ML-methods-tutorial-02.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "858bac0c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The autoreload extension is already loaded. To reload it, use:\n",
      "  %reload_ext autoreload\n"
     ]
    }
   ],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "import os\n",
    "import numpy as n\n",
    "from matplotlib import pyplot as plt\n",
    "plt.rcParams['figure.figsize'] = (10,6)\n",
    "import pickle\n",
    "import numpy.lib.recfunctions as rfn\n",
    "import copy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "21c85ee8",
   "metadata": {},
   "outputs": [],
   "source": [
    "from modules import utils"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1991d924",
   "metadata": {},
   "source": [
    "### Load the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "177bbe00",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_file = 'data/all_data.p'\n",
    "\n",
    "f = open(data_file, 'rb')\n",
    "all_data = pickle.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "86059b2e",
   "metadata": {},
   "outputs": [],
   "source": [
    "experiment = all_data['Female1']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "15fab73f",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save timestamps into variable\n",
    "ts = experiment['t']\n",
    "# Save annotations into variable\n",
    "annotations = experiment['ann']"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "51e7dbd0",
   "metadata": {},
   "source": [
    "## Calculate some features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "4a4be90e",
   "metadata": {},
   "outputs": [],
   "source": [
    "vectors = {'rear_1' : ('ImplantedTailBase_1', 'BackCenter_1'),\n",
    "           'rear_2' : ('InteracteeTailBase_2', 'BackCenter_2'),\n",
    "           'front_1': ('BackCenter_1', 'NapeCenter_1'),\n",
    "           'front_2': ('BackCenter_2', 'NapeCenter_2'),\n",
    "           'head_1' : ('NapeCenter_1', ('GreenTape_1', 'RedTape_1')),\n",
    "           'head_2' : ('NapeCenter_2', ('YellowEar_2', 'OrangeEar_2')),\n",
    "           'front_1_to_front_2' : ('NapeCenter_1', 'NapeCenter_2'),\n",
    "           'front_2_to_front_1' : ('NapeCenter_2', 'NapeCenter_1'),\n",
    "           'front_1_to_rear_2' : ('NapeCenter_1', 'InteracteeTailBase_2'),\n",
    "           'front_2_to_rear_1' : ('NapeCenter_2', 'ImplantedTailBase_1'),\n",
    "           'rear_1_to_rear_2' : ('BackCenter_1','BackCenter_2'),\n",
    "           'rear_2_to_rear_1' : ('BackCenter_2','BackCenter_1')}\n",
    "velocities_to_calculate = [\n",
    "    'BackCenter_1',\n",
    "    'BackCenter_2',\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "32a26c86",
   "metadata": {},
   "outputs": [],
   "source": [
    "exp = utils.compute_and_add_vectors(experiment,vectors)\n",
    "exp = utils.compute_and_add_velocities(exp, velocities_to_calculate)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6b2cfe7e",
   "metadata": {},
   "source": [
    "## Format your dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "f4f633d2",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn import preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "3a3e2eb5",
   "metadata": {},
   "outputs": [],
   "source": [
    "labels = exp['ann']\n",
    "unique_labels = n.unique(labels)\n",
    "ts = exp['t']\n",
    "features_to_use = [\n",
    "    'rear_1_ang',\n",
    "    'front_1_ang',\n",
    "    'head_1_ang',\n",
    "    'rear_1_ang',\n",
    "    'front_1_ang',\n",
    "    'head_1_ang',\n",
    "    'BackCenter_1_vel_mag',\n",
    "    'BackCenter_1_vel_ang',\n",
    "    'rear_2_ang',\n",
    "    'front_2_ang',\n",
    "    'head_2_ang',\n",
    "    'rear_2_ang',\n",
    "    'front_2_ang',\n",
    "    'head_2_ang',\n",
    "    'BackCenter_2_vel_mag',\n",
    "    'BackCenter_2_vel_ang',\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "cf9be276",
   "metadata": {},
   "outputs": [],
   "source": [
    "n_timepoints = len(ts)\n",
    "n_features = len(features_to_use)\n",
    "n_labels = len(unique_labels)\n",
    "Xs = n.zeros((n_timepoints, n_features))\n",
    "Ys = n.zeros(n_timepoints)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "13c28cb5",
   "metadata": {},
   "outputs": [],
   "source": [
    "label_encoder = preprocessing.LabelEncoder()\n",
    "label_encoder.fit(unique_labels)\n",
    "Ys = label_encoder.transform(labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "64b7149d",
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(n_features):\n",
    "    Xs[:,i] = exp[features_to_use[i]]\n",
    "Xs[n.isnan(Xs)] = 0"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5f582bd5",
   "metadata": {},
   "source": [
    "## Split into train and test (using cross-validation)\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "e589e1de",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import StratifiedKFold, KFold"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "cb9f7dfe",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "n_splits = 2\n",
    "splitter = StratifiedKFold(n_splits=n_splits, shuffle=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "56955e0e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x72 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "# splitter = KFold(n_splits=5, shuffle=False)\n",
    "\n",
    "\n",
    "i = 0\n",
    "plt.figure(figsize=(10,1))\n",
    "for train, test in splitter.split(Xs, Ys):\n",
    "    plt.scatter(train, [i]*len(train),s=1, color='navy')\n",
    "    plt.scatter(test, [i]*len(test),s=1, color='salmon')\n",
    "    i += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d4fd8eb2",
   "metadata": {},
   "source": [
    "## Build Architecture\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "3a62246e",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "from torch import nn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "bbda15e3",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set the architecture parameters\n",
    "# N is batch size; D_in is input dimension;\n",
    "# H is hidden dimension; D_out is output dimension.\n",
    "D_in = Xs.shape[1] \n",
    "D_out = n_labels\n",
    "hidden_dims = [300,300,300]\n",
    "\n",
    "\n",
    "# Set the learning parameter\n",
    "N_batch = 16\n",
    "n_epochs = 3\n",
    "learning_rate = 1e-4\n",
    "print_interval = 50\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "a0353b7c",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "### EPOCH  0 ###\n",
      "Batch   0, Train Loss: 7.053131, Test Loss: 7.533041, Test Correct Frac: 0.007\n",
      "Batch  50, Train Loss: 0.834274, Test Loss: 1.140828, Test Correct Frac: 0.671\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_7325/1480301380.py:64: DeprecationWarning: `np.int` is a deprecated alias for the builtin `int`. To silence this warning, use `int` by itself. Doing this will not modify any behavior and is safe. When replacing `np.int`, you may wish to use e.g. `np.int64` or `np.int32` to specify the precision. If you wish to review your current use, check the release note link for additional information.\n",
      "Deprecated in NumPy 1.20; for more details and guidance: https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations\n",
      "  correct_preds = n.array(pred_labels == true_labels, n.int)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Batch 100, Train Loss: 0.989083, Test Loss: 1.153818, Test Correct Frac: 0.636\n",
      "Batch 150, Train Loss: 0.267104, Test Loss: 1.009258, Test Correct Frac: 0.669\n",
      "Batch 200, Train Loss: 0.568707, Test Loss: 0.905184, Test Correct Frac: 0.660\n",
      "Batch 250, Train Loss: 0.805399, Test Loss: 0.963437, Test Correct Frac: 0.692\n",
      "Batch 300, Train Loss: 0.756275, Test Loss: 1.051639, Test Correct Frac: 0.658\n",
      "Batch 350, Train Loss: 0.630345, Test Loss: 0.928667, Test Correct Frac: 0.679\n",
      "Batch 400, Train Loss: 1.447714, Test Loss: 1.043342, Test Correct Frac: 0.668\n",
      "Batch 450, Train Loss: 0.497316, Test Loss: 0.966567, Test Correct Frac: 0.680\n",
      "Batch 500, Train Loss: 0.353750, Test Loss: 0.918962, Test Correct Frac: 0.680\n",
      "Batch 550, Train Loss: 0.682372, Test Loss: 0.913197, Test Correct Frac: 0.684\n",
      "Batch 600, Train Loss: 0.344556, Test Loss: 0.989302, Test Correct Frac: 0.677\n",
      "Batch 650, Train Loss: 0.155167, Test Loss: 0.997071, Test Correct Frac: 0.666\n",
      "Batch 700, Train Loss: 0.647025, Test Loss: 0.927700, Test Correct Frac: 0.663\n",
      "Batch 750, Train Loss: 0.766181, Test Loss: 1.005191, Test Correct Frac: 0.691\n",
      "Batch 800, Train Loss: 0.291289, Test Loss: 1.010886, Test Correct Frac: 0.675\n",
      "Batch 850, Train Loss: 0.662434, Test Loss: 1.019074, Test Correct Frac: 0.652\n",
      "Batch 900, Train Loss: 0.679949, Test Loss: 0.988218, Test Correct Frac: 0.670\n",
      "### EPOCH  1 ###\n",
      "Batch   0, Train Loss: 0.674798, Test Loss: 0.997313, Test Correct Frac: 0.676\n",
      "Batch  50, Train Loss: 0.504107, Test Loss: 1.000436, Test Correct Frac: 0.685\n",
      "Batch 100, Train Loss: 1.004174, Test Loss: 1.010003, Test Correct Frac: 0.640\n",
      "Batch 150, Train Loss: 0.254108, Test Loss: 1.004783, Test Correct Frac: 0.692\n",
      "Batch 200, Train Loss: 0.706711, Test Loss: 0.971843, Test Correct Frac: 0.663\n",
      "Batch 250, Train Loss: 0.165733, Test Loss: 1.011380, Test Correct Frac: 0.664\n",
      "Batch 300, Train Loss: 0.650555, Test Loss: 1.073341, Test Correct Frac: 0.621\n",
      "Batch 350, Train Loss: 0.775193, Test Loss: 1.100643, Test Correct Frac: 0.632\n",
      "Batch 400, Train Loss: 0.303471, Test Loss: 1.012295, Test Correct Frac: 0.650\n",
      "Batch 450, Train Loss: 0.463418, Test Loss: 1.121654, Test Correct Frac: 0.622\n",
      "Batch 500, Train Loss: 0.684342, Test Loss: 1.065341, Test Correct Frac: 0.699\n",
      "Batch 550, Train Loss: 0.315848, Test Loss: 0.978730, Test Correct Frac: 0.679\n",
      "Batch 600, Train Loss: 0.126164, Test Loss: 0.947678, Test Correct Frac: 0.674\n",
      "Batch 650, Train Loss: 0.178064, Test Loss: 0.996721, Test Correct Frac: 0.667\n",
      "Batch 700, Train Loss: 0.635803, Test Loss: 0.947761, Test Correct Frac: 0.673\n",
      "Batch 750, Train Loss: 0.289336, Test Loss: 0.952254, Test Correct Frac: 0.665\n",
      "Batch 800, Train Loss: 0.595669, Test Loss: 0.894067, Test Correct Frac: 0.703\n",
      "Batch 850, Train Loss: 0.535652, Test Loss: 0.927285, Test Correct Frac: 0.681\n",
      "Batch 900, Train Loss: 0.415819, Test Loss: 0.941468, Test Correct Frac: 0.649\n",
      "### EPOCH  2 ###\n",
      "Batch   0, Train Loss: 1.047182, Test Loss: 0.941182, Test Correct Frac: 0.681\n",
      "Batch  50, Train Loss: 0.244706, Test Loss: 1.008008, Test Correct Frac: 0.668\n",
      "Batch 100, Train Loss: 0.508520, Test Loss: 0.978413, Test Correct Frac: 0.668\n",
      "Batch 150, Train Loss: 0.342068, Test Loss: 0.952412, Test Correct Frac: 0.671\n",
      "Batch 200, Train Loss: 0.410630, Test Loss: 0.918745, Test Correct Frac: 0.686\n",
      "Batch 250, Train Loss: 0.343465, Test Loss: 1.003853, Test Correct Frac: 0.667\n",
      "Batch 300, Train Loss: 0.281805, Test Loss: 0.926468, Test Correct Frac: 0.676\n",
      "Batch 350, Train Loss: 0.283757, Test Loss: 0.950644, Test Correct Frac: 0.686\n",
      "Batch 400, Train Loss: 0.321368, Test Loss: 0.967549, Test Correct Frac: 0.669\n",
      "Batch 450, Train Loss: 0.323049, Test Loss: 0.944875, Test Correct Frac: 0.690\n",
      "Batch 500, Train Loss: 0.180179, Test Loss: 0.983044, Test Correct Frac: 0.696\n",
      "Batch 550, Train Loss: 0.468326, Test Loss: 1.047196, Test Correct Frac: 0.692\n",
      "Batch 600, Train Loss: 0.182801, Test Loss: 1.019563, Test Correct Frac: 0.658\n",
      "Batch 650, Train Loss: 0.350483, Test Loss: 0.925305, Test Correct Frac: 0.678\n",
      "Batch 700, Train Loss: 0.462745, Test Loss: 0.972522, Test Correct Frac: 0.685\n",
      "Batch 750, Train Loss: 0.657442, Test Loss: 0.952815, Test Correct Frac: 0.686\n",
      "Batch 800, Train Loss: 0.855147, Test Loss: 0.969401, Test Correct Frac: 0.679\n",
      "Batch 850, Train Loss: 0.731871, Test Loss: 0.943347, Test Correct Frac: 0.688\n",
      "Batch 900, Train Loss: 0.708938, Test Loss: 1.028806, Test Correct Frac: 0.645\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "### EPOCH  0 ###\n",
      "Batch   0, Train Loss: 4.908206, Test Loss: 8.005853, Test Correct Frac: 0.003\n",
      "Batch  50, Train Loss: 0.657169, Test Loss: 1.167734, Test Correct Frac: 0.686\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_7325/1480301380.py:64: DeprecationWarning: `np.int` is a deprecated alias for the builtin `int`. To silence this warning, use `int` by itself. Doing this will not modify any behavior and is safe. When replacing `np.int`, you may wish to use e.g. `np.int64` or `np.int32` to specify the precision. If you wish to review your current use, check the release note link for additional information.\n",
      "Deprecated in NumPy 1.20; for more details and guidance: https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations\n",
      "  correct_preds = n.array(pred_labels == true_labels, n.int)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Batch 100, Train Loss: 0.815621, Test Loss: 1.088732, Test Correct Frac: 0.663\n",
      "Batch 150, Train Loss: 0.846230, Test Loss: 0.885262, Test Correct Frac: 0.729\n",
      "Batch 200, Train Loss: 0.568076, Test Loss: 0.910429, Test Correct Frac: 0.697\n",
      "Batch 250, Train Loss: 0.288455, Test Loss: 1.063106, Test Correct Frac: 0.714\n",
      "Batch 300, Train Loss: 0.852365, Test Loss: 1.123410, Test Correct Frac: 0.695\n",
      "Batch 350, Train Loss: 0.390555, Test Loss: 0.920546, Test Correct Frac: 0.717\n",
      "Batch 400, Train Loss: 1.119537, Test Loss: 0.879542, Test Correct Frac: 0.702\n",
      "Batch 450, Train Loss: 0.709614, Test Loss: 0.919154, Test Correct Frac: 0.665\n",
      "Batch 500, Train Loss: 0.975222, Test Loss: 1.015522, Test Correct Frac: 0.678\n",
      "Batch 550, Train Loss: 0.381768, Test Loss: 0.883237, Test Correct Frac: 0.698\n",
      "Batch 600, Train Loss: 0.431939, Test Loss: 0.891518, Test Correct Frac: 0.702\n",
      "Batch 650, Train Loss: 0.372947, Test Loss: 0.889185, Test Correct Frac: 0.685\n",
      "Batch 700, Train Loss: 0.240341, Test Loss: 0.980172, Test Correct Frac: 0.718\n",
      "Batch 750, Train Loss: 0.684692, Test Loss: 1.149584, Test Correct Frac: 0.712\n",
      "Batch 800, Train Loss: 0.341448, Test Loss: 0.994521, Test Correct Frac: 0.664\n",
      "Batch 850, Train Loss: 0.638970, Test Loss: 0.975689, Test Correct Frac: 0.680\n",
      "Batch 900, Train Loss: 0.915990, Test Loss: 0.975103, Test Correct Frac: 0.668\n",
      "### EPOCH  1 ###\n",
      "Batch   0, Train Loss: 0.824960, Test Loss: 0.901120, Test Correct Frac: 0.714\n",
      "Batch  50, Train Loss: 0.527783, Test Loss: 0.926288, Test Correct Frac: 0.695\n",
      "Batch 100, Train Loss: 0.766010, Test Loss: 0.892149, Test Correct Frac: 0.728\n",
      "Batch 150, Train Loss: 0.585051, Test Loss: 0.883555, Test Correct Frac: 0.725\n",
      "Batch 200, Train Loss: 0.335090, Test Loss: 0.885937, Test Correct Frac: 0.697\n",
      "Batch 250, Train Loss: 0.362156, Test Loss: 0.935049, Test Correct Frac: 0.697\n",
      "Batch 300, Train Loss: 0.560741, Test Loss: 0.894463, Test Correct Frac: 0.724\n",
      "Batch 350, Train Loss: 0.611956, Test Loss: 0.928632, Test Correct Frac: 0.705\n",
      "Batch 400, Train Loss: 0.271582, Test Loss: 0.929906, Test Correct Frac: 0.703\n",
      "Batch 450, Train Loss: 0.573084, Test Loss: 0.945474, Test Correct Frac: 0.702\n",
      "Batch 500, Train Loss: 0.467061, Test Loss: 0.934035, Test Correct Frac: 0.665\n",
      "Batch 550, Train Loss: 0.815096, Test Loss: 0.898881, Test Correct Frac: 0.684\n",
      "Batch 600, Train Loss: 0.343586, Test Loss: 0.934031, Test Correct Frac: 0.685\n",
      "Batch 650, Train Loss: 0.540266, Test Loss: 0.902927, Test Correct Frac: 0.694\n",
      "Batch 700, Train Loss: 0.705661, Test Loss: 0.937668, Test Correct Frac: 0.695\n",
      "Batch 750, Train Loss: 0.670125, Test Loss: 0.910629, Test Correct Frac: 0.723\n",
      "Batch 800, Train Loss: 0.659415, Test Loss: 0.906071, Test Correct Frac: 0.683\n",
      "Batch 850, Train Loss: 0.071305, Test Loss: 0.964598, Test Correct Frac: 0.682\n",
      "Batch 900, Train Loss: 0.539454, Test Loss: 0.999761, Test Correct Frac: 0.676\n",
      "### EPOCH  2 ###\n",
      "Batch   0, Train Loss: 0.361855, Test Loss: 0.997446, Test Correct Frac: 0.705\n",
      "Batch  50, Train Loss: 0.277077, Test Loss: 0.850299, Test Correct Frac: 0.719\n",
      "Batch 100, Train Loss: 0.421805, Test Loss: 0.935070, Test Correct Frac: 0.717\n",
      "Batch 150, Train Loss: 0.767353, Test Loss: 0.955387, Test Correct Frac: 0.716\n",
      "Batch 200, Train Loss: 0.540108, Test Loss: 0.910450, Test Correct Frac: 0.715\n",
      "Batch 250, Train Loss: 0.287330, Test Loss: 0.954853, Test Correct Frac: 0.672\n",
      "Batch 300, Train Loss: 0.472837, Test Loss: 0.936267, Test Correct Frac: 0.697\n",
      "Batch 350, Train Loss: 0.300252, Test Loss: 0.936973, Test Correct Frac: 0.683\n",
      "Batch 400, Train Loss: 0.123264, Test Loss: 0.841394, Test Correct Frac: 0.732\n",
      "Batch 450, Train Loss: 0.277395, Test Loss: 0.942254, Test Correct Frac: 0.690\n",
      "Batch 500, Train Loss: 0.299425, Test Loss: 0.856974, Test Correct Frac: 0.743\n",
      "Batch 550, Train Loss: 0.385955, Test Loss: 0.881766, Test Correct Frac: 0.739\n",
      "Batch 600, Train Loss: 0.350040, Test Loss: 0.957414, Test Correct Frac: 0.725\n",
      "Batch 650, Train Loss: 0.469661, Test Loss: 0.983461, Test Correct Frac: 0.715\n",
      "Batch 700, Train Loss: 0.394673, Test Loss: 0.906141, Test Correct Frac: 0.684\n",
      "Batch 750, Train Loss: 0.263567, Test Loss: 0.884116, Test Correct Frac: 0.692\n",
      "Batch 800, Train Loss: 0.203600, Test Loss: 0.878844, Test Correct Frac: 0.728\n",
      "Batch 850, Train Loss: 0.464651, Test Loss: 0.951232, Test Correct Frac: 0.688\n",
      "Batch 900, Train Loss: 0.287498, Test Loss: 0.987250, Test Correct Frac: 0.663\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAkkAAAF1CAYAAAD80H5/AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8/fFQqAAAACXBIWXMAAAsTAAALEwEAmpwYAABXcElEQVR4nO3dd3hUVf7H8fdJZpJJSGgh9Bak917EhhVEUVfF3rtr27X/XFfdXXdd+7p214qKBXvHAqJIC733AAFCQiCk1zm/P+4kBEiZJDMJEz+v55lnJpM7M2dubuZ+5nvOPddYaxERERGRA4U1dANEREREDkcKSSIiIiIVUEgSERERqYBCkoiIiEgFFJJEREREKqCQJCIiIlIBhSQR+V0wxlxujPm1odsBYIx50Bjzdj2/5jfGmMsCvaxIY6aQJAIYYy40xiQaY7KNMTt9O4mjGrA9ScaYPF97Si/P+vnYmcaYq4PdRn80VDAxxnQ1xlhjjKu+XzsYfO+le12ew1o7wVr7ZqCXFWnMGsUHiEhdGGP+DNwDXA98BxQC44EzgEN28MYYl7W2uB6adrq19odAP2k9tj9kGWMMYKy13oZuiz/0NxUJDlWS5HfNGNMM+BvwR2vtx9baHGttkbX2C2vtnb5lHjTGTDPGvG2MyQQuN8a0N8Z8bozZY4zZYIy5ptxzjvRVpTKNMbuMMU/67vf4niPdGJNhjFlgjGlTizZfboz51RjzuDFmrzFmszFmgu93DwNHA8+Wrz75KhF/NMasB9b77rvG1/Y9vvfSvtxrWGPMLcaYTcaY3caYx4wxYcaYSN/yA8ot29pX9Yqv4fs40rcO9vmujzzoPW4yxmT53t9Fvvu7G2N+9j1mtzHm/UqefpbvOsO3HsaUe+5D1pvv/pnGmIeNMbOBXKBbNW1MMsacWO7nA7rQjDGXGmO2+P7e9x+8PBBhjHnL9x5XGmOGV7KeSt/LUt97Oc8Yc5wxJtkYc7cxJgV43RjTwhjzpTEmzff+vjTGdDzo/V1dbv1WuA3VYtkEY8ws3/v4wRjznKnnrkSRYFFIkt+7MYAH+KSa5c4ApgHNgXeAqUAy0B44B/inMeYE37L/Af5jrW0KHAF84Lv/MqAZ0AmIw6lc5dWy3aOAtUAr4FHgVWOMsdbeB/wC3GStjbHW3lTuMWf6HtfXGHM88C9gMtAO2AK8d9BrnAUMB4b63v+V1toC33IXl1vuAuAHa22av403xrQEvgKewVkXTwJfGWPijDFNfPdPsNbGAkcCS3wP/TswHWgBdAT+W8lLHOO7bu5bD3N8P1e43so97hLgWiAWyKqsjX68v77A88BFOOu3GdDhoMUm4azL5sDnQIXdqdba0vcyyPdeSoNhW6Al0MXX5jDgdd/PnXG2raq6aKtbF/4u+y4wH2cdPYizDkUaBYUk+b2LA3b70VUxx1r7qa/7pRVwFHC3tTbfWrsE+B/7dw5FQHdjTCtrbba1dm65++OA7tbaEmvtQmttZhWv+alxKk6ll2vK/W6LtfYVa20J8CbOjri6qtS/rLV7rLV5ODvv16y1i3zB515gjDGma7nl/+1bfivwNE4Ywvd6FxpjSj8/LgGmVPPaB5sIrLfWTrHWFltrpwJrgNN9v/cC/Y0xUdbandbalb77i3BCQHvfuq/peKfq1tsb1tqVvu3h5GraWJVzgC+stb9aawuBvwIHnyjzV2vt1762TAEG1fC9eIEHrLUF1to8a226tfYja22utTYLeBg4torH12QbqnBZY0xnYATwV2ttoe/v8XkN34fIYUshSX7v0oFWpvoBvtvK3W4P7PHtiEptYX+l4CqgJ7DG10Vzmu/+KThjnt4zxuwwxjxqjHFX8ZpnWmubl7u8Uu53KaU3rLW5vpsxNXwPW8o9RzbOuuhQyfJbfI/BWjsPyAGONcb0BrpT8x3jAa9f7jU6WGtzgPNwKm07jTFf+V4H4C7AAPN9XVRX1vB1q1tvla6j8m3043Xal38u32ulV9YWnO49jx/bYXlp1tr80h+MMdHGmJd8XXyZOF2OzY0x4ZU8vibbUGXLlv4v5JZbtvw6FAlpCknyezcHyMfpiqpK+SrADqClMSa23H2dge0A1tr11toLgNbAv4FpxpgmvrFOD1lr++J0IZ0GXBqYt1FpWyu7fwdORQYAXxdXXOl78OlU7nZn32NKvYnT5XYJMK38ztpPB7x+udcoXYffWWtPwqlYrAFe8d2fYq29xlrbHrgOeN5UfNRXZeugOpWuo4PbiBMUo8v9rm252ztxugMBMMZE4azfQDr4Pd4O9AJG+bp6S7vpKutCC4SdOP8L5ddDp8oWFgk1Cknyu2at3YfTFfKcMeZM37dxtzFmgjHm0Uoesw34DfiXcQZjD8SpHr0DYIy52BgT7+uay/A9rMQYM84YM8D3zT4Tp+uoJAhvaxfQrZpl3gWuMMYMNsZEAv8E5llrk8otc6dvMHAn4Fag/CDpKThjli4G3qrmtYxvPZVdgK+BnsaZesFljDkP6At8aYxpY4yZ5AtuBUA2vvVkjDm33GDkvThBoaJ1mIbTHVXdeqhKpW30/X4JcL5vexmO08VWahpwunEGfkcAD1G3sOLP3zQWZxxShm/M1wN1eD2/WGu3AInAg8aYCOMMkPenO1IkJCgkye+etfZJ4M/AX3B2rtuAm4BPq3jYBUBXnGrDJzhjQ773/W48sNIYk40ziPt8X6WlLc7OMxNYDfwMVHUU0BfmwHmSqhtcXuo/wDm+I5GeqWgBa+2PwP3ARzjVgCOA8w9a7DNgIU4Y+Ap4tdzjk4FFOCHll2racyTOzrv8ZR9OJe12nG6ou4DTrLW7cT6XbsdZt3twxtXc6HuuEcA837r9HLjVWru5gveXizMmZ7ZvPNfoatp4CGttehVtBGf9HYET1h7CCZ6lj10J3IwzMHsnziDwVJzQVxsPAm/63svkSpZ5GogCdgNzgW9r+Vo1dRHOARDpwD9wwnRt36fIYcVYW9uqtIg0VsYYC/Sw1m6oYpnXgB3W2r/UX8tCkzEmBqeq2KOiUNeYGGdahjXW2qBXskSCTZUkEakx31Fwf6BcdUkOZIw53dd92wR4HFgOJDVsqwLPGDPCGHOEcebRKp2E9dMGbpZIQCgkiUiNGGP+DqwAHmvsVZE6OgOny3AH0AOn27Uxlu7bAjNxxo49A9xgrV3coC0SCRB1t4mIiIhUQJUkERERkQooJImIiIhUoCazu/qtVatWtmvXrsF4ahEREZGAWrhw4W5r7SEn6Q5KSOratSuJiYnBeGoRERGRgDLGHHwKIkDdbSIiIiIVUkgSERERqYBCkoiIiEgFgjImSUREREJDUVERycnJ5OfnN3RTgs7j8dCxY0fcbrdfyyskiYiI/I4lJycTGxtL165dMcY0dHOCxlpLeno6ycnJJCQk+PUYdbeJiIj8juXn5xMXF9eoAxKAMYa4uLgaVcwUkkRERH7nGntAKlXT96mQJCIiIg0mIyOD559/vsaPO/XUU8nIyAh8g8pRSBIREZEGU1lIKikpqfJxX3/9Nc2bNw9Sqxx+hSRjzJ+MMSuNMSuMMVONMZ6gtkpERER+F+655x42btzI4MGDGTFiBOPGjePCCy9kwIABAJx55pkMGzaMfv368fLLL5c9rmvXruzevZukpCT69OnDNddcQ79+/Tj55JPJy8sLSNuqPbrNGNMBuAXoa63NM8Z8AJwPvBGQFoiIiMhh4aEvVrJqR2ZAn7Nv+6Y8cHq/Sn//yCOPsGLFCpYsWcLMmTOZOHEiK1asKDsC7bXXXqNly5bk5eUxYsQIzj77bOLi4g54jvXr1zN16lReeeUVJk+ezEcffcTFF19c57b7293mAqKMMS4gGthR51eui61zYfuiBm2CiIiIBN7IkSMPOET/mWeeYdCgQYwePZpt27axfv36Qx6TkJDA4MGDARg2bBhJSUkBaUu1lSRr7XZjzOPAViAPmG6tnR6QV6+tr+6A5p3hgncbtBkiIiKNSVUVn/rSpEmTstszZ87khx9+YM6cOURHR3PcccdVeAh/ZGRk2e3w8PCAdbdVW0kyxrQAzgASgPZAE2PMITUsY8y1xphEY0xiWlpaQBpXKbcHigOzAkRERKThxMbGkpWVVeHv9u3bR4sWLYiOjmbNmjXMnTu3Xtvmz4zbJwKbrbVpAMaYj4EjgbfLL2StfRl4GWD48OE2wO08kMsDRQpJIiIioS4uLo6xY8fSv39/oqKiaNOmTdnvxo8fz4svvsjAgQPp1asXo0ePrte2+ROStgKjjTHRON1tJwCJQW1VddxRkJ3aoE0QERGRwHj33YqHz0RGRvLNN99U+LvScUetWrVixYoVZfffcccdAWtXtd1t1tp5wDRgEbDc95iXq3xQsLk8UNz4T8QnIiIiDcevE9xaax8AHghyW/znjlJ3m4iIiARVaM647Y5SJUlERESCKjRDkisKihSSREREJHhCMyRpCgAREREJstAMSa4oKCkEb9UnvxMRERGprdAMSW7f+XU1LklERCSkZWRk8Pzzz9fqsU8//TS5ubkBbtF+oRmSXFHOtcYliYiIhLTDOST5NQXAYae0klSUC8RVuaiIiIgcvu655x42btzI4MGDOemkk2jdujUffPABBQUFnHXWWTz00EPk5OQwefJkkpOTKSkp4f7772fXrl3s2LGDcePG0apVK2bMmBHwtoVmSCqtJKm7TUREJHC+uQdSlgf2OdsOgAmPVPrrRx55hBUrVrBkyRKmT5/OtGnTmD9/PtZaJk2axKxZs0hLS6N9+/Z89dVXgHNOt2bNmvHkk08yY8YMWrVqFdg2+4Rmd1tZJUlHuImIiDQW06dPZ/r06QwZMoShQ4eyZs0a1q9fz4ABA/jhhx+4++67+eWXX2jWrFm9tEeVJBEREXFUUfGpD9Za7r33Xq677rpDfrdw4UK+/vpr7r33Xk4++WT++te/Br09IVpJKh24rUqSiIhIKIuNjSUrKwuAU045hddee43s7GwAtm/fTmpqKjt27CA6OpqLL76YO+64g0WLFh3y2GAIzUqSpgAQERFpFOLi4hg7diz9+/dnwoQJXHjhhYwZMwaAmJgY3n77bTZs2MCdd95JWFgYbrebF154AYBrr72WCRMm0K5du6AM3DbW2oA/6fDhw21iYmLAn7fMrlXwwhg4903od2bwXkdERKSRW716NX369GnoZtSbit6vMWahtXb4wcuGaHebKkkiIiISXKEZklwakyQiIiLBFZohSZUkERERCbLQDElllaTgTUUuIiLyexGM8cmHo5q+zxANSZGA0bnbRERE6sjj8ZCent7og5K1lvT0dDwej9+PCc0pAIwBlweKNSZJRESkLjp27EhycjJpaWkN3ZSg83g8dOzY0e/lQzMkgTMuSZUkERGROnG73SQkJDR0Mw5LodndBuCOViVJREREgiZ0Q5JLlSQREREJntANSe4oTQEgIiIiQRO6Icnl0WSSIiIiEjShG5JUSRIREZEgCt2Q5PJoMkkREREJmtANSZoCQERERIIodEOSK0pTAIiIiEjQhG5IUiVJREREgqjakGSM6WWMWVLukmmMua0e2lY1VZJEREQkiKo9LYm1di0wGMAYEw5sBz4JbrP84I5SJUlERESCpqbdbScAG621W4LRmBpxR0FJAXi9Dd0SERERaYRqGpLOB6ZW9AtjzLXGmERjTGK9nEnY5XGuNVeSiIiIBIHfIckYEwFMAj6s6PfW2pettcOttcPj4+MD1b7KuaOca4UkERERCYKaVJImAIustbuC1ZgaKa0k6dQkIiIiEgQ1CUkXUElXW4MorSQpJImIiEgQ+BWSjDHRwEnAx8FtTg2UjUlSSBIREZHAq3YKAABrbS4QF+S21ExZJUljkkRERCTwQnfGbVWSREREJIhCNySpkiQiIiJBFPohSZUkERERCYLQDUllUwCokiQiIiKBF7ohSZUkERERCaLQDUmqJImIiEgQhW5IUiVJREREgih0Q5JOSyIiIiJBFLohyRgnKCkkiYiISBCEbkgCJyQVa0ySiIiIBF5ohyR3lCpJIiIiEhShHZJUSRIREZEgCe2Q5I5WJUlERESCIsRDkipJIiIiEhyhHZJcUZpMUkRERIIitEOS26PJJEVERCQoQjskuTyqJImIiEhQhHZIckdBUW5Dt0JEREQaodAOSZoCQERERIIktEOSJpMUERGRIAntkKRKkoiIiARJaIckd5QTkqxt6JaIiIhIIxPaIcnlca5VTRIREZEAC+2Q5I52rjUuSURERAIsxEOSKkkiIiISHKEdklxRzrUqSSIiIhJgoR2SSitJCkkiIiISYKEdkkorSepuExERkQDzKyQZY5obY6YZY9YYY1YbY8YEu2F+USVJREREgsTl53L/Ab611p5jjIkAooPYJv+pkiQiIiJBUm1IMsY0BY4BLgew1hYChcFtlp9USRIREZEg8ae7rRuQBrxujFlsjPmfMabJwQsZY641xiQaYxLT0tIC3tAKqZIkIiIiQeJPSHIBQ4EXrLVDgBzgnoMXsta+bK0dbq0dHh8fH+BmVkKVJBEREQkSf0JSMpBsrZ3n+3kaTmhqeKUzbquSJCIiIgFWbUiy1qYA24wxvXx3nQCsCmqr/OVSJUlERESCw9+j224G3vEd2bYJuCJ4TaoBneBWREREgsSvkGStXQIMD25TaiEsDMIjoSi3oVsiIiIijUxoz7gNzuDtIlWSREREJLBCPyS5oqBYY5JEREQksEI/JKmSJCIiIkEQ+iFJlSQREREJgtAPSaokiYiISBCEfkhyRWkKABEREQm40A9J7ihNJikiIiIB1zhCkipJIiIiEmChH5JcHlWSREREJOBCPyS5FZJEREQk8EI/JGkKABEREQmC0A9JmgJAREREgiD0Q1JpJcnahm6JiIiINCKhH5LcHue6uKBh2yEiIiKNSuiHJFeUc61xSSIiIhJAoR+SSitJGpckIiIiAdQIQlK0c61KkoiIiARQ6IcklypJIiIiEnihH5LcGpMkIiIigRf6IamskqSQJCIiIoET+iGptJKk7jYREREJoNAPSaWVJHW3iYiISACFfkhSJUlERESCIPRDkipJIiIiEgShH5JUSRIREZEgCP2QpEqSiIiIBEHohyRVkkRERCQIQj8khYVDeIQqSSIiIhJQLn8WMsYkAVlACVBsrR0ezEbVmCtKlSQREREJKL9Cks84a+3uoLWkLtweKMpt6FaIiIhIIxL63W3gDN4uViVJREREAsffkGSB6caYhcaYa4PZoFpxR+ncbSIiIhJQ/na3jbXW7jDGtAa+N8assdbOKr+ALzxdC9C5c+cAN7MaqiSJiIhIgPlVSbLW7vBdpwKfACMrWOZla+1wa+3w+Pj4wLayOqokiYiISIBVG5KMMU2MMbGlt4GTgRXBblhVXp+9mY8WJu+/Q5UkERERCTB/KkltgF+NMUuB+cBX1tpvg9usqn28aDtfLd+5/w63pgAQERGRwKp2TJK1dhMwqB7a4reYSBdZ+UX773BHaTJJERERCaiQnAIg1uMiK794/x2aTFJEREQCLCRDUszBIUmTSYqIiEiAhWRIaupxH9jdpoHbIiIiEmAhGZJiPS6yC4qx1jp3lE4BUPqziIiISB2FZEiKiXThtZBbWOLc4fIAFkoKG7RdIiIi0niEZEiK9bgB9o9Lckc515pQUkRERAIkREOSM3NB2bgkl8e51rgkERERCZCQDEkxpSGpQJUkERERCY6QDElNyypJvpCkSpKIiIgEWEiGpP1jknzdbe5o51qVJBEREQmQkAxJMZFOJSm7bOC2KkkiIiISWCEZkmIP6W4rHZOkWbdFREQkMEIyJDWJcGFM+YHbvkqSzt8mIiIiARKSISkszBAT4So3BYCvkqTuNhEREQmQkAxJ4HS5ZR08JkkDt0VERCRAQjYkxXhc+wduq5IkIiIiARayISnW4yaroHQKAFWSREREJLBCOCS5Dj26TZUkERERCZCQDUkxkeW628JdEOZSJUlEREQCJmRDUqzHTWZpSAJn1m1VkkRERCRAQjYkNfWUmwIAnPO3qZIkIiIiARKyISkm0kVBsZfCYq9zh1shSURERAInZENS6alJsgvKDd4uVkgSERGRwAjZkBTjcQPs73Jze3RaEhEREQmYkA1JFZ7kVpUkERERCZDGE5JUSRIREZEACt2QFOl0t2lMkoiIiARD6IakskqSxiSJiIhI4DWCkFS+kqSQJCIiIoHhd0gyxoQbYxYbY74MZoP8FXPwFADuKM2TJCIiIgFTk0rSrcDqYDWkpiJd4US4wsgs625TJUlEREQCx6+QZIzpCEwE/hfc5tSMc2qS0u42DxTlgrUN2ygRERFpFPytJD0N3AV4K1vAGHOtMSbRGJOYlpYWiLZVKybSRXb5KQCsF0qKqn6QiIiIiB+qDUnGmNOAVGvtwqqWs9a+bK0dbq0dHh8fH7AGViXW495/dJsryrnWNAAiIiISAP5UksYCk4wxScB7wPHGmLeD2io/xUS6DpxMEjQNgIiIiAREtSHJWnuvtbajtbYrcD7wk7X24qC3zA+xHteBk0mCKkkiIiISECE7TxKUdrepkiQiIiKB56rJwtbamcDMoLSkFmI9rv1TAKiSJCIiIgEU4pUkp7vNWqtKkoiIiARUyIckayGnsATc0c6dqiSJiIhIAIR0SIqJdAM4cyW5VEkSERGRwAnpkLT/JLdFzmlJwJl1W0RERKSOGkVIyixfSdL520RERCQAGkVIyi4oLldJ0pgkERERqbsQD0nOmKSs/CJVkkRERCSgQjokxUSWjkkqX0lSSBIREZG6C+mQVNbdll8M4W4w4ZoCQERERAIipENSkwgXxvi628CpJqmSJCIiIgEQ0iEpLMwQE+Fyjm4DZ1ySKkkiIiISACEdkmD/qUkAZ9ZtVZJEREQkABpBSHKX627zaDJJERERCYiQD0kx5StJLo+mABAREZGACPmQFOtxOVMAgG/gtsYkiYiISN01gpDk3h+SVEkSERGRAAn5kBQTqUqSiIiIBF7Ih6SmHtf+gduqJImIiEiAhHxIiol0UVDspbDYq8kkRUREJGBCPiSVnZqkoFiTSYqIiEjANIKQ5AZ8pyZRJUlEREQCJORDUoyvkpSVX+yEJFWSREREJABCPiTFlg9JrijwFkNJUQO3SkREREJdyIekpgd0t3mcOzUNgIiIiNRRyIekmMjylSRfSNI0ACIiIlJHIR+SDji6zR3l3KlKkoiIiNRRyIek/QO3i1RJEhERkYAJ+ZAU6QonwhVGlipJIiIiEkDVhiRjjMcYM98Ys9QYs9IY81B9NKwmnFOTaEySiIiIBI7Lj2UKgOOttdnGGDfwqzHmG2vt3CC3zW9lJ7lVJUlEREQCpNqQZK21QLbvR7fvYoPZqJqK9bjJ1pgkERERCSC/xiQZY8KNMUuAVOB7a+28oLaqhmJLu9vc0c4dqiSJiIhIHfkVkqy1JdbawUBHYKQxpv/ByxhjrjXGJBpjEtPS0gLczKrt727TZJIiIiISGDU6us1amwHMBMZX8LuXrbXDrbXD4+PjA9M6P8V63M48SS7fmCSdv01ERETqyJ+j2+KNMc19t6OAE4E1QW5XjcR6XGQecFoSjUkSERGRuvHn6LZ2wJvGmHCcUPWBtfbL4DarZmI9LrILivGGe5zUp0qSiIiI1JE/R7ctA4bUQ1tqLdbjwlrILQkjxoSpkiQiIiJ1FvIzboMzJglwZt12RWkKABEREamzRhGSYiJLz9/mO8JNR7eJiIhIHTWKkBTrKReSVEkSERGRAGhkIalIlSQREREJiEYSkpwxSdkakyQiIiIB0khCUvkxSVFQlNvALRIREZFQ1yhC0v6B26XdbaokiYiISN00ipDUJMKFMZBdNnBbY5JERESkbhpFSAoLM8REusgsmwJAlSQRERGpm0YRkgBiI13lpgBQJUlERETqpvGEJI+b7AKNSRIREZHAaEQhyaXJJEVERCRgGk1IiikNSZpMUkRERAKg0YQkp7vNV0nyFkFJcUM3SUREREJYowlJMZGu/fMkgQZvi4iISJ00mpDU1FM6BUC0c4cGb4uIiEgdNJqQFOtxUVjspSgswrlDlSQRERGpg0YTkkpPTZJvfSFJlSQRERGpg0YTkmI9bgDyrHOtSpKIiIjURSMKSU4lKderSpKIiIjUXaMJSTG+kJTtda5VSRIREZG6aDQhqamvuy27xBeSVEkSERGROmg0Iam0uy2rWGOSREREpO4aTUgqPbotszjcuUOVJBEREamDxhOSPKUhqbS7LbcBWyMiIiKhrtGEpEhXOBGuMPYWlQ7cViVJREREaq/RhCRwTk2yp7C0u01jkkRERKT2GlVIivW42VdoAKNKkoiIiNRJtSHJGNPJGDPDGLPaGLPSGHNrfTSsNmIiXWQVFIM7SpUkERERqROXH8sUA7dbaxcZY2KBhcaY7621q4LcthqL9bjIzi8Gl0eVJBEREamTaitJ1tqd1tpFvttZwGqgQ7AbVhuxHhdZ+aWVJIUkERERqb0ajUkyxnQFhgDzgtKaOoqJdJOVX+SrJKm7TURERGrP75BkjIkBPgJus9ZmVvD7a40xicaYxLS0tEC20W+xnvJjklRJEhERkdrzKyQZY9w4Aekda+3HFS1jrX3ZWjvcWjs8Pj4+kG30W1OPi+yCYqwqSSIiIlJH/hzdZoBXgdXW2ieD36Tai/G4sBZKwiN1dJuIiIjUiT+VpLHAJcDxxpglvsupQW5XrcR6nJPbFod7FJJERESkTqqdAsBa+ytg6qEtdVZ6ktsiE4lHUwCIiIhIHTSyGbedkFRoIlRJEhERkTppZCHJ6W4rIFKTSYqIiEidNLKQ5FSS8onQFAAiIiJSJ40yJOVZt6YAEBERkTppZCHJ6W7Ls24oKQRvSQO3SEREREJVowpJ0e5wjIFcrxOWNC5JREREaqtRhaSwMENMpIvsEl9I0rgkERERqaVGFZIAmnrc5UJSbsM2RkREREJWowtJMZEuMkt8c2Squ01ERERqqdGFpFiPi8xiX0jShJIiIiJSS40uJMV4XOwrCnd+UCVJREREaqnRhaRYj5uM0pCkSpKIiIjUUiMMSa79IUmVJBEREamlxheSIl2kF6iSJCIiInXT+EKSx0WWjm4TERGROmqEIclNvo1wflAlSURERGqp0YWkmEgX+fhCkipJIiIiUkuNLiTFesqFJM24LSIiIrXUCEOSmwJ07jYRERGpm0YYklyAoSTcA8UakyQiIiK100hDEhSHRaqSJCIiIrXW6EJSTGS5kKRKkoiIiNRSowtJsR5nPFKRUSVJREREaq/RhaQIVxiRrjAKTYSmABAREZFaa3QhCZxxSQVEaDJJERERqbVGGpLczlxJu9fB6i8Vln5vrIWSooZuhYiIhLhGGpJc/BZ1LBRkwfsXwaNHwIeXw4qPoSC7oZtXOWuhuLChWxHa9iXDW2fAY91h0VvOOhUREamFRhmSYiJdfO6eAHdugEs+gYGTIelXmHYFPHYEvHcRLH0f8vc1bEOthfSNzs7842vhqf7wz3Yw57k67dzTsws48cmfmbUuLYCNDQHLPoTnj4TkRGjZDT6/Gd483VnHIiIiNeRq6AYEQ6zHxe7dORDuhiOOdy4Tn4Ctc2DV57D6C1jzJYS5of8fYOKTEBkT/IaVhqItvzqhLWk2ZO1wftckHrqMhcIc+O7/IG0NnPoEuCJq/DJfL9/JhtRs7vt0Od//6Vg87vAAv5HDTO4e+PoOWPERdBoFZ70IzbvCojfh+wfg+TFw3N1w5C3ONiEiIuKHakOSMeY14DQg1VrbP/hNqruYSDfZ+cUH3hkWDl2Pci7jH4HtC52d6vyXYNcquGAqNO8UnAZ5S2DeS/DbM5C109fINk4oKm1Tq55gDHi9MONh+OVxSN8Ek9+CJnE1ermvlu+kWZSbbXvyeHnWJm45oUcQ3tRhYuMM+PRGyEmF4++HsbdBuG+zHn4F9JoAX98JP/7N6W49/RnoOKxBmywiIqHBn+62N4DxQW5HQMV6XGQdHJLKCwuDTiNgwiNw0YeQsQVeOR62LQh8Y3YscZ77u3shvhec9jTclAi3r4VzX4cRVzn3G7O/bSfcD3/4HyQvgP8dD6lr/H651Kx85m/ew2VHdmXigHY8N2MD2/bU4US/h+uYnqI8+OZumHImRMbC1T/CMXfsD0ilYtvCeVPgvHcgNx1ePRG+vffwHpsmIiKHhWorSdbaWcaYrvXQloBp6nGRXViM12sJCzNVL9z9RLj6B3h3MrwxEc54DgaeW/dGFGTDjH/CvBecrrRzXod+Z+0PQ9UZeC606ArvXQivngTnvAY9Tqr2Yd+tSMFr4bSB7WgS6eKnNak8/NVqXrykBtUTa51K28LXYcUn0KYfnPQ36DLG/+cIph1LnDFcu9fCqOvhxAfBHVX1Y/qcBglHww8PwdznnaMeT3zACahRLSE6Dtye+mi9SP3L3eN09e/ZCJk7nM8hE+5U2E04mDDnC5oJh/AI57MmpnVDt1oECnMhIrrBXj5gY5KMMdcC1wJ07tw5UE9bKzEeF9ZCTmFx2QzcVYrvBdfMgPcvgY+vdna+x/2f86FRG2u/dcbI7NsGw65wduJRzWv+PJ1GwDU/wdQLnBB38sMw+oYqg9ZXy3fSvXUMPdvEAnDT8d157Lu1zFqXxjE946t+vby9sOwDWPgmpK7EupuwNf44OuxdiOv18dBr4v5g0RB2LoPFb0Piq07wvOQTZ7yZvzzN4LQnnYH8n98CH1114O/dTZywFO0LTdFxzmOMbzsoW+/mwNtuj9OOzmOcnY5IQyjKh9RVsGfT/kCUvsG5nZ9Rs+eKbAbH/8WpdGubDg1er1Mtz9zuDOvI3A6ZO51QnLUDslKgVQ/ofTr0PKV2+6RgKcqHvZt92+uG/dtt+gZnOpe7k/wvMASYsX50p/gqSV/6OyZp+PDhNjExsY5Nq72p87dy78fL+e2e42nfvJoKQ3nFhfDVn2HxFOgzyRkAHNHE/8dn7oRv74ZVn0F8bzj9P9B5dM3fwMEKsuGT65zB5kMvg1Mfr3BAd2pWPqP/+SM3Hd+DP5/U03locQmnPDWLsDDDt7ceQ4TroOBnLWydCwvfgFWfOrOUtxsMwy7nmV0DefKXFE7sHsv/es6HX5+GohwYcgkcdy80bVf391ad7DRY/iEseRd2LXe+5fY/B0552AkztVVcCNsTIWe388GSm+6ExNLbpZf8fb4uRwtl/yq+G6X/O8V54C2GJq2hz+nQ70xnvFko71ysdd5TfQ9093qhMMtZ73kZznX+PmcnX5AFrkiIiHH+LyOaVHzbFVn3dlhb9w9layEnzRmTGNM68NtDfiZsmw9bf4MtvznV35LSKUQMNOvoHOUZdwS0PGL/dbOOvvaVgPU67bPe/bezd8EPD8CmmdB2oHNgS6cRNW9fSZHzXDm7ff9nvuucNN/tdOc6Igba9ndeq01/Z0de3XbnLYE9m53PhJQVsGul8/8b3dK5lFaHy3/hiWrpdM1jyw0jsOX+v333RcbW7bMFnG11x2JoPzR4BwV5S5zP7jVfwfrpsDcJvAfND2fCIbad81ndJN5pU9ZOCHNBwjHO51WviRDbJjhtrLDdXti5GNZ95wwpSd8AGdso9wHrfJbGdXe22bjuMOamQ4dSBJgxZqG1dvgh9zfGkPTlsh3c9O5ipv/pmLKKit+sdQ7Bn/4XaDcQLngPmravevmCTKcC8+PfoLgAjr3LOZKqFkemVcrrhZ/+Dr8+6eyAh17qBLH4XmVdTVPmJHH/ZysPed8z1qRyxRsLuHdCb647ppvzDSNtDexc6kyFsHstRMQ6XXxDL4P2g3nt18387ctVtG/mYce+fH6+8zi6ePJg1mOw4FXnQ2zMH5336WkauPcJToBZ/x0smepce4udD5vBF0L/s+v+ARZARSVe5q/dyvDCRCLXfeF8WBXlOh9IvU/zBaajnH9wa50P8qwU55td5s79t7NSnL9jy24HXprEB/8blLXO/FI7lzgfojsWO12a+RlOl298H2c7a93H2eZa9ai4e7Ok2Kme7tnku2x2rjOTne3XeinbGR1825Y4O/2CTN/v6qBJ63LBoFu5gNDtwC89Xq+z7suqLhv3V2H2JoHLAy06Q/MuzqWF77p5Z+d2RJP9Qaj0seWfZ89mJ/CBs1OKbQdNOzifJ03bO2GlaXvnvuiWztG2YS7nf6vs2u1cG+MEjK1znEC05TdIWeasqzCX88WmyxjoONL5+7RIqFv3sbWw8hPnSNusnc7nzYkPVf+/V1IMm2fC8mlOl3bp+y8vzO1s103iILoV5O2B1NX7A154JLTuDW0GQNsBToAy4bBrBaQsdwJR6irn/wyc37XqCU1aOcG69AtOSUHt3rsJc4ZhDL3MqbjU5ItC6mrn83Hpe857d0c7B48MOBeOOKHu+4TCXNg0wwlG67513md4BCQc66ynWN+21dS3rTWJPzCce71OmF7zhXOU955NgHGOCu5zmvOZ1TKhbm2sSEG20+5138K66c6BNibMF4p7+gJRaSg6wqng17PfVUiauTaVy19fwEc3jGFYl1ruUNd+63THRMTAqY85O+ryO7TMnc6HR1aKU10BZ0M97SnnjxwsS9+DL/+8/zUxzkYd34dPtseyqqQD9132B6cNWTshbR2krWH23Nk0zdpIv4gUwsoeC3Qc4XwY9P9D2Q7k08Xbue39JYzv15b7T+/LMY/O4OqjE7h3Qh/nMXs2wY9/h5UfO9/Qjr3b+VApznfKpsV55W77LkV5+7+dh/nGQJiwcuMhfNfJC5zAmbfHOQJw4HlOOGrdJ3jrtJastdw5bRnTFiYT63Fx9tCOXDS0FT32zXGqieu+c/5O0XHOt9OslIrPJxjV0hlgXpjjhIzyISEixvn7loWm1s7OzxVV7tp3cXmc67BwXyWopIJqQYnzu+xdThAqDUW5u53XM+HQui+0H+ys//T1zoEDezY6/wPg/J1Kw1NsG+db4J5NzgEQpcuA07aW3ZwwULqjx/j+7qW3zf5tIbKp0wXgaQYe33X5nyNjnS8hhTnOei0svWTvv52fCRlJzpGhezY677O82HZOgMjPcEJMcbnZ+MMjy1VeEpztN2Or8772bjlwWXB28MUFBwYBE+4EqJblwllYuNPlsW+7rwtku/NzTc4tacL2bxcuj/N/2+VI59JxRM0q3jVRkAUzH4G5Lzhfhk58EIZceuBQBK8Xts2DFdNg5afOthTZzKlSdBzuC0TxTohp0sr5Ox8c/EuKYPd6XxBa5lSHUpbv3y5LeZo7walNfycUtOnvBPeDA6G1TojKTXfGY5VeF2axv7vc14by2yLG2W6WvOt8fsa0cT5/hl7q/E0rUlLkBI4FrzrTu4RHQL8/OOFo889O2Mzb67S935lOYOp8pH/DOax1wvH66U4w2viTsx1GNnMCXO+J0P0EX4Wshqx1Qt3qL5zQlLLcuT+qpVP5bBLvvP+y262dz5+Y1s7rhbud91oa5Mt+9oWyvUlOIFr3jTPlTUmh0+7uJ0DP8c4+o4ZHbgdTrUOSMWYqcBzQCtgFPGCtfbWqxzR0SFq4ZS9nv/Abr18xgnG96jD4cNdKePd82Ld1/33hkU5Kjy13adrO+UftfmL99JuWFDk7pdRVzg4sbTXFKashfQMuU/G38OImbZmf1YriuJ4cc+RRTmUgvrfzoVXOjLWpXPNmIiO6tuT1K0bgcYdz3ZREFiTtZc69xxPpKvetZPtCZx6ipF8C997CI5x//MEXQbdxQS+x1sVT36/jPz+u59IxXcjMK+Lr5SkUlngZ2bUlF43uzPheTYnc7PvW5y2ueLuJaXvgB3xx4UHVmHKXvUkHhpC6MmFO0Gk/xAlF7Yc4g/QrqhIVFzo7j9TVThUybY2z7WXvciorB1fAWnZzgl8DjSMoU5B10Bgd33r0NNtfXSqtOjXtUPmOq7RilLHVeXzGFud2eMSBlarmnf2rPFjr7LRLQ1NehtNVUlLk/I29xb7bRU51xlvkhIvOY5y/VSC6FGti1yr46nana6/DcGfeORPmBKMVHzvbrMvj7PwGnOt8Ftb1QIjSMJ+ywgn3bfo7VZJ6+Ywthg3fO+Mz13/nBNSuR8Owy51qi9vjhN5FbzpDFUr/D4ZfBUMuPvBztbjQqaIs/xDWfO0E/Nj2zhfTPpOc95bp+/Jd+sW7/O3SL7VNOzifjb0nOr0Jge4K35vktC99g1PpyU7zdZemOV9E/Gac/4vSSl5cd2e76DneGX5ymM5VV6dKUk01dEhavyuLk56axX8vGMLpg6roKjtIflEJ2zPySN6bR/LeXJL35pG+O5VBYZuYfNww3M07QFSLoP2TLtmWwS1TF/P6FSM4Ir5m/dhT5iTx98+WMP3S9nQt2ersGGLb+rpHekJUc56cvpZnftrA+9eOZlS3QxP8wi17ueh/czkiPob3rh1dNuj9l/VpXPLqfJ4+bzBnDulw4IOsdb4lZG7fX8lweXy3fZUOV6RzvwkvV82ooLphvc63lAYotdbUBwu2cddHyzh3WEcePWcgxhjSswuYtjCZd+dvZUt6Li2bRHDu8I5cOLIzXeIC8E2/pNjpjiqtzJWv3JW/9hYfVK0LO+gopnDnW22bfg161IiEGGth2fvOUIQc32z+YS7noIX+50DvU2tX0TjcZe6AJe84Z0bI2OrsA9oNhs2znM+sHifBiKudYFjduLPCHFj7jdMdueH7Q7/0uDzlvki13f9lquvR0G5Qw33pKMzdH5xyUp3us5LC/cG+pMj5uSzYFzpfAHucDK26N0yba+h3FZJ27stjzL9+4p9nDeDCURUfaWet5ZsVKXy7IqUsEKVmHdiH7Q43tG3mYduePM4b3olHzh6ACeJGet2URL5buYtTB7Tl+YtqNuHh+S/PYXd2IT/8+dhKl8krLOHEJ38m1uPiy5uPwhW+/1vzul1ZnPviHFpEu/nw+iOJj93/TdXrtRz/xExaxUQy7YYja/7GGpmf16Vx5RsLOPKIOF67fATu8AOrD16vZfbG3bwzdyvfr95Fiddy6wk9+JNvML00HGstU+dv46juregcp4BYK3kZsOAVp1um75mHVZdJUHm9znirRW/B9kXQ9wxnwtrKuuGqk7vHqcJHNt0fijzNGr76+jtVWUg6fPsy6qC0ApKVX/GZ4Hdl5nPfJyv4YfUu2jb10C2+Ccf1iqdji2g6tYyiY4toOraIonWsh/AwU1aB6dgiipuDNHv1tj25fL9qF22aRvL18hSWJ+9jQEf/KiqlE0jedHzVbYuKCOcvE/twwzuLeGfeVi47sisAyXtzufTV+US4wphy1agDAhJAWJjholFdePjr1axJyaR32wAP1A4hK3fs48a3F9KzTSzPXzT0kIAEzvo6ukc8R/eIJ2VfPv/6ZjX/+XE93eKbcMbgDhU8q9SXqfO38X+fLGdkQkvev3Z0UL/0NFpRzeGYOxu6FfUvLGz/aa4CIbqlE7TksNYoQ1K0OxxjILvgwFKmtZYPErfxj69WU1js5b5T+3DlUQmEVzPh5J9O6kny3jye+H4dHVtGcdaQjgFv85S5WzDG8PZVo5j80hwe/W4NU64a5ddjy08gWZ3x/dtyVPdWPDF9LRMHtsMAl746n5zCYj64bgydWlb87fqcYR15bPpa3p67hX+cOaAmb63R2J6RxxWvL6BZlJs3rhjh1xxcbZt5eOycQezcl89d05aR0KoJAzs2r/FrW2t59qcNlFjLH8d1rzCcSdU2787h71+uomWTCOZv3sPsDekc1aNV9Q+UoHn8u7VEusK46fjuCqxyWGqUn7RhYYaYyANPTbJtTy4XvzqPuz9aTt92TfnutmO45phu1QYkAGMMj5w9kCOPiOOuacv4bcPuah9TE7mFxbw3fyvj+7elR5tYbjyuO7+s382cjel+Pf7gCSSrYozhwUl9yS0s4e9fruKKNxawPSOP1y4fQZ92lVeIWjSJ4LQB7fhk0fZDwufvwb68Iq54fT55RSW8fsVI2jT1f1BqhCuMFy4aSquYSK59ayGpmTU4qgmn++7+z1bwxPfrePqH9Ux+aQ7Je+twqpnfoeISL396fwnucMOnN46lfTMPj09fSzCGG4h/PluynWdnbOCJ79dxz0fLKfHqbyGHn0YZkgCaetxk5hdR4rW89utmTn5qFku37eMfZ/Zn6jWj6dqqZgNpI1xhvHDxMBJaNeG6txeyblcF83/U0seLtpOZX8yVY7sCcMmYLrRt6uGx79ZU+yGempXPvM17OHWA/xM7dm8dy5VHJfDZkh2s3JHJcxcOZUTX6qdKuGh0F3IKS/hsyXa/X6sxKCgu4fopC9m8O4eXLh5Gr7Y1H5waFxPJK5cOZ19eEde9vZCC4hK/Huf1Wv7vk+W8PXcr1x3Tjf9eMIT1u7KZ+MyvTF+ZUuN2/F49O2MDS7Zl8PBZA+gcF80tJ/RgybYMflqTGtTX/XX9bqbMSWLWujS27clVEPBJ3pvLXz5dwbAuLbhpXHfeT9zGLVMXU1hcxzmyRAKsUXa3AcREutiQms3kl+awcMtejusVzz/PGlCzGbgP0izKzetXjOSs52Zz+Wvz+eSPY2tUUaiItZY3fktiQIdmDO3cAgCPO5xbTujB/32ynB9Xp3Ji38pnQ/1uRQrWz6628m45oQdrU7L4w9AOVT5/eUM7N6dPu6a8PXcrF47sfNiWxxck7eGX9bu57YQe1Z+7rxrWWu6etow5m9J56rxBHNm99t0zfds35cnJg7jhnUXc98kKHvMdFVeZEq/l7o+ceZhuGted20/uiTGGAR2acdPURVw7ZSFXjO3KvRP6HDqTupRZsi2D//60gTMHty872vXsYR154eeNPDF9HeN6ta7zdnKwfXlF/O2LVXy0KPmA+yPCw+gcF01CqyYktGpC1zjnul0zD9ER4URHuohyh/tV4Q5VJV7L7R8sxeu1PDV5MJ3jomke7eYfX60mq6CYFy8eSnREo901SYhptFtirMdF4pa9NI928+TkQZw1pENAduodmkfx2uUjOO+lOVz5xgLev24MMZG1X42zN6SzITWbJ84ddED7zh3ekZdnbeTx6Ws5vnflH+I16WorLybSxZtXjqzRY4wxXDy6M/d9soJFWzMY1qVFjR5fHwqKS7jtvSVsz8ijWZSbq46q2+yxj09fy6dLdnDnKb0CMhZtwoB23HJCD575cT192jWttH3FJV7unLaMTxZv57YTe3DrCT3Kto+urZrw0Q1H8q+v1/D67CQWbtnLsxcM1dFaFcgtLOZP7y+hTWwkD52xfy5cd3gYt53Ygz+9v5RvV6bUqBJbnV/X7+bOaUtJzSrg5uO7c8HIzmzdk8vm3Tkk7c5hk+/653VplVZOIl1hTmiKcPmuwxnapQUPnN4vYO1sKK/8sol5m/fw2DkDy7bZq4/uRlOPm3s+Xsalr87n1ctH0Czq8JxPR35fGm1ImjS4Pd3im3DnKb0POVqrrvp3aMZzFw3lqjcT+eM7i3j1suEHHE5fE6/P3kyrmAhOG3Tgh7Q7PIw/n9yLW6Yu5otlOyo8Kqq0q+3mao5qC6QzB3fgX1+v4Z25Ww7LkDRlzha2Z+TRu20s//5mDUceEVflWKuqfLBgG8/N2MgFIztz43GBm0X9thN6sDYlk4e/WkWP1jGHnHi4qMTLnz9YyhdLd3DHyT0rPGox0hXOg5P6MbpbHHdNW8rEZ37h3+cMDOjOvjF4+KvVJKXn8M7Vow7Z6U4a1IHnZmzkye/XcUq/tnWu3uQWFvOvr9cwZe4Wjoh3guzgTs0BaN88itEHzU1W4rXsyMhj8+4c0rIKyC0qIbegmNzCEvKKSsgtLCa3oITcQmf+ttdnJ3HOsI70a3/4zyNWmRXb9/HE9LWcOqAt5ww78EvH5BGdiPG4uPW9xVzw8lzevHJkwD+7RWqqUc6TVF9KT6R7wchO/POsms+hlLQ7h3FPzOTmciekLc/rtUz876/kFhbzw5+PPeSIpsrO1RZs93+6gvcTtzHv3hNo0SSA56ero325RRzz2AwGdWrOU5MHMf4/v9AyOoLPbhqLx12zk4smJu3hglfmMrpbHK9fPqLWIbgyOQXFnP3Cb+zIyOOzm44iwTdGrrDYy63vLeabFSncM6E31x9bfTjbtieXm6YuZum2DC4Z3YX7Jvap8fsNlBKvZX1qFku3ZbBk2z6WbssgLbuAd64eVa/bKMBPa3Zx5RuJXHN0AvdN7FvhMl8v38mN7yziqfMG1alSmJi0h9s/XMrWPblcNTaBO07pFdC/wb7cIkb96wfOGtKRf/0hNI8uzSss4fRnfyUrv4hvbz2m0s+On9elcf2UhbRt5uHtq0fRoQ5DJET8Vdk8SRrIUAcXjOzMTeO6M3X+Np6fubHGj39zThKuMMPFlUx4GRZmuPOUnmxJz+X9BdsO+X1tu9rq6uLRXSgs9vLhwkPb1JCen7mBzPwi7hnfm7iYSB47ZyBrd2Xx72/X1Oh5dmTkcf3bC+nQPIpnLxga8IAE0CTSxSuXDic8zHDNW4lk5hdRUFzCH99dxDcrUvjLxD5+BSSATi2j+fC6MVxzdAJT5m7h/JfnklnJHGGBlrw3l6+W7eSfX69m8ktzGPDgd4x/+hfu/mg5Xy7bQVxMBF6v5Zapi8kv8m+wekUKikt49dfNrNi+z6/l07MLuGvacnq3jeWOU3pVutz4fm3p264pT/+wnqKSmg8azi8q4V9fr+bcl+bgtZb3rhnNX07rG/CQ2izazaRB7fl08fZ6+9sG2iPfrGZDajaPnzuoyi9Xx/aM5+2rR7I7u4BzX/iNjWk1OSWGSGApJNXR7Sf35KwhHXjsu7U1Ouoru6CYDxOTmTigHa2rGPw9rldrhndpwTM/rievcP9OpjZHtQVKr7axjOjagnfmbcV7mBytk7w3l9d/S+IPQzrSt73TvXZcr9ZcfmRXXp+dxM/r0vx6nrzCEq6dkkh+kZf/XTacZtHBGxfRqWU0z180jKTdOdw6dTE3vL2I71ft4qFJ/bj66JrN4hvhCuO+iX158eKhrNi+jyteXxDUqRoKiku4e9oyjvr3DP747iLemJ1EYbGXycM78dR5g/jp9mNZ+teTmXLVKB6fPIg1KTUPq6Wstdz/6Qr+/uUqTvvvr0x+cQ7frthZ6ZFi1lru/Xg5mXlFPH3+4APPN3iQsDDD7Sc7X0Q+Wphc6XIVWbF9H6f/91demrWJC0d25ttbj6nwdD+BcsnoruQVlfBxDdtZXonXNkjomLE2lTfnbOHKsQkc3SO+2uWHdWnJ+9eOobDEy+QX5/gdjkUCTSGpjpw5lAYwultL7vxwGfM2+Te30UcLk8kuKObysVUPLDbGcNf43qRmFfDWnKSy+2t7VFugXDSqC1vSc5m9MbBzRtXWk9PXYXBCa3n3TOhNzzYx3PHhUtKzCyp+sI+1lrs+WsbKHZn85/zBdG8d/ArdmCPieGBSP2asTeOnNak8fFb/spnQa2N8/3Y8e+EQlmzL4Ko3FhwQrAMlNSufC16ey/uJ27jm6AQ+v2ksKx46hU//OJYHJ/XjrCEd6RYfU3awwbhyYXXm2pofcj9l7hY+SEzmumO68ZeJfdixL4/r317EsY/N4JVZm9iXd2Bl5cPEZKav2sWdp/Tya3b443u3ZnCn5jzz43q/p2b4IHEbf3j+N7Lyi3nzypE8fNYAmtThAA5/DOjYjEGdmvP2vK21nt/pye/XcsITP/OPL1dRXIvKWW04Vb1l9GwTw13jK6/qHaxv+6Z8eP2ReNzhXPS/eazakRnEVopUTCEpACJd4bx08XA6tYzi2ikL2ZBa9Tc1r9c57H9wp+ZlAzurMjKhJcf2jOeFnzeWldobqqut1IQBbWnZJIK3525pkNcvb8X2fXyyZDtXHpVwyBQPHnc4/zl/CPtyi7jn4+VV7lxe+HkjXyx1jmQ7oY9/0yIEwiWju3D/aX157sKhXDSqS52fb3z/djx13mAWJO3hmrcS69TNdbBlyRlM+u9sVu/M4rkLh3LfxL4M7Ni82ikI7pnQm15tYrnjw2Xsriasljd3Uzp/+2IVJ/Ruzd3je3P10d34+c5xvHjxMNo3j+Lhr1cz5l8/8sBnK9i8O4et6bk89MVKxnSL8/vIRmMMd5zcix378nlvftVdyEUlXh78fCV3TVvGiIQWfHPr0Rzbs/rKSKBcMroLG1KzmbtpT40fm5qVz2u/JtG+mYf//bqZS16dX+0Xh7oqrertyy3i6fOG1LgbMqFVE967djRNIsK55NV5bEgN3Px0Iv5QSAqQZtFu3rhiJO5ww+Wvzyctq/IPn5/Xp7F5dw5X+CaP9Medp/QiI7eIV2ZtatCutlKRrnDOHd6RH1ankrKvZjNIB9q/v11D8yg3N1RyBFqfdk25a3wvvl+1i6mV7AR/WLWLx75by6RB7bnBz7FAgXTVUQlMDGBVcNKg9jx2ziBmb9zN9TWYvLIqny3ZzrkvziE8zDDthjE1aq/HHc4zFwwhM7+Iu6Yt86sSsj0jjxvfWUTnuGieOn9wWWUqPMwwvn9bPrhuDF/efBQT+rdj6vxtjHt8Jmc9P5uwMMMTkwfVaO6jsd3jGJXQkmdnbKi0+rYnp5BLX53PG78lcdVRCbx5xch6P3DhtIHtaBblrtWXk+dnbKSwxMvbV4/iycmDWLR1L6f/91eWJWcEvqE+HyRuK6vqlXaD11SnltG8c81owsIMF74yj6TdOTV+Dmstr8/ezC1TFzf451VD2JKew8NfreLZn9azNiXrsJhpfm1KFtNXphwWbamKQlIAdWoZzauXjWB3dgFXv5VY6YftG7OTaB0byYT+/u9k+ndoxsSB7Xj11828M3drg3a1lbpoZBe81jJ1/tYqlysu8bImJZOk3TkB/4eYtS6NX9bv5ubje9C0inOpXTk2gaO6t+LvX646ZEzG+l1Z3Pb+Evq1b8q/z656gsdQcvawjvzrrAHMXJvGTe8urtXAZHDGsfzrm9Xc+t4SBnVszmc3ja3VYei92sbyfxN689OaVKZUs5PPKyzh2rcSKSr28sqlwyv92/bv0IwnJg9i9j3Hc9uJPWga5ebfZw+s8aSxxhhuP7kXaVkFTJmbdMjvV+3IZNKzv7Jw616eOHcQ95/WNygD+qvjcYczeXhHvluZwq4anN5mR0Ye787bytlDO9AtPoY/DO3IRzcciTGGc16cwweJgT8II2l3Dg99sYojj/C/qleZhFZNePfqURR7LRf9b16NTsuzL6+I699eyENfrOLLZTsY/59ZfLvi9zFb/aa0bG7/YCnHP/Ezb/yWxOPT13HK07M49rGZ/O2LVczZmF5v3a7lrd6Zybkv/sa1UxZy89TFh/XBCJoCIAimr0zhurcXclKfNrxw8bAD5l/ZkJrNiU/+zJ9P6sktJ9RsfqNNadmc9NQsvNZyRHwMP/z52EA3vcYue20+a1Iy+fXu43GHh+H1WpLSc1iWvI+lyRksT97Hih37yC9y/hFbx0YyMqEloxJaMjIhjh6tY2o923GJ13Laf38lp8CZIqG6Lp9dmfmc8vQsOrWI5qMbjiTCFca+3CLOeO5XsgtK+PymsXWakf1w9dacJP762UpOHdCWZ84fUqOde2Z+EbdOXcyMtWlcOKozD57er06ze1trufKNBfy2MZ0vbj6qwu5iay23vb+Ez5fu4NXLhnN87/rr+rz0tfksT87gl7uPL5sk9stlO7jzw2U0i3Lz0iXDGORHF3kwJe3O4bjHZ/KnE3ty64n+fYbc+7Ezc/uMO46jY4v9k47uySnk5qmLmL0hvazbNxCzty9LzuCOD5eSsi+f7/50DO2aBeb/asX2fVz4ylxaNIngg+vGVHvGg5U79nHjO4vYvjePeyb0Zlzv1tz23hKWb9/H+SM6cf9pfYM+lqwhbEjN4tmfNvD50h1EuMK4cGQXrjvWORjkh9W7+H7VLn7bkE5hiZfm0W6O79Wak/q24Zie8UFfH0m7czjnxTm4wgx/GNqBl2Zton1zD8+cP4QhnRtu7r3KpgBQSAqSN2Zv5sEvVnHF2K4HzJL7189W8N78bfx27/G0iqn5RGn3fLSM9xZs45YTKp5bqb59v2oX17yVyKRB7UnPKWBZ8r6yEwt73GH0a9+MgR2dS25hCfM372Hepj2k+L4FN492M6JraWhqSd92Tf3eiX+0MJnbP1zKfy8YUna6iep8u2In17+9iBuPO4I/n9STK95YwNxN6Uy9ZjTD/Th/Xaj63y+b+MdXqzljcHuenDzYr4kTN6Vlc/VbiWxNz+WBSf24ZHTdx0sBpGUVMOE/s2gVE8mnfzx0DquXZ23kn1+v4c5TevHHcd0D8pr+WrotgzOem83tJ/XkxnHdeWL6Wp6fuZFhXVrwwsVDaR1bt9MQBcqlr81nXUoWv949rtr/l6TdOZzw5M9cPKrzAbOOlyou8fLYd2t5adYmhndpwfMXDa3yiNuq7M0p5NHv1vLegq3ENYnkycmDDpksta4Wb93Lxf+bR9tmHt6/bkyFn6PWWt5bsI0HPl9Jy+gInr1wSNn/d2Gxl6d/WMcLP2+kS8to/nP+kAYPvhWx1rJqZyYz1qTSLMpNV9+pbNo3i6r0i+XalCz++9N6vlq+E48rnEvHdOHqo7tVOClndkExs9al8f2qXfy0JpV9eUVEuMI4fWB7/nxyz6DMT5WyL59zXvyNnIJiPrx+DN1bx7Jwy15umbqYXZn5/Pnknlx/zBEBP02QPxSSGsDfvljFa7M389fT+nLlUQlk5hcx+p8/MqF/O56YPKhWz7krM5/7PlnBQ2f0OywmWSvxWk54YibJe/Po3S6WgR2bM6hjMwZ2bE6P1jEVfoBba0nem8e8zXuYvzmd+Zv3kJTulM87NI/izlN6MWlQ+yr/UfKLSjj+8ZnEx0byyY1ja/RPdfe0ZXywcBvH92rNj2tS+ffZAzhvRMVzVTUmz8/cwKPfruWcYR159OyBZeussNjLrsx8du7LZ+e+POc6I4+PF2/HHR7G8xcNPWS26LqasSaVK95YcMiXiFnr0rj89fmM79+W5y4c2iBdn1e/mci8zekM69KCmWvTuGBkJx6c1K/KqQTqW+mXkxcvHsb4/m2rXPa29xbz7coUZt01rsqQ98XSHdw1bRmxHhe3ndiTCf3b+j3mqsTrdLs/Pn0tWfnFXDamK7edVHUXeF3M25TOZa/Pp2ucM7C7efT+duYWFvOXT1fw8aLtHN2jFU+fN5i4CoLU3E3p/Pn9JaRmFXDbiT244bjuh8U587btyeXzpTv4dPF21ldwEFCkK6zsnH8J8U1IiGtCfGwkHyRu45sVKTSJCOeyI7ty1VEJFb7vihSVeFmQtIdvlqfwvq/r9bIxXbjxuO4BG3e3N6eQyS/NYee+fN69ZhQDOzYv+92+vCL+7+PlfLV8J2O7x/HU5MG1Duq1pZDUAEq8lhvfWcj0Vbt48eJhbNuTyz++Ws2XNx9F/w6he2qBg5UePVWXCfR2ZeYzd1M6L8/axModmfRr35R7J/ThqB4Vn1D2hZkb+fe3a5h6zWjGHFGzHXhOQTETn/mFpPRcLj+yKw9OCv3zYfnr6R/W8fQP6xnSuTler2XHvnx2Zxdw8MdAbKSLfh2a8tg5g+jUMjjnhHvw85W88VsSb1wxguN6tWZLeg6Tnp1Nu2YePrrhyAbrBlm1I5NTn/kFV5jhgUn9uHjU4Xcy5xKv5ZhHZ5DQqglvXz2q0uXWpmQx/j+zuPaYbtw7oU+1z7smJZPb3lvCmpQsXGGGY3rGc8bg9pzYp02lf4+FW/bywOcrWLE9k1EJLfnbGf3p1Tb4R93+sj6Nq95IpHe7WN6+ehRNPW42pGZz4zsLWZ+aza0n9ODm43tUGXz25RZx36fL+XLZTkZ2bcmT5w06oDuyvuzNKeSr5Tv5bMl2FiTtBWB4lxacMaQDEwe0o6jEy6a0HDbvzmHz7mzfdQ5b9+RSVOL888ZGurhibFeuPCrhgNBYU9sz8njq+3V8tCiZmEgXNxx3BFccmUBURO0/37MLirnolbmsTsnizStGVviZXVr9e+iLlTSJcPH45EGM69W61q9ZUwpJDSSvsIQLXpnLmpRMmnrcdImL5sPrj2zoZh22vF7L50t38Nh3a9mekccxPeO5Z3zvA46M2ZNTyLGPzmBkQktevXxErV5n/a4svl2RwvXHHXHI6V4aM2stz890pjqIj42kfbMo2jbz0L65h7bNomjfzEPbZh5ig1QBKC+/qIQznp1Nek4hH99wJNe8lciurHw+/+NRDX6y3s+X7qBjiyiGNuAYieo8+9N6Hp++jh9vP5Yj4mMqXOa6KYn8tiGdWXeN87siYK1l5Y5Mvli6g8+X7mDnvnw87jBO7NOGSYPac2yveCJd4ezOLuDf36zhw4XJtGkayX0T+3L6wHb1Gih/WLWL699eyOBOzTlvRCce/Hwlke5wnj5vsN/dfNZaPlm8nb9+thID/OOs/hWeKzNQ8otK2J1dQFpWAVvSc/ly2Q5+XpdGUYmle+sYzhrSgUmD2vv15aS4xMv2jDyS9+bRv0OzgJ4UeG1KFo99t4YfVqfStqmHP53Ug7OHdqzxAQv5RSVc+cYC5m3ew4sXD+OkvlWPMVy/K4ubpy5mTUoWVx2VwF3je9VLFVchqQHtzi7grOdns21PHs9dODSgh3o3VvlFJbw9dwv//ck51chZQzpw+8m96NA8ir99sYo3ftvMt7fV7znrJPDWpGQy6dnZuMIM+UUlvHXlqEqrh3Kg1Kx8xj7yE5eM7spfTz/03HTLkjOY9OxsbjuxB7edWLvxi16vJXHLXj5fup2vl6ewJ6eQph4Xx/Zqzcy1qc4O8KgEbjm+R4NV/r5evpOb3l2E18KwLi149sIhtRoovm1PLre9v4SFW/byh6Ed+PsZ/Wv9nnZnF/BB4jZSfJXa3VmFpGUXsDurgKyDZsJv0zSSSYPac8bgDvRr3/Swq1rO25TOI9+uYfHWDLq3juGuU3pxUt82frWzuMTLje8sYvqqXTU6P2J+UQn//Ho1b83ZQv8OTXn7qlF1qo75QyGpgW1Jz+HbFSlcdVRCgxw6HKr25Rbx/M8beH12EgAXjOjEu/O3cvbQjjxy9sCGbZwEROlBDn+Z2KfGp2P5vbt56mJ+XpvKvP878ZDukNIj9WbdNS4glcGiEi+/btjNF0t28MPqXQzq1JwHTu9H99YVV7Hq0/SVKaxPzebaY7rVqTJcXOLlmR/X898ZG+ga14T/XjCkRkMjrLV8mJjMw1+vZl9eEc2i3LSKiaBVTCStYiOJj4kkPjay7L42TT30adf0sBgLVRVrLd+t3MWj361hU1oOXeOiGd0tjlHdWjIqIa7Co4K9Xsud05bx0aJkHjy9b7Vnl6jIdytTmL5yF4+fG/ypWRSSJKRtz8jjielr+WTxdiJdYfx857hqD/+V0LE9I++wOBAh1MzfvIfJL83h0bMHMnlEp0Puv3dCb65rgMlRQ92cjen86f0lpOcUcPf43lw5NqHag0M2787h3o+XMXfTHkZ2bck//9C/Xk5tVJ+KS7x8vGg7361MYX7SnrIjmTu1jGJUgjMh6+hucXRsEcXfvlzF67OTajRVRUNSSJJGYW1KFnlFJX6dzkWksbPWMv7pX3C7DF/cdBTGGKy1nPfyXDbvzmHWnePqNOD292xvTiF3fbSM71ft4rhe8Tx+7qAKpxsoLPbyyi+b+M+P64l0hfF/p/bhvOGdGuQw9vpU4rWs3pnJvM17mLcpnflJe8jIdSaFbBUTye7sAq4cm8D9p/U57LoQK6KQJCLSCE2Zk8T9n63k0z+OZXCn5sxal8alr83nb2f049IxXRu6eSHNWsuUuVv4x1eraRbl5qnJgw8YM7do617u/Wg5a3dlMXFAOx44vW+9H7p+uPB6LetSs5i3aQ/zN+/hiNYx3HZCj5AJiwpJIiKNUJZv/rXx/dvx+LkDOeO52aRnFzLjjuMCMnu2OKfRuHnqYjamZXPdMUdw3THdeOqHdUyZu4W2TT38/Yz+nFjNUVtyeKssJDW++dhFRH5HYj1uzhragQ8SkxmV0JJlyft49JyBCkgB1KddUz6/aSx//3IVL/68kf/9sokSa7lsTFfuOKVX2SlspPFRJUlEJMStSclk/NO/EB5m6BIXzfTbjtFRtEHy1bKdfLI4mT+O696g5xqTwKqskuTXf5ExZrwxZq0xZoMx5p7AN09ERGqrd9umjOzakhKv5U8n9lRACqKJA9vxv8tGKCD9TlRbIzTGhAPPAScBycACY8zn1tpVwW6ciIj4567xvfhy2U4mDtBktSKB4k9H6khgg7V2E4Ax5j3gDEAhSUTkMDG8a8uyM92LSGD4U5PtAGwr93Oy7z4RERGRRsufkFTRJAeHjPY2xlxrjEk0xiSmpaXVvWUiIiIiDcifkJQMdCr3c0dgx8ELWWtfttYOt9YOj4/37+zLIiIiIocrf0LSAqCHMSbBGBMBnA98HtxmiYiIiDSsagduW2uLjTE3Ad8B4cBr1tqVQW+ZiIiISAPya5pQa+3XwNdBbouIiIjIYUMzjomIiIhUQCFJREREpAIKSSIiIiIVUEgSERERqYBCkoiIiEgFFJJEREREKqCQJCIiIlIBY+0hp2Gr+5MakwZsCfgTH6gVsDvIryEOrev6o3VdP7Se64/Wdf3Ruq69LtbaQ86pFpSQVB+MMYnW2uEN3Y7fA63r+qN1XT+0nuuP1nX90boOPHW3iYiIiFRAIUlERESkAqEckl5u6Ab8jmhd1x+t6/qh9Vx/tK7rj9Z1gIXsmCQRERGRYArlSpKIiIhI0IRcSDLGjDfGrDXGbDDG3NPQ7WkMjDFJxpjlxpglxphE330tjTHfG2PW+65blFv+Xt/6X2uMOaXhWn74M8a8ZoxJNcasKHdfjdetMWaY72+0wRjzjDHG1Pd7OdxVsq4fNMZs923bS4wxp5b7ndZ1LRhjOhljZhhjVhtjVhpjbvXdr+06wKpY19qu64u1NmQuQDiwEegGRABLgb4N3a5QvwBJQKuD7nsUuMd3+x7g377bfX3rPRJI8P09whv6PRyuF+AYYCiwoi7rFpgPjAEM8A0woaHf2+F2qWRdPwjcUcGyWte1X8/tgKG+27HAOt/61HZdf+ta23U9XUKtkjQS2GCt3WStLQTeA85o4DY1VmcAb/puvwmcWe7+96y1BdbazcAGnL+LVMBaOwvYc9DdNVq3xph2QFNr7RzrfNq9Ve4x4lPJuq6M1nUtWWt3WmsX+W5nAauBDmi7Drgq1nVltK4DLNRCUgdgW7mfk6l6gxH/WGC6MWahMeZa331trLU7wflHBVr77tffoO5qum47+G4ffL/45yZjzDJfd1xpF5DWdQAYY7oCQ4B5aLsOqoPWNWi7rhehFpIq6kPV4Xl1N9ZaOxSYAPzRGHNMFcvqbxA8la1brfPaewE4AhgM7ASe8N2vdV1HxpgY4CPgNmttZlWLVnCf1nUNVLCutV3Xk1ALSclAp3I/dwR2NFBbGg1r7Q7fdSrwCU732S5fiRbfdapvcf0N6q6m6zbZd/vg+6Ua1tpd1toSa60XeIX9XcNa13VgjHHj7LTfsdZ+7Ltb23UQVLSutV3Xn1ALSQuAHsaYBGNMBHA+8HkDtymkGWOaGGNiS28DJwMrcNbrZb7FLgM+893+HDjfGBNpjEkAeuAMCBT/1Wjd+rousowxo31HpFxa7jFShdKdts9ZONs2aF3Xmm+9vAqsttY+We5X2q4DrLJ1re26HjX0yPGaXoBTcUb4bwTua+j2hPoF50jBpb7LytJ1CsQBPwLrfdctyz3mPt/6X4uOkKhu/U7FKYcX4Xybu6o26xYYjvNBuBF4Ft9EsLpUu66nAMuBZTg7kHZa13Vez0fhdNUsA5b4Lqdqu67Xda3tup4umnFbREREpAKh1t0mIiIiUi8UkkREREQqoJAkIiIiUgGFJBEREZEKKCSJiIiIVEAhSURERKQCCkkiIiIiFVBIEhEREanA/wOqcFXQXJwQ+gAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "train_scores = []\n",
    "test_scores = []\n",
    "Ys_pred = n.zeros(Ys.shape)\n",
    "for train_idxs,test_idxs in splitter.split(Xs,Ys):\n",
    "    \n",
    "    # Load scaled data\n",
    "    X_train = Xs[train_idxs]\n",
    "    Y_train = Ys[train_idxs]\n",
    "    \n",
    "    X_test = Xs[test_idxs]\n",
    "    Y_test = Ys[test_idxs]\n",
    "    \n",
    "    scaler = preprocessing.StandardScaler().fit(X_train)\n",
    "    X_train_scaled = scaler.transform(X_train)\n",
    "    X_test_scaled = scaler.transform(X_test)\n",
    "    \n",
    "    \n",
    "    # Set up the model\n",
    "    trackers = {'test_loss' : [], 't' : [], 'train_loss' : [], 'test_frac_correct': []}\n",
    "    best_model = None\n",
    "    def track():\n",
    "        for key in trackers.keys():\n",
    "            trackers[key].append(globals()[key])\n",
    "    layers = []\n",
    "    prev_dim = D_in\n",
    "    for dim in hidden_dims:\n",
    "        layers += [torch.nn.Linear(prev_dim, dim),\n",
    "                   torch.nn.ReLU()]\n",
    "        prev_dim = dim\n",
    "    layers += [torch.nn.Linear(prev_dim, D_out)]\n",
    "\n",
    "    loss_fn = torch.nn.CrossEntropyLoss()\n",
    "    model = nn.Sequential(*layers)\n",
    "    optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate)\n",
    "    t = 0\n",
    "    end = False\n",
    "    \n",
    "    # Convert data to tensors\n",
    "    features_train = torch.tensor(X_train).float(); targets_train = torch.tensor(Y_train);\n",
    "    features_test = torch.tensor(X_test).float(); targets_test = torch.tensor(Y_test);\n",
    "    for epoch_idx in range(n_epochs):\n",
    "        print(\"### EPOCH {:2d} ###\".format(epoch_idx))\n",
    "        # Figure out what batch to look at\n",
    "        indices = n.random.choice(range(targets_train.shape[0]), targets_train.shape[0], False)\n",
    "        num_batches = len(indices) // N_batch + 1\n",
    "        for batch_idx in range(num_batches):\n",
    "            # extract the batch\n",
    "            batch_train_x = features_train[indices[batch_idx*N_batch :(batch_idx+1)*N_batch]]\n",
    "            batch_train_y = targets_train[indices[batch_idx*N_batch : (batch_idx+1)*N_batch]]\n",
    "            # Forward pass: compute predicted y by passing x to the model.\n",
    "            batch_train_y_pred = model(batch_train_x)\n",
    "\n",
    "            # Compute and print loss.\n",
    "            loss = loss_fn(batch_train_y_pred, batch_train_y)\n",
    "            if batch_idx % print_interval == 0:\n",
    "                prediction_test = model(features_test)\n",
    "                train_loss = loss.item()\n",
    "                test_loss= loss_fn(prediction_test, targets_test).item()\n",
    "\n",
    "\n",
    "                pred_labels = n.argmax(prediction_test.detach().numpy(),axis=1)\n",
    "                true_labels = targets_test.detach().numpy()\n",
    "                correct_preds = n.array(pred_labels == true_labels, n.int)\n",
    "                test_frac_correct = n.mean(correct_preds)\n",
    "\n",
    "                if len(trackers['test_loss']) == 0 or test_loss < min(trackers['test_loss']):\n",
    "                    best_model = copy.deepcopy(model)\n",
    "                print(\"Batch {:3d}, Train Loss: {:5f}, Test Loss: {:5f}, Test Correct Frac: {:.3f}\".format(batch_idx, train_loss, test_loss, test_frac_correct))\n",
    "                track()\n",
    "\n",
    "            t += 1\n",
    "            optimizer.zero_grad()\n",
    "            loss.backward()\n",
    "            optimizer.step()\n",
    "        if end: break\n",
    "    \n",
    "    \n",
    "    train_score = 0\n",
    "    test_score = test_frac_correct\n",
    "    prediction_test = best_model(features_test)\n",
    "\n",
    "    pred = n.argmax(prediction_test.detach().numpy(),axis=1)\n",
    "\n",
    "#     true = targets_test.detach().numpy()\n",
    "    Ys_pred[test_idxs] = pred\n",
    "    \n",
    "    train_scores.append(train_score)\n",
    "    test_scores.append(test_score)\n",
    "    \n",
    "    plt.plot(trackers['t'],trackers['train_loss'], label='train')\n",
    "    plt.plot(trackers['t'],trackers['test_loss'], label='test')\n",
    "    plt.title(\"Cross Entropy Loss through training\")\n",
    "    plt.legend()\n",
    "    plt.show()\n",
    "    plt.plot(trackers['t'], trackers['test_frac_correct'])\n",
    "    plt.title(\"Fraction of correct labels on test set through training\")\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "f35f152f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.0, 1.0)"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 288x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(4,6))\n",
    "plt.bar([1,2],[n.mean(train_scores), n.mean(test_scores)], color=['blue','orange'])\n",
    "plt.scatter([1]*n_splits, train_scores, label='train')\n",
    "plt.scatter([2]*n_splits, test_scores, label='test')\n",
    "plt.legend()\n",
    "plt.ylim(0,1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "a7c282d6",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "de51c077",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(confusion_matrix((Ys),(Ys_pred), normalize='true'), vmin=0)\n",
    "# plt.xticks(unique_labels)\n",
    "plt.colorbar()\n",
    "plt.xlabel(\"Predicted label\")\n",
    "plt.ylabel(\"True Label\")\n",
    "plt.xticks(n.arange(len(unique_labels)),label_encoder.inverse_transform(n.arange(len(unique_labels))), rotation=90);\n",
    "plt.yticks(n.arange(len(unique_labels)),label_encoder.inverse_transform(n.arange(len(unique_labels))), rotation=0);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1b4b373a",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}