Tensorflow #3 - Confusion Matrix.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "#3 - Confusion Matrix",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyMi1w6Bz2mEPJulFn6PGtWM",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU",
    "gpuClass": "standard"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/EmmanueleVilla/Tensorflow---Serie-youtube/blob/main/Tensorflow%20%233%20-%20Confusion%20Matrix.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "id": "TxhKl80mbicO",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "402c8925-6631-49ed-88a7-cb830dc1371e"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'2.8.2'"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            }
          },
          "metadata": {},
          "execution_count": 2
        }
      ],
      "source": [
        "import tensorflow as tf\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "import random as random\n",
        "tf.__version__"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# 1. Import the dataset\n",
        "mnist = tf.keras.datasets.mnist\n",
        "(train_data, train_labels), (test_data, test_labels) = mnist.load_data()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Uk6JkaiXc3OR",
        "outputId": "c506930c-64ac-4f17-858f-a598b74d304b"
      },
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/mnist.npz\n",
            "11493376/11490434 [==============================] - 0s 0us/step\n",
            "11501568/11490434 [==============================] - 0s 0us/step\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "len(train_data), len(test_data)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7k6nIPK8dHfh",
        "outputId": "17698e0f-436d-4339-e7f8-eda773ee8901"
      },
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(60000, 10000)"
            ]
          },
          "metadata": {},
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "train_data.shape"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jlrge4hUdaKI",
        "outputId": "f908b85b-706b-47bf-b82c-ccc044e0dd0b"
      },
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(60000, 28, 28)"
            ]
          },
          "metadata": {},
          "execution_count": 5
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# 2. Show some data example\n",
        "plt.figure()\n",
        "plt.imshow(random.choice(train_data))\n",
        "plt.colorbar()\n",
        "plt.grid(False)\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "id": "fTtWdv0ndniZ",
        "outputId": "68c94908-9bec-4a2e-a762-9d40df2f5039"
      },
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAATEAAAD4CAYAAACE9dGgAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAXlUlEQVR4nO3dfbBd1Xnf8e8PIaRYvInKkQVSDMFyU2E3gtxiF2MHl8YG6qnA6VDIBMuY9OIOSoDxNMXMdMBNmTIdI+J4YmoRsEWKTTQBjGzTyLLC1AXH2JIqC70Uo3FEkRCSBR4k4xh07336x9nXnKtzzzr7nte9rn4fz557zn723mvNkf14rbXXXlsRgZlZro4bdAXMzDrhJGZmWXMSM7OsOYmZWdacxMwsa8f3s7ATNCtmM6efRZodU37Ba7wRr6uTa3z4g3Pi5VdGSx27aevr6yLikk7K61RHSUzSJcDngBnAX0TEnanjZzOH9+jiToo0s4SnY0PH13j5lVG+v+7XSh07Y8Fz8zousENtdyclzQD+HLgUWAJcLWlJtypmZoMRwFjJ/7QiaZGkJyTtkLRd0o3F/tsl7ZW0pdguqzvn05J2SXpW0odbldFJS+x8YFdE/Lgo+CFgGbCjg2ua2YAFwZEo150sYQT4VERslnQSsEnS+iJ2d0R8tv7goiF0FXAOcDrwbUnvjGheoU4G9s8AXqj7vqfYN4GkYUkbJW08wusdFGdm/dKtllhE7IuIzcXnw8BOJskTdZYBD0XE6xHx98Auag2mpnp+dzIiVkXEUEQMzWRWr4szsw4FwWiU24B5442UYhtudl1JZwLnAk8Xu1ZI2irpfklzi32lGkf1Oklie4FFdd8XFvvMLHNjRKkNODjeSCm2VZNdT9KJwMPATRFxCLgHOBtYCuwD7mq3rp0ksR8AiyWdJekEav3YtR1cz8wqIIBRotRWhqSZ1BLYgxHxCEBE7I+I0YgYA+7lzS7jlBtHbSexiBgBVgDrqPVz10TE9navZ2bVMYWWWJIkAfcBOyNiZd3+BXWHXQFsKz6vBa6SNEvSWcBi4PupMjqaJxYRjwOPd3INM6uWAI50b4mu9wHXAM9I2lLsu5XalKylRXG7gesBImK7pDXUZjmMADek7kxCn2fsm1n1xRS6ii2vFfEkMNkTBE0bPxFxB3BH2TKcxMxsooDRjNZKdRIzswlqM/bz4SRmZkcRo5P2AKvJSczMJqgN7DuJmVmmavPEnMTMLGNjbomZWa7cEjOzrAViNKOV653EzKyBu5Nmlq1AvBEzBl2N0pzEzGyC2mRXdyfNLGMe2DcrvPjHFzSNPXPTFzq69tl/e20yPnZ4ZtPYkv+yJ3nuyN4X26rTdBAhRsMtMTPL2JhbYmaWq9rAfj6pIZ+amllfeGDfzLI36nliZpYrz9g3s+yN+e6kmeWq9gC4k5hNE8e/bX4y/vwX5iXjT53f/J2oR+KEtuo07qNLtiTjm//DeU1jx/I8sFYCccSPHZlZriLwZFczy5k82dXM8hW4JWZmmfPAvpllK5AXRTSzfNVe2ZZPasinpmbWJ355rmVkxsknJ+NzH3k9Gf/a2x9oUULzuWDf/PkpyTP/0xc/lowvWv1cMn78TzYl4za54BiasS9pN3AYGAVGImKoG5Uys8E61lpiH4yIg124jplVQISOnZaYmU0/tYH9Y+exowC+JSmAL0bEqqMPkDQMDAPM5i0dFmdmvZfXGvud1vTCiDgPuBS4QdIHjj4gIlZFxFBEDM1kVofFmVmv1Qb2VWprRdIiSU9I2iFpu6Qbi/2nSVov6bni79xivyT9maRdkrZKav4Uf6GjJBYRe4u/B4BHgfM7uZ6ZVcMox5XaShgBPhURS4D3UmvsLAFuATZExGJgQ/Edag2ixcU2DNzTqoC2k5ikOZJOGv8MfAjY1u71zKwaxmfsd6MlFhH7ImJz8fkwsBM4A1gGrC4OWw1cXnxeBjwQNd8DTpW0IFVGJ2Ni84FHJY1f5ysR8TcdXM96YMZb35qML/j6PyTj/33REx2V/+6nPt40dtYdI8lzT//hd5Px0XYqZKVM4UUh8yRtrPu+arKxcQBJZwLnAk8D8yNiXxF6iVo+gVqCe6HutD3Fvn000XYSi4gfA7/Z7vlmVk0RcGSsdBI7WGZ+qKQTgYeBmyLiUNH4KcqLKG4OtsVTLMxsglp3snt3JyXNpJbAHoyIR4rd+yUtiIh9RXfxQLF/L7Co7vSFxb6m8rmPamZ9M1o8P9lqa0W1Jtd9wM6IWFkXWgssLz4vBx6r2/+x4i7le4FX67qdk3JLzMwmGJ9i0SXvA64BnpE0/lKEW4E7gTWSrgOeB64sYo8DlwG7gJ8D17YqwEnMzI7Sve5kRDwJTZtsF09yfAA3TKUMJzEza+A19q0y9t8/Nxl/bFFns2J+44k/SMYXX7u9aWzsyBsdlW29Ubs7eew8O2lm04yXpzaz7Lk7aWbZ6vLdyZ5zEjOzBl4U0cyyFSFGnMTMLGfuTppZtjwmZn03elHzxS/XLf18i7NnJ6O3vPTPkvF3/sHOZNxzwfLkJGZm2fI8MTPLnueJmVm2ImCk/KKIA+ckZmYN3J00s2x5TMzMshdOYmaWMw/sW1fp+PQ/08itrzSNnXJceh7YvtH0K9t2fOIfJ+Njv0jPE7P8RHhMzMyyJkZ9d9LMcuYxMTPLlp+dNLO8RW1cLBdOYmbWwHcnzSxb4YF9M8udu5PWVcedOCcZX3/Ow21f+18+lX7Z8lk//GHb17Z85XR3smWbUdL9kg5I2la37zRJ6yU9V/xNv6HVzLIRUUtiZbYqKNPx/TJwyVH7bgE2RMRiYEPx3cymibFQqa0KWiaxiPgOcPRzLcuA1cXn1cDlXa6XmQ1QRLmtCtodE5sfEfuKzy8B85sdKGkYGAaYzVvaLM7M+iUQYxndney4phER1Cb5NouvioihiBiayaxOizOzPoiSWxW0m8T2S1oAUPw90L0qmdlATcOB/cmsBZYXn5cDj3WnOmZWCRk1xVqOiUn6KnARME/SHuA24E5gjaTrgOeBK3tZyenuuNnpNb923bKkxRX+tu2y44XqjlMe95Z03bRwQTL+o+G3tl32O//zjmR89NChtq+dg6q0sspomcQi4uomoYu7XBczq4AAxsa6k8Qk3Q98BDgQEe8q9t0O/DvgJ8Vht0bE40Xs08B1wCjwRxGxrlUZ+dyCMLP+CCBUbmvtyzTOMwW4OyKWFtt4AlsCXAWcU5zzBUkzWhXgJGZmDbo1T6zJPNNmlgEPRcTrEfH3wC7g/FYnOYmZWaPyA/vzJG2s24ZLlrBC0tbiscbxxxbPAF6oO2ZPsS/JD4Cb2VGmNH3iYEQMTbGAe4A/oZYG/wS4C/jEFK/xS26JmVmjHk6xiIj9ETEaEWPAvbzZZdwLLKo7dGGxL8ktsQr4fzefl4xvv+bzbV/7d7b/bjL+js9sTcbH2i65ZuTi32oa2/2vZibPvfT9/ycZv/v0NW3VqYx3v/aHyfjbb/tuz8oeuIDo0t3JyUhaUPfY4hXA+Ao5a4GvSFoJnA4sBr7f6npOYmY2ia5NsZhsnulFkpZSa8vtBq4HiIjtktYAO4AR4IaIGG1VhpOYmTXq0mz8JvNM70scfwdwx1TKcBIzs0YVeaSoDCcxM5tofLJrJpzEzKxBVRY8LMNJzMwa9fDuZLc5iZlZA7klZvVmzP/VZPyTv//Njq6//Y2RprHZn06/7m3std0dlf38Zy5Ixjdet7JpbJbS88RaWfXqmcn48Cm7m8a+dGhR0xjAr//lvmS85X3/nFVorbAynMTM7CilV6ioBCcxM2vklpiZZa3T5836yEnMzCbyPDEzy53vTppZ3jJKYl5PzMyy5pZYH+y8Mz0n6bFT/2cyPtZilHX5525uGnvbphbrXh2Xfg/D7s+klzjfdG3zeWCQngvWap7Xl+76SDL+8lB6ttbwv76naWzNv5/s3RVvmrFrczI+3bk7aWb5CvzYkZllzi0xM8uZu5NmljcnMTPLmpOYmeVK4e6kmeXOdyePLZp5QjL+mX/+WEfX/x8t1r5625+2/w7EA598TzK+7ROt3nmZXhMsNRfsm1eky/5Hz/5dMn75jb9IxlNm355eL2zs+nck46PP7mq77Bzk1BJrOWNf0v2SDkjaVrfvdkl7JW0ptst6W00z66sevgG828o8dvRlYLLpzXdHxNJie7y71TKzgYk3x8VabVXQMolFxHeAV/pQFzOrimnWEmtmhaStRXdzbrODJA1L2ihp4xFe76A4M+sXjZXbqqDdJHYPcDawFNgH3NXswIhYFRFDETE0k1ltFmdmNrm2klhE7I+I0YgYA+4F0ksdmFlepnt3UtKCuq9XANuaHWtmmclsYL/lPDFJXwUuAuZJ2gPcBlwkaSm1XLwbuL6Hday80QvOScavPik936mVz9/z0WR8Pu3PE9MlL7d9bhnf+L33N42NPbujo2t//NSNLY74laaRRxd/I3nmu3/vD5Pxt982veeJVaWVVUbLJBYRV0+y+74e1MXMqmI6JTEzO7aI6tx5LMNJzMwmqtB4Vxl+UYiZNerS3ckmjy2eJmm9pOeKv3OL/ZL0Z5J2FXNQzytTVScxM2vUvSkWX6bxscVbgA0RsRjYUHwHuBRYXGzD1OajtuQkZmYNujXFoslji8uA1cXn1cDldfsfiJrvAaceNZ1rUh4T64Kf3XKop9d/21OvJuOp/y6Nvf/c5Llrl7Zaaqf5NAWAT77w2+nTn3u+xfWbm3HqKcn4RU+uSMZ3/nbzm+h7Rv4hee7M15Lh6a+3Y2LzI2J8LaSXgPnF5zOAF+qO21PsS66b5CRmZhPFlO5OzpNUP2FvVUSsKl1UREid3UZwEjOzRuXTysGIGJri1fdLWhAR+4ru4oFi/16gfgXQhcW+JI+JmVmDHj92tBZYXnxeDjxWt/9jxV3K9wKv1nU7m3JLzMwadWlMrMlji3cCayRdBzwPXFkc/jhwGbAL+DlwbZkynMTMbKIurlDR5LFFgIsnOTaAG6ZahpOYmU0g8pqx7yRmZg2cxI4x7zrtpZ5e/wNfSi8587/+afO5XPtufiN57vwZ6XlgX2rxuriXLp+TjI+9tr9pLC74zeS5J/zX9O+68x3pxVRWvvIbTWNfv/1fJM89/eH2lzeaFpzEzCxrTmJmlq3MVrFwEjOzRk5iZpYzL4poZllzd9LM8lWh17GV4SRmZo2cxI4tL340ve7VBV9s9uRFzXfP/Woy/vunbkrGn1r4b5vG/njJuuS5rZx8XHrdrd3Xnp2M/5ur9jSNXXVKeuHOd8xMvzH+t+5Kv1Zt4QPPNo3NOfh08txjmWfsm1n2NJZPFnMSM7OJPCZmZrlzd9LM8uYkZmY5c0vMzPLmJGZm2Zra244GzkmsC0b2vpiMH9xzfvoC6VdDsqDFml9/9b2Hm8ZmaWb64i387okH0/EV6fdWHonRprGbX2xYoXiCrXen1xtb8NDfJeOjkVFzokJymyfW8m1HkhZJekLSDknbJd1Y7D9N0npJzxV/5/a+umbWFxHltgoo88q2EeBTEbEEeC9wg6QlwC3AhohYDGwovpvZNNDjV7Z1VcskFhH7ImJz8fkwsJPaq8WXAauLw1YDl/eqkmbWRzGFrQKmNCYm6UxqIzhPA/PrXmz5EjC/yTnDwDDAbN7Sbj3NrI+m5cC+pBOBh4GbIuKQpF/GIiKkyRuXEbEKWAVwsk6rSO42s5SckliZMTEkzaSWwB6MiEeK3fslLSjiC4ADvamimfVVkNXAfsuWmGpNrvuAnRGxsi60FlhO7ZXky4HHelLDaeCffP7VZHz7h0eS8XNOSP8zdTqNIuVHR9KvfPvIhvRyOKeva173E9d8L3nuyaTj1jtVGbQvo0x38n3ANcAzkrYU+26llrzWSLoOeB64sjdVNLO+m05JLCKepDb/bTLp2Ypmlp3cJrt6xr6ZTRThRRHNLHP55DAnMTNr5O6kmeUrAHcnzSxr+eQwJ7F+GN3e/NVhAH90c3qu1erPrUzGFx6fXqon5bzPpcte9PWfJOPv3Lmx7bKtutydNLOsdfPupKTdwGFgFBiJiCFJpwF/BZwJ7AaujIiftnP9Uo8dmdkxpDerWHwwIpZGxFDxvWtLeTmJmdkEtcmuUWrrQNeW8nISM7NGYyU3mCdpY902PMnVAviWpE118VJLeZXhMTEzazCFVtbBui5iMxdGxF5Jvwqsl/R/64OppbzKcEvMzCbq8phYROwt/h4AHgXOp4tLeTmJmdlRas9OltlakTRH0knjn4EPAdt4cykv6HApL3cnK+BXvvb9ZPyTX7uwZ2WfzneT8eYvXLNprXsLHs4HHi1Wgj4e+EpE/I2kH9ClpbycxMxsoi6+PDcifgw0vEA0Il6mS0t5OYmZWaOKLD1dhpOYmTXKJ4c5iZlZI43l87ojJzEzmygYn8iaBScxM5tAdPxIUV85iZlZIycxM8uak5iZZctjYmaWO9+dNLOMhbuTZpaxwEnMzDKXT2/SSczMGnmemJnlLaMk1nJRREmLJD0haYek7ZJuLPbfLmmvpC3Fdlnvq2tmPRcBo2Pltgoo0xIbAT4VEZuLFRo3SVpfxO6OiM/2rnpmNhAZtcRaJrHijST7is+HJe0Ezuh1xcxsgDJKYlNaY1/SmcC5wNPFrhWStkq6X9LcJucMj7/O6Qivd1RZM+uDAMai3FYBpZOYpBOBh4GbIuIQcA9wNrCUWkvtrsnOi4hVETEUEUMzmdWFKptZbwXEWLmtAkrdnZQ0k1oCezAiHgGIiP118XuBb/SkhmbWX0FlBu3LKHN3UsB9wM6IWFm3f0HdYVdQew2TmU0HEeW2CijTEnsfcA3wjKQtxb5bgaslLaWWt3cD1/ekhmbWfxVJUGWUuTv5JKBJQo93vzpmNnjVaWWV4Rn7ZjZRAF6Kx8yy5paYmeUrsro76SRmZhMFREXmgJXhJGZmjSoyG78MJzEza+QxMTPLVoTvTppZ5twSM7N8BTE6OuhKlOYkZmYTjS/FkwknMTNrlNEUiyktimhm018AMRaltjIkXSLpWUm7JN3S7fo6iZnZRNG9RRElzQD+HLgUWEJt9Zsl3ayuu5Nm1qCLA/vnA7si4scAkh4ClgE7ulVAX5PYYX568Nvx18/X7ZoHHOxnHaagqnWrar3AdWtXN+v29k4vcJifrvt2/PW8kofPlrSx7vuqiFhV9/0M4IW673uA93Rax3p9TWIR8db675I2RsRQP+tQVlXrVtV6gevWrqrVLSIuGXQdpsJjYmbWS3uBRXXfFxb7usZJzMx66QfAYklnSToBuApY280CBj2wv6r1IQNT1bpVtV7gurWrynXrSESMSFoBrANmAPdHxPZulqHI6BkpM7OjuTtpZllzEjOzrA0kifX6MYROSNot6RlJW46a/zKIutwv6YCkbXX7TpO0XtJzxd+5Farb7ZL2Fr/dFkmXDahuiyQ9IWmHpO2Sbiz2D/S3S9SrEr9brvo+JlY8hvAj4HeoTXz7AXB1RHRtBm8nJO0GhiJi4BMjJX0A+BnwQES8q9j334BXIuLO4v8A5kbEf6xI3W4HfhYRn+13fY6q2wJgQURslnQSsAm4HPg4A/ztEvW6kgr8brkaREvsl48hRMQbwPhjCHaUiPgO8MpRu5cBq4vPq6n9j6DvmtStEiJiX0RsLj4fBnZSmzk+0N8uUS/rwCCS2GSPIVTpHzKAb0naJGl40JWZxPyI2Fd8fgmYP8jKTGKFpK1Fd3MgXd16ks4EzgWepkK/3VH1gor9bjnxwH6jCyPiPGpP3d9QdJsqKWpjAVWaI3MPcDawFNgH3DXIykg6EXgYuCkiDtXHBvnbTVKvSv1uuRlEEuv5YwidiIi9xd8DwKPUur9Vsr8YWxkfYzkw4Pr8UkTsj4jRqL208F4G+NtJmkktUTwYEY8Uuwf+201Wryr9bjkaRBLr+WMI7ZI0pxhwRdIc4EPAtvRZfbcWWF58Xg48NsC6TDCeIApXMKDfTpKA+4CdEbGyLjTQ365Zvaryu+VqIDP2i1vIf8qbjyHc0fdKTELSr1NrfUHtkayvDLJukr4KXERtqZb9wG3A14A1wK8BzwNXRkTfB9ib1O0ial2iAHYD19eNQfWzbhcC/xt4Bhhfue9WauNPA/vtEvW6mgr8brnyY0dmljUP7JtZ1pzEzCxrTmJmljUnMTPLmpOYmWXNSczMsuYkZmZZ+/9RWGVrojVe1AAAAABJRU5ErkJggg==\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# 3. Create the neural network model\n",
        "\n",
        "model = tf.keras.models.Sequential(\n",
        "    [\n",
        "        tf.keras.layers.Flatten(input_shape=(28,28)),\n",
        "        tf.keras.layers.Lambda(lambda x : x / 255.0),\n",
        "        tf.keras.layers.Dense(56, activation=\"sigmoid\"),\n",
        "        tf.keras.layers.Dense(10, activation=\"softmax\")\n",
        "    ]\n",
        ")"
      ],
      "metadata": {
        "id": "RS79uMp3d6MI"
      },
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# 4. Compile the model\n",
        "model.compile(\n",
        "    loss=tf.keras.losses.SparseCategoricalCrossentropy(),\n",
        "    optimizer=tf.keras.optimizers.SGD(),\n",
        "    metrics=[\"accuracy\"]\n",
        ")"
      ],
      "metadata": {
        "id": "3ANs0x9jeZf6"
      },
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# 5. Fit the model\n",
        "history = model.fit(train_data, train_labels, epochs=10, verbose=0)"
      ],
      "metadata": {
        "id": "rFd9CYGdfv91"
      },
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# 6. Evaluate the model\n",
        "loss, accuracy = model.evaluate(test_data, test_labels)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "F-Cp6WOkgEbv",
        "outputId": "17187c7c-4600-4cd8-bcca-5ed3985d2e9f"
      },
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 2s 3ms/step - loss: 0.3098 - accuracy: 0.9146\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# 7. Plot the training history\n",
        "pd.DataFrame(history.history)[[\"accuracy\"]].plot()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "id": "Ra1k-TkLgew2",
        "outputId": "544598e4-8612-4a0e-9a4d-570e3dc170a8"
      },
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7f99771a0a90>"
            ]
          },
          "metadata": {},
          "execution_count": 11
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# 8. Predict the test data\n",
        "test_preds = model.predict(test_data)"
      ],
      "metadata": {
        "id": "LgBc0zQPhHAF"
      },
      "execution_count": 12,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.metrics import confusion_matrix"
      ],
      "metadata": {
        "id": "_NyQK584_aOi"
      },
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "test_labels.shape, test_preds.shape"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "WshJ_h-F_xKB",
        "outputId": "0deb25fb-0ca9-40f8-ed47-ac1c13b224f7"
      },
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((10000,), (10000, 10))"
            ]
          },
          "metadata": {},
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "test_labels[0], test_preds[0]"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7Diz7qOt_1lp",
        "outputId": "e821c943-f515-4cbd-b803-31ba1abe579e"
      },
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(7, array([3.6866590e-04, 5.8298992e-06, 5.8434927e-04, 2.7033514e-03,\n",
              "        3.1450265e-05, 1.9571121e-04, 3.1361535e-06, 9.9341702e-01,\n",
              "        5.8305257e-05, 2.6323241e-03], dtype=float32))"
            ]
          },
          "metadata": {},
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "test_preds_sparse = tf.argmax(test_preds, axis=1)"
      ],
      "metadata": {
        "id": "JPXKl6ou_7Ih"
      },
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "test_preds_sparse[:10]"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "cnIcI8rIAVfJ",
        "outputId": "3ef69866-253c-41bb-a3fe-c1f98eab3a5b"
      },
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(10,), dtype=int64, numpy=array([7, 2, 1, 0, 4, 1, 4, 9, 6, 9])>"
            ]
          },
          "metadata": {},
          "execution_count": 17
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "matrix = confusion_matrix(test_labels, test_preds_sparse)\n",
        "matrix"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "GJFQf1AtAWmh",
        "outputId": "f1ab4f5c-6981-4483-bfd4-c4368cfb6d33"
      },
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[ 958,    0,    3,    2,    0,    4,    9,    1,    3,    0],\n",
              "       [   0, 1105,    2,    3,    0,    1,    4,    1,   19,    0],\n",
              "       [  13,    5,  908,   14,   16,    2,   14,   15,   39,    6],\n",
              "       [   3,    0,   19,  913,    0,   31,    3,   15,   18,    8],\n",
              "       [   1,    3,    5,    0,  917,    1,   11,    2,    5,   37],\n",
              "       [  11,    3,    6,   49,    9,  764,   13,    6,   24,    7],\n",
              "       [  15,    3,    3,    1,   13,   17,  900,    1,    5,    0],\n",
              "       [   4,   10,   28,    6,    8,    0,    0,  934,    3,   35],\n",
              "       [   9,   11,    8,   23,   11,   21,   13,   11,  854,   13],\n",
              "       [  14,    6,    5,   11,   44,   10,    0,   21,    5,  893]])"
            ]
          },
          "metadata": {},
          "execution_count": 18
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.metrics import ConfusionMatrixDisplay"
      ],
      "metadata": {
        "id": "ZVnD6ldQAdQg"
      },
      "execution_count": 19,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "ConfusionMatrixDisplay.from_predictions(test_labels, test_preds_sparse)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        },
        "id": "kZK2RYmAAnjo",
        "outputId": "0cfdf875-d1c0-41be-9bcb-e777ee4cd7bc"
      },
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7f98fa39b350>"
            ]
          },
          "metadata": {},
          "execution_count": 20
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def plot_wrong_labeled_data(X_test, y_true, y_preds, label_to_check, wrong_label = None):\n",
        "  # Search for the wrong data\n",
        "  wrongs = []\n",
        "  for i in range(len(X_test)):\n",
        "    true_label_found = y_true[i] == label_to_check\n",
        "    wrong_label_found = wrong_label != None and y_preds[i] == wrong_label\n",
        "    wrong_preds_label = wrong_label == None and y_preds[i] != label_to_check\n",
        "    if true_label_found and (wrong_label_found or wrong_preds_label):\n",
        "      wrongs.append((X_test[i], y_preds[i]))\n",
        "  \n",
        "  # Shuffle result\n",
        "  random.shuffle(wrongs)\n",
        "\n",
        "  # Plot result\n",
        "  plt.figure(figsize=(10,10))\n",
        "  for i in range(25):\n",
        "    wrong = wrongs[i]\n",
        "    plt.subplot(5, 5, i+1)\n",
        "    plt.xticks([])\n",
        "    plt.yticks([])\n",
        "    plt.imshow(wrong[0], cmap=plt.cm.binary)\n",
        "    plt.text(14, 32, wrong[1].numpy())\n",
        "  plt.show()"
      ],
      "metadata": {
        "id": "8wJXHprdAs5I"
      },
      "execution_count": 21,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plot_wrong_labeled_data(test_data, test_labels, test_preds_sparse, 5, 3)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 592
        },
        "id": "QBTXoFx5BeHQ",
        "outputId": "8ce5f118-3fb9-4f47-cc8d-8d48018038c9"
      },
      "execution_count": 22,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 720x720 with 25 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAj0AAAI/CAYAAACcWWgKAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3debxV0//H8bXlppGoaO4WEhrQLVL9Ko0ylgr5CplSvtKskJIvMpe5b0kDUiIVJUSISlE0R4MozUWlQZ3fH/h8P2u75zj33DPcs9fr+Xh8H4/3ttfZd/1andv67TV5oVDIAAAABN1Rqa4AAABAMtDpAQAATqDTAwAAnECnBwAAOIFODwAAcAKdHgAA4ISjc1K4RIkSoczMzARVBdlZt26d2bZtmxfv59KWqbFw4cJtoVCoZLyfS3smH9/NYEnEd5O2TI1IbZmjTk9mZqZZsGBBfGqFqGRlZSXkubRlanietz4Rz6U9k4/vZrAk4rtJW6ZGpLZkeAsAADiBTg8AAHACnR4AAOAEOj0AAMAJdHoAAIAT6PQAAAAn0OkBAABOoNMDAACckKPNCYG87sCBA5LnzJkjefDgwVa5hQsXSn7ggQck33HHHQmsHQAglXjTAwAAnECnBwAAOMGZ4a3XXntN8hNPPGHdO+OMMySPHj06aXVC7j388MPW9Ycffih51qxZYT9XoUIFycuWLYt/xQDAMfPmzbOu77rrLsldu3aV3LZt26TVyY83PQAAwAl0egAAgBPo9AAAACcEek6PnsejlyIXKVLEKtevX7+k1Qm5d/nll0ueMmWKdc/zPMnly5eX3KVLF6vctddeK7l06dLxrmKgPP3009b1nj17JD///PPWvR9//DGqZ1522WWSd+/eLfmjjz6yyun21M4880zreubMmZJpz7xBbwthjDHDhg2T3K1bN8nnnHNO0uqE6M2YMUPyoUOHwpbbu3ev5DvvvNO6t3nzZsn693Yq8aYHAAA4gU4PAABwQqCHt9566y3J27dvl/zoo49a5apWrZq0OiE206ZNkzx9+nTJRYsWtcoNGDBA8vXXXy+5ePHiYZ+tX8PXqlUrN9UMjE6dOkkeO3asde/IkSNhPxduOMrPPywZ7vPhnuffZuC0006TPHXqVMkNGzaMqj6Ij6+//lqyHsI0xphNmzZJ/vnnnyX37NnTKte8eXPJv/76q+Qvvvgi6nroZyB6+nvVunVryfv374/peR07dpScV3a7500PAABwAp0eAADghEAPb02cOFHyKaecIvmKK66wyq1Zs0ayPqTST7/imzRpkmT/6oMHH3ww55WFZd26ddb1LbfcIlmvJGjfvr1Vzv+qPBoMaf3h22+/lTxhwgTJkYaz8gq9guTw4cMprIl7lixZIrlly5aSt23bFvYzH3zwQbbZGGOqV68uWR8gvGrVqrDP86/m++abbyLUGH/Zt2+fda0PX9b/3umVsMYYc9xxx0k+//zzJfv/LWzXrp3kaIe+E403PQAAwAl0egAAgBPo9AAAACcEak7PkCFDwt777rvvJJ911lnWPT2nJ1ply5aVzI6i8aeXtxpj7+ypx4b5s48fvS3Ab7/9FtVnGjVqZF0XK1YsV3UIhULWtW5rvdvzggULwtajZMmSuaoDItNzeIwxplmzZpK3bt0qOdY5HHo+TrVq1ST7fydohQoViulnuW7y5MnWtT7FQPNv83LllVcmrE6JxpseAADgBDo9AADACYEa3vIfVqidcMIJkvPnz2/d08Nd9erVk+wfOrnoooskFyhQQLJevofkWr16daqr4JwTTzxR8nPPPWfd0zsjx5teAr127Vrr3hlnnCG5cOHCCasDjBk6dKh1rYee9fBkRkaGVU7/PvVvG6L16tUrt1VEBPp7dNNNN4Utp//N1EOY6Y43PQAAwAl0egAAgBPo9AAAACcEak5PJLNmzZJcpUoV617BggWTXR38g9KlS4e91ktXR4wYEbbcvffem6DauW3kyJGS4z2H55FHHrGu/UcU/EXP4THGmA4dOkiuWrWq5GOPPTaOtYMxf98uINzS9N69e1vX+ogDJNfOnTslDxgwQHKkrSl27Ngh2f9900ctderUKR5VTBre9AAAACfQ6QEAAE5wZnirVKlSkhnOyvsyMzOt6xdeeEFymzZtJOsT140x5sUXX5TcunVryXpnV2SvePHikvPlyyfZf2q5Hs7wvx4P991asWKFdd2/f/9sy/l3iA03dPLhhx9a108//bTkV199VXI67xybV1WsWNG6Dneiud7WA6n13nvvSX7++eej+ozeouWdd96x7unv7w8//CB54MCBMdYweXjTAwAAnECnBwAAOCFQw1v+wwq1OXPmSNbDI0gPF198sWS9s+v8+fOtcnplV/fu3SXrwzSNMeaYY46JdxXT3i233CK5T58+kn/99VernG6LVq1aWfeuvvpqyTNnzpT8+eefW+W+//773FU2gm7duknWu0cbY0zjxo0T9nNd4d9NeerUqdmWe/jhh61rvepL77pcv379ONYO2Rk3bly2/71SpUrW9ZQpUyTrVZAzZsywyl1yySWSBw8eLDkrK8sqp39X5BW86QEAAE6g0wMAAJxApwcAADghUHN6mjRpYl3rOQXPPPOMZD1Waczfd5tE3vbWW29J9s/R0Keu6124/XNKmNsRmZ7z5J/To7377rsRr+OpaNGikv3L6Pft2yd569atkrt06WKV08ur/aeAIzr+35cNGzaU/PHHH0v2b2eg5/7onbb1/DFj7B2DER/637+rrrpKcr169axy/jk+f6lTp451fe6550qeN2+eZP8WInkRb3oAAIAT6PQAAAAnBGp4y/+adOHChZInTpwo2b/MVi/H8w99Ie/Rh4rec8891r3rrrsu28/o3ZmNMWbXrl3xr1iA6GGqRx991Lqnv0vx1qhRI+u6RYsWkrt27Sp548aNVrnHHntMsj4QddWqVVa5G2+8UfKYMWNyVVdX1apVy7rWw8iDBg2SrHdRN8b+zunhSP8uvnrrkfvuuy9XdcUf9A73/t3uo+Hf+uHmm2+WrIe3/L8b/L938wLe9AAAACfQ6QEAAE4I1PCW33PPPZftf/e/gmvatKnkTz75RHLlypUTUzHEzWWXXWZd16hRQ/LixYsl//LLL1Y5vSJJrwrCH/QQhn/YWH8v9GodY+wdsdevXx/2+Xp1mB5e1iuBIqlSpYp1PXz4cMlr166V7D+YVK8g0sPfxvx92AY5p4ej/ENTQ4cOldyjR4+wz9DDYgxv5U3+lXl/+fnnn5Nck5zjTQ8AAHACnR4AAOAEOj0AAMAJgZ7TU7x4ccmvv/56tv/dGGOef/55yXp5rv7vyJv883H0XA+9+67fiBEjJOvT2PF3+lR7//WBAwese0eOHJH8+++/h32m53mSixQpktsqWp599lnJ/lOe9enueqmtMczpiQf992H27NnWPT1fMpJYllQjsXbv3m1dDxs2LNtyLVu2TEZ1coU3PQAAwAl0egAAgBMCPbwVztVXX21dv/rqq5JHjRol+fLLL7fK6d1hYcyGDRsk66HABx98MBXVyZFTTz011VUIBL30PK/QQ5zVqlWz7unhrSDbuXOn5P3790vWu5nHy5IlSyTrwyyXL18e1efPPPNM6/ree++NT8UQN3p6iDH2wc6FCxeWrLd/yat40wMAAJxApwcAADjByeGt888/37o+++yzJesdZvVKFPx9B9sLL7xQ8vbt2yXrV+vGJHYVnP/gSX1QZiT+VT0IjvHjx0vWh2H6RTv8ko7efPNNyRUqVJAcj+Gtu+66y7rWOyjrnc/1Cj1jjKlYsaLkdu3aSe7Zs6dVzn+4JVJD76785JNPhi2ndzn3r/TMi3jTAwAAnECnBwAAOIFODwAAcIIzc3r0PJMHHnjAuqd3Ds3KypLcpEmTxFcsjemTykOhkGQ9xm+MPeb71ltvWff00vFixYqF/Vn79u2TrOcP+bcV2Lt3b7afZw5Pctx+++2S9Snrffv2tcrVr18/rj93x44dkvWu6vrvqN+4ceOs66effjqudUqmLl26WNcjR46UPHPmTMn+07H1Trv+HZRnzJghefTo0VHVQ8/jue6666x7AwYMkMyuy3mT/k785z//kbxixQqrnJ4bVrdu3cRXLI540wMAAJxApwcAADgh0MNb+pWsfuW9dOlSq1zZsmUl650n8+fPn8DapR//gYx6GaM+tNN/COWmTZskn3feedY9vXtujRo1wv5svfuzPihSD6sZY79er169umR9wCgSZ9WqVZI//PDDbLMx9vBWs2bNJDds2NAqp//O5cuXT7L/MFk9dLJo0aKo6tqpU6eoyqUD/5Dy8ccfL/mll16S7N/teM6cOZIjfZd0PuGEE6xybdq0kazbsm3btlHVHcn18ssvS54yZYp1b/r06ZL1Tt6lSpUKW65AgQJxrmFi8aYHAAA4gU4PAABwQloMb/3www/WtV6JpVcGTZo0ySqnX3Pr17NnnXWWVe61116TXLly5dxV1iGdO3eWXKdOHckdOnSwyunD6fz0cMjKlSsl+3dzjdYVV1whWa9gKVq0aEzPQ86UKFEi2//uH/IMN/TlH2K55JJLJOvhLf9Oy5FWaWkZGRmS9UrNdKd3ODbGmDfeeEOyf5VaOP7DYxs1aiRZf6/8Q9T+Q10RX3oIaujQoZJr1qwZ9jP+4WS9+/ihQ4fCfq5IkSKS9cHc/mHR008/PUKN8zbe9AAAACfQ6QEAAE6g0wMAAJyQZ+b0HD582Lp+8MEHJftPeN21a1e2z9DLNI2xT0/v06eP5Msuu8wqV7BgwZxVFn+jT9edO3eudW/Pnj2S9QnYxhizefNmyfqE+6+++irsz7rlllsk65PejbGXzNKuyae/Z7o9dTvnxLRp03JVHz2HxxhjevfuLVnPWUh3Tz31lHXdokULyXpnbP+WACVLlpSs50wZY8wZZ5wRzyoiRt26dZO8bt06yf55bdHS34kLLrjAuqe3HtF/h4KENz0AAMAJdHoAAIAT8szw1tatW63r++67T7J/GWvt2rUl69fV/ld1xYsXj2cVESX/waH6ulevXsmuDpJIbwehdz7X20IYY281oQ/EjIejj/7frzX/st7BgwfH9WflFfoASGOCtdu06/TBrXpbliVLlljlbrjhBsn+A12vuuoqyXoY8+STT45XNdMGb3oAAIAT6PQAAAAn0OkBAABOyDNzevynuB45ciRFNQEQD3oLiS5dulj3brzxRsl6uXW/fv1i+lldu3aVrI+X6NixY0zPA/KKgQMHZpsRG970AAAAJ9DpAQAATsgzw1sA3KFP9O7bt2+2GQDijTc9AADACXR6AACAE+j0AAAAJ9DpAQAATqDTAwAAnECnBwAAOIFODwAAcAKdHgAA4AQ6PQAAwAleKBSKvrDnbTXGrE9cdZCNiqFQqGS8H0pbpgztGRy0ZbDEvT1py5QJ25Y56vQAAACkK4a3AACAE+j0AAAAJ9DpAQAATqDTAwAAnECnBwAAOMG5To/neQU8z5vved5iz/OWep43KNV1Qmxoy2ChPYODtgyWILWnc0vWPc/zjDGFQ6HQHs/zMowxnxljuoVCobkprhpyiLYMFtozOGjLYAlSex6d6gokW+iPXt6ePy8z/vyfWz2/gKAtg4X2DA7aMliC1J7ODW8ZY4znefk8z1tkjNlijHk/FArNS3WdEBvaMlhoz+CgLYMlKO3pZKcnFAodDoVCZxljyhlj6nieVy3VdUJsaMtgoT2Dg7YMlqC0p5Odnr+EQqFdxpiPjDEtU10X5A5tGSy0Z3DQlsGS7u3pXKfH87ySnucV+zMXNMY0M8asSG2tEAvaMlhoz+CgLYMlSO3p3ERmY0xpY8xoz/PymT86fRNCodC0FNcJsaEtg4X2DA7aMlgC057OLVkHAABucm54CwAAuIlODwAAcAKdHgAA4IQcTWQuUaJEKDMzM0FVQXbWrVtntm3b5sX7ubRlaixcuHBbKBQqGe/n0p7Jx3czWBLx3aQtUyNSW+ao05OZmWkWLFgQn1ohKllZWQl5Lm2ZGp7nrU/Ec2nP5OO7GSyJ+G7SlqkRqS0Z3gIAAE6g0wMAAJxApwcAADiBTg8AAHACnR4AAOAEOj0AAMAJdHoAAIAT6PQAAAAn0OkBAABOoNMDAACckKNjKAAAQHrZvHmz5LFjx0pevXq1VW748OGS27ZtK/nQoUNWufPOO0/yCSecILlZs2ZWuUqVKsVY48ThTQ8AAHACnR4AAOAEOj0AAMAJgZ7TM3HiRMnt27cPW65z586Sn3zySckFChRITMUgBg4cKHnQoEEpqcN9990nWdcHybFu3TrJzZs3t+755xz8pWnTptZ17969wz4D8fXiiy9a10WLFpXcoUMHyRs2bLDKvfzyy9k+b/v27db10KFDJR977LGSZ82aZZWrVatWdBWGGTVqlOT+/fuHLed5nuRJkyaFLTdlypRs//vZZ59tXX/66aeSCxUq9I/1TAbe9AAAACfQ6QEAAE5Iy+Et/fpz/fr1kitXrmyV69Gjh+ROnTpJPvXUU61ykydPlqyX3OlXc4iPxo0bW9cff/xxaiqi6GG1Ro0aWff814iPfv36SX7hhRck79692yqnX7drH374oXU9Z84cycuXL5dcsWLFXNUTf6hevbrkZcuWhS2nf8+GQiHrnn/Zczi6zVu1aiX5rLPOiurzMGbu3LnW9aOPPpptuXz58lnXQ4YMier548aNk7xo0SLJ/jY+cuRIVM9LJt70AAAAJ9DpAQAATkjL4a2OHTtKvuSSSyQ/9dRTYT+jh1UqVKhg3dOv5/RKEsRfXhjOisRfP4a3EkO/HvcPaWl6BWWTJk0kv/POO1a5/fv3Sz58+HA8qgjl999/l+wfttIOHjyY42cff/zx1nW3bt0k62FQ/1AMwvv222+t6507d2Zbzv9nqqeERKJX6S1dulRyiRIlrHJFihSJ6nnJxJseAADgBDo9AADACXR6AACAE9JyTo8eA9Zji/PmzbPK6R07u3fvLlmPTxtjTP369SU/9thjcasn0g87MifGG2+8YV1v27Yt23L+nZYHDx4suXbt2pJvvfVWq9zIkSMl6x2Do12Ci8jq1KkjeeXKlVF9plSpUtb15ZdfLln/3q5WrZpVrlixYrFUEYp/a5DjjjtOsp5D55+ftXfvXsmFCxcO+3zdtv52zut40wMAAJxApwcAADghLYe3fvnlF8nPPfec5CuuuMIqN378eMmbN2+W7N8lsnTp0vGuIuLso48+khxp2XuqDi3F3+3YsUPyQw89ZN07cOBAtp/p06ePdX3uuedmW65nz57W9YUXXih58eLFOaon/lmZMmWiKnfHHXdIvueee6x7/uXMSJxTTjnFur7qqqsk6+Ff/w7KeqqHPrDb/7x0xpseAADgBDo9AADACXR6AACAE9JyTs+mTZsk6xN/b7jhhrCf0dth33333da9WbNmSdZb3Q8YMMAq598uHTnnP9Yh3PycSKedRzoaItySc//P0dccNRE/v/32m+RmzZpJ/vrrr8N+JiMjQ/LRR0f3K6lq1aphr9u0aRPVMxC9SpUqRVWuffv2kpnDk3fcf//9kvUJ7P75b/r6k08+kVy5cmWr3FFHpe/7kvStOQAAQA7Q6QEAAE5Iy+Et/wnLf6levbp1rV+pX3nllZLz589vlcvKypL82muvSf7ggw+scu+++67k8uXL56DG+EvDhg2t63DDW/7/HstOyfozkYbLED96CWykIS1NL2enXfKmMWPGpLoKyIWSJUtKnjBhgmT/tgJ6mfrUqVMl+7d10VtEpBve9AAAACfQ6QEAAE5Iy+GtcPyv0x9//HHJeujr1VdftcqddNJJktevXy+5QYMGVrlWrVpJnj59uuRy5crFWGP3+Iepot1BOZadlvVn7rvvvoj1QOps2bJF8g8//GDdq1ChQrKrg2xs3749qnJjx46VfNttt4Utpw+p9O/CrVfQIv48z5O8aNEi616+fPkkv/3225L1ii9jjPnXv/4l+dFHH413FROKNz0AAMAJdHoAAIAT6PQAAAAnpOWcHj3uqPXr18+6rlKliuRXXnlFsp7D41exYkXJn3/+uXWvVq1akrt16yZ50qRJ/1BjpJp/TtDs2bMl6xPckTt65/MRI0ZIvvXWW61yhw8flvzII49InjFjhlVu8uTJkjMzM+NVTURh48aNkvfu3RvVZ/QJ3pF8++23kv3zuJYvXx7VMxAbfWK6/9+ujh07StZbu3zzzTdWOb2dy+jRoyVfd911catnovCmBwAAOIFODwAAcEJaDm+1bdtWcvfu3SXr4Sxj7GELvUQyWv6l6I899pjkO++8U/LWrVutcnr3S+RNesfnxo0bS2aoK3f0QYSdOnWSvH//fqucHgbRQx3+1+iXX365ZP39a9q0ae4ri4j0Qcw//vhjjj9/3HHHWdf69/OXX34pecOGDVa5mTNnSm7evHmOfy6id8YZZ1jXehrIkCFDJPu/l2vWrJHcu3dvyf7tBvLidi686QEAAE6g0wMAAJyQlsNbZcuWlfzcc89J1ocdGvP3Q9Jyq27dupJ37Ngh+aWXXrLK9e3bN64/N8j0Tsmx7LocD+EOPUX8dOnSxbquVKmSZL0S8rvvvrPK6dfq+pDD//73v1a566+/Ph7VhKKHEPUBy/7hKE0PR/oPs9SrhvTu9v5VsnqHX4a3EksPRxtjzOmnny65V69ekgsVKmSVe+GFFyT/+uuvkvWQtjH24eAZGRm5q2yc8KYHAAA4gU4PAABwAp0eAADghLSc06N17tw5JT9Xn1Trn0uE6EV72nmjRo2yzX56fo6eIxTtvB29fN0YlrAnip6fc/bZZ0tu0aKFVU4vZ9e7OK9duzaBtYMx9jYferf7UaNGWeX072A9p+f444+3yuldnY8+Ovw/Pd9//33OK4u4q1atmuQbbrjBurdr1y7Jb775pmS9U7MxxvTo0UNynz59JOs5YsnGmx4AAOAEOj0AAMAJaT+8lUx65+VjjjlG8qWXXpqK6gROtENdkYQbBvMPW4Ub7mL5evLpYZSbb77ZunfHHXdk+5n33nvPutbDLwUKFIhj7WCMPYQV65QCPVT8ySef5LpOSJ7atWtb16+++qrkJUuWSK5Ro4ZV7tlnn5Wsl73r3Z6TjTc9AADACXR6AACAExjeyoEnnnhC8mmnnSbZ/0rPRZGGhSKttkoW/yosvfoOeYf/YNJw/ENYtGfes2LFCuvaP3T5Fz1VwBhjbrnlloTVCfFXpEgRyZmZmda9devWSf7666+TVKPIeNMDAACcQKcHAAA4gU4PAABwAnN6Ipg/f751PW3aNMkvvvhisquT5+h5PP4l4eHoU9WNiX6n5dxiKXp6mDlzZlTlGjZsaF3754Ug5/zfkTPPPFNyyZIlc/w8/+/IzZs3Z1vOPw/krrvuyvHPQt7gX9qut3mpVatWsquTLd70AAAAJ9DpAQAATkjp8NayZcskz54927p30003Sc7IyEhanfSBaf/+97+te23btpXcrl27pNUpr4plOEofAprd9V/iMQymd3gO93P+6eciMX744QfJ11xzjeRIy1pr1qwp+fTTT09MxRyjDwF95JFHrHvPPfec5GiHt/SUgNdffz1sufz580vu0KFDVM9GZPqgVv+BvPqAV/090u1gjDGFCxeO6meNHz9e8pgxYyTPmDEj7GeysrKienai8aYHAAA4gU4PAABwQkqHt0aNGiX58ccft+7pV2H+GeHx9tprr0nu0aOHZP8Oovfff39C65HO9I7H/qGkWFZORTsMFm/xOPTUJb/99pt17V/x+Je7777but6yZYvk7777LuzzTzrpJMnTp0+XrA8pRewiDU1MnTpVsh7qP3DggFXuq6++ktymTRvJ4VZrGWNMnz59JN977705qLHbdu/eLbl3797WvXfffVfyxo0brXv58uWTrL87+r8bY0zz5s0l64NE9ffQGPvAX72Lun8V5TPPPCO5RYsWJi/gTQ8AAHACnR4AAOAEOj0AAMAJKZ3Tc88990jWy+iMMebGG2+UXKFCBeuennfToEEDydEubffPIejfv7/kc889V3K/fv2ieh6iX0aeF3dGzgunwKeTlStXSr7tttuse7G0rz6l+bzzzrPu9e3bVzLzeJJLL1lfs2ZNttkYe+6PppdJG2NM586dJV9//fVxqKF7jjvuOMn++XSHDx8O+zl976effgpbTrdlpDlZWrVq1STreXfGGFO2bNmonpFMvOkBAABOoNMDAACckNLhLf2q7l//+pd1T7/mvv322617eumbPsTssssus8odPHhQ8hdffCF56dKlVrk6depIfuKJJyQXLFgw8v8ByJZ/uEhf+5eEJ2spup9eYs/wVs7s3LlTcrTDWYUKFbKu69evL1kPV+sls0g8/5JlTQ9j6hyJ3uFXT1/I7hq58/TTT1vXehuIp556yrqnd71+++23JfsP+NVbx3z22WfZPtsYY0455RTJ+t/uE088Maq6pxJvegAAgBPo9AAAACfQ6QEAAE7wQqFQ1IWzsrJCCxYsSGB1oqO3Sx8wYIDkSHUbOnSo5IoVK1r3WrZsKdl/6myqZWVlmQULFngJeG6eaEtNzw/xzxWJZe6PPjE90jyjZPI8b2EoFIr7ccPJbM+5c+dKPv/886P6jN623hhjmjVrFtc6pULQvpvXXXeddT127Nhsy1WtWtW61vNF9FL0cuXKxa9ySZCI72Ze/D3rgkhtyZseAADgBDo9AADACSldsh4rPRylM9KbHnLyDz9x+nneobeTOHLkSAprgngaPXp0xGsgCHjTAwAAnECnBwAAOIFODwAAcAKdHgAA4AQ6PQAAwAl0egAAgBPo9AAAACfQ6QEAAE6g0wMAAJyQowNHPc/baoxZn7jqIBsVQ6FQyXg/lLZMGdozOGjLYIl7e9KWKRO2LXPU6QEAAEhXDG8BAAAn0OkBAABOoNMDAACcQKcHAAA4gU4PAABwgnOdHs/zCnieN9/zvMWe5y31PG9QquuE2NCWwUJ7BgdtGSxBak/nlqx7nucZYwqHQqE9nudlGGM+M8Z0C4VCc1NcNeQQbRkstGdw0JbBEqT2PK7FHpoAACAASURBVDrVFUi20B+9vD1/Xmb8+T+3en4BQVsGC+0ZHLRlsASpPZ0b3jLGGM/z8nmet8gYs8UY834oFJqX6johNrRlsNCewUFbBktQ2tPJTk8oFDocCoXOMsaUM8bU8TyvWqrrhNjQlsFCewYHbRksQWlPJzs9fwmFQruMMR8ZY1qmui7IHdoyWGjP4KAtgyXd29O5To/neSU9zyv2Zy5ojGlmjFmR2lohFrRlsNCewUFbBkuQ2tO5iczGmNLGmNGe5+Uzf3T6JoRCoWkprhNiQ1sGC+0ZHLRlsASmPZ1bsg4AANzk3PAWAABwE50eAADgBDo9AADACTmayFyiRIlQZmZmgqqC7Kxbt85s27bNi/dzacvUWLhw4bZQKFQy3s+lPZOP72awJOK7SVumRqS2zFGnJzMz0yxYsCA+tUJUsrKyEvJc2jI1PM9bn4jn0p7Jx3czWBLx3aQtUyNSWzK8BQAAnECnBwAAOIFODwAAcAKdHgAA4AQ6PQAAwAl0egAAgBPo9AAAACfQ6QEAAE7I0eaEQeE/Wf7111+XfP/990tevnx52GeMGDFC8o033hjH2gHuaty4sXX98ccfS/7oo48kN2rUKEk1AhAkvOkBAABOoNMDAACc4Mzw1q5duyT37NnTujdq1KhsP1O4cGHr+thjj5U8f/58yQxvAbHTQ1p6OMuPIa3Uad26teQvv/zSujdw4EDJN910U7KqhCTyt3n//v0lf/DBB2E/d9ppp0m+5JJLJA8YMMAqV7Ro0dxWMWq86QEAAE6g0wMAAJxApwcAADgh0HN6tm3bJvnSSy+VPHfu3LCfOf/88yU/9NBD1r3KlStLHj58eDyqGBgHDhyQfN9991n3xowZI7lZs2ZRPc+/rYDneVF9rm7dupJ1e0VSrVo1yWXKlInqM4ifSPN4ND33Ry9fj/S8QYMGhS0X6Rkw5ueff5a8dOlSyT/99JNVTm/zUb9+fclVq1ZNYO2QaLr99b+fxhizefNmyZF+N69atUry448/Ltn/dyOZ82J50wMAAJxApwcAADghUMNbW7Zssa7btm0rWQ9pHX20/X92mzZtJA8dOlTySSedFPZnsUzdtnPnTsmPPPKIdU8PVY0dOzaq58U6vKWH0qL9jF5WqbciMCa5SyldEu2QVrjP+Hduzu2z8Xd6esDq1avDltuwYYPkO+64Q/LMmTMTUzEkxZ133ilZD2elO970AAAAJ9DpAQAATgjU8Fbnzp2t688++yzbcsOGDYv4uWhUqFAhx58JMr17tX9mvj64NX/+/NY9vVqqXbt2kvUwlTHGZGVlSdarR04++WSrXKVKlSRHO7z1/vvvR1UO8RNpVVU0Yh3CYlfn6L377rs5/syKFSskL1iwwLpXtmxZySVKlJCsh8aNMebw4cPZPvuTTz6xri+88ELJerd8RM//Z9+lSxfJ+iDuY445xir3/PPPSy5XrpzkadOmWeUmTJggOSMjQ7L+fZ5svOkBAABOoNMDAACcQKcHAAA4Ie3n9Lz44ouSp0+fHrZczZo1Jd98880JrZOL9NJu/27VDRo0kHz66adb977++utsnzdkyJA41i6yZP4sV/nn4ISbk+PfzVuLZR6Q/3n6RHDYtm/fbl1Hu72Eppev165dO2w5vU3Ihx9+aN3bvXt3VD+rRo0akmfNmmXdK168eFTPcJ3e1sUYe5fyAgUKSH7mmWescjfccEO2z/PvuK+3gMkreNMDAACcQKcHAAA4Ie2Ht/SySn3opZ9eUn3w4EHrnl5GnS9fvjjWzk360EG/X375xbr+9ddfJevD6aZOnWqVW7t2bVQ/W+/krF/dHnfccVa5hg0bRvU8xEekHZT1MvJIw0+Rhrf0MzhINDb+IaEzzzxT8pIlS8J+7q677pI8e/ZsyV988UXYz7z55puxVNHyzTffSPYPo+hDUGH7/vvvJX/11Vdhy3Xv3l1ykE4g4E0PAABwAp0eAADgBDo9AADACWk/p0fP1Ynk888/D/sZPd9AzynQS60RH/65OfoYCT2/Z//+/Va5aI+U0HN6xo0bJ9k/V0svp9XlKleuHNXPwT+LNI9Hi3YOjm5b5B1NmjSR3K1bN8nXX3+9VU7P9/F/v2Oht7+44oorcv08V+gtASJtD7Bu3TrJe/bsse4VKlRI8lFHpde7k/SqLQAAQIzo9AAAACek/fDWU089Jblly5bWvZEjR0rWu40uW7bMKqdfry9atEjy6NGjrXL6NW7BggVjrDG0bdu2ST7++OMlX3zxxVY5/fr6jDPOyPHP8Z/+q4e0zj777LDlGOLMGb3TcrQnoetynIKed7Rr106yPnHbb9OmTZKbNm0qecaMGVY5/Xt2zpw5kt96662wz9bLposUKWLd078jjj467f8pSxq9NUgkr732muTx48db9wYMGJBtToehrrxfQwAAgDig0wMAAJyQ9u8ETzzxRMkdO3a07vmv/+J/ndq1a1fJP//8s+RLL73UKjdv3jzJkQ7TQ/T69u0ruVevXpLjfWCgPpzQGPvAvDJlykju06ePVU7v+K2H35C9aIe0tGhXeWmRDiblUNH4OO2006Iqp6cRXHvttWHL6XbW+Z577omhdohV3bp1JZcsWdK6t3Xr1qieoXe81itrI30v8wre9AAAACfQ6QEAAE6g0wMAAJyQ9nN6YtG6dWvrWu882bNnz7Cfe/XVVyUzpyd6Y8aMkbxmzRrr3h133CE5mXNmSpcuLVnPKxoyZIhVbsOGDZKZ0/PP9K67iRTpxHV9zz/HgPk+0cvMzJR89dVXS9ZLmY0xZvHixZL1iesPP/xw4iqHmFWsWFGybjtjjHnhhRckR/qOaQ888IBk/5L1e++9N5YqJhRvegAAgBPo9AAAACc4Obzlp5e2P/3005L1sJcxxvz000/JqlKgRFrGmhdccMEFkh955BHr3qRJkyT7l73j7xo2bJij/25M9K/RY5Gs4bYg0jsgv/jii5IvueQSq9ztt9+ebTn/liGx7KSOxCpVqpR1rYd/9W7Yd999t1Xu2Weflfz7779LfvPNN61yvXv3llygQIFc1TVeeNMDAACcQKcHAAA4geEtY0xGRobkYsWKpbAmyGv279+f6iqklVhWR8XjM+GGyCINqyF6RYsWlaxXchljH2Cp2+XKK6+0yr3xxhuSo93tGalz3HHHSfavgtQHRespAP7VYPr63HPPjXcVY8KbHgAA4AQ6PQAAwAl0egAAgBOY02PsU34XLVoUtlz9+vWTUR0g5fTcjLy4i3G0c3qQeNdcc43klStXSvbv3HzbbbdJfv311yX7T/pG3uNvo/Hjx0vW7aq3LMireNMDAACcQKcHAAA4wcnhre+//966Drdr6zHHHGNds/w1mMaNGyc5FApZ9/Sr+6D7+OOPJevhorw4vNW4ceOoyjVq1CixFYE55ZRTJA8fPlzy0qVLrXILFy6UrHc+f/TRRxNYOyTC9u3bJS9fvlxyhQoVrHLly5dPWp2ixZseAADgBDo9AADACQkZ3vrvf/8rWe/qaIwxrVq1kqwPtEu0LVu2SPYffHfo0KFsP9O6dWvrumbNmvGvWBobO3as5Fq1aklOh4MFd+7cKVnvFOt5XiqqkyeEWwGlh72MSe6Qkf7Z0Q5pffTRR5IZ3kquX375RfKePXvC3vvxxx+TVqegmj9/vnVduXJlySVKlIjrz9q9e7d1XadOHclr166V7N91uUyZMnGtRzzwpgcAADiBTg8AAHACnR4AAOCEhMzpyczMlNyiRQvr3uOPPy5Zz5E5//zzrXIFChTIdT1GjBiR7c8NN4fHGHvewLBhw3JdhyDTY/YPPfSQZD3XJ6/Sy2d/++03yfrvrjHsFmvM3+fS+E9c1sItb4922Xs8dlZ2cR7PddddZ1336NFDcrznIm7evNm6njp1quRly5ZJXrNmTdhn6OXsO3bssO6dcMIJua2iE5o0aWJdly5dWvLDDz9s3WvTpk2On6/nxnXt2tW6p+fxaFdeeWWOf06y8aYHAAA4gU4PAABwQkKGt6pXry7Z/2q8Z8+e2X7G/wq2T58+kjMyMsL+rJ9++knyq6++at379ttvJe/fvz/sM/ROyxMmTJBcvHjxsJ+BMZUqVZKsX6d3797dKnfOOeckrU7h6OEsY4zp0KFDtuVuv/1261q/Mg46/9L0cCINQSXr4E//EJZ+Fe+iZs2aWdcNGjSQrIc2/H9u+t6YMWPCPl9v+fHss89a9/zDU9FYvXq1ZP/QJ9MKwtO7x+/du9e6991330kePHiwdW/FihVRPf/ll1+WrLcViPTvp96GpnPnzlH9nFTiTQ8AAHACnR4AAOCEhAxvlSpVSvI777xj3Xvsscck61fS/tfTiTzo0b9SbPLkyZL9O0gjPD30o19/LlmyxCqXzOEtfRCe3hlcH3BojL0jc/PmzSV36dIlgbXL2/R3MNrdj5NJD81EWkHmonbt2lnX1157reTRo0dL9g9h6aHoXbt2Jah2f6d/X9x8881J+7nprmDBgpL9u8frw5IXL15s3fNfR0M/z/+zLr/8csn6uxiPVdeJxpseAADgBDo9AADACXR6AACAExIyp0fzj/Hdc889kvWydP/yuyeffFLye++9Z9378ssvo/rZeum8nlvkX4qux0kRPX2aul4i61+2qLcOaNu2rXUv2jHgr776SvLy5cslv//++1a59evXS9ZzFPLnz2+V69Wrl+T7779fsst/F/ScGT2/x78MPdql7bH8XP9cHRd3V46F/++33nqjffv2kvU8DWOSO4+nSpUqkmfMmCG5YsWKSatDurviiiskFypUyLrn/zc0twoXLiz5hhtusO7pHfiLFCkS15+baLzpAQAATqDTAwAAnJDw4a1I9CtZ/+tZPeSgM/IOvVO23inUPyShtynQB79G4n8N718yGY7ecuCiiy6S3K9fP6ucf9sC2HQbMsSU9/m/H3oYWR8I6h+qXLBgQa5/th6e6tixo+ROnTpZ5cqVKyf56KNT+k9PIMyfP9+6Hj58uOTx48db9/yHxP5FD5cZY8zFF18s+YILLpBcvnz5mOuZ1/CmBwAAOIFODwAAcAKdHgAA4AQGVhEXFSpUkPzGG29Y94YOHSrZP/fggw8+kFy5cmXJa9assco1bdo025+bmZlpXd96662SXTohHdD090zP09AZ6e3000+3rvU2LzrDxpseAADgBDo9AADACQxvIe78p6rrU54BAEgV3vQAAAAn0OkBAABOoNMDAACcQKcHAAA4gU4PAABwAp0eAADgBDo9AADACXR6AACAE+j0AAAAJ3ihUCj6wp631RizPnHVQTYqhkKhkvF+KG2ZMrRncNCWwRL39qQtUyZsW+ao0wMAAJCuGN4CAABOoNMDAACcQKcHAAA4gU4PAABwAp0eAADgBOc6PZ7nFfA8b77neYs9z1vqed6gVNcJsaEtg4X2DA7aMliC1J7OLVn3PM8zxhQOhUJ7PM/LMMZ8ZozpFgqF5qa4asgh2jJYaM/goC2DJUjteXSqK5BsoT96eXv+vMz4839u9fwCgrYMFtozOGjLYAlSezo3vGWMMZ7n5fM8b5ExZosx5v1QKDQv1XVCbGjLYKE9g4O2DJagtKeTnZ5QKHQ4FAqdZYwpZ4yp43letVTXCbGhLYOF9gwO2jJYgtKeTnZ6/hIKhXYZYz4yxrRMdV2QO7RlsNCewUFbBku6t6dznR7P80p6nlfsz1zQGNPMGLMitbVCLGjLYKE9g4O2DJYgtadzE5mNMaWNMaM9z8tn/uj0TQiFQtNSXCfEhrYMFtozOGjLYAlMezq3ZB0AALjJueEtAADgJjo9AADACXR6AACAE3I0kblEiRKhzMzMBFUF2Vm3bp3Ztm2bF+/n0papsXDhwm2hUKhkvJ9LeyYf381gScR3k7ZMjUhtmaNOT2ZmplmwYEF8aoWoZGVlJeS5tGVqeJ63PhHPpT2Tj+9msCTiu0lbpkaktmR4CwAAOIFODwAAcAKdHgAA4AQ6PQAAwAl0egAAgBPo9AAAACfQ6QEAAE5Iy1PWFy5cKLl+/fqSr732Wqvc8OHDk1YnAEB0Dh8+LHnZsmXWvUsvvVTyunXrJNeuXdsqN3/+/MRUDoHGmx4AAOAEOj0AAMAJaTm8tXr1asn79++XPGHCBKtcz549JZ922mk5/jlffvmldV21alXJRYsWzfHzgKA5cuSIdT1y5EjJ/mEL7amnnpJcsWJFyXv37rXKXXXVVZKPPvp/v66qVKlilWvRooXkkiX/d+QO39O8Y8uWLZJHjRoluV+/fmE/43n/O9rs+uuvT0i98D/HHnus5F9//VWybge/1q1bSx40aJB1r1q1anGsXXzwpgcAADiBTg8AAHACnR4AAOCEtJzTE87u3but608//VRytHN69DNuvPFG616zZs0kP/bYY5IjjXfCpudyGGPMu+++K/n9998P+7mLLrpIcizjxPnz57eue/ToIblYsWI5fh7+sG/fPuu6c+fOUX1Of2c2bNgQttxzzz2X4zrVqFFD8ujRo8PeQ/zp7USefPJJ696CBQskr1y5UrL/92etWrUkd+/eXXLbtm3jVk/8wT+HTtPtEunfuMmTJ0v2z4PV8306deok+cwzz7TK6fl6icabHgAA4AQ6PQAAwAmBGt7yW7NmjWS9A6he5m6MMWvXrpV8ww03SP7222+tcvr64YcflpyRkZH7yjrCP9ywePFiyZFeoephMJ1j9dJLL0levny5ZJY458yiRYtSXYW/+eabbyRv27YthTUJpt9++8261tMA3nnnHcl6yXMkeld9Y+zfwR06dIilioiS/j1oTPjhLv/wU6VKlST//PPPkn/66Ser3DPPPJNtfuKJJ6xy3bp1i7LGucebHgAA4AQ6PQAAwAmBHt566KGHJH/22WeS9+zZY5X7+uuvo3qefqV31FH0F6OlV8T5V/ukysaNGyXrV/J6B2D8s7POOiumzw0bNkyyXtURrddff9261q/YmzZtKvmcc86JoXYwxpglS5ZInjFjhuRHHnnEKhduCNG/QqdVq1aSGzRoIFm3lzHGFChQIOeVRdS2b98ueeLEiVF9Ru9ybowxK1askDx37lzJX3zxhVVu3LhxkletWiVZ/91KNv7lBgAATqDTAwAAnECnBwAAOCEt5/R89dVXOf6M3p05Wqeccop1/d5770nOly9fjp/nKj3Ou3r16qT9XH169+mnn27d06cJN27cOGl1Cho9lyYnbr/99lz9XL2sGfFx4MAB67p9+/aS9bYORYoUscrdd999kg8dOiTZfyq6//cpUmPKlCmS58yZE9VnypcvH/beeeedl202xt5R+/PPP5ecyq1BeNMDAACcQKcHAAA4IS2Ht/RrsnirV6+e5FdeecW6p4dLEH+33nqr5GgProxEL7MsU6ZMrp+Hv5swYUJU5a688soE1wS5NXv2bOtaL0vWu6X//vvvVjk9XLJp0ybJr732mlVO/27t06eP5FgOEEZyXXPNNbl+xvnnnx+HmuQeb3oAAIAT6PQAAAAn0OkBAABOSMs5PdFq1qyZ5OLFi4ctp+cbXHLJJZJZlp5cJ554ouSaNWtG9ZlJkyZZ12+//bbkW265RTJzelJr6tSp1nXLli0lL168OKpn6M9cfPHF1j39vc2fP38sVXSe/3ekXlasT0zfv3+/VS7aY3zWrl0rWc8Fe/jhh61y1157bdg6AbnFmx4AAOAEOj0AAMAJaTm8deONN0qOtKNk3759JV9wwQWS9fJLJJ4+kd7/Zx8KhSQfPnw47DP0Mtknn3xScv/+/a1y+hn6JO7p06db5fTfByTevn37rOuZM2dKjvb7OGbMmGyzMcbUqVNHsv+kZ0SnVq1a1vWXX34pWQ9p6f9ujDG1a9fO9nnLli2zrvV3cOHChZJ79OhhldN/H7p16/ZP1UaCFC5cWLIeckx3vOkBAABOoNMDAACckJbDW+XKlcs2//jjj1a5pk2bSh42bJjkf//73wmsHfyaN28uuUaNGtY9vXJnyJAhkiMNh/hfm4ejDz9s1aqVdU+v8mrRokVUz8PfnXrqqdZ1hQoVJP/www9hP6dfnevvqV/dunUlb9++XfKIESOscvoQYj2McuGFF4Z9NiKrUqVKtv/d/x0Ox1/uqquukrx582bJTZo0sco9//zzkjt06CBZ77CO2OnDlxs2bGjd07ty69/B/lWynTp1SlDtEo83PQAAwAl0egAAgBPo9AAAACek5ZwevdOyHoP0zw3QO4D27NlT8rZt26xyelnkCSecELd6Imf0cvOnnnoqrs8+ePCgdd2uXTvJer7QeeedF9efG3T6z9EYYxo3bixZ7+Lrl5GRIVnPy4tW5cqVresuXbpIvv766yVPnjzZKqfnCKW7vXv3StZzpNLBSSedJPnss8+27o0bN07y+PHjJTMXMz7077j33nvPuqd3wN6zZ4/ku+66yyqnf5927tw53lVMKN70AAAAJ9DpAQAATkjL4S1Nv+b+4IMPrHuXX3655CVLlki+//77rXL6FZ9+HV6qVKm41RN/8A9lRHvYZLT0a3O9nNq/c7d+dav/PrzyyitWueOPPz6u9Qu6EiVKZJvj7dhjj7Wu9c7eW7dulayXRgfBxo0bJetDO++8885UVCcu/Dty6+v3339fsn8YRQ+RIjb+P8NffvlFsh6q1tNIjDHm7rvvlly/fn3J1apVi3cV4443PQAAwAl0egAAgBPSfnhL86/o+OabbyTrQ+30gZXGGDNv3rxsn/Hqq69a5fRwGWLTu3dv6/rDDz+UrA81jESvMOjYsaN1T78C18NbDz74oFXunnvukTxjxgzJ/sMU9W7SyLvCHVo6evRo6zrdv8P16tWTPHjw4BTWJDmmTZsmeceOHdY9PZSN+NOrpD/55BPr3q5duyQ/8cQTkvVu2sYYc8wxxySodrHjTQ8AAHACnR4AAOAEOj0AAMAJgZrTE4k+wdu/c/PDDz8s+dNPP5Xsny+ilzNfcskl8a6iE/7v//7Pun755Zcl63ZYtGiRVa5BgwaS9ano/fv3j+rn3nHHHda1ntOjjRw50rpmTg/yknXr1qW6CnGXlZVlXesdmZE6ev6lf8m63h5Gz5vT84CMMebqq69OUO1ix5seAADgBDo9AADACc4Mb+mdJ1u1amXd00MYZcuWlbxlyxar3MSJEyVffPHFksMtl8U/a9++vWTdLvowRWPsHXgLFiyYsPr4D6NFekvk3xXEx4IFC1JdBWTjqKP+905EbxMSif+AX4a3AAAAUoRODwAAcIIzw1uRHH30//4Y9Oxz/+GTY8eOlaxX+XDwXXwUKVIk2xwPBw8ejOvzgkjvYH7o0CHJtWrVSkV1ItKHBEdy0003JbgmyaWH68aPHy/5zDPPtMqdffbZSatTLH7++WfJ/uEtfXhsgQIFJOvhFiTeqlWrJL/++utRfeaGG25IVHXihr9FAADACXR6AACAE+j0AAAAJyR9Ts/ixYslr1y5UrJ/V86ZM2dG9Tw936BChQrWvbfeeivbz/jnA6xevVrysmXLovq5S5YskZzXx89TbejQoZL1n7Ux9q6fFStWjOvP/fHHHyW3bds2rs8Ogj59+ljX+oRkPaenU6dOVrkqVapIvuCCC6x7NWrUiGcVLQMGDJAc7rttjDHHHXec5BNPPDFh9UkF/bvrsssuk+zfIf7222+X3LNnT8nJnH+4c+dO6/qHH36QrJcyr1ixwiqnt6fQO6mXLFky3lWEz/fffy+5ZcuWUX1Gt1c6tBFvegAAgBPo9AAAACckfXhLH1TWq1cvyf7liEeOHInqefpz/mf8/vvvsVQxrLp160quXr16XJ8dJPoVqTH2QaKbN2+27um/D/qVvM7G2K/Gd+zYEfZn660ERowYIfnAgQNhP1OoUCHJ11xzTdhyQbN27Vrret++fdmWe/HFF8M+o3Dhwta1HmLWy6snTZpklTvppJMk66E0P33o74wZMyT72zN//vyS3377bcnVqlUL++x0pA/e1X+m/qFKfRCvHhJr0qSJVa5OnTq5rtP06dMl6+kBmzZtssrpKQF6Wbp/R3s99DV48OBc1w/hDRw40LrWvz9/+uknyZFOHdCnE+TF7S38eNMDAACcQKcHAAA4IenDW61bt5a8ceNGyaNGjbLK+Wf+h6OHwaIdEovWWWedZV0/9NBDkvUuzrD5hysiDS3pXT+7desm+Z133rHK6Vfj+rVrPOhduP0rlYJMr6ozxpjvvvtOst6dORL/wbDLly/Pttxdd91lXev2XLp0aVQ/K5Jrr71Wsh4CCrLGjRtLnjVrlnVPD1t88sknku+9916rXCyHJeuhqVifoXda9g9l61V6sE2ZMkWyXvnq5x+mL1++vGQ9XP31119b5aKdEqJX2upVn+mANz0AAMAJdHoAAIAT6PQAAAAnJH1iSuXKlSU//vjjkrt3726V++WXXyS/8MIL1j29q69exhqt0qVLW9d6t159urde9um/h/CqVq1qXT/xxBOSI82Z0XMFoj1FO1Z6mbresdYlZcqUsa71PCo9v2revHlWuVjmVPlPaY60ZDka/tPTn3766Rw/I0iKFi1qXevfrXrelX+u3MKFCyVPmDDBurd+/fqofrbe9fq6666L6jN6u5J02MU3r9BbQug5eH6DBg3K9c/S20Do9jLGmDvvvFNyuv27yJseAADgBDo9AADACXlm3XW5cuXC3hs2bFgSa4J4a9++veQSJUpY94YMGSJ5zpw5CatD7dq1ret+/fpJrl+/fsJ+bjrRw10TJ06UvHXrVqvcV199JfmBBx6w7n3++edxrVPXrl0lX3TRRZIbNWpkldOv4mHTu2br76L/Wn8XkTfpA331dh/G2LtXjx07Nuwz6tWrJ/ngwYPWPT3sP3XqVMn+ndfTGW96AACAE+j0AAAAJ9DpAQAATsgzc3oQXHqcWJ/Ia4x9BIRepu5fxepkigAAEnRJREFUgqxPZ9dHGJx66qlWuTZt2kjOysqS3KpVK6ucPgEckfmXFLdo0SLbDCCxjjrqf+8pTj75ZOveyy+/nG2GjTc9AADACXR6AACAExjeQkodc8wxki+99NJsMwAA8cCbHgAA4AQ6PQAAwAl0egAAgBPo9AAAACfQ6QEAAE6g0wMAAJxApwcAADiBTg8AAHACnR4AAOAELxQKRV/Y87YaY9YnrjrIRsVQKFTyn4vlDG2ZMrRncNCWwRL39qQtUyZsW+ao0wMAAJCuGN4CAABOoNMDAACcQKcHAAA4gU4PAABwAp0eAADgBOc6PZ7nFfA8b77neYs9z1vqed6gVNcJsaEtg4X2DA7aMliC1J7OLVn3PM8zxhQOhUJ7PM/LMMZ8ZozpFgqF5qa4asgh2jJYaM/goC2DJUjteXSqK5BsoT96eXv+vMz4839u9fwCgrYMFtozOGjLYAlSezo3vGWMMZ7n5fM8b5ExZosx5v1QKDQv1XVCbGjLYKE9g4O2DJagtKeTnZ5QKHQ4FAqdZYwpZ4yp43letVTXCbGhLYOF9gwO2jJYgtKeTnZ6/hIKhXYZYz4yxrRMdV2QO7RlsNCewUFbBku6t6dznR7P80p6nlfsz1zQGNPMGLMitbVCLGjLYKE9g4O2DJYgtadzE5mNMaWNMaM9z8tn/uj0TQiFQtNSXCfEhrYMFtozOGjLYAlMezq3ZB0AALjJueEtAADgJjo9AADACXR6AACAE3I0kblEiRKhzMzMBFUF2Vm3bp3Ztm2bF+/n0papsXDhwm2hUKhkvJ9LeyYf381gScR3k7ZMjUhtmaNOT2ZmplmwYEF8aoWoZGVlJeS5tGVqeJ63PhHPpT2Tj+9msCTiu0lbpkaktmR4CwAAOIFODwAAcAKdHgAA4AQ6PQAAwAl0egAAgBPo9AAAACfQ6QEAAE6g0wMAAJyQo80JAQBA8nzxxReSzz//fMmeZ28GfuTIkaTVKZ3xpgcAADiBTg8AAHACw1tIqtWrV1vXzz//vOTdu3dH9blPP/001/WoV6+e5M8++yzXz0Pspk2bJvmtt96y7m3evFnyscceK7lz585WucmTJ0t+4okn4l1FxNnvv/8uec+ePda9X3/9VfJXX30lefbs2VE//+abb5Z8+umnx1LFPGPo0KGS9ZBWvnz5UlGdtMebHgAA4AQ6PQAAwAl0egAAgBPy7Jye//u//5McaQ5Hy5YtJTdu3DhsuczMTMnt27fPXeWQI19++aXk22+/Pey9WPiXbUarWLFiufq5yJ2OHTtKHjt2bI4/P336dOtazwupVKmS5OrVq4ctd9FFF1n3Yv275Lorr7wyx5/Zt2+f5O+++866t2rVqmw/U6FCBeu6bNmykrdv327d0/NgDh8+nOP65SXjx4+X/Prrr0v2/9+ll7bXrVs38RVLU7zpAQAATqDTAwAAnJBnhrfee+8963revHmSI7121p/zP0PLnz+/ZP8QSzhNmza1rkePHi05IyMjqmfAmB49ekieP3++dS+3Qwr+NjrjjDMk6+XOTZo0scrddNNNufq5yJ01a9bk+DPly5eXfMopp4Qtp5e9L1u2zLq3cuXKbJ9njDE1a9bMcZ1gzMSJEyVH+30OhUKS9S7Dxhhz9913S27VqpXkihUrWuVKly4tWQ/tGGNM165do6pHuunVq5fkp556yrp31VVXSdbDYMYYc9555yW2YmmENz0AAMAJdHoAAIAT8szw1qFDhyJe59bBgwclb9u2LarP6FnzxhhTpUoVyQMHDoxLvVygd9w999xzrXt6pYbecde/i2qbNm0kd+rUSbJ/FRa7lOZN33//vXW9YMECyYULF5ashyeNMeb++++XXLt2bcnFixePdxURZxdffLF1/eyzz2Zbzt+WBQsWzPHP8q9WysnuzelE/x7csGGDdU8PafmHDOfMmSPZ9ZVdvOkBAABOoNMDAACcQKcHAAA4Ic/M6ckJvXRx/fr1Sfu5W7ZsSdrPChK9o7ZeMmyMPX4/YcIEyc2bN098xZA0ek6dMcYcOHBA8mWXXSb51ltvtcpdcMEFia0Yckyfgm6MvfxcK1GihHVdrly5hNXJr2jRokn7Wcmk5+P45+boLRiiXc7u4lJ23vQAAAAn0OkBAABOSIvhrZNOOsm61jsvL1myJMfP27hxo3WtD6eLpGTJkjn+WYjst99+k6yHNkaOHGmVy8rKkqyXtiM9LFy4MOy9c845R3KDBg2SUR3kwogRI6xrvQuz/m527949aXWCMY8++mi22Rh7KEwvZ/dvIaJ3107mcGQy8aYHAAA4gU4PAABwAp0eAADghLSY07N//37ret26dZIPHz4sedasWTE9/4orrpDcv39/ycwdiQ+9HX29evWse/rUdb39gP/0dL1NwSuvvCLZPybNMRR506RJk8Le09+zjIyMZFQHOaSPPBg1alTYctWqVcs2I7X0XJ158+ZJbteunVVOL23XR1706NEjgbVLLt70AAAAJ9DpAQAATsgzw1unnXaadR1p1+WWLVsmrB56+bpeTo34+Oyzz6zrd999V/LgwYMl61ewxth/B/QQmX9ZbJ8+fSSXKlUqd5VFrvzyyy+S33//feue/r4XK1YsaXVCbObOnSt56dKlYcvprQn69u1r3bvnnnskB3XH5LxKLz/XWZ++boz971/Pnj0lb9q0ySqnh77S7dR23vQAAAAn0OkBAABOyDPDW6eeeqp1napDRfVBiHrVkTH2DHYOQoyPVq1aSdYrtoYMGWKVe/755yXrV63+g/VeeuklycOHD5fcvn373FcWOaIPGT1y5Ih1Tx88W7NmzaTVCbnnP2BUX+vfn/5dgfX1jz/+KLlMmTLxriKi5B+a0teRDjDVh0On2wGmvOkBAABOoNMDAACcQKcHAAA4Ic/M6fFr0qSJZL0DszHG/PDDD0mpg15ObYwxFSpUkMycnvjLnz+/5Hvvvde6d8stt0i++eabJU+bNs0qp5dJd+rUSfKaNWuscnfddVfuKot/VKJECcknnXSSdU9/p/WO6P5210vbzzzzTMkFChSIVzURBT3XI9J355xzzpH83HPPWfdmz54tuUqVKpI/+OADq1w6zAtxQSyntvuXwOfF5ey86QEAAE6g0wMAAJyQZ4e39Gvua6+91rqnD7zz79wbjn9JvLZ69WrJemmt/7XrG2+8IVkf1Na4ceOo6oDY6eGRKVOmSB44cKBV7umnn5a8Y8cOyf369bPK6WW2/nuIv8cee8y6/ve//y15xYoVkq+55pqwz9Dfv9atW1v3jjqK//8tkfQuvg8++GBUn/EvbdfDW/v27ZOsl68jPejtWzp06CBZH1hqTN5czs5vCgAA4AQ6PQAAwAl5dnhLy8zMtK4HDRqUsJ+ld5GtXLmydW/jxo2S9Su9jz76yCpXtWrVBNUOfv7hrUqVKknWr2B37txpldOHm+pVQZdeemmcawhj7BVaxhhTqFAhyf3795e8aNGisM9o27atZL0jrDH2cDOAxNLfN509z7PK6ZVdupwe9ko23vQAAAAn0OkBAABOoNMDAACckBZzepJJ7wo8Y8YM616dOnUkb968WfKGDRuscq7P6fEvQdXLk7t27Wrd0yerx8N1110nWe/66p+rs23bNsl6fg9zepLjwgsvlKy3fPDvnD1y5EjJL730kmT/DrH6GXonaOQdegm7fzk7gqFXr17WtT6dfe7cudlmY5K7nJ03PQAAwAl0egAAgBMY3oqgWrVq1rVeMjtu3DjJN954o1VOH1Tqf0ZQ/frrr5IbNGhg3du9e7fk+++/P2l10ofdDR061Lqnd/5dvny55C+++CLsM5AY+vDQM844w7r3+OOPS37nnXckf/nll1a5YcOGSU7m3zGEt2nTJuvav5wZwaOngBhjzOHDhyXrg8InTZpklWN4CwAAIM7o9AAAACcwvJUDendJPbzlX620bt06ya4Mb+lZ++vXr7fuvfnmm5KrV6+etDppegdtY+whkFWrVkn2ry77/PPPJethGCSGfh1ujDH/+c9/JG/dujXZ1QkE/ffbv9JUD9/qXbJjpb/rAwYMCFuufv36kjmwObjy5csnWX+3UznUyZseAADgBDo9AADACXR6AACAE9JiTs9rr71mXX/88ceSX3zxxaTVY968eUn7WelA72o8fPjwsOVOPPHEZFQnR+rVqyd55cqVkr/++mur3N69eyUzpycxdu3aJfmZZ56x7t13331RPSMzMzOeVUp7es6a3hW3fPnyVrlPPvkk1z9Lb1ehdzfX/91Pbz9QtGjRXNcBqaO3+Wjfvr11T8/dKVeunOQ2bdokvmJh8KYHAAA4gU4PAABwQloMb73xxhvW9ZQpUyTrJXG33nqrVa5mzZo5/lkLFy6U7H9Vp5eiw5jjjz9esj449MMPP7TK6d2Pb7vtNutejRo1JGdlZUk+4YQTrHJHHZW7/vn+/fut64MHD0pmp9i/e/LJJyX7lxTrdj906JB1T2/l4P8zD+fnn3+WPHr06Kg+06dPH+taHzQLe1hQ//3WS8WNMaZw4cI5frZ/2KpRo0aSFy9eLNm/BH78+PGSGdJKb3pI6+qrr5bs/12q/32eMGGC5GTuwOzHmx4AAOAEOj0AAMAJaTG85ad3dnzhhRckv/XWW1Y5vaLo2GOPlbxx40arnD6sUA9hbd68Oar6XHHFFda1/xVyUIUbWtQHyxljzOrVqyX37dvXuhduaKlly5bW9THHHCP5wgsvDPt5fdir5t8lWq/S0s/wD2m6+hp+9uzZkv2He86dO1fy2rVrk1an1q1bS+7Zs6d1T/9dhDFly5aVrA/+1KumjLFX0fTr1y/s8/Suzno1mDH2kJYe6nrwwQetcqkc0nDBxIkTJV955ZXWvXPPPVdyjx49onqeHsLyt3koFJIcboWWMXlnSEvjTQ8AAHACnR4AAOAEOj0AAMAJaTGnx78UXe/IvHPnTsn+OTiXXXZZwuqkd/T973//a90rVqxYwn5uXtW2bVvJ+jR6Y4x55ZVXJH/22WfWvZkzZ0o+cOCA5BkzZoT9WZMnT5Ycj+XmJ598suQxY8ZY9/Lnz5/r56cjvUz9o48+su7p+SKR/nz0TteR6J2uTzvtNOueXsJevXp1ybndwiDo5syZI/mmm26SPGvWLKucnuMzbdo061603y29i7aeS1K8ePHoKou4qFu3rmQ9v8cYex5ehw4dJOv5scaEPxXdP2dOn5Kg/57o3w3G5J15PBq/OQAAgBPo9AAAACekxfBW8+bNrWu9Q/Oll14qWR8OmQh6Oea9994r2cXhrEjKlCljXffu3TvbbIy9dFwPh7z55ptWOf+u3NH417/+JVkveTfGmJIlS0rWw6euDmf56WGKatWqWfcGDhwo2T+EpXdn1X/G/h2T9VCV3hXY/7MQm4oVK0rWOyH/9ttvVrkRI0ZI9u+0rIcQTznlFMkXXXSRVc6/TBmpodvB3yZ6W5VHH300aXXKi3jTAwAAnECnBwAAOIFODwAAcEJazOnx08tpp0+fLvnxxx+3yr399ts5fnZGRobk//znP9Y9fWq7/+RpxObss8/ONl911VWpqA7+VKpUqWyzMcY0adIk2dVBLkRaOq7nZwEu4E0PAABwAp0eAADghLQc3tL0ieaunG4OAAByjjc9AADACXR6AACAE+j0AAAAJ9DpAQAATqDTAwAAnECnBwAAOIFODwAAcAKdHgAA4AQ6PQAAwAleKBSKvrDnbTXGrE9cdZCNiqFQqGS8H0pbpgztGRy0ZbDEvT1py5QJ25Y56vQAAACkK4a3AACAE+j0AAAAJ9DpAQAATqDTAwAAnECnBwAAOMG5To/neQU8z5vved5iz/OWep43KNV1Qmxoy2ChPYODtgyWILWnc0vWPc/zjDGFQ6HQHs/zMowxnxljuoVCobkprhpyiLYMFtozOGjLYAlSex6d6gokW+iPXt6ePy8z/vyfWz2/gKAtg4X2DA7aMliC1J7ODW8ZY4znefk8z1tkjNlijHk/FArNS3WdEBvaMlhoz+CgLYMlKO3pZKcnFAodDoVCZxljyhlj6nieVy3VdUJsaMtgoT2Dg7YMlqC0p5Odnr+EQqFdxpiPjDEtU10X5A5tGSy0Z3DQlsGS7u3pXKfH87ySnucV+zMXNMY0M8asSG2tEAvaMlhoz+CgLYMlSO3p3ERmY0xpY8xoz/PymT86fRNCodC0FNcJsaEtg4X2DA7aMlgC057OLVkHAABucm54CwAAuIlODwAAcAKdHgAA4AQ6PQAAwAl0egAAgBPo9AAAACfQ6QEAAE6g0wMAAJzw//LvYIklZJuvAAAAAElFTkSuQmCC\n"
          },
          "metadata": {}
        }
      ]
    }
  ]
}