Base Model-checkpoint.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Base Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Importing Necessary Dependencies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Python Standard Libraries for importing data from binary file\n",
    "import os.path #for accessing the file path\n",
    "import struct  #for unpacking the binary data\n",
    "\n",
    "import time    #for calculating time\n",
    "\n",
    "#core packages\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "#custom module\n",
    "from dataPrep import load_dataset\n",
    "from dataPrep import visualize_dataset\n",
    "from dataPrep import prep_dataset\n",
    "\n",
    "\n",
    "np.random.seed(1)\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Dataset Preparation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Loading Complete Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Data\t\t\t Datatype\t\t Shape\n",
      "=================================================================\n",
      "Training Set Images:\t<class 'numpy.ndarray'>\t (60000, 28, 28)\n",
      "Training Set Labels:\t<class 'numpy.ndarray'>\t (60000, 1)\n",
      "Dev Set Images:\t\t<class 'numpy.ndarray'>\t (5000, 28, 28)\n",
      "Dev Set Labels:\t\t<class 'numpy.ndarray'>\t (5000, 1)\n",
      "Test Set Images:\t<class 'numpy.ndarray'>\t (5000, 28, 28)\n",
      "Test Set Labels:\t<class 'numpy.ndarray'>\t (5000, 1)\n",
      "=================================================================\n"
     ]
    }
   ],
   "source": [
    "#loading complete dataset\n",
    "train_x_orig, train_y_orig, dev_x_orig,dev_y_orig,test_x_orig,test_y_orig = load_dataset()\n",
    "\n",
    "print(\"Data\\t\\t\\t\",\"Datatype\\t\\t\",\"Shape\")\n",
    "print(\"=================================================================\")\n",
    "print(\"Training Set Images:\\t\" + str(type(train_x_orig))+\"\\t\",str(train_x_orig.shape))\n",
    "print(\"Training Set Labels:\\t\" + str(type(train_y_orig))+\"\\t\",str(train_y_orig.shape))\n",
    "print(\"Dev Set Images:\\t\\t\" + str(type(dev_x_orig))+\"\\t\",str(dev_x_orig.shape))\n",
    "print(\"Dev Set Labels:\\t\\t\" + str(type(dev_y_orig))+\"\\t\",str(dev_y_orig.shape))\n",
    "print(\"Test Set Images:\\t\" + str(type(test_x_orig))+\"\\t\",str(test_x_orig.shape))\n",
    "print(\"Test Set Labels:\\t\" + str(type(test_y_orig))+\"\\t\",str(test_y_orig.shape))\n",
    "print(\"=================================================================\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Preparing the Dataset (Flattening and Normalizing)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Data\t\t\t Before Processing\t After Processing\n",
      "=================================================================\n",
      "Training Set Images:\t(60000, 28, 28)\t\t(784, 60000)\n",
      "Training Set Labels:\t(60000, 1)\t\t(10, 60000)\n",
      "Dev Set Images:\t\t(5000, 28, 28)\t\t(784, 5000)\n",
      "Dev Set Labels:\t\t(5000, 1)\t\t(10, 5000)\n",
      "Test Set Images:\t(5000, 28, 28)\t\t(784, 5000)\n",
      "Test Set Labels:\t(5000, 1)\t\t(10, 5000)\n",
      "=================================================================\n"
     ]
    }
   ],
   "source": [
    "# Preparing the Dataset (Flattening and Normalizing)\n",
    "train_x_norm,train_y_encoded, dev_x_norm,dev_y_encoded, test_x_norm, test_y_encoded = prep_dataset(train_x_orig, train_y_orig, dev_x_orig,dev_y_orig,test_x_orig,test_y_orig)\n",
    "print(\"Data\\t\\t\\t\",\"Before Processing\\t\",\"After Processing\")\n",
    "print(\"=================================================================\")\n",
    "print(\"Training Set Images:\\t\" + str(train_x_orig.shape)+\"\\t\\t\"+ str(train_x_norm.shape))\n",
    "print(\"Training Set Labels:\\t\" + str(train_y_orig.shape)+\"\\t\\t\"+ str(train_y_encoded.shape))\n",
    "print(\"Dev Set Images:\\t\\t\" + str(dev_x_orig.shape)+\"\\t\\t\"+ str(dev_x_norm.shape))\n",
    "print(\"Dev Set Labels:\\t\\t\" + str(dev_y_orig.shape)+\"\\t\\t\"+ str(dev_y_encoded.shape))\n",
    "print(\"Test Set Images:\\t\" + str(test_x_orig.shape)+\"\\t\\t\"+ str(test_x_norm.shape))\n",
    "print(\"Test Set Labels:\\t\" + str(test_y_orig.shape)+\"\\t\\t\"+ str(test_y_encoded.shape))\n",
    "print(\"=================================================================\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Visualizing Datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x576 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "visualize_dataset(train_x_sample, train_y_sample, dataset = \"training\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x576 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "visualize_dataset(dev_x_orig, dev_y_orig, dataset = \"dev\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA54AAAHxCAYAAAAFsmNBAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nOzdd5zU5dX///fZZeldBBSQZkGNnRjbbYk1JkZNNLemWGKCeqtRo4nEmJjkTjHGXhK7qLGkKGoMGkv0NnaxF8SKoiggIiJNdvf8/pjB3369roHZmfnszlzzej4ePNg9XPv5XAPvHebMZ+asubsAAAAAAMhKQ2dvAAAAAACQNhpPAAAAAECmaDwBAAAAAJmi8QQAAAAAZIrGEwAAAACQKRpPAAAAAECmaDwBAAAAAJmi8QQAdCoz+7jNr1YzW9Lm82+VcdxHzOzbK/nzcWbmbc71npndamY7teMcR5jZ3WXs0czsVDObkd/DTDO7uiPODQBAR6LxBAB0KnfvveKXpLck7dWmdm3Gp29pc+7NJN0v6TYzOyDj864wQdLXJe2U38MX8nsAACApNJ4AgKpmZo1m9jMze93M3jeza82sf/7PepnZDWb2gZl9aGaPmtkAMztT0uclXZa/knjmqs7j7u+6+xmSfifpD23O/3Mze8PMFprZ82b25Xx9M0nnSNpxxRXTfH1fM3vGzD4yszfN7OSVnPbzkqa4+xv5Pcxy98vanHugmV2dvxo7M391tKHQuQEAqFY0ngCAavcjSbtJ2k7ScEnLJZ2d/7PvSeoiaZikQZKOlvSJu58g6XFJ38tf0TyhHee7SdJwMxud/3y6pG0k9ZP0e0k3mNkgd39K0nGS7sufY2h+/UeSvimpv6R9JZ1oZnsUONcjkg4zsx+a2eZm1viZP79W0gJJYyRtKWkfSd9ZybkBAKhKNJ4AgGp3uKSJ+auBSyX9UtJ/m5kp14SuLmmsuze7++PuvqjM883K/z5Qktz9L/mroa3ufo2kdyRtUeiL3f0ed38hv/5JSX+VtEOB5ZdLOlHSXpIekDTbzI6XJDMbKWl7ST9098Xu/q6k8yR11MuAAQComC6dvQEAAArJN5cjJE0xM2/zRw2SVlOucRsq6e9m1lvS1ZJ+5u4tZZx2WP73D/J7OEzSsZLWytd7K3d1tdCet5X0W0kbSOoqqZuka2Jr3d0lXSXpKjPrKmm//MdPSnJJ3SXNzf01SMrd7ldLvWEAAHQWrngCAKpWvjF7R9IX3b1/m1/d3f19d1/m7j9393HKXR3cX///FUEvdNxV2FfS2+7+hpmtK+l85YYADXT3/so1fis6wdg5/irpL5JGuHs/SZParF/Zbf3E3a9T7qW9n5M0U9LHkga0ud193X3zMm8fAAAdjsYTAFDtLpJ0mpmNkCQzG2xme+U/3sXMNjCzBuXeW9ksacXVztnKvTeyKGY2NP8y159Impgv95bUKmmupAYzO0LS2m2+bLakEWbWlD+G5b9mnrsvNbNtlGuGC53ze2a2h5n1zg8N+mr++I/lBw49Iul0M+uT//N1zGy72LkBAKhmNJ4AgGp3uqS7Jf3bzBZKekjSiqt+wyTdImmhpOclTVHuiqOUG0B0kJnNN7PTCxy7MT8VdpGkZyTtLGnvFT/GJf8ezYskTZX0rqTR+Y9XuEPSDElzzOzt/BXaIySdkd/rjyX9bSW3baGkUyW9LWm+pP+VdJi7P57/8wOVG1L0knIv/f2LpCGxc6/kHAAAdDrL/R8JAAAAAEA2uOIJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjWcnMrP7zOx7Hf21QEch40gZ+UbqyDhSR8Y7Fo1nBZjZDDPbpbP3UYiZHWJmLWb2cZtfO3b2vlA7qj3jkmRmY8zsNjNbaGbvm9npnb0n1IZqz7eZdTOzs81slpnNN7M/mllTZ+8LtaPaMy5JZna8mb1nZgvM7Aoz69bZe0LtqJGM1/3jFBrP+vGwu/du8+u+zt4QUClm1lXSXZL+LWmopOGS/typmwIqZ6Kk8ZI+J2ldSZtLOqVTdwRUkJntrlzOd5Y0StIYSb/szD0BlcTjlBwazwyZ2YD8Mxtz889S32Zmwz+zbKyZPZZ/hu8WMxvY5uu3MrOHzOxDM3uGq5SoNlWU8UMkzXL3s9x9kbsvdfdnSzwWIKmq8r2XpPPc/QN3nyvpPEnfLfFYwKeqKOMHS7rc3V9w9/mS/le5+3WgLFWU8UPE4xQaz4w1SLpS0khJa0laIumCz6w5SLkHEGtKalbuAYXMbJikf0r6taSBkk6UdKOZrf7Zk5jZWvlviLVWspfN8pf1Xzazn5lZl/JuGiCpejK+laQZZnZ7Puf3mdlGZd861Ltqybflf7X9fLiZ9SvxdgErVEvGN5T0TJvPn5E0xMxWK/F2AStUS8Z5nCIaz0y5+zx3v9HdF7v7Qkm/kbTDZ5Zd4+7Pu/siST+T9A0za5T0bUlT3H2Ku7e6+12SpkraM3Ket9y9v7u/VWAr9yv3Eq3Bkr4u6UBJP6rIjURdq6KMD5d0gHL/Wayp3H8Ut+Rf2gKUpIryfbukY81sdTMbKukH+XrPCtxM1LEqynhvSQvafL7i4z5l3DygmjLO4xTReGbKzHqa2cVm9qaZfaRcA9g/H+YVZrb5+E1JTZIGKffMzP75Z08+NLMPJW0naY327sPdX3f3N/LfNM9J+pWk/Uq9XcAK1ZJx5Z7BfMDdb3f3TySdIWk1SeuXcCxAUlXl+zeSnpL0tKSHJN0sabmkOSUcC/hUFWX8Y0l923y+4uOFJRwL+FQVZZzHKaLxzNoJktaT9AV37ytp+3y97UumRrT5eC3lHky8r9w3wTX5Z09W/Orl7qdVYF/+mT0ApaqWjD+rXK6BSqqKfLv7Enc/2t2HufsYSfMkPeHuLaXcKKCNqsi4pBckbdLm800kzXb3eSUcC2irWjLO4xTReFZSk5l1b/Ori3IvEVki6cP8G5VPjXzdt81sAzPrqdyVyL/nH0z8WdJeZra7mTXmj7lj5A3Rq2RmXzKzIfmPxyn3MoJbSrydqF9Vm/H8sbYys13yz2Iep9x/GtNKuaGoS1WbbzMbZmZrWs5Wyt2Hx/YCrEzVZlzS1ZIOy59ngHJTmyeVciNR16o54zxOEY1nJU1RLtgrfv1C0jmSeigXrEck3RH5umuUu3N9T1J35d+74+4zJe0t6WRJc5V71uVHivybWe4NzR9b4Tc07yzpWTNblN/nTZJ+W8JtRH2r2oy7+3Tl3otxkaT5+eN+Nf9yFqAYVZtvSWOVe4ntIklXSZro7neWcBtR36o24+5+h6TTJd2r3Esd3xRPrqD9qjnjPE6RZO51f9UXAAAAAJAhrngCAAAAADJF4wkAAAAAyBSNJwAAAAAgUzSeAAAAAIBMldV4mtkeZjbdzF41s4mV2hRQLcg4UkfGkTLyjdSRcdSSkqfa5n8GzcuSdpX0tqTHJR3o7i8W+pqu1s27q1dJ5wOKtVDz33f31cs9DhlHteqsjJNvdATuw5E6Mo7UFcp4lzKOuaWkV939dUkysxuU+5k0BcPeXb30Bdu5jFMCq3a3//3NCh2KjKMqdVbGyTc6AvfhSB0ZR+oKZbycl9oOU+4Hqa7wdr4GpIKMI3VkHCkj30gdGUdNKeeKp0Vqwet2zWyCpAmS1F09yzgd0OHIOFK3yoyTb9Qw7sOROjKOmlLOFc+3JY1o8/lwSbM+u8jdL3H38e4+vkndyjgd0OHIOFK3yoyTb9Qw7sOROjKOmlJO4/m4pHXMbLSZdZV0gKRbK7MtoCqQcaSOjCNl5BupI+OoKSW/1Nbdm83saEn/ktQo6Qp3f6FiOwM6GRlH6sg4Uka+kToyjlpTzns85e5TJE2p0F6AqkPGkToyjpSRb6SOjKOWlPNSWwAAAAAAVonGEwAAAACQKRpPAAAAAECmynqPJwAAAACk6r3jtwlqvz/q8ujaPXouC2pf3uar0bXNM94qb2M1iCueAAAAAIBM0XgCAAAAADJF4wkAAAAAyBSNJwAAAAAgUwwXAupYY9++Qe2Epx6Mrt25R0tQ23XaXtG1Tcf0CGotL77czt0BAAB0jIbu3aP19fd7Kajt3GNxdO1yr+iWksMVTwAAAABApmg8AQAAAACZovEEAAAAAGSKxhMAAAAAkCkaTwAAAABApphqC9Sx2QdsGNS2635PdO1yD5+num3cTdG12231g6A28MV2bg4AAKCDtG66brR+zagriz7GwTN2CY87d17Je0oNVzwBAAAAAJmi8QQAAAAAZIrGEwAAAACQKRpPAAAAAECmyhouZGYzJC2U1CKp2d3HV2JTqF2+9SZBbeHoHtG1fa97JOvtlI2Ml2beJh7UBnbCPrBqZBypq7eMN/bvV+APGoPS2wePiy495Yhrg9q+vT4oeg+zW5ZE6wced0JQ6/vQjOja5vdmF32+elZv+a6Uhp49g9rrx5d/3BnnrxfU+iyq/se7HaUSU213cvf3K3AcoFqRcaSOjCN1ZBwpI9+oCbzUFgAAAACQqXIbT5d0p5k9YWYTYgvMbIKZTTWzqcu1rMzTAR2OjCN1K804+UYCyDhSxuMU1IxyX2q7rbvPMrPBku4ys5fc/f62C9z9EkmXSFJfGxi+8QuobmQcqVtpxsk3EkDGkTIep6BmlHXF091n5X+fI2mypC0rsSmgWpBxpI6MI3VkHCkj36glJV/xNLNekhrcfWH+490k/apiO0OmugwfFtRaVu8fXTtrp3BC3r+OOz26tmfDw0Gtu8Vj9sFp4cs9Dllru+jazlAPGR9ywwtB7f9+HE56k6Sdeiwt+rhNQxeXvCd0nHrIeDl8m3BKtyStfe70oPbHYcVPLbxvSfw53227Lw9qh765c3Tt/AP7BLXmN2cWvYd6UY8Zn/a7cKqmJB27/Z1B7cj+/yr6uK3t2MOQxvg0+7vPvyCoHT/rv6Jr3/himPHWhQvbsYv01WO+K6VhyOpB7fntriz66+9c0ita7//EnKDWUvy2klfOS22HSJpsZiuOc52731GRXQHVgYwjdWQcqSPjSBn5Rk0pufF099clxZ8OBhJAxpE6Mo7UkXGkjHyj1vDjVAAAAAAAmaLxBAAAAABkqtwfp4ISfLL7+KD21u7xf4o1HgqnXvd5aUF07dytBwS1ZXvG1/5iw9uC2j69PoyujYsPoGmPwY3lHwPlmX3AhkFtpx73Rtc2yILagtZPomu7Tu1d3saArFiYY0mae/hWQW3SxLOjazds6hrUWgr8gILXmpcEtbWb4osbIverV438d3TtFzc6Iqh1Z7hQ3Wlce3RQm7zH+dG1G3YNH2e0Z2DQD97ZPlp/d0k4gPDGtf9Z9HHPXvM/0frx/w6HDr1y4mbRtQ3/91TR5wMkad42a5T19T+49ZBofe1Xih80V4+44gkAAAAAyBSNJwAAAAAgUzSeAAAAAIBM0XgCAAAAADJF4wkAAAAAyBRTbTP08f5fiNZvOyuclNi3oXt07ZJvhFND57Y0R9eu1aXzp8Q+siysLWrtFl17ybs7RKrvV3ZDWKnFQ8MJn60F5xyGz1P9/L2doyvX/MND5WwLyMyMX4fTayXpxUMujFTD6bWS9Nwny4Pa1277QXTt+me9F9Re+33f6NoXtr0qqL3bsji6tueMj4JaeyaUIg29J4U5WL9r/JrCJg8fHNSe2vrKos/1xvHrRev24NNBbc/tDouuffPLPYLaeftfEV0bm3Y79cr4/y3/c97RQW3oOfw/BGnxvvHH4v/6fWxqefw+/8aPBwW19S6dF13bUvTO6hNXPAEAAAAAmaLxBAAAAABkisYTAAAAAJApGk8AAAAAQKYYLpShdU94MVovNEgopoeFb3Req0v8zc/lig0GkqSTXt4vqH1y/ZDo2kG3hLe55cMFBc7IIKGO0mXYmtH6sQfc0sE7ATrOh9/ZOqg9e8h5BVY3BpXj340PpXj5yHFBbZ3HH42ujY+C26jAHkL9G+L/TS9fPRwmF94CpG7jvu8UvXbJ+8UPIDx0xm5BreGx+GMaj9QaHggHDknS6AfC2pl3fDO6ts+Vlwe18d3io1uWDortApBG/3hatN4z8vi6kLN/fUBQ6z/t4ZL3VM+44gkAAAAAyBSNJwAAAAAgUzSeAAAAAIBM0XgCAAAAADK1ysbTzK4wszlm9nyb2kAzu8vMXsn/PiDbbQLZIeNIHRlH6sg4Uka+kYpiptpOknSBpKvb1CZKusfdTzOzifnPT6r89mrIluGUwhOHXlJgcbeiD3vynM2D2iEDHir66/d74vvRes/b+ga1wXfPjK7tNfP1sKawJknxeXNVb5ISz/iijYdF64f2q62ptnP+Z5ug1rLb/Oja331ucrjWi3+Rx29/eVC03u/PjxR9jCoySQlnvHH11aP13/4ivA/uUmD264/fGx/UXj5ivehan/pcO3YXOm/zG4peG5tsLknLe4X/fdf5VNtJSjjjhfxn43BK/jW//kF07Rvf/VNQW17gPvHxB8LJzWOWZzPFs+H/norWf3DaUUHtPz8/N7r2noP+ENQOuSf+99B475Pt2F3VmKQ6zHd7zT0inGT+1xFnFVgd3rcuaF0aX/lxaznbQhurfBTm7vdL+uAz5b0lXZX/+CpJ+1R4X0CHIeNIHRlH6sg4Uka+kYpS3+M5xN3flaT874MrtyWgKpBxpI6MI3VkHCkj36g5xbzUtixmNkHSBEnqruJ/eDFQK8g4Uka+kToyjtSRcVSLUq94zjazNSQp//ucQgvd/RJ3H+/u45va8d5GoJORcaSuqIyTb9QwMo6U8TgFNafUK563SjpY0mn532trQkkZuqwxNFo/9rpwWMS4puK/uf+5uHe0/uzB4Rv8j3u26MNquF4oem1z8YetB3WR8YZ2PPfUIAtqsxb3K7A6/gb9mPmHhMMAvndS/K/70L7nF33c2H5b5UV//Zd+f2G0vtvBXwtqXXd9s+jjVpFkMr5oq9HR+o7dlxd9jMkPfz6orTP10ZL3tMIr530hqO3U44kCq8PMoizJZLwSlns4/m92y5Lo2uH3dv4jgkGXhMOMHjspHKYkSdt2D0drzT02ftuG3lvevqpI3ea7y4jh0fo5Pw4HaPUsMKAt5nuvh/+/S1LPyeX/X4CcYn6cyvWSHpa0npm9bWaHKRfyXc3sFUm75j8HahIZR+rIOFJHxpEy8o1UrPKKp7sfWOCPdq7wXoBOQcaROjKO1JFxpIx8IxWlvscTAAAAAICi0HgCAAAAADJF4wkAAAAAyFTmP8czNa8dGZ+euFvPcHpiS/EDNDV53hbR+nvbDQxqQ1+OT3VrXbYsLHo7NoG606rWdqwOn6d6+89joitX0+yijxqbYHtw3/iU2PbsdvKiQUHt1aXxqdQnrPZ80cf95dhwv78ftmd0bfM7s4o+LkrXfXZ8euWclsVBbXBj/GfYjd/01aC2aMyo6NrlQ8Jpzq98vym69uXd/xjUYhOXC5nfGr9tPSK3mXv7+jPrR9sEteO/Fh9uuqA1nDa+57k/jq5d446HyttYRm7+cPNofduh4dTRr41+Jrr2IRU/5RTVafqx8am2W3cLJzcX8uwn4drlB5GNrHHFEwAAAACQKRpPAAAAAECmaDwBAAAAAJmi8QQAAAAAZIrhQu3UvHZ80EOLt2fsSeiyEf8X/4OfRuo/jS9d56Yjw9ox4RvugVLMbgmz32dmc9Ff/9qZW0Xrh/a9MKi157vpsgXxAUdTvhIO7PIFH0XXNt4XnvG4gS9G18aGF8zec2R07WqXMlyoQzz2XLR80DePDmrXXh/mTZKuH31XWPxPWbvKK36QUMyTy/pH6/54/Dajvvz3d/4d1A7tNyO69uGlfYLaGmdW5xChQm6//fPR+h8ODR/rHDbgseja+/Y4Pqh1vePx8jaGzDSuE/4ff8befy77uL+Z+eWg1vzmzLKPG+NbbxKtz92iV1AbfEFtfU+2F1c8AQAAAACZovEEAAAAAGSKxhMAAAAAkCkaTwAAAABApmg8AQAAAACZYqptOzVN7xn/gx06dh8xj+xzVlDb/aUfRdcOvjDtqVmovG+/9O2g1q3AJEDbYsOgdtPXzylw5KZytqV/HLxjtO6vFz/1c+bSgWXtYfi3X4/Wl1xa1mFRpoYHng5q39k/nP4tSa8d0xjU9h73THTtfw8MJ2i+19wvunajrnOC2lpdCvw/EvHj874frQ8R9+GoP303nVf02tebe0fr3WYvDmpe8o6QtfnjBwe1L/dcUPTXf9y6LFpf+pV4vVyxCbYn//ma6Nqtu4d7+PNRI6Jr//adXcJzTX2+nbvrfFzxBAAAAABkisYTAAAAAJApGk8AAAAAQKZoPAEAAAAAmVrlcCEzu0LSVyTNcffP5Wu/kPR9SXPzy0529ylZbbKarPWrh6P1bd44KqgtHGnRtUuHtgS1SV+6OLp2226tRe9ttYYeQe2xk8+Prt37vm8GtZYXphd9rpTUc8Yb2vHc09vPDQ1qYzUjunbdi18Oaus3xYcINSj8PlnQ+kl07QEHHRPUGh9/Mrq2PRos/D4r9HcT2+/wnh9G175S3rYqpp4zHnjk2Wh57CNhrdDYhpeGfiWoNb83O7r25SsPDGqv7lb81Kke7xf/f0C9qod8xwa2SdIuva8OarNb4kNTfnX4EUGtSU+Ut7EOduYGf43WmywcDnb/x+Oia/2pFyq6p45QDxkvpLUp/li6WFtMPj5aX+ejcEhce7Rut2m0/rOrJwW1rbuFj/tzwscZB/V9J7pyxA1/CWpn77d/fG9Pv1jgfJ2vmEedkyTtEamf7e6b5n8lF3TUlUki40jbJJFxpGuSyDfSNklkHAlYZePp7vdL+qAD9gJ0CjKO1JFxpIx8I3VkHKko5z2eR5vZs2Z2hZkNKLTIzCaY2VQzm7pc2fzMHCAjZBypW2XGyTdqGPfhSB0ZR00ptfH8k6SxkjaV9K6kMwstdPdL3H28u49vUrcSTwd0ODKO1BWVcfKNGsV9OFJHxlFzSmo83X22u7e4e6ukSyVtWdltAZ2LjCN1ZBwpI99IHRlHLVrlVNsYM1vD3d/Nf7qvCg//S497tNz/6nDabf92HPa0db4Wrc/ZcUhQu+KnZ0fXbtjUNajFJnBK0rwtBga1/rU36C0z9ZLxVhU/MfOKfS8Kaj+/+/vRtd9e7Y+RcxUSPv/18/d2jq5svLf8CbYxrR7uofDfTWxteVP3OkO9ZDwLhSbYlmt+65KgNuD2+LTxQjMSkZNavj9au0+0vllk8v24f/wwunbdux+r6J46Q+y+WpKW+/Kg9uSHIwocZW6Bem1JLeOFjD38pbK+3lrL//+5oWfPoLbp+U9H1xaeYFue5QonN9uyMPfVrpgfp3K9pB0lDTKztyWdKmlHM9tUkkuaIenwDPcIZIqMI3VkHCkj30gdGUcqVtl4unv4Q8ikyzPYC9ApyDhSR8aRMvKN1JFxpKKcqbYAAAAAAKwSjScAAAAAIFMlDReqBQ3du0frr1y+flCbst0F0bXXf/j5oPbYPutE1za/8WY7dhcxf0G03GXJ4KC2XlP4BuP26rqw+KEyqH3d5yyO1p/5JKxtEs6okhR/w3zfiTOja+/9eIPwuANfLLzBz9i4V/y4bw7bJKg1vzOr6ON2GbZmtP65Xs8UfQygPXZc/+Wi18bulVvmz6/cZlAXer6VxkO7WT/aJqht0PXB6NoFkW+eOeeNia7tlchwoXox/YPVw+Ko4r++2/vFX2Nr6NUrWn/lknWD2q2DLyt+E+3wcWv856yecO0xQW3ktIcy2UOWuOIJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMhUGqPPIl67cr1offr2sZ+32yO69pRBzwe1a25/P7r299fvF9SG/3tJdO2CseHE3S2PejK69h9r/jFSLX6q7UYPHxStrzXl6aDmRR8VtcafeCFaP+D+w4PatF0uLvq4l425sR27KDAuN2JU1/j3mffrHRbfKX4HizYeFq0f2u+W4g8S8eB1m0frQ1V7E+dQWfsNeryzt4AEzF+/+OsEo66NTwVvrtRmKmzZl8KfICBJtx9zelDr19AtunbcLUcFtXX//mh5G0NV6HrdwLAY/y83qkt8qL9ad9gsqI36w0vRtbcOy2aC7TJfHtS2ufzE6NqRv0jj8QRXPAEAAAAAmaLxBAAAAABkisYTAAAAAJApGk8AAAAAQKaSHS7U+FKv+B9sX95xv9PnvWj9kMPDIUBvH/ZxdO0ajT3L20Q7DLomfi5ftqzD9oDqte55YQ5e3SE+gmJcUzjUYUBDOCirvRpkQW3nHvFpAD/5XfhcWdPkraNrB0x6OKjN2iF+l9fQjufg7lwS3rescf+C6FoGdtWPLsPWjNaHNcYGnDRluxkkZ9nAlmj9vPnjgpp/FH/sUQ26jBwR1LY6LT40ZfXG+CChmBF3lrwlJO6M/7k0Wu9q4ffUtt3DYT9ZWtgaPt4aNXl+dG1r1pvpIFzxBAAAAABkisYTAAAAAJApGoRfIAwAACAASURBVE8AAAAAQKZoPAEAAAAAmVpl42lmI8zsXjObZmYvmNmx+fpAM7vLzF7J/z4g++0ClUfGkToyjpSRb6SOjCMVxUy1bZZ0grs/aWZ9JD1hZndJOkTSPe5+mplNlDRR0knZbbV9Rl/0arS+2WbfCmqPf/7q6Nouaiz6fC0ezpvqyOm1krTds/sHtYFPvBNdG59bWrdqMuOV4E+8ENT2/vvx0bXTDrwwqLVWZM5a+PxXoeM+Mv6aoHblOqOia88a99WgdsbXr4qubc/tOOHa7wa1kU/EpzJWkbrNeEdZsv4a0fpGXcubYNvQPT45unXp0rKOm5jk8z30wXD6tyQd9/WXg9oN39gtunbQxeGk76w0rjYwWp/3p3BS7SmDni1wlPD/hv866ajoyn43P1L03mpU8hkvpM+MJUHt2U/iU5437ho+bt+pR/XeV27zr/Dx1vqzZnT8RjrQKq94uvu77v5k/uOFkqZJGiZpb0krHsVdJWmfrDYJZImMI3VkHCkj30gdGUcq2vUeTzMbJWkzSY9KGuLu70q5bwhJgyu9OaCjkXGkjowjZeQbqSPjqGVFN55m1lvSjZKOc/eP2vF1E8xsqplNXa7wh9UD1YKMI3WlZJx8o1ZwH47UkXHUuqIaTzNrUi7o17r7TfnybDNbI//na0iaE/tad7/E3ce7+/gmha/tB6oBGUfqSs04+UYt4D4cqSPjSMEqhwuZmUm6XNI0dz+rzR/dKulgSaflf78lkx2WqGV29HtPa+wT1vfZ5KDo2je+Fg4Hu+Ggs6Nrh3cJx/UMaOgRXftRa/hG51/O2T669pYnNgtqIydHl6rvHVODWrN7fDE+VasZz8rYE+NDGjZcfnRQO32/cNiPJH2p5/yK7mllDu03I1o/7DsXBLX2DBG6bMGYaH3sVbOCWrUP6yLjtWFAQzhIaNEeG0fX9rj5say3UzPqId9zN48PF4rdp33rmH9F1959zyZBzT5eHF3b/N7soPbBd7eOrl00LNzbhAOmRNce2T/cW6F75f+ZuVNQGzD5uejaSoy5q2b1kPFC7OFngtqJR8aHTF188TlBbXSX+IC2b70RDuF6/Nmx0bWjbwqHGb23dbyBX7JG+IhgtSfjw0rH/eX5oNaycGF0bSqKmWq7raTvSHrOzJ7O105WLuR/NbPDJL0lKRypCtQGMo7UkXGkjHwjdWQcSVhl4+nuD0iKP9Um7VzZ7QAdj4wjdWQcKSPfSB0ZRyraNdUWAAAAAID2ovEEAAAAAGSKxhMAAAAAkKlihgslr/WZadH6yHCQlk469QvRtY3rhBMw39lzaHTtmvd8EO7h+Zeia9fV49E60NFG/+ThoHbpBfFpzCcdOTKoff6L8e+zK0feU97GCnireUlQ2/PqH0XXjr0mMgV7fvxHpLXMnVHOtoCC5kcmnjO9FpK0+pMFJtR/OywdNWB6dOnwKfOC2hvLBkfXPvxB+Jjm2lFnRNeO7NI1qDUUuK7xRvMnQW3PB8KJ6ZI05sJwVq0tijwwQ93pekf8sfExI7dtx1HC74d1I7VCRtzdjlMVkPo05hiueAIAAAAAMkXjCQAAAADIFI0nAAAAACBTNJ4AAAAAgEwxXKhCWl55PagNPTesSfX5ZmKkqfmdWdH6qFPC+twCx/iKtqjgjlZulMIBSZLU0mE7AID263/3y9H6lQtGBbVD+82Irt23V2SwYa/3o2uPG/hiUDtv/sbRtfv1CQf+XPrBNtG1d58bDn8ZOyl+vwwgPVzxBAAAAABkisYTAAAAAJApGk8AAAAAQKZoPAEAAAAAmaLxBAAAAABkiqm2AABUgV/P3jFSXd7R20AVapkXTqSVpH/suklQm3z1ZtG1t427pejzbffUt4Ja662rRdf+cfOdg9r6E6dH1w74kAm2QD3jiicAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMjUKocLmdkISVdLGiqpVdIl7n6umf1C0vclzc0vPdndp2S1USArZBwpI98do+t9z0Tr6//nkKD2u81vjq69c8r4oDZSDGNZlXrOePM7s8JiOOtHkvQVbVH0cQfq5aLXDorUWor+ahSjnjOOtBQz1bZZ0gnu/qSZ9ZH0hJndlf+zs939jOy2B3QIMo6UkW+kjowjdWQcSVhl4+nu70p6N//xQjObJmlY1hsDOgoZR8rIN1JHxpE6Mo5UtOs9nmY2StJmkh7Nl442s2fN7AozG1DgayaY2VQzm7pcy8raLJA1Mo6UkW+kjowjdWQctazoxtPMeku6UdJx7v6RpD9JGitpU+WehTkz9nXufom7j3f38U3qVoEtA9kg40gZ+UbqyDhSR8ZR64pqPM2sSbmgX+vuN0mSu8929xZ3b5V0qaQts9smkC0yjpSRb6SOjCN1ZBwpKGaqrUm6XNI0dz+rTX2N/GvOJWlfSc9ns0UgW2QcKSPfHcObm6P10Qc8G9Qu0ZjoWibYloaMI3VkHKkoZqrttpK+I+k5M3s6XztZ0oFmtqkklzRD0uGZ7BDIHhlHysg3UkfGkToyjiQUM9X2AUkW+SN+ThCSQMaRMvKN1JFxpI6MIxXtmmoLAAAAAEB70XgCAAAAADJF4wkAAAAAyBSNJwAAAAAgUzSeAAAAAIBM0XgCAAAAADJF4wkAAAAAyBSNJwAAAAAgU+buHXcys7mS3sx/OkjS+x128o7FbetcI9199c44cZuM18LfU6m4bZ2vUzLOfXgSauG2VcN9uFQbf1el4rZ1LjKePW5b54pmvEMbz//nxGZT3X18p5w8Y9w2pPz3xG2DlPbfFbcNUtp/V9w2SGn/XXHbqhMvtQUAAAAAZIrGEwAAAACQqc5sPC/pxHNnjduGlP+euG2Q0v674rZBSvvvitsGKe2/K25bFeq093gCAAAAAOoDL7UFAAAAAGSqwxtPM9vDzKab2atmNrGjz19JZnaFmc0xs+fb1Aaa2V1m9kr+9wGducdSmdkIM7vXzKaZ2Qtmdmy+nsTtyxIZrw1kvHRkvDaQ8dKklG8p3YyT79KllPFU8y2lmfEObTzNrFHShZK+JGkDSQea2QYduYcKmyRpj8/UJkq6x93XkXRP/vNa1CzpBHdfX9JWko7K/1ulcvsyQcZrChkvARmvKWS8nRLMt5Ruxsl3CRLM+CSlmW8pwYx39BXPLSW96u6vu/snkm6QtHcH76Fi3P1+SR98pry3pKvyH18laZ8O3VSFuPu77v5k/uOFkqZJGqZEbl+GyHiNIOMlI+M1goyXJKl8S+lmnHyXLKmMp5pvKc2Md3TjOUzSzDafv52vpWSIu78r5QIjaXAn76dsZjZK0maSHlWCt6/CyHgNIuPtQsZrEBkvWj3kW0osA+S7Xeoh48llIJWMd3TjaZEaY3WrmJn1lnSjpOPc/aPO3k8NIOM1hoy3GxmvMWS8Xch3jSHf7UbGa0xKGe/oxvNtSSPafD5c0qwO3kPWZpvZGpKU/31OJ++nZGbWpFzQr3X3m/LlZG5fRsh4DSHjJSHjNYSMt1s95FtKJAPkuyT1kPFkMpBaxju68Xxc0jpmNtrMuko6QNKtHbyHrN0q6eD8xwdLuqUT91IyMzNJl0ua5u5ntfmjJG5fhsh4jSDjJSPjNYKMl6Qe8i0lkAHyXbJ6yHgSGUgx4+besVfXzWxPSedIapR0hbv/pkM3UEFmdr2kHSUNkjRb0qmSbpb0V0lrSXpL0v7u/tk3PVc9M9tO0n8kPSepNV8+WbnXltf87csSGa8NZLx0ZLw2kPHSpJRvKd2Mk+/SpZTxVPMtpZnxDm88AQAAAAD1paNfagsAAAAAqDM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFM0ngAAAACATNF4diIzu8/MvtfRXwt0FDKOlJFvpI6MI3VkvGPReFaAmc0ws106ex+FmFk3MzvbzGaZ2Xwz+6OZNXX2vlA7yDhSVgP5PsDMppvZAjObY2ZXmVnfzt4Xake1Z1ySzGyMmd1mZgvN7H0zO72z94TaUe0Z53FKDo1nfZgoabykz0laV9Lmkk7p1B0BlUXGkbIHJW3r7v0kjZHURdKvO3dLQOWYWVdJd0n6t6ShkoZL+nOnbgqoLB6niMYzU2Y2IP/s3dz8sxu3mdnwzywba2aP5Z/JvsXMBrb5+q3M7CEz+9DMnjGzHUvcyl6SznP3D9x9rqTzJH23xGMBnyLjSFm15NvdZ7r7+21KLZLWLuVYQFvVknFJh0ia5e5nufsid1/q7s+WeCzgU1WUcR6niMYzaw2SrpQ0UtJakpZIuuAzaw5SLnhrSmpWLogys2GS/qncs9oDJZ0o6UYzW/2zJzGztfLfEGsV2Iflf7X9fLiZ9SvxdgErkHGkrFryLTPbzswWSFoo6euSzinvpgGSqifjW0maYWa3519me5+ZbVT2rQOqJ+M8ThGNZ6bcfZ673+jui919oaTfSNrhM8uucffn3X2RpJ9J+oaZNUr6tqQp7j7F3Vvd/S5JUyXtGTnPW+7e393fKrCV2yUda2arm9lQST/I13tW4GaijpFxpKyK8i13fyD/Utvhkv4gaUZFbiTqWhVlfLikA5R7wL+mcg/2b8m/BBcoWRVlnMcpovHMlJn1NLOLzexNM/tI0v2S+ufDvMLMNh+/KalJ0iDlnpnZP//syYdm9qGk7SStUcJWfiPpKUlPS3pI0s2SlkuaU8KxgE+RcaSsivL9KXd/R9Idkm4o5ziAVFUZXyLpAXe/3d0/kXSGpNUkrV/CsYBPVVHGeZwiGs+snSBpPUlfcPe+krbP19teah/R5uO1lAvh+8p9E1yTf/Zkxa9e7n5aezfh7kvc/Wh3H+buYyTNk/SEu7eUcqOANsg4UlYV+Y7oImlsBY4DVEvGn5XkJXwdsCpVkXEep+TQeFZOk5l1b/Ori6Q+yj2L92H+jcqnRr7u22a2gZn1lPQrSX/Ph/DPkvYys93NrDF/zB0jb4heJTMbZmZrWs5Wyr2MILYXYGXIOFJWzfn+Vv79Q2ZmI5V75vyekm8p6lXVZjx/rK3MbJf8lajjlHvgP62UG4q6VbUZ53FKDo1n5UxRLtgrfv1CueEPPZS783xEuZdHfdY1kiZJek9Sd+Vf8+3uMyXtLelkSXOVe9blR4r8m+UfkHxshd/QPFa5y/qLJF0laaK731nCbUR9I+NIWTXnewPl8v2xcj9aZbqk77f/JqLOVW3G3X26cu+nu0jS/Pxxv5p/2S1QrKrNuHicIkkyd17ZAAAAAADIDlc8AQAAAACZovEEAAAAAGSKxhMAAAAAkCkaTwAAAABApspqPM1sDzObbmavmtnESm0KqBZkHKkj40gZ+UbqyDhqSclTbfM/Z+llSbtKelvS45IOdPcXC31NV+vm3dWrpPMBxVqo+e+7++rlHoeMo1p1VsbJNzoC9+FIHRlH6gplvEsZx9xS0qvu/rokmdkNyv2sm4Jh765e+oLtXMYpgVW72//+ZoUORcZRlTor4+QbHYH7cKSOjCN1hTJezktthyn3g1RXeDtfA1JBxpE6Mo6UkW+kjoyjppRzxdMiteB1u2Y2QdIESequnmWcDuhwZBypW2XGyTdqGPfhSB0ZR00p54rn25JGtPl8uKRZn13k7pe4+3h3H9+kbmWcDuhwZBypW2XGyTdqGPfhSB0ZR00pp/F8XNI6ZjbazLpKOkDSrZXZFlAVyDhSR8aRMvKN1JFx1JSSX2rr7s1mdrSkf0lqlHSFu79QsZ0BnYyMI3VkHCkj30gdGUetKec9nnL3KZKmVGgvQNUh40gdGUfKyDdSR8ZRS8p5qS0AAAAAAKtE4wkAAAAAyBSNJwAAAAAgU2W9xxMAAADpaxwyOFrf577niz7Gny7cJ6gNvuChkvcEoLZwxRMAAAAAkCkaTwAAAABApmg8AQAAAACZovEEAAAAAGSK4UIAAABYqekTx0TrB/f9Z9HHOPgn5wa1r17w+ZL3BKC2cMUTAAAAAJApGk8AAAAAQKZoPAEAAAAAmaLxBAAAAABkisYTAAAAAJApptoCAOpe46DVgtqHu6wTXbv2sS8GtZPW+Fd07d4P/k9QG/utp9q5O6DzvfSNC6P11g7eB4DaxRVPAAAAAECmaDwBAAAAAJmi8QQAAAAAZKqs93ia2QxJCyW1SGp29/GV2BRQLcg4UkfGkToyjpSRb9SSSgwX2snd36/AcYBqRcaRurrJeGPfvtF6z8kW1KaMvagdR+4RrU7f8fKgtv7vjoquHf2Th9txPrRT3WS8El67drNI9Ymyj/vzOZ8v+xiIIt+oCbzUFgAAAACQqXIbT5d0p5k9YWYTKrEhoMqQcaSOjCN1ZBwpI9+oGeW+1HZbd59lZoMl3WVmL7n7/W0X5L8JJkhSd/Us83RAhyPjSN1KM06+kQAyjpTxOAU1o6wrnu4+K//7HEmTJW0ZWXOJu4939/FN6lbO6YAOR8aRulVlnHyj1pFxpIzHKaglJTeeZtbLzPqs+FjSbpKer9TGgM5GxpE6Mo7UkXGkjHyj1pTzUtshkiab2YrjXOfud1RkV4lr6N49qC38yibRte98qSWovbrHJdG1u7y4b3iu0wZF13Z7fmZQa5k9J7q2jpHxEtkWGwa1vue+F117/ei7glqjxZ8Te/aTpUHt2MOPjq7t+q+pK9sicuou46/8NMymJP1t+DlBbfzPj4uu/aRfOAF34eeWRde+sXs41Xbklm+vbIuorLrLeHu8P2HraH3ajhcEtSZrjK5d7sWf79eDw8m4X9EWxR8An0W+UVNKbjzd/XVJ8W4JSAAZR+rIOFJHxpEy8o1aw49TAQAAAABkisYTAAAAAJApGk8AAAAAQKbK/TmeWImGXr2i9QF3dg1qN48K38hfSGuB5wvu2ODvYfHq+DE2+Hf4M4bX/g7DhdA+jeuvE60f9Zcbg9puPRZF1167cEhQ+83f9o+u/f0B1wS1w88LzyVJV643MlpHfWv6OBwMJEk//mZ4n7jaww8Xfdzmn24T/4Pdiz4E0OGW7v5RtN6q1qBWaIhQbG0h9y7pXfRaoL3eOz68H/7BhJuiaw/rFw48bPHis1zIOy2Lg9rujx4ZXTv83HBgV8MDT5e9h2rGFU8AAAAAQKZoPAEAAAAAmaLxBAAAAABkisYTAAAAAJApGk8AAAAAQKaYalshDZusH9RemxhOr5Wk50ZdlvV2VunJnf4Y1Lb89Q+ja0edUvxkR6Sr+YtbBLU/XHFhdO2IxnAy3EaXnRhdO+a8l4LaqHnxzP1wrW8EtSu2mxRdC8SM+N+HMjnukrWXFb12wdLu0fqASm0GiPBtNw1qk8fH78Ol+OOXmB2eOTCo9TgvnuYeby2IVF8u+lyAJL0yKXw8IknTdz0/qDUoPsm8JTKl+Y8fjo6uPfvfewS1fi+FE2kl6ZMvhhk/fIP/RNd+7brng9o3jz8hurbXjY9G67WGK54AAAAAgEzReAIAAAAAMkXjCQAAAADIFI0nAAAAACBTDBeqkOnH9wxq07a/qOzjHvjankHt3YvGRtfO++rioPbcf10eXdvdwn/6XXd/Mrr2lVNWtkOkpqFPn2h9t3PvD2obNsUHUKx92+FBbd1T40NdWtqxNwYJIQUffNg7Wme4ELL01rHhve3ILsUPESqkefLqQa3rHfEBce25vwcKuXnH+FCshnYMxfrbx6sFtdu/uF507Tqz2zHYJ7K1m/bZLbr0mAtfD2p3nHtudO1GexwT1Nb9/uPF76tKcMUTAAAAAJApGk8AAAAAQKZoPAEAAAAAmaLxBAAAAABkapWNp5ldYWZzzOz5NrWBZnaXmb2S/52ZCKhZZBypI+NIHRlHysg3UlHMVNtJki6QdHWb2kRJ97j7aWY2Mf/5SZXfXvVZuteW0foNO8QnbBXrKy99LVpv2u/joNb/k+eia+d9NT7ttlzL9vx8UOs+O5ygK0n+xAuZ7CFjk0TGP/XmsRtF6ycO/L+gtt59h0XXjjvxpaDWWt62JEkNFh6l1XnhRhEmiYxnasBq4X21JC3z5UFtyM3lTxJFYJLIuCTJCkwbH7na/KDW0I4Xvr26fFm0PujJj4KaF31UFGmSyPen/rN4nWh9w35vBrWXCuT29WWDw+KSpWXtS5KWfSl8zPzn884qsDr8iRg9LP79O3mXC4LaSfpCu/ZWDVZ5j+Pu90v64DPlvSVdlf/4Kkn7VHhfQIch40gdGUfqyDhSRr6RilIvFQxx93clKf975GkDoKaRcaSOjCN1ZBwpI9+oOcW81LYsZjZB0gRJ6h65pAzUOjKOlJFvpI6MI3VkHNWi1Cues81sDUnK/z6n0EJ3v8Tdx7v7+CZ1K/F0QIcj40hdURkn36hhZBwp43EKak6pVzxvlXSwpNPyv99SsR1VkYZevYLahj9/Nrp2k3bMipi+vCWo2SkDo2tb5s8Mah8duFV07XP/dX7xm2iHBaOagtpbe/SOrl3niUy20BnqIuONa48Oapd+N3wDuyS1RsZFjD0/zLIktS5cWNa+Fu8bf8P8+K6PBbXHlnUv61x1rC4ynoWPvxHeBz+6RXzA3FvN4WCL3n97tOJ7QlRdZnzmieOj9SfHnRvU2jP07bJ520XrNTpUMAV1mW9JuujKvaL1I44LH7+M7GLRtfOWh4/xl0/uF13btG9Ya/koHKolSe/sGLZWwxrjV5ljg4+O/9aR0bUzjg4fg43WM9G11ayYH6dyvaSHJa1nZm+b2WHKhXxXM3tF0q75z4GaRMaROjKO1JFxpIx8IxWrvOLp7gcW+KOdK7wXoFOQcaSOjCN1ZBwpI99IBT8ADwAAAACQKRpPAAAAAECmaDwBAAAAAJnK/Od41rKXf71RULt5zeInxy5s/SRaP+aY44Na94fDaZ2SNPOn2wS1Ow8/vcAZO25E9riL5kfr8RmnqFavTBga1LbsFk5Ok6RtngrfYjLwkfiU5xjrEr+7mX3ElkHtpGOuj65tssaizwdkZd7nwimJhbJ58LSDglovvV7xPQErnPe9izM57gPnxKeN99fDmZwPKGTELbOj9b99b7Wgtn/vedG1vbuEE2WvXfeG6NovXXtoUOt6/QbRtbt/8cloPead5r5BzR58Orp29INFH7aqccUTAAAAAJApGk8AAAAAQKZoPAEAAAAAmaLxBAAAAABkiuFCK3H3186IVIsf4LPY40Naun0QDh169aytomtv//ofgtqQxh7Rta1qLXpvMdM/GhytD30wHCTkM98t61yoDoM3jr9BP6br1QPLOldsiJAkPfaTcGBXg8LhLZLKTDhQGSO3nRnU5rQsiq6d+8SQoMZwIWRp++7xwYbl3n/2v6b4IUKvnRl/THPWV68Oao0W39kPHgkH2o37eXywYfPrM4reG2pfy8uvRetX/feXwrV/+Vd07e1nbR/Utv7Zq9G1P1z3nqA296d9omuP6V/8/fsRU74b1NbRo0V/fS3iiicAAAAAIFM0ngAAAACATNF4AgAAAAAyReMJAAAAAMgUjScAAAAAIFNMtc3QkMb4BNx//PXSdhyla2U2U4R5fx8era/+TPGT7FBbfrfuTUGt0ETZ9vhk9/FB7aoTz4qubWhHxiuxNyDGmsIcvn1CmGNJumXt04PaVQs2ia4dc9rzQY3pzMhS4fvJ4q81rPvPI4Laek3PRNf6HeFE/GnjLiz6XIX2+9JOlwW1PS/dJ7q2cY/w+9eXx6f7Il2tT78Y1K7bIX4/3nXb8J541x5LChw5Vp9b9L7uWhL/aRTrTXwuqKX+/wNXPAEAAAAAmaLxBAAAAABkisYTAAAAAJCpVTaeZnaFmc0xs+fb1H5hZu+Y2dP5X3tmu00gO2QcqSPjSBn5RurIOFJRzHChSZIukHT1Z+pnu/sZFd9RFfnte7sHtQuG39fxG0HWJqlOM/7c0hFBbetuM6JrF4wJn6faYmpTdO2Jg88Jake89o3o2iW/HxbUZh60PLp22g6XR+tYpUmq04wX6/VfbRHUXj74jwVW9w4qPxr4WnTlmlM/DGoXz9g+uvbdp4YGtXXOjh+3ZfacAnurS5NEvj/VKi9Qb8fYksghXjl98+jSF8edHzlXe8SvgcT2e9u4cCCeJG1xwrFBbdhpD7VrF1Vuksh4SZrfmx2t97l9YVD7x+K+0bV79fyorD2c+eZu0XrD4pllHbcWrfKKp7vfL+mDDtgL0CnIOFJHxpEy8o3UkXGkopz3eB5tZs/mL/8PqNiOgOpBxpE6Mo6UkW+kjoyjppTaeP5J0lhJm0p6V9KZhRaa2QQzm2pmU5drWYmnAzocGUfqiso4+UaN4j4cqSPjqDklNZ7uPtvdW9y9VdKlkrZcydpL3H28u49vUrdS9wl0KDKO1BWbcfKNWsR9OFJHxlGLSmo8zWyNNp/uK+n5QmuBWkTGkToyjpSRb6SOjKMWrXKqrZldL2lHSYPM7G1Jp0ra0cw2VW7u2QxJh2e4x07z1hcWBbUdv3lMdO0Wxz8V1A4d9J/o2pc+WSOonXLv16Nrh416P6g9sHF8qtvy+CC7qIeXNQa11f/0cPEHSEg9Z/zMu74c1Cbsd2F07VPHhJMLrs311wAADW1JREFUG2TRtZs//v2gNvyHS6Jru77+eFAb3G+r6FrtEJbea+4XX4tP1XPGi9X7rbD24NL4bM5tuxf/nO23+swLaxtNji/eKCxtuM63oktHHhHe4bfMnVv0vlJCvivv4p0nBbWdeiyNrm3PBNvLFowJahP6zWjHEeIWjY5PQk8FGa+81sWLg9qpFx4UXbvdCX8IagMaehR9rn5d449/lowZFdSaX59R9HFr0SobT3c/MFLmZxogGWQcqSPjSBn5RurIOFJRzlRbAAAAAABWicYTAAAAAJApGk8AAAAAQKZW+R5P/L/6XvdItP7KdWHtlI3jb1JuffaloLauHouufevUbYLa8o1a4sdtx1v8D3vksKA2Rk8X/fVIw7hfvRzU1u0Tn0/Qs1/45nh7PD7YZ8Q5Twa15qXxwRTl+sm9+0frhb6ngJjYcLXf3rFvdO2SsYOKPm5r1/D53UPOuiW69pC+c4LaC1tfG127xdeODGqDLq7P4UKovB16hINX2nOtIjZESJL+uU/4Ez8m3Duj6OMCWVrz0mei9buPHB7U9u8dDo6TpI9aw8c67y/pHV3bdUj4GMpeX9kOax9XPAEAAAAAmaLxBPD/tXP3sXWWZRzHf9dK3cBNNt7n2BhunWMEw6ASwjARiYoEGIaMwB8ElIiJKIzgyyRG/QfFRGAJ4SXDIY1BFDOEJRh11oEQGLKN19kA432sdkPQbQw22l7+0UPS8dzdnvP03M/pc5/vJ1naXr3Pc6779NeTXT3tDQAAAETF4AkAAAAAiIrBEwAAAAAQFYMnAAAAACAqTrWNKHR67Zjx+v7N7gBjwMB/3s7U5nw9W6tX/vOV6zNOlqmN7+NpDHH0v/JasN4+Qj2vWw49L1i/5Oe3juq6gBR+nvzwM2Vd91cvLgiuPK5r86iuO5JxO9tyrwVG0t85J1hfNPHh3Ne4+Z0TMrXxX3q1aEvJ4RVPAAAAAEBUDJ4AAAAAgKgYPAEAAAAAUTF4AgAAAACi4lSOFtVx/cZMbaAJfQAhWxe+H6z3DuzM1Gau/F9wrTe0IwCohsERnv0GR33sW/i1itB113T+JvdVB+u47lWbPxdcO+fHGwK3B+qz7agJudfu8F3B+t9+mM3oeD1RuKfU8IonAAAAACAqBk8AAAAAQFQMngAAAACAqBg8AQAAAABR7XPwNLPpZrbazHrMbIOZXVmrH2Rmq8zsxdrbKfHbBRqPjCN1ZBwpI99IHRlHKvKcatsv6Wp3X29mkyStM7NVki6R1O3u15nZEklLJP0gXqutadZpr8S58K7waVwtiow30bhJkzK1FafcFly7/J2TMjVflz3NEBlkvFnMMqX/fjr/zXcMhk94nrxxd9GOUkS+h5l7z+XB+r/Ov6nkTkbnyV3Z10ZWP3BCcO2M7Y/GbqfZyHiDjZuQPcH2rfn5z8O/e9vsYH38A5xguzf7fMXT3XvdfX3t/e2SeiRNk7RQUldtWZekc2M1CcRExpE6Mo6UkW+kjowjFXX9jaeZzZQ0X9Ljkg53915p6BtC0mGNbg4oGxlH6sg4Uka+kToyjirLPXia2URJKyQtdvdtddzuMjNba2ZrPxC/3omxi4wjdUUyTr5RFTyHI3VkHFWXa/A0s3YNBf0ud7+3Vu4zs6m1z0+VtCV0W3df5u6d7t7ZrvGN6BloODKO1BXNOPlGFfAcjtSRcaRgn4cLmZlJWi6px91vGPaplZIulnRd7e39UTpscRdO/eeobv/F584P1g/Y+caorpsSMt5cvV87LlM7tv2h4No/xm4mUWQ8vraDDwrWX1qcPUnohUtuDa7d1L8jU1u05LvBtZ/oXlNHd2kj33vq+P76YH2evpOplX3g0LEPXpap+ZbwIDR36ZuZ2ozXkj9EKIiMN97uBcdmai+cf0vu27/Vnz0YEfuW51TbBZIukvSsmT1Vq12joZDfY2aXSnpd0qI4LQLRkXGkjowjZeQbqSPjSMI+B093f0RS9jz4Iac3th2gfGQcqSPjSBn5RurIOFJR16m2AAAAAADUi8ETAAAAABAVgycAAAAAIKo8hwuhBNsvODlYX/jx0MmF7bmvu6lvSrDe0f9K7msAZWuz8M/Eup7Ofp90KHyCIxBL2+QDM7WXr5gbXPv8pflPSTzvR9/L1Cb/9rH8jQGS/IPdwfrsq7L/nzjnqs/GbmcPs/Rk7rX9EftA69hv5oxg/chrnx/Vdbv+fFqw/inxnL03vOIJAAAAAIiKwRMAAAAAEBWDJwAAAAAgKgZPAAAAAEBUHC40Ruz4ZPhnAO3WNqrrzr5tYFS3B2KbcvabmdqADwbXzr1uR3ZtwzsChvSffmKwfs5Nf83ULp/8UHDtjsH3M7XPrFgcXDvn99mDsnxvDQIA9urdYw4P1pdNvy/3NTYEDuzqWPbv4Fr+T7J3vOIJAAAAAIiKwRMAAAAAEBWDJwAAAAAgKgZPAAAAAEBUDJ4AAAAAgKg41XaMGJjQ7A6AuPq/ED4h9Bezb8vUut87IHyRre80siVgr/brXhesr5x3cLambG0kHVoTrHOCLQCMPZv7D8wWt79bfiMJ4BVPAAAAAEBUDJ4AAAAAgKgYPAEAAAAAUe1z8DSz6Wa22sx6zGyDmV1Zq//UzN40s6dq/86M3y7QeGQcKSPfSB0ZR+rIOFKR53ChfklXu/t6M5skaZ2Zrap97kZ3/2W89lrHtAfDf6R86VlfzNR+fVR37HZaDRkvwX5/Dx/U8q2fXZGpnfiNp2K300rIN1JHxpE6Ml5Q76mjP0f19P13Zmo3Hn1EeHHfllHfX8r2+dVw915JvbX3t5tZj6RpsRsDykLGkTLyjdSRcaSOjCMVdf2Np5nNlDRf0uO10rfN7Bkzu8PMpjS4N6B0ZBwpI99IHRlH6sg4qiz34GlmEyWtkLTY3bdJulXSLEnHa+inMNePcLvLzGytma39QLsa0DIQBxlHysg3UkfGkToyjqrLNXiaWbuGgn6Xu98rSe7e5+4D7j4o6XZJJ4Vu6+7L3L3T3TvbNb5RfQMNRcaRMvKN1JFxpI6MIwV5TrU1Scsl9bj7DcPqU4ct+6qk5xrfHhAfGUfKyDdSR8aROjKOVOQ56mmBpIskPWtmHx41eY2kC83seEku6VVJ34zSYYuwR58O1reekq2dpRPzX1fh62IPZLyJDr79sUzt1dtHWv1e1F4SRb6ROjKO1JHxgiZstWB90UtfztT+MOsvwbULXzg7W1zzzKj6alV5TrV9RFLoq/anxrcDlI+MI2XkG6kj40gdGUcq6jrVFgAAAACAejF4AgAAAACiYvAEAAAAAESV53AhAAAAAKiUI5Y+Gqy/uzRbO1MnjHCVzY1rqMXxiicAAAAAICoGTwAAAABAVAyeAAAAAICoGDwBAAAAAFExeAIAAAAAojJ3L+/OzLZKeq324SGS3irtzsvF3prrKHc/tBl3PCzjVXicimJvzdeUjPMcnoQq7G0sPIdL1XisimJvzUXG42NvzRXMeKmD5x53bLbW3TubcueRsTek/DixN0hpP1bsDVLajxV7g5T2Y8XexiZ+1RYAAAAAEBWDJwAAAAAgqmYOnsuaeN+xsTek/DixN0hpP1bsDVLajxV7g5T2Y8XexqCm/Y0nAAAAAKA18Ku2AAAAAICoSh88zewMM3vezDaa2ZKy77+RzOwOM9tiZs8Nqx1kZqvM7MXa2ynN7LEoM5tuZqvNrMfMNpjZlbV6EvuLiYxXAxkvjoxXAxkvJqV8S+lmnHwXl1LGU823lGbGSx08zaxN0s2SviJpnqQLzWxemT002J2SzvhIbYmkbnfvkNRd+7iK+iVd7e7HSDpZ0uW1r1Uq+4uCjFcKGS+AjFcKGa9TgvmW0s04+S4gwYzfqTTzLSWY8bJf8TxJ0kZ3f9ndd0v6naSFJffQMO7+D0lvf6S8UFJX7f0uSeeW2lSDuHuvu6+vvb9dUo+kaUpkfxGR8Yog44WR8Yog44UklW8p3YyT78KSyniq+ZbSzHjZg+c0SW8M+3hTrZaSw929VxoKjKTDmtzPqJnZTEnzJT2uBPfXYGS8gsh4Xch4BZHx3Foh31JiGSDfdWmFjCeXgVQyXvbgaYEax+qOYWY2UdIKSYvdfVuz+6kAMl4xZLxuZLxiyHhdyHfFkO+6kfGKSSnjZQ+emyRNH/bxkZI2l9xDbH1mNlWSam+3NLmfwsysXUNBv8vd762Vk9lfJGS8Qsh4IWS8Qsh43Voh31IiGSDfhbRCxpPJQGoZL3vwfEJSh5kdbWYfk3SBpJUl9xDbSkkX196/WNL9TeylMDMzScsl9bj7DcM+lcT+IiLjFUHGCyPjFUHGC2mFfEsJZIB8F9YKGU8iAylm3NzLfXXdzM6UtFRSm6Q73P3aUhtoIDO7W9LnJR0iqU/STyTdJ+keSTMkvS5pkbt/9I+exzwzO1XSw5KelTRYK1+jod8tr/z+YiLj1UDGiyPj1UDGi0kp31K6GSffxaWU8VTzLaWZ8dIHTwAAAABAayn7V20BAAAAAC2GwRMAAAAAEBWDJwAAAAAgKgZPAAAAAEBUDJ4AAAAAgKgYPAEAAAAAUTF4AgAAAACiYvAEAAAAAET1f4nDbARm5O9iAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 1152x576 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "visualize_dataset(test_x_orig, test_y_orig, dataset = \"test\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Utility Functions"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### ReLU Function and Its derivative"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def relu(Z):\n",
    "    \"\"\"\n",
    "        Compute the ReLU activation of Z\n",
    "        \n",
    "        Argument:\n",
    "            - Z -- Array of the Sum of the product of Weights and input\n",
    "        \n",
    "        Returns:\n",
    "            - A -- Array of Activation obtained by applying ReLU function. same size as that of Z\n",
    "    \"\"\"\n",
    "    A = np.maximum(0.0,Z)\n",
    "    \n",
    "    cache = Z\n",
    "    assert(A.shape == Z.shape)\n",
    "    return A, cache"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def relu_grad(dA, cache):\n",
    "    \"\"\"\n",
    "        Compute the gradient of dA\n",
    "        \n",
    "        Arguments:\n",
    "            - dA -- Array of the gradient of activation of the previous layer\n",
    "            - cache -- list of other useful variables like Z\n",
    "            \n",
    "        Returns:\n",
    "            - dZ -- array of gradient/derivative of the dA, Same size of dA\n",
    "    \"\"\"\n",
    "    Z = cache\n",
    "    dZ = np.array(dA, copy=True) # just converting dz to a correct object.\n",
    "    \n",
    "    dZ[Z < 0] = 0\n",
    "    \n",
    "    assert(dZ.shape == Z.shape)\n",
    "    return dZ"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Softmax Function and its derivative"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def softmax(Z):\n",
    "    \"\"\"\n",
    "        Compute the softmax activtion of Z\n",
    "        \n",
    "        Argument:\n",
    "            - Z -- Array of the Sum of the product of Weights and input\n",
    "        \n",
    "        Returns:\n",
    "            - A -- Array of Activation obtained by applying Softmax function. same size as that of Z\n",
    "    \"\"\"\n",
    "    shift = Z - np.max(Z) #Avoiding underflow or overflow errors due to floating point instability in softmax\n",
    "    t = np.exp(shift)\n",
    "    A = np.divide(t,np.sum(t,axis = 0))\n",
    "    \n",
    "    cache = Z\n",
    "    assert(A.shape == Z.shape)\n",
    "    return A, cache"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Deep Learning Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1. Creating NN Architecture"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### initializing layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "def init_layers():\n",
    "    \"\"\"\n",
    "        Initializes the layers of NN with certain neural units in each layers\n",
    "        \n",
    "        Returns:\n",
    "            - layer_dim -- list of the units of each layer of the network\n",
    "    \"\"\"\n",
    "    layers_dim = [784,32,16,10]\n",
    "    return layers_dim"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Initializing Parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "def init_params(layers_dim):\n",
    "    \"\"\"\n",
    "        Initializes the parameters of the Neural Network\n",
    "        \n",
    "        Argument:\n",
    "            - layers_dim -- Layer Dimensions of the NN\n",
    "        \n",
    "        Returns:\n",
    "            - params -- Dictionary of the initialized parameters: Weights and biases\n",
    "    \"\"\"\n",
    "    L = len(layers_dim)\n",
    "    params = {}\n",
    "        \n",
    "    for l in range(1,L):\n",
    "        params['W' + str(l)] = np.random.randn(layers_dim[l],layers_dim[l-1]) *0.01\n",
    "        params['b' + str(l)] = np.zeros((layers_dim[l],1))\n",
    "     \n",
    "        assert(params['W' + str(l)].shape == (layers_dim[l],layers_dim[l-1]))\n",
    "        assert(params['b' + str(l)].shape == (layers_dim[l],1))\n",
    "    return params"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Initializing Hyper Parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "def init_hyperParams(alpha = 0.01, num_iteration = 5000):\n",
    "    \"\"\"\n",
    "        Initializes the hyper parameters\n",
    "        \n",
    "        Arguments:\n",
    "            - alpha -- learning rate\n",
    "            - num_iteration -- number of iteration the gradient descent will run\n",
    "        Returns:\n",
    "            - Dictionary of hyper parameters\n",
    "        \n",
    "    \"\"\"\n",
    "    hyperParams = {}\n",
    "    hyperParams['learning_rate'] = alpha\n",
    "    hyperParams['num_iterations'] = num_iteration\n",
    "    \n",
    "    \n",
    "    return hyperParams"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2. Forward Propogation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Calculating sum of product of inputs and weights (Z) for individual layer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "def forward_sum(A,W,b):\n",
    "    \"\"\"\n",
    "        Calculates the forward sum \n",
    "        \n",
    "        Arguments:\n",
    "            - A -- array of activation from the previous layer\n",
    "            - W -- weights of the current layer\n",
    "            - b -- bias of the current layer\n",
    "    \"\"\"\n",
    "    \n",
    "    Z = np.dot(W,A) + b\n",
    "    \n",
    "    cache = (A,W,b)\n",
    "    assert(Z.shape == (W.shape[0],Z.shape[1]))\n",
    "    \n",
    "    return Z, cache"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Calculating Activation for individual Layer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "def forward_activation(A,W,b,activation):\n",
    "    \n",
    "    if activation == 'relu':\n",
    "        Z, sum_cache = forward_sum(A,W,b)\n",
    "        A, activation_cache = relu(Z)\n",
    "        \n",
    "    if activation == 'softmax':\n",
    "        Z, sum_cache = forward_sum(A,W,b)\n",
    "        A, activation_cache = softmax(Z)\n",
    "    \n",
    "    cache = (sum_cache,activation_cache)\n",
    "    assert(A.shape == Z.shape)\n",
    "    \n",
    "    return A, cache"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Complete Forward Propagation for L layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "def forward_prop(X,parameters):\n",
    "    caches = []\n",
    "    A = X\n",
    "    L = len(parameters) // 2\n",
    "    for l in range(1, L):\n",
    "        A_prev = A \n",
    "        A, cache = forward_activation(A_prev,parameters['W' + str(l)],parameters['b' + str(l)],activation='relu')\n",
    "        caches.append(cache)\n",
    "\n",
    "    AL, cache = forward_activation(A,parameters['W' + str(L)],parameters['b' + str(L)],activation='softmax')\n",
    "    caches.append(cache)\n",
    "    \n",
    "    assert(AL.shape == (10,X.shape[1]))\n",
    "    \n",
    "    return AL,caches\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3. Cost Function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "def compute_cost(AL,Y):\n",
    "    m = Y.shape[1]\n",
    "\n",
    "    cost = -(1./m) * np.sum(np.sum(np.multiply(Y,np.log(AL)), axis = 0,keepdims=True))\n",
    "    \n",
    "    cost = np.squeeze(cost)      # Making sure your cost's shape is not returned as ndarray\n",
    "    assert(cost.shape == ())\n",
    "    \n",
    "    return cost"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4. Backward Propagation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Calculating Gradients for individual Layer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "def backward_grad(dZ, cache):\n",
    "    A_prev, W, b = cache\n",
    "    m = A_prev.shape[1]\n",
    "    \n",
    "    dW = (1/m) * np.dot(dZ,A_prev.T)\n",
    "    db = (1/m) * np.sum(dZ, axis = 1, keepdims=True )\n",
    "    dA_prev = np.dot(W.T, dZ)\n",
    "\n",
    "    \n",
    "    assert (dA_prev.shape == A_prev.shape)\n",
    "    assert (dW.shape == W.shape)\n",
    "    assert (db.shape == b.shape)\n",
    "    \n",
    "    \n",
    "    return dA_prev, dW, db"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Calculating Backward Activation for individual layer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "def backward_activation(dA,cache,activation):\n",
    "    sum_cache, activation_cache = cache\n",
    "    \n",
    "    if activation == \"relu\":\n",
    "        dZ = relu_grad(dA,activation_cache)\n",
    "        dA_prev, dW, db = backward_grad(dZ, sum_cache)\n",
    "        \n",
    "    elif activation == \"softmax\":\n",
    "        dZ = dA\n",
    "        dA_prev, dW, db = backward_grad(dA, sum_cache)\n",
    "    \n",
    "    return dA_prev, dW, db\n",
    "    \n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Complete Backward Propagation for L layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "def backward_prop(AL, Y,caches):\n",
    "    grads = {}\n",
    "    L = len(caches) # the number of layers\n",
    "    m = AL.shape[1]\n",
    "    Y = Y.reshape(AL.shape) # after this line, Y is the same shape as AL\n",
    "    \n",
    "    dA = np.subtract(AL,Y)\n",
    "    \n",
    "    current_cache = caches[L-1]\n",
    "    grads[\"dA\" + str(L-1)], grads[\"dW\" + str(L)], grads[\"db\" + str(L)] = backward_activation(dA, current_cache, activation = 'softmax')\n",
    "    \n",
    "    for l in reversed(range(L-1)):\n",
    "        current_cache = caches[l]\n",
    "        dA_prev_temp, dW_temp, db_temp = backward_activation(grads[\"dA\" + str(l + 1)], current_cache, activation = 'relu')\n",
    "        grads[\"dA\" + str(l)] = dA_prev_temp\n",
    "        grads[\"dW\" + str(l + 1)] = dW_temp\n",
    "        grads[\"db\" + str(l + 1)] = db_temp\n",
    "\n",
    "    return grads"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5. Update Parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "def update_parameters(parameters, grads, learning_rate):\n",
    "    L = len(parameters) // 2\n",
    "    for l in range(L):\n",
    "        parameters[\"W\" + str(l+1)] = parameters[\"W\" + str(l+1)] - (learning_rate * grads[\"dW\" + str(l+1)])\n",
    "        parameters[\"b\" + str(l+1)] = parameters[\"b\" + str(l+1)] - (learning_rate * grads[\"db\" + str(l+1)])\n",
    "    return parameters"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6.  Prediction "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "def predict(X,y,parameters):\n",
    "    m = y.shape[1]\n",
    "    n = len(parameters) // 2 # number of layers in the neural network\n",
    "    \n",
    "    probas, caches = forward_prop(X, parameters)\n",
    "   \n",
    "    assert(probas.shape == y.shape)\n",
    "        \n",
    "    predicted_labels = np.argmax(probas,axis=0).reshape(1,probas.shape[1])\n",
    "    predicted_prob = np.max(probas,axis = 0).reshape(1,m)\n",
    "    \n",
    "    Y = np.argmax(y,axis=0).reshape(1,y.shape[1])\n",
    "    \n",
    "    true_prediction = np.equal(predicted_labels,Y)\n",
    "    \n",
    "    num_correct_labels = np.sum(true_prediction)\n",
    "    accuracy = num_correct_labels / m\n",
    "        \n",
    "    return predicted_labels, predicted_prob, accuracy\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Visualizing the costs and accuracy for model analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "def visualize_results(attr, attr_type):\n",
    "    \n",
    "    plt.plot(np.squeeze(attr))\n",
    "    if attr_type == 'costs':\n",
    "        plt.ylabel(\"cost\")\n",
    "        plt.title(\"Cost\")\n",
    "        \n",
    "    elif attr_type == 'train_accs':\n",
    "        plt.ylabel(\"accuracy\")\n",
    "        plt.title(\"Training Accuracy\")\n",
    "        \n",
    "    elif attr_type == 'val_accs':\n",
    "        plt.ylabel(\"accuracy\")\n",
    "        plt.title(\"Validation Accuracy\")\n",
    "        \n",
    "    else:\n",
    "        raise ValueError(\"Dataset set must be training or dev or test set\")\n",
    "        \n",
    "    plt.xlabel('iterations (per hundreds)')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 7. Train Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train(X_train, Y_train, X_dev, Y_dev, layers_dim, hyperParams):\n",
    "\n",
    "    learning_rate = hyperParams['learning_rate']\n",
    "    num_iterations = hyperParams['num_iterations']\n",
    "    \n",
    "    np.random.seed(1)\n",
    "    costs = []      # keep track of cost\n",
    "    train_accs = []  # keep track of training accuracy\n",
    "    val_accs = []     # keep track of Validation accuracy\n",
    "    \n",
    "    parameters = init_params(layers_dim)\n",
    "    \n",
    "    for i in range(0, num_iterations):\n",
    "\n",
    "        AL, caches = forward_prop(X_train, parameters)\n",
    "        \n",
    "        cost = compute_cost(AL, Y_train)\n",
    "    \n",
    "        grads = backward_prop(AL, Y_train, caches)\n",
    " \n",
    "        parameters = update_parameters(parameters, grads, learning_rate)\n",
    "                \n",
    "        _,_,train_acc = predict(X_train, Y_train,parameters)\n",
    "        _,_,val_acc= predict(X_dev, Y_dev,parameters)        \n",
    "        \n",
    "        if i == 0 or (i+1) % 200 == 0:\n",
    "            print (\"Iteration: %d == Cost: %f || Training acc: %f || Val acc: %f\"%(i,cost,train_acc,val_acc))\n",
    "        if i == 0 or (i+1) % 100 == 0:\n",
    "            costs.append(cost)\n",
    "            train_accs.append(train_acc)\n",
    "            val_accs.append(val_acc)\n",
    "            \n",
    "            \n",
    "    visualize_results(costs, attr_type='costs')  \n",
    "    visualize_results(train_accs, attr_type='train_accs')       \n",
    "    visualize_results(val_accs, attr_type='val_accs')       \n",
    "    \n",
    "    return parameters"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Running Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 0 == Cost: 2.302584 || Training acc: 0.112367 || Val acc: 0.119000\n",
      "Iteration: 199 == Cost: 2.301098 || Training acc: 0.112367 || Val acc: 0.119000\n",
      "Iteration: 399 == Cost: 2.300742 || Training acc: 0.112367 || Val acc: 0.119000\n",
      "Iteration: 599 == Cost: 2.292399 || Training acc: 0.125850 || Val acc: 0.135400\n",
      "Iteration: 799 == Cost: 1.562681 || Training acc: 0.483900 || Val acc: 0.493800\n",
      "Iteration: 999 == Cost: 0.728470 || Training acc: 0.775083 || Val acc: 0.780200\n",
      "Iteration: 1199 == Cost: 0.574855 || Training acc: 0.828917 || Val acc: 0.838400\n",
      "Iteration: 1399 == Cost: 0.499987 || Training acc: 0.856733 || Val acc: 0.858200\n",
      "Iteration: 1599 == Cost: 0.440030 || Training acc: 0.875033 || Val acc: 0.875000\n",
      "Iteration: 1799 == Cost: 0.370123 || Training acc: 0.894600 || Val acc: 0.895400\n",
      "Iteration: 1999 == Cost: 0.318842 || Training acc: 0.908633 || Val acc: 0.908200\n",
      "Iteration: 2199 == Cost: 0.283214 || Training acc: 0.918433 || Val acc: 0.917200\n",
      "Iteration: 2399 == Cost: 0.254448 || Training acc: 0.926933 || Val acc: 0.926200\n",
      "Iteration: 2599 == Cost: 0.231414 || Training acc: 0.933350 || Val acc: 0.933000\n",
      "Iteration: 2799 == Cost: 0.212893 || Training acc: 0.939117 || Val acc: 0.938000\n",
      "Iteration: 2999 == Cost: 0.197592 || Training acc: 0.943617 || Val acc: 0.942800\n",
      "Iteration: 3199 == Cost: 0.184544 || Training acc: 0.947267 || Val acc: 0.947000\n",
      "Iteration: 3399 == Cost: 0.173426 || Training acc: 0.950333 || Val acc: 0.948400\n",
      "Iteration: 3599 == Cost: 0.163896 || Training acc: 0.953117 || Val acc: 0.951800\n",
      "Iteration: 3799 == Cost: 0.155610 || Training acc: 0.955517 || Val acc: 0.952800\n",
      "Iteration: 3999 == Cost: 0.148316 || Training acc: 0.957200 || Val acc: 0.954400\n",
      "Iteration: 4199 == Cost: 0.141858 || Training acc: 0.959117 || Val acc: 0.955600\n",
      "Iteration: 4399 == Cost: 0.136066 || Training acc: 0.960850 || Val acc: 0.957600\n",
      "Iteration: 4599 == Cost: 0.130795 || Training acc: 0.962117 || Val acc: 0.958800\n",
      "Iteration: 4799 == Cost: 0.125917 || Training acc: 0.963417 || Val acc: 0.959000\n",
      "Iteration: 4999 == Cost: 0.121376 || Training acc: 0.964650 || Val acc: 0.960200\n",
      "Iteration: 5199 == Cost: 0.117199 || Training acc: 0.966017 || Val acc: 0.961200\n",
      "Iteration: 5399 == Cost: 0.113238 || Training acc: 0.967300 || Val acc: 0.961600\n",
      "Iteration: 5599 == Cost: 0.109496 || Training acc: 0.968433 || Val acc: 0.961200\n",
      "Iteration: 5799 == Cost: 0.105943 || Training acc: 0.969383 || Val acc: 0.961800\n",
      "Iteration: 5999 == Cost: 0.102583 || Training acc: 0.970483 || Val acc: 0.962400\n",
      "Iteration: 6199 == Cost: 0.099382 || Training acc: 0.971433 || Val acc: 0.962600\n",
      "Iteration: 6399 == Cost: 0.096305 || Training acc: 0.972433 || Val acc: 0.962600\n",
      "Iteration: 6599 == Cost: 0.093368 || Training acc: 0.973233 || Val acc: 0.962400\n",
      "Iteration: 6799 == Cost: 0.090576 || Training acc: 0.973967 || Val acc: 0.962600\n",
      "Iteration: 6999 == Cost: 0.087906 || Training acc: 0.974883 || Val acc: 0.963000\n",
      "Iteration: 7199 == Cost: 0.085373 || Training acc: 0.975700 || Val acc: 0.963600\n",
      "Iteration: 7399 == Cost: 0.082962 || Training acc: 0.976450 || Val acc: 0.964400\n",
      "Iteration: 7599 == Cost: 0.080692 || Training acc: 0.977167 || Val acc: 0.964200\n",
      "Iteration: 7799 == Cost: 0.078519 || Training acc: 0.977767 || Val acc: 0.964200\n",
      "Iteration: 7999 == Cost: 0.076441 || Training acc: 0.978517 || Val acc: 0.964200\n",
      "Iteration: 8199 == Cost: 0.074428 || Training acc: 0.979133 || Val acc: 0.964600\n",
      "Iteration: 8399 == Cost: 0.072483 || Training acc: 0.979800 || Val acc: 0.964000\n",
      "Iteration: 8599 == Cost: 0.070624 || Training acc: 0.980383 || Val acc: 0.963800\n",
      "Iteration: 8799 == Cost: 0.068838 || Training acc: 0.980967 || Val acc: 0.964600\n",
      "Iteration: 8999 == Cost: 0.067108 || Training acc: 0.981483 || Val acc: 0.965200\n",
      "Iteration: 9199 == Cost: 0.065451 || Training acc: 0.982183 || Val acc: 0.965400\n",
      "Iteration: 9399 == Cost: 0.063847 || Training acc: 0.982700 || Val acc: 0.965400\n",
      "Iteration: 9599 == Cost: 0.062305 || Training acc: 0.983017 || Val acc: 0.966000\n",
      "Iteration: 9799 == Cost: 0.060818 || Training acc: 0.983433 || Val acc: 0.965800\n",
      "Iteration: 9999 == Cost: 0.059388 || Training acc: 0.983900 || Val acc: 0.966000\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nO3deXwcd33/8ddnD2m1OizJkh3HR2wnhhAgiYNzESgpya8laSAQoE0oUI6SlgeUs782tIW2/FpaeqRtCoRHCElICRAIEBKg3EcIkBA5sZ07US7b8SXbsiVb0kq7+/n9MSNlLUuyLO1oNbvv5yP72N2Z2d3PaJx973xn5vs1d0dERGpXotIFiIhIZSkIRERqnIJARKTGKQhERGqcgkBEpMYpCEREapyCQESkxikIRI7AzN5kZl1mdsDMtpvZ/5rZy2bxfk+b2fnlrFFkNhQEIlMwsw8B/wl8AlgMrAA+A1xcybpEysl0ZbHIxMxsAfAs8HZ3/9oE8+uBTwK/H076KvCX7p4zsw7gBuBlQBF4EHgF8AXgD4EcUAA+7u7/EvGqiExJewQikzsbyADfnGT+XwNnAacCpwBnAH8TzvswsBXoJNiT+CvA3f0twGbg1e7epBCQ+UBBIDK5hcBud89PMv8PCX7R73L3HuDvgbeE80aAJcBx7j7i7r9w7X7LPKUgEJncHqDDzFKTzD8WeKbk+TPhNIB/BbqBH5jZk2Z2RXRlisyOgkBkcr8GhoDXTjJ/G3BcyfMV4TTcvd/dP+zuq4FXAx8ys/PC5bRnIPPKZL90RGqeu+83s48BnzazPPADgiaf84HfBr4M/I2Z3UPw5f4x4IsAZnYR8AjwBNBHcGC4EL71TmD1HK6KyJS0RyAyBXe/EvgQwUHgHmAL8F7gVuAfgC5gE3A/cG84DWAN8CPgAMGexWfc/WfhvH8iCJB9Zvbnc7MmIpPT6aMiIjVOewQiIjVOQSAiUuMUBCIiNU5BICJS42J3+mhHR4evXLmy0mWIiMTK+vXrd7t750TzYhcEK1eupKurq9JliIjEipk9M9k8NQ2JiNQ4BYGISI1TEIiI1DgFgYhIjVMQiIjUOAWBiEiNUxCIiNS42F1HMFOP7ezn2xu3UZ9O0pBOUp9OkDQjYQbBf5gZBiQSkDDDzEgnjLpUgvpUkjWLm1jckqn0qoiIlFXNBMHjOw9w1U+6Z/0+SxZkOG1FG3/5qhNZsTBbhspERCqrZoLg905ewoUvvpBcvsjQSIFcvkjRnaJDsRiMyeAOjuNOOM8ZKTgjhSIHcwUe2t7Hhi37uH3jNk48ppk/O29NhddKRGT2aiYIIGj6yaSTZNLJGb3+7OMXAnDXk3vY0jtQztJERCpGB4tnYEV7li17BytdhohIWSgIZmB5WwOb92qPQESqg4JgBla0Z9m+f5CRQrHSpYiIzJqCYAaWtWcpOmzbp+YhEYk/BcEMrGgPThvVcQIRqQYKghlYHgaBjhOISDVQEMzAMS0Z0knTKaQiUhUUBDOQTBhLW3XmkIhUBwXBDC1vz7JVQSAiVUBBMEPL27PaIxCRqqAgmKEV7Vl6B0boHxqpdCkiIrOiIJih5W06hVREqoOCYIbGriXQmUMiEnMKghla3t4AwBYdJxCRmFMQzNCChjTNmZSCQERiT0EwQ2bG8jadOSQi8acgmIUV7Vm29OpgsYjEm4JgFpa3N7Bl7wDuXulSRERmTEEwCyvas+TyRXr6c5UuRURkxhQEs7BMvZCKSBVQEMzCCgWBiFSByILAzJab2U/N7GEze9DM3j/BMmZmV5lZt5ltMrPToqonCktbg2sJNFKZiMRZKsL3zgMfdvd7zawZWG9mP3T3h0qWuQBYE97OBK4O72Mhk07SWJdk70H1NyQi8RXZHoG7b3f3e8PH/cDDwNJxi10M3OiBu4BWM1sSVU1RaM3WsW9guNJliIjM2JwcIzCzlcBa4O5xs5YCW0qeb+XwsMDMLjezLjPr6unpiarMGWlrTNOrIBCRGIs8CMysCfg68AF37xs/e4KXHHZSvrtf4+7r3H1dZ2dnFGXOWFu2jn2DahoSkfiKNAjMLE0QAje5+zcmWGQrsLzk+TJgW5Q1lVvQNKQgEJH4ivKsIQM+Dzzs7ldOsthtwFvDs4fOAva7+/aoaopCW1ZNQyISb1GeNXQO8BbgfjPbEE77K2AFgLt/FvgucCHQDQwAb4+wnki0ZuvYPzhCoegkExO1dImIzG+RBYG738nExwBKl3HgPVHVMBfasmncoW9whLbGukqXIyJy1HRl8Sy1ZYMvfzUPiUhcKQhmqTWbBqBXB4xFJKYUBLM0ukegi8pEJK4UBLP0XNOQ9ghEJJ4UBLPU2hg0DWmPQETiSkEwS831KZIJ08FiEYktBcEsmRmtDWk1DYlIbCkIyqA1m1bTkIjEloKgDNqydfRqTAIRiSkFQRm0Zut0jEBEYktBUAZt2bR6IBWR2FIQlEFbo/YIRCS+FARl0JpNk8sXGRwuVLoUEZGjpiAoA3U8JyJxpiAog7axjucUBCISPwqCMmgd63hOB4xFJH4UBGWgpiERiTMFQRm0aUwCEYkxBUEZjDUNHdQegYjEj4KgDOpSCRrrktojEJFYUhCUSWu2Th3PiUgsKQjKpK0xrYPFIhJLCoIyacvWqWlIRGJJQVAmahoSkbhSEJRJW1ajlIlIPCkIyqQ1W0ff0AiFole6FBGRo6IgKJO2bBp32D+ovQIRiRcFQZmomwkRiSsFQZm0ht1M6ICxiMSNgqBMxvYINIi9iMSMgqBM1DQkInGlICiT1sbRpiHtEYhIvCgIyqS5PkUyYewb1B6BiMSLgqBMzIzmTIr+oXylSxEROSoKgjJSEIhIHEUWBGZ2nZntMrMHJpl/rpntN7MN4e1jUdUyV5rq0woCEYmdVITvfQPwKeDGKZb5hbtfFGENcyrYI9DBYhGJl8j2CNz9DmBvVO8/HzXXq2lIROKn0scIzjazjWb2v2b2wskWMrPLzazLzLp6enrmsr6j0pxJcSCnIBCReKlkENwLHOfupwD/Ddw62YLufo27r3P3dZ2dnXNW4NFqzqTVNCQisVOxIHD3Pnc/ED7+LpA2s45K1VMOTeEegbu6ohaR+KhYEJjZMWZm4eMzwlr2VKqecmjOpBgpOLl8sdKliIhMW2RnDZnZl4FzgQ4z2wr8LZAGcPfPAm8A3m1meWAQuNRj/lO6ORN0M9E3NEImnaxwNSIi0xNZELj7ZUeY/ymC00urRnN98Oc8MJRnUXOFixERmaZKnzVUVZozQRDoFFIRiRMFQRk11SsIRCR+FARlNHqM4EBOp5CKSHwoCMpotGmoT3sEIhIjCoIy0jECEYkjBUEZNZWcNSQiEhcKgjJKJRNk65LqZkJEYkVBUGZN9ep4TkTiRUFQZhqlTETiRkFQZs2ZNH1qGhKRGFEQlJnGJBCRuFEQlJmahkQkbhQEZdZUr3GLRSReFARl1pxJ6zoCEYmVaQWBmb1xOtMkaBo6OFygUIz10AoiUkOmu0fwkWlOq3m6ulhE4mbKgWnM7ALgQmCpmV1VMqsF0DfdBFrCHkj7cyMsyKYrXI2IyJEdaYSybUAX8Bpgfcn0fuCDURUVZ+p4TkTiZsogcPeNwEYz+5K7jwCYWRuw3N1756LAuGkKg0DXEohIXEz3GMEPzazFzNqBjcD1ZnZlhHXF1ujgNDqFVETiYrpBsMDd+4BLgOvd/SXA+dGVFV8arlJE4ma6QZAysyXA7wPfjrCe2GvRMQIRiZnpBsHHge8DT7j7PWa2Gng8urLi67mmIQWBiMTDkc4aAsDdvwZ8reT5k8DroyoqzjLpBMmE6RiBiMTGdK8sXmZm3zSzXWa208y+bmbLoi4ujsxMPZCKSKxMt2noeuA24FhgKXB7OE0moB5IRSROphsEne5+vbvnw9sNQGeEdcVaU31aTUMiEhvTDYLdZvZmM0uGtzcDe6IsLM60RyAicTLdIHgHwamjO4DtwBuAt0dVVNw11ysIRCQ+phsE/w/4I3fvdPdFBMHwd5FVFXM6WCwicTLdIDi5tG8hd98LrI2mpPhrzugYgYjEx3SDIBF2NgdA2OfQtK5BqEVN4TECdw1OIyLz33S/zP8d+JWZ3QI4wfGCf4ysqphrzqTIF51cvkgmnax0OSIiU5rulcU3mlkX8ErAgEvc/aFIK4ux0W4m+oZGFAQiMu9Nu3kn/OLXl/80NJf0QLqoucLFiIgcwXSPERw1M7su7JLigUnmm5ldZWbdZrbJzE6Lqpa5NjpKmcYtFpE4iCwIgBuAV00x/wJgTXi7HLg6wlrmlMYkEJE4iSwI3P0OYO8Ui1wM3OiBu4DWcMyD2NMoZSISJ1HuERzJUmBLyfOt4bTDmNnlZtZlZl09PT1zUtxsjA1gr4vKRCQGKhkENsG0CU+8d/dr3H2du6/r7Jz/fd01a5QyEYmRSgbBVmB5yfNlwLYK1VJWo8cIdLBYROKgkkFwG/DW8Oyhs4D97r69gvWUTSqZoLEuyf5BHSMQkfkvsm4izOzLwLlAh5ltBf4WSAO4+2eB7wIXAt3AAFXWm+nStgY27x2odBkiIkcUWRC4+2VHmO/Ae6L6/Epb1dFI964DlS5DROSIKtk0VNVWdzaxee8A+UKx0qWIiExJQRCRVR2NjBScrb2DlS5FRGRKCoKIHN/ZCMBTuw9WuBIRkakpCCKyqqMJgCd6dJxAROY3BUFE2rJpFjSktUcgIvOegiAiZsbqzkae7FEQiMj8piCI0KqORu0RiMi8pyCI0PGdTezoG+KgOp8TkXlMQRChVR06c0hE5j8FQYRW6xRSEYkBBUGEVi5sxAwdMBaReU1BEKFMOsmxCxp4areuJRCR+UtBELHVnY08qaYhEZnHFAQRW93RyFM9Bwk6WxURmX8UBBFb1dFIfy5Pz4FcpUsREZmQgiBiqzuDPoee0gFjEZmnFAQRG72WQMcJRGS+UhBE7NjWBupSCZ7QaGUiMk8pCCKWTBinLm/lZ4/16ICxiMxLCoI58OpTjqV71wEe2dFf6VJERA6jIJgDF77oGJIJ4/aN2ypdiojIYRQEc2BhUz3nnNDB7Zu2qXlIROYdBcEcefXJS9iyd5ANW/ZVuhQRkUMoCObI77zwGOqSCW7fuL3SpYiIHEJBMEcWNKQ59/mdfHvTNgpFNQ+JyPyhIJhDrz7lWHb15/jNU3srXYqIyBgFwRw67wWLaEgnufmezZUuRURkjIJgDmXrUrztnJXcumEbv+zeXelyREQABcGce/95a1jV0chHvnE/g8OFSpcjIqIgmGuZdJJ/uuTFbN47wJU/fLTS5YiIKAgq4azVC3nTmSv4/J1PsVHXFYhIhSkIKuSKC05kUXOG93/lPvYPjFS6HBGpYQqCCmnJpPnUm9by7L5B3vOle8kXipUuSURqlIKggtatbOcTr3sxd3bv5h++83ClyxGRGpWqdAG17o3rlvPojn6uvfMpjl/UxFvOOq7SJYlIjYl0j8DMXmVmj5pZt5ldMcH8t5lZj5ltCG9/HGU989VHLnwBrzxxER+99QH+59dPV7ocEakxkQWBmSWBTwMXACcBl5nZSRMserO7nxrero2qnvksmTCufvNpnP+CxXz0Ww9yzR1PVLokEakhUe4RnAF0u/uT7j4MfAW4OMLPi7X6VJKr33waF528hE989xGu/MGjGrtAROZElEGwFNhS8nxrOG2815vZJjO7xcyWT/RGZna5mXWZWVdPT08Utc4L6WSC/7p0Lb+/bhlX/aSbD311I7m8rj4WkWhFGQQ2wbTxP3FvB1a6+8nAj4AvTPRG7n6Nu69z93WdnZ1lLnN+SSaMT77+ZP7v7z6fb973LG++9m72HhyudFkiUsWiDIKtQOkv/GXAIYP2uvsed8+FTz8HvCTCemLDzHjPb5/Ap960lo1b9/O6z/ySp3cfrHRZIlKlogyCe4A1ZrbKzOqAS4HbShcwsyUlT18D6GT6EhedfCxfufws+ofyXHL1r7hvc2+lSxKRKhRZELh7Hngv8H2CL/ivuvuDZvZxM3tNuNj7zOxBM9sIvA94W1T1xNVpK9r4+rtfSlN9iss+dxc/emhnpUsSkSpjcTszZd26dd7V1VXpMubc7gM53nnDPTywrY+rLl3L75285MgvEhEJmdl6d1830Tx1MRETHU313PSus1i7vJX3feU+bt+47cgvEhGZBgVBjDTVp7jhHWfwkhVtvP8r9/GtDc9WuiQRqQIKgphpqk9x/dtP5/SV7Xzw5g18Z9P2SpckIjGnIIihxvoU173tdE4L9wx+qAPIIjILCoKYagz3DF54bAvvuelefv5Y9V5xLSLRUhDEWHMmzY3vOJMTFjXxri908Y17t1a6JBGJIQVBzC3IpvnSu87kJce18aGvbuRfvvcIxWK8TgkWkcpSEFSB1mwdX3jHGVx2xnI+87Mn+JMvrmfPgdyRXygigoKgatSlEnzidS/moxedxM8e3cX5V/6cW+97Vl1Zi8gRKQiqiJnxzpet4jvveznHLWzkAzdv4G3X30P3rv5KlyYi85iCoAo9b3EzX3/3S/noRSex/plefuc/7uAj37ifXX1DlS5NROYh9TVU5fYcyPHfP+nmprufIZkw3nTGcfzJK1azuCVT6dJEZA5N1deQgqBGPLPnIFf9uJtbNzxL0ow3rlvGH798Nas6GitdmojMAQWBjNm8Z4Crf97NLeu3ki865524iHecs4qzj1+I2USDyolINVAQyGF29Q/xxV8/wxfv3szeg8Os6mjkD05fzutPW0Znc32lyxORMlMQyKSGRgp8e9N2br5nM/c83UsqYbx8TQcXn7qU/3PSYhrrU5UuUUTKQEEg09K96wBfW7+F2zdsY9v+ITLpBK94Xie/+8JjOO/ExSzIpitdoojMkIJAjkqx6HQ908vtG7fxg4d2sLMvRzJhvGRFG694fie/taaTFx7bQiKhYwoicaEgkBkrFp1Nz+7nhw/t4OeP9fDAs30AtGbTnLmqnbNXL+T0Ve2ceEwLSQWDyLylIJCy6enPcWd3D7/q3sOvn9zD1t5BIBgwZ+2KVtauaOPU5Qs4ZVkrC5t00FlkvlAQSGS27B1g/TO93PP0XtY/08tjO/sZ7fx0aWsDL1rawouXLuCkY1s48ZgWlizI6DRVkQqYKgh0SojMyvL2LMvbs7x27VIADubyPPDsfjZs2ccD2/p44Nn9fP/B50ZQa8mkeP4xzaxZ3MzzFjVxwqJmjl/UyDEtCgiRSlEQSFk11qc4c/VCzly9cGxa/9AIj+7o5+Ed/Ty8vY/Hd/bznU3b+dLgyNgy2bokKxc2sqqjkeMWZlm5sJHl7VlWLMxyTEtGxx9EIqQgkMg1Z9KsW9nOupXtY9PcnZ7+HN09B3iy5yDduw7w9J6DPLhtP997cAeFksF10kljyYIGlrU1sLS1gSWtDSxtzbBkQQNLFmQ4ZkGG5oxObRWZKQWBVISZsaglw6KWDC89vuOQeSOFItv3DbF57wCb9w6wtXeArb2DbO0d4BeP72Zn/xDjD2011iVZ3JJhUUs9i1sydDbVs6ilns7mejqbMnQ019HRVE9btk57FyLjKAhk3kknE6xYGDQLTWSkUGTH/iG27x9i+/5BdvYFj3f15djZN8S9m3vZ1Zcjly8e9tqEQXtj3dhtYWM9bY1p2hvrac+maWusozVbR1s2TVu2jtZsmqb6lI5fSFVTEEjspJOJsYPUk3F3+nN5evpz7O7PsfvAMLsP5NhzIMfug8Ps7s/ROzDMwzv62HtwmP2DI4ftZYxKJYzWbJoFDWlas3UsaEiP3VpKHi9oSNOcSdGSCe8b0jTXp3Thncx7CgKpSmZGSyZNSybN8Z1NR1y+UHT2D46w9+Aw+waG6R0YoXdgmP3hfe/ACH2DI+wbHGZn3xCP7exn/+AI/UP5I753c32KpkyK5kyK5jAkmuoPfdwULjP6uLE+WL6xPkVTXYrG+iSppMaRkmgoCESAZMLGmouORr5Q5EAuT99gnv2DI/QNjdA/NDIWEn1DefqHgscHhvL050boPTjM5j0D9A3lOZjLMzhSmNZnZdIJmupTZOuCgGisS5IN78c/z4bh0VAXPG8YnVbyOFuXpD6VULOXKAhEZiOVTNCaDY4rzNRIociBoTwHcnkODueD0MgFITE2PVfg4HAw7WAuz4FcgYHhIHy27RtkIJdnYKTAQK7AcOHwYyOTMYOGdJLsaECkUzTUJcemZcLHDenk2PSGuiSZVCK4Tz93a0gnyaQT4X2S+nQimJdKkk6aAmceUxCIVFg6maCtsY62o9wbmcxwvsjgcBAcA8N5BoYL4S18HIbK4EiBwbF5BQbD+YMjBYZGCuzoGwkeDxcYCKcNjUw/ZEoljLHAqE8lxu7rR+9TCepTQZDUp4IQGZ0WLJegLhkun0yMza8Ll6lLBfPrUodPrw/n6VjN5BQEIlWmLvwijKLb8GLRyeWLDI0EgTEwHARELl9gcDiYPpQPAmYoXyRXEiCj83IjRYbC9xgO7/uG8gznh8mF83P54tjjo9nDmUoqYWN/m3QyDJbweem00cf1qQTppD03LwyUdPK554fMTyZIpxLUJY10MkEqGc4/5DVGKhEsl04a6fBxKhG8plKnNisIRGTaEgkLmojqkrTN0WcWi85woRgERCEIj1y+OHafGwmaw0qnD+eL5AqHzhu7FSZ4XCgyEj4eGCyE8wuMFPyQ+aPLFCPqoi1hQXNjXTJBKgyUdMLGQuWyM1bwxy9fXfbPVRCIyLyWSBiZRNCsBPPjCvJC0RkpBMFTGhAjBR97nC8WGc4Hz0cfDxeK5EeXLzj5QpF8YXR6sOxIseRxIXjPfKHISNEjG0ZWQSAicpSSCSM5Fk7xF+mJyWb2KjN71My6zeyKCebXm9nN4fy7zWxllPWIiMjhIgsCM0sCnwYuAE4CLjOzk8Yt9k6g191PAP4D+GRU9YiIyMSi3CM4A+h29yfdfRj4CnDxuGUuBr4QPr4FOM90srGIyJyKMgiWAltKnm8Np024jLvngf3AwnHLYGaXm1mXmXX19PREVK6ISG2KMggm+mU//qSr6SyDu1/j7uvcfV1nZ2dZihMRkUCUQbAVWF7yfBmwbbJlzCwFLAD2RliTiIiME2UQ3AOsMbNVZlYHXArcNm6Z24A/Ch+/AfiJ+2SdAYuISBQiu47A3fNm9l7g+0ASuM7dHzSzjwNd7n4b8Hngf8ysm2BP4NKo6hERkYlZ3H6Am1kP8MwMX94B7C5jOXGgda4NWufaMJt1Ps7dJzzIGrsgmA0z63L3dZWuYy5pnWuD1rk2RLXOGvJIRKTGKQhERGpcrQXBNZUuoAK0zrVB61wbIlnnmjpGICIih6u1PQIRERlHQSAiUuNqJgiONDZCNTCz5Wb2UzN72MweNLP3h9PbzeyHZvZ4eD9XowzOCTNLmtl9Zvbt8PmqcHyLx8PxLsozKvw8YWatZnaLmT0Sbuuza2AbfzD8N/2AmX3ZzDLVtp3N7Doz22VmD5RMm3C7WuCq8Ptsk5mdNpvProkgmObYCNUgD3zY3V8AnAW8J1zPK4Afu/sa4Mfh82ryfuDhkuefBP4jXN9egnEvqsl/Ad9z9xOBUwjWvWq3sZktBd4HrHP3FxH0VHAp1bedbwBeNW7aZNv1AmBNeLscuHo2H1wTQcD0xkaIPXff7u73ho/7Cb4glnLouA9fAF5bmQrLz8yWAb8HXBs+N+CVBONbQPWtbwvwWwTds+Duw+6+jyrexqEU0BB2TpkFtlNl29nd7+DwTjcn264XAzd64C6g1cyWzPSzayUIpjM2QlUJh/1cC9wNLHb37RCEBbCocpWV3X8CfwEUw+cLgX3h+BZQfdt6NdADXB82h11rZo1U8TZ292eBfwM2EwTAfmA91b2dR022Xcv6nVYrQTCtcQ+qhZk1AV8HPuDufZWuJypmdhGwy93Xl06eYNFq2tYp4DTgandfCxykipqBJhK2i18MrAKOBRoJmkbGq6btfCRl/XdeK0EwnbERqoKZpQlC4CZ3/0Y4eefobmN4v6tS9ZXZOcBrzOxpgua+VxLsIbSGTQhQfdt6K7DV3e8On99CEAzVuo0Bzgeecvcedx8BvgG8lOrezqMm265l/U6rlSCYztgIsRe2j38eeNjdryyZVTruwx8B35rr2qLg7h9x92XuvpJgm/7E3f8Q+CnB+BZQResL4O47gC1m9vxw0nnAQ1TpNg5tBs4ys2z4b3x0nat2O5eYbLveBrw1PHvoLGD/aBPSjLh7TdyAC4HHgCeAv650PRGt48sIdg83ARvC24UE7eY/Bh4P79srXWsE634u8O3w8WrgN0A38DWgvtL1lXldTwW6wu18K9BW7dsY+HvgEeAB4H+A+mrbzsCXCY6BjBD84n/nZNuVoGno0+H32f0EZ1TN+LPVxYSISI2rlaYhERGZhIJARKTGKQhERGqcgkBEpMYpCEREapyCQCJhZr8K71ea2ZvK/N5/NdFnRcXMXmtmH4vovQ9E9L7njvbGOov3uMHM3jDF/Pea2dtn8xkyPygIJBLu/tLw4UrgqIIg7C12KocEQclnReUvgM/M9k2msV6RK7kStxyuI+gVVGJOQSCRKPml+8/Ay81sQ9infNLM/tXM7gn7Uf+TcPlzw7EUvkRwgQxmdquZrQ/7ob88nPbPBL1QbjCzm0o/K7zK8l/DPuvvN7M/KHnvn5X04X9TeIUqZvbPZvZQWMu/TbAezwNy7r47fH6DmX3WzH5hZo+F/R2NjokwrfWa4DP+0cw2mtldZra45HPeULLMgZL3m2xdXhVOuxO4pOS1f2dm15jZD4Abp6jVzOxT4d/jO5R0XDfR38ndB4CnzeyM6fybkPmrnL8ORCZyBfDn7j76hXk5weXwp5tZPfDL8AsKgu7CX+TuT4XP3+Hue82sAbjHzL7u7leY2Xvd/dQJPusSgqtuTwE6wtfcEc5bC7yQoD+WXwLnmNlDwOuAE93dzax1gvc8B7h33LSVwCuA44GfmtkJwFuPYr1KNQJ3uftfm9m/AO8C/mGC5UpNtC5dwOcI+lvqBm4e95qXAC9z98EptsFa4PnAi4HFBN04XGdm7VP8nQhxh+YAAAKISURBVLqAlxNc4SsxpT0CmWu/Q9BHygaCLrIXEgyuAfCbcV+W7zOzjcBdBB1srWFqLwO+7O4Fd98J/Bw4veS9t7p7kaDrjZVAHzAEXGtmlwADE7znEoJun0t91d2L7v448CRw4lGuV6lhYLQtf31Y15FMtC4nEnTM9rgH3QV8cdxrbnP3wfDxZLX+Fs/9/bYBPwmXn+rvtIugR1CJMe0RyFwz4M/c/fuHTDQ7l6BL5dLn5wNnu/uAmf0MyEzjvSeTK3lcAFLung+bNc4j6LTuvQS/qEsNAgvGTRvfL4szzfWawIg/189Lgef+n8wT/lALm35Kh2E8bF0mqatUaQ2T1XrhRO9xhL9ThuBvJDGmPQKJWj/QXPL8+8C7LeguGzN7ngUDq4y3AOgNQ+BEgqE3R42Mvn6cO4A/CNvAOwl+4U7aZGHBuA0L3P27wAcImpXGexg4Ydy0N5pZwsyOJ+j47NGjWK/pepqgOQeCvvgnWt9SjwCrwpoALpti2clqvQO4NPz7LQF+O5w/1d/peQQdwUmMaY9AorYJyIdNPDcQjLe7Erg3/KXbw8RDDH4P+FMz20TwRXtXybxrgE1mdq8H3U6P+iZwNrCR4JftX7j7jjBIJtIMfMvMMgS/kj84wTJ3AP9uZlbyy/1RgmanxcCfuvuQmV07zfWars+Ftf2GoNfJqfYqCGu4HPiOme0G7gReNMnik9X6TYJf+vcT9NT783D5qf5O5xD0DCoxpt5HRY7AzP4LuN3df2RmNxB0d33LEV5W9cxsLfAhd39LpWuR2VHTkMiRfYJgwHQ5VAfw0UoXIbOnPQIRkRqnPQIRkRqnIBARqXEKAhGRGqcgEBGpcQoCEZEa9/8Bew5EqPeranIAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "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": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "hyperParams = init_hyperParams(alpha = 0.1,num_iteration = 10000)\n",
    "layers_dim = init_layers()\n",
    "parameters = train(train_x_norm, train_y_encoded,dev_x_norm, dev_y_encoded,layers_dim, hyperParams)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Accuracy: 0.9839\n",
      "\n",
      "Error:\t0.016100000000000003\n"
     ]
    }
   ],
   "source": [
    "predicted_labels_train, prediction_prob_train,train_acc = predict(train_x_norm, train_y_encoded,parameters)\n",
    "print(\"\\nAccuracy: \"  + str(train_acc))\n",
    "print(\"\\nError:\\t\"+str(1-train_acc))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Accuracy: 0.966\n",
      "\n",
      "Error:\t0.03400000000000003\n"
     ]
    }
   ],
   "source": [
    "predicted_labels_dev, prediction_prob_prob,dev_acc = predict(dev_x_norm, dev_y_encoded,parameters)\n",
    "print(\"\\nAccuracy: \"  + str(dev_acc))\n",
    "print(\"\\nError:\\t\"+str(1-dev_acc))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Accuracy: 0.964\n",
      "\n",
      "Error:\t0.03600000000000003\n"
     ]
    }
   ],
   "source": [
    "predicted_labels_test, prediction_prob_prob,test_acc = predict(test_x_norm, test_y_encoded,parameters)\n",
    "print(\"\\nAccuracy: \"  + str(test_acc))\n",
    "print(\"\\nError:\\t\"+str(1-test_acc))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def visualize_prediction(x_orig, y_orig, predicted_labels, prediction_prob, dataset):\n",
    "#     print(x_orig.shape,y_orig.shape, predicted_labels.shape)\n",
    "    if(dataset == \"training\"):\n",
    "        visual_title = \"Sample Training Data Set\"\n",
    "        rng = range(30,40)\n",
    "    elif(dataset == \"dev\"):\n",
    "        visual_title = \"Sample Dev Data Set\"\n",
    "        rng = range(110,120)\n",
    "    elif(dataset == \"test\"):\n",
    "        visual_title = \"Sample Test Data Set\"\n",
    "        rng = range(110,120)        \n",
    "    else:\n",
    "        raise ValueError(\"Dataset set must be training or dev or test set\")\n",
    "    fig, axes = plt.subplots(nrows=2, ncols=5,figsize=(16,8))\n",
    "    fig.subplots_adjust(hspace=1)\n",
    "    fig.suptitle(visual_title)\n",
    "\n",
    "    for ax,i in zip(axes.flatten(),rng):\n",
    "        ax.imshow(x_orig[i].squeeze(),interpolation='nearest', cmap='Greys')\n",
    "        ax.set(title = \"True: \"+ str(y_orig[0,i])+\" | Predicted: \"+str(predicted_labels[0,i]))\n",
    "        ax.set(xlabel= \"Prediction Prob: %f\"%(prediction_prob[0,i]))\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x576 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "visualize_prediction(train_x_sample, train_y_sample.T,predicted_labels_train, prediction_prob_train,dataset = \"training\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x576 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "visualize_prediction(dev_x_orig, dev_y_orig.T, predicted_labels_dev, prediction_prob_prob,dataset = \"dev\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x576 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "visualize_prediction(test_x_orig, test_y_orig.T, predicted_labels_test, prediction_prob_prob,dataset = \"test\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Predicting Real Time images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "from PIL import Image\n",
    "from dataPrep import one_hot_encoding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f37a42a6550>"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAEXCAYAAABrgzLrAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAZfklEQVR4nO3debRcZZ3u8e9jwpAwNCABAoJhahQBI5wOAURRFJGZtArplkFpkauALAav4l2LqZfSAnr7CmLTzLQEEAlTsxgCyixwDEMSpjAEiXBJYlQS4BKS/O4f+z1YHE69dc6pOqlK3uez1lmnzv7tXfutfeqpPb61FRGY2YrvA+1ugJktGw67WSEcdrNCOOxmhXDYzQrhsJsVwmG3QZH0r5IuXdbT2uA57G0maWHNz1JJb9X8/c/LYP7/JenUoZ5PMyRNlPS0pAWSpkvat91tWh4Nb3cDShcRq/c8ljQL+JeImFJvfEnDI2LxsmhbJ5C0CXAZsA9wB7AfcLWkjSPiT21t3HLGa/YOlzZ5r5Y0SdIC4Ku918aSPpc+KHr+/pCkyZLmSnpR0rcHOe9zJc2W9LqkRyTt3GuUEZJ+lda43ZK2bXUbgI2BeRFxe1RuAN4GNhvk8xXLYV8+HAhcCfwdcHVuREnDgJuBR4CNgM8DJ0nafRDzfQjYDlgHuBb4laRVauoTUrt66pMlDR9oGyTNkPSVTBuel7S3pGGS/hFYAEwfxOspmsO+fLgvIm6KiKUR8VaDcccDa0bEDyNiUUQ8B1wEHDzQmUbEFRExP+02/BhYE9iiZpSHImJyRLwDnJXq/zDQNkTExyLimjq1xcDlwDVUa/TLgSP7sRysF++zLx9eHsC4HwY2kfSXmmHDgN8OdKaSvgt8HRgNBLAasG5f7YqIJZL+CGwIrNLCNnwR+CGwK/AY1YfJDZI+HxHTBvp8JXPYlw+9uya+AYys+XuDmscvAzMj4qPNzFDSZ4Djgd2BJ9PgvwKqGW3jmvE/QLXJ/grV+6rpNiQfB34bEVPT3w9J6k7tctgHwJvxy6fHgL0lrS1pNHBsTe1BYJGkEyStmvZzt5W0Q+b5hqdxe35WBtYAFgPzgJWAU6nW7LXGSdpf0krAiVT70o8Msg31PAJ8WtJ2AJK6gF2AJwbxXEVz2JdPlwJPAS8BtwJX9RTSPu5ewDhgFlVY/4Nqf7qeHwBv1fzcDtwCTAFmpud5HXi113STga8C84GDgAkRsXigbZD0jKSD+qpFxJ3Av1Id/FtAdYDytIi4K/N6rA/yl1eYlcFrdrNCOOxmhXDYzQrhsJsVwmE3K0RTF9VI2hP4d6qroy6MiDNz46+77roxZsyYZmZpZhmzZs1i3rx56qs26LCnzg7nUXVymA08IunGiHiy3jRjxoyhu7t7sLM0swa6urrq1prZjB8HPBcRL0TEIqoLO/Zv4vnMbAg1E/aNeG8HjdlpmJl1oGbC3td+wfsux5N0ZPpig+65c+c2MTsza0YzYZ9NTa8n4ENUPZ7eIyIuiIiuiOgaNWpUE7Mzs2Y0E/ZHgC0lbZp6SR0M3NiaZplZqw36aHxELJZ0NHAb1am3iyNiRstaZmYt1dR59oi4haorpJl1OF9BZ1YIh92sEA67WSEcdrNCOOxmhXDYzQrhsJsVwmE3K4TDblYIh92sEA67WSEcdrNCOOxmhXDYzQrhsJsVwmE3K4TDblYIh92sEA67WSEcdrNCOOxmhWjq22XNnn322Wz9zjvvrFubNWtWdtpXXnnfPUfeY9NNN83W99lnn7q1cePGZaddEXnNblYIh92sEA67WSEcdrNCOOxmhXDYzQrhsJsVwufZO0BEZOuSsvWFCxfWrc2ZMyc77RprrJGtr7TSStn66aefnq3fdtttdWuLFi3KTrt06dJsvZErrriibu2qq67KTrvjjjs2Ne9O1FTYJc0CFgBLgMUR0dWKRplZ67Vizf6ZiJjXgucxsyHkfXazQjQb9gBul/R7SUe2okFmNjSa3YzfJSJekbQecIekpyPintoR0ofAkQCbbLJJk7Mzs8Fqas0eEa+k33OAycD7uhJFxAUR0RURXaNGjWpmdmbWhEGHXdJqktboeQzsAUxvVcPMrLWa2YxfH5iczgEPB66MiFtb0qrCNDqP/vLLL2frxxxzTN3aAw88kJ12nXXWydaPOuqobH3rrbfO1tdcc826tbXWWis77bRp07L1Rq9t3rz6J4m+//3vZ6e96aabsvXVVlstW+9Egw57RLwAfLyFbTGzIeRTb2aFcNjNCuGwmxXCYTcrhMNuVgh3ce0Ab775ZrZ+7LHHZuu33HJL3Vqj03pz587N1i+88MJsferUqdn6yiuvnK3nvPPOO9n6hAkTsvUpU6bUrc2YMSM77TPPPJOtb7/99tl6J/Ka3awQDrtZIRx2s0I47GaFcNjNCuGwmxXCYTcrhM+zd4Du7u5s/dZb8z2HR4wYUbc2cuTI7LSrrrpqtp7rogqNz4XnzrM3+grtRl9j3dWV/zLj3Hn2BQsWZKd98cUXs3WfZzezjuWwmxXCYTcrhMNuVgiH3awQDrtZIRx2s0L4PHsHaHROt9Gti3O3Pp44cWJ22rPPPjtbb3Q+upmvVG7U176RRl8lnXv+YcOGZadtdCvr5ZHX7GaFcNjNCuGwmxXCYTcrhMNuVgiH3awQDrtZIXyevQOMHj26qelz5+Hvv//+7LRvvPFGtt5s23Ia9WefNGlStv7www9n68OH1397b7fddtlpd9ppp2x9edRwzS7pYklzJE2vGbaOpDskzUy/1x7aZppZs/qzGX8psGevYd8D7oyILYE7099m1sEahj0i7gHm9xq8P3BZenwZcECL22VmLTbYA3TrR8SrAOn3evVGlHSkpG5J3Y3uK2ZmQ2fIj8ZHxAUR0RURXaNGjRrq2ZlZHYMN+2uSRgOk33Na1yQzGwqDDfuNwGHp8WHADa1pjpkNlYbn2SVNAnYD1pU0GzgFOBO4RtIRwB+ALw9lI1d0n/70p7P18ePHZ+sPPvhg3drMmTOz0x5yyCHZ+iWXXJKtb7TRRtl6zvnnn5+tn3zyydl6o/P0uf7sxx9/fHbaFbE/e8OwR0S9bz/YvcVtMbMh5MtlzQrhsJsVwmE3K4TDblYIh92sEO7i2gFWWWWVbP3cc8/N1vfee++6tfnze3dreK977rknW99vv/2y9Z///OfZ+rRp0+rWTj311Oy0jTS6XfQZZ5xRtzZhwoTstM2c1utUXrObFcJhNyuEw25WCIfdrBAOu1khHHazQjjsZoXwefYO0Oic7rbbbput/+xnP6tb+8Y3vpGd9u23387Wn3zyyWx93333zdZzry13q2mAESNGZOunn356tn700UcPql0rKq/ZzQrhsJsVwmE3K4TDblYIh92sEA67WSEcdrNC+Dx7B2i2b/T+++9ft7bmmmtmp/3Wt76Vrb/wwgvZ+sKFC7P1YcOG1a01et0/+tGPsvWvfe1r2XruXPry2B+9WV6zmxXCYTcrhMNuVgiH3awQDrtZIRx2s0I47GaF8Hn2FUDufPKuu+6anXbnnXfO1hudZ2/03e25tg0fnn/7Nbpd9J577pmtb7DBBtl6aRqu2SVdLGmOpOk1w06V9EdJj6WfvYa2mWbWrP5sxl8K9PUR+tOIGJt+bmlts8ys1RqGPSLuAfL3EDKzjtfMAbqjJT2RNvPXrjeSpCMldUvqnjt3bhOzM7NmDDbs5wObA2OBV4Fz6o0YERdERFdEdI0aNWqQszOzZg0q7BHxWkQsiYilwH8C41rbLDNrtUGFXdLomj8PBKbXG9fMOkPD8+ySJgG7AetKmg2cAuwmaSwQwCzgm0PYRmtg6dKldWsnnXRSdtrrrrsuW1977bqHY4DG5+l/85vf1K0tXrw4O+19992XrR9xxBHZeu61rbrqqtlpV0QNwx4RE/sYfNEQtMXMhpAvlzUrhMNuVgiH3awQDrtZIRx2s0K4i2sHaHT74EZfe3zllVfWrV10Uf7ESaNTUGeccUa23uiW0FdddVXd2nHHHZedtlH32bvuuitbv/vuu+vWvvCFL2SnXRF5zW5WCIfdrBAOu1khHHazQjjsZoVw2M0K4bCbFcLn2ZcDjb7O67zzzqtbW7BgQXbaRue6G51Hb2T8+PF1a42+uWj+/PxXHzbqIvv000/Xrfk8u5mtsBx2s0I47GaFcNjNCuGwmxXCYTcrhMNuVgifZ+8Ajfqr33zzzdn6448/Xre2+uqrZ6d95plnsvXTTjstW3/++eez9Vyf8kbn0Rvd0rnR11zvvvvu2XppvGY3K4TDblYIh92sEA67WSEcdrNCOOxmhXDYzQrRn1s2bwxcDmwALAUuiIh/l7QOcDUwhuq2zV+JiD8PXVPLteuuu2brY8aMqVt76aWXstNef/31TdUbfef9yiuvnK0389wnnnhitr7NNtsMet4rov6s2RcDJ0TER4HxwLclbQ18D7gzIrYE7kx/m1mHahj2iHg1IqamxwuAp4CNgP2By9JolwEHDFUjzax5A9pnlzQG+ATwELB+RLwK1QcCsF6rG2dmrdPvsEtaHfg1cFxEvD6A6Y6U1C2pu9F3qZnZ0OlX2CWtRBX0X0bEdWnwa5JGp/poYE5f00bEBRHRFRFdjb5g0MyGTsOwq+qSdRHwVET8pKZ0I3BYenwYcEPrm2dmrdKfLq67AIcA0yQ9loadDJwJXCPpCOAPwJeHpokrvkanmLbYYotsfdKkSXVr55xzTnbaKVOmZOt/+ctfsvVFixZl6yNHjqxb22GHHbLTHnroodn6xIkTs/Xccm3UrXhF1DDsEXEfUG/JuMOw2XLCV9CZFcJhNyuEw25WCIfdrBAOu1khHHazQvirpDtAs+d8x44dW7d2xRVXZKd98cUXs/WZM2dm60uWLMnWt9pqq7q1zTbbLDtts0o8l57jNbtZIRx2s0I47GaFcNjNCuGwmxXCYTcrhMNuVgifZy/cpptu2lTdlh9es5sVwmE3K4TDblYIh92sEA67WSEcdrNCOOxmhXDYzQrhsJsVwmE3K4TDblYIh92sEA67WSEcdrNCOOxmhWgYdkkbS/qNpKckzZD0nTT8VEl/lPRY+tlr6JtrZoPVny+vWAycEBFTJa0B/F7SHan204g4e+iaZ2at0jDsEfEq8Gp6vEDSU8BGQ90wM2utAe2zSxoDfAJ4KA06WtITki6WtHaL22ZmLdTvsEtaHfg1cFxEvA6cD2wOjKVa859TZ7ojJXVL6p47d24Lmmxmg9GvsEtaiSrov4yI6wAi4rWIWBIRS4H/BMb1NW1EXBARXRHRNWrUqFa128wGqD9H4wVcBDwVET+pGT66ZrQDgemtb56ZtUp/jsbvAhwCTJP0WBp2MjBR0lgggFnAN4ekhWbWEv05Gn8f0NeNrm9pfXPMbKj4CjqzQjjsZoVw2M0K4bCbFcJhNyuEw25WCIfdrBAOu1khHHazQjjsZoVw2M0K4bCbFcJhNyuEw25WCEXEspuZNBd4qWbQusC8ZdaAgenUtnVqu8BtG6xWtu3DEdHnV0It07C/b+ZSd0R0ta0BGZ3atk5tF7htg7Ws2ubNeLNCOOxmhWh32C9o8/xzOrVtndoucNsGa5m0ra377Ga27LR7zW5my4jDblaItoRd0p6SnpH0nKTvtaMN9UiaJWlaug11d5vbcrGkOZKm1wxbR9Idkmam3225x16dtnXEbbwztxlv67Jr9+3Pl/k+u6RhwLPA54HZwCPAxIh4cpk2pA5Js4CuiGj7BRiSPgUsBC6PiG3SsB8D8yPizPRBuXZE/M8OadupwMJ238Y73a1odO1txoEDgMNp47LLtOsrLIPl1o41+zjguYh4ISIWAVcB+7ehHR0vIu4B5vcavD9wWXp8GdWbZZmr07aOEBGvRsTU9HgB0HOb8bYuu0y7lol2hH0j4OWav2fTWfd7D+B2Sb+XdGS7G9OH9SPiVajePMB6bW5Pbx11G+9etxnvmGXXjtuftyPsfd1KqpPO/+0SEdsDXwS+nTZXrX/6dRvvZaWP24x3hMHe/rxZ7Qj7bGDjmr8/BLzShnb0KSJeSb/nAJOpcyvqNnqt5w666fecNrfnXf29jfey0NdtxumAZdfM7c+b1Y6wPwJsKWlTSSsDBwM3tqEd7yNptXTgBEmrAXvQebeivhE4LD0+DLihjW15j065jXe924zT5mXX9tufR8Qy/wH2ojoi/zzwg3a0oU67NgMeTz8z2t02YBLVZt07VFtERwAfBO4EZqbf63RQ264ApgFPUAVrdJva9kmqXcMngMfSz17tXnaZdi2T5ebLZc0K4SvozArhsJsVwmE3K4TDblYIh92sEG0Pu6QlqafPdEm/kjSyiefaTdLN6fF+uR51ktaS9K1BzONUSSc2GGclSZel3nNPSfp+nfEulbRbr2GSNK/nkklJoyWFpE/WjDNX0gclHSXp0DTscEkb1owzS9K6Ddq5m6S/Sno0tfOUhgvg/dPfPMBpVpF0taoejw+ly0b7Gu876T0xQ9JxvWrHqOo1OSN1DKq7zOv1NKvzWi7NtPvA9H/4yEBeb39IOjotj8j9zyQdpqrH3kxJh9Ubr562hx14KyLGRtVzahFwVG0xvfkH3M6IuDEizsyMshYw4LD305eBVSJiW2AH4Jv13tS9RXUu9CFgpzRoZ+DR9BtJWwHzIuJPEfGLiLg8jXc4sCEDd29EfALoAr4qaYfaoqThg3jOnCOAP0fEFsBPgX/rPYKkbYBvUF1J9nFgH0lbptpnqDq0bBcRHwN6eorVW+aLgRMi4qPAeKpLoLceRLsnAvdRXQTWavcDn+O9X7P+HpLWAU4BdqRaLqcM9Br6Tgh7rXuBLSSNSZ/EPwemAhtL2kPSg5Kmpi2A1eHdvvFPS7oPmNDzRGlNd256vL6kyZIeTz87A2cCm6etirPSeCdJekRVh4TTap7rB2lNMgXYqh+vI4DVUlBGUH2IDeTa7PtJ4U6/f8J7w/9Aatepkk6U9CWqsP4yvZ4Radxj0vKa1miNFBFvUHW53Dwtu19JuomqU5AknZXWtNMkHVQz6Zpp2T4p6Rf9+GCu7Xl2LbC7pN79JT4K/C4i3oyIxcDdVFeWAfwP4MyIeDu1u+eS1z6XebSgp1l6r+1C9UF1cK/ad9MyeVzSmWnYFpKmpGFTJW2ee/6IeDQiZjVoxheAOyJifkT8GbgD2HMgr6Njwp7+SV+kupIIqlBdntY6bwD/C/hcVJ1UuoHjJa1KdS3xvsCuwAZ1nv7/AHdHxMeB7amujvse8HzaqjhJ0h7AllSfmmOBHSR9Kq3pDqbqoTQB+IeaNh8l6Sje79rU5leBPwBnR8RAuoM+wN/CPg64nr/1J9iZ6sPgXRFxLdUy+ef0et5KpXlpeZ0PNNr1+CDVmm9GGrQTcFhEfJbqdY+lWst+DjhLf7vEcxxwArAtVWeOCen5LpTU13ehv9vrMQX5r1RXttWaDnxK1a7KSKqrzHpe/98Du6ZdgLsl9fw/Gi5zvb+nWX8dANwaEc8C8yVtn57vi6m2Y3pv/TiN/0vgvDRs59QmJD02wPnWarq3aKs30QZjRM1CuJfq2uENgZci4ndp+Hhga+D+tBJYGXgQ+AjwYkTMBJD0X0Bf3VI/CxwKEBFLgL/2sQm0R/p5NP29OlX41wAmR8SbaR7vXscfEb+o85rGAUvS61gbuFfSlIh4Ib8o3vUw8AlV1+evFBELJb0gaQuqN09/e0X1dAD5PTVbPb3sKulRYCnVGnNGCtAdNWH5JDApLbvXJN1N9aH3OvBwz+uSNCmNe21E/Eud+TXs9RgRT0n6N6q110Kqy5cXp/JwqmU6PrXhGkmb0WCZq7kecBOB/50eX5X+nkr1wXdJz3sjIuar6luxUURMTsP+X83rGjvA+dZqurdoJ4T9rd4LIQX6jdpBVG++ib3GG0vruscK+FFE/EeveRw3iHn8E9Wa4B1gjqT7qTaz+xX2iHhT0nPA16neVAC/o1rDrQc80892vJ1+L6H+//reiNinj+G9l3/d5jb4u7eeXo+z09bc39HHl2BExEVUH/xI+mGarmf669KxjYclLaW6fVLdZa6+e8D1S9ri+SywjaQAhgEh6btUy6X3680tq2bMBnar+ftDwG8H8gQdsxnfwO+AXdKaDUkjJf098DSwac0+0cQ6099Jta+HpGGS1gQWUK21e9wGfL3mWMBGktYD7gEOlDQifWrv24/2/gH4bNrXXY1qLfT0AF4vVJvqx1FtwZB+f4dqX7avQPV+Pa10D3BQWnajgE9RbX0AjFPVg/EDwEFUB7FyanuefQm4q6/Xk5Y9kjah2iqZlErXU4WP9B5Ymeo+aX0u83Q8oK8ecP31JardyQ9HxJiI2Bh4kWoL5naq98zI1J510lbDbEkHpGGrqIkzTDVuA/aQtHbaKt0jDeu35SLsETGX6mjzJElPUIX/I2kT6Ujgv1UdoKt3NPM7wGckTaPapP1YRPyJardguqSzIuJ24ErgwTTetcAa6eDO1VQ9lH5NtasBZPfZz6PaDZhO1aX3koh4YoAv+36qXng9YZ9K9Wn+QJ3xLwV+ofceoGuVyVQ9sh4H7gK+GxH/N9UepDrYOZ0qBJMhu89+EfDBtOVyPNWxEyRtKOmWmvF+LelJ4Cbg2+mgFMDFwGaqvujyKqrjCkH9Zb4LcAjVB8FgvtBxYs9rqm0b8E8RcSvVh1d32hXtOS5yCHBseq8+QDqWVG+fXdKxkmZT/X+fkHRhGt7V8zjtUp2RXtsjwOkDPA7kXm/tpOq87qUR8ds2N6V4qq53ODwiDm9zU4bMcrFmN7PmOeztdT0wq92NMKD6P1zf7kYMJW/GmxXCa3azQjjsZoVw2M0K4bCbFcJhNyvE/wf2LXmsuMsJUgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "image_name = \"8_1.jpg\" \n",
    "label = np.array([8]).reshape(1,1)\n",
    "\n",
    "fname = \"dataset/\" + image_name\n",
    "\n",
    "image_data =255 - np.asarray(Image.open(fname).convert('L').resize((28,28)))\n",
    "image_flattened = image_data.reshape(image_data.shape[0]*image_data.shape[1],-1)\n",
    "image_norm =(image_flattened/255.)\n",
    "\n",
    "label_encoded = one_hot_encoding(label)\n",
    "\n",
    "pridected_label,pred_prob,acc = predict(image_norm, label_encoded, parameters)\n",
    "\n",
    "plt.title(\"True Label: \"+ str(label.squeeze()))\n",
    "plt.xlabel(\"Predicted: %d | With Prob: %.4f | Acc: %.1f\"%(pridected_label.squeeze(), pred_prob.squeeze(),acc))\n",
    "plt.imshow(image_data, interpolation ='nearest',cmap='binary')\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}