Generalization Analysis.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "from os.path import join, curdir\n",
    "from os import listdir\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "import matplotlib"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "params = {'backend': 'ps',\n",
    "#           'text.latex.preamble': [r'\\usepackage{gensymb}'],\n",
    "          'axes.labelsize': 8, # fontsize for x and y labels (was 10)\n",
    "          'axes.titlesize': 8,\n",
    "          'font.size': 8, # was 10\n",
    "          'legend.fontsize': 8, # was 10\n",
    "          'xtick.labelsize': 8,\n",
    "          'ytick.labelsize': 8,\n",
    "          'font.family': 'serif',\n",
    "}\n",
    "\n",
    "matplotlib.rcParams.update(params)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_all_csv_files():\n",
    "    files = listdir(curdir)\n",
    "    csv_files = [file for file in files  if file.endswith('.csv') ]\n",
    "    return csv_files\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['10imbalanced_skew1.0_VGG11_A0.csv', '10imbalanced_skew1.0_VGG11_S_A0.csv', '10imbalanced_skew1.0_VGG11_XS_A0.csv', '10imbalanced_skew1.0_VGG11_XXS_A0.csv', '10imbalanced_skew1.0_VGG11_XXXS_A0.csv', '10imbalanced_skew1.0_VGG11_XXXS_A1.csv', '10imbalanced_skew1.0_VGG13_A0.csv', '10imbalanced_skew1.0_VGG13_S_A0.csv', '10imbalanced_skew1.0_VGG13_XS_A0.csv', '10imbalanced_skew1.0_VGG13_XXS_A0.csv', '10imbalanced_skew1.0_VGG13_XXXS_A0.csv', '10imbalanced_skew1.0_VGG16_A0.csv', '10imbalanced_skew1.0_VGG16_S_A0.csv', '10imbalanced_skew1.0_VGG16_XS_A0.csv', '10imbalanced_skew1.0_VGG16_XXS_A0.csv', '10imbalanced_skew1.0_VGG16_XXXS_A0.csv', '10imbalanced_skew1.0_VGG19_A0.csv', '10imbalanced_skew1.0_VGG19_S_A0.csv', '10imbalanced_skew1.0_VGG19_XS_A0.csv', '10imbalanced_skew1.0_VGG19_XXS_A0.csv', '10imbalanced_skew1.0_VGG19_XXXS_A0.csv', '10_VGG11_A0.csv', '10_VGG11_A2.csv', '10_VGG11_S_2.csv', '10_VGG11_S_A0.csv', '10_VGG11_XS_1.csv', '10_VGG11_XS_2.csv', '10_VGG11_XS_A0.csv', '10_VGG11_XXS_1.csv', '10_VGG11_XXS_A0.csv', '10_VGG11_XXXS_1.csv', '10_VGG11_XXXS_A0.csv', '10_VGG13_A0.csv', '10_VGG13_A2.csv', '10_VGG13_S_A0.csv', '10_VGG13_S_A1.csv', '10_VGG13_XS_1.csv', '10_VGG13_XS_2.csv', '10_VGG13_XS_A0.csv', '10_VGG13_XXS_1.csv', '10_VGG13_XXS_A0.csv', '10_VGG13_XXXS_1.csv', '10_VGG13_XXXS_A0.csv', '10_VGG16_A0.csv', '10_VGG16_S.csv', '10_VGG16_S_A0.csv', '10_VGG16_XS_1.csv', '10_VGG16_XS_2.csv', '10_VGG16_XS_A0.csv', '10_VGG16_XXS_1.csv', '10_VGG16_XXS_A0.csv', '10_VGG16_XXXS_1.csv', '10_VGG16_XXXS_A0.csv', '10_VGG19_A0.csv', '10_VGG19_A2.csv', '10_VGG19_S_2.csv', '10_VGG19_S_A0.csv', '10_VGG19_XS_2.csv', '10_VGG19_XS_A0.csv', '10_VGG19_XXS_1.csv', '10_VGG19_XXS_A0.csv', '10_VGG19_XXXS_1.csv', '10_VGG19_XXXS_A0.csv', 'catdog_VGG11_A0.csv', 'catdog_VGG11_A2.csv', 'catdog_VGG11_A3.csv', 'catdog_VGG11_A4.csv', 'catdog_VGG11_S_2.csv', 'catdog_VGG11_S_A0.csv', 'catdog_VGG11_S_A3.csv', 'catdog_VGG11_S_A4.csv', 'catdog_VGG11_XS_1.csv', 'catdog_VGG11_XS_A0.csv', 'catdog_VGG11_XS_A1.csv', 'catdog_VGG11_XS_A2.csv', 'catdog_VGG11_XS_A3.csv', 'catdog_VGG11_XS_A4.csv', 'catdog_VGG11_XXS_1.csv', 'catdog_VGG11_XXS_A0.csv', 'catdog_VGG11_XXS_A1.csv', 'catdog_VGG11_XXS_A3.csv', 'catdog_VGG11_XXS_A4.csv', 'catdog_VGG11_XXXS_A0.csv', 'catdog_VGG11_XXXS_A1.csv', 'catdog_VGG11_XXXS_A3.csv', 'catdog_VGG11_XXXS_A4.csv', 'catdog_VGG13_A0.csv', 'catdog_VGG13_A2.csv', 'catdog_VGG13_A3.csv', 'catdog_VGG13_A4.csv', 'catdog_VGG13_S_2.csv', 'catdog_VGG13_S_A0.csv', 'catdog_VGG13_S_A3.csv', 'catdog_VGG13_S_A4.csv', 'catdog_VGG13_XS_1.csv', 'catdog_VGG13_XS_A0.csv', 'catdog_VGG13_XS_A1.csv', 'catdog_VGG13_XS_A2.csv', 'catdog_VGG13_XS_A3.csv', 'catdog_VGG13_XS_A4.csv', 'catdog_VGG13_XXS_1.csv', 'catdog_VGG13_XXS_A0.csv', 'catdog_VGG13_XXS_A1.csv', 'catdog_VGG13_XXS_A3.csv', 'catdog_VGG13_XXS_A4.csv', 'catdog_VGG13_XXXS_A0.csv', 'catdog_VGG13_XXXS_A1.csv', 'catdog_VGG13_XXXS_A3.csv', 'catdog_VGG13_XXXS_A4.csv', 'catdog_VGG16_A0.csv', 'catdog_VGG16_A2.csv', 'catdog_VGG16_A3.csv', 'catdog_VGG16_S_A0.csv', 'catdog_VGG16_S_A2.csv', 'catdog_VGG16_S_A3.csv', 'catdog_VGG16_S_A4.csv', 'catdog_VGG16_XS_1.csv', 'catdog_VGG16_XS_A0.csv', 'catdog_VGG16_XS_A1.csv', 'catdog_VGG16_XS_A2.csv', 'catdog_VGG16_XS_A3.csv', 'catdog_VGG16_XS_A4.csv', 'catdog_VGG16_XXS_1.csv', 'catdog_VGG16_XXS_A0.csv', 'catdog_VGG16_XXS_A1.csv', 'catdog_VGG16_XXS_A3.csv', 'catdog_VGG16_XXS_A4.csv', 'catdog_VGG16_XXXS_A0.csv', 'catdog_VGG16_XXXS_A1.csv', 'catdog_VGG16_XXXS_A3.csv', 'catdog_VGG16_XXXS_A4.csv', 'catdog_VGG19_A0.csv', 'catdog_VGG19_A2.csv', 'catdog_VGG19_A3.csv', 'catdog_VGG19_A4.csv', 'catdog_VGG19_S_2.csv', 'catdog_VGG19_S_A0.csv', 'catdog_VGG19_S_A3.csv', 'catdog_VGG19_S_A4.csv', 'catdog_VGG19_XS_A0.csv', 'catdog_VGG19_XS_A2.csv', 'catdog_VGG19_XS_A3.csv', 'catdog_VGG19_XS_A4.csv', 'catdog_VGG19_XXS_1.csv', 'catdog_VGG19_XXS_A0.csv', 'catdog_VGG19_XXS_A1.csv', 'catdog_VGG19_XXS_A3.csv', 'catdog_VGG19_XXS_A4.csv', 'catdog_VGG19_XXXS_A0.csv', 'catdog_VGG19_XXXS_A1.csv', 'catdog_VGG19_XXXS_A3.csv', 'catdog_VGG19_XXXS_A4.csv', 'cd_128_VGG11VGG19VGG19_XXXS_A0.csv', 'cd_128_VGG11VGG19VGG19_XXXS_A1.csv', 'cd_128_VGG11_XXXS_A0.csv', 'cd_128_VGG11_XXXS_A1.csv', 'cd_256_VGG11VGG19VGG19_XXXS_A0.csv', 'cd_256_VGG11VGG19VGG19_XXXS_A1.csv', 'cd_256_VGG11_XXXS_A0.csv', 'cd_256_VGG11_XXXS_A1.csv', 'cd_32_VGG11VGG19VGG19_XXXS_A1.csv', 'cd_32_VGG11VGG19VGG19_XXXS_A2.csv', 'cd_32_VGG11_XXXS_A0.csv', 'cd_32_VGG11_XXXS_A1.csv', 'cd_32_VGG11_XXXS_A2.csv', 'cd_64_VGG11VGG19VGG19_XXXS_A0.csv', 'cd_64_VGG11VGG19VGG19_XXXS_A1.csv', 'cd_64_VGG11_XXXS_A0.csv', 'cd_64_VGG11_XXXS_A1.csv']\n"
     ]
    }
   ],
   "source": [
    "csv_files = get_all_csv_files()\n",
    "#csv_files = [file for file in csv_files if ('100_VGG' in file or '10_VGG' in file or 'catdog_VGG' in file) and not 'bnb' in file and not'nbn' in file]# or 'frog' in file)]\n",
    "#csv_files = [file for file in csv_files if ('100_VGG' in file or '10_VGG' in file or 'catdog_VGG' in file) and not 'bnb' in file and not'nbn' in file]# or 'frog' in file)]\n",
    "#csv_files = [file for file in csv_files if ('100_VGG' in file or '10_VGG' in file or 'catdog_VGG' in file) and not 'bnb' in file and not'nbn' in file]# or 'frog' in file)]\n",
    "csv_files = [file for file in csv_files if '_VGG' in file and not'nbn' in file and not 'BIG' in file and not '100' in file]# or 'frog' in file)]\n",
    "\n",
    "print(csv_files)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import re\n",
    "\n",
    "regex_ft = 'train_features\\d'\n",
    "regex_clf = 'train_classifier\\d'\n",
    "regex_cnv = 'train_conv\\d\\d'\n",
    "regex_fc = 'train_fc.*'\n",
    "\n",
    "\n",
    "def filter_columns(columns):\n",
    "    res = []\n",
    "    highest_clf = higher_clf_0\n",
    "    for col in columns:\n",
    "        if re.match(regex_ft, col) or re.match(regex_cnv, col) or re.match(regex_fc):\n",
    "            res.append(col)\n",
    "        elif(re.match(regex_clf, col)):\n",
    "            if int(highest-clf[-1]) < int(col[-1]):\n",
    "                highest_clf = col\n",
    "            res.append(col)\n",
    "    col = col.remove(highest_clf)\n",
    "    return col\n",
    "\n",
    "def compute_average_saturation(epoch_df):\n",
    "    cols = list(epoch_df.columns)\n",
    "    val = 0\n",
    "    #print(cols)\n",
    "    c = 0\n",
    "    for col in cols:\n",
    "        if 'eval' in col or 'accuracy' in col or 'loss' in col or 'epoch' in col or 'time_per_step' in col or 'train_classifier6' in col: \n",
    "            continue\n",
    "        print(col)\n",
    "        c += 1\n",
    "        val += epoch_df[col].values\n",
    "    return val / c\n",
    "\n",
    "\n",
    "def get_final_epoch_accuracies(files):\n",
    "    result = []\n",
    "    for csv_file in files:\n",
    "        print(csv_file)\n",
    "        res = {}\n",
    "        file = pd.DataFrame.from_csv(csv_file, sep=';')\n",
    "        try:\n",
    "            \n",
    "            divisor = 2 if 'catdog' in csv_file else 10\n",
    "            divisor = 1\n",
    "            \n",
    "            last_epoch = file[file['epoch'] == 20]\n",
    "\n",
    "            res['test_acc'] = last_epoch['test_accuracy'].values[0]\n",
    "            res['test_loss'] = last_epoch['test_loss'].values[0] / divisor\n",
    "            res['train_acc'] = last_epoch['train_accuracy'].values[0]\n",
    "            res['train_loss'] = last_epoch['train_loss'].values[0] / divisor\n",
    "            res['average_sat'] = compute_average_saturation(last_epoch)[0]\n",
    "            res['name'] = csv_file.split('.csv')[0]\n",
    "        except:\n",
    "            continue\n",
    "        result.append(res)\n",
    "    return result\n",
    "\n",
    "def get_all_epoch_data(files):\n",
    "    result = []\n",
    "    for csv_file in files:\n",
    "        print(csv_file)\n",
    "        res = {}\n",
    "        file = pd.DataFrame.from_csv(csv_file, sep=';')\n",
    "        try:\n",
    "            \n",
    "            divisor = 2 if 'catdog' in csv_file else 10\n",
    "            if '100' in csv_file:\n",
    "                divisor = 100\n",
    "            divisor = 1\n",
    "            \n",
    "            last_epoch = file\n",
    "\n",
    "            res['test_acc'] = last_epoch['test_accuracy'].values\n",
    "            res['test_loss'] = last_epoch['test_loss'].values / divisor\n",
    "            res['train_acc'] = last_epoch['train_accuracy'].values\n",
    "            res['train_loss'] = last_epoch['train_loss'].values \n",
    "            res['average_sat'] = [compute_average_saturation(last_epoch[last_epoch['epoch']==i]) for i in range(20)]\n",
    "            res['name'] = csv_file.split('.csv')\n",
    "        except:\n",
    "            continue\n",
    "        result.append(res)\n",
    "    return result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10imbalanced_skew1.0_VGG11_A0.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "10imbalanced_skew1.0_VGG11_S_A0.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "10imbalanced_skew1.0_VGG11_XS_A0.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "10imbalanced_skew1.0_VGG11_XXS_A0.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "10imbalanced_skew1.0_VGG11_XXXS_A0.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "10imbalanced_skew1.0_VGG11_XXXS_A1.csv\n",
      "10imbalanced_skew1.0_VGG13_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "10imbalanced_skew1.0_VGG13_S_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "10imbalanced_skew1.0_VGG13_XS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "10imbalanced_skew1.0_VGG13_XXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "10imbalanced_skew1.0_VGG13_XXXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "10imbalanced_skew1.0_VGG16_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "10imbalanced_skew1.0_VGG16_S_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "10imbalanced_skew1.0_VGG16_XS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "10imbalanced_skew1.0_VGG16_XXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "10imbalanced_skew1.0_VGG16_XXXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "10imbalanced_skew1.0_VGG19_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "10imbalanced_skew1.0_VGG19_S_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "10imbalanced_skew1.0_VGG19_XS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "10imbalanced_skew1.0_VGG19_XXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "10imbalanced_skew1.0_VGG19_XXXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "10_VGG11_A0.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "10_VGG11_A2.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "10_VGG11_S_2.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "10_VGG11_S_A0.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "10_VGG11_XS_1.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "10_VGG11_XS_2.csv\n",
      "10_VGG11_XS_A0.csv\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\ipykernel_launcher.py:41: FutureWarning: from_csv is deprecated. Please use read_csv(...) instead. Note that some of the default arguments are different, so please refer to the documentation for from_csv when changing your function calls\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "10_VGG11_XXS_1.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "10_VGG11_XXS_A0.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "10_VGG11_XXXS_1.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "10_VGG11_XXXS_A0.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "10_VGG13_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "10_VGG13_A2.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "10_VGG13_S_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "10_VGG13_S_A1.csv\n",
      "10_VGG13_XS_1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "10_VGG13_XS_2.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "10_VGG13_XS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "10_VGG13_XXS_1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "10_VGG13_XXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "10_VGG13_XXXS_1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "10_VGG13_XXXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "10_VGG16_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "10_VGG16_S.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "10_VGG16_S_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "10_VGG16_XS_1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "10_VGG16_XS_2.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "10_VGG16_XS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "10_VGG16_XXS_1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "10_VGG16_XXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "10_VGG16_XXXS_1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "10_VGG16_XXXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "10_VGG19_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "10_VGG19_A2.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "10_VGG19_S_2.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "10_VGG19_S_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "10_VGG19_XS_2.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "10_VGG19_XS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "10_VGG19_XXS_1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "10_VGG19_XXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "10_VGG19_XXXS_1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "10_VGG19_XXXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG11_A0.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_A2.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG11_A3.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_A4.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_S_2.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_S_A0.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_S_A3.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_S_A4.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_XS_1.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_XS_A0.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_XS_A1.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_XS_A2.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_XS_A3.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_XS_A4.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_XXS_1.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_XXS_A0.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_XXS_A1.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_XXS_A3.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_XXS_A4.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_XXXS_A0.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_XXXS_A1.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_XXXS_A3.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG11_XXXS_A4.csv\n",
      "train_features0\n",
      "train_features4\n",
      "train_features8\n",
      "train_features11\n",
      "train_features15\n",
      "train_features18\n",
      "train_features22\n",
      "train_features25\n",
      "train_classifier2\n",
      "catdog_VGG13_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_A2.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG13_A3.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_A4.csv\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_S_2.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_S_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_S_A3.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_S_A4.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_XS_1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_XS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_XS_A1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_XS_A2.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_XS_A3.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_XS_A4.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_XXS_1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_XXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_XXS_A1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_XXS_A3.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_XXS_A4.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_XXXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_XXXS_A1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_XXXS_A3.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG13_XXXS_A4.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features21\n",
      "train_features24\n",
      "train_features28\n",
      "train_features31\n",
      "train_classifier2\n",
      "catdog_VGG16_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_A2.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_A3.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_S_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_S_A2.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_S_A3.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_S_A4.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_XS_1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_XS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_XS_A1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_XS_A2.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_XS_A3.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_XS_A4.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_XXS_1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_XXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_XXS_A1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_XXS_A3.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_XXS_A4.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_XXXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_XXXS_A1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_XXXS_A3.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG16_XXXS_A4.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features24\n",
      "train_features27\n",
      "train_features30\n",
      "train_features34\n",
      "train_features37\n",
      "train_features40\n",
      "train_classifier2\n",
      "catdog_VGG19_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_A2.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_A3.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_A4.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_S_2.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_S_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_S_A3.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_S_A4.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_XS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_XS_A2.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_XS_A3.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_XS_A4.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_XXS_1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_XXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_XXS_A1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_XXS_A3.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_XXS_A4.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_XXXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_XXXS_A1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_XXXS_A3.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "catdog_VGG19_XXXS_A4.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "cd_128_VGG11VGG19VGG19_XXXS_A0.csv\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "cd_128_VGG11VGG19VGG19_XXXS_A1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "cd_128_VGG11_XXXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "cd_128_VGG11_XXXS_A1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "cd_256_VGG11VGG19VGG19_XXXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "cd_256_VGG11VGG19VGG19_XXXS_A1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "cd_256_VGG11_XXXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "cd_256_VGG11_XXXS_A1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "cd_32_VGG11VGG19VGG19_XXXS_A1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "cd_32_VGG11VGG19VGG19_XXXS_A2.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "cd_32_VGG11_XXXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "cd_32_VGG11_XXXS_A1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "cd_32_VGG11_XXXS_A2.csv\n",
      "cd_64_VGG11VGG19VGG19_XXXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "cd_64_VGG11VGG19VGG19_XXXS_A1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "cd_64_VGG11_XXXS_A0.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n",
      "cd_64_VGG11_XXXS_A1.csv\n",
      "train_features0\n",
      "train_features3\n",
      "train_features7\n",
      "train_features10\n",
      "train_features14\n",
      "train_features17\n",
      "train_features20\n",
      "train_features23\n",
      "train_features27\n",
      "train_features30\n",
      "train_features33\n",
      "train_features36\n",
      "train_features40\n",
      "train_features43\n",
      "train_features46\n",
      "train_features49\n",
      "train_classifier2\n"
     ]
    }
   ],
   "source": [
    "result = get_final_epoch_accuracies(csv_files)\n",
    "#all_epoch_results = get_all_epoch_data(csv_files) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\"\\nfor res in all_epoch_results:\\n    try:\\n    #if True:\\n        plt.figure(figsize=(10,5))\\n        plt.title(res['name'][0])\\n        plt.plot(list(range(20)),res['test_acc'][1:], label='acc')\\n        plt.show()\\n        plt.figure(figsize=(10,5))\\n        plt.plot(list(range(20)),np.array(res['average_sat']), label='sat')\\n        plt.legend()\\n        \\n        plt.show()\\n    except:\\n        continue\\n\""
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "\"\"\"\n",
    "for res in all_epoch_results:\n",
    "    try:\n",
    "    #if True:\n",
    "        plt.figure(figsize=(10,5))\n",
    "        plt.title(res['name'][0])\n",
    "        plt.plot(list(range(20)),res['test_acc'][1:], label='acc')\n",
    "        plt.show()\n",
    "        plt.figure(figsize=(10,5))\n",
    "        plt.plot(list(range(20)),np.array(res['average_sat']), label='sat')\n",
    "        plt.legend()\n",
    "        \n",
    "        plt.show()\n",
    "    except:\n",
    "        continue\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[{'test_acc': 0.8665, 'test_loss': 0.003320103514194489, 'train_acc': 0.93314, 'train_loss': 0.0015503320200741294, 'average_sat': 28.14666666666667, 'name': '10imbalanced_skew1.0_VGG11_A0'}, {'test_acc': 0.8357, 'test_loss': 0.00403153568506241, 'train_acc': 0.90066, 'train_loss': 0.0022582134519517428, 'average_sat': 42.623333333333335, 'name': '10imbalanced_skew1.0_VGG11_S_A0'}, {'test_acc': 0.8083, 'test_loss': 0.00454896314740181, 'train_acc': 0.8475600000000001, 'train_loss': 0.0035090531387925146, 'average_sat': 54.513333333333335, 'name': '10imbalanced_skew1.0_VGG11_XS_A0'}, {'test_acc': 0.7298, 'test_loss': 0.0062475737214088434, 'train_acc': 0.7648199999999999, 'train_loss': 0.005433967106342316, 'average_sat': 63.88666666666667, 'name': '10imbalanced_skew1.0_VGG11_XXS_A0'}, {'test_acc': 0.6164, 'test_loss': 0.008456563538312912, 'train_acc': 0.6339199999999999, 'train_loss': 0.008354220584630966, 'average_sat': 66.66666666666667, 'name': '10imbalanced_skew1.0_VGG11_XXXS_A0'}, {'test_acc': 0.887, 'test_loss': 0.0028485461115837093, 'train_acc': 0.94626, 'train_loss': 0.001220825977176428, 'average_sat': 34.94363636363636, 'name': '10imbalanced_skew1.0_VGG13_A0'}, {'test_acc': 0.8572, 'test_loss': 0.003459156645834446, 'train_acc': 0.9192799999999999, 'train_loss': 0.00186619205608964, 'average_sat': 48.615454545454554, 'name': '10imbalanced_skew1.0_VGG13_S_A0'}, {'test_acc': 0.8327, 'test_loss': 0.0039048209860920906, 'train_acc': 0.86736, 'train_loss': 0.003103066331148148, 'average_sat': 58.23909090909092, 'name': '10imbalanced_skew1.0_VGG13_XS_A0'}, {'test_acc': 0.7662, 'test_loss': 0.005363530206680298, 'train_acc': 0.7879, 'train_loss': 0.0049162977540493005, 'average_sat': 64.77181818181818, 'name': '10imbalanced_skew1.0_VGG13_XXS_A0'}, {'test_acc': 0.6223, 'test_loss': 0.008262405389547348, 'train_acc': 0.64354, 'train_loss': 0.00814807932138443, 'average_sat': 68.46454545454546, 'name': '10imbalanced_skew1.0_VGG13_XXXS_A0'}, {'test_acc': 0.8797, 'test_loss': 0.0032080261424183845, 'train_acc': 0.93906, 'train_loss': 0.0014731433972716332, 'average_sat': 28.139285714285712, 'name': '10imbalanced_skew1.0_VGG16_A0'}, {'test_acc': 0.8618, 'test_loss': 0.0033706476002931594, 'train_acc': 0.9129999999999999, 'train_loss': 0.002046670450717211, 'average_sat': 41.63142857142857, 'name': '10imbalanced_skew1.0_VGG16_S_A0'}, {'test_acc': 0.8075, 'test_loss': 0.0046469519853591925, 'train_acc': 0.86056, 'train_loss': 0.003245978153347969, 'average_sat': 52.40000000000001, 'name': '10imbalanced_skew1.0_VGG16_XS_A0'}, {'test_acc': 0.7559999999999999, 'test_loss': 0.005719660440087319, 'train_acc': 0.77894, 'train_loss': 0.005150727593898773, 'average_sat': 62.94642857142857, 'name': '10imbalanced_skew1.0_VGG16_XXS_A0'}, {'test_acc': 0.639, 'test_loss': 0.008057727736234664, 'train_acc': 0.6683, 'train_loss': 0.007674482513666153, 'average_sat': 67.40928571428572, 'name': '10imbalanced_skew1.0_VGG16_XXXS_A0'}, {'test_acc': 0.8812, 'test_loss': 0.003122223375737667, 'train_acc': 0.9290200000000001, 'train_loss': 0.0017646702511608602, 'average_sat': 23.10411764705883, 'name': '10imbalanced_skew1.0_VGG19_A0'}, {'test_acc': 0.8598, 'test_loss': 0.003642730489373207, 'train_acc': 0.90038, 'train_loss': 0.002499729889035225, 'average_sat': 33.27176470588234, 'name': '10imbalanced_skew1.0_VGG19_S_A0'}, {'test_acc': 0.8154, 'test_loss': 0.004577692070603371, 'train_acc': 0.8554, 'train_loss': 0.0034510183018445968, 'average_sat': 45.542352941176475, 'name': '10imbalanced_skew1.0_VGG19_XS_A0'}, {'test_acc': 0.7429, 'test_loss': 0.006021276444196701, 'train_acc': 0.7676, 'train_loss': 0.005412630167007446, 'average_sat': 55.97411764705882, 'name': '10imbalanced_skew1.0_VGG19_XXS_A0'}, {'test_acc': 0.6095, 'test_loss': 0.008564063137769699, 'train_acc': 0.6351, 'train_loss': 0.00831241609096527, 'average_sat': 62.50058823529412, 'name': '10imbalanced_skew1.0_VGG19_XXXS_A0'}, {'test_acc': 0.8804, 'test_loss': 0.0030059443414211275, 'train_acc': 0.93022, 'train_loss': 0.0015811696387827395, 'average_sat': 28.386666666666663, 'name': '10_VGG11_A0'}, {'test_acc': 0.8618, 'test_loss': 0.003449261014163494, 'train_acc': 0.9138799999999999, 'train_loss': 0.001951062341183424, 'average_sat': 48.29545454545455, 'name': '10_VGG11_A2'}, {'test_acc': 0.8595, 'test_loss': 0.003415983256697655, 'train_acc': 0.8965200000000001, 'train_loss': 0.0023666250401735305, 'average_sat': 42.446666666666665, 'name': '10_VGG11_S_2'}, {'test_acc': 0.8565, 'test_loss': 0.0035161548644304275, 'train_acc': 0.89828, 'train_loss': 0.0023453720557689672, 'average_sat': 40.842222222222226, 'name': '10_VGG11_S_A0'}, {'test_acc': 0.8078, 'test_loss': 0.004392937165498733, 'train_acc': 0.83868, 'train_loss': 0.0036893108302354814, 'average_sat': 55.20888888888888, 'name': '10_VGG11_XS_1'}, {'test_acc': 0.8258, 'test_loss': 0.004066813966631889, 'train_acc': 0.8418200000000001, 'train_loss': 0.003629473238587379, 'average_sat': 55.20666666666667, 'name': '10_VGG11_XS_A0'}, {'test_acc': 0.7553, 'test_loss': 0.005663011080026628, 'train_acc': 0.7554, 'train_loss': 0.005662724031805992, 'average_sat': 65.79777777777778, 'name': '10_VGG11_XXS_1'}, {'test_acc': 0.754, 'test_loss': 0.005557730048894882, 'train_acc': 0.75568, 'train_loss': 0.0056240694564580924, 'average_sat': 63.88777777777778, 'name': '10_VGG11_XXS_A0'}, {'test_acc': 0.6281, 'test_loss': 0.008323376202583313, 'train_acc': 0.6027600000000001, 'train_loss': 0.00898552043914795, 'average_sat': 71.17888888888889, 'name': '10_VGG11_XXXS_1'}, {'test_acc': 0.6288, 'test_loss': 0.00822241576910019, 'train_acc': 0.60402, 'train_loss': 0.008976143237352371, 'average_sat': 67.35888888888888, 'name': '10_VGG11_XXXS_A0'}, {'test_acc': 0.8879, 'test_loss': 0.0029417147070169448, 'train_acc': 0.9452, 'train_loss': 0.0012194141113013026, 'average_sat': 34.693636363636365, 'name': '10_VGG13_A0'}, {'test_acc': 0.8667, 'test_loss': 0.00335091445595026, 'train_acc': 0.9068799999999999, 'train_loss': 0.0022017125114798548, 'average_sat': 40.70928571428572, 'name': '10_VGG13_A2'}, {'test_acc': 0.8939, 'test_loss': 0.002698920160531997, 'train_acc': 0.9453600000000001, 'train_loss': 0.001237413241788745, 'average_sat': 35.157272727272726, 'name': '10_VGG13_S_A0'}, {'test_acc': 0.8362, 'test_loss': 0.003840552979707718, 'train_acc': 0.85882, 'train_loss': 0.003271747159063816, 'average_sat': 60.29818181818182, 'name': '10_VGG13_XS_1'}, {'test_acc': 0.8367, 'test_loss': 0.003826314589381218, 'train_acc': 0.8583, 'train_loss': 0.003268085570037365, 'average_sat': 60.51181818181819, 'name': '10_VGG13_XS_2'}, {'test_acc': 0.8337, 'test_loss': 0.003914144262671471, 'train_acc': 0.85514, 'train_loss': 0.003353869499266148, 'average_sat': 61.50545454545456, 'name': '10_VGG13_XS_A0'}, {'test_acc': 0.7658, 'test_loss': 0.005358525905013086, 'train_acc': 0.76688, 'train_loss': 0.005399670512080193, 'average_sat': 66.61999999999999, 'name': '10_VGG13_XXS_1'}, {'test_acc': 0.7747, 'test_loss': 0.005206653934717178, 'train_acc': 0.7722600000000001, 'train_loss': 0.005249527580738067, 'average_sat': 66.0509090909091, 'name': '10_VGG13_XXS_A0'}, {'test_acc': 0.6554, 'test_loss': 0.007741486471891403, 'train_acc': 0.6332399999999999, 'train_loss': 0.008358000195026397, 'average_sat': 69.88636363636364, 'name': '10_VGG13_XXXS_1'}, {'test_acc': 0.6545, 'test_loss': 0.007767208981513976, 'train_acc': 0.6280600000000001, 'train_loss': 0.008463159784078599, 'average_sat': 69.88636363636364, 'name': '10_VGG13_XXXS_A0'}, {'test_acc': 0.8825, 'test_loss': 0.002986784648895264, 'train_acc': 0.93714, 'train_loss': 0.001512837432026863, 'average_sat': 27.985000000000003, 'name': '10_VGG16_A0'}, {'test_acc': 0.865, 'test_loss': 0.0032702807977795607, 'train_acc': 0.90784, 'train_loss': 0.0021960915344953536, 'average_sat': 40.23428571428571, 'name': '10_VGG16_S'}, {'test_acc': 0.8859, 'test_loss': 0.003097249273955822, 'train_acc': 0.9383799999999999, 'train_loss': 0.0014641700659692287, 'average_sat': 27.942857142857143, 'name': '10_VGG16_S_A0'}, {'test_acc': 0.8270000000000001, 'test_loss': 0.0041146620541810995, 'train_acc': 0.85426, 'train_loss': 0.0034174102574586872, 'average_sat': 53.18071428571428, 'name': '10_VGG16_XS_1'}, {'test_acc': 0.8357, 'test_loss': 0.003887486907839775, 'train_acc': 0.8549399999999999, 'train_loss': 0.0034285132145881647, 'average_sat': 52.34428571428572, 'name': '10_VGG16_XS_2'}, {'test_acc': 0.8351, 'test_loss': 0.00397021451741457, 'train_acc': 0.8557600000000001, 'train_loss': 0.003386295389533043, 'average_sat': 51.78642857142857, 'name': '10_VGG16_XS_A0'}, {'test_acc': 0.7709, 'test_loss': 0.005274969577789307, 'train_acc': 0.77282, 'train_loss': 0.005232830882072449, 'average_sat': 62.945714285714295, 'name': '10_VGG16_XXS_1'}, {'test_acc': 0.7692, 'test_loss': 0.0053257405370473855, 'train_acc': 0.7662800000000001, 'train_loss': 0.005363836864233017, 'average_sat': 61.94142857142857, 'name': '10_VGG16_XXS_A0'}, {'test_acc': 0.6316, 'test_loss': 0.008241455793380738, 'train_acc': 0.6150399999999999, 'train_loss': 0.008661547809839248, 'average_sat': 69.19642857142857, 'name': '10_VGG16_XXXS_1'}, {'test_acc': 0.6379, 'test_loss': 0.008092593294382095, 'train_acc': 0.6241399999999999, 'train_loss': 0.008564249309301377, 'average_sat': 67.63357142857143, 'name': '10_VGG16_XXXS_A0'}, {'test_acc': 0.8676, 'test_loss': 0.0035951254978775975, 'train_acc': 0.9268, 'train_loss': 0.0018311631272733208, 'average_sat': 23.333529411764705, 'name': '10_VGG19_A0'}, {'test_acc': 0.8758, 'test_loss': 0.0033266779631376266, 'train_acc': 0.92322, 'train_loss': 0.0019193500845134253, 'average_sat': 22.874117647058824, 'name': '10_VGG19_A2'}, {'test_acc': 0.8555, 'test_loss': 0.00360090516358614, 'train_acc': 0.8927799999999999, 'train_loss': 0.0026590359604358677, 'average_sat': 33.86941176470589, 'name': '10_VGG19_S_2'}, {'test_acc': 0.8571, 'test_loss': 0.003575297597795725, 'train_acc': 0.9012600000000001, 'train_loss': 0.002358041264116764, 'average_sat': 34.03, 'name': '10_VGG19_S_A0'}, {'test_acc': 0.8105, 'test_loss': 0.004748665735125542, 'train_acc': 0.8353200000000001, 'train_loss': 0.003998426278829574, 'average_sat': 43.291176470588226, 'name': '10_VGG19_XS_2'}, {'test_acc': 0.8276, 'test_loss': 0.004186698150634766, 'train_acc': 0.8440799999999999, 'train_loss': 0.003689096356034279, 'average_sat': 44.11823529411765, 'name': '10_VGG19_XS_A0'}, {'test_acc': 0.7563, 'test_loss': 0.005804333865642547, 'train_acc': 0.75444, 'train_loss': 0.005772608959078789, 'average_sat': 54.962941176470586, 'name': '10_VGG19_XXS_1'}, {'test_acc': 0.7503, 'test_loss': 0.005908278012275696, 'train_acc': 0.753, 'train_loss': 0.005756411521434785, 'average_sat': 55.69764705882352, 'name': '10_VGG19_XXS_A0'}, {'test_acc': 0.6331, 'test_loss': 0.008346724969148636, 'train_acc': 0.60246, 'train_loss': 0.00904909703373909, 'average_sat': 64.33882352941177, 'name': '10_VGG19_XXXS_1'}, {'test_acc': 0.6289, 'test_loss': 0.008463240754604341, 'train_acc': 0.61578, 'train_loss': 0.008749859375953675, 'average_sat': 63.970588235294116, 'name': '10_VGG19_XXXS_A0'}, {'test_acc': 0.8685, 'test_loss': 0.002694634012877941, 'train_acc': 0.8952, 'train_loss': 0.0019315169364213945, 'average_sat': 11.152222222222225, 'name': 'catdog_VGG11_A0'}, {'test_acc': 0.8635, 'test_loss': 0.002621495984494686, 'train_acc': 0.8949, 'train_loss': 0.0019586543768644327, 'average_sat': 27.1309090909091, 'name': 'catdog_VGG11_A2'}, {'test_acc': 0.8440000000000001, 'test_loss': 0.0029504293128848075, 'train_acc': 0.8965, 'train_loss': 0.0019527736246585843, 'average_sat': 10.893333333333336, 'name': 'catdog_VGG11_A3'}, {'test_acc': 0.8315, 'test_loss': 0.003122882641851902, 'train_acc': 0.8947, 'train_loss': 0.0019851779416203497, 'average_sat': 10.893333333333336, 'name': 'catdog_VGG11_A4'}, {'test_acc': 0.843, 'test_loss': 0.003159349583089352, 'train_acc': 0.8871, 'train_loss': 0.0020762361511588103, 'average_sat': 10.980000000000002, 'name': 'catdog_VGG11_S_2'}, {'test_acc': 0.84, 'test_loss': 0.002821375988423824, 'train_acc': 0.8801, 'train_loss': 0.0021853141784667967, 'average_sat': 20.283333333333335, 'name': 'catdog_VGG11_S_A0'}, {'test_acc': 0.8420000000000001, 'test_loss': 0.003123454570770264, 'train_acc': 0.8809999999999999, 'train_loss': 0.002199879877269268, 'average_sat': 19.543333333333337, 'name': 'catdog_VGG11_S_A3'}, {'test_acc': 0.8505, 'test_loss': 0.002927114509046077, 'train_acc': 0.8854, 'train_loss': 0.0021499499306082727, 'average_sat': 19.892222222222227, 'name': 'catdog_VGG11_S_A4'}, {'test_acc': 0.812, 'test_loss': 0.0032540484815835953, 'train_acc': 0.8432, 'train_loss': 0.002707408933341503, 'average_sat': 30.555555555555557, 'name': 'catdog_VGG11_XS_1'}, {'test_acc': 0.812, 'test_loss': 0.003366310626268387, 'train_acc': 0.8434, 'train_loss': 0.002720622482895851, 'average_sat': 34.11333333333333, 'name': 'catdog_VGG11_XS_A0'}, {'test_acc': 0.82, 'test_loss': 0.0032648295909166337, 'train_acc': 0.8577, 'train_loss': 0.002553858995437622, 'average_sat': 32.638888888888886, 'name': 'catdog_VGG11_XS_A1'}, {'test_acc': 0.813, 'test_loss': 0.0031264190524816515, 'train_acc': 0.8497, 'train_loss': 0.002663262373209, 'average_sat': 35.59, 'name': 'catdog_VGG11_XS_A2'}, {'test_acc': 0.809, 'test_loss': 0.0035266731306910514, 'train_acc': 0.8516, 'train_loss': 0.002643510335683823, 'average_sat': 30.90444444444445, 'name': 'catdog_VGG11_XS_A3'}, {'test_acc': 0.8059999999999999, 'test_loss': 0.0032725931257009512, 'train_acc': 0.8499, 'train_loss': 0.002642316530644893, 'average_sat': 32.465555555555554, 'name': 'catdog_VGG11_XS_A4'}, {'test_acc': 0.7909999999999999, 'test_loss': 0.0035595538020133967, 'train_acc': 0.8021, 'train_loss': 0.003297457775473595, 'average_sat': 47.91555555555556, 'name': 'catdog_VGG11_XXS_1'}, {'test_acc': 0.7965, 'test_loss': 0.0034204698950052264, 'train_acc': 0.8066, 'train_loss': 0.0032473419070243835, 'average_sat': 48.78333333333333, 'name': 'catdog_VGG11_XXS_A0'}, {'test_acc': 0.794, 'test_loss': 0.0034867883771657945, 'train_acc': 0.7995, 'train_loss': 0.003340817755460739, 'average_sat': 48.61, 'name': 'catdog_VGG11_XXS_A1'}, {'test_acc': 0.7609999999999999, 'test_loss': 0.0037985174655914312, 'train_acc': 0.8024, 'train_loss': 0.003264449995756149, 'average_sat': 47.74444444444444, 'name': 'catdog_VGG11_XXS_A3'}, {'test_acc': 0.7745, 'test_loss': 0.003655799061059952, 'train_acc': 0.7936, 'train_loss': 0.0033786171197891235, 'average_sat': 50.86666666666667, 'name': 'catdog_VGG11_XXS_A4'}, {'test_acc': 0.7559999999999999, 'test_loss': 0.004022237867116928, 'train_acc': 0.7428, 'train_loss': 0.004049529632925986, 'average_sat': 62.153333333333336, 'name': 'catdog_VGG11_XXXS_A0'}, {'test_acc': 0.746, 'test_loss': 0.004075590044260025, 'train_acc': 0.7459, 'train_loss': 0.004019846785068512, 'average_sat': 64.58444444444444, 'name': 'catdog_VGG11_XXXS_A1'}, {'test_acc': 0.7345, 'test_loss': 0.004052538141608239, 'train_acc': 0.7426, 'train_loss': 0.004077369254827499, 'average_sat': 59.376666666666665, 'name': 'catdog_VGG11_XXXS_A3'}, {'test_acc': 0.7555, 'test_loss': 0.004045585677027702, 'train_acc': 0.7402, 'train_loss': 0.0040807071685791015, 'average_sat': 64.23555555555555, 'name': 'catdog_VGG11_XXXS_A4'}, {'test_acc': 0.8740000000000001, 'test_loss': 0.0024774231314659122, 'train_acc': 0.9108, 'train_loss': 0.0016929133042693136, 'average_sat': 15.959999999999997, 'name': 'catdog_VGG13_A0'}, {'test_acc': 0.85, 'test_loss': 0.003101175479590893, 'train_acc': 0.9026, 'train_loss': 0.001853799730539322, 'average_sat': 22.182857142857152, 'name': 'catdog_VGG13_A2'}, {'test_acc': 0.8635, 'test_loss': 0.0025618661865592003, 'train_acc': 0.9189, 'train_loss': 0.0016111065991222858, 'average_sat': 15.72909090909091, 'name': 'catdog_VGG13_A3'}, {'test_acc': 0.8785, 'test_loss': 0.002419160284101963, 'train_acc': 0.9177, 'train_loss': 0.0015917012020945551, 'average_sat': 15.606363636363637, 'name': 'catdog_VGG13_A4'}, {'test_acc': 0.8809999999999999, 'test_loss': 0.0023667931407690052, 'train_acc': 0.9122, 'train_loss': 0.0016455898158252238, 'average_sat': 15.640909090909089, 'name': 'catdog_VGG13_S_2'}, {'test_acc': 0.831, 'test_loss': 0.003177961990237236, 'train_acc': 0.9023, 'train_loss': 0.001937880353629589, 'average_sat': 27.91181818181818, 'name': 'catdog_VGG13_S_A0'}, {'test_acc': 0.8735, 'test_loss': 0.002632187902927399, 'train_acc': 0.9005, 'train_loss': 0.0018407063692808148, 'average_sat': 27.628181818181815, 'name': 'catdog_VGG13_S_A3'}, {'test_acc': 0.862, 'test_loss': 0.0026278874799609184, 'train_acc': 0.8975, 'train_loss': 0.0018734912097454074, 'average_sat': 25.851818181818178, 'name': 'catdog_VGG13_S_A4'}, {'test_acc': 0.8305, 'test_loss': 0.003119363121688366, 'train_acc': 0.8641, 'train_loss': 0.002399623741209507, 'average_sat': 39.841818181818184, 'name': 'catdog_VGG13_XS_1'}, {'test_acc': 0.8255, 'test_loss': 0.003086755715310573, 'train_acc': 0.8605, 'train_loss': 0.0024758649431169035, 'average_sat': 40.90818181818182, 'name': 'catdog_VGG13_XS_A0'}, {'test_acc': 0.833, 'test_loss': 0.0030465585440397265, 'train_acc': 0.8557, 'train_loss': 0.0025216143399477003, 'average_sat': 38.06818181818182, 'name': 'catdog_VGG13_XS_A1'}, {'test_acc': 0.8240000000000001, 'test_loss': 0.003209416277706623, 'train_acc': 0.8575, 'train_loss': 0.002499591769278049, 'average_sat': 38.84909090909091, 'name': 'catdog_VGG13_XS_A2'}, {'test_acc': 0.836, 'test_loss': 0.0030663911700248717, 'train_acc': 0.8621, 'train_loss': 0.0024395530581474306, 'average_sat': 39.27636363636364, 'name': 'catdog_VGG13_XS_A3'}, {'test_acc': 0.8235, 'test_loss': 0.003141145184636116, 'train_acc': 0.8629, 'train_loss': 0.0024561676859855648, 'average_sat': 39.41636363636364, 'name': 'catdog_VGG13_XS_A4'}, {'test_acc': 0.7879999999999999, 'test_loss': 0.0035656539052724844, 'train_acc': 0.8053, 'train_loss': 0.003253913900256157, 'average_sat': 51.41909090909091, 'name': 'catdog_VGG13_XXS_1'}, {'test_acc': 0.8005, 'test_loss': 0.0034210907742381093, 'train_acc': 0.8073, 'train_loss': 0.00322254194021225, 'average_sat': 49.14727272727273, 'name': 'catdog_VGG13_XXS_A0'}, {'test_acc': 0.805, 'test_loss': 0.0033116612881422037, 'train_acc': 0.8125, 'train_loss': 0.003141328290104866, 'average_sat': 51.42181818181818, 'name': 'catdog_VGG13_XXS_A1'}, {'test_acc': 0.8055, 'test_loss': 0.003420772381126881, 'train_acc': 0.8199, 'train_loss': 0.003086591449379921, 'average_sat': 52.98181818181818, 'name': 'catdog_VGG13_XXS_A3'}, {'test_acc': 0.746, 'test_loss': 0.0039894977137446405, 'train_acc': 0.8079, 'train_loss': 0.003273702898621559, 'average_sat': 50.70909090909092, 'name': 'catdog_VGG13_XXS_A4'}, {'test_acc': 0.7415, 'test_loss': 0.004097677379846573, 'train_acc': 0.735, 'train_loss': 0.004125016662478447, 'average_sat': 65.05727272727273, 'name': 'catdog_VGG13_XXXS_A0'}, {'test_acc': 0.753, 'test_loss': 0.00406969353556633, 'train_acc': 0.7474, 'train_loss': 0.004013991624116898, 'average_sat': 58.52181818181818, 'name': 'catdog_VGG13_XXXS_A1'}, {'test_acc': 0.7265, 'test_loss': 0.004312678650021553, 'train_acc': 0.7431, 'train_loss': 0.00405831329524517, 'average_sat': 64.49, 'name': 'catdog_VGG13_XXXS_A3'}, {'test_acc': 0.758, 'test_loss': 0.003991135179996491, 'train_acc': 0.7449, 'train_loss': 0.0040021948724985115, 'average_sat': 65.90818181818182, 'name': 'catdog_VGG13_XXXS_A4'}, {'test_acc': 0.867, 'test_loss': 0.002629607528448105, 'train_acc': 0.9085, 'train_loss': 0.0017024948820471765, 'average_sat': 13.461428571428566, 'name': 'catdog_VGG16_A0'}, {'test_acc': 0.8545, 'test_loss': 0.0026699738875031478, 'train_acc': 0.9101, 'train_loss': 0.001667993824183941, 'average_sat': 13.586428571428572, 'name': 'catdog_VGG16_A2'}, {'test_acc': 0.8290000000000001, 'test_loss': 0.0036543388999998572, 'train_acc': 0.9117, 'train_loss': 0.0016190816707909107, 'average_sat': 13.474285714285712, 'name': 'catdog_VGG16_A3'}, {'test_acc': 0.8175, 'test_loss': 0.003768643796443939, 'train_acc': 0.9006, 'train_loss': 0.0018419273525476453, 'average_sat': 22.461428571428577, 'name': 'catdog_VGG16_S_A0'}, {'test_acc': 0.8585, 'test_loss': 0.002684751987457275, 'train_acc': 0.9104, 'train_loss': 0.0016639040306210518, 'average_sat': 13.182142857142853, 'name': 'catdog_VGG16_S_A2'}, {'test_acc': 0.855, 'test_loss': 0.00273121777176857, 'train_acc': 0.8967, 'train_loss': 0.0019008837729692454, 'average_sat': 22.462142857142858, 'name': 'catdog_VGG16_S_A3'}, {'test_acc': 0.843, 'test_loss': 0.003062607228755951, 'train_acc': 0.9068, 'train_loss': 0.001765730431675911, 'average_sat': 22.711428571428577, 'name': 'catdog_VGG16_S_A4'}, {'test_acc': 0.8290000000000001, 'test_loss': 0.003006260700523853, 'train_acc': 0.8641, 'train_loss': 0.0024122435607016085, 'average_sat': 36.82928571428572, 'name': 'catdog_VGG16_XS_1'}, {'test_acc': 0.8325, 'test_loss': 0.003084098294377327, 'train_acc': 0.8655, 'train_loss': 0.002367827166616917, 'average_sat': 36.38357142857143, 'name': 'catdog_VGG16_XS_A0'}, {'test_acc': 0.8185, 'test_loss': 0.003229932174086571, 'train_acc': 0.863, 'train_loss': 0.002484482754766941, 'average_sat': 33.92857142857143, 'name': 'catdog_VGG16_XS_A1'}, {'test_acc': 0.826, 'test_loss': 0.0031214293390512465, 'train_acc': 0.8561, 'train_loss': 0.002553989844024181, 'average_sat': 36.38285714285715, 'name': 'catdog_VGG16_XS_A2'}, {'test_acc': 0.848, 'test_loss': 0.002887876927852631, 'train_acc': 0.8527, 'train_loss': 0.0026231952369213107, 'average_sat': 33.70428571428571, 'name': 'catdog_VGG16_XS_A3'}, {'test_acc': 0.8265, 'test_loss': 0.0030514897331595423, 'train_acc': 0.8704, 'train_loss': 0.002426468749344349, 'average_sat': 34.76642857142858, 'name': 'catdog_VGG16_XS_A4'}, {'test_acc': 0.807, 'test_loss': 0.003271414056420326, 'train_acc': 0.8114, 'train_loss': 0.0031806650191545487, 'average_sat': 45.48571428571429, 'name': 'catdog_VGG16_XXS_1'}, {'test_acc': 0.772, 'test_loss': 0.003828900329768658, 'train_acc': 0.8137, 'train_loss': 0.0031961119055747985, 'average_sat': 46.98571428571428, 'name': 'catdog_VGG16_XXS_A0'}, {'test_acc': 0.7985, 'test_loss': 0.0034104646891355518, 'train_acc': 0.8006, 'train_loss': 0.003309540662169457, 'average_sat': 47.20928571428571, 'name': 'catdog_VGG16_XXS_A1'}, {'test_acc': 0.81, 'test_loss': 0.0032649461925029753, 'train_acc': 0.8051, 'train_loss': 0.003263369795680046, 'average_sat': 48.9957142857143, 'name': 'catdog_VGG16_XXS_A3'}, {'test_acc': 0.7785, 'test_loss': 0.003667614087462426, 'train_acc': 0.8024, 'train_loss': 0.0032701441824436192, 'average_sat': 50.22285714285715, 'name': 'catdog_VGG16_XXS_A4'}, {'test_acc': 0.7295, 'test_loss': 0.0042751260101795184, 'train_acc': 0.7499, 'train_loss': 0.003991067180037499, 'average_sat': 62.94714285714286, 'name': 'catdog_VGG16_XXXS_A0'}, {'test_acc': 0.737, 'test_loss': 0.004082210406661034, 'train_acc': 0.7396, 'train_loss': 0.0041464025318622595, 'average_sat': 61.83, 'name': 'catdog_VGG16_XXXS_A1'}, {'test_acc': 0.713, 'test_loss': 0.0042377296686172475, 'train_acc': 0.7476, 'train_loss': 0.003969203984737396, 'average_sat': 58.48285714285714, 'name': 'catdog_VGG16_XXXS_A3'}, {'test_acc': 0.7285, 'test_loss': 0.0042198471128940575, 'train_acc': 0.7294, 'train_loss': 0.0042175237715244285, 'average_sat': 63.839285714285715, 'name': 'catdog_VGG16_XXXS_A4'}, {'test_acc': 0.8745, 'test_loss': 0.002697319455444813, 'train_acc': 0.9049, 'train_loss': 0.0017897516377270222, 'average_sat': 11.798235294117646, 'name': 'catdog_VGG19_A0'}, {'test_acc': 0.868, 'test_loss': 0.0025764219164848326, 'train_acc': 0.9125, 'train_loss': 0.001644399444013834, 'average_sat': 11.429999999999998, 'name': 'catdog_VGG19_A2'}, {'test_acc': 0.8740000000000001, 'test_loss': 0.002539345122873783, 'train_acc': 0.9088, 'train_loss': 0.001699266306310892, 'average_sat': 12.074117647058822, 'name': 'catdog_VGG19_A3'}, {'test_acc': 0.865, 'test_loss': 0.002767819829285145, 'train_acc': 0.9026, 'train_loss': 0.0018331575497984886, 'average_sat': 11.568823529411763, 'name': 'catdog_VGG19_A4'}, {'test_acc': 0.8585, 'test_loss': 0.0025251737534999845, 'train_acc': 0.8936, 'train_loss': 0.0019407542586326603, 'average_sat': 21.461176470588242, 'name': 'catdog_VGG19_S_2'}, {'test_acc': 0.8335, 'test_loss': 0.0033522522076964373, 'train_acc': 0.8974, 'train_loss': 0.0019015515074133876, 'average_sat': 20.865294117647057, 'name': 'catdog_VGG19_S_A0'}, {'test_acc': 0.841, 'test_loss': 0.0031415996812283998, 'train_acc': 0.8956, 'train_loss': 0.00193530947715044, 'average_sat': 21.50882352941176, 'name': 'catdog_VGG19_S_A3'}, {'test_acc': 0.8565, 'test_loss': 0.0025708468481898308, 'train_acc': 0.8947, 'train_loss': 0.0019279314428567887, 'average_sat': 20.7964705882353, 'name': 'catdog_VGG19_S_A4'}, {'test_acc': 0.84, 'test_loss': 0.0028997346088290217, 'train_acc': 0.8628, 'train_loss': 0.0024673070460557943, 'average_sat': 31.803529411764703, 'name': 'catdog_VGG19_XS_A0'}, {'test_acc': 0.8285, 'test_loss': 0.002985629230737686, 'train_acc': 0.8477, 'train_loss': 0.0026621660515666013, 'average_sat': 31.250000000000007, 'name': 'catdog_VGG19_XS_A2'}, {'test_acc': 0.8320000000000001, 'test_loss': 0.0031234656497836108, 'train_acc': 0.8651, 'train_loss': 0.002418072694540024, 'average_sat': 31.892941176470586, 'name': 'catdog_VGG19_XS_A3'}, {'test_acc': 0.8170000000000001, 'test_loss': 0.003215162478387356, 'train_acc': 0.8517, 'train_loss': 0.0025538194820284845, 'average_sat': 30.700588235294124, 'name': 'catdog_VGG19_XS_A4'}, {'test_acc': 0.8079999999999999, 'test_loss': 0.0032894854545593264, 'train_acc': 0.8045, 'train_loss': 0.0033372373431921004, 'average_sat': 43.65764705882353, 'name': 'catdog_VGG19_XXS_1'}, {'test_acc': 0.8009999999999999, 'test_loss': 0.003364966355264187, 'train_acc': 0.8178, 'train_loss': 0.003147915044426918, 'average_sat': 45.128823529411775, 'name': 'catdog_VGG19_XXS_A0'}, {'test_acc': 0.804, 'test_loss': 0.003335661232471466, 'train_acc': 0.8131, 'train_loss': 0.003209082412719727, 'average_sat': 44.30058823529412, 'name': 'catdog_VGG19_XXS_A1'}, {'test_acc': 0.767, 'test_loss': 0.003799564518034458, 'train_acc': 0.8029999999999999, 'train_loss': 0.003293837490677833, 'average_sat': 43.014705882352935, 'name': 'catdog_VGG19_XXS_A3'}, {'test_acc': 0.7829999999999999, 'test_loss': 0.0037329939231276507, 'train_acc': 0.8054, 'train_loss': 0.0032319936990737913, 'average_sat': 44.20823529411764, 'name': 'catdog_VGG19_XXS_A4'}, {'test_acc': 0.7345, 'test_loss': 0.004149810791015625, 'train_acc': 0.7343, 'train_loss': 0.004161357006430626, 'average_sat': 58.45470588235294, 'name': 'catdog_VGG19_XXXS_A0'}, {'test_acc': 0.752, 'test_loss': 0.0040563028901815425, 'train_acc': 0.7405, 'train_loss': 0.004083720451593399, 'average_sat': 56.248823529411766, 'name': 'catdog_VGG19_XXXS_A1'}, {'test_acc': 0.7495, 'test_loss': 0.004021619945764542, 'train_acc': 0.7472, 'train_loss': 0.004029232785105705, 'average_sat': 56.98529411764706, 'name': 'catdog_VGG19_XXXS_A3'}, {'test_acc': 0.7315, 'test_loss': 0.004230681315064429, 'train_acc': 0.7363, 'train_loss': 0.004061273249983789, 'average_sat': 56.06588235294117, 'name': 'catdog_VGG19_XXXS_A4'}, {'test_acc': 0.977, 'test_loss': 0.000530758123844862, 'train_acc': 0.9666, 'train_loss': 0.0007876014675945043, 'average_sat': 36.029411764705884, 'name': 'cd_128_VGG11VGG19VGG19_XXXS_A0'}, {'test_acc': 0.971, 'test_loss': 0.0007012429870665072, 'train_acc': 0.9638, 'train_loss': 0.0007869602128863334, 'average_sat': 41.361176470588234, 'name': 'cd_128_VGG11VGG19VGG19_XXXS_A1'}, {'test_acc': 0.9925, 'test_loss': 0.00018529360252432525, 'train_acc': 0.9888, 'train_loss': 0.00025930717014707624, 'average_sat': 8.799999999999997, 'name': 'cd_128_VGG11_XXXS_A0'}, {'test_acc': 0.9915, 'test_loss': 0.0002026885882951319, 'train_acc': 0.9819, 'train_loss': 0.00043969581220299007, 'average_sat': 8.880588235294114, 'name': 'cd_128_VGG11_XXXS_A1'}, {'test_acc': 0.9690000000000001, 'test_loss': 0.0006819703113287687, 'train_acc': 0.9606, 'train_loss': 0.0008815486811101438, 'average_sat': 41.72764705882353, 'name': 'cd_256_VGG11VGG19VGG19_XXXS_A0'}, {'test_acc': 0.97, 'test_loss': 0.0006760791088454424, 'train_acc': 0.9677, 'train_loss': 0.0006928824875503777, 'average_sat': 37.68411764705883, 'name': 'cd_256_VGG11VGG19VGG19_XXXS_A1'}, {'test_acc': 0.99, 'test_loss': 0.00022883394430391488, 'train_acc': 0.9911, 'train_loss': 0.00021640934241004287, 'average_sat': 9.041764705882352, 'name': 'cd_256_VGG11_XXXS_A0'}, {'test_acc': 0.988, 'test_loss': 0.0004164109574630856, 'train_acc': 0.9934, 'train_loss': 0.00016901938999071717, 'average_sat': 8.98470588235294, 'name': 'cd_256_VGG11_XXXS_A1'}, {'test_acc': 0.971, 'test_loss': 0.0006070871520787478, 'train_acc': 0.965, 'train_loss': 0.0007527809351682662, 'average_sat': 38.97117647058823, 'name': 'cd_32_VGG11VGG19VGG19_XXXS_A1'}, {'test_acc': 0.97, 'test_loss': 0.0007007738519459963, 'train_acc': 0.9623, 'train_loss': 0.0008510698556900023, 'average_sat': 40.99294117647059, 'name': 'cd_32_VGG11VGG19VGG19_XXXS_A2'}, {'test_acc': 0.988, 'test_loss': 0.0002515608079265803, 'train_acc': 0.9883, 'train_loss': 0.0002593355044256896, 'average_sat': 9.662352941176469, 'name': 'cd_32_VGG11_XXXS_A0'}, {'test_acc': 0.9945, 'test_loss': 0.00015805791853927076, 'train_acc': 0.9812, 'train_loss': 0.0004399410227779299, 'average_sat': 9.075294117647054, 'name': 'cd_32_VGG11_XXXS_A1'}, {'test_acc': 0.9665, 'test_loss': 0.0008241188344545662, 'train_acc': 0.9651, 'train_loss': 0.0007229028768837452, 'average_sat': 41.911764705882355, 'name': 'cd_64_VGG11VGG19VGG19_XXXS_A0'}, {'test_acc': 0.9755, 'test_loss': 0.0005903222225606443, 'train_acc': 0.9688, 'train_loss': 0.0007760775502771138, 'average_sat': 40.80823529411765, 'name': 'cd_64_VGG11VGG19VGG19_XXXS_A1'}, {'test_acc': 0.9875, 'test_loss': 0.00030730583338299766, 'train_acc': 0.9937, 'train_loss': 0.00015526653984561565, 'average_sat': 8.938823529411762, 'name': 'cd_64_VGG11_XXXS_A0'}, {'test_acc': 0.9925, 'test_loss': 0.0002353996434248984, 'train_acc': 0.9925, 'train_loss': 0.000162417875463143, 'average_sat': 8.754117647058823, 'name': 'cd_64_VGG11_XXXS_A1'}]\n"
     ]
    }
   ],
   "source": [
    "print(result)\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[28.14666667 42.62333333 54.51333333 63.88666667 66.66666667 34.94363636\n",
      " 48.61545455 58.23909091 64.77181818 68.46454545 28.13928571 41.63142857\n",
      " 52.4        62.94642857 67.40928571 23.10411765 33.27176471 45.54235294\n",
      " 55.97411765 62.50058824 28.38666667 48.29545455 42.44666667 40.84222222\n",
      " 55.20888889 55.20666667 65.79777778 63.88777778 71.17888889 67.35888889\n",
      " 34.69363636 40.70928571 35.15727273 60.29818182 60.51181818 61.50545455\n",
      " 66.62       66.05090909 69.88636364 69.88636364 27.985      40.23428571\n",
      " 27.94285714 53.18071429 52.34428571 51.78642857 62.94571429 61.94142857\n",
      " 69.19642857 67.63357143 23.33352941 22.87411765 33.86941176 34.03\n",
      " 43.29117647 44.11823529 54.96294118 55.69764706 64.33882353 63.97058824\n",
      " 11.15222222 27.13090909 10.89333333 10.89333333 10.98       20.28333333\n",
      " 19.54333333 19.89222222 30.55555556 34.11333333 32.63888889 35.59\n",
      " 30.90444444 32.46555556 47.91555556 48.78333333 48.61       47.74444444\n",
      " 50.86666667 62.15333333 64.58444444 59.37666667 64.23555556 15.96\n",
      " 22.18285714 15.72909091 15.60636364 15.64090909 27.91181818 27.62818182\n",
      " 25.85181818 39.84181818 40.90818182 38.06818182 38.84909091 39.27636364\n",
      " 39.41636364 51.41909091 49.14727273 51.42181818 52.98181818 50.70909091\n",
      " 65.05727273 58.52181818 64.49       65.90818182 13.46142857 13.58642857\n",
      " 13.47428571 22.46142857 13.18214286 22.46214286 22.71142857 36.82928571\n",
      " 36.38357143 33.92857143 36.38285714 33.70428571 34.76642857 45.48571429\n",
      " 46.98571429 47.20928571 48.99571429 50.22285714 62.94714286 61.83\n",
      " 58.48285714 63.83928571 11.79823529 11.43       12.07411765 11.56882353\n",
      " 21.46117647 20.86529412 21.50882353 20.79647059 31.80352941 31.25\n",
      " 31.89294118 30.70058824 43.65764706 45.12882353 44.30058824 43.01470588\n",
      " 44.20823529 58.45470588 56.24882353 56.98529412 56.06588235 36.02941176\n",
      " 41.36117647  8.8         8.88058824 41.72764706 37.68411765  9.04176471\n",
      "  8.98470588 38.97117647 40.99294118  9.66235294  9.07529412 41.91176471\n",
      " 40.80823529  8.93882353  8.75411765]\n",
      "['Imb. CIFAR10' 'Imb. CIFAR10' 'Imb. CIFAR10' 'Imb. CIFAR10'\n",
      " 'Imb. CIFAR10' 'Imb. CIFAR10' 'Imb. CIFAR10' 'Imb. CIFAR10'\n",
      " 'Imb. CIFAR10' 'Imb. CIFAR10' 'Imb. CIFAR10' 'Imb. CIFAR10'\n",
      " 'Imb. CIFAR10' 'Imb. CIFAR10' 'Imb. CIFAR10' 'Imb. CIFAR10'\n",
      " 'Imb. CIFAR10' 'Imb. CIFAR10' 'Imb. CIFAR10' 'Imb. CIFAR10' 'CIFAR10'\n",
      " 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10'\n",
      " 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10'\n",
      " 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10'\n",
      " 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10'\n",
      " 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10'\n",
      " 'CIFAR10' 'CIFAR10' 'CIFAR10' 'CIFAR10' 'cat-vs-dog' 'cat-vs-dog'\n",
      " 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog'\n",
      " 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog'\n",
      " 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog'\n",
      " 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog'\n",
      " 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog'\n",
      " 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog'\n",
      " 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog'\n",
      " 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog'\n",
      " 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog'\n",
      " 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog'\n",
      " 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog'\n",
      " 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog'\n",
      " 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog'\n",
      " 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog'\n",
      " 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog'\n",
      " 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog'\n",
      " 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog' 'cat-vs-dog'\n",
      " 'cat-vs-dog' 'cat-vs-dog' 'cd' 'cd' 'cd' 'cd' 'cd' 'cd' 'cd' 'cd' 'cd'\n",
      " 'cd' 'cd' 'cd' 'cd' 'cd' 'cd' 'cd']\n",
      "['cat-vs-dog' 'Imb. CIFAR10' 'cd' 'CIFAR10']\n",
      "['VGG16', 'VGG19', 'VGG11', 'VGG13']\n"
     ]
    }
   ],
   "source": [
    "all_test_acc = np.array([res['test_acc'] for res in result])\n",
    "all_train_acc = np.array([res['train_acc'] for res in result])\n",
    "\n",
    "all_test_loss = np.array([res['test_loss'] for res in result])\n",
    "all_train_loss = np.array([res['train_loss'] for res in result])\n",
    "\n",
    "all_average_sat = np.array([res['average_sat'] for res in result])\n",
    "print(all_average_sat)\n",
    "all_names = np.array([res['name'] for res in result])\n",
    "\n",
    "all_problem = np.array([name.split('_')[0] for name in all_names])\n",
    "all_problem[all_problem == '10'] = 'CIFAR10'\n",
    "all_problem[all_problem == '10imbalanced'] = 'Imb. CIFAR10'\n",
    "all_problem[all_problem == 'catdog'] = 'cat-vs-dog'\n",
    "all_problem[all_problem == '100'] = 'CIFAR100'\n",
    "\n",
    "all_networks = [name.split('_')[1] for name in all_names if 'VGG' in name.split('_')[1]]\n",
    "all_networks_filtered  = []\n",
    "filter_sizes = []\n",
    "depth = []\n",
    "for i in range(len(all_names)):\n",
    "    if 'VGG11_A' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG11')\n",
    "        filter_sizes.append('full Filtersize')\n",
    "    elif 'VGG13_A' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG13')\n",
    "        filter_sizes.append('full Filtersize')\n",
    "    elif 'VGG16_A' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG16')\n",
    "        filter_sizes.append('full Filtersize')\n",
    "    elif 'VGG19_A' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG19')\n",
    "        filter_sizes.append('full Filtersize')\n",
    "    elif 'VGG11_S' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG11')\n",
    "        filter_sizes.append('halfed Filtersize')\n",
    "    elif 'VGG11_XS' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG11')\n",
    "        filter_sizes.append('quarter Filtersize')\n",
    "    elif 'VGG11_XXS' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG11')\n",
    "        filter_sizes.append('eigths Filtersize')\n",
    "    elif 'VGG11_XXXS' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG11')\n",
    "        filter_sizes.append('sixteenth Filtersize')\n",
    "    elif 'VGG13_S' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG13')\n",
    "        filter_sizes.append('halfed Filtersize')\n",
    "    elif 'VGG13_XS' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG13')\n",
    "        filter_sizes.append('quarter Filtersize')\n",
    "    elif 'VGG13_XXS' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG13')\n",
    "        filter_sizes.append('eigths Filtersize')\n",
    "    elif 'VGG13_XXXS' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG13')\n",
    "        filter_sizes.append('sixteenth Filtersize')\n",
    "    elif 'VGG16_S' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG16')\n",
    "        filter_sizes.append('halfed Filtersize')\n",
    "    elif 'VGG16_XS' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG16')\n",
    "        filter_sizes.append('quarter Filtersize')\n",
    "    elif 'VGG16_XXS' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG16')\n",
    "        filter_sizes.append('eigths Filtersize')\n",
    "    elif 'VGG16_XXXS' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG16')\n",
    "        filter_sizes.append('sixteenth Filtersize')\n",
    "    elif 'VGG19_S' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG19')\n",
    "        filter_sizes.append('halfed Filtersize')\n",
    "    elif 'VGG19_XS' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG19')\n",
    "        filter_sizes.append('quarter Filtersize')\n",
    "    elif 'VGG19_XXS' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG19')\n",
    "        filter_sizes.append('eigths Filtersize')\n",
    "    elif 'VGG19_XXXS' in all_names[i]:\n",
    "        all_networks_filtered.append('VGG19')\n",
    "        filter_sizes.append('sixteenth Filtersize')\n",
    "    else:\n",
    "        all_networks_filtered.append('LEL')\n",
    "        print(all_names[i], 'LEL')\n",
    "\n",
    "all_networks_filtered = np.array(all_networks_filtered)\n",
    "unique_networks_filtered = list(set(all_networks_filtered))\n",
    "filter_sizes = np.array(filter_sizes)\n",
    "unique_networks_filtersize = np.array(list(set(filter_sizes)))\n",
    "\n",
    "unique_problem = np.array(list(set(all_problem)))\n",
    "unique_problem = unique_problem[unique_problem != 'BIG10']\n",
    "print(all_problem)\n",
    "print(unique_problem)\n",
    "print(unique_networks_filtered)\n",
    "\n",
    "all_acc_gap = all_train_acc - all_test_acc\n",
    "all_loss_gap = (all_test_loss - all_train_loss) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "from matplotlib import pyplot as plt\n",
    "from numpy.polynomial.polynomial import polyfit\n",
    "from scipy import optimize as opt\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Test Accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from matplotlib.patches import Patch\n",
    "from matplotlib.lines import Line2D\n",
    "#b, m, n = polyfit(all_average_sat[all_problem == 'CIFAR100'], all_test_acc[all_problem == 'CIFAR100'], 2)\n",
    "b10, m10, n10 = polyfit(all_average_sat[all_problem == 'CIFAR10'], all_test_acc[all_problem == 'CIFAR10'], 2)\n",
    "b2, m2, n2 = polyfit(all_average_sat[all_problem == 'cat-vs-dog'], all_test_acc[all_problem == 'cat-vs-dog'], 2)\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(10,5))\n",
    "\n",
    "cmap = plt.get_cmap('viridis')\n",
    "\n",
    "network_legend = [Line2D([0], [0], marker='o', label='VGG11',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),\n",
    "                 Line2D([0], [0], marker='^',  label='VGG13',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),\n",
    "                 Line2D([0], [0], marker='s', label='VGG16',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),\n",
    "                 Line2D([0], [0], marker='p', label='VGG19',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),]\n",
    "\n",
    "depth_legend  = [Line2D([0], [0], color=cmap(0.0), lw=4),\n",
    "                Line2D([0], [0], color=cmap(0.25), lw=4),\n",
    "                Line2D([0], [0], color=cmap(0.5), lw=4),\n",
    "                Line2D([0], [0], color=cmap(0.75), lw=4),\n",
    "                Line2D([0], [0], color=cmap(1.0), lw=4)]\n",
    "\n",
    "x = np.arange(0, 100, 1)\n",
    "\n",
    "colors = {\n",
    "    'full Filtersize': cmap(0.0),\n",
    "    'halfed Filtersize' : cmap(0.25),\n",
    "    'quarter Filtersize' : cmap(0.5),\n",
    "    'eigths Filtersize': cmap(0.75),\n",
    "    'sixteenth Filtersize': cmap(1.0)\n",
    "}\n",
    "\n",
    "\n",
    "network_depth = {\n",
    "    'VGG11': 'o',\n",
    "    'VGG13': '^',\n",
    "    'VGG16': 's',\n",
    "    'VGG19': 'p'\n",
    "}\n",
    "\n",
    "filter_legend = []\n",
    "\n",
    "\n",
    "\n",
    "for i in range(len(all_average_sat)):\n",
    "    if 'cat-vs-dog' in all_problem[i]:\n",
    "        continue\n",
    "    ax.scatter(x=all_average_sat[i], y=all_test_acc[i], color=colors[filter_sizes[i]], marker=network_depth[all_networks_filtered[i]])\n",
    "#plt.plot(x, n*(x**2)+m*x+b, c='orange', label='CIFAR100 regression parabola')\n",
    "#ax.plot(x, n2*(x**2)+m2*x+b2, c='green')\n",
    "ax.plot(x, n10*(x**2)+m10*x+b10, c='grey')\n",
    "\n",
    "l1 = plt.legend(network_legend, ['VGG11', 'VGG13', 'VGG16', 'VGG19'], loc=1, prop={'size': 11})\n",
    "ax = plt.gca().add_artist(l1)\n",
    "plt.legend(depth_legend, ['full Filtersize', \n",
    "                           'halfed Filtersize', \n",
    "                           'quarter Filtersize', \n",
    "                          'eigths Filtersize',\n",
    "                          'sixteenth Filtersize'], loc=4, prop={'size': 11})\n",
    "plt.xlim((0,100))\n",
    "\n",
    "#for i, xy in enumerate(zip(all_average_sat, all_test_acc)):\n",
    "#    plt.annotate(all_names[i], xy)\n",
    "\n",
    "#plt.legend(prop={'size': 13})\n",
    "plt.grid()\n",
    "\n",
    "\n",
    "\n",
    "    \n",
    "#plt.title('Average Layer Saturation at Training Time versus Test Accuracy')\n",
    "plt.xlabel('Saturation in %', fontsize=16)\n",
    "plt.ylabel('Test Accuracy', fontsize=16)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from matplotlib.patches import Patch\n",
    "from matplotlib.lines import Line2D\n",
    "#b, m, n = polyfit(all_average_sat[all_problem == 'CIFAR100'], all_test_acc[all_problem == 'CIFAR100'], 2)\n",
    "b10, m10, n10 = polyfit(all_average_sat[all_problem == 'CIFAR10'], all_test_acc[all_problem == 'CIFAR10'], 2)\n",
    "b2, m2, n2 = polyfit(all_average_sat[all_problem == 'cat-vs-dog'], all_test_acc[all_problem == 'cat-vs-dog'], 2)\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(10,5))\n",
    "\n",
    "cmap = plt.get_cmap('viridis')\n",
    "\n",
    "\n",
    "network_legend = [Line2D([0], [0], marker='o', label='VGG11',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),\n",
    "                 Line2D([0], [0], marker='^',  label='VGG13',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),\n",
    "                 Line2D([0], [0], marker='s', label='VGG16',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),\n",
    "                 Line2D([0], [0], marker='p', label='VGG19',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),]\n",
    "\n",
    "depth_legend  = [Line2D([0], [0], color=cmap(0.0), lw=4),\n",
    "                Line2D([0], [0], color=cmap(0.25), lw=4),\n",
    "                Line2D([0], [0], color=cmap(0.5), lw=4),\n",
    "                Line2D([0], [0], color=cmap(0.75), lw=4),\n",
    "                Line2D([0], [0], color=cmap(1.0), lw=4)]\n",
    "\n",
    "x = np.arange(0, 100, 1)\n",
    "\n",
    "colors = {\n",
    "    'full Filtersize': cmap(0.0),\n",
    "    'halfed Filtersize' : cmap(0.25),\n",
    "    'quarter Filtersize' : cmap(0.5),\n",
    "    'eigths Filtersize': cmap(0.75),\n",
    "    'sixteenth Filtersize': cmap(1.0)\n",
    "}\n",
    "\n",
    "network_depth = {\n",
    "    'VGG11': 'o',\n",
    "    'VGG13': '^',\n",
    "    'VGG16': 's',\n",
    "    'VGG19': 'p'\n",
    "}\n",
    "\n",
    "filter_legend = []\n",
    "\n",
    "\n",
    "\n",
    "for i in range(len(all_average_sat)):\n",
    "    if 'CIFAR' in all_problem[i]:\n",
    "        continue\n",
    "    ax.scatter(x=all_average_sat[i], y=all_test_acc[i], color=colors[filter_sizes[i]], marker=network_depth[all_networks_filtered[i]])\n",
    "#plt.plot(x, n*(x**2)+m*x+b, c='orange', label='CIFAR100 regression parabola')\n",
    "ax.plot(x, n2*(x**2)+m2*x+b2, c='grey')\n",
    "#.plot(x, n10*(x**2)+m10*x+b10, c='grey')\n",
    "\n",
    "l1 = plt.legend(network_legend, ['VGG11', 'VGG13', 'VGG16', 'VGG19'], loc=1, prop={'size': 11})\n",
    "ax = plt.gca().add_artist(l1)\n",
    "plt.legend(depth_legend, ['full Filtersize', \n",
    "                           'halfed Filtersize', \n",
    "                           'quarter Filtersize', \n",
    "                          'eigths Filtersize',\n",
    "                          'sixteenth Filtersize'], loc=4, prop={'size': 11})\n",
    "\n",
    "for i, xy in enumerate(zip(all_average_sat, all_test_acc)):\n",
    "    plt.annotate(all_names[i], xy)\n",
    "plt.xlim((0,100))\n",
    "\n",
    "#plt.legend(prop={'size': 13})\n",
    "plt.grid()\n",
    "\n",
    "\n",
    "\n",
    "    \n",
    "#plt.title('Average Layer Saturation at Training Time versus Test Accuracy')\n",
    "plt.xlabel('Saturation in %', fontsize=16)\n",
    "plt.ylabel('Test Accuracy', fontsize=16)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Test Loss "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "from matplotlib.patches import Patch\n",
    "from matplotlib.lines import Line2D\n",
    "#b, m, n = polyfit(all_average_sat[all_problem == 'CIFAR100'], all_test_acc[all_problem == 'CIFAR100'], 2)\n",
    "b10, m10, n10 = polyfit(all_average_sat[all_problem == 'CIFAR10'], all_test_loss[all_problem == 'CIFAR10'], 2)\n",
    "b2, m2, n2 = polyfit(all_average_sat[all_problem == 'cat-vs-dog'], all_test_loss[all_problem == 'cat-vs-dog'], 2)\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(10,5))\n",
    "\n",
    "cmap = plt.get_cmap('viridis')\n",
    "\n",
    "network_legend = [Line2D([0], [0], marker='o', label='VGG11',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),\n",
    "                 Line2D([0], [0], marker='^',  label='VGG13',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),\n",
    "                 Line2D([0], [0], marker='s', label='VGG16',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),\n",
    "                 Line2D([0], [0], marker='p', label='VGG19',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),]\n",
    "\n",
    "depth_legend  = [Line2D([0], [0], color=cmap(0.0), lw=4),\n",
    "                Line2D([0], [0], color=cmap(0.25), lw=4),\n",
    "                Line2D([0], [0], color=cmap(0.5), lw=4),\n",
    "                Line2D([0], [0], color=cmap(0.75), lw=4),\n",
    "                Line2D([0], [0], color=cmap(1.0), lw=4)]\n",
    "\n",
    "x = np.arange(0, 100, 1)\n",
    "\n",
    "colors = {\n",
    "    'full Filtersize': cmap(0.0),\n",
    "    'halfed Filtersize' : cmap(0.25),\n",
    "    'quarter Filtersize' : cmap(0.5),\n",
    "    'eigths Filtersize': cmap(0.75),\n",
    "    'sixteenth Filtersize': cmap(1.0)\n",
    "}\n",
    "\n",
    "\n",
    "network_depth = {\n",
    "    'VGG11': 'o',\n",
    "    'VGG13': '^',\n",
    "    'VGG16': 's',\n",
    "    'VGG19': 'p'\n",
    "}\n",
    "\n",
    "filter_legend = []\n",
    "\n",
    "\n",
    "\n",
    "for i in range(len(all_average_sat)):\n",
    "    if 'cat-vs-dog' in all_problem[i]:\n",
    "        continue\n",
    "    ax.scatter(x=all_average_sat[i], y=all_test_loss[i], color=colors[filter_sizes[i]], marker=network_depth[all_networks_filtered[i]])\n",
    "#plt.plot(x, n*(x**2)+m*x+b, c='orange', label='CIFAR100 regression parabola')\n",
    "#ax.plot(x, n2*(x**2)+m2*x+b2, c='green')\n",
    "ax.plot(x, n10*(x**2)+m10*x+b10, c='grey')\n",
    "\n",
    "l1 = plt.legend(network_legend, ['VGG11', 'VGG13', 'VGG16', 'VGG19'], loc=1, prop={'size': 11})\n",
    "ax = plt.gca().add_artist(l1)\n",
    "plt.legend(depth_legend, ['full Filtersize', \n",
    "                           'halfed Filtersize', \n",
    "                           'quarter Filtersize', \n",
    "                          'eigths Filtersize',\n",
    "                          'sixteenth Filtersize'], loc=4, prop={'size': 11})\n",
    "\n",
    "#for i, xy in enumerate(zip(all_average_sat, all_test_acc)):\n",
    "#    plt.annotate(all_names[i], xy)\n",
    "plt.xlim((0,100))\n",
    "\n",
    "#plt.legend(prop={'size': 13})\n",
    "plt.grid()\n",
    "\n",
    "\n",
    "\n",
    "    \n",
    "#plt.title('Average Layer Saturation at Training Time versus Test Accuracy')\n",
    "plt.xlabel('Saturation in %', fontsize=16)\n",
    "plt.ylabel('Test Loss', fontsize=16)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "from matplotlib.patches import Patch\n",
    "from matplotlib.lines import Line2D\n",
    "#b, m, n = polyfit(all_average_sat[all_problem == 'CIFAR100'], all_test_acc[all_problem == 'CIFAR100'], 2)\n",
    "b10, m10, n10 = polyfit(all_average_sat[all_problem == 'CIFAR10'], all_test_loss[all_problem == 'CIFAR10'], 2)\n",
    "b2, m2, n2 = polyfit(all_average_sat[all_problem == 'cat-vs-dog'], all_test_loss[all_problem == 'cat-vs-dog'], 2)\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(10,5))\n",
    "\n",
    "cmap = plt.get_cmap('viridis')\n",
    "\n",
    "\n",
    "network_legend = [Line2D([0], [0], marker='o', label='VGG11',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),\n",
    "                 Line2D([0], [0], marker='^',  label='VGG13',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),\n",
    "                 Line2D([0], [0], marker='s', label='VGG16',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),\n",
    "                 Line2D([0], [0], marker='p', label='VGG19',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),]\n",
    "\n",
    "depth_legend  = [Line2D([0], [0], color=cmap(0.0), lw=4),\n",
    "                Line2D([0], [0], color=cmap(0.25), lw=4),\n",
    "                Line2D([0], [0], color=cmap(0.5), lw=4),\n",
    "                Line2D([0], [0], color=cmap(0.75), lw=4),\n",
    "                Line2D([0], [0], color=cmap(1.0), lw=4)]\n",
    "\n",
    "x = np.arange(0, 100, 1)\n",
    "\n",
    "colors = {\n",
    "    'full Filtersize': cmap(0.0),\n",
    "    'halfed Filtersize' : cmap(0.25),\n",
    "    'quarter Filtersize' : cmap(0.5),\n",
    "    'eigths Filtersize': cmap(0.75),\n",
    "    'sixteenth Filtersize': cmap(1.0)\n",
    "}\n",
    "\n",
    "network_depth = {\n",
    "    'VGG11': 'o',\n",
    "    'VGG13': '^',\n",
    "    'VGG16': 's',\n",
    "    'VGG19': 'p'\n",
    "}\n",
    "\n",
    "filter_legend = []\n",
    "\n",
    "\n",
    "\n",
    "for i in range(len(all_average_sat)):\n",
    "    if 'CIFAR' in all_problem[i]:\n",
    "        continue\n",
    "    ax.scatter(x=all_average_sat[i], y=all_test_loss[i], color=colors[filter_sizes[i]], marker=network_depth[all_networks_filtered[i]])\n",
    "#plt.plot(x, n*(x**2)+m*x+b, c='orange', label='CIFAR100 regression parabola')\n",
    "ax.plot(x, n2*(x**2)+m2*x+b2, c='grey')\n",
    "#plt.plot(x, n10*(x**2)+m10*x+b10, c='grey')\n",
    "\n",
    "l1 = plt.legend(network_legend, ['VGG11', 'VGG13', 'VGG16', 'VGG19'], loc=1, prop={'size': 11})\n",
    "ax = plt.gca().add_artist(l1)\n",
    "plt.legend(depth_legend, ['full Filtersize', \n",
    "                           'halfed Filtersize', \n",
    "                           'quarter Filtersize', \n",
    "                          'eigths Filtersize',\n",
    "                          'sixteenth Filtersize'], loc=4, prop={'size': 11})\n",
    "\n",
    "plt.xlim((0,100))\n",
    "\n",
    "#for i, xy in enumerate(zip(all_average_sat, all_test_acc)):\n",
    "#    plt.annotate(all_names[i], xy)\n",
    "\n",
    "#plt.legend(prop={'size': 13})\n",
    "plt.grid()\n",
    "\n",
    "\n",
    "\n",
    "    \n",
    "#plt.title('Average Layer Saturation at Training Time versus Test Accuracy')\n",
    "plt.xlabel('Saturation in %', fontsize=16)\n",
    "plt.ylabel('Test Loss', fontsize=16)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Train Loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "from matplotlib.patches import Patch\n",
    "from matplotlib.lines import Line2D\n",
    "#b, m, n = polyfit(all_average_sat[all_problem == 'CIFAR100'], all_test_acc[all_problem == 'CIFAR100'], 2)\n",
    "b10, m10, n10 = polyfit(all_average_sat[all_problem == 'CIFAR10'], all_train_loss[all_problem == 'CIFAR10'], 2)\n",
    "b2, m2, n2 = polyfit(all_average_sat[all_problem == 'cat-vs-dog'], all_train_loss[all_problem == 'cat-vs-dog'], 2)\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(10,5))\n",
    "\n",
    "cmap = plt.get_cmap('viridis')\n",
    "\n",
    "network_legend = [Line2D([0], [0], marker='o', label='VGG11',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),\n",
    "                 Line2D([0], [0], marker='^',  label='VGG13',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),\n",
    "                 Line2D([0], [0], marker='s', label='VGG16',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),\n",
    "                 Line2D([0], [0], marker='p', label='VGG19',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),]\n",
    "\n",
    "depth_legend  = [Line2D([0], [0], color=cmap(0.0), lw=4),\n",
    "                Line2D([0], [0], color=cmap(0.25), lw=4),\n",
    "                Line2D([0], [0], color=cmap(0.5), lw=4),\n",
    "                Line2D([0], [0], color=cmap(0.75), lw=4),\n",
    "                Line2D([0], [0], color=cmap(1.0), lw=4)]\n",
    "\n",
    "x = np.arange(0, 100, 1)\n",
    "\n",
    "colors = {\n",
    "    'full Filtersize': cmap(0.0),\n",
    "    'halfed Filtersize' : cmap(0.25),\n",
    "    'quarter Filtersize' : cmap(0.5),\n",
    "    'eigths Filtersize': cmap(0.75),\n",
    "    'sixteenth Filtersize': cmap(1.0)\n",
    "}\n",
    "\n",
    "\n",
    "network_depth = {\n",
    "    'VGG11': 'o',\n",
    "    'VGG13': '^',\n",
    "    'VGG16': 's',\n",
    "    'VGG19': 'p'\n",
    "}\n",
    "\n",
    "filter_legend = []\n",
    "\n",
    "\n",
    "\n",
    "for i in range(len(all_average_sat)):\n",
    "    if 'cat-vs-dog' in all_problem[i]:\n",
    "        continue\n",
    "    ax.scatter(x=all_average_sat[i], y=all_train_loss[i], color=colors[filter_sizes[i]], marker=network_depth[all_networks_filtered[i]])\n",
    "#plt.plot(x, n*(x**2)+m*x+b, c='orange', label='CIFAR100 regression parabola')\n",
    "#ax.plot(x, n2*(x**2)+m2*x+b2, c='green')\n",
    "ax.plot(x, n10*(x**2)+m10*x+b10, c='grey')\n",
    "\n",
    "l1 = plt.legend(network_legend, ['VGG11', 'VGG13', 'VGG16', 'VGG19'], loc=1, prop={'size': 11})\n",
    "ax = plt.gca().add_artist(l1)\n",
    "plt.legend(depth_legend, ['full Filtersize', \n",
    "                           'halfed Filtersize', \n",
    "                           'quarter Filtersize', \n",
    "                          'eigths Filtersize',\n",
    "                          'sixteenth Filtersize'], loc=4, prop={'size': 11})\n",
    "\n",
    "#for i, xy in enumerate(zip(all_average_sat, all_test_acc)):\n",
    "#    plt.annotate(all_names[i], xy)\n",
    "plt.xlim((0,100))\n",
    "\n",
    "#plt.legend(prop={'size': 13})\n",
    "plt.grid()\n",
    "\n",
    "\n",
    "\n",
    "    \n",
    "#plt.title('Average Layer Saturation at Training Time versus Test Accuracy')\n",
    "plt.xlabel('Saturation in %', fontsize=16)\n",
    "plt.ylabel('Test Loss', fontsize=16)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "from matplotlib.patches import Patch\n",
    "from matplotlib.lines import Line2D\n",
    "#b, m, n = polyfit(all_average_sat[all_problem == 'CIFAR100'], all_test_acc[all_problem == 'CIFAR100'], 2)\n",
    "b10, m10, n10 = polyfit(all_average_sat[all_problem == 'CIFAR10'], all_train_loss[all_problem == 'CIFAR10'], 2)\n",
    "b2, m2, n2 = polyfit(all_average_sat[all_problem == 'cat-vs-dog'], all_train_loss[all_problem == 'cat-vs-dog'], 2)\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(10,5))\n",
    "\n",
    "cmap = plt.get_cmap('viridis')\n",
    "\n",
    "\n",
    "network_legend = [Line2D([0], [0], marker='o', label='VGG11',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),\n",
    "                 Line2D([0], [0], marker='^',  label='VGG13',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),\n",
    "                 Line2D([0], [0], marker='s', label='VGG16',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),\n",
    "                 Line2D([0], [0], marker='p', label='VGG19',\n",
    "                         markerfacecolor='black', \n",
    "                         markersize=10),]\n",
    "\n",
    "depth_legend  = [Line2D([0], [0], color=cmap(0.0), lw=4),\n",
    "                Line2D([0], [0], color=cmap(0.25), lw=4),\n",
    "                Line2D([0], [0], color=cmap(0.5), lw=4),\n",
    "                Line2D([0], [0], color=cmap(0.75), lw=4),\n",
    "                Line2D([0], [0], color=cmap(1.0), lw=4)]\n",
    "\n",
    "x = np.arange(0, 100, 1)\n",
    "\n",
    "colors = {\n",
    "    'full Filtersize': cmap(0.0),\n",
    "    'halfed Filtersize' : cmap(0.25),\n",
    "    'quarter Filtersize' : cmap(0.5),\n",
    "    'eigths Filtersize': cmap(0.75),\n",
    "    'sixteenth Filtersize': cmap(1.0)\n",
    "}\n",
    "\n",
    "network_depth = {\n",
    "    'VGG11': 'o',\n",
    "    'VGG13': '^',\n",
    "    'VGG16': 's',\n",
    "    'VGG19': 'p'\n",
    "}\n",
    "\n",
    "filter_legend = []\n",
    "\n",
    "\n",
    "\n",
    "for i in range(len(all_average_sat)):\n",
    "    if 'CIFAR' in all_problem[i]:\n",
    "        continue\n",
    "    ax.scatter(x=all_average_sat[i], y=all_train_loss[i], color=colors[filter_sizes[i]], marker=network_depth[all_networks_filtered[i]])\n",
    "#plt.plot(x, n*(x**2)+m*x+b, c='orange', label='CIFAR100 regression parabola')\n",
    "ax.plot(x, n2*(x**2)+m2*x+b2, c='grey')\n",
    "#plt.plot(x, n10*(x**2)+m10*x+b10, c='grey')\n",
    "\n",
    "l1 = plt.legend(network_legend, ['VGG11', 'VGG13', 'VGG16', 'VGG19'], loc=1, prop={'size': 11})\n",
    "ax = plt.gca().add_artist(l1)\n",
    "plt.legend(depth_legend, ['full Filtersize', \n",
    "                           'halfed Filtersize', \n",
    "                           'quarter Filtersize', \n",
    "                          'eigths Filtersize',\n",
    "                          'sixteenth Filtersize'], loc=4, prop={'size': 11})\n",
    "\n",
    "plt.xlim((0,100))\n",
    "\n",
    "#for i, xy in enumerate(zip(all_average_sat, all_test_acc)):\n",
    "#    plt.annotate(all_names[i], xy)\n",
    "\n",
    "#plt.legend(prop={'size': 13})\n",
    "plt.grid()\n",
    "\n",
    "\n",
    "\n",
    "    \n",
    "#plt.title('Average Layer Saturation at Training Time versus Test Accuracy')\n",
    "plt.xlabel('Saturation in %', fontsize=16)\n",
    "plt.ylabel('Train Loss', fontsize=16)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "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": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}