trainer.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
    "tf.keras.backend.clear_session()\n",
    "from tensorflow.keras.layers import Input, Softmax, Dense, Dropout, BatchNormalization\n",
    "from tensorflow.keras.models import Model\n",
    "import numpy as np\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from tensorflow.keras import models, layers, optimizers, regularizers, initializers\n",
    "import matplotlib.pyplot as plt\n",
    "import random\n",
    "from IPython.display import clear_output\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "BATCH_SIZE = 256\n",
    "SEED = 23582068\n",
    "tf.random.set_seed(SEED)\n",
    "np.random.seed(SEED)\n",
    "random.seed(SEED)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "47258\n",
      "48745\n",
      "47499\n",
      "143502\n",
      "559\n",
      "559\n",
      "697\n",
      "697\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'eddie_good0_scaled_kpts_3D_1989.npy': 0,\n",
       " 'chris_ok_scaled_kpts_3D_486.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_513.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_1629.npy': 1,\n",
       " 'connor_bad_scaled_kpts_3D_3073.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_929.npy': 1,\n",
       " 'chris_buff_3d_8177_ok.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_627.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_3131.npy': 0,\n",
       " 'daniel_bad_scaled_kpts_3D_1915.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_4093.npy': 1,\n",
       " 'connor_good_scaled_kpts_3D_599.npy': 0,\n",
       " 'connor_good_scaled_kpts_3D_603.npy': 0,\n",
       " 'isaac_ok_scaled_kpts_3D_3289.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_3808.npy': 0,\n",
       " 'michael_bad_scaled_kpts_3D_1550.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_1182.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_1199.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_1673.npy': 0,\n",
       " 'connor_bad_scaled_kpts_3D_771.npy': 1,\n",
       " 'connor_bad_scaled_kpts_3D_2474.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_3390.npy': 0,\n",
       " 'eddie_ok_scaled_kpts_3D_759.npy': 1,\n",
       " 'eddie_bad0_scaled_kpts_3D_3431.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_1596.npy': 0,\n",
       " 'eddie_bad0_scaled_kpts_3D_1639.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_3234.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_1277.npy': 1,\n",
       " 'daniel_ok_scaled_kpts_3D_1500.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_76.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_23.npy': 1,\n",
       " 'connor_bad_scaled_kpts_3D_1296.npy': 1,\n",
       " 'connor_bad_scaled_kpts_3D_3021.npy': 1,\n",
       " 'amruth_bad1_3D_3822.npy': 1,\n",
       " 'chris_bad0_scaled_kpts_3D_2431.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_1617.npy': 0,\n",
       " 'amruthgood2_scaled_kpts_3D_3031.npy': 0,\n",
       " 'michael_good_scaled_kpts_3D_2845.npy': 0,\n",
       " 'connor_ok_scaled_kpts_3D_1408.npy': 1,\n",
       " 'chris_buff_3d_3016_ok.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_4450.npy': 0,\n",
       " 'chris_bad1_scaled_kpts_3D_3627.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_422.npy': 0,\n",
       " 'isaac_ok_scaled_kpts_3D_4498.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_3479.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_281.npy': 0,\n",
       " 'daniel_bad_scaled_kpts_3D_3046.npy': 1,\n",
       " 'chris_buff_3d_9680_ok.npy': 1,\n",
       " 'daniel_ok_scaled_kpts_3D_2488.npy': 1,\n",
       " 'chris_buff_3d_5225_ok.npy': 1,\n",
       " 'chris_buff_3d_1622_ok.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_2728.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_4334.npy': 0,\n",
       " 'daniel_bad_scaled_kpts_3D_3254.npy': 1,\n",
       " 'chris_bad1_scaled_kpts_3D_1367.npy': 1,\n",
       " 'connor_bad_scaled_kpts_3D_1681.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_113.npy': 0,\n",
       " 'michael_good_scaled_kpts_3D_580.npy': 0,\n",
       " 'amruthok1_scaled_kpts_3D_2069.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_3107.npy': 1,\n",
       " 'daniel_ok_scaled_kpts_3D_2650.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_2552.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_4018.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_965.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_82.npy': 0,\n",
       " 'amruth_bad1_3D_2572.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_305.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_1027.npy': 0,\n",
       " 'connor_bad_scaled_kpts_3D_20.npy': 1,\n",
       " 'chris_buff_3d_7350_ok.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_2749.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_3793.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_2843.npy': 0,\n",
       " 'eddie_good0_scaled_kpts_3D_4021.npy': 0,\n",
       " 'chris_buff_3d_1312_ok.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_2283.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_497.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_3111.npy': 0,\n",
       " 'eddie_good1_scaled_kpts_3D_3032.npy': 0,\n",
       " 'amruthok2_scaled_kpts_3D_3066.npy': 1,\n",
       " 'chris_good1_scaled_kpts_3D_1953.npy': 0,\n",
       " 'amruthok2_scaled_kpts_3D_2364.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_3358.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_1215.npy': 1,\n",
       " 'chris_good1_scaled_kpts_3D_2510.npy': 0,\n",
       " 'chris_bad0_scaled_kpts_3D_1492.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_3205.npy': 0,\n",
       " 'isaac_good_2_scaled_kpts_3D_1747.npy': 0,\n",
       " 'chris_bad0_scaled_kpts_3D_1355.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_2039.npy': 1,\n",
       " 'eddie_ok_scaled_kpts_3D_1943.npy': 1,\n",
       " 'daniel_ok_scaled_kpts_3D_982.npy': 1,\n",
       " 'eddie_bad0_scaled_kpts_3D_2318.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_3632.npy': 0,\n",
       " 'amruthok1_scaled_kpts_3D_3083.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_4061.npy': 0,\n",
       " 'eddie_bad0_scaled_kpts_3D_856.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_2238.npy': 0,\n",
       " 'chris_good0_scaled_kpts_3D_3935.npy': 0,\n",
       " 'michael_bad_scaled_kpts_3D_2420.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_3354.npy': 0,\n",
       " 'michael_ok_scaled_kpts_3D_2538.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_3327.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_1432.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_1836.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_400.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_3351.npy': 1,\n",
       " 'chris_buff_3d_1236_ok.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_1276.npy': 0,\n",
       " 'isaac_ok_scaled_kpts_3D_1570.npy': 1,\n",
       " 'eddie_bad1_scaled_kpts_3D_1614.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_1358.npy': 1,\n",
       " 'eddie_bad0_scaled_kpts_3D_3444.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_3910.npy': 0,\n",
       " 'chris_ok_scaled_kpts_3D_4263.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_1490.npy': 0,\n",
       " 'connor_ok_scaled_kpts_3D_3240.npy': 1,\n",
       " 'eddie_ok_scaled_kpts_3D_180.npy': 1,\n",
       " 'chris_bad1_scaled_kpts_3D_641.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_2550.npy': 0,\n",
       " 'michael_good_scaled_kpts_3D_2051.npy': 0,\n",
       " 'connor_bad_scaled_kpts_3D_2883.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_4064.npy': 0,\n",
       " 'daniel_good_scaled_kpts_3D_210.npy': 0,\n",
       " 'amruthok2_scaled_kpts_3D_2328.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_3124.npy': 0,\n",
       " 'eddie_good1_scaled_kpts_3D_3156.npy': 0,\n",
       " 'chris_bad0_scaled_kpts_3D_1998.npy': 1,\n",
       " 'eddie_ok_scaled_kpts_3D_457.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_1486.npy': 0,\n",
       " 'chris_good1_scaled_kpts_3D_687.npy': 0,\n",
       " 'daniel_bad_scaled_kpts_3D_682.npy': 1,\n",
       " 'eddie_bad1_scaled_kpts_3D_3416.npy': 1,\n",
       " 'chris_good1_scaled_kpts_3D_1003.npy': 0,\n",
       " 'isaac_ok_scaled_kpts_3D_2200.npy': 1,\n",
       " 'eddie_ok_scaled_kpts_3D_240.npy': 1,\n",
       " 'isaac_good_scaled_kpts_3D_1195.npy': 0,\n",
       " 'eddie_bad1_scaled_kpts_3D_3510.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_323.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_1772.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_3049.npy': 0,\n",
       " 'isaac_good_scaled_kpts_3D_467.npy': 0,\n",
       " 'amruthbad2_scaled_kpts_3D_2230.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_3015.npy': 0,\n",
       " 'amruthgood1_scaled_kpts_3D_4071.npy': 0,\n",
       " 'chris_ok_scaled_kpts_3D_3975.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_3760.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_168.npy': 0,\n",
       " 'isaac_ok_scaled_kpts_3D_316.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_1948.npy': 1,\n",
       " 'isaac_bad_2_scaled_kpts_3D_3139.npy': 1,\n",
       " 'daniel_bad_scaled_kpts_3D_409.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_3272.npy': 1,\n",
       " 'connor_bad_scaled_kpts_3D_2666.npy': 1,\n",
       " 'chris_buff_3d_6764_ok.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_2714.npy': 0,\n",
       " 'chris_good0_scaled_kpts_3D_688.npy': 0,\n",
       " 'chris_bad0_scaled_kpts_3D_184.npy': 1,\n",
       " 'isaac_good_scaled_kpts_3D_1373.npy': 0,\n",
       " 'isaac_ok_2_scaled_kpts_3D_3865.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_2938.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_380.npy': 0,\n",
       " 'eddie_good1_scaled_kpts_3D_2199.npy': 0,\n",
       " 'isaac_good_2_scaled_kpts_3D_2488.npy': 0,\n",
       " 'chris_good1_scaled_kpts_3D_3378.npy': 0,\n",
       " 'eddie_good0_scaled_kpts_3D_638.npy': 0,\n",
       " 'isaac_bad_2_scaled_kpts_3D_2345.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_357.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_3264.npy': 0,\n",
       " 'eddie_good1_scaled_kpts_3D_2476.npy': 0,\n",
       " 'eddie_bad0_scaled_kpts_3D_2494.npy': 1,\n",
       " 'chris_bad1_scaled_kpts_3D_1423.npy': 1,\n",
       " 'eddie_bad0_scaled_kpts_3D_3738.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_787.npy': 0,\n",
       " 'eddie_bad1_scaled_kpts_3D_2988.npy': 1,\n",
       " 'chris_buff_3d_1117_ok.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_604.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_1561.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_751.npy': 1,\n",
       " 'chris_good1_scaled_kpts_3D_3128.npy': 0,\n",
       " 'chris_good0_scaled_kpts_3D_2634.npy': 0,\n",
       " 'isaac_good_scaled_kpts_3D_1183.npy': 0,\n",
       " 'amruthok1_scaled_kpts_3D_4109.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_2377.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_3132.npy': 0,\n",
       " 'eddie_good1_scaled_kpts_3D_2002.npy': 0,\n",
       " 'daniel_ok_scaled_kpts_3D_2784.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_1442.npy': 0,\n",
       " 'amruthok1_scaled_kpts_3D_4087.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_874.npy': 0,\n",
       " 'isaac_ok_2_scaled_kpts_3D_1904.npy': 1,\n",
       " 'chris_buff_3d_1993_ok.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_976.npy': 1,\n",
       " 'chris_bad1_scaled_kpts_3D_1927.npy': 1,\n",
       " 'isaac_good_scaled_kpts_3D_4089.npy': 0,\n",
       " 'connor_ok_scaled_kpts_3D_3167.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_398.npy': 1,\n",
       " 'amruth_bad1_3D_1690.npy': 1,\n",
       " 'chris_buff_3d_6318_ok.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_2284.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_2353.npy': 0,\n",
       " 'chris_good0_scaled_kpts_3D_2486.npy': 0,\n",
       " 'chris_bad1_scaled_kpts_3D_2833.npy': 1,\n",
       " 'daniel_bad_scaled_kpts_3D_695.npy': 1,\n",
       " 'chris_bad0_scaled_kpts_3D_651.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_3930.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_3189.npy': 0,\n",
       " 'isaac_good_scaled_kpts_3D_3243.npy': 0,\n",
       " 'eddie_bad1_scaled_kpts_3D_465.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_2569.npy': 0,\n",
       " 'amruthok2_scaled_kpts_3D_3534.npy': 1,\n",
       " 'isaac_bad_2_scaled_kpts_3D_4206.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_1866.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_2877.npy': 0,\n",
       " 'amruthgood1_scaled_kpts_3D_334.npy': 0,\n",
       " 'daniel_ok_scaled_kpts_3D_1631.npy': 1,\n",
       " 'isaac_good_scaled_kpts_3D_20.npy': 0,\n",
       " 'amruthgood1_scaled_kpts_3D_5.npy': 0,\n",
       " 'chris_ok_scaled_kpts_3D_3627.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_1257.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_2931.npy': 1,\n",
       " 'amruth_bad1_3D_1709.npy': 1,\n",
       " 'isaac_good_scaled_kpts_3D_2151.npy': 0,\n",
       " 'eddie_bad0_scaled_kpts_3D_1358.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_4158.npy': 1,\n",
       " 'chris_bad0_scaled_kpts_3D_2428.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_388.npy': 0,\n",
       " 'amruthok1_scaled_kpts_3D_3435.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_3234.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_3361.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_2391.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_3520.npy': 0,\n",
       " 'eddie_ok_scaled_kpts_3D_3096.npy': 1,\n",
       " 'eddie_bad0_scaled_kpts_3D_2085.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_2088.npy': 1,\n",
       " 'eddie_ok_scaled_kpts_3D_4280.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_4244.npy': 1,\n",
       " 'chris_bad1_scaled_kpts_3D_333.npy': 1,\n",
       " 'connor_good_scaled_kpts_3D_1851.npy': 0,\n",
       " 'isaac_bad_2_scaled_kpts_3D_3506.npy': 1,\n",
       " 'daniel_ok_scaled_kpts_3D_2380.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_2272.npy': 0,\n",
       " 'michael_bad_scaled_kpts_3D_872.npy': 1,\n",
       " 'connor_bad_scaled_kpts_3D_496.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_1442.npy': 0,\n",
       " 'isaac_bad_2_scaled_kpts_3D_768.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_3517.npy': 0,\n",
       " 'daniel_ok_scaled_kpts_3D_2815.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_1177.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_2374.npy': 0,\n",
       " 'daniel_bad_scaled_kpts_3D_1640.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_3148.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_4074.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_1711.npy': 0,\n",
       " 'isaac_ok_scaled_kpts_3D_3525.npy': 1,\n",
       " 'daniel_ok_scaled_kpts_3D_3065.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_3321.npy': 0,\n",
       " 'chris_buff_3d_255_ok.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_3643.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_3101.npy': 1,\n",
       " 'chris_buff_3d_7177_ok.npy': 1,\n",
       " 'isaac_good_scaled_kpts_3D_1018.npy': 0,\n",
       " 'amruthgood2_scaled_kpts_3D_2070.npy': 0,\n",
       " 'eddie_bad0_scaled_kpts_3D_3800.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_1308.npy': 0,\n",
       " 'chris_good1_scaled_kpts_3D_4377.npy': 0,\n",
       " 'amruthgood2_scaled_kpts_3D_2143.npy': 0,\n",
       " 'chris_bad0_scaled_kpts_3D_2663.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_3354.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_3727.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_3209.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_1443.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_749.npy': 0,\n",
       " 'connor_bad_scaled_kpts_3D_221.npy': 1,\n",
       " 'amruth_bad1_3D_1086.npy': 1,\n",
       " 'chris_good1_scaled_kpts_3D_1685.npy': 0,\n",
       " 'michael_good_scaled_kpts_3D_624.npy': 0,\n",
       " 'amruthgood2_scaled_kpts_3D_2685.npy': 0,\n",
       " 'chris_buff_3d_8614_ok.npy': 1,\n",
       " 'chris_bad1_scaled_kpts_3D_2138.npy': 1,\n",
       " 'chris_buff_3d_3671_ok.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_2250.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_3776.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_2139.npy': 1,\n",
       " 'eddie_bad1_scaled_kpts_3D_3298.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_4102.npy': 0,\n",
       " 'connor_good_scaled_kpts_3D_2484.npy': 0,\n",
       " 'connor_good_scaled_kpts_3D_1310.npy': 0,\n",
       " 'amruthok1_scaled_kpts_3D_4093.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_1317.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_778.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_3688.npy': 1,\n",
       " 'connor_good_scaled_kpts_3D_1877.npy': 0,\n",
       " 'connor_ok_scaled_kpts_3D_515.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_717.npy': 1,\n",
       " 'connor_good_scaled_kpts_3D_230.npy': 0,\n",
       " 'isaac_ok_2_scaled_kpts_3D_1240.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_1902.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_2714.npy': 0,\n",
       " 'isaac_good_2_scaled_kpts_3D_2522.npy': 0,\n",
       " 'daniel_bad_scaled_kpts_3D_2136.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_1022.npy': 0,\n",
       " 'eddie_bad1_scaled_kpts_3D_1236.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_1632.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_97.npy': 0,\n",
       " 'michael_ok_scaled_kpts_3D_3122.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_1156.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_1513.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_2913.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_4247.npy': 0,\n",
       " 'michael_ok_scaled_kpts_3D_692.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_2471.npy': 0,\n",
       " 'chris_good0_scaled_kpts_3D_881.npy': 0,\n",
       " 'chris_ok_scaled_kpts_3D_3210.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_2767.npy': 0,\n",
       " 'amruthbad2_scaled_kpts_3D_321.npy': 1,\n",
       " 'chris_bad0_scaled_kpts_3D_1961.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_3328.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_3815.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_1136.npy': 0,\n",
       " 'isaac_good_scaled_kpts_3D_3188.npy': 0,\n",
       " 'eddie_good1_scaled_kpts_3D_3476.npy': 0,\n",
       " 'chris_buff_3d_8034_ok.npy': 1,\n",
       " 'daniel_ok_scaled_kpts_3D_1984.npy': 1,\n",
       " 'amruth_bad1_3D_4242.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_280.npy': 0,\n",
       " 'eddie_good1_scaled_kpts_3D_1052.npy': 0,\n",
       " 'connor_ok_scaled_kpts_3D_2139.npy': 1,\n",
       " 'isaac_good_scaled_kpts_3D_3631.npy': 0,\n",
       " 'eddie_bad1_scaled_kpts_3D_3376.npy': 1,\n",
       " 'eddie_ok_scaled_kpts_3D_742.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_2859.npy': 0,\n",
       " 'amruth_bad1_3D_934.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_1500.npy': 0,\n",
       " 'chris_buff_3d_4232_ok.npy': 1,\n",
       " 'eddie_bad0_scaled_kpts_3D_984.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_490.npy': 0,\n",
       " 'amruthok2_scaled_kpts_3D_588.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_3072.npy': 1,\n",
       " 'isaac_bad_2_scaled_kpts_3D_3414.npy': 1,\n",
       " 'chris_buff_3d_4254_ok.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_3679.npy': 0,\n",
       " 'isaac_ok_2_scaled_kpts_3D_2595.npy': 1,\n",
       " 'connor_bad_scaled_kpts_3D_64.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_3937.npy': 1,\n",
       " 'chris_bad1_scaled_kpts_3D_2791.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_2755.npy': 0,\n",
       " 'eddie_bad0_scaled_kpts_3D_2550.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_3860.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_731.npy': 1,\n",
       " 'daniel_bad_scaled_kpts_3D_1036.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_153.npy': 0,\n",
       " 'michael_bad_scaled_kpts_3D_3353.npy': 1,\n",
       " 'eddie_ok_scaled_kpts_3D_456.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_2081.npy': 1,\n",
       " 'chris_buff_3d_4734_ok.npy': 1,\n",
       " 'isaac_good_scaled_kpts_3D_794.npy': 0,\n",
       " 'chris_bad1_scaled_kpts_3D_543.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_1078.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_2940.npy': 0,\n",
       " 'isaac_bad_scaled_kpts_3D_2569.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_2298.npy': 0,\n",
       " 'amruth_bad1_3D_1231.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_3710.npy': 0,\n",
       " 'chris_good0_scaled_kpts_3D_2045.npy': 0,\n",
       " 'chris_buff_3d_8166_ok.npy': 1,\n",
       " 'chris_buff_3d_549_ok.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_3377.npy': 0,\n",
       " 'eddie_ok_scaled_kpts_3D_3688.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_1334.npy': 1,\n",
       " 'daniel_bad_scaled_kpts_3D_159.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_832.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_2362.npy': 1,\n",
       " 'daniel_bad_scaled_kpts_3D_1692.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_1769.npy': 0,\n",
       " 'chris_good1_scaled_kpts_3D_2563.npy': 0,\n",
       " 'eddie_ok_scaled_kpts_3D_1775.npy': 1,\n",
       " 'amruth_bad1_3D_915.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_4538.npy': 0,\n",
       " 'daniel_bad_scaled_kpts_3D_1041.npy': 1,\n",
       " 'chris_good1_scaled_kpts_3D_1387.npy': 0,\n",
       " 'amruthbad2_scaled_kpts_3D_3573.npy': 1,\n",
       " 'connor_bad_scaled_kpts_3D_1938.npy': 1,\n",
       " 'chris_bad0_scaled_kpts_3D_3629.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_3944.npy': 0,\n",
       " 'isaac_ok_scaled_kpts_3D_4579.npy': 1,\n",
       " 'chris_buff_3d_2465_ok.npy': 1,\n",
       " 'chris_buff_3d_7051_ok.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_191.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_3333.npy': 1,\n",
       " 'connor_good_scaled_kpts_3D_3300.npy': 0,\n",
       " 'eddie_ok_scaled_kpts_3D_4209.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_447.npy': 1,\n",
       " 'chris_bad0_scaled_kpts_3D_4124.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_4174.npy': 0,\n",
       " 'isaac_ok_scaled_kpts_3D_2848.npy': 1,\n",
       " 'eddie_bad1_scaled_kpts_3D_1266.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_2813.npy': 0,\n",
       " 'michael_bad_scaled_kpts_3D_3068.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_1322.npy': 0,\n",
       " 'connor_bad_scaled_kpts_3D_1932.npy': 1,\n",
       " 'daniel_ok_scaled_kpts_3D_1117.npy': 1,\n",
       " 'eddie_bad0_scaled_kpts_3D_3944.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_1640.npy': 0,\n",
       " 'amruthok2_scaled_kpts_3D_2081.npy': 1,\n",
       " 'eddie_ok_scaled_kpts_3D_3601.npy': 1,\n",
       " 'eddie_bad0_scaled_kpts_3D_2671.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_1017.npy': 1,\n",
       " 'eddie_ok_scaled_kpts_3D_3600.npy': 1,\n",
       " 'eddie_bad1_scaled_kpts_3D_4123.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_2132.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_885.npy': 0,\n",
       " 'amruthgood2_scaled_kpts_3D_1635.npy': 0,\n",
       " 'isaac_bad_2_scaled_kpts_3D_273.npy': 1,\n",
       " 'eddie_bad0_scaled_kpts_3D_3241.npy': 1,\n",
       " 'chris_good1_scaled_kpts_3D_2791.npy': 0,\n",
       " 'connor_bad_scaled_kpts_3D_2211.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_503.npy': 0,\n",
       " 'chris_buff_3d_10089_ok.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_1746.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_866.npy': 0,\n",
       " 'eddie_good0_scaled_kpts_3D_2000.npy': 0,\n",
       " 'chris_good1_scaled_kpts_3D_2567.npy': 0,\n",
       " 'daniel_bad_scaled_kpts_3D_3548.npy': 1,\n",
       " 'chris_buff_3d_10437_ok.npy': 1,\n",
       " 'daniel_ok_scaled_kpts_3D_2132.npy': 1,\n",
       " 'connor_good_scaled_kpts_3D_2661.npy': 0,\n",
       " 'eddie_ok_scaled_kpts_3D_3311.npy': 1,\n",
       " 'chris_buff_3d_3552_ok.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_3479.npy': 0,\n",
       " 'connor_ok_scaled_kpts_3D_552.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_3104.npy': 0,\n",
       " 'eddie_bad1_scaled_kpts_3D_1766.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_4467.npy': 0,\n",
       " 'isaac_ok_scaled_kpts_3D_2281.npy': 1,\n",
       " 'chris_buff_3d_9705_ok.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_1258.npy': 0,\n",
       " 'isaac_good_2_scaled_kpts_3D_3666.npy': 0,\n",
       " 'eddie_bad1_scaled_kpts_3D_1803.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_2769.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_2659.npy': 1,\n",
       " 'chris_buff_3d_10312_ok.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_3387.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_1462.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_307.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_1198.npy': 0,\n",
       " 'chris_good0_scaled_kpts_3D_1860.npy': 0,\n",
       " 'michael_bad_scaled_kpts_3D_1246.npy': 1,\n",
       " 'chris_buff_3d_7480_ok.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_189.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_2216.npy': 0,\n",
       " 'chris_bad1_scaled_kpts_3D_1919.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_4382.npy': 0,\n",
       " 'eddie_ok_scaled_kpts_3D_3309.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_2343.npy': 0,\n",
       " 'connor_ok_scaled_kpts_3D_140.npy': 1,\n",
       " 'isaac_good_scaled_kpts_3D_1525.npy': 0,\n",
       " 'michael_ok_scaled_kpts_3D_3072.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_749.npy': 1,\n",
       " 'chris_buff_3d_3565_ok.npy': 1,\n",
       " 'chris_bad0_scaled_kpts_3D_2681.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_293.npy': 0,\n",
       " 'amruthok1_scaled_kpts_3D_2975.npy': 1,\n",
       " 'eddie_ok_scaled_kpts_3D_480.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_560.npy': 0,\n",
       " 'isaac_good_scaled_kpts_3D_2518.npy': 0,\n",
       " 'amruthok2_scaled_kpts_3D_1332.npy': 1,\n",
       " 'eddie_ok_scaled_kpts_3D_391.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_565.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_86.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_3321.npy': 0,\n",
       " 'amruthgood2_scaled_kpts_3D_359.npy': 0,\n",
       " 'chris_good0_scaled_kpts_3D_2387.npy': 0,\n",
       " 'connor_good_scaled_kpts_3D_365.npy': 0,\n",
       " 'isaac_bad_2_scaled_kpts_3D_114.npy': 1,\n",
       " 'isaac_bad_2_scaled_kpts_3D_967.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_1034.npy': 0,\n",
       " 'isaac_good_2_scaled_kpts_3D_4343.npy': 0,\n",
       " 'chris_buff_3d_2596_ok.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_3422.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_255.npy': 0,\n",
       " 'isaac_bad_2_scaled_kpts_3D_3816.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_1482.npy': 0,\n",
       " 'connor_ok_scaled_kpts_3D_2348.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_1605.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_2897.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_1430.npy': 1,\n",
       " 'chris_good1_scaled_kpts_3D_2660.npy': 0,\n",
       " 'michael_good_scaled_kpts_3D_294.npy': 0,\n",
       " 'eddie_good1_scaled_kpts_3D_3693.npy': 0,\n",
       " 'isaac_bad_scaled_kpts_3D_2994.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_3046.npy': 0,\n",
       " 'connor_bad_scaled_kpts_3D_3352.npy': 1,\n",
       " 'eddie_bad1_scaled_kpts_3D_3033.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_3101.npy': 0,\n",
       " 'michael_ok_scaled_kpts_3D_2679.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_1691.npy': 0,\n",
       " 'isaac_ok_scaled_kpts_3D_4449.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_1087.npy': 1,\n",
       " 'chris_bad1_scaled_kpts_3D_858.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_457.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_1638.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_3512.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_1002.npy': 0,\n",
       " 'connor_good_scaled_kpts_3D_1166.npy': 0,\n",
       " 'amruthgood1_scaled_kpts_3D_2231.npy': 0,\n",
       " 'amruthgood1_scaled_kpts_3D_780.npy': 0,\n",
       " 'connor_bad_scaled_kpts_3D_1661.npy': 1,\n",
       " 'eddie_ok_scaled_kpts_3D_3157.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_3690.npy': 0,\n",
       " 'michael_good_scaled_kpts_3D_1040.npy': 0,\n",
       " 'chris_good1_scaled_kpts_3D_3842.npy': 0,\n",
       " 'isaac_good_scaled_kpts_3D_506.npy': 0,\n",
       " 'chris_buff_3d_7488_ok.npy': 1,\n",
       " 'daniel_bad_scaled_kpts_3D_3452.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_3329.npy': 0,\n",
       " 'amruthok2_scaled_kpts_3D_3619.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_4031.npy': 1,\n",
       " 'daniel_bad_scaled_kpts_3D_2973.npy': 1,\n",
       " 'daniel_bad_scaled_kpts_3D_3110.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_2849.npy': 1,\n",
       " 'connor_good_scaled_kpts_3D_2058.npy': 0,\n",
       " 'isaac_ok_2_scaled_kpts_3D_162.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_1657.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_2773.npy': 1,\n",
       " 'chris_bad1_scaled_kpts_3D_2558.npy': 1,\n",
       " 'isaac_good_scaled_kpts_3D_3297.npy': 0,\n",
       " 'isaac_good_2_scaled_kpts_3D_164.npy': 0,\n",
       " 'chris_bad0_scaled_kpts_3D_1490.npy': 1,\n",
       " 'chris_buff_3d_1398_ok.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_85.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_3964.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_2856.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_1518.npy': 0,\n",
       " 'amruthbad2_scaled_kpts_3D_3331.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_4189.npy': 1,\n",
       " 'chris_buff_3d_4007_ok.npy': 1,\n",
       " 'connor_bad_scaled_kpts_3D_942.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_2741.npy': 0,\n",
       " 'eddie_bad1_scaled_kpts_3D_1973.npy': 1,\n",
       " 'eddie_ok_scaled_kpts_3D_2015.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_535.npy': 0,\n",
       " 'eddie_good0_scaled_kpts_3D_3239.npy': 0,\n",
       " 'amruthok1_scaled_kpts_3D_518.npy': 1,\n",
       " 'daniel_bad_scaled_kpts_3D_2638.npy': 1,\n",
       " 'daniel_ok_scaled_kpts_3D_2723.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_2102.npy': 0,\n",
       " 'isaac_good_2_scaled_kpts_3D_3786.npy': 0,\n",
       " 'chris_buff_3d_3907_ok.npy': 1,\n",
       " 'eddie_bad1_scaled_kpts_3D_1404.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_2518.npy': 0,\n",
       " 'connor_bad_scaled_kpts_3D_2408.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_78.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_52.npy': 1,\n",
       " 'chris_bad1_scaled_kpts_3D_2087.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_3603.npy': 0,\n",
       " 'amruth_bad1_3D_3869.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_2414.npy': 1,\n",
       " 'daniel_bad_scaled_kpts_3D_277.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_146.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_2168.npy': 0,\n",
       " 'michael_bad_scaled_kpts_3D_2939.npy': 1,\n",
       " 'connor_good_scaled_kpts_3D_2695.npy': 0,\n",
       " 'chris_buff_3d_7028_ok.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_3651.npy': 0,\n",
       " 'connor_ok_scaled_kpts_3D_214.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_2925.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_899.npy': 0,\n",
       " 'isaac_bad_2_scaled_kpts_3D_1558.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_301.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_4229.npy': 0,\n",
       " 'michael_ok_scaled_kpts_3D_1880.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_3851.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_2104.npy': 0,\n",
       " 'michael_bad_scaled_kpts_3D_470.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_6.npy': 0,\n",
       " 'amruthok1_scaled_kpts_3D_634.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_312.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_1892.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_1083.npy': 0,\n",
       " 'amruthok2_scaled_kpts_3D_1434.npy': 1,\n",
       " 'eddie_ok_scaled_kpts_3D_2172.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_4393.npy': 0,\n",
       " 'chris_ok_scaled_kpts_3D_164.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_2904.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_3254.npy': 0,\n",
       " 'eddie_good0_scaled_kpts_3D_4385.npy': 0,\n",
       " 'amruthbad2_scaled_kpts_3D_856.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_1337.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_4407.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_593.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_4000.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_3444.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_4506.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_4399.npy': 0,\n",
       " 'eddie_good1_scaled_kpts_3D_2216.npy': 0,\n",
       " 'eddie_good0_scaled_kpts_3D_1547.npy': 0,\n",
       " 'amruthgood1_scaled_kpts_3D_4244.npy': 0,\n",
       " 'michael_bad_scaled_kpts_3D_3788.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_2317.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_3517.npy': 1,\n",
       " 'eddie_ok_scaled_kpts_3D_3631.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_2015.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_1552.npy': 1,\n",
       " 'connor_bad_scaled_kpts_3D_292.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_1420.npy': 1,\n",
       " 'eddie_ok_scaled_kpts_3D_4053.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_1958.npy': 0,\n",
       " 'chris_bad0_scaled_kpts_3D_2944.npy': 1,\n",
       " 'chris_buff_3d_8770_ok.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_4054.npy': 0,\n",
       " 'connor_bad_scaled_kpts_3D_911.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_3691.npy': 0,\n",
       " 'isaac_good_scaled_kpts_3D_4337.npy': 0,\n",
       " 'isaac_ok_scaled_kpts_3D_4303.npy': 1,\n",
       " 'connor_bad_scaled_kpts_3D_2686.npy': 1,\n",
       " 'isaac_bad_2_scaled_kpts_3D_2260.npy': 1,\n",
       " 'eddie_bad1_scaled_kpts_3D_1371.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_451.npy': 0,\n",
       " 'isaac_good_scaled_kpts_3D_2884.npy': 0,\n",
       " 'chris_buff_3d_529_ok.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_3791.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_4156.npy': 1,\n",
       " 'chris_buff_3d_8041_ok.npy': 1,\n",
       " 'isaac_bad_2_scaled_kpts_3D_1195.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_91.npy': 0,\n",
       " 'amruthbad2_scaled_kpts_3D_2448.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_653.npy': 0,\n",
       " 'connor_bad_scaled_kpts_3D_1022.npy': 1,\n",
       " 'chris_buff_3d_985_ok.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_439.npy': 0,\n",
       " 'isaac_good_scaled_kpts_3D_1824.npy': 0,\n",
       " 'connor_bad_scaled_kpts_3D_3372.npy': 1,\n",
       " 'daniel_bad_scaled_kpts_3D_3008.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_3467.npy': 0,\n",
       " 'isaac_bad_2_scaled_kpts_3D_2800.npy': 1,\n",
       " 'chris_buff_3d_7291_ok.npy': 1,\n",
       " 'eddie_bad1_scaled_kpts_3D_2902.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_3056.npy': 0,\n",
       " 'eddie_bad1_scaled_kpts_3D_4087.npy': 1,\n",
       " 'chris_bad0_scaled_kpts_3D_3446.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_1956.npy': 0,\n",
       " 'chris_bad0_scaled_kpts_3D_698.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_617.npy': 1,\n",
       " 'daniel_bad_scaled_kpts_3D_2414.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_1343.npy': 0,\n",
       " 'amruthgood2_scaled_kpts_3D_991.npy': 0,\n",
       " 'eddie_good1_scaled_kpts_3D_258.npy': 0,\n",
       " 'amruthok2_scaled_kpts_3D_1809.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_1232.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_215.npy': 0,\n",
       " 'amruthok2_scaled_kpts_3D_3958.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_1213.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_3873.npy': 1,\n",
       " 'isaac_good_scaled_kpts_3D_3986.npy': 0,\n",
       " 'chris_bad0_scaled_kpts_3D_1053.npy': 1,\n",
       " 'chris_buff_3d_9884_ok.npy': 1,\n",
       " 'chris_good1_scaled_kpts_3D_3223.npy': 0,\n",
       " 'connor_good_scaled_kpts_3D_176.npy': 0,\n",
       " 'chris_bad0_scaled_kpts_3D_3074.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_4374.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_1782.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_3965.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_3656.npy': 0,\n",
       " 'amruthgood2_scaled_kpts_3D_3852.npy': 0,\n",
       " 'daniel_ok_scaled_kpts_3D_2322.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_1361.npy': 0,\n",
       " 'amruthbad2_scaled_kpts_3D_2038.npy': 1,\n",
       " 'isaac_bad_2_scaled_kpts_3D_2153.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_3787.npy': 1,\n",
       " 'connor_good_scaled_kpts_3D_1977.npy': 0,\n",
       " 'daniel_ok_scaled_kpts_3D_409.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_1766.npy': 1,\n",
       " 'isaac_bad_2_scaled_kpts_3D_345.npy': 1,\n",
       " 'isaac_good_scaled_kpts_3D_235.npy': 0,\n",
       " 'daniel_bad_scaled_kpts_3D_202.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_2326.npy': 0,\n",
       " 'isaac_ok_2_scaled_kpts_3D_3510.npy': 1,\n",
       " 'chris_good1_scaled_kpts_3D_3281.npy': 0,\n",
       " 'chris_bad0_scaled_kpts_3D_3623.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_1505.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_2527.npy': 0,\n",
       " 'amruth_bad1_3D_1890.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_3241.npy': 1,\n",
       " 'connor_bad_scaled_kpts_3D_2829.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_3399.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_2857.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_2257.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_2648.npy': 0,\n",
       " 'daniel_ok_scaled_kpts_3D_3480.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_2704.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_305.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_1126.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_1051.npy': 0,\n",
       " 'chris_buff_3d_6157_ok.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_3694.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_2409.npy': 0,\n",
       " 'chris_buff_3d_7607_ok.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_1936.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_1738.npy': 1,\n",
       " 'eddie_bad0_scaled_kpts_3D_3438.npy': 1,\n",
       " 'chris_bad0_scaled_kpts_3D_2452.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_659.npy': 0,\n",
       " 'chris_buff_3d_2908_ok.npy': 1,\n",
       " 'chris_buff_3d_8842_ok.npy': 1,\n",
       " 'chris_buff_3d_1581_ok.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_1585.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_879.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_2610.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_919.npy': 0,\n",
       " 'daniel_ok_scaled_kpts_3D_2728.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_1613.npy': 0,\n",
       " 'isaac_bad_2_scaled_kpts_3D_4421.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_2199.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_3310.npy': 0,\n",
       " 'daniel_ok_scaled_kpts_3D_2592.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_3617.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_355.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_2603.npy': 1,\n",
       " 'chris_buff_3d_8686_ok.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_2366.npy': 0,\n",
       " 'amruth_bad1_3D_307.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_1053.npy': 0,\n",
       " 'eddie_good1_scaled_kpts_3D_1405.npy': 0,\n",
       " 'eddie_good0_scaled_kpts_3D_2624.npy': 0,\n",
       " 'amruthok2_scaled_kpts_3D_3691.npy': 1,\n",
       " 'amruth_bad1_3D_2548.npy': 1,\n",
       " 'eddie_bad1_scaled_kpts_3D_1770.npy': 1,\n",
       " 'chris_ok_scaled_kpts_3D_4445.npy': 1,\n",
       " 'eddie_bad1_scaled_kpts_3D_779.npy': 1,\n",
       " 'connor_good_scaled_kpts_3D_49.npy': 0,\n",
       " 'chris_bad1_scaled_kpts_3D_580.npy': 1,\n",
       " 'daniel_bad_scaled_kpts_3D_860.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_1632.npy': 0,\n",
       " 'chris_bad1_scaled_kpts_3D_29.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_1323.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_3612.npy': 1,\n",
       " 'eddie_ok_scaled_kpts_3D_3331.npy': 1,\n",
       " 'chris_buff_3d_6517_ok.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_2131.npy': 0,\n",
       " 'amruthgood1_scaled_kpts_3D_1698.npy': 0,\n",
       " 'amruth_bad1_3D_2300.npy': 1,\n",
       " 'chris_bad1_scaled_kpts_3D_2170.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_1448.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_2486.npy': 0,\n",
       " 'michael_bad_scaled_kpts_3D_516.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_878.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_1318.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_2422.npy': 1,\n",
       " 'chris_bad1_scaled_kpts_3D_2181.npy': 1,\n",
       " 'eddie_ok_scaled_kpts_3D_4190.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_1786.npy': 0,\n",
       " 'amruthgood2_scaled_kpts_3D_2475.npy': 0,\n",
       " 'isaac_bad_2_scaled_kpts_3D_1521.npy': 1,\n",
       " 'daniel_ok_scaled_kpts_3D_2968.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_747.npy': 0,\n",
       " 'amruthbad2_scaled_kpts_3D_2191.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_2807.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_1414.npy': 0,\n",
       " 'chris_buff_3d_843_ok.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_2120.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_3776.npy': 1,\n",
       " 'chris_buff_3d_6747_ok.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_2099.npy': 0,\n",
       " 'connor_good_scaled_kpts_3D_922.npy': 0,\n",
       " 'eddie_good1_scaled_kpts_3D_2256.npy': 0,\n",
       " 'chris_good1_scaled_kpts_3D_975.npy': 0,\n",
       " 'eddie_bad0_scaled_kpts_3D_4164.npy': 1,\n",
       " 'daniel_bad_scaled_kpts_3D_939.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_1335.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_1548.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_129.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_1095.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_3221.npy': 0,\n",
       " 'chris_ok_scaled_kpts_3D_4227.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_1687.npy': 0,\n",
       " 'daniel_bad_scaled_kpts_3D_2154.npy': 1,\n",
       " 'daniel_ok_scaled_kpts_3D_1230.npy': 1,\n",
       " 'daniel_ok_scaled_kpts_3D_1392.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_3909.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_2670.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_2226.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_2414.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_2400.npy': 0,\n",
       " 'isaac_bad_scaled_kpts_3D_2542.npy': 1,\n",
       " 'eddie_bad0_scaled_kpts_3D_2999.npy': 1,\n",
       " 'isaac_good_scaled_kpts_3D_2783.npy': 0,\n",
       " 'eddie_bad0_scaled_kpts_3D_2836.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_2202.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_406.npy': 0,\n",
       " 'eddie_ok_scaled_kpts_3D_1016.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_1839.npy': 1,\n",
       " 'eddie_bad0_scaled_kpts_3D_4382.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_3260.npy': 1,\n",
       " 'amruth_bad1_3D_830.npy': 1,\n",
       " 'eddie_bad1_scaled_kpts_3D_1034.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_7.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_1881.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_1188.npy': 1,\n",
       " 'chris_buff_3d_5412_ok.npy': 1,\n",
       " 'isaac_bad_2_scaled_kpts_3D_3905.npy': 1,\n",
       " 'connor_good_scaled_kpts_3D_1680.npy': 0,\n",
       " 'isaac_ok_scaled_kpts_3D_939.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_3162.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_4261.npy': 0,\n",
       " 'michael_ok_scaled_kpts_3D_3054.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_2824.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_1298.npy': 1,\n",
       " 'daniel_ok_scaled_kpts_3D_271.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_2863.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_3104.npy': 0,\n",
       " 'chris_buff_3d_5076_ok.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_3724.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_1571.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_967.npy': 1,\n",
       " 'amruth_bad1_3D_3103.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_297.npy': 0,\n",
       " 'daniel_good_scaled_kpts_3D_581.npy': 0,\n",
       " 'amruth_bad1_3D_3886.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_2763.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_857.npy': 0,\n",
       " 'isaac_good_scaled_kpts_3D_166.npy': 0,\n",
       " 'eddie_good1_scaled_kpts_3D_3205.npy': 0,\n",
       " 'isaac_bad_scaled_kpts_3D_652.npy': 1,\n",
       " 'eddie_bad1_scaled_kpts_3D_1714.npy': 1,\n",
       " 'connor_bad_scaled_kpts_3D_2498.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_2272.npy': 0,\n",
       " 'eddie_good0_scaled_kpts_3D_2047.npy': 0,\n",
       " 'amruthgood1_scaled_kpts_3D_1345.npy': 0,\n",
       " 'michael_bad_scaled_kpts_3D_3962.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_1262.npy': 0,\n",
       " 'amruthgood1_scaled_kpts_3D_3267.npy': 0,\n",
       " 'eddie_good0_scaled_kpts_3D_2177.npy': 0,\n",
       " 'eddie_bad1_scaled_kpts_3D_0.npy': 1,\n",
       " 'isaac_bad_2_scaled_kpts_3D_4199.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_874.npy': 1,\n",
       " 'chris_bad1_scaled_kpts_3D_1048.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_4308.npy': 1,\n",
       " 'eddie_bad1_scaled_kpts_3D_4057.npy': 1,\n",
       " 'isaac_bad_2_scaled_kpts_3D_4357.npy': 1,\n",
       " 'eddie_bad0_scaled_kpts_3D_1299.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_890.npy': 1,\n",
       " 'eddie_bad0_scaled_kpts_3D_2735.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_2203.npy': 0,\n",
       " 'eddie_good0_scaled_kpts_3D_4070.npy': 0,\n",
       " 'connor_good_scaled_kpts_3D_884.npy': 0,\n",
       " 'daniel_good_scaled_kpts_3D_530.npy': 0,\n",
       " 'michael_bad_scaled_kpts_3D_2315.npy': 1,\n",
       " 'eddie_bad1_scaled_kpts_3D_4433.npy': 1,\n",
       " 'chris_bad0_scaled_kpts_3D_3449.npy': 1,\n",
       " 'connor_good_scaled_kpts_3D_3176.npy': 0,\n",
       " 'amruthok2_scaled_kpts_3D_3061.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_1089.npy': 0,\n",
       " 'amruthok2_scaled_kpts_3D_886.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_3414.npy': 0,\n",
       " 'chris_bad0_scaled_kpts_3D_2390.npy': 1,\n",
       " 'chris_bad0_scaled_kpts_3D_1421.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_3834.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_37.npy': 0,\n",
       " 'chris_ok_scaled_kpts_3D_1561.npy': 1,\n",
       " 'connor_good_scaled_kpts_3D_3140.npy': 0,\n",
       " 'amruthgood2_scaled_kpts_3D_44.npy': 0,\n",
       " 'isaac_good_scaled_kpts_3D_3788.npy': 0,\n",
       " 'isaac_good_2_scaled_kpts_3D_539.npy': 0,\n",
       " 'connor_bad_scaled_kpts_3D_869.npy': 1,\n",
       " 'eddie_bad1_scaled_kpts_3D_534.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_3963.npy': 1,\n",
       " 'eddie_bad0_scaled_kpts_3D_2071.npy': 1,\n",
       " 'chris_buff_3d_4119_ok.npy': 1,\n",
       " 'connor_ok_scaled_kpts_3D_425.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_3928.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_680.npy': 0,\n",
       " 'eddie_ok_scaled_kpts_3D_1257.npy': 1,\n",
       " 'chris_bad0_scaled_kpts_3D_2862.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_559.npy': 1,\n",
       " 'daniel_ok_scaled_kpts_3D_3178.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_2005.npy': 0,\n",
       " 'chris_bad0_scaled_kpts_3D_4272.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_779.npy': 0,\n",
       " 'connor_good_scaled_kpts_3D_3229.npy': 0,\n",
       " 'chris_buff_3d_1437_ok.npy': 1,\n",
       " 'eddie_bad0_scaled_kpts_3D_115.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_2082.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_1514.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_4434.npy': 1,\n",
       " 'isaac_bad_2_scaled_kpts_3D_2215.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_3708.npy': 0,\n",
       " 'connor_good_scaled_kpts_3D_1452.npy': 0,\n",
       " 'chris_bad1_scaled_kpts_3D_636.npy': 1,\n",
       " 'michael_good_scaled_kpts_3D_2290.npy': 0,\n",
       " 'chris_good1_scaled_kpts_3D_807.npy': 0,\n",
       " 'amruthgood1_scaled_kpts_3D_4270.npy': 0,\n",
       " 'amruthgood2_scaled_kpts_3D_2295.npy': 0,\n",
       " 'chris_bad0_scaled_kpts_3D_2260.npy': 1,\n",
       " 'daniel_ok_scaled_kpts_3D_476.npy': 1,\n",
       " 'chris_buff_3d_9824_ok.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_959.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_2168.npy': 0,\n",
       " 'eddie_bad1_scaled_kpts_3D_533.npy': 1,\n",
       " 'chris_buff_3d_8502_ok.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_3196.npy': 0,\n",
       " 'isaac_bad_scaled_kpts_3D_124.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_2339.npy': 1,\n",
       " 'chris_buff_3d_3676_ok.npy': 1,\n",
       " 'chris_buff_3d_5972_ok.npy': 1,\n",
       " 'connor_good_scaled_kpts_3D_2461.npy': 0,\n",
       " 'chris_ok_scaled_kpts_3D_4065.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_2889.npy': 0,\n",
       " 'isaac_bad_scaled_kpts_3D_70.npy': 1,\n",
       " 'chris_good1_scaled_kpts_3D_1468.npy': 0,\n",
       " 'amruthok1_scaled_kpts_3D_3025.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_4214.npy': 0,\n",
       " 'eddie_ok_scaled_kpts_3D_1315.npy': 1,\n",
       " 'chris_bad1_scaled_kpts_3D_2078.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_1857.npy': 1,\n",
       " 'chris_buff_3d_5507_ok.npy': 1,\n",
       " 'amruthgood2_scaled_kpts_3D_1929.npy': 0,\n",
       " 'eddie_good1_scaled_kpts_3D_1806.npy': 0,\n",
       " 'amruthok2_scaled_kpts_3D_4133.npy': 1,\n",
       " 'eddie_good1_scaled_kpts_3D_3771.npy': 0,\n",
       " 'isaac_bad_2_scaled_kpts_3D_2093.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_1812.npy': 0,\n",
       " 'chris_bad0_scaled_kpts_3D_3210.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_933.npy': 0,\n",
       " 'chris_buff_3d_1571_ok.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_254.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_3011.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_3838.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_1402.npy': 1,\n",
       " 'connor_good_scaled_kpts_3D_2492.npy': 0,\n",
       " 'eddie_ok_scaled_kpts_3D_1023.npy': 1,\n",
       " 'chris_buff_3d_2632_ok.npy': 1,\n",
       " 'daniel_ok_scaled_kpts_3D_2570.npy': 1,\n",
       " 'amruth_bad1_3D_2896.npy': 1,\n",
       " 'daniel_bad_scaled_kpts_3D_2282.npy': 1,\n",
       " 'eddie_ok_scaled_kpts_3D_1590.npy': 1,\n",
       " 'chris_bad1_scaled_kpts_3D_3167.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_1892.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_2869.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_2245.npy': 1,\n",
       " 'amruth_bad1_3D_884.npy': 1,\n",
       " 'isaac_ok_scaled_kpts_3D_4639.npy': 1,\n",
       " 'chris_buff_3d_2165_ok.npy': 1,\n",
       " 'amruthgood1_scaled_kpts_3D_4429.npy': 0,\n",
       " 'chris_ok_scaled_kpts_3D_247.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_1548.npy': 1,\n",
       " 'chris_good1_scaled_kpts_3D_2022.npy': 0,\n",
       " 'eddie_bad1_scaled_kpts_3D_728.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_4132.npy': 1,\n",
       " 'chris_buff_3d_7152_ok.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_1230.npy': 0,\n",
       " 'chris_bad1_scaled_kpts_3D_2921.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_181.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_1737.npy': 0,\n",
       " 'chris_ok_scaled_kpts_3D_2994.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_2140.npy': 0,\n",
       " 'chris_buff_3d_3529_ok.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_3237.npy': 0,\n",
       " 'eddie_good0_scaled_kpts_3D_3292.npy': 0,\n",
       " 'eddie_bad1_scaled_kpts_3D_4239.npy': 1,\n",
       " 'eddie_good0_scaled_kpts_3D_747.npy': 0,\n",
       " 'chris_good1_scaled_kpts_3D_2527.npy': 0,\n",
       " 'daniel_ok_scaled_kpts_3D_416.npy': 1,\n",
       " 'connor_good_scaled_kpts_3D_1165.npy': 0,\n",
       " 'eddie_bad0_scaled_kpts_3D_1757.npy': 1,\n",
       " 'chris_bad1_scaled_kpts_3D_1598.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_925.npy': 0,\n",
       " 'amruthbad2_scaled_kpts_3D_2928.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_1486.npy': 0,\n",
       " 'michael_ok_scaled_kpts_3D_300.npy': 1,\n",
       " 'chris_buff_3d_520_ok.npy': 1,\n",
       " 'daniel_bad_scaled_kpts_3D_972.npy': 1,\n",
       " 'isaac_bad_scaled_kpts_3D_1523.npy': 1,\n",
       " 'amruth_bad1_3D_1707.npy': 1,\n",
       " 'isaac_ok_2_scaled_kpts_3D_4259.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_790.npy': 1,\n",
       " 'daniel_good_scaled_kpts_3D_2290.npy': 0,\n",
       " 'chris_good0_scaled_kpts_3D_1239.npy': 0,\n",
       " 'eddie_good1_scaled_kpts_3D_1656.npy': 0,\n",
       " 'michael_bad_scaled_kpts_3D_1286.npy': 1,\n",
       " 'connor_bad_scaled_kpts_3D_2641.npy': 1,\n",
       " 'chris_bad1_scaled_kpts_3D_3053.npy': 1,\n",
       " 'isaac_good_2_scaled_kpts_3D_2004.npy': 0,\n",
       " 'chris_bad0_scaled_kpts_3D_607.npy': 1,\n",
       " 'amruth_bad1_3D_3980.npy': 1,\n",
       " 'eddie_bad1_scaled_kpts_3D_13.npy': 1,\n",
       " 'michael_bad_scaled_kpts_3D_810.npy': 1,\n",
       " 'amruthok1_scaled_kpts_3D_184.npy': 1,\n",
       " 'amruthbad2_scaled_kpts_3D_503.npy': 1,\n",
       " 'michael_ok_scaled_kpts_3D_3440.npy': 1,\n",
       " 'connor_good_scaled_kpts_3D_3282.npy': 0,\n",
       " 'eddie_bad0_scaled_kpts_3D_9.npy': 1,\n",
       " 'amruthok2_scaled_kpts_3D_2801.npy': 1,\n",
       " 'chris_good0_scaled_kpts_3D_3057.npy': 0,\n",
       " 'connor_good_scaled_kpts_3D_479.npy': 0,\n",
       " 'amruthok2_scaled_kpts_3D_2002.npy': 1,\n",
       " 'amruth_bad1_3D_2314.npy': 1,\n",
       " 'daniel_ok_scaled_kpts_3D_962.npy': 1,\n",
       " 'chris_good1_scaled_kpts_3D_3078.npy': 0,\n",
       " 'chris_bad0_scaled_kpts_3D_1086.npy': 1,\n",
       " ...}"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset_path = \"/home/jeff/Documents/Code/EvoSkeleton-master/examples/keypoints_normalized_clipped/\"\n",
    "jeff_val = \"/home/jeff/Documents/Code/EvoSkeleton-master/examples/jeff_kpts_normalized_clipped/\"\n",
    "daniel_test = \"/home/jeff/Documents/Code/EvoSkeleton-master/examples/daniel_kpts_normalized_clipped/\"\n",
    "val_datapath = daniel_test\n",
    "test_datapath = jeff_val\n",
    "\n",
    "partition = {\n",
    "    \"train\": [],\n",
    "    \"val\": [],\n",
    "    \"test\": [],\n",
    "}\n",
    "labels = {}\n",
    "classes = []\n",
    "val_classes = []\n",
    "good = 0 \n",
    "ok = 0\n",
    "bad = 0\n",
    "test = False\n",
    "for f in os.listdir(dataset_path):\n",
    "    rand = random.random()\n",
    "    # if (rand < 0.05):\n",
    "        # test = True\n",
    "        # partition[\"test\"].append(f)\n",
    "    # elif (rand > 0.1 and rand < 0.3):\n",
    "    #     test = False\n",
    "    #     partition[\"val\"].append(f)\n",
    "    # else:\n",
    "        # test = False\n",
    "    partition[\"train\"].append(f)\n",
    "    if (\"ok\") in f:\n",
    "        labels[f] = 1\n",
    "        ok += 1\n",
    "    elif (\"good\") in f:\n",
    "        labels[f] = 0\n",
    "        good += 1\n",
    "    elif (\"bad\") in f:\n",
    "        labels[f] = 1\n",
    "        bad += 1\n",
    "\n",
    "for f in os.listdir(val_datapath):\n",
    "    partition[\"val\"].append(f)\n",
    "    if (\"ok\") in f:\n",
    "        labels[f] = 1\n",
    "        # ok += 1\n",
    "        val_classes.append(1)\n",
    "    elif (\"good\") in f:\n",
    "        labels[f] = 0\n",
    "        # good += 1\n",
    "        val_classes.append(0)\n",
    "    elif (\"bad\") in f:\n",
    "        labels[f] = 1\n",
    "        # bad += 1\n",
    "        val_classes.append(1)\n",
    "    else:\n",
    "        print(\"something went wrong!\")\n",
    "\n",
    "    # label is not being added because you haven't labelled test set with good, bad, ok, yet\n",
    "\n",
    "\n",
    "for f in os.listdir(test_datapath):\n",
    "    # print(f)\n",
    "    partition[\"test\"].append(f)\n",
    "    if (\"ok\") in f:\n",
    "        labels[f] = 1\n",
    "        classes.append(1)\n",
    "    elif (\"good\") in f:\n",
    "        labels[f] = 0\n",
    "        classes.append(0)\n",
    "    elif (\"bad\") in f:\n",
    "        labels[f] = 1\n",
    "        classes.append(1)\n",
    "print(good)\n",
    "print(ok)\n",
    "print(bad)\n",
    "print(len(partition[\"train\"]))\n",
    "print(len(partition[\"val\"]))\n",
    "print(len(val_classes))\n",
    "print(len(partition[\"test\"]))\n",
    "print(len(classes))\n",
    "# print(len(labels))\n",
    "# labels\n",
    "\n",
    "# labels = to_categorical(labels)\n",
    "# print(partition[\"test\"])\n",
    "labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "class DataGenerator(keras.utils.Sequence):\n",
    "    'Generates data for Keras'\n",
    "    def __init__(self, list_IDs, labels, dataset_dir, batch_size=64, dim=(17,3), n_classes=3, shuffle=True,):\n",
    "        'Initialization'\n",
    "        self.dim = dim\n",
    "        self.batch_size = batch_size\n",
    "        self.labels = labels\n",
    "        self.dataset_dir = dataset_dir\n",
    "        self.list_IDs = list_IDs\n",
    "        self.n_classes = n_classes\n",
    "        self.shuffle = shuffle\n",
    "        self.on_epoch_end()\n",
    "\n",
    "    def __len__(self):\n",
    "        'Denotes the number of batches per epoch'\n",
    "        return int(np.floor(len(self.list_IDs) / self.batch_size))\n",
    "        # return len(self.list_IDs)\n",
    "\n",
    "    def __getitem__(self, index):\n",
    "        'Generate one batch of data'\n",
    "        # Generate indexes of the batch\n",
    "        indexes = self.indexes[index*self.batch_size:(index+1)*self.batch_size]\n",
    "\n",
    "        # Find list of IDs\n",
    "        list_IDs_temp = [self.list_IDs[k] for k in indexes]\n",
    "\n",
    "        # Generate data\n",
    "        X, y = self.__data_generation(list_IDs_temp)\n",
    "\n",
    "        return X, y\n",
    "\n",
    "    def on_epoch_end(self):\n",
    "        'Updates indexes after each epoch'\n",
    "        self.indexes = np.arange(len(self.list_IDs))\n",
    "        if self.shuffle == True:\n",
    "            np.random.shuffle(self.indexes)\n",
    "\n",
    "    def __data_generation(self, list_IDs_temp):\n",
    "        'Generates data containing batch_size samples' # X : (n_samples, *dim, n_channels)\n",
    "        # Initialization\n",
    "        X = np.empty((self.batch_size, *self.dim))\n",
    "        y = np.empty((self.batch_size), dtype=int)\n",
    "\n",
    "        # Generate data\n",
    "        for i, ID in enumerate(list_IDs_temp):\n",
    "            # Store sample\n",
    "            f = np.load(self.dataset_dir + ID)\n",
    "            # reshaped = f.reshape(-1,3,1)\n",
    "            # X[i,] = reshaped\n",
    "            X[i] = f\n",
    "\n",
    "            # Store class\n",
    "            y[i] = self.labels[ID]\n",
    "\n",
    "        return X, keras.utils.to_categorical(y, num_classes=self.n_classes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "560\n",
      "2\n",
      "697\n"
     ]
    }
   ],
   "source": [
    "params = {\n",
    "    'dim': (10,3),\n",
    "    'n_classes': 2,\n",
    "}\n",
    "\n",
    "train_datagen = DataGenerator(\n",
    "    partition[\"train\"], \n",
    "    labels, \n",
    "    dataset_dir=dataset_path,\n",
    "    batch_size=BATCH_SIZE,\n",
    "    shuffle=True,\n",
    "    **params\n",
    ")\n",
    "val_datagen = DataGenerator(\n",
    "    partition[\"val\"],\n",
    "    labels,\n",
    "    dataset_dir=val_datapath,\n",
    "    batch_size=BATCH_SIZE,\n",
    "    # batch_size=1,\n",
    "    shuffle=True,\n",
    "    **params\n",
    ")\n",
    "test_datagen = DataGenerator(\n",
    "    partition[\"test\"],\n",
    "    labels,\n",
    "    dataset_dir=test_datapath,\n",
    "    # dataset_dir=jeff_val,\n",
    "    shuffle=False,\n",
    "    batch_size=1,\n",
    "    **params\n",
    ")\n",
    "print(len(train_datagen))\n",
    "print(len(val_datagen))\n",
    "print(len(test_datagen))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " dense (Dense)               (None, 10, 12)            48        \n",
      "                                                                 \n",
      " dropout (Dropout)           (None, 10, 12)            0         \n",
      "                                                                 \n",
      " dense_1 (Dense)             (None, 10, 12)            156       \n",
      "                                                                 \n",
      " dropout_1 (Dropout)         (None, 10, 12)            0         \n",
      "                                                                 \n",
      " dense_2 (Dense)             (None, 10, 12)            156       \n",
      "                                                                 \n",
      " dropout_2 (Dropout)         (None, 10, 12)            0         \n",
      "                                                                 \n",
      " flatten (Flatten)           (None, 120)               0         \n",
      "                                                                 \n",
      " dense_3 (Dense)             (None, 2)                 242       \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 602\n",
      "Trainable params: 602\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "\n",
    "kernel_rate = 0.005\n",
    "bias_rate =   0.005\n",
    "dropout_rate = 0.05\n",
    "model = models.Sequential()\n",
    "\n",
    "model.add(layers.InputLayer(input_shape=(10,3)))\n",
    "model.add(\n",
    "    layers.Dense(\n",
    "        units=12, activation='relu',\n",
    "        kernel_regularizer=regularizers.L1(kernel_rate),\n",
    "        bias_regularizer=regularizers.L1(bias_rate),\n",
    "    )\n",
    ")\n",
    "model.add(layers.Dropout(dropout_rate))\n",
    "model.add(\n",
    "    layers.Dense(\n",
    "        units=12, activation='relu',\n",
    "        kernel_regularizer=regularizers.L1(kernel_rate),\n",
    "        bias_regularizer=regularizers.L1(bias_rate),\n",
    "    )\n",
    ")\n",
    "model.add(layers.Dropout(dropout_rate))\n",
    "model.add(\n",
    "    layers.Dense(\n",
    "        units=12, activation='relu',\n",
    "        kernel_regularizer=regularizers.L1(kernel_rate),\n",
    "        bias_regularizer=regularizers.L1(bias_rate),\n",
    "    )\n",
    ")\n",
    "model.add(layers.Dropout(dropout_rate))\n",
    "model.add(layers.Flatten())\n",
    "model.add(layers.Dense(2, activation = 'softmax'))\n",
    "model.summary()\n",
    "\n",
    "# best performance: 3x12, 0.005 bias kernel 0.01 dropout 0.0008 lr"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "class PlotLearning(keras.callbacks.Callback):\n",
    "    \"\"\"\n",
    "    Callback to plot the learning curves of the model during training.\n",
    "    \"\"\"\n",
    "    def on_train_begin(self, logs={}):\n",
    "        self.metrics = {}\n",
    "        for metric in logs:\n",
    "            self.metrics[metric] = []\n",
    "            \n",
    "\n",
    "    def on_epoch_end(self, epoch, logs={}):\n",
    "        # Storing metrics\n",
    "        for metric in logs:\n",
    "            if metric in self.metrics:\n",
    "                self.metrics[metric].append(logs.get(metric))\n",
    "            else:\n",
    "                self.metrics[metric] = [logs.get(metric)]\n",
    "        \n",
    "        # Plotting\n",
    "        metrics = [x for x in logs if 'val' not in x]\n",
    "        \n",
    "        f, axs = plt.subplots(1, len(metrics), figsize=(15,5))\n",
    "        clear_output(wait=True)\n",
    "\n",
    "        for i, metric in enumerate(metrics):\n",
    "            axs[i].plot(range(1, epoch + 2), \n",
    "                        self.metrics[metric], \n",
    "                        label=metric)\n",
    "            if logs['val_' + metric]:\n",
    "                axs[i].plot(range(1, epoch + 2), \n",
    "                            self.metrics['val_' + metric], \n",
    "                            label='val_' + metric)\n",
    "                \n",
    "            axs[i].legend()\n",
    "            axs[i].grid()\n",
    "\n",
    "        plt.tight_layout()\n",
    "        plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "# from tensorflow.keras.callbacks import Callback\n",
    "# import tensorflow.keras.backend as K\n",
    "class SGDRScheduler(tf.keras.callbacks.Callback):\n",
    "    '''Cosine annealing learning rate scheduler with periodic restarts.\n",
    "    # Usage\n",
    "        ```python\n",
    "            schedule = SGDRScheduler(min_lr=1e-5,\n",
    "                                     max_lr=1e-2,\n",
    "                                     steps_per_epoch=np.ceil(epoch_size/batch_size),\n",
    "                                     lr_decay=0.9,\n",
    "                                     cycle_length=5,\n",
    "                                     mult_factor=1.5)\n",
    "            model.fit(X_train, Y_train, epochs=100, callbacks=[schedule])\n",
    "        ```\n",
    "    # Arguments\n",
    "        min_lr: The lower bound of the learning rate range for the experiment.\n",
    "        max_lr: The upper bound of the learning rate range for the experiment.\n",
    "        steps_per_epoch: Number of mini-batches in the dataset. Calculated as `np.ceil(epoch_size/batch_size)`. \n",
    "        lr_decay: Reduce the max_lr after the completion of each cycle.\n",
    "                  Ex. To reduce the max_lr by 20% after each cycle, set this value to 0.8.\n",
    "        cycle_length: Initial number of epochs in a cycle.\n",
    "        mult_factor: Scale epochs_to_restart after each full cycle completion.\n",
    "    # References\n",
    "        Blog post: jeremyjordan.me/nn-learning-rate\n",
    "        Original paper: http://arxiv.org/abs/1608.03983\n",
    "    '''\n",
    "    def __init__(self,\n",
    "                 min_lr,\n",
    "                 max_lr,\n",
    "                 steps_per_epoch,\n",
    "                 lr_decay=1,\n",
    "                 cycle_length=10,\n",
    "                 mult_factor=2):\n",
    "\n",
    "        self.min_lr = min_lr\n",
    "        self.max_lr = max_lr\n",
    "        self.lr_decay = lr_decay\n",
    "\n",
    "        self.batch_since_restart = 0\n",
    "        self.next_restart = cycle_length\n",
    "\n",
    "        self.steps_per_epoch = steps_per_epoch\n",
    "\n",
    "        self.cycle_length = cycle_length\n",
    "        self.mult_factor = mult_factor\n",
    "\n",
    "        self.history = {}\n",
    "\n",
    "    def clr(self):\n",
    "        '''Calculate the learning rate.'''\n",
    "        fraction_to_restart = self.batch_since_restart / (self.steps_per_epoch * self.cycle_length)\n",
    "        lr = self.min_lr + 0.5 * (self.max_lr - self.min_lr) * (1 + np.cos(fraction_to_restart * np.pi))\n",
    "        return lr\n",
    "\n",
    "    def on_train_begin(self, logs={}):\n",
    "        '''Initialize the learning rate to the minimum value at the start of training.'''\n",
    "        logs = logs or {}\n",
    "        tf.keras.backend.set_value(self.model.optimizer.lr, self.max_lr)\n",
    "\n",
    "    def on_batch_end(self, batch, logs={}):\n",
    "        '''Record previous batch statistics and update the learning rate.'''\n",
    "        logs = logs or {}\n",
    "        self.history.setdefault('lr', []).append(tf.keras.backend.get_value(self.model.optimizer.lr))\n",
    "        for k, v in logs.items():\n",
    "            self.history.setdefault(k, []).append(v)\n",
    "\n",
    "        self.batch_since_restart += 1\n",
    "        tf.keras.backend.set_value(self.model.optimizer.lr, self.clr())\n",
    "\n",
    "    def on_epoch_end(self, epoch, logs={}):\n",
    "        '''Check for end of current cycle, apply restarts when necessary.'''\n",
    "        if epoch + 1 == self.next_restart:\n",
    "            self.batch_since_restart = 0\n",
    "            self.cycle_length = np.ceil(self.cycle_length * self.mult_factor)\n",
    "            self.next_restart += self.cycle_length\n",
    "            self.max_lr *= self.lr_decay\n",
    "            self.best_weights = self.model.get_weights()\n",
    "\n",
    "    def on_train_end(self, logs={}):\n",
    "        '''Set weights to the values from the end of the most recent cycle for best performance.'''\n",
    "        self.model.set_weights(self.best_weights)\n",
    "\n",
    "# lr_schedule = SGDRScheduler(min_lr=0.0008, max_lr=0.0063, steps_per_epoch=140, lr_decay=0.95, cycle_length=100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "# from tensorflow.keras.callbacks import Callback\n",
    "# import tensorflow.keras.backend as K\n",
    "class PlateauDestroyer(tf.keras.callbacks.Callback):\n",
    "    def __init__(self,\n",
    "                 og_lr,\n",
    "                 steps_per_epoch,\n",
    "                 epoch_to_jump_at,\n",
    "                 number_of_high_epochs,\n",
    "                 lr_new, lr_post):\n",
    "\n",
    "        self.og_lr = og_lr\n",
    "        self.steps_per_epoch = steps_per_epoch\n",
    "        self.epoch_to_jump_at = epoch_to_jump_at\n",
    "        self.number_of_high_epochs = number_of_high_epochs\n",
    "        self.lr_new = lr_new\n",
    "        self.lr_post = lr_post\n",
    "        self.history = {}\n",
    "\n",
    "    def on_train_begin(self, logs={}):\n",
    "        logs = logs or {}\n",
    "        tf.keras.backend.set_value(self.model.optimizer.lr, self.og_lr)\n",
    "\n",
    "    def on_batch_end(self, batch, logs={}):\n",
    "        '''Record previous batch statistics and update the learning rate.'''\n",
    "        logs = logs or {}\n",
    "        self.history.setdefault('lr', []).append(tf.keras.backend.get_value(self.model.optimizer.lr))\n",
    "        for k, v in logs.items():\n",
    "            self.history.setdefault(k, []).append(v)\n",
    "\n",
    "    def on_epoch_end(self, epoch, logs={}):\n",
    "        if (epoch+1 >= self.epoch_to_jump_at and epoch+1 <= self.epoch_to_jump_at + self.number_of_high_epochs):\n",
    "            lr = self.lr_new\n",
    "            # print(epoch+1, lr)\n",
    "        elif (epoch+1 > self.epoch_to_jump_at + self.number_of_high_epochs):\n",
    "            lr = self.lr_post\n",
    "        else: \n",
    "            lr = self.og_lr\n",
    "            # print(epoch+1, lr)\n",
    "        tf.keras.backend.set_value(self.model.optimizer.lr, lr)\n",
    "        print(tf.keras.backend.get_value(self.model.optimizer.lr))\n",
    "\n",
    "    def on_train_end(self, logs={}):\n",
    "        '''Set weights to the values from the end of the most recent cycle for best performance.'''\n",
    "        self.model.set_weights(self.best_weights)\n",
    "\n",
    "plat_schedule = PlateauDestroyer(\n",
    "    og_lr=0.0063, \n",
    "    steps_per_epoch=140, \n",
    "    epoch_to_jump_at=10, \n",
    "    number_of_high_epochs=10, \n",
    "    lr_new=10, \n",
    "    lr_post=0.0063\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:`period` argument is deprecated. Please use `save_freq` to specify the frequency in number of batches seen.\n"
     ]
    }
   ],
   "source": [
    "model_filepath = 'best_model.h5'\n",
    "checkpoint = keras.callbacks.ModelCheckpoint('model_saves/model{epoch:05d}.h5', period=10) \n",
    "model_save = tf.keras.callbacks.ModelCheckpoint(\n",
    "    model_filepath,\n",
    "    monitor = \"accuracy\",\n",
    "    verbose = 0,\n",
    "    save_best_only = True,\n",
    "    save_weights_only = False,\n",
    "    mode = \"max\",\n",
    "    save_freq = \"epoch\"\n",
    ")\n",
    "val_save = tf.keras.callbacks.ModelCheckpoint(\n",
    "    \"best_val.h5\",\n",
    "    monitor = \"val_accuracy\",\n",
    "    verbose = 0,\n",
    "    save_best_only = True,\n",
    "    save_weights_only = False,\n",
    "    mode = \"max\",\n",
    "    save_freq = \"epoch\"\n",
    ")\n",
    "callback = [model_save, checkpoint, val_save, PlotLearning()]\n",
    "# exp_sch = tf.keras.optimizers.schedules.ExponentialDecay(\n",
    "#     initial_learning_rate=0.006,\n",
    "#     decay_steps=50*140,\n",
    "#     decay_rate=0.9,\n",
    "#     staircase=True\n",
    "# )\n",
    "model.compile(\n",
    "    optimizer = tf.keras.optimizers.Adam(),\n",
    "    loss = \"binary_crossentropy\",\n",
    "    metrics = [\"accuracy\"]\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1080x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "140/140 [==============================] - 3s 17ms/step - loss: 0.5455 - accuracy: 0.7020 - val_loss: 0.5540 - val_accuracy: 0.7480\n"
     ]
    },
    {
     "ename": "NameError",
     "evalue": "name 'pickle' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32m/home/jeff/Documents/Code/StridedTransformer-Pose3D/datatrainer_halfdecent.ipynb Cell 12'\u001b[0m in \u001b[0;36m<cell line: 13>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     <a href='vscode-notebook-cell:/home/jeff/Documents/Code/StridedTransformer-Pose3D/datatrainer_halfdecent.ipynb#ch0000011?line=11'>12</a>\u001b[0m \u001b[39m# save history to disk\u001b[39;00m\n\u001b[1;32m     <a href='vscode-notebook-cell:/home/jeff/Documents/Code/StridedTransformer-Pose3D/datatrainer_halfdecent.ipynb#ch0000011?line=12'>13</a>\u001b[0m \u001b[39mwith\u001b[39;00m \u001b[39mopen\u001b[39m(\u001b[39m'\u001b[39m\u001b[39mhistory.pkl\u001b[39m\u001b[39m'\u001b[39m, \u001b[39m'\u001b[39m\u001b[39mwb\u001b[39m\u001b[39m'\u001b[39m) \u001b[39mas\u001b[39;00m f:\n\u001b[0;32m---> <a href='vscode-notebook-cell:/home/jeff/Documents/Code/StridedTransformer-Pose3D/datatrainer_halfdecent.ipynb#ch0000011?line=13'>14</a>\u001b[0m     pickle\u001b[39m.\u001b[39mdump(history\u001b[39m.\u001b[39mhistory, f)\n",
      "\u001b[0;31mNameError\u001b[0m: name 'pickle' is not defined"
     ]
    }
   ],
   "source": [
    "history = model.fit(\n",
    "    train_datagen,\n",
    "    epochs=1000,\n",
    "    steps_per_epoch=len(train_datagen) // 4,\n",
    "    validation_data=val_datagen,\n",
    "    validation_steps=len(val_datagen),\n",
    "    shuffle=False,\n",
    "    callbacks = callback,\n",
    "    use_multiprocessing=True,\n",
    "    workers=4\n",
    ") \n",
    "# save history to disk\n",
    "with open('history.pkl', 'wb') as f:\n",
    "    pickle.dump(history.history, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.figure(figsize = (15,7))\n",
    "\n",
    "tr_losses = history.history['loss']\n",
    "val_losses = history.history['val_loss']\n",
    "\n",
    "tr_accs = history.history['accuracy']\n",
    "val_accs = history.history['val_accuracy']\n",
    "\n",
    "plt.plot(tr_losses, label = \"train_loss\")\n",
    "plt.plot(val_losses, label = \"val_loss\")\n",
    "plt.ylim(0.52, 0.56)\n",
    "plt.xlabel(\"Number of epochs\")\n",
    "plt.ylabel(\"Cost (J)\")\n",
    "plt.grid()\n",
    "plt.legend()\n",
    "plt.show()\n",
    "\n",
    "plt.figure(figsize = (15,7))\n",
    "\n",
    "plt.plot(tr_accs, label = \"acc_train\")\n",
    "plt.plot(val_accs, label = \"acc_val\")\n",
    "plt.xlabel(\"Number of epochs\")\n",
    "# plt.xlim(0, 1000)\n",
    "plt.ylabel(\"Accuracy\")\n",
    "plt.grid()\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "697/697 [==============================] - 1s 1ms/step\n",
      "[[0.1428269  0.8571731 ]\n",
      " [0.00295965 0.99704033]\n",
      " [0.580567   0.41943303]\n",
      " ...\n",
      " [0.57679313 0.42320693]\n",
      " [0.5970938  0.40290618]\n",
      " [0.00172906 0.998271  ]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 504x432 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "559/559 [==============================] - 1s 1ms/step\n",
      "[[0.01 0.99]\n",
      " [0.44 0.56]\n",
      " [0.51 0.49]\n",
      " ...\n",
      " [0.38 0.62]\n",
      " [0.52 0.48]\n",
      " [0.23 0.77]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 504x432 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "697/697 [==============================] - 1s 1ms/step\n",
      "[[0.47 0.53]\n",
      " [0.01 0.99]\n",
      " [0.6  0.4 ]\n",
      " ...\n",
      " [0.59 0.41]\n",
      " [0.62 0.38]\n",
      " [0.01 0.99]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 504x432 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "559/559 [==============================] - 1s 1ms/step\n",
      "[[0.06 0.94]\n",
      " [0.6  0.4 ]\n",
      " [0.52 0.48]\n",
      " ...\n",
      " [0.37 0.63]\n",
      " [0.51 0.49]\n",
      " [0.37 0.63]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 504x432 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "test_model = tf.keras.models.load_model('prettygoodtest.h5')\n",
    "val_model = tf.keras.models.load_model('prettygoodval.h5')\n",
    "# epoch_model = tf.keras.models.load_model('model_saves/model00100.h5')\n",
    "\n",
    "import matplotlib.patheffects as PathEffects\n",
    "import itertools\n",
    "def plot_confusion_matrix(cm, classes, normalise = False, title = 'Confusion matrix', cmap = plt.cm.RdYlGn):\n",
    "    \n",
    "    plt.imshow(cm, interpolation = 'nearest', cmap = cmap)\n",
    "    plt.title(title)\n",
    "    plt.colorbar()\n",
    "    tick_marks = np.arange(len(classes))\n",
    "    plt.xticks(tick_marks, classes, rotation = 45)\n",
    "    plt.yticks(tick_marks, classes)\n",
    "\n",
    "    if normalise:\n",
    "        cm = cm.astype('float') / cm.sum(axis = 1)[:, np.newaxis]\n",
    "        cm = cm.round(2)\n",
    "\n",
    "    thresh = cm.max() / 1.1\n",
    "    \n",
    "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
    "        txt =  plt.text(\n",
    "            j, i, cm[i, j],\n",
    "            horizontalalignment = \"center\",\n",
    "            color = \"white\",\n",
    "            size = 16\n",
    "        )\n",
    "        txt.set_path_effects([PathEffects.withStroke(linewidth=3, foreground='black')])\n",
    "    plt.tight_layout()\n",
    "    plt.ylabel('True label')\n",
    "    plt.xlabel('Predicted label')\n",
    "\n",
    "val_datagen = DataGenerator(\n",
    "    partition[\"val\"],\n",
    "    labels,\n",
    "    dataset_dir=val_datapath,\n",
    "    # batch_size=BATCH_SIZE,\n",
    "    batch_size=1,\n",
    "    shuffle=False,\n",
    "    **params\n",
    ")\n",
    "\n",
    "def graph_predictions(model, datagen, classes, title): \n",
    "    predictions = model.predict(\n",
    "        datagen,\n",
    "        verbose = 1,\n",
    "    )\n",
    "    print(predictions)\n",
    "    predictions.squeeze().argmax(axis = -1)\n",
    "    classification__report = classification_report(\n",
    "        classes, \n",
    "        predictions.squeeze().argmax(axis = 1)\n",
    "    )\n",
    "    confusion__matrix = confusion_matrix(\n",
    "        classes,\n",
    "        predictions.squeeze().argmax(axis = 1)\n",
    "    )\n",
    "    np.set_printoptions(precision = 2)\n",
    "    fig2 = plt.figure(figsize = (7,6))\n",
    "    plot_confusion_matrix(\n",
    "        confusion__matrix,\n",
    "        classes = np.unique(classes),\n",
    "        normalise = True,\n",
    "        title = title\n",
    "    )\n",
    "    plt.show()\n",
    "\n",
    "graph_predictions(test_model, test_datagen, classes,  \"Best Test Model on Test Set\")\n",
    "graph_predictions(test_model, val_datagen, val_classes, \"Best Test Model on Val Set\")\n",
    "graph_predictions(val_model, test_datagen, classes,  \"Best Val Model on Test Set\")\n",
    "graph_predictions(val_model, val_datagen, val_classes, \"Best Val Model on Val Set\")\n",
    "# DONT FORGET TO CHANGE WHICH MODEL YOU'RE LOADING"
   ]
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "cc378a7666fd3eee2c39db92d67f0f5595e9050382902d6af70c22ea2e477f8d"
  },
  "kernelspec": {
   "display_name": "Python 3.9.7 ('opencv-course')",
   "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": "3.10.4"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}