example-stl10.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "f611aeec",
   "metadata": {},
   "source": [
    "# LAVA Data Value\n",
    "## For Mislabeled Data Detection (STL10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "dcf3bcff",
   "metadata": {},
   "outputs": [],
   "source": [
    "import lava"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "4e34f958",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import torchvision\n",
    "# print(torch.__version__)\n",
    "# print(torchvision.__version__)\n",
    "# print(torch.version.cuda)\n",
    "# print(torch.cuda.is_available())\n",
    "\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "import torchvision.models as models\n",
    "from torch.autograd import Variable\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "from torch import tensor\n",
    "from torchvision import datasets, transforms\n",
    "import pandas as pd\n",
    "import numpy as n\n",
    "\n",
    "from torch.utils.data import Dataset, TensorDataset, DataLoader"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b4b3bded",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "! nvidia-smi"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a54003f7",
   "metadata": {},
   "outputs": [],
   "source": [
    "cuda_num = 0\n",
    "import torchvision\n",
    "print(torchvision.__version__)\n",
    "import torch\n",
    "print(torch.__version__)\n",
    "import os\n",
    "os.environ[\"CUDA_VISIBLE_DEVICES\"]=str(cuda_num)\n",
    "print(os.environ[\"CUDA_VISIBLE_DEVICES\"])\n",
    "torch.cuda.set_device(cuda_num)\n",
    "print(\"Cuda device: \", torch.cuda.current_device())\n",
    "print(\"cude devices: \", torch.cuda.device_count())\n",
    "device = torch.device('cuda:' + str(cuda_num) if torch.cuda.is_available() else 'cpu')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "defdeae8",
   "metadata": {},
   "source": [
    "### Set Parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "23633651",
   "metadata": {},
   "outputs": [],
   "source": [
    "training_size = 5000\n",
    "valid_size = 2000\n",
    "resize = 32\n",
    "portion = 0.3"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7e0b7fb1",
   "metadata": {},
   "source": [
    "### Loading Corrupted Data into DataLoaders\n",
    "### Train and Test Loaders with Indices of Corrupted Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "0f7319d3",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "seeed\n",
      "Files already downloaded and verified\n",
      "Files already downloaded and verified\n",
      "Train Type:  <class 'torchvision.datasets.stl10.STL10'>  Train:  Dataset STL10\n",
      "    Number of datapoints: 5000\n",
      "    Root location: /home/just/Programming/otdd-main/data\n",
      "    Split: train\n",
      "    StandardTransform\n",
      "Transform: Compose(\n",
      "               Resize(size=(32, 32), interpolation=bilinear, max_size=None, antialias=None)\n",
      "               ToTensor()\n",
      "               Normalize(mean=(0.485, 0.456, 0.406), std=(0.229, 0.224, 0.225))\n",
      "           )\n",
      "Teest Type:  <class 'torchvision.datasets.stl10.STL10'>  Train:  Dataset STL10\n",
      "    Number of datapoints: 8000\n",
      "    Root location: /home/just/Programming/otdd-main/data\n",
      "    Split: test\n",
      "    StandardTransform\n",
      "Transform: Compose(\n",
      "               Resize(size=(32, 32), interpolation=bilinear, max_size=None, antialias=None)\n",
      "               ToTensor()\n",
      "               Normalize(mean=(0.485, 0.456, 0.406), std=(0.229, 0.224, 0.225))\n",
      "           )\n",
      "FOLD IDSSS:  [   0    1    2 ... 4997 4998 4999]\n",
      "K:  train  IDXS:  [   0    1    2 ... 4997 4998 4999]\n",
      "SAMPLER:  <class 'otdd.pytorch.datasets.SubsetSampler'> \n",
      "\n",
      "FOLD SAMPLER:  {'train': <otdd.pytorch.datasets.SubsetSampler object at 0x7fa1a41a2fa0>} \n",
      "\n",
      "TRAIN:  Dataset STL10\n",
      "    Number of datapoints: 5000\n",
      "    Root location: /home/just/Programming/otdd-main/data\n",
      "    Split: train\n",
      "    StandardTransform\n",
      "Transform: Compose(\n",
      "               Resize(size=(32, 32), interpolation=bilinear, max_size=None, antialias=None)\n",
      "               ToTensor()\n",
      "               Normalize(mean=(0.485, 0.456, 0.406), std=(0.229, 0.224, 0.225))\n",
      "           ) \n",
      "\n",
      "IDXS:  [   0    1    2 ... 4997 4998 4999] \n",
      "\n",
      "\n",
      "STL11\n",
      "len classes:  10\n",
      "len train:  5000\n",
      "i = 0 \n",
      "i = 1000 \n",
      "i = 2000 \n",
      "i = 3000 \n",
      "i = 4000 \n",
      "MAX TEST:  2000\n",
      "Fold Sizes: 5000/8000 (train/test)\n"
     ]
    }
   ],
   "source": [
    "loaders, shuffle_ind = lava.load_data_corrupted(corrupt_type='shuffle', dataname='STL10', resize=resize,\n",
    "                                        training_size=training_size, test_size=valid_size, currupt_por=portion)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f59b1c38",
   "metadata": {},
   "source": [
    "### Load Feature Extractor for a Given Dataset\n",
    "### Checkpoints provided and training files provided"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "58be73c8",
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_extractor = lava.load_pretrained_feature_extractor('preresnet18_test_stl10.pth', device)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d345bcac",
   "metadata": {},
   "source": [
    "### Compute Dual Solution of Optimal Transport Problem"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "5abb7145",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WROGNG: unexpected line 659\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/79 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/32 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/45 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/45 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/100 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finished first label distance in: 26.34044051170349 s.\n",
      "load datasets again?\n",
      "finisehd load datasets in 0.7925167083740234s.\n",
      "Calculate the same thing again...\n",
      "This took 0.005351066589355469 s.\n",
      "before Loss\n",
      "x:  tensor([[[1332.5459, 3423.4160,  490.6534,  ..., 1607.5631, 2216.5532,\n",
      "          1078.6294],\n",
      "         [2925.5969, 3242.3452,  991.0903,  ..., 1027.2827, 3962.2500,\n",
      "          2039.5393],\n",
      "         [1888.4187, 3969.7878,  406.4285,  ..., 1373.5833, 3360.2104,\n",
      "          1685.6584],\n",
      "         ...,\n",
      "         [2476.4060, 3421.6416, 1880.0450,  ..., 3688.2124, 2613.2915,\n",
      "          1599.5713],\n",
      "         [1480.9888, 3007.9954, 1007.4749,  ..., 2048.5557, 2138.5688,\n",
      "          1008.3782],\n",
      "         [2817.9932, 3489.5110, 1094.2736,  ...,  794.5531, 3329.8154,\n",
      "          1659.2576]]], device='cuda:0')\n",
      "x shape:  torch.Size([1, 5000, 2000])\n",
      "y:  tensor([[[ 558.8251, 1381.0925,  467.7468,  ..., 1158.1200, 2032.8157,\n",
      "          2032.8157],\n",
      "         [2104.1140, 1897.9562, 2559.4919,  ..., 3450.8323, 1225.1885,\n",
      "          1225.1885],\n",
      "         [ 629.5283, 1137.2161,  819.1660,  ..., 1697.6150, 1958.6172,\n",
      "          1958.6172],\n",
      "         ...,\n",
      "         [ 629.5283, 1137.2161,  819.1660,  ..., 1697.6150, 1958.6172,\n",
      "          1958.6172],\n",
      "         [ 466.6659, 1486.4958,  773.7169,  ..., 1349.8907, 2112.9458,\n",
      "          2112.9458],\n",
      "         [ 466.6659, 1486.4958,  773.7169,  ..., 1349.8907, 2112.9458,\n",
      "          2112.9458]]], device='cuda:0')\n",
      "distance calculation takes 28.4696 seconds\n"
     ]
    }
   ],
   "source": [
    "dual_sol, trained_with_flag = lava.compute_dual(feature_extractor, loaders['train'], loaders['test'], \n",
    "                                                training_size, shuffle_ind, resize=resize)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3598f5b7",
   "metadata": {},
   "source": [
    "### Compute the Data Values from Calibrated Gradient\n",
    "### Visualize Detection Rate Based on LAVA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "b8f745db",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "inspected: 10, found: 10 detection rate: 0.01 baseline: 1.8\n",
      "inspected: 20, found: 20 detection rate: 0.01 baseline: 3.6\n",
      "inspected: 30, found: 29 detection rate: 0.02 baseline: 5.4\n",
      "inspected: 40, found: 39 detection rate: 0.03 baseline: 7.2\n",
      "inspected: 50, found: 49 detection rate: 0.03 baseline: 9.0\n",
      "inspected: 60, found: 59 detection rate: 0.04 baseline: 10.8\n",
      "inspected: 70, found: 69 detection rate: 0.05 baseline: 12.6\n",
      "inspected: 80, found: 79 detection rate: 0.05 baseline: 14.4\n",
      "inspected: 90, found: 89 detection rate: 0.06 baseline: 16.2\n",
      "inspected: 100, found: 99 detection rate: 0.07 baseline: 18.0\n",
      "inspected: 110, found: 109 detection rate: 0.07 baseline: 19.8\n",
      "inspected: 120, found: 119 detection rate: 0.08 baseline: 21.6\n",
      "inspected: 130, found: 129 detection rate: 0.09 baseline: 23.400000000000002\n",
      "inspected: 140, found: 139 detection rate: 0.09 baseline: 25.2\n",
      "inspected: 150, found: 149 detection rate: 0.10 baseline: 27.0\n",
      "inspected: 160, found: 159 detection rate: 0.11 baseline: 28.8\n",
      "inspected: 170, found: 168 detection rate: 0.11 baseline: 30.6\n",
      "inspected: 180, found: 178 detection rate: 0.12 baseline: 32.4\n",
      "inspected: 190, found: 188 detection rate: 0.13 baseline: 34.2\n",
      "inspected: 200, found: 198 detection rate: 0.13 baseline: 36.0\n",
      "inspected: 210, found: 208 detection rate: 0.14 baseline: 37.800000000000004\n",
      "inspected: 220, found: 218 detection rate: 0.15 baseline: 39.6\n",
      "inspected: 230, found: 228 detection rate: 0.15 baseline: 41.4\n",
      "inspected: 240, found: 237 detection rate: 0.16 baseline: 43.2\n",
      "inspected: 250, found: 245 detection rate: 0.16 baseline: 45.0\n",
      "inspected: 260, found: 255 detection rate: 0.17 baseline: 46.800000000000004\n",
      "inspected: 270, found: 265 detection rate: 0.18 baseline: 48.6\n",
      "inspected: 280, found: 275 detection rate: 0.18 baseline: 50.4\n",
      "inspected: 290, found: 285 detection rate: 0.19 baseline: 52.2\n",
      "inspected: 300, found: 294 detection rate: 0.20 baseline: 54.0\n",
      "inspected: 310, found: 303 detection rate: 0.20 baseline: 55.800000000000004\n",
      "inspected: 320, found: 312 detection rate: 0.21 baseline: 57.6\n",
      "inspected: 330, found: 322 detection rate: 0.21 baseline: 59.4\n",
      "inspected: 340, found: 331 detection rate: 0.22 baseline: 61.2\n",
      "inspected: 350, found: 341 detection rate: 0.23 baseline: 63.0\n",
      "inspected: 360, found: 351 detection rate: 0.23 baseline: 64.8\n",
      "inspected: 370, found: 361 detection rate: 0.24 baseline: 66.60000000000001\n",
      "inspected: 380, found: 370 detection rate: 0.25 baseline: 68.4\n",
      "inspected: 390, found: 379 detection rate: 0.25 baseline: 70.2\n",
      "inspected: 400, found: 389 detection rate: 0.26 baseline: 72.0\n",
      "inspected: 410, found: 397 detection rate: 0.26 baseline: 73.8\n",
      "inspected: 420, found: 406 detection rate: 0.27 baseline: 75.60000000000001\n",
      "inspected: 430, found: 413 detection rate: 0.28 baseline: 77.4\n",
      "inspected: 440, found: 421 detection rate: 0.28 baseline: 79.2\n",
      "inspected: 450, found: 431 detection rate: 0.29 baseline: 81.0\n",
      "inspected: 460, found: 441 detection rate: 0.29 baseline: 82.8\n",
      "inspected: 470, found: 448 detection rate: 0.30 baseline: 84.60000000000001\n",
      "inspected: 480, found: 458 detection rate: 0.31 baseline: 86.4\n",
      "inspected: 490, found: 467 detection rate: 0.31 baseline: 88.2\n",
      "inspected: 500, found: 475 detection rate: 0.32 baseline: 90.0\n",
      "inspected: 510, found: 483 detection rate: 0.32 baseline: 91.8\n",
      "inspected: 520, found: 493 detection rate: 0.33 baseline: 93.60000000000001\n",
      "inspected: 530, found: 503 detection rate: 0.34 baseline: 95.4\n",
      "inspected: 540, found: 509 detection rate: 0.34 baseline: 97.2\n",
      "inspected: 550, found: 519 detection rate: 0.35 baseline: 99.0\n",
      "inspected: 560, found: 528 detection rate: 0.35 baseline: 100.8\n",
      "inspected: 570, found: 535 detection rate: 0.36 baseline: 102.60000000000001\n",
      "inspected: 580, found: 545 detection rate: 0.36 baseline: 104.4\n",
      "inspected: 590, found: 553 detection rate: 0.37 baseline: 106.2\n",
      "inspected: 600, found: 560 detection rate: 0.37 baseline: 108.0\n",
      "inspected: 610, found: 566 detection rate: 0.38 baseline: 109.8\n",
      "inspected: 620, found: 571 detection rate: 0.38 baseline: 111.60000000000001\n",
      "inspected: 630, found: 576 detection rate: 0.38 baseline: 113.4\n",
      "inspected: 640, found: 584 detection rate: 0.39 baseline: 115.2\n",
      "inspected: 650, found: 591 detection rate: 0.39 baseline: 117.0\n",
      "inspected: 660, found: 600 detection rate: 0.40 baseline: 118.8\n",
      "inspected: 670, found: 609 detection rate: 0.41 baseline: 120.60000000000001\n",
      "inspected: 680, found: 619 detection rate: 0.41 baseline: 122.4\n",
      "inspected: 690, found: 625 detection rate: 0.42 baseline: 124.2\n",
      "inspected: 700, found: 635 detection rate: 0.42 baseline: 126.0\n",
      "inspected: 710, found: 641 detection rate: 0.43 baseline: 127.8\n",
      "inspected: 720, found: 650 detection rate: 0.43 baseline: 129.6\n",
      "inspected: 730, found: 656 detection rate: 0.44 baseline: 131.4\n",
      "inspected: 740, found: 666 detection rate: 0.44 baseline: 133.20000000000002\n",
      "inspected: 750, found: 673 detection rate: 0.45 baseline: 135.0\n",
      "inspected: 760, found: 681 detection rate: 0.45 baseline: 136.8\n",
      "inspected: 770, found: 686 detection rate: 0.46 baseline: 138.6\n",
      "inspected: 780, found: 696 detection rate: 0.46 baseline: 140.4\n",
      "inspected: 790, found: 703 detection rate: 0.47 baseline: 142.20000000000002\n",
      "inspected: 800, found: 710 detection rate: 0.47 baseline: 144.0\n",
      "inspected: 810, found: 717 detection rate: 0.48 baseline: 145.8\n",
      "inspected: 820, found: 723 detection rate: 0.48 baseline: 147.6\n",
      "inspected: 830, found: 730 detection rate: 0.49 baseline: 149.4\n",
      "inspected: 840, found: 735 detection rate: 0.49 baseline: 151.20000000000002\n",
      "inspected: 850, found: 742 detection rate: 0.49 baseline: 153.0\n",
      "inspected: 860, found: 746 detection rate: 0.50 baseline: 154.8\n",
      "inspected: 870, found: 751 detection rate: 0.50 baseline: 156.6\n",
      "inspected: 880, found: 758 detection rate: 0.51 baseline: 158.4\n",
      "inspected: 890, found: 766 detection rate: 0.51 baseline: 160.20000000000002\n",
      "inspected: 900, found: 774 detection rate: 0.52 baseline: 162.0\n",
      "inspected: 910, found: 780 detection rate: 0.52 baseline: 163.8\n",
      "inspected: 920, found: 787 detection rate: 0.52 baseline: 165.6\n",
      "inspected: 930, found: 795 detection rate: 0.53 baseline: 167.4\n",
      "inspected: 940, found: 802 detection rate: 0.53 baseline: 169.20000000000002\n",
      "inspected: 950, found: 806 detection rate: 0.54 baseline: 171.0\n",
      "inspected: 960, found: 811 detection rate: 0.54 baseline: 172.8\n",
      "inspected: 970, found: 817 detection rate: 0.54 baseline: 174.6\n",
      "inspected: 980, found: 823 detection rate: 0.55 baseline: 176.4\n",
      "inspected: 990, found: 828 detection rate: 0.55 baseline: 178.20000000000002\n",
      "inspected: 1000, found: 834 detection rate: 0.56 baseline: 180.0\n",
      "inspected: 1010, found: 839 detection rate: 0.56 baseline: 181.8\n",
      "inspected: 1020, found: 845 detection rate: 0.56 baseline: 183.6\n",
      "inspected: 1030, found: 849 detection rate: 0.57 baseline: 185.4\n",
      "inspected: 1040, found: 856 detection rate: 0.57 baseline: 187.20000000000002\n",
      "inspected: 1050, found: 862 detection rate: 0.57 baseline: 189.0\n",
      "inspected: 1060, found: 865 detection rate: 0.58 baseline: 190.8\n",
      "inspected: 1070, found: 870 detection rate: 0.58 baseline: 192.6\n",
      "inspected: 1080, found: 878 detection rate: 0.59 baseline: 194.4\n",
      "inspected: 1090, found: 882 detection rate: 0.59 baseline: 196.20000000000002\n",
      "inspected: 1100, found: 887 detection rate: 0.59 baseline: 198.0\n",
      "inspected: 1110, found: 892 detection rate: 0.59 baseline: 199.8\n",
      "inspected: 1120, found: 896 detection rate: 0.60 baseline: 201.6\n",
      "inspected: 1130, found: 899 detection rate: 0.60 baseline: 203.4\n",
      "inspected: 1140, found: 904 detection rate: 0.60 baseline: 205.20000000000002\n",
      "inspected: 1150, found: 909 detection rate: 0.61 baseline: 207.0\n",
      "inspected: 1160, found: 911 detection rate: 0.61 baseline: 208.8\n",
      "inspected: 1170, found: 917 detection rate: 0.61 baseline: 210.6\n",
      "inspected: 1180, found: 921 detection rate: 0.61 baseline: 212.4\n",
      "inspected: 1190, found: 928 detection rate: 0.62 baseline: 214.20000000000002\n",
      "inspected: 1200, found: 935 detection rate: 0.62 baseline: 216.0\n",
      "inspected: 1210, found: 941 detection rate: 0.63 baseline: 217.8\n",
      "inspected: 1220, found: 944 detection rate: 0.63 baseline: 219.6\n",
      "inspected: 1230, found: 949 detection rate: 0.63 baseline: 221.4\n",
      "inspected: 1240, found: 953 detection rate: 0.64 baseline: 223.20000000000002\n",
      "inspected: 1250, found: 959 detection rate: 0.64 baseline: 225.0\n",
      "inspected: 1260, found: 963 detection rate: 0.64 baseline: 226.8\n",
      "inspected: 1270, found: 968 detection rate: 0.65 baseline: 228.6\n",
      "inspected: 1280, found: 972 detection rate: 0.65 baseline: 230.4\n",
      "inspected: 1290, found: 978 detection rate: 0.65 baseline: 232.20000000000002\n",
      "inspected: 1300, found: 985 detection rate: 0.66 baseline: 234.0\n",
      "inspected: 1310, found: 989 detection rate: 0.66 baseline: 235.8\n",
      "inspected: 1320, found: 993 detection rate: 0.66 baseline: 237.6\n",
      "inspected: 1330, found: 995 detection rate: 0.66 baseline: 239.4\n",
      "inspected: 1340, found: 1002 detection rate: 0.67 baseline: 241.20000000000002\n",
      "inspected: 1350, found: 1006 detection rate: 0.67 baseline: 243.0\n",
      "inspected: 1360, found: 1010 detection rate: 0.67 baseline: 244.8\n",
      "inspected: 1370, found: 1012 detection rate: 0.67 baseline: 246.6\n",
      "inspected: 1380, found: 1013 detection rate: 0.68 baseline: 248.4\n",
      "inspected: 1390, found: 1017 detection rate: 0.68 baseline: 250.20000000000002\n",
      "inspected: 1400, found: 1020 detection rate: 0.68 baseline: 252.0\n",
      "inspected: 1410, found: 1025 detection rate: 0.68 baseline: 253.8\n",
      "inspected: 1420, found: 1029 detection rate: 0.69 baseline: 255.6\n",
      "inspected: 1430, found: 1032 detection rate: 0.69 baseline: 257.40000000000003\n",
      "inspected: 1440, found: 1037 detection rate: 0.69 baseline: 259.2\n",
      "inspected: 1450, found: 1039 detection rate: 0.69 baseline: 261.0\n",
      "inspected: 1460, found: 1044 detection rate: 0.70 baseline: 262.8\n",
      "inspected: 1470, found: 1049 detection rate: 0.70 baseline: 264.6\n",
      "inspected: 1480, found: 1054 detection rate: 0.70 baseline: 266.40000000000003\n",
      "inspected: 1490, found: 1057 detection rate: 0.70 baseline: 268.2\n",
      "inspected: 1500, found: 1060 detection rate: 0.71 baseline: 270.0\n",
      "inspected: 1510, found: 1066 detection rate: 0.71 baseline: 271.8\n",
      "inspected: 1520, found: 1070 detection rate: 0.71 baseline: 273.6\n",
      "inspected: 1530, found: 1072 detection rate: 0.71 baseline: 275.40000000000003\n",
      "inspected: 1540, found: 1078 detection rate: 0.72 baseline: 277.2\n",
      "inspected: 1550, found: 1080 detection rate: 0.72 baseline: 279.0\n",
      "inspected: 1560, found: 1085 detection rate: 0.72 baseline: 280.8\n",
      "inspected: 1570, found: 1088 detection rate: 0.73 baseline: 282.6\n",
      "inspected: 1580, found: 1092 detection rate: 0.73 baseline: 284.40000000000003\n",
      "inspected: 1590, found: 1094 detection rate: 0.73 baseline: 286.2\n",
      "inspected: 1600, found: 1095 detection rate: 0.73 baseline: 288.0\n",
      "inspected: 1610, found: 1097 detection rate: 0.73 baseline: 289.8\n",
      "inspected: 1620, found: 1099 detection rate: 0.73 baseline: 291.6\n",
      "inspected: 1630, found: 1104 detection rate: 0.74 baseline: 293.40000000000003\n",
      "inspected: 1640, found: 1106 detection rate: 0.74 baseline: 295.2\n",
      "inspected: 1650, found: 1108 detection rate: 0.74 baseline: 297.0\n",
      "inspected: 1660, found: 1111 detection rate: 0.74 baseline: 298.8\n",
      "inspected: 1670, found: 1114 detection rate: 0.74 baseline: 300.6\n",
      "inspected: 1680, found: 1117 detection rate: 0.74 baseline: 302.40000000000003\n",
      "inspected: 1690, found: 1119 detection rate: 0.75 baseline: 304.2\n",
      "inspected: 1700, found: 1121 detection rate: 0.75 baseline: 306.0\n",
      "inspected: 1710, found: 1126 detection rate: 0.75 baseline: 307.8\n",
      "inspected: 1720, found: 1128 detection rate: 0.75 baseline: 309.6\n",
      "inspected: 1730, found: 1133 detection rate: 0.76 baseline: 311.40000000000003\n",
      "inspected: 1740, found: 1137 detection rate: 0.76 baseline: 313.2\n",
      "inspected: 1750, found: 1139 detection rate: 0.76 baseline: 315.0\n",
      "inspected: 1760, found: 1142 detection rate: 0.76 baseline: 316.8\n",
      "inspected: 1770, found: 1146 detection rate: 0.76 baseline: 318.6\n",
      "inspected: 1780, found: 1152 detection rate: 0.77 baseline: 320.40000000000003\n",
      "inspected: 1790, found: 1154 detection rate: 0.77 baseline: 322.2\n",
      "inspected: 1800, found: 1157 detection rate: 0.77 baseline: 324.0\n",
      "inspected: 1810, found: 1160 detection rate: 0.77 baseline: 325.8\n",
      "inspected: 1820, found: 1162 detection rate: 0.77 baseline: 327.6\n",
      "inspected: 1830, found: 1165 detection rate: 0.78 baseline: 329.40000000000003\n",
      "inspected: 1840, found: 1168 detection rate: 0.78 baseline: 331.2\n",
      "inspected: 1850, found: 1172 detection rate: 0.78 baseline: 333.0\n",
      "inspected: 1860, found: 1177 detection rate: 0.78 baseline: 334.8\n",
      "inspected: 1870, found: 1180 detection rate: 0.79 baseline: 336.6\n",
      "inspected: 1880, found: 1183 detection rate: 0.79 baseline: 338.40000000000003\n",
      "inspected: 1890, found: 1189 detection rate: 0.79 baseline: 340.2\n",
      "inspected: 1900, found: 1193 detection rate: 0.80 baseline: 342.0\n",
      "inspected: 1910, found: 1195 detection rate: 0.80 baseline: 343.8\n",
      "inspected: 1920, found: 1198 detection rate: 0.80 baseline: 345.6\n",
      "inspected: 1930, found: 1198 detection rate: 0.80 baseline: 347.40000000000003\n",
      "inspected: 1940, found: 1200 detection rate: 0.80 baseline: 349.2\n",
      "inspected: 1950, found: 1200 detection rate: 0.80 baseline: 351.0\n",
      "inspected: 1960, found: 1203 detection rate: 0.80 baseline: 352.8\n",
      "inspected: 1970, found: 1203 detection rate: 0.80 baseline: 354.6\n",
      "inspected: 1980, found: 1207 detection rate: 0.80 baseline: 356.40000000000003\n",
      "inspected: 1990, found: 1211 detection rate: 0.81 baseline: 358.2\n",
      "inspected: 2000, found: 1212 detection rate: 0.81 baseline: 360.0\n",
      "inspected: 2010, found: 1214 detection rate: 0.81 baseline: 361.8\n",
      "inspected: 2020, found: 1216 detection rate: 0.81 baseline: 363.6\n",
      "inspected: 2030, found: 1217 detection rate: 0.81 baseline: 365.40000000000003\n",
      "inspected: 2040, found: 1218 detection rate: 0.81 baseline: 367.2\n",
      "inspected: 2050, found: 1220 detection rate: 0.81 baseline: 369.0\n",
      "inspected: 2060, found: 1225 detection rate: 0.82 baseline: 370.8\n",
      "inspected: 2070, found: 1226 detection rate: 0.82 baseline: 372.6\n",
      "inspected: 2080, found: 1229 detection rate: 0.82 baseline: 374.40000000000003\n",
      "inspected: 2090, found: 1233 detection rate: 0.82 baseline: 376.2\n",
      "inspected: 2100, found: 1236 detection rate: 0.82 baseline: 378.0\n",
      "inspected: 2110, found: 1236 detection rate: 0.82 baseline: 379.8\n",
      "inspected: 2120, found: 1238 detection rate: 0.83 baseline: 381.6\n",
      "inspected: 2130, found: 1240 detection rate: 0.83 baseline: 383.40000000000003\n",
      "inspected: 2140, found: 1241 detection rate: 0.83 baseline: 385.2\n",
      "inspected: 2150, found: 1244 detection rate: 0.83 baseline: 387.0\n",
      "inspected: 2160, found: 1249 detection rate: 0.83 baseline: 388.8\n",
      "inspected: 2170, found: 1249 detection rate: 0.83 baseline: 390.6\n",
      "inspected: 2180, found: 1250 detection rate: 0.83 baseline: 392.40000000000003\n",
      "inspected: 2190, found: 1250 detection rate: 0.83 baseline: 394.2\n",
      "inspected: 2200, found: 1253 detection rate: 0.84 baseline: 396.0\n",
      "inspected: 2210, found: 1256 detection rate: 0.84 baseline: 397.8\n",
      "inspected: 2220, found: 1257 detection rate: 0.84 baseline: 399.6\n",
      "inspected: 2230, found: 1257 detection rate: 0.84 baseline: 401.40000000000003\n",
      "inspected: 2240, found: 1258 detection rate: 0.84 baseline: 403.2\n",
      "inspected: 2250, found: 1259 detection rate: 0.84 baseline: 405.0\n",
      "inspected: 2260, found: 1260 detection rate: 0.84 baseline: 406.8\n",
      "inspected: 2270, found: 1264 detection rate: 0.84 baseline: 408.6\n",
      "inspected: 2280, found: 1267 detection rate: 0.84 baseline: 410.40000000000003\n",
      "inspected: 2290, found: 1268 detection rate: 0.85 baseline: 412.2\n",
      "inspected: 2300, found: 1272 detection rate: 0.85 baseline: 414.0\n",
      "inspected: 2310, found: 1274 detection rate: 0.85 baseline: 415.8\n",
      "inspected: 2320, found: 1276 detection rate: 0.85 baseline: 417.6\n",
      "inspected: 2330, found: 1278 detection rate: 0.85 baseline: 419.40000000000003\n",
      "inspected: 2340, found: 1278 detection rate: 0.85 baseline: 421.2\n",
      "inspected: 2350, found: 1281 detection rate: 0.85 baseline: 423.0\n",
      "inspected: 2360, found: 1281 detection rate: 0.85 baseline: 424.8\n",
      "inspected: 2370, found: 1283 detection rate: 0.86 baseline: 426.6\n",
      "inspected: 2380, found: 1284 detection rate: 0.86 baseline: 428.40000000000003\n",
      "inspected: 2390, found: 1286 detection rate: 0.86 baseline: 430.2\n",
      "inspected: 2400, found: 1290 detection rate: 0.86 baseline: 432.0\n",
      "inspected: 2410, found: 1291 detection rate: 0.86 baseline: 433.8\n",
      "inspected: 2420, found: 1293 detection rate: 0.86 baseline: 435.6\n",
      "inspected: 2430, found: 1297 detection rate: 0.86 baseline: 437.40000000000003\n",
      "inspected: 2440, found: 1299 detection rate: 0.87 baseline: 439.2\n",
      "inspected: 2450, found: 1300 detection rate: 0.87 baseline: 441.0\n",
      "inspected: 2460, found: 1301 detection rate: 0.87 baseline: 442.8\n",
      "inspected: 2470, found: 1302 detection rate: 0.87 baseline: 444.6\n",
      "inspected: 2480, found: 1305 detection rate: 0.87 baseline: 446.40000000000003\n",
      "inspected: 2490, found: 1307 detection rate: 0.87 baseline: 448.2\n",
      "inspected: 2500, found: 1310 detection rate: 0.87 baseline: 450.0\n",
      "inspected: 2510, found: 1314 detection rate: 0.88 baseline: 451.8\n",
      "inspected: 2520, found: 1316 detection rate: 0.88 baseline: 453.6\n",
      "inspected: 2530, found: 1319 detection rate: 0.88 baseline: 455.40000000000003\n",
      "inspected: 2540, found: 1320 detection rate: 0.88 baseline: 457.2\n",
      "inspected: 2550, found: 1322 detection rate: 0.88 baseline: 459.0\n",
      "inspected: 2560, found: 1325 detection rate: 0.88 baseline: 460.8\n",
      "inspected: 2570, found: 1327 detection rate: 0.88 baseline: 462.6\n",
      "inspected: 2580, found: 1328 detection rate: 0.89 baseline: 464.40000000000003\n",
      "inspected: 2590, found: 1331 detection rate: 0.89 baseline: 466.2\n",
      "inspected: 2600, found: 1331 detection rate: 0.89 baseline: 468.0\n",
      "inspected: 2610, found: 1333 detection rate: 0.89 baseline: 469.8\n",
      "inspected: 2620, found: 1336 detection rate: 0.89 baseline: 471.6\n",
      "inspected: 2630, found: 1341 detection rate: 0.89 baseline: 473.40000000000003\n",
      "inspected: 2640, found: 1344 detection rate: 0.90 baseline: 475.2\n",
      "inspected: 2650, found: 1346 detection rate: 0.90 baseline: 477.0\n",
      "inspected: 2660, found: 1349 detection rate: 0.90 baseline: 478.8\n",
      "inspected: 2670, found: 1350 detection rate: 0.90 baseline: 480.6\n",
      "inspected: 2680, found: 1353 detection rate: 0.90 baseline: 482.40000000000003\n",
      "inspected: 2690, found: 1353 detection rate: 0.90 baseline: 484.2\n",
      "inspected: 2700, found: 1353 detection rate: 0.90 baseline: 486.0\n",
      "inspected: 2710, found: 1353 detection rate: 0.90 baseline: 487.8\n",
      "inspected: 2720, found: 1355 detection rate: 0.90 baseline: 489.6\n",
      "inspected: 2730, found: 1355 detection rate: 0.90 baseline: 491.40000000000003\n",
      "inspected: 2740, found: 1355 detection rate: 0.90 baseline: 493.2\n",
      "inspected: 2750, found: 1356 detection rate: 0.90 baseline: 495.0\n",
      "inspected: 2760, found: 1362 detection rate: 0.91 baseline: 496.8\n",
      "inspected: 2770, found: 1362 detection rate: 0.91 baseline: 498.6\n",
      "inspected: 2780, found: 1364 detection rate: 0.91 baseline: 500.40000000000003\n",
      "inspected: 2790, found: 1366 detection rate: 0.91 baseline: 502.2\n",
      "inspected: 2800, found: 1367 detection rate: 0.91 baseline: 504.0\n",
      "inspected: 2810, found: 1369 detection rate: 0.91 baseline: 505.8\n",
      "inspected: 2820, found: 1370 detection rate: 0.91 baseline: 507.6\n",
      "inspected: 2830, found: 1371 detection rate: 0.91 baseline: 509.40000000000003\n",
      "inspected: 2840, found: 1371 detection rate: 0.91 baseline: 511.2\n",
      "inspected: 2850, found: 1374 detection rate: 0.92 baseline: 513.0\n",
      "inspected: 2860, found: 1374 detection rate: 0.92 baseline: 514.8000000000001\n",
      "inspected: 2870, found: 1374 detection rate: 0.92 baseline: 516.6\n",
      "inspected: 2880, found: 1374 detection rate: 0.92 baseline: 518.4\n",
      "inspected: 2890, found: 1375 detection rate: 0.92 baseline: 520.2\n",
      "inspected: 2900, found: 1376 detection rate: 0.92 baseline: 522.0\n",
      "inspected: 2910, found: 1377 detection rate: 0.92 baseline: 523.8000000000001\n",
      "inspected: 2920, found: 1377 detection rate: 0.92 baseline: 525.6\n",
      "inspected: 2930, found: 1378 detection rate: 0.92 baseline: 527.4\n",
      "inspected: 2940, found: 1379 detection rate: 0.92 baseline: 529.2\n",
      "inspected: 2950, found: 1379 detection rate: 0.92 baseline: 531.0\n",
      "inspected: 2960, found: 1379 detection rate: 0.92 baseline: 532.8000000000001\n",
      "inspected: 2970, found: 1382 detection rate: 0.92 baseline: 534.6\n",
      "inspected: 2980, found: 1384 detection rate: 0.92 baseline: 536.4\n",
      "inspected: 2990, found: 1386 detection rate: 0.92 baseline: 538.2\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXgAAAEWCAYAAABsY4yMAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8qNh9FAAAACXBIWXMAAAsTAAALEwEAmpwYAAAazUlEQVR4nO3de5wcZZ3v8c93JhkSEiBAIhACCZHAGnSFkJU7yhrA6wFXPNxUXNcb4upZlV28rIt42aMeOa/FZUVxWUEFL7CsgYXFwHIJysWEiyFBCASQW8JEAiEQJpf57R/1TNKZdE9q0l1zefi+X69+TVd1ddWvq5/6TvXT1VWKCMzMLD9tg12AmZlVwwFvZpYpB7yZWaYc8GZmmXLAm5llygFvZpYpB7wNOZIelTQr3f+8pB8Mdk1Vk7SnpFWS2ge7lh6174MNTw546xdJJ0m6Q9KLkp5J9z8uSVUsLyK+HhEfanY+kqZICkkjWlFXieWsSrdHJZ21pedFxB8iYmxErK+yPntlccBbaZI+A/wT8C1gV2AX4GPAYUBHg+cMmT3SATYuIsYCJwNfkvSWwS7IXnkc8FaKpB2Ac4CPR8TlEfFCFO6OiFMjoitN90NJ35V0jaQXgaMkvV3S3ZJWSnpc0tm95v0+SY9J+qOkL/R67GxJP64ZPljSbyQ9J+leSW+qeewmSV+R9GtJL0j6laTx6eFb0t/n0p71Ib2WM1HSakk71Yw7QNJySSMl7S3pZknPp3E/K7PeIuI2YCHwWkltkr6YXuszki5J63WzTxiSPiBpSXodj0g6NY0vM4/TJP0h1blhfabnniXp4bSuf97r9TZ8H2x4csBbWYcA2wC/LDHtKcDXgO2AW4EXgfcD44C3A6dLOh5A0nTgu8D7gInAzsCkejOVtDvwn8BXgZ2AzwJXSJrQa9l/CbyK4lPFZ9P4I9Pfcakr5LbaeUfEU8BtwLt7zevyiFgLfAX4FbBjqu87W1oJKhwG7AfcDXwg3Y4CpgJjgX+u87wxwHnAWyNiO+BQ4J70cJl5HA7sC7yZ4tPDa9L4vwaOB95Isa5XAOenZZZ+H2z4cMBbWeOB5RGxrmdEzZ70aklH1kz7y4j4dUR0R8TLEXFTRCxIw78DLqMIGYATgKsj4pb0KeDvge4GNbwXuCYirknzmgPMA95WM82/RcSDEbEa+Dmwfz9e46UUXSqk7xROSuMA1gKTgYnpNd26hXktB54FfgCcFRE3AKcC50bEkohYBXwOOKnB9wLdFHv9oyPi6YhYmMaXmceXI2J1RNwL3Au8Po3/GPCFiHgireuzgRPSc/vzPtgw4YC3sv4IjK8Nkog4NCLGpcdq29LjtU+UdJCkGyV1SnqeImh6uk4m1k4fES+m+dUzGXhP+qfynKTnKPZWd6uZZmnN/Zco9nDLugI4RNJuFHv83cDc9NjfAgLulLRQ0ge3MK/xEbFjRLwmIs5L4yYCj9VM8xgwguK7jA3SOjiRYj09Lek/Jf1JP+bRaB1MBq6sWXf3A+vTc/vzPtgw4YC3sm4DuoDjSkzb+xSllwKzgT0iYgfgAoqwBHga2KNnQknbUnQP1PM48KOIGFdzGxMR/3cratp8gogVFN0wJ1J0z/w00ulWI2JpRHw4IiYCHwX+RdLeJZZb6ymKkO2xJ7AOWFanlusi4miKf16/By7s7zzqeJyi26d2/Y2KiCfp3/tgw4QD3kqJiOeAL1ME2wmStktf2u0PjNnC07cDno2IlyW9gSI8e1wOvEPS4ZI6KL7IbdQufwy8U9KxktoljZL0Jkll+oo7KfbIp25hukspvi84gY3dM0h6T81yVlD8w+hvF8ZlwN9I2kvSWODrwM9qu73SsnaRdFzqi+8CVtUsq9Q8GrgA+JqkyWk5EyT1/MPuz/tgw4TfQCstIr4JfJqiu2JZun0P+DvgN3089ePAOZJeAL5E0TfeM8+FwBkUYfo0RXg+0WD5j1N8gvg8RWA/DpxJiXYcES9RfPH769RFcXCDSWcD04ClqQ+7x58Bd0halab5VEQs2dJye7kI+BHFET2PAC9TfPHZWxvFen6Koh//jcDp/ZxHPf+Uav9Vei9uBw6C/r0PNnzIF/wwM8uT9+DNzDLlgDczy5QD3swsUw54M7NMVXpmvf4aP358TJkyZbDLMDMbNubPn788IibUe2xIBfyUKVOYN2/eYJdhZjZsSHqs0WPuojEzy5QD3swsUw54M7NMOeDNzDLlgDczy1RlAS/ponRJsfuqWoaZmTVW5WGSP6S4lNglFS4je3MWLWPu4k6OmDaBo6dvcl0IvnXdA1y/aCmzpu/K/nuM49I7iqOlpk/cgUVPPc/yVV2MH7sNK19ex++fXklbm4iA1WvW0R0wamQb7W1trFvfzZr13XT7vHOvOO2C7th4svy2NNwsASPaRHcE3QHbjRrBuG07WNW1lnXrg1VdRRsUMKajndMO22tDG36480VWda1l93Gj+eSb9wHYMP7ZF9cAwc5jOpg6YSynHFScGn/u4k4WPrVyQzvffpsRHD9jEmceuy9zFi3bsG2cctBkjp6+y2bb1cd/chc3P/AM23a0s+sOoxg/dhtOOWgyV979JDc/8Azjx3bwxXfst2EbrN32zjx236bWVV/beLMqPZukpCkUlwF7bZnpZ86cGT4OfqM5i5bxycvuZvXa9Ywe2c55Jx+wSQM7/8aHNkzbqg3TbLDUa8NtgjaJdQ0a94g20SaxZn39U/O/7XW7cf2iZRse72hv48NHTuWiWx/ZsF1Nn7g98x9bsXk9bHrC/zbB9943k3sef26Tbe+Mo/be6pDvaxsvS9L8iJhZ77FB74OX9BFJ8yTN6+zsHOxyhpS5iztZvXY9AKvXrmfu4o3r5/pFSzeZ1uFuw129NtwdNAx3KB5rFO4Av3moc5PH16zv5vpFSzfZrhY88Vz9eurUMndx52bbXu/h/uhrG2+FQQ/4iPh+RMyMiJkTJtT9te0r1hHTJjB6ZDsAo0e2c8S0jetn1vRdN5m2TZgNa/XacJuKvfRGRrSJjvbGMXbo3hM2ebyjvY1Z03fdZLt63aRx9eupU8sR0yZstu31Hu6PvrbxVhhSpyqwTR09fRfOO/mAuv1zPR8J3QdvzXil9sHvv8e4re6D79kOW9EH39c23grugzczG8YGpQ9e0mXAbcC+kp6Q9FdVLcvMzDZXWRdNRJxc1bzNzGzLBv1LVjMzq4YD3swsUw54M7NMOeDNzDLlgDczy5QD3swsUw54M7NMOeDNzDLlgDczy5QD3swsUw54M7NMOeDNzDLlgDczy5QD3swsUw54M7NMOeDNzDLlgDczy5QD3swsUw54M7NMOeDNzDLlgDczy5QD3swsUw54M7NMOeDNzDLlgDczy5QD3swsUw54M7NMOeDNzDLlgDczy5QD3swsUw54M7NMOeDNzDLlgDczy5QD3swsUw54M7NMOeDNzDLlgDczy5QD3swsUw54M7NMOeDNzDLlgDczy5QD3swsUw54M7NMOeDNzDLlgDczy5QD3swsUw54M7NMOeDNzDLlgDczy5QD3swsUw54M7NMOeDNzDLlgDczy5QD3swsUw54M7NMlQp4SZMlzUr3R0vartqyzMysWVsMeEkfBi4HvpdGTQL+o8KazMysBcrswZ8BHAasBIiIxcCrqizKzMyaVybguyJiTc+ApBFAVFeSmZm1QpmAv1nS54HRko4GfgFcVW1ZZmbWrDIBfxbQCSwAPgpcA3yxyqLMzKx5I7Y0QUR0Axemm5mZDRNbDHhJC9i8z/15YB7w1Yj4YxWFmZlZc7YY8MC1wHrg0jR8ErAtsBT4IfDOSiozM7OmlAn4WRExo2Z4gaS7ImKGpPdWVZiZmTWnzJes7ZLe0DMg6c+A9jS4rpKqzMysaWX24D8EXCRpLCCKHzx9SNIY4B+rLM7MzLZemaNofgu8TtIOafj5mod/XlVhZmbWnDJ78Eh6O7AfMEoSABFxToV1mZlZk8qcbOwC4ETgrym6aN4DTK64LjMza1KZL1kPjYj3Aysi4svAIcA+1ZZlZmbNKhPwq9PflyRNBNYCu1VXkpmZtUKZPvirJY0DvgXcRfGr1h9UWZSZmTWvzFE0X0l3r5B0NTCq15E0ZmY2BJU5F0078HZgSs/0koiIc6stzczMmlGmi+Yq4GWK0wV3V1uOmZm1SpmAnxQRf1p5JWZm1lJljqK5VtIxlVdiZmYtVWYP/nbgSkltFIdICoiI2L7SyszMrCllAv5cih83LYgIX2zbzGyYKNNF8zhwn8PdzGx4KbMHvwS4SdK1QFfPSB8maWY2tJUJ+EfSrSPdzMxsGCjzS9YvD0QhZmbWWg0DXtJVFOedqSsi/lclFZmZWUv0tQf//wasCjMza7mGAR8RNw9kIWZm1lplDpM0M7NhyAFvZpYpB7yZWaZ8FI2ZWabKHEXzF8CuwI/T8MnAsiqLMjOz5m3xKBpJ346ImTUPXSVpXuWVmZlZU8r0wY+RNLVnQNJewJjqSjIzs1Yocy6av6E42dgSinPBTwY+WmlVZmbWtDLnovkvSdOAP0mjfh8RXX09x8zMBt8Wu2gkbQucCXwiIu4F9pT0jsorMzOzppTpg/83YA3FVZ0AngS+WllFZmbWEmUC/tUR8U2K67ESES9R9MWbmdkQVibg10gaTfrRk6RXU3NlJzMzG5rKHEXzD8B/AXtI+glwGPCBKosyM7PmlTmKZo6ku4CDKbpmPhURyyuvzMzMmlJmDx5gFLAiTT9dEhFxS3VlmZlZs7YY8JK+AZwILAS60+gAHPBmZkNYmT3444F9/eMmM7PhpcxRNEuAkVUXYmZmrVVmD/4l4B5JN1BzeGREfLKyqszMrGllAn52upmZ2TBS5jDJiyV1APukUQ9ExNpqyzIzs2aVOYrmTcDFwKMUx8HvIek0HyZpZja0lemi+TZwTEQ8ACBpH+Ay4MAqCzMzs+aUOYpmZE+4A0TEg/ioGjOzIa/MHvw8ST9g40W33wv4mqxmZkNcmYA/HTgD6Dksci7wL5VVZGZmLVHmKJou4FzgXEk7AZP8q1Yzs6GvzCX7bpK0fQr3+cCFkv5/9aWZmVkzynzJukNErAT+ArgkIg4C3lxtWWZm1qwyAT9C0m7A/waurrgeMzNrkTIBfw5wHfBQRPxW0lRgcbVlmZlZs8p8yfoL4Bc1w0uAd1dZlJmZNa9hwEv624j4pqTvkC64XctnkzQzG9r62oO/P/31j5rMzIahhgEfEVelvxcDSBqbhlcNTGlmZtaMMsfBv1bS3RTXZF0kab6k/aovzczMmlHmKJrvA5+OiMkRsSfwGeDCassyM7NmlQn4MRFxY89ARNwEjKmsIjMza4kyJxtbIunvgR+l4fdSXIjbzMyGsDJ78B8EJgD/DlwBjE/jzMxsCOvrOPhRwMeAvYEFwGd8LVYzs+Gjrz34i4GZFOH+VuBbA1KRmZm1RF998NMj4nUAkv4VuHNgSjIzs1boaw9+Q3dMRKwbgFrMzKyF+tqDf72klem+gNFpWEBExPaVV2dmZlutr1MVtA9kIWZm1lplDpM0M7NhyAFvZpYpB7yZWaYc8GZmmXLAm5llygFvZpYpB7yZWaYc8GZmmXLAm5llygFvZpYpB7yZWaYc8GZmmXLAm5llygFvZpYpB7yZWaYc8GZmmXLAm5llygFvZpYpB7yZWaYc8GZmmXLAm5llygFvZpYpB7yZWaYc8GZmmXLAm5llygFvZpYpB7yZWaYc8GZmmXLAm5llygFvZpYpB7yZWaYc8GZmmXLAm5llygFvZpYpB7yZWaYc8GZmmXLAm5llygFvZpYpB7yZWaYc8GZmmXLAm5llygFvZpYpB7yZWaYc8GZmmXLAm5llygFvZpYpB7yZWaYc8GZmmXLAm5llygFvZpYpB7yZWaYc8GZmmXLAm5llygFvZpYpB7yZWaYqDXhJb5H0gKSHJJ1V5bLMzGxTI6qasaR24HzgaOAJ4LeSZkfEolYva86iZcxd3MkR0yZw9PRdSk9Xb/jSOx4D4JSDJm8ybvmqLsaP3YbpE3fg5gc7eXT5i3S0i65163lxTTdtgu5o9SuzZowe2UbX2m66e40f0Va8VzuP6WDPnccw/7EVmz13bEc7R+77KhY++TzLV73MyPY2VnWtY2R7G+1tbXS0i4NfPZ4/PPsSK1Z18fo9d2Thk8+zdOXL7Lr9Nrzj9btz84OdrFjVxfEzJrH/HuOYu7iT7UaN5IWX126xDfa0t5522KNn+uWr1vDwMy8wa/qunHnsvqXXSdltZTjJ8TW1iiKqSSVJhwBnR8SxafhzABHxj42eM3PmzJg3b16/ljNn0TI+edndrF67ntEj2znv5APqvsm9p/vg4Xtx0a2PbDJ84S1LWLO+iIOO9jY+fOTUTcaZba0RbWJdzR5AmTYIRTs8/9QZG8K/pw3XOuOovUuFfNltZTjJ8TX1l6T5ETGz3mNVdtHsDjxeM/xEGrcJSR+RNE/SvM7Ozn4vZO7izg0NfvXa9cxdXH8evae7ftHSzYZrN6w167s3G2e2tdb1+nhXpg1C0Q572nRtG651/aKlpWoou60MJzm+plYa9C9ZI+L7ETEzImZOmDCh388/YtoERo9sB4q9oiOm1Z9H7+lmTd91s+GO9o2ro6O9bbNxZltrRJs2GS7TBqFohz1turYN15o1fddSNZTdVoaTHF9TKw37LhpwH7zV5z74zeXYX53ja+qPvrpoqgz4EcCDwJuBJ4HfAqdExMJGz9nagDcze6XqK+ArO4omItZJ+gRwHdAOXNRXuJuZWWtVFvAAEXENcE2VyzAzs/r8DaKZWaYc8GZmmXLAm5llygFvZpapyg6T3BqSOoHHgPHA8kEupx7X1X9DtTbX1T+uq38Gsq7JEVH3F15DKuB7SJrX6LjOweS6+m+o1ua6+sd19c9QqctdNGZmmXLAm5llaqgG/PcHu4AGXFf/DdXaXFf/uK7+GRJ1Dck+eDMza95Q3YM3M7MmOeDNzDI1KAEv6T2SFkrqljSzZvwUSasl3ZNuF9Q8dqCkBekC3udJUhq/k6Q5khanvzu2uq702OfSsh+QdGzN+LoXFpe0l6Q70vifSerY2rp61XG2pCdr1tHbtrbGKg32BdclPZrayz2S5qVxdduKCuelWn8naUYL67hI0jOS7qsZ1+86JJ2Wpl8s6bSK6hr0tiVpD0k3SlqUtsVPpfGDus76qGvQ11mfImLAb8BrgH2Bm4CZNeOnAPc1eM6dwMGAgGuBt6bx3wTOSvfPAr5RQV3TgXuBbYC9gIcpToHcnu5PBTrSNNPTc34OnJTuXwCc3qJ1dzbw2Trj+11jhe/vgC+zTg2PAuN7javbVoC3pTal1MbuaGEdRwIzatt1f+sAdgKWpL87pvs7VlDXoLctYDdgRrq/HcU1JaYP9jrro65BX2d93QZlDz4i7o+IB8pOL2k3YPuIuD2KtXcJcHx6+Djg4nT/4prxrazrOOCnEdEVEY8ADwFvSLeHImJJRKwBfgocJ0nAnwOXt6KukvpVY8W1DMYyy2jUVo4DLonC7cC41OaaFhG3AM82WcexwJyIeDYiVgBzgLdUUFcjA9a2IuLpiLgr3X8BuJ/iWs6Dus76qKuRIbE9DsU++L0k3S3pZklHpHG7U1y0u0ftBbx3iYin0/2lQBXX7Gp0AfFG43cGnouIdXXqbYVPpI+jF2ljl1R/a6zSYCyztwB+JWm+pI+kcY3aykDX2986BrK+IdO2JE0BDgDuYAits151wRBaZ71VFvCSrpd0X51bX/+tngb2jIgDgE8Dl0ravuwy0959n8d9bmVdA2oLNX4XeDWwP8X6+vZg1jqEHR4RM4C3AmdIOrL2wTJtZSAMlTqSIdO2JI0FrgD+T0SsrH1sMNdZnbqGzDqrp8pL9s3aiud0AV3p/nxJDwP7UFzTdVLNpJPSOIBlknaLiKfTR7NnWl1XWtYeDZZfb/wfKT4qjkh78bXTb1HZGiVdCFy9lTVWqa9aBkREPJn+PiPpSoqPxo3aykDX2986ngTe1Gv8Ta0uKiKW9dwfzLYlaSRFiP4kIv49jR70dVavrqGyzhoZUl00kiZIak/3pwLTgCXpo9lKSQen/u33A79MT5sN9HxDflrN+FaaDZwkaRtJe6W67qS4kPg0FUfMdAAnAbPTHsaNwAmtrqtX3/C7gJ6jIPpVYytq6cNgLHMDSWMkbddzHziGYj01aiuzgfenIzIOBp6v6Q6oQn/ruA44RtKOqQvgmDSupYZC20rb978C90fEuTUPDeo6a1TXUFhnfarq29u+bmlFPEGxt74MuC6NfzewELgHuAt4Z81zZqaV9zDwz2z8Fe7OwA3AYuB6YKdW15Ue+0Ja9gOkI3hi47f4D6bHvlAzfirFG/oQ8Atgmxatux8BC4DfUTSM3ba2xorf4wFfZq91f2+6LexZfqO2QnEExvmp1gXUHEHVglouo/jovja1rb/amjqAD6a29BDwlxXVNehtCzicovvldykH7knLGNR11kddg77O+rr5VAVmZpkaUl00ZmbWOg54M7NMOeDNzDLlgDczy5QD3swsUw54y5KkXSRdKmlJOl3BbZLe1cT8zpb02XT/HElb84M5JO2vmjMOmlXJAW/ZST9K+Q/gloiYGhEHUvygZFKv6bbql9wR8aWIuH4ry9uf4jhos8o54C1Hfw6siYgN1xOIiMci4juSPiBptqT/Bm6QNFbSDZLuUnH++A3nJJL0BUkPSrqV4jTSPeN/KOmEdP9AFSfGmy/pup5fNkq6SdI3JN2Z5nFE+uXiOcCJKs4dfuIArQ97harsXDRmg2g/il9CNzID+NOIeDbtxb8rIlZKGg/cLml2muYkij3uEWl+82tnks5N8h3guIjoTIH9NYpfUAKMiIg3pC6Zf4iIWZK+RPFry0+07NWaNeCAt+xJOp/ip+ZrKH7WPicies6FLuDrKs422U1x6tZdgCOAKyPipTSPeucL2Rd4LTCn6BWineLn/z16TpQ1n+JiNmYDygFvOVpIcV4jACLijLR3Pi+NerFm2lOBCcCBEbFW0qPAqJLLEbAwIg5p8HhX+rseb2s2CNwHbzn6b2CUpNNrxm3bYNodgGdSuB8FTE7jbwGOlzQ6nZnynXWe+wAwQdIhUHTZSNpvC7W9QHHJN7PKOeAtO1GcQe944I2SHpF0J8Vl3v6uzuQ/AWZKWkBxGurfp3ncBfyM4oyU11Kc5rX3ctZQnBL6G5LupTjD4KFbKO9GYLq/ZLWB4LNJmpllynvwZmaZcsCbmWXKAW9mlikHvJlZphzwZmaZcsCbmWXKAW9mlqn/AULmtO+HRN3nAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of poisoned images:  1499  out of 10000.\n",
      "last index of poison 4675\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "calibrated_gradient = lava.compute_values_and_visualize(dual_sol, trained_with_flag, training_size, portion)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "a72963e8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[2663.899,\n",
       " 2418.3823,\n",
       " 2408.9248,\n",
       " 2350.5876,\n",
       " 2338.0808,\n",
       " 2329.4268,\n",
       " 2326.6265,\n",
       " 2297.5244,\n",
       " 2287.1013,\n",
       " 2263.9346,\n",
       " 2258.9753,\n",
       " 2253.043,\n",
       " 2252.0122,\n",
       " 2201.0159,\n",
       " 2147.6462,\n",
       " 2133.2493,\n",
       " 2081.9287,\n",
       " 2079.318,\n",
       " 2073.779,\n",
       " 2061.6963,\n",
       " 2053.6355,\n",
       " 2051.4255,\n",
       " 2036.6858,\n",
       " 2031.4175,\n",
       " 1995.8184,\n",
       " 1971.8591,\n",
       " 1971.0344,\n",
       " 1963.4426,\n",
       " 1947.4111,\n",
       " 1947.0498,\n",
       " 1945.0283,\n",
       " 1925.0142,\n",
       " 1884.5908,\n",
       " 1874.2131,\n",
       " 1867.8916,\n",
       " 1858.4092,\n",
       " 1850.8071,\n",
       " 1813.2207,\n",
       " 1803.3213,\n",
       " 1799.1411,\n",
       " 1789.0459,\n",
       " 1788.5667,\n",
       " 1784.4619,\n",
       " 1783.9531,\n",
       " 1781.6553,\n",
       " 1767.4937,\n",
       " 1763.6116,\n",
       " 1762.5593,\n",
       " 1750.5083,\n",
       " 1738.9062,\n",
       " 1736.5793,\n",
       " 1727.5774,\n",
       " 1714.363,\n",
       " 1705.2625,\n",
       " 1704.6936,\n",
       " 1701.2852,\n",
       " 1697.2239,\n",
       " 1694.096,\n",
       " 1687.8237,\n",
       " 1676.8035,\n",
       " 1669.9917,\n",
       " 1666.7358,\n",
       " 1662.8354,\n",
       " 1661.73,\n",
       " 1652.4321,\n",
       " 1650.4236,\n",
       " 1646.2275,\n",
       " 1641.9666,\n",
       " 1638.2361,\n",
       " 1631.4382,\n",
       " 1624.0493,\n",
       " 1622.6262,\n",
       " 1622.4265,\n",
       " 1622.2219,\n",
       " 1621.0908,\n",
       " 1619.897,\n",
       " 1619.4084,\n",
       " 1615.0515,\n",
       " 1612.9219,\n",
       " 1610.9905,\n",
       " 1609.1643,\n",
       " 1608.0625,\n",
       " 1597.667,\n",
       " 1595.4299,\n",
       " 1595.0217,\n",
       " 1594.2593,\n",
       " 1591.8801,\n",
       " 1590.6313,\n",
       " 1586.8433,\n",
       " 1585.0679,\n",
       " 1577.3813,\n",
       " 1576.6333,\n",
       " 1571.4861,\n",
       " 1569.4502,\n",
       " 1569.3916,\n",
       " 1564.5869,\n",
       " 1563.0879,\n",
       " 1561.2795,\n",
       " 1552.4768,\n",
       " 1547.9126,\n",
       " 1544.9043,\n",
       " 1544.02,\n",
       " 1543.3743,\n",
       " 1542.5935,\n",
       " 1540.04,\n",
       " 1535.8984,\n",
       " 1535.1047,\n",
       " 1529.2646,\n",
       " 1528.8088,\n",
       " 1525.8662,\n",
       " 1520.9766,\n",
       " 1518.9983,\n",
       " 1517.73,\n",
       " 1516.8323,\n",
       " 1512.9878,\n",
       " 1509.8906,\n",
       " 1508.8042,\n",
       " 1505.9153,\n",
       " 1499.7556,\n",
       " 1495.5806,\n",
       " 1493.7957,\n",
       " 1492.8123,\n",
       " 1490.3396,\n",
       " 1487.2336,\n",
       " 1480.8591,\n",
       " 1474.2617,\n",
       " 1472.8625,\n",
       " 1471.8333,\n",
       " 1470.8904,\n",
       " 1465.5349,\n",
       " 1464.8186,\n",
       " 1463.8066,\n",
       " 1461.0259,\n",
       " 1460.2793,\n",
       " 1459.6179,\n",
       " 1458.4612,\n",
       " 1454.7158,\n",
       " 1453.0215,\n",
       " 1448.5115,\n",
       " 1447.9233,\n",
       " 1443.7229,\n",
       " 1435.6265,\n",
       " 1433.5774,\n",
       " 1432.6968,\n",
       " 1424.9639,\n",
       " 1415.76,\n",
       " 1412.5178,\n",
       " 1410.4724,\n",
       " 1409.936,\n",
       " 1407.4138,\n",
       " 1406.1782,\n",
       " 1399.9534,\n",
       " 1398.0647,\n",
       " 1385.959,\n",
       " 1385.9431,\n",
       " 1383.1135,\n",
       " 1370.4912,\n",
       " 1368.9897,\n",
       " 1364.6028,\n",
       " 1355.8042,\n",
       " 1353.8494,\n",
       " 1352.5593,\n",
       " 1345.5139,\n",
       " 1344.8209,\n",
       " 1344.8202,\n",
       " 1344.1777,\n",
       " 1333.495,\n",
       " 1333.2297,\n",
       " 1330.2782,\n",
       " 1327.4403,\n",
       " 1321.0449,\n",
       " 1320.9287,\n",
       " 1318.9495,\n",
       " 1311.6697,\n",
       " 1309.6028,\n",
       " 1302.8479,\n",
       " 1300.1674,\n",
       " 1296.874,\n",
       " 1296.7354,\n",
       " 1294.0991,\n",
       " 1290.5173,\n",
       " 1288.8652,\n",
       " 1287.6316,\n",
       " 1287.4701,\n",
       " 1281.884,\n",
       " 1281.022,\n",
       " 1279.3862,\n",
       " 1278.9757,\n",
       " 1276.1699,\n",
       " 1274.2076,\n",
       " 1270.3828,\n",
       " 1270.0415,\n",
       " 1265.3506,\n",
       " 1258.82,\n",
       " 1257.717,\n",
       " 1253.6329,\n",
       " 1251.3843,\n",
       " 1251.0995,\n",
       " 1250.0282,\n",
       " 1248.6414,\n",
       " 1245.4634,\n",
       " 1241.8962,\n",
       " 1241.1239,\n",
       " 1239.3059,\n",
       " 1234.169,\n",
       " 1233.7899,\n",
       " 1231.5303,\n",
       " 1230.698,\n",
       " 1227.1699,\n",
       " 1226.6295,\n",
       " 1225.2931,\n",
       " 1218.2384,\n",
       " 1217.7433,\n",
       " 1217.6711,\n",
       " 1211.0923,\n",
       " 1210.9381,\n",
       " 1210.8464,\n",
       " 1210.0364,\n",
       " 1205.9957,\n",
       " 1205.6212,\n",
       " 1203.9235,\n",
       " 1202.8687,\n",
       " 1191.6522,\n",
       " 1184.575,\n",
       " 1184.1488,\n",
       " 1181.2472,\n",
       " 1175.4486,\n",
       " 1174.0198,\n",
       " 1172.7352,\n",
       " 1172.1118,\n",
       " 1170.4711,\n",
       " 1170.2284,\n",
       " 1167.6708,\n",
       " 1166.517,\n",
       " 1165.8794,\n",
       " 1155.8302,\n",
       " 1154.6438,\n",
       " 1153.9292,\n",
       " 1153.8081,\n",
       " 1153.1477,\n",
       " 1152.7694,\n",
       " 1150.071,\n",
       " 1148.7996,\n",
       " 1144.2443,\n",
       " 1135.2875,\n",
       " 1133.6198,\n",
       " 1131.0701,\n",
       " 1130.8463,\n",
       " 1129.9513,\n",
       " 1124.8054,\n",
       " 1122.7885,\n",
       " 1119.658,\n",
       " 1118.1664,\n",
       " 1118.0905,\n",
       " 1116.0891,\n",
       " 1115.1405,\n",
       " 1114.9282,\n",
       " 1107.61,\n",
       " 1105.5348,\n",
       " 1104.3372,\n",
       " 1103.6382,\n",
       " 1102.7383,\n",
       " 1099.6337,\n",
       " 1098.4093,\n",
       " 1095.632,\n",
       " 1090.8948,\n",
       " 1089.1511,\n",
       " 1089.1212,\n",
       " 1088.9426,\n",
       " 1083.7148,\n",
       " 1083.5466,\n",
       " 1081.3716,\n",
       " 1080.5405,\n",
       " 1078.1995,\n",
       " 1074.0426,\n",
       " 1073.9324,\n",
       " 1072.5017,\n",
       " 1072.2646,\n",
       " 1065.5371,\n",
       " 1063.801,\n",
       " 1062.5417,\n",
       " 1058.3298,\n",
       " 1054.4844,\n",
       " 1052.3193,\n",
       " 1051.7814,\n",
       " 1049.7252,\n",
       " 1049.1086,\n",
       " 1048.0952,\n",
       " 1048.0922,\n",
       " 1046.658,\n",
       " 1046.6533,\n",
       " 1044.6914,\n",
       " 1044.2473,\n",
       " 1043.7323,\n",
       " 1043.3058,\n",
       " 1041.1985,\n",
       " 1040.8118,\n",
       " 1038.8436,\n",
       " 1036.8083,\n",
       " 1031.0182,\n",
       " 1030.3387,\n",
       " 1030.2136,\n",
       " 1028.2576,\n",
       " 1026.1758,\n",
       " 1025.5194,\n",
       " 1022.5399,\n",
       " 1019.57056,\n",
       " 1018.1738,\n",
       " 1017.5022,\n",
       " 1015.63184,\n",
       " 1014.75415,\n",
       " 1010.4082,\n",
       " 1010.152,\n",
       " 1009.6083,\n",
       " 1003.38586,\n",
       " 997.4502,\n",
       " 994.38,\n",
       " 994.2002,\n",
       " 988.0239,\n",
       " 987.02954,\n",
       " 986.24023,\n",
       " 984.73193,\n",
       " 983.48206,\n",
       " 980.8484,\n",
       " 980.63464,\n",
       " 979.2776,\n",
       " 979.1256,\n",
       " 978.44055,\n",
       " 978.10364,\n",
       " 976.5177,\n",
       " 976.45715,\n",
       " 974.9298,\n",
       " 973.6881,\n",
       " 971.1615,\n",
       " 970.56726,\n",
       " 970.4335,\n",
       " 970.27515,\n",
       " 969.8839,\n",
       " 969.79126,\n",
       " 968.8131,\n",
       " 966.6482,\n",
       " 966.3568,\n",
       " 963.7926,\n",
       " 963.38513,\n",
       " 963.19775,\n",
       " 959.1194,\n",
       " 958.8291,\n",
       " 958.1738,\n",
       " 957.5725,\n",
       " 954.4133,\n",
       " 954.2838,\n",
       " 951.0459,\n",
       " 945.67456,\n",
       " 943.8081,\n",
       " 940.7892,\n",
       " 938.8722,\n",
       " 938.2727,\n",
       " 937.3617,\n",
       " 936.63916,\n",
       " 934.35815,\n",
       " 933.1559,\n",
       " 928.67456,\n",
       " 924.5698,\n",
       " 923.6261,\n",
       " 921.98145,\n",
       " 921.96533,\n",
       " 918.8264,\n",
       " 916.7001,\n",
       " 914.916,\n",
       " 912.0117,\n",
       " 909.01465,\n",
       " 907.6251,\n",
       " 907.5635,\n",
       " 907.49097,\n",
       " 905.8613,\n",
       " 905.397,\n",
       " 903.6316,\n",
       " 900.91003,\n",
       " 900.8938,\n",
       " 900.4967,\n",
       " 900.04724,\n",
       " 892.74414,\n",
       " 891.6416,\n",
       " 888.9929,\n",
       " 888.98645,\n",
       " 886.7097,\n",
       " 882.2129,\n",
       " 878.6483,\n",
       " 878.6194,\n",
       " 873.71643,\n",
       " 873.4133,\n",
       " 871.37915,\n",
       " 868.23535,\n",
       " 867.67126,\n",
       " 865.4374,\n",
       " 862.3972,\n",
       " 860.3264,\n",
       " 857.9928,\n",
       " 857.81067,\n",
       " 857.4342,\n",
       " 846.0005,\n",
       " 845.1819,\n",
       " 843.8137,\n",
       " 843.1338,\n",
       " 842.6821,\n",
       " 842.1737,\n",
       " 841.4591,\n",
       " 840.25903,\n",
       " 839.78406,\n",
       " 839.14075,\n",
       " 835.422,\n",
       " 833.8324,\n",
       " 829.91614,\n",
       " 828.2988,\n",
       " 826.2992,\n",
       " 819.4398,\n",
       " 817.1433,\n",
       " 815.6626,\n",
       " 813.17065,\n",
       " 812.2661,\n",
       " 811.9354,\n",
       " 810.39075,\n",
       " 808.81726,\n",
       " 802.2743,\n",
       " 801.3441,\n",
       " 799.6803,\n",
       " 799.4315,\n",
       " 799.35986,\n",
       " 794.19495,\n",
       " 793.3876,\n",
       " 792.6616,\n",
       " 791.99304,\n",
       " 791.6411,\n",
       " 791.32544,\n",
       " 787.0282,\n",
       " 784.6156,\n",
       " 783.5984,\n",
       " 782.27625,\n",
       " 782.18994,\n",
       " 781.9271,\n",
       " 781.43274,\n",
       " 778.0597,\n",
       " 777.9508,\n",
       " 777.6033,\n",
       " 777.05054,\n",
       " 774.8196,\n",
       " 774.7185,\n",
       " 773.04065,\n",
       " 772.5045,\n",
       " 770.7899,\n",
       " 770.5547,\n",
       " 770.43164,\n",
       " 765.58936,\n",
       " 765.5487,\n",
       " 763.39636,\n",
       " 763.2743,\n",
       " 763.1903,\n",
       " 761.5967,\n",
       " 759.9712,\n",
       " 757.56555,\n",
       " 755.76355,\n",
       " 755.7063,\n",
       " 754.1964,\n",
       " 753.5576,\n",
       " 751.31287,\n",
       " 748.396,\n",
       " 744.75903,\n",
       " 744.1644,\n",
       " 742.72864,\n",
       " 739.51306,\n",
       " 737.8944,\n",
       " 737.6376,\n",
       " 736.9938,\n",
       " 735.31116,\n",
       " 730.3843,\n",
       " 730.08167,\n",
       " 729.5432,\n",
       " 728.95593,\n",
       " 727.3773,\n",
       " 726.7732,\n",
       " 726.3849,\n",
       " 722.93823,\n",
       " 720.23645,\n",
       " 719.4729,\n",
       " 717.0386,\n",
       " 716.9343,\n",
       " 716.6621,\n",
       " 715.8983,\n",
       " 714.1494,\n",
       " 713.7262,\n",
       " 712.2512,\n",
       " 711.47217,\n",
       " 710.7146,\n",
       " 709.48486,\n",
       " 709.041,\n",
       " 708.6201,\n",
       " 705.65063,\n",
       " 702.7783,\n",
       " 702.2532,\n",
       " 701.1804,\n",
       " 700.7239,\n",
       " 699.22546,\n",
       " 698.64795,\n",
       " 695.7804,\n",
       " 694.63965,\n",
       " 694.46277,\n",
       " 694.0165,\n",
       " 693.44763,\n",
       " 692.7385,\n",
       " 692.62354,\n",
       " 691.5741,\n",
       " 690.6748,\n",
       " 689.5603,\n",
       " 688.76404,\n",
       " 687.933,\n",
       " 687.60876,\n",
       " 687.13684,\n",
       " 686.34937,\n",
       " 686.26624,\n",
       " 686.0852,\n",
       " 684.63257,\n",
       " 684.4928,\n",
       " 683.38745,\n",
       " 683.2269,\n",
       " 683.20764,\n",
       " 682.7495,\n",
       " 682.0989,\n",
       " 680.2654,\n",
       " 679.93945,\n",
       " 679.8119,\n",
       " 679.1531,\n",
       " 679.1106,\n",
       " 677.42615,\n",
       " 676.3812,\n",
       " 673.9763,\n",
       " 673.2666,\n",
       " 671.7792,\n",
       " 667.74194,\n",
       " 665.0824,\n",
       " 664.0762,\n",
       " 662.75256,\n",
       " 658.984,\n",
       " 658.3624,\n",
       " 657.4253,\n",
       " 656.458,\n",
       " 656.0917,\n",
       " 654.2572,\n",
       " 653.96497,\n",
       " 653.9602,\n",
       " 653.8727,\n",
       " 653.5408,\n",
       " 652.8146,\n",
       " 652.1887,\n",
       " 650.6255,\n",
       " 648.53906,\n",
       " 648.4885,\n",
       " 648.47546,\n",
       " 648.1378,\n",
       " 648.0383,\n",
       " 646.65967,\n",
       " 646.2826,\n",
       " 646.03687,\n",
       " 645.14185,\n",
       " 642.5685,\n",
       " 642.39575,\n",
       " 641.72095,\n",
       " 640.81604,\n",
       " 637.21216,\n",
       " 635.73376,\n",
       " 635.16895,\n",
       " 631.5996,\n",
       " 630.80066,\n",
       " 630.2853,\n",
       " 627.7517,\n",
       " 627.5592,\n",
       " 625.7689,\n",
       " 623.8508,\n",
       " 623.32263,\n",
       " 623.1472,\n",
       " 621.6372,\n",
       " 619.31885,\n",
       " 618.0785,\n",
       " 615.4812,\n",
       " 614.31067,\n",
       " 613.3816,\n",
       " 612.4663,\n",
       " 612.3229,\n",
       " 610.90076,\n",
       " 610.0753,\n",
       " 608.0624,\n",
       " 603.4142,\n",
       " 603.13513,\n",
       " 601.7355,\n",
       " 600.5969,\n",
       " 599.2189,\n",
       " 598.7318,\n",
       " 598.4326,\n",
       " 596.08484,\n",
       " 595.56213,\n",
       " 591.7235,\n",
       " 591.06274,\n",
       " 589.646,\n",
       " 589.32837,\n",
       " 589.08044,\n",
       " 588.5332,\n",
       " 587.86194,\n",
       " 584.71375,\n",
       " 583.97595,\n",
       " 581.95105,\n",
       " 581.60376,\n",
       " 581.2826,\n",
       " 577.5139,\n",
       " 575.5558,\n",
       " 574.7926,\n",
       " 574.1311,\n",
       " 573.44556,\n",
       " 572.34314,\n",
       " 569.87317,\n",
       " 569.5261,\n",
       " 568.6161,\n",
       " 566.95593,\n",
       " 566.7949,\n",
       " 565.948,\n",
       " 565.59436,\n",
       " 564.5049,\n",
       " 563.5563,\n",
       " 562.8616,\n",
       " 562.6483,\n",
       " 562.23706,\n",
       " 560.4159,\n",
       " 559.86206,\n",
       " 559.4098,\n",
       " 558.7633,\n",
       " 558.2865,\n",
       " 557.7363,\n",
       " 557.71106,\n",
       " 556.2733,\n",
       " 555.7036,\n",
       " 555.5093,\n",
       " 555.04565,\n",
       " 554.88464,\n",
       " 553.10315,\n",
       " 551.3375,\n",
       " 551.25366,\n",
       " 550.7135,\n",
       " 550.30164,\n",
       " 550.26404,\n",
       " 550.1747,\n",
       " 548.8999,\n",
       " 548.1113,\n",
       " 546.9938,\n",
       " 544.6559,\n",
       " 540.6465,\n",
       " 540.1765,\n",
       " 540.073,\n",
       " 539.96875,\n",
       " 538.95435,\n",
       " 538.14514,\n",
       " 537.9354,\n",
       " 533.25916,\n",
       " 530.5409,\n",
       " 529.3231,\n",
       " 526.3428,\n",
       " 525.3042,\n",
       " 525.15405,\n",
       " 524.4342,\n",
       " 524.2446,\n",
       " 522.13025,\n",
       " 521.9098,\n",
       " 521.0045,\n",
       " 520.4662,\n",
       " 519.87195,\n",
       " 519.2246,\n",
       " 519.01855,\n",
       " 517.7189,\n",
       " 517.02026,\n",
       " 515.09265,\n",
       " 514.9236,\n",
       " 513.89343,\n",
       " 513.4707,\n",
       " 511.5686,\n",
       " 510.8064,\n",
       " 509.4254,\n",
       " 507.86646,\n",
       " 507.70605,\n",
       " 507.11804,\n",
       " 506.5431,\n",
       " 505.91467,\n",
       " 505.06555,\n",
       " 504.1991,\n",
       " 501.93506,\n",
       " 498.88147,\n",
       " 496.53088,\n",
       " 495.11487,\n",
       " 494.34033,\n",
       " 494.1726,\n",
       " 492.1853,\n",
       " 491.9485,\n",
       " 491.7262,\n",
       " 490.96814,\n",
       " 489.7749,\n",
       " 486.81287,\n",
       " 485.61536,\n",
       " 485.099,\n",
       " 482.97546,\n",
       " 482.04175,\n",
       " 481.94226,\n",
       " 481.40442,\n",
       " 480.8357,\n",
       " 480.79602,\n",
       " 479.09937,\n",
       " 478.80933,\n",
       " 478.48303,\n",
       " 478.38318,\n",
       " 477.73633,\n",
       " 477.71228,\n",
       " 477.60522,\n",
       " 477.2782,\n",
       " 474.65857,\n",
       " 474.6371,\n",
       " 474.04565,\n",
       " 473.6681,\n",
       " 473.25647,\n",
       " 472.99268,\n",
       " 472.75806,\n",
       " 472.5796,\n",
       " 472.0204,\n",
       " 471.5293,\n",
       " 470.45996,\n",
       " 470.13538,\n",
       " 469.19397,\n",
       " 468.94934,\n",
       " 467.78516,\n",
       " 467.33997,\n",
       " 466.0327,\n",
       " 465.30383,\n",
       " 464.93225,\n",
       " 463.8734,\n",
       " 460.0808,\n",
       " 459.31543,\n",
       " 457.22644,\n",
       " 457.01392,\n",
       " 456.22205,\n",
       " 454.33362,\n",
       " 454.25403,\n",
       " 454.10767,\n",
       " 453.6958,\n",
       " 452.57166,\n",
       " 449.66162,\n",
       " 448.8379,\n",
       " 448.3828,\n",
       " 447.54163,\n",
       " 447.0149,\n",
       " 446.58142,\n",
       " 446.25977,\n",
       " 445.72998,\n",
       " 443.0995,\n",
       " 439.76404,\n",
       " 438.58826,\n",
       " 438.5769,\n",
       " 437.03174,\n",
       " 435.46924,\n",
       " 434.56592,\n",
       " 433.99023,\n",
       " 432.6433,\n",
       " 432.54077,\n",
       " 428.06213,\n",
       " 427.67358,\n",
       " 427.13196,\n",
       " 425.7201,\n",
       " 425.14978,\n",
       " 425.12878,\n",
       " 421.5818,\n",
       " 421.56006,\n",
       " 420.3003,\n",
       " 419.83557,\n",
       " 416.7318,\n",
       " 416.58435,\n",
       " 415.95312,\n",
       " 414.86426,\n",
       " 412.76526,\n",
       " 412.64,\n",
       " 410.9087,\n",
       " 410.55054,\n",
       " 409.7743,\n",
       " 407.6128,\n",
       " 406.75745,\n",
       " 406.74585,\n",
       " 405.0327,\n",
       " 403.3551,\n",
       " 402.48315,\n",
       " 400.2633,\n",
       " 399.84863,\n",
       " 397.89954,\n",
       " 394.60864,\n",
       " 392.89185,\n",
       " 392.41113,\n",
       " 392.3109,\n",
       " 392.07104,\n",
       " 389.07495,\n",
       " 388.35864,\n",
       " 387.00684,\n",
       " 385.9939,\n",
       " 385.1714,\n",
       " 384.56128,\n",
       " 383.98572,\n",
       " 383.91724,\n",
       " 382.68298,\n",
       " 382.53052,\n",
       " 382.406,\n",
       " 381.53906,\n",
       " 381.09302,\n",
       " 380.71143,\n",
       " 379.615,\n",
       " 376.6366,\n",
       " 375.92566,\n",
       " 373.2926,\n",
       " 373.27844,\n",
       " 371.25806,\n",
       " 370.5487,\n",
       " 369.90027,\n",
       " 369.72803,\n",
       " 368.6975,\n",
       " 367.6018,\n",
       " 365.85632,\n",
       " 365.7661,\n",
       " 362.56873,\n",
       " 360.64136,\n",
       " 359.68872,\n",
       " 359.5404,\n",
       " 358.33325,\n",
       " 356.1422,\n",
       " 354.96814,\n",
       " 354.42297,\n",
       " 353.48132,\n",
       " 352.83557,\n",
       " 352.06653,\n",
       " 350.50586,\n",
       " 350.42468,\n",
       " 348.88245,\n",
       " 348.05493,\n",
       " 347.9204,\n",
       " 347.88354,\n",
       " 347.60535,\n",
       " 346.10156,\n",
       " 345.8523,\n",
       " 344.4359,\n",
       " 344.1842,\n",
       " 344.16272,\n",
       " 344.01282,\n",
       " 343.59143,\n",
       " 342.77124,\n",
       " 342.45605,\n",
       " 341.906,\n",
       " 341.4043,\n",
       " 341.36462,\n",
       " 339.14197,\n",
       " 338.69714,\n",
       " 337.4574,\n",
       " 337.08423,\n",
       " 336.91528,\n",
       " 336.56458,\n",
       " 335.85986,\n",
       " 335.43152,\n",
       " 335.37378,\n",
       " 335.29016,\n",
       " 335.00513,\n",
       " 334.8269,\n",
       " 333.7915,\n",
       " 333.05408,\n",
       " 332.83496,\n",
       " 332.1455,\n",
       " 330.85974,\n",
       " 330.511,\n",
       " 330.38086,\n",
       " 329.0714,\n",
       " 328.78003,\n",
       " 328.30896,\n",
       " 328.15186,\n",
       " 327.828,\n",
       " 327.5979,\n",
       " 327.33423,\n",
       " 326.9364,\n",
       " 326.2605,\n",
       " 325.2495,\n",
       " 325.02515,\n",
       " 324.2024,\n",
       " 322.58435,\n",
       " 321.7118,\n",
       " 321.3352,\n",
       " 319.06165,\n",
       " 317.64685,\n",
       " 317.08765,\n",
       " 316.30902,\n",
       " 315.97333,\n",
       " 315.81213,\n",
       " 315.69775,\n",
       " 313.80347,\n",
       " 313.58484,\n",
       " 313.28076,\n",
       " 311.76917,\n",
       " 309.979,\n",
       " 309.17218,\n",
       " 308.80945,\n",
       " 308.16736,\n",
       " 308.01764,\n",
       " 306.19965,\n",
       " 306.19403,\n",
       " 304.28058,\n",
       " 302.72925,\n",
       " 302.0005,\n",
       " 301.77252,\n",
       " 300.44763,\n",
       " 300.1792,\n",
       " 299.4417,\n",
       " 298.52124,\n",
       " 296.39307,\n",
       " 296.23798,\n",
       " 296.18164,\n",
       " 296.0865,\n",
       " 295.99585,\n",
       " 295.8177,\n",
       " 294.30017,\n",
       " 294.12177,\n",
       " 294.0819,\n",
       " 292.4134,\n",
       " 291.16833,\n",
       " 290.9162,\n",
       " 290.87225,\n",
       " 290.18152,\n",
       " 288.6944,\n",
       " 288.1938,\n",
       " 288.07074,\n",
       " 287.96777,\n",
       " 287.85815,\n",
       " 287.3664,\n",
       " 286.9831,\n",
       " 286.89746,\n",
       " 286.01727,\n",
       " 285.7337,\n",
       " 285.61792,\n",
       " 285.4837,\n",
       " 285.3824,\n",
       " 285.15393,\n",
       " 284.34216,\n",
       " 283.88092,\n",
       " 282.74548,\n",
       " 282.4707,\n",
       " 282.372,\n",
       " 281.91382,\n",
       " 281.29742,\n",
       " 280.46655,\n",
       " 279.85492,\n",
       " 279.49304,\n",
       " 278.26117,\n",
       " 277.61047,\n",
       " 276.98572,\n",
       " 276.62183,\n",
       " 276.0066,\n",
       " 275.74707,\n",
       " 274.3576,\n",
       " 274.1809,\n",
       " 274.01062,\n",
       " 271.5113,\n",
       " 271.1494,\n",
       " 270.8523,\n",
       " 268.24683,\n",
       " 267.78088,\n",
       " 267.68256,\n",
       " 267.64142,\n",
       " 265.69427,\n",
       " 262.614,\n",
       " 262.16418,\n",
       " 262.0078,\n",
       " 261.95074,\n",
       " 261.0147,\n",
       " 260.21124,\n",
       " 259.22906,\n",
       " 259.16864,\n",
       " 258.16553,\n",
       " 257.21112,\n",
       " 256.11285,\n",
       " 255.578,\n",
       " 255.12506,\n",
       " 254.97644,\n",
       " 254.54468,\n",
       " 253.79132,\n",
       " 253.34753,\n",
       " 253.19922,\n",
       " 252.89435,\n",
       " 251.67499,\n",
       " 250.41339,\n",
       " 249.07642,\n",
       " 247.24133,\n",
       " 247.00641,\n",
       " 246.79785,\n",
       " 245.62463,\n",
       " 244.67957,\n",
       " 244.46143,\n",
       " 244.25385,\n",
       " ...]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "calibrated_gradient"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5b4759d8",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "otdd",
   "language": "python",
   "name": "otdd"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}