 0.990.. Train AUC Triangle: 0.848.. Train AUC Starmarker: 0.853..  \n",
      "Val AUC HexStar: 0.975.. Val AUC EllStar: 0.977.. Val AUC TEStarmarker: 0.961.. Val AUC Hexagon: 0.976.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.983.. Val AUC Triangle: 0.841.. Val AUC Starmarker: 0.854.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 84/100.. Train loss: 0.211.. Valid loss: 0.228.. Train acc: 0.585.. Valid acc: 0.573.. Train AUC: 0.949.. Valid AUC: 0.946.. \n",
      "Train AUC HexStar: 0.985.. Train AUC EllStar: 0.989.. Train AUC TEStarmarker: 0.956.. Train AUC Hexagon: 0.983.. Train AUC Star: 0.996.. Train AUC Ellipse: 0.990.. Train AUC Triangle: 0.850.. Train AUC Starmarker: 0.845..  \n",
      "Val AUC HexStar: 0.978.. Val AUC EllStar: 0.982.. Val AUC TEStarmarker: 0.964.. Val AUC Hexagon: 0.979.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.987.. Val AUC Triangle: 0.833.. Val AUC Starmarker: 0.850.. \n",
      "Validation loss decreased (0.235322 --> 0.227822).  Saving model ...\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 85/100.. Train loss: 0.204.. Valid loss: 0.238.. Train acc: 0.589.. Valid acc: 0.541.. Train AUC: 0.953.. Valid AUC: 0.944.. \n",
      "Train AUC HexStar: 0.984.. Train AUC EllStar: 0.990.. Train AUC TEStarmarker: 0.961.. Train AUC Hexagon: 0.986.. Train AUC Star: 0.997.. Train AUC Ellipse: 0.992.. Train AUC Triangle: 0.856.. Train AUC Starmarker: 0.859..  \n",
      "Val AUC HexStar: 0.973.. Val AUC EllStar: 0.974.. Val AUC TEStarmarker: 0.960.. Val AUC Hexagon: 0.976.. Val AUC Star: 0.994.. Val AUC Ellipse: 0.983.. Val AUC Triangle: 0.848.. Val AUC Starmarker: 0.843.. \n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 86/100.. Train loss: 0.223.. Valid loss: 0.264.. Train acc: 0.567.. Valid acc: 0.539.. Train AUC: 0.950.. Valid AUC: 0.945.. \n",
      "Train AUC HexStar: 0.983.. Train AUC EllStar: 0.989.. Train AUC TEStarmarker: 0.959.. Train AUC Hexagon: 0.985.. Train AUC Star: 0.995.. Train AUC Ellipse: 0.994.. Train AUC Triangle: 0.845.. Train AUC Starmarker: 0.847..  \n",
      "Val AUC HexStar: 0.977.. Val AUC EllStar: 0.969.. Val AUC TEStarmarker: 0.962.. Val AUC Hexagon: 0.976.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.981.. Val AUC Triangle: 0.847.. Val AUC Starmarker: 0.858.. \n",
      "EarlyStopping counter: 2 out of 20\n",
      "Epoch 87/100.. Train loss: 0.220.. Valid loss: 0.245.. Train acc: 0.578.. Valid acc: 0.530.. Train AUC: 0.951.. Valid AUC: 0.945.. \n",
      "Train AUC HexStar: 0.977.. Train AUC EllStar: 0.988.. Train AUC TEStarmarker: 0.961.. Train AUC Hexagon: 0.979.. Train AUC Star: 0.996.. Train AUC Ellipse: 0.991.. Train AUC Triangle: 0.848.. Train AUC Starmarker: 0.864..  \n",
      "Val AUC HexStar: 0.969.. Val AUC EllStar: 0.975.. Val AUC TEStarmarker: 0.963.. Val AUC Hexagon: 0.974.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.984.. Val AUC Triangle: 0.845.. Val AUC Starmarker: 0.856.. \n",
      "EarlyStopping counter: 3 out of 20\n",
      "Epoch 88/100.. Train loss: 0.216.. Valid loss: 0.250.. Train acc: 0.589.. Valid acc: 0.533.. Train AUC: 0.951.. Valid AUC: 0.947.. \n",
      "Train AUC HexStar: 0.982.. Train AUC EllStar: 0.987.. Train AUC TEStarmarker: 0.960.. Train AUC Hexagon: 0.982.. Train AUC Star: 0.996.. Train AUC Ellipse: 0.993.. Train AUC Triangle: 0.851.. Train AUC Starmarker: 0.857..  \n",
      "Val AUC HexStar: 0.978.. Val AUC EllStar: 0.977.. Val AUC TEStarmarker: 0.964.. Val AUC Hexagon: 0.981.. Val AUC Star: 0.994.. Val AUC Ellipse: 0.986.. Val AUC Triangle: 0.840.. Val AUC Starmarker: 0.857.. \n",
      "EarlyStopping counter: 4 out of 20\n",
      "Epoch 89/100.. Train loss: 0.209.. Valid loss: 0.236.. Train acc: 0.587.. Valid acc: 0.558.. Train AUC: 0.951.. Valid AUC: 0.944.. \n",
      "Train AUC HexStar: 0.985.. Train AUC EllStar: 0.990.. Train AUC TEStarmarker: 0.957.. Train AUC Hexagon: 0.983.. Train AUC Star: 0.996.. Train AUC Ellipse: 0.992.. Train AUC Triangle: 0.852.. Train AUC Starmarker: 0.857..  \n",
      "Val AUC HexStar: 0.974.. Val AUC EllStar: 0.977.. Val AUC TEStarmarker: 0.963.. Val AUC Hexagon: 0.975.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.985.. Val AUC Triangle: 0.846.. Val AUC Starmarker: 0.842.. \n",
      "EarlyStopping counter: 5 out of 20\n",
      "Epoch 90/100.. Train loss: 0.200.. Valid loss: 0.238.. Train acc: 0.589.. Valid acc: 0.537.. Train AUC: 0.954.. Valid AUC: 0.942.. \n",
      "Train AUC HexStar: 0.985.. Train AUC EllStar: 0.991.. Train AUC TEStarmarker: 0.963.. Train AUC Hexagon: 0.986.. Train AUC Star: 0.997.. Train AUC Ellipse: 0.991.. Train AUC Triangle: 0.863.. Train AUC Starmarker: 0.854..  \n",
      "Val AUC HexStar: 0.970.. Val AUC EllStar: 0.975.. Val AUC TEStarmarker: 0.964.. Val AUC Hexagon: 0.973.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.984.. Val AUC Triangle: 0.843.. Val AUC Starmarker: 0.834.. \n",
      "EarlyStopping counter: 6 out of 20\n",
      "Epoch 91/100.. Train loss: 0.218.. Valid loss: 0.230.. Train acc: 0.567.. Valid acc: 0.543.. Train AUC: 0.952.. Valid AUC: 0.943.. \n",
      "Train AUC HexStar: 0.985.. Train AUC EllStar: 0.988.. Train AUC TEStarmarker: 0.958.. Train AUC Hexagon: 0.983.. Train AUC Star: 0.996.. Train AUC Ellipse: 0.992.. Train AUC Triangle: 0.855.. Train AUC Starmarker: 0.855..  \n",
      "Val AUC HexStar: 0.977.. Val AUC EllStar: 0.979.. Val AUC TEStarmarker: 0.964.. Val AUC Hexagon: 0.978.. Val AUC Star: 0.994.. Val AUC Ellipse: 0.987.. Val AUC Triangle: 0.838.. Val AUC Starmarker: 0.829.. \n",
      "EarlyStopping counter: 7 out of 20\n",
      "Epoch 92/100.. Train loss: 0.205.. Valid loss: 0.242.. Train acc: 0.593.. Valid acc: 0.571.. Train AUC: 0.951.. Valid AUC: 0.947.. \n",
      "Train AUC HexStar: 0.984.. Train AUC EllStar: 0.991.. Train AUC TEStarmarker: 0.958.. Train AUC Hexagon: 0.983.. Train AUC Star: 0.996.. Train AUC Ellipse: 0.993.. Train AUC Triangle: 0.852.. Train AUC Starmarker: 0.848..  \n",
      "Val AUC HexStar: 0.977.. Val AUC EllStar: 0.973.. Val AUC TEStarmarker: 0.964.. Val AUC Hexagon: 0.981.. Val AUC Star: 0.994.. Val AUC Ellipse: 0.982.. Val AUC Triangle: 0.846.. Val AUC Starmarker: 0.857.. \n",
      "EarlyStopping counter: 8 out of 20\n",
      "Epoch 93/100.. Train loss: 0.208.. Valid loss: 0.240.. Train acc: 0.590.. Valid acc: 0.534.. Train AUC: 0.953.. Valid AUC: 0.943.. \n",
      "Train AUC HexStar: 0.983.. Train AUC EllStar: 0.987.. Train AUC TEStarmarker: 0.962.. Train AUC Hexagon: 0.985.. Train AUC Star: 0.995.. Train AUC Ellipse: 0.990.. Train AUC Triangle: 0.862.. Train AUC Starmarker: 0.861..  \n",
      "Val AUC HexStar: 0.975.. Val AUC EllStar: 0.976.. Val AUC TEStarmarker: 0.961.. Val AUC Hexagon: 0.975.. Val AUC Star: 0.995.. Val AUC Ellipse: 0.983.. Val AUC Triangle: 0.835.. Val AUC Starmarker: 0.846.. \n",
      "EarlyStopping counter: 9 out of 20\n",
      "Epoch 94/100.. Train loss: 0.200.. Valid loss: 0.233.. Train acc: 0.588.. Valid acc: 0.571.. Train AUC: 0.953.. Valid AUC: 0.948.. \n",
      "Train AUC HexStar: 0.984.. Train AUC EllStar: 0.990.. Train AUC TEStarmarker: 0.960.. Train AUC Hexagon: 0.984.. Train AUC Star: 0.997.. Train AUC Ellipse: 0.993.. Train AUC Triangle: 0.862.. Train AUC Starmarker: 0.852..  \n",
      "Val AUC HexStar: 0.978.. Val AUC EllStar: 0.973.. Val AUC TEStarmarker: 0.964.. Val AUC Hexagon: 0.979.. Val AUC Star: 0.996.. Val AUC Ellipse: 0.983.. Val AUC Triangle: 0.850.. Val AUC Starmarker: 0.859.. \n",
      "EarlyStopping counter: 10 out of 20\n",
      "Epoch 95/100.. Train loss: 0.199.. Valid loss: 0.244.. Train acc: 0.587.. Valid acc: 0.565.. Train AUC: 0.954.. Valid AUC: 0.946.. \n",
      "Train AUC HexStar: 0.985.. Train AUC EllStar: 0.990.. Train AUC TEStarmarker: 0.960.. Train AUC Hexagon: 0.984.. Train AUC Star: 0.997.. Train AUC Ellipse: 0.994.. Train AUC Triangle: 0.858.. Train AUC Starmarker: 0.864..  \n",
      "Val AUC HexStar: 0.973.. Val AUC EllStar: 0.970.. Val AUC TEStarmarker: 0.962.. Val AUC Hexagon: 0.977.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.983.. Val AUC Triangle: 0.845.. Val AUC Starmarker: 0.870.. \n",
      "EarlyStopping counter: 11 out of 20\n",
      "Epoch 96/100.. Train loss: 0.197.. Valid loss: 0.227.. Train acc: 0.595.. Valid acc: 0.584.. Train AUC: 0.954.. Valid AUC: 0.949.. \n",
      "Train AUC HexStar: 0.984.. Train AUC EllStar: 0.991.. Train AUC TEStarmarker: 0.963.. Train AUC Hexagon: 0.984.. Train AUC Star: 0.997.. Train AUC Ellipse: 0.994.. Train AUC Triangle: 0.859.. Train AUC Starmarker: 0.861..  \n",
      "Val AUC HexStar: 0.976.. Val AUC EllStar: 0.981.. Val AUC TEStarmarker: 0.965.. Val AUC Hexagon: 0.976.. Val AUC Star: 0.995.. Val AUC Ellipse: 0.987.. Val AUC Triangle: 0.850.. Val AUC Starmarker: 0.864.. \n",
      "EarlyStopping counter: 12 out of 20\n",
      "Epoch 97/100.. Train loss: 0.201.. Valid loss: 0.224.. Train acc: 0.605.. Valid acc: 0.576.. Train AUC: 0.953.. Valid AUC: 0.950.. \n",
      "Train AUC HexStar: 0.984.. Train AUC EllStar: 0.990.. Train AUC TEStarmarker: 0.960.. Train AUC Hexagon: 0.983.. Train AUC Star: 0.996.. Train AUC Ellipse: 0.992.. Train AUC Triangle: 0.863.. Train AUC Starmarker: 0.855..  \n",
      "Val AUC HexStar: 0.976.. Val AUC EllStar: 0.979.. Val AUC TEStarmarker: 0.969.. Val AUC Hexagon: 0.980.. Val AUC Star: 0.992.. Val AUC Ellipse: 0.985.. Val AUC Triangle: 0.854.. Val AUC Starmarker: 0.860.. \n",
      "Validation loss decreased (0.227822 --> 0.223806).  Saving model ...\n",
      "Epoch 98/100.. Train loss: 0.202.. Valid loss: 0.225.. Train acc: 0.608.. Valid acc: 0.578.. Train AUC: 0.955.. Valid AUC: 0.950.. \n",
      "Train AUC HexStar: 0.984.. Train AUC EllStar: 0.991.. Train AUC TEStarmarker: 0.961.. Train AUC Hexagon: 0.983.. Train AUC Star: 0.997.. Train AUC Ellipse: 0.993.. Train AUC Triangle: 0.866.. Train AUC Starmarker: 0.863..  \n",
      "Val AUC HexStar: 0.977.. Val AUC EllStar: 0.980.. Val AUC TEStarmarker: 0.966.. Val AUC Hexagon: 0.976.. Val AUC Star: 0.995.. Val AUC Ellipse: 0.986.. Val AUC Triangle: 0.848.. Val AUC Starmarker: 0.873.. \n",
      "EarlyStopping counter: 1 out of 20\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 99/100.. Train loss: 0.189.. Valid loss: 0.227.. Train acc: 0.620.. Valid acc: 0.587.. Train AUC: 0.957.. Valid AUC: 0.950.. \n",
      "Train AUC HexStar: 0.990.. Train AUC EllStar: 0.992.. Train AUC TEStarmarker: 0.964.. Train AUC Hexagon: 0.989.. Train AUC Star: 0.998.. Train AUC Ellipse: 0.993.. Train AUC Triangle: 0.864.. Train AUC Starmarker: 0.867..  \n",
      "Val AUC HexStar: 0.981.. Val AUC EllStar: 0.974.. Val AUC TEStarmarker: 0.962.. Val AUC Hexagon: 0.982.. Val AUC Star: 0.995.. Val AUC Ellipse: 0.982.. Val AUC Triangle: 0.846.. Val AUC Starmarker: 0.876.. \n",
      "EarlyStopping counter: 2 out of 20\n",
      "Epoch 100/100.. Train loss: 0.210.. Valid loss: 0.236.. Train acc: 0.578.. Valid acc: 0.533.. Train AUC: 0.954.. Valid AUC: 0.945.. \n",
      "Train AUC HexStar: 0.984.. Train AUC EllStar: 0.989.. Train AUC TEStarmarker: 0.965.. Train AUC Hexagon: 0.983.. Train AUC Star: 0.995.. Train AUC Ellipse: 0.992.. Train AUC Triangle: 0.857.. Train AUC Starmarker: 0.863..  \n",
      "Val AUC HexStar: 0.975.. Val AUC EllStar: 0.977.. Val AUC TEStarmarker: 0.961.. Val AUC Hexagon: 0.976.. Val AUC Star: 0.993.. Val AUC Ellipse: 0.986.. Val AUC Triangle: 0.835.. Val AUC Starmarker: 0.857.. \n",
      "EarlyStopping counter: 3 out of 20\n"
     ]
    }
   ],
   "source": [
    "trainer.train_simultaneous_model(['HexStar', 'EllStar', 'TEStarmarker', 'Hexagon', 'Star', 'Ellipse', 'Triangle', 'Starmarker'], \n",
    "                                 20, [160, 80, 40, 20], [20, 1])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "75e423d5",
   "metadata": {},
   "source": [
    "### Evaluating the mapping network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "e5e1cac3",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using cache found in C:\\Users\\aaagafonov/.cache\\torch\\hub\\pytorch_vision_v0.10.0\n"
     ]
    }
   ],
   "source": [
    "from revelionn.utils.model import load_mapping_model\n",
    "\n",
    "main_module, mapping_module, activation_extractor, transformation, img_size = load_mapping_model(\n",
    "    os.path.join(root_path, 'trained_models', 'mapping_models', 'C1_20_[160, 80, 40, 20]_[20, 1].rvl'), \n",
    "    cur_path, os.path.join(root_path, 'main_net_classes'), device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "b6f40fd1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test loss: 0.2134.. Test acc: 0.5975.. Test AUC: 0.9481.. \n",
      "Test concepts AUC: [0.9813480533138458, 0.9878792414908505, 0.9534641305338418, 0.9862450972327712, 0.9957535388742688, 0.9927575355754008, 0.8373931843602047, 0.8498644982142298]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "([0.9813480533138458,\n",
       "  0.9878792414908505,\n",
       "  0.9534641305338418,\n",
       "  0.9862450972327712,\n",
       "  0.9957535388742688,\n",
       "  0.9927575355754008,\n",
       "  0.8373931843602047,\n",
       "  0.8498644982142298],\n",
       " 0.9480881599494266)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from revelionn.datasets import create_dataloader\n",
    "\n",
    "test_loader = create_dataloader('D:\\\\Datasets\\\\scdb_custom\\\\scdb_partition_corrected\\\\C1_test.csv', \n",
    "                                'D:\\\\Datasets\\\\scdb_custom\\\\images', 'name',\n",
    "                                mapping_module.get_class_labels(), 100, 6, transformation)\n",
    "mapping_module.evaluate_model(test_loader)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6c0c5dd4",
   "metadata": {},
   "source": [
    "## Interpretation of the convolutional network"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c1406419",
   "metadata": {},
   "source": [
    "### Formation of logical explanations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "bd6c7e90",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "The image is classified as ['C1'].\n",
      "\n",
      "The following concepts were extracted from the image:\n",
      "['HexStar', 'EllStar', 'NotTEStarmarker', 'Hexagon', 'Star', 'Ellipse', 'NotTriangle', 'NotStarmarker']\n",
      "with the following probabilities:\n",
      "[0.9892287, 0.9961117, 0.7307813167572021, 0.9938647, 0.9998809, 0.9967223, 0.751079648733139, 0.6663315296173096]\n",
      "\n",
      "Read Observations: \n",
      "\t__input__ Type HexStar\n",
      "\t__input__ Type EllStar\n",
      "\t__input__ Type not (TEStarmarker)\n",
      "\t__input__ Type has some Hexagon\n",
      "\t__input__ Type has some Star\n",
      "\t__input__ Type has some Ellipse\n",
      "\t__input__ Type not (has some Triangle)\n",
      "\t__input__ Type not (has some Starmarker)\n",
      "\n",
      "Found Justifications:\n",
      "Justification for '__input__ Type C1':\t(Degree of Belief: 0.9966)\n",
      "\t__input__ Type has some Star\t(\"0.9998809\")\n",
      "\t__input__ Type has some Ellipse\t(\"0.9967223\")\n",
      "\t(has some Ellipse) and (has some Star) SubClassOf EllStar\n",
      "\tC1 EquivalentTo EllStar or HexStar or TEStarmarker\n",
      "\n",
      "Justification for '__input__ Type C1':\t(Degree of Belief: 0.99611)\n",
      "\t__input__ Type EllStar\t(\"0.9961117\")\n",
      "\tC1 EquivalentTo EllStar or HexStar or TEStarmarker\n",
      "\n",
      "Justification for '__input__ Type C1':\t(Degree of Belief: 0.99374)\n",
      "\t__input__ Type has some Star\t(\"0.9998809\")\n",
      "\t__input__ Type has some Hexagon\t(\"0.9938647\")\n",
      "\t(has some Hexagon) and (has some Star) SubClassOf HexStar\n",
      "\tC1 EquivalentTo EllStar or HexStar or TEStarmarker\n",
      "\n",
      "Justification for '__input__ Type C1':\t(Degree of Belief: 0.98923)\n",
      "\t__input__ Type HexStar\t(\"0.9892287\")\n",
      "\tC1 EquivalentTo EllStar or HexStar or TEStarmarker\n",
      "\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from revelionn.utils.explanation import extract_concepts_from_img, explain_target_concept\n",
    "from ontologies.scdb_ontology import concepts_map\n",
    "from PIL import Image\n",
    "\n",
    "image_path = 'D:\\\\PycharmProjects\\\\revelionn\\\\data\\\\001236.png'\n",
    "\n",
    "image = Image.open(image_path)\n",
    "main_concepts, extracted_concepts, mapping_probabilities = extract_concepts_from_img(main_module,\n",
    "                                                                                     mapping_module,\n",
    "                                                                                     image,\n",
    "                                                                                     transformation)\n",
    "print(f'\\nThe image is classified as {main_concepts}.')\n",
    "print('\\nThe following concepts were extracted from the image:')\n",
    "print(extracted_concepts)\n",
    "print('with the following probabilities:')\n",
    "print(f'{mapping_probabilities}\\n')\n",
    "    \n",
    "justifications = explain_target_concept(extracted_concepts, mapping_probabilities, concepts_map, 'C1',\n",
    "                                        os.path.join(root_path, 'ontologies', 'SCDB.owl'), \n",
    "                                        os.path.join(root_path, 'temp'))\n",
    "print(justifications)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "959ee250",
   "metadata": {},
   "source": [
    "### Formation of visual explanations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "cd33ec27",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saliency maps: (5, 256, 256)\n",
      "Class HexStar saliency map range: [-0.014413558878004551, 1.0]\n",
      "Class EllStar saliency map range: [-0.0013842829503118992, 1.0]\n",
      "Class Hexagon saliency map range: [-0.01079257857054472, 1.0]\n",
      "Class Star saliency map range: [-8.162015728885308e-05, 1.0]\n",
      "Class Ellipse saliency map range: [-0.0030527871567755938, 1.0]\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1800x600 with 11 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "from revelionn.occlusion import perform_occlusion\n",
    "\n",
    "perform_occlusion(main_module, mapping_module, activation_extractor, transformation, img_size,\n",
    "                  image_path, window_size=20, stride=5, threads=0)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "376b62ed",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.16"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}