siamese.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "siamese.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyPCBxZdZ254Vx7303v5PJrt",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/mingrui-liu/Hyperpigmentation-autograde/blob/main/siamese.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fIKz83spVef2"
      },
      "source": [
        "import tensorflow as tf\n",
        "import glob\n",
        "import os\n",
        "import pandas as pd\n",
        "from sklearn.model_selection import train_test_split\n",
        "import cv2\n",
        "import math\n",
        "import shutil\n",
        "import random\n",
        "\n",
        "import tensorflow as tf\n",
        "import numpy as np\n",
        "\n",
        "from tensorflow.keras.applications import ResNet50\n",
        "from tensorflow.keras.models import Sequential, Model, load_model,save_model\n",
        "from tensorflow.keras.layers import Dense, Dropout, Flatten, Activation,GlobalAveragePooling2D,Lambda,Concatenate,Input,BatchNormalization\n",
        "from tensorflow.keras.metrics import categorical_accuracy\n",
        "from tensorflow.keras.optimizers import SGD,Adam,RMSprop,Nadam\n",
        "import matplotlib.pyplot as plt\n",
        "import tensorflow_addons as tfa"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UJxF1dwzVg7T"
      },
      "source": [
        "AUTOTUNE = tf.data.experimental.AUTOTUNE\n",
        "\n",
        "image_size = 224\n",
        "BATCH_SIZE = 32\n",
        "data_path = \"/content/drive/My Drive/Unilever/\""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "9GtBnMNAVg90",
        "outputId": "da61e97c-5c00-4a38-f8c0-5a7527aebcbf"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at /content/drive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PbvBjEcLVhAj"
      },
      "source": [
        "def create_generators():\n",
        "\n",
        "\tdef normalize(x):\n",
        "\t\tx /= 127.5\n",
        "\t\tx -= 1.\n",
        "\t\treturn x\n",
        "\n",
        "\tdef load_and_preprocess_image(path):\n",
        "\t\timage = tf.io.read_file(path)\n",
        "\t\timage = tf.image.decode_jpeg(image, channels=3)\n",
        "\t\timage = tf.image.resize(image, [image_size, image_size])\n",
        "\t\timage = tf.clip_by_value(image, 0.0, 255.0)\n",
        "\t\timage = tf.cast(image, dtype=tf.uint8)\n",
        "\n",
        "\t\treturn image\n",
        "\n",
        "\tdef load_file_from_disk(data_path):\n",
        "\t\tall_image_paths = sorted(glob.glob(data_path + \"image_test/*.jpg\"))\n",
        "\n",
        "\t\t# read the excel file\n",
        "\t\tMHPgrades = pd.read_excel(data_path + \"MHPgrades.xlsx\")\n",
        "\t\tgrade = MHPgrades[\"Mottled hyperpigmentation\"]\n",
        "\n",
        "\t\t#group and generate the lable\n",
        "\t\tall_image_labels = list(grade)\n",
        "\t\tall_label = []\n",
        "\t\tfor grade in all_image_labels:\n",
        "\t\t\tif grade < 3.0:\n",
        "\t\t\t\tlabel = 2.5\n",
        "\t\t\telif grade > 5.0:\n",
        "\t\t\t\tlabel = 5.5\n",
        "\t\t\telse:\n",
        "\t\t\t\tlabel = grade\n",
        "\t\t\tall_label.append(int(label * 2))\n",
        "\n",
        "\t\tlabel_to_index = dict((name, index) for index, name in enumerate(sorted(set(all_label))))\n",
        "\t\tall_labels = [label_to_index[grade] for grade in all_label]\n",
        "\t\t              \n",
        "\t\t# train test split \n",
        "\t\ttrain_paths, test_paths, train_labels, test_labels = train_test_split(all_image_paths, all_labels, test_size=0.1, random_state=1)\n",
        "\t\ttrain_paths, val_paths, train_labels, val_labels = train_test_split(train_paths, train_labels, test_size=0.1, random_state=1)\n",
        "\n",
        "\t\treturn train_paths,val_paths, test_paths, train_labels, val_labels, test_labels\n",
        "\n",
        "\tdef load_file_from_disk_with_additional_image(data_pat):\n",
        "\t\tall_image_paths_1 = sorted(glob.glob(data_path + \"image_test/*.jpg\"))\n",
        "\t\tall_image_paths_2 = sorted(glob.glob(data_path + \"Additional_Images/*.jpg\"))\n",
        "\n",
        "\t\t# read the excel file\n",
        "\t\tMHPgrades = pd.read_excel(data_path +\"MHPgrades.xlsx\",header = 0)\n",
        "\t\tMHPgrades2 = pd.read_excel(data_path +\"SKN-APP-0452-MHPgrades.xlsx\",header = 1)\n",
        "\n",
        "\t\tMHPgrades2 = MHPgrades2.dropna(axis = 0)\n",
        "\t\tMHPgrades2 = MHPgrades2.melt(id_vars=[\"Subject No.\", \"Side of Face\"], \n",
        "\t\t        var_name=\"VISIT\", \n",
        "\t\t        value_name=\"Mottled hyperpigmentation\").sort_values(by = ['Subject No.','Side of Face'])\n",
        "\n",
        "\t\tgrade1 = MHPgrades[\"Mottled hyperpigmentation\"]\n",
        "\n",
        "\t\t#Manipulate excel file\n",
        "\t\tMHPgrades2.loc[MHPgrades2['VISIT'] == 'Baseline' , 'VISIT'] = '00'\n",
        "\t\tMHPgrades2.loc[MHPgrades2['VISIT'] == 'Week 4' , 'VISIT'] = '04'\n",
        "\t\tMHPgrades2.loc[MHPgrades2['VISIT'] == 'Week 8' , 'VISIT'] = '08'\n",
        "\t\tMHPgrades2.loc[MHPgrades2['VISIT'] == 'Week 12' , 'VISIT'] = '12'\n",
        "\t\tMHPgrades2.loc[MHPgrades2['VISIT'] == 'Week 16' , 'VISIT'] = '16'\n",
        "\t\tMHPgrades2.loc[MHPgrades2['VISIT'] == 'Week 17' , 'VISIT'] = '17'\n",
        "\n",
        "\t\tMHPgrades2.loc[MHPgrades2['Side of Face'] == 'Left' , 'Side of Face'] = 'L'\n",
        "\t\tMHPgrades2.loc[MHPgrades2['Side of Face'] == 'Right' , 'Side of Face'] = 'R'\n",
        "\n",
        "\t\t#match grade and images and make valid pair\n",
        "\t\tvalid_path = []\n",
        "\t\tvalid_label = []\n",
        "\t\tfor index, row in MHPgrades2.iterrows():\n",
        "\t\t    id = str(row['Subject No.'] )\n",
        "\t\t    id = id.zfill(4)\n",
        "\n",
        "\t\t    side = row['Side of Face']\n",
        "\t\t   \n",
        "\t\t    visit = str(row['VISIT'])\n",
        "\n",
        "\t\t    grade = row['Mottled hyperpigmentation']\n",
        "\t\t  \n",
        "\n",
        "\n",
        "\t\t    output = [path for path in all_image_paths_2 if path.startswith('/content/drive/My Drive/Unilever/Additional_Images/P' + id + '_FACE_' + side + '_S00_VCR_S2_V00D00W' + visit)] \n",
        "\t\t    if len(output)>0:\n",
        "\t\t      valid_path.append(output[0])\n",
        "\t\t      valid_label.append(grade)\n",
        "\n",
        "\t\tall_image_paths = all_image_paths_1 + valid_path\n",
        "\t\tall_image_labels = list(grade1) +  valid_label\n",
        "\n",
        "\t\t##Create label and zip dataset\n",
        "\t\tall_label = []\n",
        "\t\tfor grade in all_image_labels:\n",
        "\t\t  if grade < 3.0:\n",
        "\t\t    label = 2.5\n",
        "\t\t  elif grade > 5.0:\n",
        "\t\t    label = 5.5\n",
        "\t\t  else:\n",
        "\t\t    label = grade\n",
        "\t\t  all_label.append(label)\n",
        "\n",
        "\t\tlabel_to_index = dict((name, index) for index, name in enumerate(sorted(set(all_label))))\n",
        "\t\tall_labels = [label_to_index[grade] for grade in all_label]\n",
        "\n",
        "\t\ttrain_paths, test_paths, train_labels, test_labels = train_test_split(all_image_paths, all_labels, test_size=0.1, random_state=1)\n",
        "\t\ttrain_paths, val_paths, train_labels, val_labels = train_test_split(train_paths, train_labels, test_size=0.1, random_state=1)\n",
        "\n",
        "\t\treturn train_paths,val_paths, test_paths, train_labels, val_labels, test_labels\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\tdef create_ds(paths,labels):\n",
        "\t\ttrain_paths_1 = tf.data.Dataset.from_tensor_slices(paths)\n",
        "\t\ttrain_paths_2 = tf.data.Dataset.from_tensor_slices(paths)\n",
        "\t\timage_ds_1 = train_paths_1.map(load_and_preprocess_image, num_parallel_calls=AUTOTUNE)\n",
        "\t\timage_ds_2 = train_paths_2.map(load_and_preprocess_image, num_parallel_calls=AUTOTUNE)\n",
        "\n",
        "\t\tlabel_ds_1 = tf.data.Dataset.from_tensor_slices(tf.cast(labels, tf.float32))\n",
        "\t\tlabel_ds_2 = tf.data.Dataset.from_tensor_slices(tf.cast(labels, tf.float32))\n",
        "\n",
        "\n",
        "\n",
        "\t\timage_label_ds_1 = tf.data.Dataset.zip((image_ds_1, label_ds_1))\n",
        "\t\timage_label_ds_1 = image_label_ds_1.shuffle(64)\n",
        "\n",
        "\t\timage_label_ds_2 = tf.data.Dataset.zip((image_ds_2, label_ds_2))\n",
        "\t\timage_label_ds_2 = image_label_ds_2.shuffle(64)\n",
        "\t\t\n",
        "\n",
        "\t\timage_label_ds = tf.data.Dataset.zip((image_label_ds_1, image_label_ds_2))\n",
        "\n",
        "\t\treturn image_label_ds\n",
        "\n",
        "\tdef detect(label1,label2):\n",
        "\n",
        "\t\tlabel = tf.cond(tf.equal(label1,label2), lambda: tf.constant(1.0), lambda: tf.constant(0.0))\n",
        "\n",
        "\t\treturn label\n",
        "\n",
        "\n",
        "\n",
        "\ttrain_paths,val_paths, test_paths, train_labels, val_labels, test_labels = load_file_from_disk_with_additional_image(data_path)\n",
        "  # here's our final training dataset\n",
        "\ttrain_ds = create_ds(train_paths,train_labels)\n",
        "\tval_ds = create_ds(val_paths,val_labels)\n",
        "\ttest_ds = create_ds(test_paths,test_labels)\n",
        "\n",
        "\n",
        "  #train\n",
        "\ttrain_ds = train_ds.cache()\n",
        "\ttrain_ds = train_ds.shuffle(buffer_size = 1024)\n",
        "\ttrain_ds = train_ds.repeat()\n",
        "\ttrain_ds = train_ds.map(lambda image_label1,image_label2: (tf.image.random_flip_left_right(image_label1[0]),tf.image.random_flip_left_right(image_label2[0]),image_label1[1],image_label2[1]),num_parallel_calls=AUTOTUNE)\n",
        "\t#train_ds = train_ds.map(lambda image1,image2,label1,label2:(randaugment.distort_image_with_randaugment(image1,num_layers = 1,magnitude = 2),\n",
        "\t#\t\t\t\t\t\t\t\t\t\t\t\t\t\t\trandaugment.distort_image_with_randaugment(image2,num_layers = 1,magnitude = 2),\n",
        "\t#\t\t\t\t\t\t\t\t\t\t\t\t\t\t\tlabel1,\n",
        "\t#\t\t\t\t\t\t\t\t\t\t\t\t\t\t\tlabel2),num_parallel_calls=AUTOTUNE)\n",
        "\ttrain_ds = train_ds.map(lambda image1,image2,label1,label2 :(normalize(tf.cast(image1,tf.float32)),normalize(tf.cast(image2,tf.float32)),label1,label2),num_parallel_calls=AUTOTUNE)\n",
        "\ttrain_ds = train_ds.map(lambda image1,image2,label1,label2 :((image1,image2),detect(label1,label2)),num_parallel_calls=AUTOTUNE)\n",
        "\ttrain_ds = train_ds.batch(BATCH_SIZE).prefetch(buffer_size=AUTOTUNE)\n",
        "\n",
        "  #val\n",
        "\tval_ds = val_ds.cache()\n",
        "\tval_ds = val_ds.shuffle(buffer_size = 64)\n",
        "\tval_ds = val_ds.repeat()\n",
        "\tval_ds = val_ds.map(lambda image_label1,image_label2: (image_label1[0],image_label2[0],image_label1[1],image_label2[1]),num_parallel_calls=AUTOTUNE)\n",
        "\tval_ds = val_ds.map(lambda image1,image2,label1,label2 :(normalize(tf.cast(image1,tf.float32)),normalize(tf.cast(image2,tf.float32)),label1,label2),num_parallel_calls=AUTOTUNE)\n",
        "\tval_ds = val_ds.map(lambda image1,image2,label1,label2 :((image1,image2),detect(label1,label2)),num_parallel_calls=AUTOTUNE)\n",
        "\tval_ds = val_ds.batch(BATCH_SIZE).prefetch(buffer_size=AUTOTUNE)\n",
        "\n",
        "  #test\n",
        "\ttest_ds = test_ds.cache()\n",
        "\ttest_ds = test_ds.shuffle(buffer_size = 64)\n",
        "\ttest_ds = test_ds.repeat()\n",
        "\ttest_ds = test_ds.map(lambda image_label1,image_label2: (image_label1[0],image_label2[0],image_label1[1],image_label2[1]),num_parallel_calls=AUTOTUNE)\n",
        "\ttest_ds = test_ds.map(lambda image1,image2,label1,label2 :(normalize(tf.cast(image1,tf.float32)),normalize(tf.cast(image2,tf.float32)),label1,label2),num_parallel_calls=AUTOTUNE)\n",
        "\ttest_ds = test_ds.map(lambda image1,image2,label1,label2 :((image1,image2),detect(label1,label2)),num_parallel_calls=AUTOTUNE)\n",
        "\ttest_ds = test_ds.batch(BATCH_SIZE).prefetch(buffer_size=AUTOTUNE)\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\treturn train_ds,val_ds, test_ds\n",
        "\n",
        "train_ds,val_ds, test_ds = create_generators()\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vCjTNukkVxEq",
        "outputId": "3a304a26-0531-41e3-e674-9c6d485e7917"
      },
      "source": [
        "def create_base_model():\n",
        "    conv_base = ResNet50(include_top = False, weights = 'imagenet',\n",
        "                          input_shape = (224, 224, 3))\n",
        "\n",
        "\n",
        "    #conv_base.trainable = False\n",
        "    x = conv_base.output\n",
        "    x = tf.keras.layers.Dropout(0.5)(x)\n",
        "    embedding = GlobalAveragePooling2D()(x)\n",
        "    embedding = Dense(128)(embedding)    \n",
        "    return Model(conv_base.input, embedding)\n",
        "\n",
        "def SiameseNetwork(base_model):\n",
        "    \"\"\"\n",
        "    Create the siamese model structure using the supplied base and head model.\n",
        "    \"\"\"\n",
        "    input_a = Input(shape=(224, 224, 3),name = \"image1\")\n",
        "    input_b = Input(shape=(224, 224, 3),name = \"image2\")\n",
        "\n",
        "    processed_a = base_model(input_a)\n",
        "    processed_b = base_model(input_b)\n",
        "\n",
        "\n",
        "\n",
        "    head = Concatenate()([processed_a, processed_b])\n",
        "    head = Dense(1)(head)\n",
        "    head = Activation(activation='sigmoid')(head)\n",
        "    return Model([input_a, input_b], head)\n",
        "\n",
        "base_model = create_base_model()\n",
        "\n",
        "\n",
        "\n",
        "siamese_network = SiameseNetwork(base_model)\n",
        "\n",
        "#siamese_network.save(\"test.h5\")\n",
        "lr_schedule  = tfa.optimizers.ExponentialCyclicalLearningRate(\n",
        "                              initial_learning_rate=1e-8,\n",
        "                              maximal_learning_rate=1e-6,\n",
        "                              step_size=240,\n",
        "                              )\n",
        "opt = Adam(learning_rate=1e-8)\n",
        "\n",
        "siamese_network.compile(optimizer= opt,\n",
        "              loss='binary_crossentropy',\n",
        "              metrics=['accuracy','RootMeanSquaredError'])\n",
        "\n",
        "\n",
        "\n",
        "history = siamese_network.fit(train_ds,\n",
        "      epochs = 100,\n",
        "      steps_per_epoch = 50,\n",
        "      validation_data = val_ds,\n",
        "      validation_steps = 20)\n",
        "      \n",
        "      \n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/100\n",
            "50/50 [==============================] - 38s 757ms/step - loss: 0.5409 - accuracy: 0.7706 - root_mean_squared_error: 0.4222 - val_loss: 2.8894 - val_accuracy: 0.1344 - val_root_mean_squared_error: 0.8966\n",
            "Epoch 2/100\n",
            "50/50 [==============================] - 35s 698ms/step - loss: 0.5422 - accuracy: 0.7656 - root_mean_squared_error: 0.4223 - val_loss: 2.9792 - val_accuracy: 0.1344 - val_root_mean_squared_error: 0.8990\n",
            "Epoch 3/100\n",
            "50/50 [==============================] - 36s 711ms/step - loss: 0.5548 - accuracy: 0.7538 - root_mean_squared_error: 0.4286 - val_loss: 3.2086 - val_accuracy: 0.1359 - val_root_mean_squared_error: 0.9040\n",
            "Epoch 4/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.5286 - accuracy: 0.7750 - root_mean_squared_error: 0.4158 - val_loss: 3.4651 - val_accuracy: 0.1328 - val_root_mean_squared_error: 0.9095\n",
            "Epoch 5/100\n",
            "50/50 [==============================] - 35s 706ms/step - loss: 0.5500 - accuracy: 0.7500 - root_mean_squared_error: 0.4263 - val_loss: 3.2807 - val_accuracy: 0.1344 - val_root_mean_squared_error: 0.9001\n",
            "Epoch 6/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.5208 - accuracy: 0.7719 - root_mean_squared_error: 0.4115 - val_loss: 2.8364 - val_accuracy: 0.1312 - val_root_mean_squared_error: 0.8855\n",
            "Epoch 7/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.5384 - accuracy: 0.7631 - root_mean_squared_error: 0.4202 - val_loss: 2.0800 - val_accuracy: 0.1453 - val_root_mean_squared_error: 0.8245\n",
            "Epoch 8/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.5135 - accuracy: 0.7912 - root_mean_squared_error: 0.4088 - val_loss: 1.0884 - val_accuracy: 0.2672 - val_root_mean_squared_error: 0.6437\n",
            "Epoch 9/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.5286 - accuracy: 0.7756 - root_mean_squared_error: 0.4159 - val_loss: 0.6879 - val_accuracy: 0.5672 - val_root_mean_squared_error: 0.4960\n",
            "Epoch 10/100\n",
            "50/50 [==============================] - 35s 706ms/step - loss: 0.5284 - accuracy: 0.7781 - root_mean_squared_error: 0.4146 - val_loss: 0.5932 - val_accuracy: 0.6953 - val_root_mean_squared_error: 0.4511\n",
            "Epoch 11/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.5230 - accuracy: 0.7825 - root_mean_squared_error: 0.4124 - val_loss: 0.5480 - val_accuracy: 0.7359 - val_root_mean_squared_error: 0.4275\n",
            "Epoch 12/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.5338 - accuracy: 0.7738 - root_mean_squared_error: 0.4182 - val_loss: 0.5177 - val_accuracy: 0.7437 - val_root_mean_squared_error: 0.4107\n",
            "Epoch 13/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.5269 - accuracy: 0.7725 - root_mean_squared_error: 0.4138 - val_loss: 0.4982 - val_accuracy: 0.7750 - val_root_mean_squared_error: 0.3991\n",
            "Epoch 14/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.5148 - accuracy: 0.7856 - root_mean_squared_error: 0.4086 - val_loss: 0.4903 - val_accuracy: 0.8078 - val_root_mean_squared_error: 0.3944\n",
            "Epoch 15/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.5257 - accuracy: 0.7837 - root_mean_squared_error: 0.4134 - val_loss: 0.4876 - val_accuracy: 0.8188 - val_root_mean_squared_error: 0.3926\n",
            "Epoch 16/100\n",
            "50/50 [==============================] - 35s 710ms/step - loss: 0.5085 - accuracy: 0.7931 - root_mean_squared_error: 0.4049 - val_loss: 0.4870 - val_accuracy: 0.8203 - val_root_mean_squared_error: 0.3923\n",
            "Epoch 17/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.5319 - accuracy: 0.7738 - root_mean_squared_error: 0.4165 - val_loss: 0.4860 - val_accuracy: 0.8250 - val_root_mean_squared_error: 0.3916\n",
            "Epoch 18/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.5227 - accuracy: 0.7744 - root_mean_squared_error: 0.4122 - val_loss: 0.4807 - val_accuracy: 0.8281 - val_root_mean_squared_error: 0.3888\n",
            "Epoch 19/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.5275 - accuracy: 0.7869 - root_mean_squared_error: 0.4131 - val_loss: 0.4830 - val_accuracy: 0.8266 - val_root_mean_squared_error: 0.3898\n",
            "Epoch 20/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.5134 - accuracy: 0.7937 - root_mean_squared_error: 0.4064 - val_loss: 0.4807 - val_accuracy: 0.8250 - val_root_mean_squared_error: 0.3887\n",
            "Epoch 21/100\n",
            "50/50 [==============================] - 35s 706ms/step - loss: 0.5237 - accuracy: 0.7725 - root_mean_squared_error: 0.4134 - val_loss: 0.4823 - val_accuracy: 0.8281 - val_root_mean_squared_error: 0.3895\n",
            "Epoch 22/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.5137 - accuracy: 0.7944 - root_mean_squared_error: 0.4071 - val_loss: 0.4817 - val_accuracy: 0.8281 - val_root_mean_squared_error: 0.3891\n",
            "Epoch 23/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.5178 - accuracy: 0.7894 - root_mean_squared_error: 0.4098 - val_loss: 0.4776 - val_accuracy: 0.8297 - val_root_mean_squared_error: 0.3869\n",
            "Epoch 24/100\n",
            "50/50 [==============================] - 35s 709ms/step - loss: 0.5174 - accuracy: 0.7900 - root_mean_squared_error: 0.4076 - val_loss: 0.4782 - val_accuracy: 0.8297 - val_root_mean_squared_error: 0.3871\n",
            "Epoch 25/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4953 - accuracy: 0.8037 - root_mean_squared_error: 0.3981 - val_loss: 0.4751 - val_accuracy: 0.8359 - val_root_mean_squared_error: 0.3853\n",
            "Epoch 26/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.5121 - accuracy: 0.7906 - root_mean_squared_error: 0.4066 - val_loss: 0.4754 - val_accuracy: 0.8297 - val_root_mean_squared_error: 0.3856\n",
            "Epoch 27/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.5221 - accuracy: 0.7887 - root_mean_squared_error: 0.4096 - val_loss: 0.4769 - val_accuracy: 0.8297 - val_root_mean_squared_error: 0.3863\n",
            "Epoch 28/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.5080 - accuracy: 0.7994 - root_mean_squared_error: 0.4026 - val_loss: 0.4779 - val_accuracy: 0.8281 - val_root_mean_squared_error: 0.3869\n",
            "Epoch 29/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4963 - accuracy: 0.8075 - root_mean_squared_error: 0.3971 - val_loss: 0.4755 - val_accuracy: 0.8359 - val_root_mean_squared_error: 0.3856\n",
            "Epoch 30/100\n",
            "50/50 [==============================] - 35s 706ms/step - loss: 0.5246 - accuracy: 0.7769 - root_mean_squared_error: 0.4115 - val_loss: 0.4729 - val_accuracy: 0.8375 - val_root_mean_squared_error: 0.3842\n",
            "Epoch 31/100\n",
            "50/50 [==============================] - 35s 705ms/step - loss: 0.4928 - accuracy: 0.8119 - root_mean_squared_error: 0.3954 - val_loss: 0.4710 - val_accuracy: 0.8391 - val_root_mean_squared_error: 0.3830\n",
            "Epoch 32/100\n",
            "50/50 [==============================] - 35s 706ms/step - loss: 0.4989 - accuracy: 0.7950 - root_mean_squared_error: 0.3994 - val_loss: 0.4713 - val_accuracy: 0.8375 - val_root_mean_squared_error: 0.3833\n",
            "Epoch 33/100\n",
            "50/50 [==============================] - 35s 705ms/step - loss: 0.5062 - accuracy: 0.8081 - root_mean_squared_error: 0.4015 - val_loss: 0.4653 - val_accuracy: 0.8453 - val_root_mean_squared_error: 0.3800\n",
            "Epoch 34/100\n",
            "50/50 [==============================] - 35s 706ms/step - loss: 0.5047 - accuracy: 0.7869 - root_mean_squared_error: 0.4017 - val_loss: 0.4700 - val_accuracy: 0.8422 - val_root_mean_squared_error: 0.3824\n",
            "Epoch 35/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.5063 - accuracy: 0.7962 - root_mean_squared_error: 0.4022 - val_loss: 0.4663 - val_accuracy: 0.8438 - val_root_mean_squared_error: 0.3803\n",
            "Epoch 36/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.5035 - accuracy: 0.7912 - root_mean_squared_error: 0.4030 - val_loss: 0.4661 - val_accuracy: 0.8438 - val_root_mean_squared_error: 0.3803\n",
            "Epoch 37/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4870 - accuracy: 0.7950 - root_mean_squared_error: 0.3932 - val_loss: 0.4683 - val_accuracy: 0.8406 - val_root_mean_squared_error: 0.3815\n",
            "Epoch 38/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.5022 - accuracy: 0.8031 - root_mean_squared_error: 0.4001 - val_loss: 0.4672 - val_accuracy: 0.8375 - val_root_mean_squared_error: 0.3809\n",
            "Epoch 39/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4917 - accuracy: 0.8100 - root_mean_squared_error: 0.3946 - val_loss: 0.4667 - val_accuracy: 0.8375 - val_root_mean_squared_error: 0.3806\n",
            "Epoch 40/100\n",
            "50/50 [==============================] - 35s 706ms/step - loss: 0.4936 - accuracy: 0.8144 - root_mean_squared_error: 0.3959 - val_loss: 0.4664 - val_accuracy: 0.8422 - val_root_mean_squared_error: 0.3803\n",
            "Epoch 41/100\n",
            "50/50 [==============================] - 35s 706ms/step - loss: 0.5084 - accuracy: 0.8012 - root_mean_squared_error: 0.4027 - val_loss: 0.4659 - val_accuracy: 0.8406 - val_root_mean_squared_error: 0.3801\n",
            "Epoch 42/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4908 - accuracy: 0.8169 - root_mean_squared_error: 0.3945 - val_loss: 0.4623 - val_accuracy: 0.8438 - val_root_mean_squared_error: 0.3779\n",
            "Epoch 43/100\n",
            "50/50 [==============================] - 35s 706ms/step - loss: 0.5140 - accuracy: 0.8019 - root_mean_squared_error: 0.4051 - val_loss: 0.4617 - val_accuracy: 0.8469 - val_root_mean_squared_error: 0.3779\n",
            "Epoch 44/100\n",
            "50/50 [==============================] - 35s 706ms/step - loss: 0.4893 - accuracy: 0.8144 - root_mean_squared_error: 0.3923 - val_loss: 0.4587 - val_accuracy: 0.8531 - val_root_mean_squared_error: 0.3761\n",
            "Epoch 45/100\n",
            "50/50 [==============================] - 35s 706ms/step - loss: 0.4855 - accuracy: 0.8150 - root_mean_squared_error: 0.3908 - val_loss: 0.4618 - val_accuracy: 0.8500 - val_root_mean_squared_error: 0.3778\n",
            "Epoch 46/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.5023 - accuracy: 0.8019 - root_mean_squared_error: 0.3994 - val_loss: 0.4564 - val_accuracy: 0.8547 - val_root_mean_squared_error: 0.3746\n",
            "Epoch 47/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4913 - accuracy: 0.8106 - root_mean_squared_error: 0.3932 - val_loss: 0.4592 - val_accuracy: 0.8531 - val_root_mean_squared_error: 0.3760\n",
            "Epoch 48/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4957 - accuracy: 0.8081 - root_mean_squared_error: 0.3958 - val_loss: 0.4612 - val_accuracy: 0.8500 - val_root_mean_squared_error: 0.3775\n",
            "Epoch 49/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4876 - accuracy: 0.8106 - root_mean_squared_error: 0.3917 - val_loss: 0.4531 - val_accuracy: 0.8562 - val_root_mean_squared_error: 0.3726\n",
            "Epoch 50/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4863 - accuracy: 0.8087 - root_mean_squared_error: 0.3925 - val_loss: 0.4557 - val_accuracy: 0.8531 - val_root_mean_squared_error: 0.3743\n",
            "Epoch 51/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4875 - accuracy: 0.8188 - root_mean_squared_error: 0.3916 - val_loss: 0.4573 - val_accuracy: 0.8516 - val_root_mean_squared_error: 0.3752\n",
            "Epoch 52/100\n",
            "50/50 [==============================] - 35s 706ms/step - loss: 0.4910 - accuracy: 0.8144 - root_mean_squared_error: 0.3937 - val_loss: 0.4530 - val_accuracy: 0.8547 - val_root_mean_squared_error: 0.3726\n",
            "Epoch 53/100\n",
            "50/50 [==============================] - 35s 706ms/step - loss: 0.4874 - accuracy: 0.8200 - root_mean_squared_error: 0.3913 - val_loss: 0.4560 - val_accuracy: 0.8516 - val_root_mean_squared_error: 0.3744\n",
            "Epoch 54/100\n",
            "50/50 [==============================] - 35s 705ms/step - loss: 0.4762 - accuracy: 0.8200 - root_mean_squared_error: 0.3867 - val_loss: 0.4565 - val_accuracy: 0.8500 - val_root_mean_squared_error: 0.3747\n",
            "Epoch 55/100\n",
            "50/50 [==============================] - 35s 706ms/step - loss: 0.5072 - accuracy: 0.8037 - root_mean_squared_error: 0.4011 - val_loss: 0.4564 - val_accuracy: 0.8500 - val_root_mean_squared_error: 0.3747\n",
            "Epoch 56/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4917 - accuracy: 0.8244 - root_mean_squared_error: 0.3926 - val_loss: 0.4495 - val_accuracy: 0.8547 - val_root_mean_squared_error: 0.3708\n",
            "Epoch 57/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4816 - accuracy: 0.8169 - root_mean_squared_error: 0.3897 - val_loss: 0.4553 - val_accuracy: 0.8516 - val_root_mean_squared_error: 0.3738\n",
            "Epoch 58/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4844 - accuracy: 0.8200 - root_mean_squared_error: 0.3885 - val_loss: 0.4534 - val_accuracy: 0.8516 - val_root_mean_squared_error: 0.3728\n",
            "Epoch 59/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.5093 - accuracy: 0.8019 - root_mean_squared_error: 0.4028 - val_loss: 0.4508 - val_accuracy: 0.8531 - val_root_mean_squared_error: 0.3713\n",
            "Epoch 60/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4593 - accuracy: 0.8369 - root_mean_squared_error: 0.3777 - val_loss: 0.4496 - val_accuracy: 0.8516 - val_root_mean_squared_error: 0.3710\n",
            "Epoch 61/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4815 - accuracy: 0.8269 - root_mean_squared_error: 0.3872 - val_loss: 0.4484 - val_accuracy: 0.8531 - val_root_mean_squared_error: 0.3703\n",
            "Epoch 62/100\n",
            "50/50 [==============================] - 35s 706ms/step - loss: 0.4870 - accuracy: 0.8188 - root_mean_squared_error: 0.3913 - val_loss: 0.4526 - val_accuracy: 0.8516 - val_root_mean_squared_error: 0.3723\n",
            "Epoch 63/100\n",
            "50/50 [==============================] - 35s 706ms/step - loss: 0.4824 - accuracy: 0.8238 - root_mean_squared_error: 0.3880 - val_loss: 0.4444 - val_accuracy: 0.8562 - val_root_mean_squared_error: 0.3677\n",
            "Epoch 64/100\n",
            "50/50 [==============================] - 35s 706ms/step - loss: 0.4711 - accuracy: 0.8238 - root_mean_squared_error: 0.3839 - val_loss: 0.4485 - val_accuracy: 0.8531 - val_root_mean_squared_error: 0.3700\n",
            "Epoch 65/100\n",
            "50/50 [==============================] - 35s 706ms/step - loss: 0.4943 - accuracy: 0.8119 - root_mean_squared_error: 0.3945 - val_loss: 0.4460 - val_accuracy: 0.8531 - val_root_mean_squared_error: 0.3689\n",
            "Epoch 66/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4668 - accuracy: 0.8263 - root_mean_squared_error: 0.3798 - val_loss: 0.4473 - val_accuracy: 0.8516 - val_root_mean_squared_error: 0.3694\n",
            "Epoch 67/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4761 - accuracy: 0.8156 - root_mean_squared_error: 0.3872 - val_loss: 0.4470 - val_accuracy: 0.8531 - val_root_mean_squared_error: 0.3690\n",
            "Epoch 68/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4816 - accuracy: 0.8213 - root_mean_squared_error: 0.3904 - val_loss: 0.4462 - val_accuracy: 0.8578 - val_root_mean_squared_error: 0.3684\n",
            "Epoch 69/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4824 - accuracy: 0.8300 - root_mean_squared_error: 0.3876 - val_loss: 0.4454 - val_accuracy: 0.8562 - val_root_mean_squared_error: 0.3683\n",
            "Epoch 70/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4732 - accuracy: 0.8281 - root_mean_squared_error: 0.3838 - val_loss: 0.4480 - val_accuracy: 0.8516 - val_root_mean_squared_error: 0.3698\n",
            "Epoch 71/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4676 - accuracy: 0.8300 - root_mean_squared_error: 0.3816 - val_loss: 0.4460 - val_accuracy: 0.8562 - val_root_mean_squared_error: 0.3684\n",
            "Epoch 72/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4822 - accuracy: 0.8206 - root_mean_squared_error: 0.3884 - val_loss: 0.4450 - val_accuracy: 0.8531 - val_root_mean_squared_error: 0.3679\n",
            "Epoch 73/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4757 - accuracy: 0.8300 - root_mean_squared_error: 0.3852 - val_loss: 0.4427 - val_accuracy: 0.8578 - val_root_mean_squared_error: 0.3669\n",
            "Epoch 74/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4620 - accuracy: 0.8344 - root_mean_squared_error: 0.3780 - val_loss: 0.4451 - val_accuracy: 0.8562 - val_root_mean_squared_error: 0.3678\n",
            "Epoch 75/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4802 - accuracy: 0.8200 - root_mean_squared_error: 0.3892 - val_loss: 0.4424 - val_accuracy: 0.8562 - val_root_mean_squared_error: 0.3665\n",
            "Epoch 76/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4796 - accuracy: 0.8256 - root_mean_squared_error: 0.3867 - val_loss: 0.4434 - val_accuracy: 0.8562 - val_root_mean_squared_error: 0.3670\n",
            "Epoch 77/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4765 - accuracy: 0.8300 - root_mean_squared_error: 0.3830 - val_loss: 0.4441 - val_accuracy: 0.8562 - val_root_mean_squared_error: 0.3675\n",
            "Epoch 78/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4774 - accuracy: 0.8300 - root_mean_squared_error: 0.3843 - val_loss: 0.4456 - val_accuracy: 0.8500 - val_root_mean_squared_error: 0.3682\n",
            "Epoch 79/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4646 - accuracy: 0.8288 - root_mean_squared_error: 0.3802 - val_loss: 0.4424 - val_accuracy: 0.8562 - val_root_mean_squared_error: 0.3664\n",
            "Epoch 80/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4584 - accuracy: 0.8356 - root_mean_squared_error: 0.3761 - val_loss: 0.4380 - val_accuracy: 0.8594 - val_root_mean_squared_error: 0.3638\n",
            "Epoch 81/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4701 - accuracy: 0.8281 - root_mean_squared_error: 0.3820 - val_loss: 0.4350 - val_accuracy: 0.8594 - val_root_mean_squared_error: 0.3626\n",
            "Epoch 82/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4617 - accuracy: 0.8306 - root_mean_squared_error: 0.3783 - val_loss: 0.4431 - val_accuracy: 0.8547 - val_root_mean_squared_error: 0.3670\n",
            "Epoch 83/100\n",
            "50/50 [==============================] - 35s 709ms/step - loss: 0.4708 - accuracy: 0.8313 - root_mean_squared_error: 0.3817 - val_loss: 0.4432 - val_accuracy: 0.8547 - val_root_mean_squared_error: 0.3670\n",
            "Epoch 84/100\n",
            "50/50 [==============================] - 35s 709ms/step - loss: 0.4531 - accuracy: 0.8469 - root_mean_squared_error: 0.3735 - val_loss: 0.4372 - val_accuracy: 0.8531 - val_root_mean_squared_error: 0.3637\n",
            "Epoch 85/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4835 - accuracy: 0.8250 - root_mean_squared_error: 0.3879 - val_loss: 0.4407 - val_accuracy: 0.8516 - val_root_mean_squared_error: 0.3652\n",
            "Epoch 86/100\n",
            "50/50 [==============================] - 35s 709ms/step - loss: 0.4619 - accuracy: 0.8238 - root_mean_squared_error: 0.3786 - val_loss: 0.4340 - val_accuracy: 0.8531 - val_root_mean_squared_error: 0.3622\n",
            "Epoch 87/100\n",
            "50/50 [==============================] - 35s 709ms/step - loss: 0.4602 - accuracy: 0.8381 - root_mean_squared_error: 0.3759 - val_loss: 0.4402 - val_accuracy: 0.8516 - val_root_mean_squared_error: 0.3650\n",
            "Epoch 88/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4663 - accuracy: 0.8313 - root_mean_squared_error: 0.3801 - val_loss: 0.4410 - val_accuracy: 0.8500 - val_root_mean_squared_error: 0.3656\n",
            "Epoch 89/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4578 - accuracy: 0.8406 - root_mean_squared_error: 0.3751 - val_loss: 0.4334 - val_accuracy: 0.8531 - val_root_mean_squared_error: 0.3616\n",
            "Epoch 90/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4790 - accuracy: 0.8250 - root_mean_squared_error: 0.3861 - val_loss: 0.4365 - val_accuracy: 0.8531 - val_root_mean_squared_error: 0.3628\n",
            "Epoch 91/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4669 - accuracy: 0.8344 - root_mean_squared_error: 0.3790 - val_loss: 0.4382 - val_accuracy: 0.8516 - val_root_mean_squared_error: 0.3639\n",
            "Epoch 92/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4847 - accuracy: 0.8263 - root_mean_squared_error: 0.3877 - val_loss: 0.4385 - val_accuracy: 0.8516 - val_root_mean_squared_error: 0.3641\n",
            "Epoch 93/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4647 - accuracy: 0.8338 - root_mean_squared_error: 0.3793 - val_loss: 0.4388 - val_accuracy: 0.8500 - val_root_mean_squared_error: 0.3644\n",
            "Epoch 94/100\n",
            "50/50 [==============================] - 35s 708ms/step - loss: 0.4668 - accuracy: 0.8363 - root_mean_squared_error: 0.3788 - val_loss: 0.4344 - val_accuracy: 0.8531 - val_root_mean_squared_error: 0.3619\n",
            "Epoch 95/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4773 - accuracy: 0.8300 - root_mean_squared_error: 0.3838 - val_loss: 0.4379 - val_accuracy: 0.8547 - val_root_mean_squared_error: 0.3638\n",
            "Epoch 96/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4572 - accuracy: 0.8400 - root_mean_squared_error: 0.3742 - val_loss: 0.4382 - val_accuracy: 0.8547 - val_root_mean_squared_error: 0.3638\n",
            "Epoch 97/100\n",
            "50/50 [==============================] - 35s 705ms/step - loss: 0.4765 - accuracy: 0.8300 - root_mean_squared_error: 0.3840 - val_loss: 0.4384 - val_accuracy: 0.8547 - val_root_mean_squared_error: 0.3639\n",
            "Epoch 98/100\n",
            "50/50 [==============================] - 35s 706ms/step - loss: 0.4637 - accuracy: 0.8269 - root_mean_squared_error: 0.3783 - val_loss: 0.4337 - val_accuracy: 0.8578 - val_root_mean_squared_error: 0.3613\n",
            "Epoch 99/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4786 - accuracy: 0.8269 - root_mean_squared_error: 0.3852 - val_loss: 0.4337 - val_accuracy: 0.8562 - val_root_mean_squared_error: 0.3613\n",
            "Epoch 100/100\n",
            "50/50 [==============================] - 35s 707ms/step - loss: 0.4539 - accuracy: 0.8431 - root_mean_squared_error: 0.3727 - val_loss: 0.4362 - val_accuracy: 0.8547 - val_root_mean_squared_error: 0.3628\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 851
        },
        "id": "2rlLipDPVy6Q",
        "outputId": "df0da787-6ed2-4ddc-a6a5-d87139145f69"
      },
      "source": [
        "def plot_acc(history):\n",
        "  \n",
        "    plt.title('Training and validation accuracy')\n",
        "    plt.plot(history.history['accuracy'][10:], label='accuracy')\n",
        "    plt.plot(history.history['val_accuracy'][10:], label = 'val_accuracy')\n",
        "    plt.xlabel('Epoch')\n",
        "    plt.ylabel('Accuracy')\n",
        "    plt.legend(loc='lower right')\n",
        "    plt.show()\n",
        "\n",
        "def plot_loss(history):\n",
        "\n",
        "    plt.title('Training and validation loss')\n",
        "    plt.plot(history.history['loss'][10:], label='loss')\n",
        "    plt.plot(history.history['val_loss'][10:], label = 'val_loss')\n",
        "    plt.xlabel('Epoch')\n",
        "    plt.ylabel('Loss')\n",
        "    plt.legend(loc='lower right')\n",
        "    plt.show()\n",
        "\n",
        "\n",
        "def plot_rmse(history):\n",
        "  \n",
        "    plt.title('Training and validation RMSE')\n",
        "    plt.plot(history.history['root_mean_squared_error'][10:], label='RMSE')\n",
        "    plt.plot(history.history['val_root_mean_squared_error'][10:], label = 'val_RMSE')\n",
        "    plt.xlabel('Epoch')\n",
        "    plt.ylabel('RMSE')\n",
        "    plt.legend(loc='lower right')\n",
        "    plt.show()\n",
        "\n",
        "plot_loss(history)\n",
        "plot_rmse(history)\n",
        "plot_acc(history)\n",
        "    \n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEWCAYAAAB8LwAVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOy9eXhkZZm/fz9VSaVSlarse6c7va9AI8giIOAKiKCiMsro4Azujo6jzncYf6KjM6OzuIwjjjoO7us4oiggywACskhDN9Ab3U2T7nT2vapSe9X7++OcUzm1JZV0qpOm3/u6cnXq1Dmn3krS76eeXZRSaDQajUaTi2OpF6DRaDSa5YkWCI1Go9EURAuERqPRaAqiBUKj0Wg0BdECodFoNJqCaIHQaDQaTUG0QGhOCCJyp4j82WKfu5SISI+IvKoM91Uiss78/hsi8qlSzl3A61wnIncvdJ2z3PcSETm22PfVnHgqlnoBmuWLiIRsDz1ADEiZj9+rlPpRqfdSSl1ejnNf7Cil3rcY9xGRbuAFoFIplTTv/SOg5N+h5tRDC4SmKEqpGut7EekBblBK3Zt7nohUWJuORqN58aBdTJp5Y7kQROT/icgg8B0RqReR34rIiIhMmN+vsF3zgIjcYH5/vYg8LCL/Zp77gohcvsBzV4vIgyISFJF7ReRmEflhkXWXssbPicgfzPvdLSJNtuffISJHRGRMRD45y8/nXBEZFBGn7dgbReQZ8/tzRORREZkUkQER+ZqIuIrc67si8g+2x58wr+kXkT/POfd1IrJTRAIi0isin7E9/aD576SIhETkfOtna7v+ZSLyhIhMmf++rNSfzWyIyGbz+kkR2SMiV9meu0JE9pr37BORj5vHm8zfz6SIjIvIQyKi96sTjP6BaxZKG9AArALeg/G39B3z8UogAnxtluvPBZ4DmoB/Af5bRGQB5/4Y+CPQCHwGeMcsr1nKGt8OvAtoAVyAtWFtAf7TvH+H+XorKIBS6nFgGnhFzn1/bH6fAj5qvp/zgVcCH5hl3ZhruMxcz6uB9UBu/GMaeCdQB7wOeL+IvMF87uXmv3VKqRql1KM5924Abge+ar63LwG3i0hjznvI+9nMseZK4DfA3eZ1fwn8SEQ2mqf8N4a70gdsA+4zj38MOAY0A63A3wG6L9AJRguEZqGkgU8rpWJKqYhSakwp9b9KqbBSKgj8I3DxLNcfUUr9l1IqBXwPaMfYCEo+V0RWAi8FblJKxZVSDwO3FXvBEtf4HaXUAaVUBPg5sN08/mbgt0qpB5VSMeBT5s+gGD8B3gYgIj7gCvMYSqknlVKPKaWSSqke4JsF1lGIt5rr262UmsYQRPv7e0Ap9axSKq2UesZ8vVLuC4agHFRK/cBc10+A/cDrbecU+9nMxnlADfAF83d0H/BbzJ8NkAC2iIhfKTWhlHrKdrwdWKWUSiilHlK6cdwJRwuEZqGMKKWi1gMR8YjIN00XTADDpVFnd7PkMGh9o5QKm9/WzPPcDmDcdgygt9iCS1zjoO37sG1NHfZ7mxv0WLHXwrAW3iQiVcCbgKeUUkfMdWww3SeD5jr+CcOamIusNQBHct7fuSJyv+lCmwLeV+J9rXsfyTl2BOi0PS72s5lzzUopu5ja73sNhngeEZHfi8j55vF/BQ4Bd4vIYRH529LehmYx0QKhWSi5n+Y+BmwEzlVK+ZlxaRRzGy0GA0CDiHhsx7pmOf941jhgv7f5mo3FTlZK7cXYCC8n270EhqtqP7DeXMffLWQNGG4yOz/GsKC6lFK1wDds953r03c/huvNzkqgr4R1zXXfrpz4Qea+SqknlFJXY7iffoVhmaCUCiqlPqaUWgNcBfy1iLzyONeimSdaIDSLhQ/Dpz9p+rM/Xe4XND+R7wA+IyIu89Pn62e55HjW+AvgShG50Awof5a5///8GPgIhhD9T846AkBIRDYB7y9xDT8HrheRLaZA5a7fh2FRRUXkHAxhshjBcImtKXLvO4ANIvJ2EakQkWuBLRjuoOPhcQxr429EpFJELsH4Hf3U/J1dJyK1SqkExs8kDSAiV4rIOjPWNIURt5nNpacpA1ogNIvFV4BqYBR4DPjdCXrd6zACvWPAPwA/w6jXKMSC16iU2gN8EGPTHwAmMIKos2HFAO5TSo3ajn8cY/MOAv9lrrmUNdxpvof7MNwv9+Wc8gHgsyISBG7C/DRuXhvGiLn8wcwMOi/n3mPAlRhW1hjwN8CVOeueN0qpOIYgXI7xc/868E6l1H7zlHcAPaar7X0Yv08wgvD3AiHgUeDrSqn7j2ctmvkjOu6jeTEhIj8D9iulym7BaDQvdrQFoTmpEZGXishaEXGYaaBXY/iyNRrNcaIrqTUnO23ALzECxseA9yuldi7tkjSaFwfaxaTRaDSagmgXk0aj0WgK8qJxMTU1Nanu7u6lXoZGo9GcVDz55JOjSqnmQs+9aASiu7ubHTt2LPUyNBqN5qRCRHIr6DNoF5NGo9FoCqIFQqPRaDQF0QKh0Wg0moJogdBoNBpNQbRAaDQajaYgWiA0Go1GUxAtEBqNRqMpiBaIaADu/zwce3KpV6LRaDTLCi0Q6ST8/gvQ+/hSr0Sj0WiWFVog3LXGv9GppV2HRqPRLDO0QDicUFUL0cmlXolGo9EsK7RAgGFFRLRAaDQajR0tEADV2oLQaDSaXLRAALjrdAxCo9FoctACAVBdp11MGo1Gk4MWCDBiENrFpNFoNFlogQDDxaQtCI1Go8lCCwQYLqZkBJKxpV6JRqPRLBu0QIBhQYAOVGs0Go2NsgqEiFwmIs+JyCER+dsCz18vIiMissv8uiHneb+IHBORr5VznRmB0G4mjUajyVBRrhuLiBO4GXg1cAx4QkRuU0rtzTn1Z0qpDxW5zeeAB8u1xgzVlgWhBUKj0WgsymlBnAMcUkodVkrFgZ8CV5d6sYicBbQCd5dpfTNoF5NGo9HkUU6B6AR6bY+PmcdyuUZEnhGRX4hIF4CIOIAvAh+f7QVE5D0iskNEdoyMjCx8pdXaxaTRaDS5LHWQ+jdAt1LqdOAe4Hvm8Q8Adyiljs12sVLqW0qps5VSZzc3Ny98FZmOrlogNBqNxqJsMQigD+iyPV5hHsuglBqzPfw28C/m9+cDF4nIB4AawCUiIaVUXqB7UdBBao1Go8mjnALxBLBeRFZjCMOfAG+3nyAi7UqpAfPhVcA+AKXUdbZzrgfOLps4AFS4oNKjLQiNRqOxUTaBUEolReRDwF2AE7hFKbVHRD4L7FBK3QZ8WESuApLAOHB9udYzJ+46LRAajUZjo5wWBEqpO4A7co7dZPv+RuDGOe7xXeC7ZVheNnomhEaj0WSx1EHq5UO1bvmt0Wg0drRAWGgXk0aj0WShBcKiug4i2oLQaDQaCy0QFraZEEqpJV6MRqPRLD1aICzcdRAL8LnbnuVt//XYUq9Go9FolhwtEBZmu42nDx7lscPjDE5Fl3hBGo1Gs7RogbAwq6nHx4YB+P2B4aVcjUaj0Sw5WiAszH5MXhUC4IHnFt787779Q3z617sXZVkajUazVGiBsDBdTLUyzbmrG3j44CiJVHpBt/rd7kG+9+gRxqfji7lCjUajOaFogbAwXUxNFWGuf1k3wViSJ49MLOhWYyFDGJ7u1XUVGo3m5EULhIVpQaz3p7hwfRMVDlmwm2nUtBx2aoHQaDQnMVogTFSVH4A1NUl87krO7q7ngecWFqgen44BsEsLhEajOYnRAmEyFHESV05WuI3N/dKNLewfDDIwFZn3vewuJl10p9FoTla0QJjsGwoyhZdWl1H/cMnGFgB+P083UySeIhxPsarRw1QkQc9YeNHXqtFoNCcCLRAm+wYCBJSXOoexoW9oraG91s3983QzjZnupVduagVgV292oDueTBNPLiw7SqPRaE4kWiBM9g0EiTp9uOIBAESESzY284dDY/Pa0C330nlrGvC4nOw6mh2HeN8Pn+R9P3xy8Rau0Wg0ZUILhMm+gQBpW8M+gIs3tBCKJXn6WOnBZsuCaPG7Oa2zNitQfWg4yH37h3nmmO4aq9Folj9aIIBoIsXhkRCV3vqsqXKnrzCqq58bDOZdo5QiEk/lHR81LYhGr4vtK+vYOxAgmjDO++FjR81zYoRiyUV/HxqNRrOYaIEADg6FSCuo9jdlTZVrr3XjdTk5NBzKu+bO3YOc/Q/3MBVJZB23qqcba1yc2VVHIqXYOxAgHE/yv08do8HrAuDI2HQZ35FGo9EcP1ogMNxLALX1pkCYqakiwrpWHweH8y2Ixw6PMR1P5W30Y6EY1ZVOPK4KtnfVA0a66227+glGk3zklesBOLLMsptSaaVTcjUaTRZaIIC9AwE8Lif++mZQKYjNCML6lhoODuVbEPsHjHP6J7PrJMZC8YyV0FbrptVfxa7eSX7w2BE2tfm45qwVAPQsIwsiEE2w/e/v5oEDC29QqNFoXnxogQD2DwbY2ObD4THabdgD1etbahgOxpgKz7iSlFLsGzSsjv7J7LkRY9Nxmmpcmcfbu+q4e88Qe/oDXHfeKmqqKmiqqeLIaL4FcdeeQfb2BxbzrZXE4FSUYCzJoQJCqNFoTl1OeYFQSrFvIMjmdn+mYZ89DrG+tQaAQyMzVkXfZIRg1Agy51kQ0zEaa6oyj7d31RNJpPC6nLzxzE4Auhs9eRZEOq346M928aV7DizemyuRSVP8AtHEHGdqNJpTiVNeIAamokxFEqZAGFlL9kym9S0+gCw3k+VeEjGut2N3MYFhQQC88SWd1FRVALCq0ZsXg3hhbJpwPMXuvhOfAjsZNgLrgYgWCI1GM8MpLxBtfjcPfPwSXndae6ajq93F1FlXjbvSwUFbJtN+0710xoo6+mwWhFKKsVCcRpuL6ezuet578Ro+eOm6zLHuRg+DgWhWmqwlDIOBKCPB2KK8t2eOTXLV1x4mOIdlMBmxLAideqvRaGYoq0CIyGUi8pyIHBKRvy3w/PUiMiIiu8yvG8zj20XkURHZIyLPiMi15VqjwyF0N3mNT/2Wi8lmQTgcwrqWmiyB2DcQZGWDh/UtNVnN/EKxJPFUmibvjIup0ungxss3015bnTnW3eQF4Oj4jBVhjz0slhVx954hnjk2xf4CdRx2rPiKtiA0Go2dsgmEiDiBm4HLgS3A20RkS4FTf6aU2m5+fds8FgbeqZTaClwGfEVE6sq11gzV+TEIMNxMh4ZmNtl9gwE2tfnoqKtmOBjLtOKw2mzYXUyF6G40BMIeh9jdP8WaJi8i8OwiCcTufuM+PaOzZ0xNRkwXk45BaDQaG+W0IM4BDimlDiul4sBPgatLuVApdUApddD8vh8YBprLtlILlw+QLBcTwLqWGvqnogSjCSLxFD2j02xu99NR50YpGAoYcQirzYbdxVSIlY0eYKZYTinFnv4A56xuYHWTd1EEQimVsUTmqrmwiv0CEe1i0mg0M5RTIDqBXtvjY+axXK4x3Ui/EJGu3CdF5BzABTxfnmXacDiMQHUkXyAAnh+Z5sBQkLSCze2GBQEzgWrLgmiyZTEVora6kgavixfMVNf+qSiT4QRbO/yc1lm7KC6moUAs0/ZjrpoLncWk0WgKsdRB6t8A3Uqp04F7gO/ZnxSRduAHwLuUUnktVUXkPSKyQ0R2jIwsUpFXdV2eBbHeFIiDQ8FMgHpzuz8TV7BSXcemS3MxAaxq9GQsCEsQtnbWclpnLQNTUUZDxxeotqyQ2urKki2IoA5SazQaG+UUiD7AbhGsMI9lUEqNKaWsnfDbwFnWcyLiB24HPqmUeqzQCyilvqWUOlspdXZz8yJ5oNx1eTGIlQ0eXE4Hh4ZD7BsI4nE56ar30FHnBqDfDFSPmZt6KQLRbUt13dMfwCGwuc3Ptk4j1fZ43Uy7+6ZwCLx6Sys9Y9OzttGwLIhQLEkypWdVaDQag3IKxBPAehFZLSIu4E+A2+wnmBaCxVXAPvO4C7gV+L5S6hdlXGM+BVxMFU4Ha5q9HBwOsW/ArLp2CB5XBXWeyiwLwldVgbvSOefLrGr00D8VIZpIsbd/irXNNVS7nGztMGZj785pCX7nswPcvWeQ/slIST2TdvcZ99zc7icYTTIRLu4+soLUgO4yq9FoMlSU68ZKqaSIfAi4C3ACtyil9ojIZ4EdSqnbgA+LyFVAEhgHrjcvfyvwcqBRRKxj1yuldpVrvRmq6yHQl3d4XUsNu3onCUaTvO70GV3rqK1mYHImBtEwR4DaorvRi1JwbCLM7r4A561pAMDnrmRNTqB6R8847//RU5nHjV4Xbz5rBTdesbno/Xf3T/GytU10mwHxnrHpopbNZDhBbXUlU5EEgUiSOk9p70Gj0by4KWsMQil1h1Jqg1JqrVLqH81jN5nigFLqRqXUVqXUGUqpS5VS+83jP1RKVdrSX7efEHEA8LVBKH/M6PoWH8cmIkbVdZsvc7yjzp0plhubjtFYgnsJDAsC4MkjEwwGomztqM08ty0nUP21+w/R4HXxs/ecx2ev3kpbrZuf7ejNu6fFcDDKUCDGts5aVpkptcXaiydTaYLRJF0NRjxFB6o1Go3FUgeplx++NogFIJbduM7qyQSwqd2f+b6jrjori6lxjgwmC6sW4vZnBwHY2jlzz22dfvqnooyFYjx7bIoHnhvhLy5czblrGnnn+d28Zksbk+EEiSLxgj19RiB9W4efroZqRKCnQHNAmKme7qo3BEsXy2k0GgstELn4TPdRaCjrsJXJBLDRZkG011YzFUkwHUsyNh0v2YKo81Tid1fwyKFRALa2Z1sQYASqb77/EH53Be88f1Xm+Saf8RpWWm0uz9qyoqoqnHTUVhe1IKw+TCsbTIHQFoRGozHRApGLr834NziQdXhVo5cKh7Civhq/uzJz3Mpk6puMMD4dn7NIzkLEaPGRTCu6Gqqp9czc0xKIXz7Vx+/2DHL9Bavx2V7TqrMolgq7u8+oyraaA3Y3eegpkupq9WHqsgRCF8tpNBoTLRC5WBZEcDDrsKvCwaZ2X6Y7q4VVLLdvIEAqrWj0luZiAjLxAbv1AOB3V9Ld6OG2p/vxuJy862XdWc9bAjEyi0BYImO9TjELwurDpC0IjUaTS9mymE5ailgQAN991zlUOrM11RIIK6hcqgUBZDKMrNRWO9s6a+kZC/OO81ZRn+O2arYsiAJdX8dCMfqnomyzxTS6Gz1MhBNMhRNZlgrMpLh21huxipMlBvGzJ45yaDjEJ19XqL2XRqNZDLQFkUuVHyo9eRYEGJ/ca6uzN9hWXxUOW4O9hVgQ9k/7Fi9b24TfXcFfXLQ6fx1mDGK0QAxit9kVNteCADgynm9FWEVyDR4XvqqKk6bl9117hrh1Z/9SL0OjeVGjLYhcRAwrooAFUYgKp4NWv5vdZubQfCyIV21u4T0vX8P5axvznnvbOV286SWdBYvuPK4KPC5nwRhEpm2HLW12pntsmNNXZLvILIHwV1fir648aSyIsVCMiXAcpRQistTL0WhelGgLohC+9oIWRDHaa92ZCuT5CESdx8XfXbG5oAiIyKwV2U01VUUFYlWjJ8vSseILRwq0/Z6KJPC7K3A6BL+78qSJQYyG4qTSSgfVNZoyogWiEPOwIGAmDgFQf4KqkJtqXAUF4tBwiA2tvqxj1S4nbX53wUymyXA8Uzntr644KTZcpVSmtfp4uHCqr0ajOX60QBTCsiBK6HkEMwJR56nMC2KXi6aaKkaD2ZujUoq+yQgr6qvzzrd3j7UzGUlQZwauTxYLIhxPEU0YRYLj04sznlWj0eSjBaIQvjZIhI2K6hLoqDVqIUotklsMmnz5LqapSIJwPEVnXb5AdDd6C1oQU5FExh11ssQg7AWCxYoFNRrN8aMFohBFaiGK0W5uyPPJYDpemrwuxsPxrPbcVk+oQgKxqsnDaCiW1611KpyYcTG5K0+KLKZRm9UwoV1MGk3Z0AJRiFlqIQrRYQ4Omk+A+nhp8lWhVLYPvt/sKttRSCAaCjftm4wkqMtYEBUnxUyILAti+sUnEP94+17u35/fMFKjOdFogSjEPC0Iq93GCRWITLHczAbZN2G4kDqLxCAgez51Oq3MIPVMDAKWz0yIaCLFCwUyr+xxh4kXmUAopfjOH3q4Z9/Q3CdrNGVGC0QhalqNf0u0IBq8LtY2ezmtQMFbuSjUj6l/KoqrwlEwFrLKNhfCIhRPklZkxSBg+YweveUPL3DFvz9ELJnKOm4VCDZ6XS86CyIQTZJMK6Lx1NwnazRlRhfKFaKqxqioLtGCEBH+72OXlHdNOTTVWNXUMwLRNxmhs666YOGYz11Js6+KwyMzAmH1YcoIhNv4c5iKJLJmxS4VB4dCRBIphqZirDQFDgwXU01VBe117mVtQSRSaeLJNN6q0v+bWe8nktACoVl6tAVRjHnWQpxomnz5FkTfRKRggNpiQ2sNB4eCmcdWFfVMHYQhFMsl1fXouOEOs2Z+W4xNx2iscdHgrWJ8GQvEF+8+wDX/+ci8rrFiSlogNMsBLRDF8LXNq5r6ROOrqsBV4cjqx9Q/GcnEQwqxvsXHweEQ6bRR32E16suNQcxWLPfMsUkO2ETGIhhN8A+/3cvOoxPzfzNFsARiIFcgQsbcjQZP5bJ2MfWMTvPcUJB4svSg/7j5+4xoF5NmGaAFohi+9mVtQYgIzTVVmY6usWSK4WCsYAaTxYZWH+F4KpMOm7EgbFlMUNyCSKcV13/nCa786sN8/9EelFlI2Dse5s3/+SjffvgF3vfDJzNDiI6HcDzJiPnerOwsi9FQjMaaKhq8VcvaxRSIJlDKEO5SsSyIqLYgNMsALRDFsCyIEqupl4KmGldmJsSgOfZ0LhcTkLEArGFBVgvwjIupSLHcnv4A49Nx2mrd3PTrPXz4p7v4/YERrr75DwxMRbjpyi2MheJ88le7M+KxUHrHZzZV671ZjE3Haapx0eCtZDqeWrabqSW0vROFhzUVQscgNMsJLRDF8LVDKg6RxXOZLDZGwz5jQ5mtSM5ivdmj6cCQMW97yvy0agWpa1wVxkyIIllMDx0aAeAX7zufv7lsI7c/08+f3fJHaqsr+dUHL+DPL1zNR1+9gdufGeBXu/qO671Z7qUKh2S5mNJpZUzu8xoWBCzfYrkpU2iPTczDgtACoVlG6CymYtiL5TwNS7uWIjTVVPGM2d57tiI5i9rqSlr9VZlA9WQ4gcflpKrC6BrrcIgxE6KIBfHQgVE2tflo8bv5wCXreMnKeu7aM8hfvXJDxgp538VruX//MDf9ag/nrG6cVbBmwyroO21FbZaLKRBNkEorGryGBQFGTKK9dmGvU06sWE7veOkWREYgdAxCswzQFkQxMsVyyzcO0eRzMT4dJ51W9JmfUttnCVKDEYc4MDzjYqrLGYDkK9KwLxJP8eSRCS5a35Q5dt6aRj79+q1ZU+qcDuHL124nrRQ3/vLZBb+33vEwvqoKtrT7syyITA2EmcUEy9OCSKcVwYyLqXQLwnovWiA0ywEtEMXIWBDLN5OpqaaKVFoxEY7TPxmh2VeVsQaKsaHVx6HhEKm0YjKcoDanPbnRsC/fxfT4C2PEU2kuWt8857q6Gjy8/dyVPPr8aCZjar4cHQ/T1eCho66aiXAis2GOmTGXppoqGsyCwOWY6moVIcL8LIgxm4vpeOM4Gs3xogWiGDXz68e0FMxUU8fpn4rM6l6y2NBaQzSRpnc8zFQknmdB+N0VBS2Ihw6O4qpwcM7q0txtKxu9JFKK4QJzs5/unWRX7+Ss1x8dD7OywUOb37CIBgOGm8naQA0LwhCI2Tq6Pj8S4mM/fzqvGrvcWG46l9MxrxiEFaROK4gv855Ymhc/WiCKUemG6vplb0GAkfbZNxlhRQkCMROoDjIZnpkFYVGs5ffDB0c5p7th1il3dqy19BVI8bzptj383Szup3Ra0TsRYVWjJ+MyGzDvY1kQjV5jPrhDZncx3bt3iP996hh7+0tr3b5YWFbYxjYfo6FYyS6j8ek4DrMQPhrXAqFZWsoqECJymYg8JyKHRORvCzx/vYiMiMgu8+sG23N/JiIHza8/K+c6izLP0aMnmmaf8Ql6JBibs0jOYn2Lkep6cDiUNSzIwu+uzOvFNByI8txQkAtt8Ye56JhFII6OTXNwuHgB2VAwSjyZNlxMZvC530x1HQ3FEYF6TyVOh1Dnmb0fk1WDsH8wv7ivnFgZTFs7/AAcKyHVNZFKE4gmaTWtJp3JpFlqyiYQIuIEbgYuB7YAbxORLQVO/ZlSarv59W3z2gbg08C5wDnAp0WkvlxrLcpyb7dhWhAHhoJEE+mSXEw+dyUdtW4ODAXNYUG5MYj8LKaHDo4CcOG60gXC6ijbl+NeCcWSTIQTJFKqYEU2wFGz4+zKBg9ttTkWxHSMeo+LCnNyX4PXNWuxXJ+ZAbV/4ARbENFsgSilFsKyhKzfoxYIzVJTTgviHOCQUuqwUioO/BS4usRrXwvco5QaV0pNAPcAl5VpncVZ5hZEbXUllU7hmWNGqmupKaXrW308c2yKeDJd2IKIJUnZgssPHxql0etiS7u/5LXVVFVQW12ZV0VsD9juLbJpWzUQKxs8uCudNHhdDFgxiFA8E3sAaJjDgrAsmH0DJ9aCsER2S4fR4beUOIQVbLcEIhxfHl11Nacu5RSITqDX9viYeSyXa0TkGRH5hYhYTURLulZE3iMiO0Rkx8jIyGKtewarmjq9PH3BIkKjt4pnjhkB31IsCDD84tachbwgtfk4ZLqZlFI8dHCUC9Y14XDkd4mdjY666jwX01G7QBSJCxwdD+OQGSukvdZtsyDiWe3M57IgLIHaNxg4oVlBlotpbbOXqgpHSZlMlkBYQr9cK8Q1pw5LHaT+DdCtlDodw0r43nwuVkp9Syl1tlLq7ObmudMv542vHVQKwqOLf+9FosnnylQ+l2xBmHEIoIAFkd2Paf9gkNFQbF7xB4vOuuo8F5O1Ua5p9s4qEB111VSabqT22moGpiwLIpZxrQE01LiKprmGYkmmIglW1FcTjCYzcYwTQSCaRMSwyFbUV2e1DinGxLTxM7eEMaKD1JolppwC0QdZYwVWmMcyKKXGlFJWHuS3gbNKvfaEMM/BQUuBtVl6XM68zb4YGzgBLawAACAASURBVMxMJqBADMK4h/UJ+O49Q4jAJRvmL8Ar6qsLuphqqiq4YG0TewcCBeskrBRXi446d+Y+Y9PxrMl9DR4XE+F4wftYVscrN7UAJzYOEYgkqKmqwOEQuho8HJssxYIw/it01ukgtWZ5UE6BeAJYLyKrRcQF/Alwm/0EEWm3PbwK2Gd+fxfwGhGpN4PTrzGPnVgaVhv/Du+b/bwlxBKIjiKDggqxzmZB1ObVQWTPhPjdnkHOWllPi3/uDKlcOurcBM1P8Ra9ExFW1FeztcNPKJYsGLw9OhbOTMADaKt1E4ga95kMJ2j02iwIr4u0Ius1LI6ZAnGpJRAnMJMpEElkfpZd9Z6SLIhx04LQQWrNcqFsAqGUSgIfwtjY9wE/V0rtEZHPishV5mkfFpE9IvI08GHgevPaceBzGCLzBPBZ89iJpWUruOug5+ET/tKlYheIUvFWVbDCdGPk10EYLqZgNMmRsWn2DQS4bFvbgtbWWWds8nY3U69pHWw1g7d7ctxMoViSsek4XXYLwkx13dNvBOOzLAirWK6Am8myOja2+ehqqGbfibQgoomM+HY1VDMVScw5iGkiHMfnrsBnCoseO6pZasoag1BK3aGU2qCUWquU+kfz2E1KqdvM729USm1VSp2hlLpUKbXfdu0tSql15td3yrnOojgcsOqCZS4QxgY536Z4lpupUBYTGJ+A79xtZHAtWCBMEbI2aqUUvRNGC431rTU4HZIXh+i1ZTBZtJuprrvNxoRNBQSiULFc/2QEp0No8bnZ1OY/oRbEVCSREdsV9cZ7mStQbQXgq81iRG1BaJaapQ5SL3+6L4SJF2Dq2FKvpCDN5ujRzhKK5Oy8ZGUdTTUzm5HFzNjRJHfuHuT0FbWZDW6+dOYUy42EYkQTabrqq3FXOlnfUpOxCiyOFhAIyzp6ts8Qk8aabBcTFG630T8Zpc3vxukQNrf5ODwSysoMmgzHM6Kz2AQiyRkLwvz5zZXqOjEdp94mEGFtQWiWGC0Qc9F9ofHvMrUiFuJiAnjvxWu5+6MX58UtfFXGTIjnBgM83TvJa7cuzHoAaPS6cFU4MgJh+eFXmvGFLe3+vFoIq0huVYM3c6zFb7zHPeZm3uAtzYLom4xkrJhN7X7SCg4NhzLP/92tz/KWbzxalnTSQNQWg2gw1jCXBTE+HafB46KqwvhvqS0IzVJTkkCIiFdEHOb3G0TkKhEpLWXmZKd1mxmHeGipV1KQ01fU8rrT2udV5QxQ6XRkbbQWDodQU1XB7c8YmVuXL9C9ZN2r01YLYW2Q1ifqLR1+hgIxRkMzDf2OjofxuyuyWohXVThpqqnisFm70eTNtyAKpbr2TUQyVsymNsOlZsUhjoxN87vdg0QSqTkbBy4Ew8VkvIfa6kp8VRVzWhDj00YRoMMhuCsdug5Cs+SUakE8CLhFpBO4G3gH8N1yLWpZ4XAYVsQytSB87kpuvu4lC8oyKobfbYzy3NjqY01zzdwXzIK9FsISiBU2gYDsgrmj4+GMhWHH6jNV4ZCMbx/AXenE43LmuZhSacVgIJq5blWjF3elIxOHuOXhF3A6BBH44wuLm/+QSKUJx1MZF5OI0FlfPasFoZRiPDxTJe5xVSzaTIidRye47tuPnfCOtpqTn1IFQpRSYeBNwNeVUm8BtpZvWcuM7gthogcme+c89cWA9cl3ocFpO/Yaht6JMM2+Kqpdho99a3t2JlMsmeLQcCjLvWRhBaoba1x5brEGryvPxTQcjJJKq4zrzekQNrb62D8YYDIc5+c7jnH19k42tfl5/IWx436fdqxmh1bRIRgzMmazIMLxFPFkmnpTIKornYvmYnrk+TH+cGgsb7a3RjMXJQuEiJwPXAfcbh4rre/ziwErDnHkD0u7jhOEtbFdftrxC0RnnYfhYIxYMmUMAaqfiZXUeirprKtm70CAaCLFu7//JH2TkYKva40UtddAWDR48/sxWaJkj81savOzbyDIDx87QiSR4oaLVnPu6gaePDJRtLPsQrBqMvy2GpOueg+9E+Gi7T4sF5llQbgrHYsmEJZ1NRmePc1Wo8mlVIH4K+BG4FazlmENcH/5lrXMaNlqzIZ4YXnGIRabjrpq1rfUsNFWcb1QrCDxwGSU3vFIVnYSGN1Od/VO8K7vPMFDB0f4l2tO58rTO/LuY7cgcinUj8n6tG6fkbGp3cf4dJxvPXiYl29oZlObn/PWNBBNpHm2b/HiEFajPnsR4or6asLxVNG2IBmBMCf8Vbuci1YHMWZWaE8WmTWu0RSjYu5TQCn1e+D3AGawelQp9eFyLmxZkamHODUE4u+v3ko8mS65Mns2rBjAkfEwA1MRuhqyey5u6fBz994h+iYifPmt23nDmYX6OUK7udHb+zBZNHhdHBwKZR3rN9t8t+dYEGCk8L77IqNK/pzVjQA8dnics1aVNi1vLqyCuCwLwhTG3olIVpquhSUQ5XAxzVgQy280q2Z5U2oW049FxC8iXmA3sFdEPlHepS0zui+CySMweXSpV1J2/O7KghvxQlhhVlPv6BknrWYymCwu3tBMnaeS/3jbS4qKA0CHaUEUyrxq8OQ37OufjFBbXUlN1cxnoM3thkW0qc2Xyfpq8LrY0FrD4/MIVAejCZKzjAOdKmBBdJuB9xdGQwWvsdbfmHExORetDsJyv2kXk2a+lOpi2qKUCgBvAO4EVmNkMp06ZOohTo04xGLRVutGBB593ggEd+W4mM5cWc/OT72a153eXujyrPtAERdTjYtIIpWV9WNM2MuuDanzuHj/JWv5zFVbs6yjc1c38mTP+KybvkUwmuA1X36Q9//oqaLnWONGrToIgNVNRtvvPX2F231YQXa7BbFYaa7WmNbZRrNqNIUoVSAqzbqHNwC3KaUSwIlrrr8caNkC1Q2w7zdLvZKTCleFgxZfFU+bMyusojE7pbiyOuuq+fhrNvD6AvEJy29v+drBLJIrUF3+/y7bxHlrGrOOnbO6gel4Kq8vVCFuvv95Bqai3LN3iEcOFW4DPxOknrFeKpwONrf72d1fuHJ7bDpupPCaCQIe1+K4mNJplbFOtAWhmS+lCsQ3gR7ACzwoIquAEzvDcalxOODc98Fzt8OuHy/1ak4qOuuqSaQUFQ7JZCPNFxHhQ69Yn2eBgK2aenpmAzQEorTXOneNEXuYK931yNg0tzz8Alee3k5nXTX/dOe+gm3GA9EElU7Ja2OyrdPPnr7CLc6tNhuWWFa7nItSBxGIJkiar6djEJr5UpJAKKW+qpTqVEpdoQyOAJeWeW3Lj5d/3IhF3P4xGDmw1Ks5aeg04w6d9dU45zmVrhQsV9LO3gnA2BSD0WTJ7UdafG7WNHl5/PDscYh/umMfFU7hU1du4eOv3cDuvgC3Pd2fd57V6jvXMtrWUUuwSItzq82GhXuRgtSjtgJCncWkmS+lBqlrReRL1nhPEfkihjVxauFwwpv+Cyqr4RfvgsTcPf41M037cgPUi8XWDj9nr6rna/cdIppIMWBmMM2nP9W5axr4Y8941ixuO488P8pde4b44KXraPW7ufqMTrZ2+PnXu57LixXY22zY2dZpFAbuLhCHsNpsWCxWDMKKP1Q6hQntYtLMk1JdTLcAQeCt5lcAWJoW3EuNvx3e+E0Y2g13fALi00u9omWPFQsoFH9YDESET7x2I8PBGN9/tKdgkdxcnLu6kWA0mdddFoy2HZ/9zV5W1FfzFxca6bEOh/B3V2ymbzLC9x/tyTo/EE0WFIj1rTVUOqVgHMLeZgMMgUikFIkSAuezYWUwdTd6mdIuJs08KVUg1iqlPq2UOmx+/T2wppwLW9asfzVc8Few8wfwz6vhh9fA49+C6eU7u3opsYrlCsUPFotz1zRy8YZmvv7A85l+S/OZkfGydY24Kx38+Xef4K49g5nj+wYCvOUbj7B/MMgnr9iM2xZXuGBdE5dsbOZr9x3K6nNkuJjyS4yqKpxsaPUVbDFuxCBmRMVqRzKbm2lgKsLzI4XTZi0sC2JdS422IDTzplSBiIjIhdYDEbkAOLX9K6/6DLzzNnjpDTD+Atz5CfjyNrjzb2HqxI/PXs6sa/YhMtNRtVx8/DUbmQwn+M8HDlHpFFp8pddytPjc3PqBC2jxuXnvD57kr366k8/fsY8r/+NhesbCfPnaM7j8tPxU3DeftYJANJnVRjwQSeSNcrXY1lHLnv5AVsuNVFoxGUnQYGsjYgnRbNXUf/XTXVz7zdnblVsWxOomL4FooqgLTaMpRKkC8T7gZhHpEZEe4GvAe8u2qpMBEVhzMVz2T/Dhp+D9j8LWN8IfvwX/fgbc+/dLvcJlw8pGDw9+4lIu3dhS1tc5bUUtV5zWRiCapK3WjWOeAfHN7X5+/aEL+Mgr1/PbZwb45oOHectZK7jvYxfzxjNXFLzGEr39AzPT6gLRwjEIMDKZxqfjDNga502G4ygFDbYW53NNlTs2EebxF8YZDcUzrdkLMRaKU+8xCh+VmmkDotGUQqlZTE8rpc4ATgdOV0qdCbyirCs72WjdAm/8T/jwTth4OTz8JRg9tNSrWjZ0NXgWpXXHXPz1qzfgkJk51vOl0ungo6/ewB0fuYjbPnQBX7jmdOo8+cV5Ft2NXlwVDp4bMgRCKUUgkswqkrOzNROonnEz5RbJgVEHAcUF4te7jOypNr+b7z7SU7QJ4Nh0jMaaqsxoWZ3JpJkP85oop5QKmBXVAH9dhvWc/NSvgsv/BcQBT/9kqVdzyrGuxcdnrtrKuy7oPq77bGj1cfqKujnPq3A6WN9SkxlEFE2kiafSRV1Mm9v8OAR224ryrF5J9k61bksgCriYlFL8elcfZ62q54OvWMezfVM8dbRws8HRkDHnut4UOV0LcWI4PBI67gSD5cDxjBwt/8fBkxV/O6x9BTz9U0if/H8kJxvvPL+by7bN3rpjMdnY5uM5MzA+06ivcB/MapeTdS01mfGpQMbdlBWknsXFtG8gyIGhEG/Y3sGbzuzE567gu4/0FHy9sVCMxhpXZkKfrqYuP5PhOK/9yoP88qnlOcd+PhyPQOho12yc8TYIHIOeB5d6JZoys7nNz3Awxvh0vGCjvly2ddRmUl1DsSRfuucAK+qrWWub3mcJRKEA9K939VHhEF53egfeqgreenYXdz47wFAgfyDQ2HScRm/VjAUR0RYEwOBUdFFngNjpHY+QSCl6xmafQX4yMKtAiEhQRAIFvoJAflMczQybXgdVtbArx800tAf6dy3NmjRlYaMVqB4MZILAxWIQYMQhhgIxhoNRPvubPRybCPPla7dnpdBaaa65HV3TacVtT/fz8g3NmbqJd56/ipRS/OixI1nnJlJpJsMJGmtc1JmCZW9HcqqSSKV51Zd+zw9zfl6LxcCUkeA5HIjNcebyZ1aBUEr5lFL+Al8+pVRJsyROWSqrYdsbYd9tEDMzXAafhf9+DdxyGfTvzL8mGYOEHgt5smHPZCo0CyKXbeYs7i/fc5Cf7zjG+y5ey0u7s2dRZFxMOQLx+AvjDExFs1qjr2r08oqNLfzo8aNZ9RhW8Luxpgp/dSUipQep02nF44fHODQcnPNcpdRJNe96KpIgFEvSM1aeIlfLZTgcPPn/Lx+Pi0kzF9uvg0QY9v7aqI340Vuhyg/eZvjJ2yBg6+Pz/P3wxY3whZXwvdfDg/8Gw/uXbu2akmn2VdHgdfHcYLAkF9MWUyB+8sejbOv081ev2pB3jruIi+lXO/vwupy8enNr1vE/PW8VY9NxHrP1k7KC301eF06H4HdXzhmkHgnG+M8HnufSLz7Atd96jA//ZG5r9+c7ejn/8/cRiiXnPHc5YP2ORkPl+YTff6pYEMeLiFwmIs+JyCER+dtZzrtGRJSInG0+rhSR74nIsyKyT0RuLOc6y8aKl0LDWtjxHfjxWw1L4rr/gbf/DGIh+PG1RquOR2+GH74JfO1G4V14HO77HHzzIqMIT7OsERE2tvrYPxS0zYIobmD73JWZ+RBfuXY7ror8/4aF0lzjyTR37B7gtVvbMi4oizNXGhlX+wcKZEeZw5/qPZWzBql7x8Nc8q/388+/20+r382lG5vZPxhgeo6Nf1fvFOPTcR46MDLrecsFSyBGguXZwAdNC2LoRWBBlM1NJCJO4Gbg1cAx4AkRuU0ptTfnPB/wEeBx2+G3AFVKqdNExIMxwe4nSqmecq23LIjA9rfBff8A4jTEoW2b8dybb4GfXAtfeykE+mDz6+EN34AqM1A5fhhuPhf+8BV4/b8v3XvQlMSmdh8//WNvZgOezcUEcNOVW0CMtNxCuDMupplAat9khGA0ycvMaXh26jwuWv1VmWwqmJmPYQ1ZqvW4Zh0adOvOPqbjKX77lxeyrbOW+/YPcf9zIzzbN5U3Q8PO0XHDVXPvvuGC1ebLjUDGgihPwN5qFjkZThBLpqiqcM5xxfKlnBbEOcAhs3dTHPgpcHWB8z4H/DNgl1sFeEWkAqgG4pys8ye2X2dYEVd9Fda9cub4htfAZV+A4ABc+kl4y/dnxAGgYQ2c+Q7Y+SPduuMkYFObj0gixbN9U3hcTiqds//XunRTy6yV5U6H4KpwZFkQlkukuUgLkQ2tvkwfKuP87DGm9Z7KzKfnXJRS/GpnH+etach0nT3DrAPZ1Vu4xsLiiJmtc/9zw7O28ogn0/zH/x3k7j2DizYtbyGU24Lon4pk2tqf7G6mcgpEJ9Bre3zMPJZBRF4CdCmlbs+59hfANDAAHAX+TSlV+tDg5YS/w2jFceaf5j937nvhxj64+G+MgUS5XPARQMEjXy37MjXHx6Y2I67wRM/4rBlM8yG35feouaE1FRi7aqzBx6GRUGZ06lgoZk6pM9ZTV11Z1IJ45tgUh0eneaMt+N1YU8XKBg+7ihThgbHp909GWNPsZXw6zs6jE0XP3dEzzhfvOcB7fvAkL/ncPXzwx0/x9BziUw4sCyIUSy7KUCY76bRiKBBlQ6thGQ6XSYROFEsWpBYRB/Al4GMFnj4HSGGk0q4GPiYied1jReQ91oyKkZGTw/+Zh2uWDqf1q+D0a+HJ70Fo+MStSTNvNrQaDQmnZmnUN1+qK52E4zP+/4wFUVPYgtjY5ieeTGfy78dCRgtxqydVncdVNAZx684+XBWOvALD7V11s1oQ/ZMR0gr+9NxVVDiEe/YNFT3Xyu75wptO4w1ndvLIoVH+/LtPMDF9YmszAtH8n+lCeKJnPJPSmrnfdIxESrG9y7DChgvUppxMlFMg+oAu2+MV5jELH7ANeMBsAHgecJsZqH478DulVEIpNQz8ATg79wWUUt9SSp2tlDq7ubm5TG9jibnwryEZNQLZFsmYnkOxzKh2OVlltjMvVkW9kHtGEjMxiBHTZWSfG2HHSre14hBWHyaLOk8lwWgyY2FYJFJpfvN0P6/a3JInbtu76hgMRDOB11yOjBtitK2zlnPXNPB/+4p/kBk0N8urt3fyT288jR+/+zymIgk+99u9Ra8pB3Y32/F8wn/393fwpbuzJ0ta8QfLPVcOCyKaSPHhn+zkSJnSdO2UUyCeANaLyGoRcQF/AtxmPamUmlJKNSmlupVS3cBjwFVKqR0YbqVXAIiIF0M8Ts2cz6Z1sO1N8MS34dGvG6my/9wNN58H0XmEZabH4KEvQd+TZVvqqY7lZlosF5O7Mnsu9VgoRr2nkooi8Y11LTU4BJ4bNP4uxqbjWe4oq5o6Nw7x8KFRxqbjvGF7lgcYgO0rrThEYdfRUXOTWtng4VWbWzk0HKJntPDGNTAVoba6MpOBtbndz/svWcsvd/bxwHMnzkKesllRC7UgIvEUk+EEB4ay60QsK2lLh58KhxSsbj9e9g4Yo24fOlj++TNlEwilVBL4EHAXsA/4uVJqj4h8VkSumuPym4EaEdmDITTfUUo9U661Lnsu+phhMdx1I4weMAQjcAzu/fTc106Pwj2fhq+cBv/39/DDN8NET9mXfCpiVVQvnovJkR2DCMVoKuJeAkNQupu8mc6yY2ajPotiHV1/tbOPOk8llxQImm9p91PpFHYWcTMdHQ9TVeGgxVfFq8zajHuLuJkGp2K017qzjn3oFetY2+zlk7fuzqTT7h8M8Lnf7p01nnE8BKIJfGYa8kID1ZawHBwOZXXStVxOnXXVNPuq5rQg7ts/xFPzfJ9HTRfi+AlwzZW1GlopdQdwR86xm4qce4nt+xBGqqsGoHUr3HAvVNdD41rjmLsOHv0abLsGui/Mv0Yp2HEL3P0po1hv2zVw+lvhl++Gn7wd/uLu7KwpzXGzud0QiLlSXEvF46rIyWKKzyoQYLiZ9pqdYo1GfXYXU35H1+lYkrv3DPHGl3QWrMdwVzrZ0u4vGqg+MhZmZYMHh0PoavCwsdXHvfuGuOGi/IGTg4EIbTkCUVXh5AvXnM5bvvEoH/+fpwlGkzx8yPhkHIgkOHNl/azvdyFMRRKsafLy9LGpBVsQ1sYfjqfon4pmphcOTEVxVTho8LpoKUEgbvzls2xq8/O9Pz8n77loIoXL6ciba3J0/MQJhK6kPllYcfaMOICRGlvfDbf9JSRyhvvFw3Dr++D2v4aV58EH/whv/m/Y8Fp483dgZB/c+l7daXaR2ZhxMS3O565cF9NoKEbTHFPyNrT6ODIeZnw6znQ8lRWvsPox2QPVd+8dJJJIZWUv5bK9q45n+6YKprAeHQ+zqnEm0eKVm1t4omciy41jMTgVzbMgAF7a3cA7zlvFnbsHOTgc5G8u28gZXXVzjlNdKFORBA1eFw1e14ItCPt1B21upgHzPYoILX73rEHq0VCMoUCsoBtKKcWl//YAt/whv1BWC4RmblweuOo/jIK6+//J2OxDw3DkEfj2q+CZnxkict0voNnWymHdK+E1/wD7fwu//+elW/+LkFUNHt5y1gou2bQ4k/OqXflprsVSXC02tflQCh47PAZQMAZhn019//4RWnxVnDXLJ/XtK+sIx1N5/nalFEfHw1mzxl+1pZVUWvHAgeyYQiyZYjQUp81feJDT/3flZn74F+fy0N+8gg9cso5tHX6eH5kuOgjpeAhEjUyz5pqqBVsQIyG7QMwI2cBkJCOCc1kQlqVXSKQmwwkGpqLsLGC5WQIxW9HjYqEb7p3MrH45vOTP4JH/gMe/ASnzD6a6Af70f7ML8+yc9wGjq+zvvwDtZ8CmK07cml/EOBzCv77ljEW7X3XlTKFcJJ5iOp6a08VkWTF/MN009iFEMzMhZjaWXb2TnLWqftbxrNu76jPnbm73Z46PhuKE46lM9hYY2Tsup4O9AwGutgW9rYKxQhYEGK6mC9fPVIivaa5hKpJgfDqe5SZbDKbChkA0+Y7PghAxRPfgcLYFcc5qo/Fiq9/N+HSceDJd0H23d2AmmSD3HCvjq5AV1Ts+k8ZcbrRAnOy85nNQ4Ta6x9auMArzus4Fb347hgwi8LovGSJx63vhPQ9ku680ywKjDsIQiLlqICxWNnhwVzp49HnDgmi0WRB+dwVOh2RcTGOhGEfHw7z93JWz3rO70UOdp5JdRyd52zkz51otNlY1ejPHnA6hq6E6L5PJyu7JjUEUY22zcc/nR6YXVSDSaUUwlsxYEMUm8c3FSDBGo9fF+hYfB4eNTTxlFsnZLQgwrA0rRmFnj22q4GgoRoftHMvtdHh0mlRaZSqzo4lURjy0i0kzN+5auOJf4NV/D+e825hDMZs4WFS64dofgKMCfvanuq5iGeJ2OTMWhOXSaPLN7mJyOoQNrT4Omxu03eIQEWqrKzNDg545Zgwt2t41+2hVEeGMFfkFc1aLDbuLCWB1k5ee0exhOVZ2TzELIhdreNJixyGCsSRKGYkETTVVx2VBNNVUsb61hkNDRibTaChGMq1oNzf6Vr/xXouluu7tn8q0dc91RVkWVzyZpm9iJsZ4bCKCUob4jIfjZXHB2dECcSpTt9IIXo/sN4LdZf5j08yP6kon8WSaVFrNtO4u4dP0xtaZBoCNOTELo92GYUHs7J3EIXCa2XtpNrZ31XFgOJjV0vvoeBgR6GrI/nTc3eilZ2yatC2oPThPC6KzrpqqCgeHF1kgMgOdqitp9lURSaTm7FZbiJFQjGZfFetbagjGkgwFYhkrqd0UBqtnVqF+TOF4ksOj01ywzmiCmCsig7bHdpG03Evbu+qIJ9NML3KrkFy0QJzqrH0FvOJTsPt/ja6zWiSWDVbL72gilXExlSQQZj1GdaUTjyvbi1znqcxkGD3dO8mGVh/eqrk9zS/tbkCpmdgGGPn47X53XrfS1c1eYsk0A7ZNbjAQxety4iuxiNDhEFY3eXl+ZHEtW/u8DutnuRArYjRoCITVjffgcJCBSdNKqjMEwrIgCg0O2j8YRCm42Kw9ybUghgJRqsyYhF0grAC1VcBY7jYlWiA0cOFHjc6xD/2bUUynRWJZkJkql0hlGvXlWgSFsCq6C7XkqDNbfiulePrYZKYlxFyct6aBphoXv9410y3nSE4Gk8VqMyZhj0MMTkVLth4s1rbULLqLyT4SttkWI5gPSilGTIFY32q4wg4Ohei3LIhaw6JqNAc1FbIgrPjDxeubcQiM5FgQQ4Eoq5u8NHpdeQJRXelkgylMY1ogNGVHBF7/VTjrXfDwl+GeTxkioRRMHoVjT2rRWALctrGjo6EYPndFSbMFNrQZm1ahlNg6c2jQkbEwk+FE5pPoXFQ4HVx5egf37hvOjFU9MpZdA2HR3WQIxAs2gTDqAwqnuBZjbZOX3vHwoo4zLWRBjM7TgghEksRTaZprqmiqMaYJHhwOMjgVoarCQb2ZLeZwCM01VQVjEHv7A9RWV9LVUE1jTRVDgVwLIkar383a5hqeH575OR4dNwoTrQ8K49Pl7RarBUJj4HDAlV+Gl77bSJv9+nnw+S6jRce3XwGH71/qFZ5yVGe5mOJzZjBZNJubVkELotrFVCSRCTiXakEAvOHMTuLJNL97dpBwPMloXcfe6gAAIABJREFUKJaVwWTR5ndTVeFYFAsirWaC4fNlV+9kpi2FRUYgPAu3IEZCxoZvXb+upSZjQXTUVSMykzLc4i9cC7G3f4ot7X6joM5XleeGGgxEafO7WdvizbYgxgyrzfrdjk+XNmN8oWiB0MwgAlf8q1FgV9NiTMN73ZfA5YPdv1zq1Z1yWC6mcDzFyBx9mOyICJ+6cnPBdhf1nkpCsSRP9IwbrorW0tutnLGilu5GD7/a1ZfxhRdyMTkckglUAyRTaYaDhauoZyOTyTRc3M0UTaSyWqJbKKW44XtP8M93Zff4nMq4mCqMVugyfwvC2vAtgVjfUsPB4RADkxHa/NnvscXnzrMgkqk0+weDbDVnk7f63VkikkylGQ3FaPVXsba5hrHpOBPT8Uxh4sosgdAWhOZEImIMMPqz3xhi8dK/gI2XGZXXqZxPK3tvg6++xGgIqFl0smIQodicKa523njmCi4oOJrUcH/8/sAIp62oLdoZthAiwtXbO3n08Bh/fMGY37WqgEAAdDd5Mi6mkVCMtCo9g8lidZNVC1FcIG785bNcf8sTecd7xyOMhuJZKaJgVFE7HUJNlVET0uCtmr8FYW7mLTaBmIok2D8YzASoLVr8+am0h0eniSXTbDEFIrfiejQURylorXWzxqwHOTwaYjQUJ5JIsbKhmpqqCiqdoi0IzTJgyxsgMgE9D80cUwoe/FcYfx4e+uLSre1FjNuVHaQu1YKYDath37GJyJz1D4V4w5mdKAXf/P1hgIIxCDDiEL3jEVJpNZP+OU+B8FZV0F7r5vAsmUxPHplgx5FxgtHsjXKn2Z48d6DPVCSB312RcQM11bgYCc4v0Gtt+M01xvtZb6YVh+OpvPfY4qtibDpOwjaDw2qxsbWjNnPOaCiWmdNhpbi2+tw2K2o6Y7WtavQiIjR4XdqC0CwD1r0SKr2w99czx449AYPPgK/DmFUxeXT2eyRP7tGLS4FlQQQiCQLR5CIJxEya6UIEYnWTlzO66uibjOB3V2QEJ++8Ri/xlDGONFMDUaQP02ysbS6eyRSJp+idCJNW5PUssh4PB2NZm/NUJJnVbbfZtwALIhTD5XRkBkOtb5lx0+UG4q1UV7sVsXcggKvCkbEOmv1ulJrJSLJcUq1+NyvqPbicDp4fCWVqICy3XoO3quzV1FogNHNTWW10gt33W0iZ/t4/fguq/PDOX4M4jIaBxdh/B3x+BfzkbTByoPh5miysOohjpptkMQSi3rahn7EAgQB4w/YOAFYWsR4gO5NpcIEWBMCaZm/Rpn2HhkOZ5LodPdkj660gvFLZNQaBnJGwzTVV845BWCmulhXS7KvKdPDtyHUxWcVyttfY0z/FpjYflaZ7rzWnoC4jELVVODP1IKGMBbGi3hChBm+lFgjNMmHL1RAehaOPQHAI9vwKtl9ndIo95z3w9E+N3k65HHkUfvEuqFsFLzxkZEfd/nEIDp7493CSYVkQxyaMjWGuTq6lYG2Ozb4qOhawYQNceXoHToewqiE/g8nCih/0jE0zaBZ92a2XUlnbXEMoliyYCWQNRqr3VPJHm0DEkin29gcyFeWDNjdT7sxwy4KYT8uKkWB223URybiZcq2k3HYbSin29gfYYmt62JJzzlAgitMhmUaLRibTNEfGwrT53Zn0Z21BaJYP618DlR5DGJ76HqQT8NIbjOcu/KhhTfzfZ7OvGdoDP7kWarvgz++CD++Es99lDDL64kb4xkVw72fg8AMw1afnU+RgxSB6x00LYo5ZEKVQb2a/nLGiLisdcz40+6r4/JtO44aLVhc9p8VXhcfl5IXRaQbMFNeFvN5sPZkODgVxOR28/owOdvVOEk8afz97+wPEU2kuP60NgP7JmSyiQCSR5WJqqqkinkwTnEe7jZFgLC/l2MoGm8uC6J+KMhFOZDKYCp0zOBWjxVeVadC3trmGo+NhDo2EWGlLCmjwlN+C0N1cNaXh8sD6V8O+34DDCeteZczLBvA0wIUfMQTizv8H/k6jieADnzdE5R2/BK/Rc4bXfRHOfT/svRUO3WfUXDz8ZeM5p8sQk/bTYfXFRjvzhjVGZtUpiGVB9JoWRKl1ELPhdTnZ2OrjNVtaj+s+bz27a9bnRYRVjV56RqcJxZJ56Z+lssbW1fVla7Ozsg4MBVnT7OW8NY18/9Ej7Omf4syV9Rn30hWntfOVew9mXFxgZDHZZ4ZnaiGCsYKzxH/7TD9OES4/rT1zbDQUy5t0d9UZnaTSKm/cbGNNFQ6B4UCUaCLFJ/7naRwC56xuzFuDVQsxHIxmrAowBCKVVjx7bJI3vWRF5niDt4pANEkilc64qxYbLRCa0tly9Uyg+vX/nv3cue+H535nBKzT5qcxdy28606jKaCdpnXw8k8YX9GAEfCe6DEC3RMvwNHHYM+txrnt2+Fdd4CruDvjxUql00GFQzKpmqW02ZgLEeGuj778uO9TCqubPOwbCJJIpTl71cJGh7b53XhczoK1EAeGQpy1qj5z7x09ExmBaPUbjfS8Lif9potJKZXnYrL3Y7KsFYtkKs1Nv96D312REYhkKs3YdDyzqVucv7aR89c2kovTITTVVNE3GeFDP36KR54f44tvOSPTLwuM33Oj15Wpph6cimaEEWasqLQi24Iw/x4mpuNZgrKYaIHQlM761xqzJ3xthgVhx+WBG+4xooKxoJEWW11niMRsuP35g42UgrFD8NwdcM9N8NjXDTE5BamudBKMJfG48hvvLXe6G73cvWcIEWibZ5sNC4dDWNPszbQvtwjFkvRNRnjbOV20+N2savTwRM847375Gnb1TrK9y3ChtddVZyyISCJFIqXyYhBAwclyf+wZZ3w6zvh0PBOYHp82ahRyBWI2Wv1ubt3Zh1Lwuau3cs1ZK/LOafZVMRKciUHYxWa1TSyyXUxmsVy4fAKhYxCa0qmqMSyH13/VcDMVQsTY9OtXzS0OxRCBpvVwwUdg05Xw8L+fssV4VhxiMTKYTjTdTd7/v73zjq+qPB/498m6CZnMBBIgoGzCMiBDhkrdiooWEbSi1bqxQ6VqW7XWOrps5YdardWCgyLWgYALB4qWYRBwoFIiCXuFmf3+/njuzbxZkJvLTZ7v55PPvec97z3nvYfDee6zKS51FJW4I4pg8tG9XVw1DcLXB7qn1zk8NL0NK7L3sOtAAdm7DpWZgDomRpcV0dt3WDVbX3gqlDv+/VV0XbS2PJBiZbY6wcuyqBvw75Gc4ME5mHFmby4bkV7DHM2mPlxYwr784jLnNkCcJ6LMRFcxc70smzqAneVMQBgNY+Al0H1s053v1N9A0SF4/6GmO+cxRKsyAXH05qWmpnu78l++Dc2irkj/1ARy9x4ui+aC8j7Q5QKiNbsPFvLyZ1pt1pfj0TExuiyKqWKhPh+tW2nF1aoaRGmpY9HarZzauwOeiDCWb9TEO1/OREM0iKtHd+eBCzO4dmzNXRs7xHvYvq+gUg5ERY7roNeyiz8BEcDe1CYgjGOb9j1hyOWw4inY9V2wV9Pk+BzVoapB+DgaDeIHfTUaafG6bWVj67ftxxMRVvaLOjNd+0A//dFGwgQGpKn2mpIYU5Ys509AhIUJbWOr96b+bNMetu8v4NyBnRjUOaksz6JqmY36cGL3tlwyrPa2rh0SNNzW5y9JTqh8/L4dE2jdKrLSD4XyekwmIIyWzLhfQrgH3v1tsFfS5Phi3hsjxLWpaRsbRby3GdHRaBDd2sXSOyWexRVMPuu3H+D4DnFloaDd28XSJjaK3L2H6ZWSUOav6ZQYXZYsV7EXREXax3vYWcVMs3DNViLDhVP6dCAzvTVrN+/jUGFxmYBobIGdnBBNSanjqy1qOqsa9TV9fE9evfGkSqHCvrLiu8zEZLRo4pNh5I0a2fThHzUbu4X0pwhlDUJESG8XS0SY0C726NZ/er8UlmfvLntAf7Ntf5l5yXcuXzRTxRIiPsG0Ze9hvxoEUK03tXOOReu2ctLx7UiIjiQzvQ0lpY6s7/eyY38B8Z6IslLsjYVPI1mTq33Cqzqd4zwR1SrnRoSHkRgTyR4zMRktnpE3QdowzbWYORQeGag5FFVxDl66Gpb8vunXGAB8D6L2IeiDAG1/2qVtK8LCji6X5Yz+KTgHb3+5jbzDRWzJy68kIEAd1QCDKzRB6pSk0VNb8vJrFBDJCR7+t/MgX23VInrrNu8jZ89hzuyvoa1DurRGBFZk7ynrRd3YtI9XgfB5zl5iIsPLSnfURdvYqIB2lQuogBCRM0TkaxH5VkRm1DJvoog4EcmsMDZARJaJyDoRWSMigYnjMkIDT7yG0d6yRntUJHaGN++Cb96qPC/rOVgzF5bNhMI6Gs0sfwpevTlwa24EQlmDALjzrD48M23YUR+nd0o8Xdq0YtHarXy73RfBVDlv4bR+yQzsnMTYnu3Lxso0iLzDZZ3w4qs8fK8e3Z1YTzgXzVrG++t3sHDtFsLDhPHeZMLEmEh6JcezfOPuamU2GgufBrFh50GSEzz1zjpvHRsV0L7UARMQIhIOzATOBPoCk0Wkr5958cB04NMKYxHAbOBa51w/YBwQ2MLnRmiQ1EV7VFw2H9r30Qf8YW8lzwM74M07NZO7cL/mUdTEJ4/Bgp9p2ZC9m5pm7UdAKPsgQB9g/poKNRQR4Yz+KXz83U5WZmtEUVUNomvbWF65YVSlCKCE6EjiPBFlGkScJ6JaD4weyfH854ZRdG7Tiiv/uZzZn3zPid3aVOrIl5nemlXZe9ial98gB3V96eB1SjtXPYKpNrTkdwgKCGAY8K1zboNzrhB4AZjgZ95vgQeBim2XTgM+d86tBnDO7XLONV5jWiP0ifDA+f8HB7bB4jt1bNEMKDgAU19SDWP18/4/u/wpWHQ7pHoV1uyPmmbNR0CrEM6DaGxO75dCUYnjH0s3EhMZTmpS/ZLvOiZGs2VvfrUs6spzYvj3tSMY3aMdeYeLKpXWADVfHSws4fvdhwJiYvJEhJcVM2yIgAhlE1MqUPGnWY53rAwRGQJ0ds4tqPLZnoATkcUiskpEbvN3AhG5RkRWiMiKHTt2NObajVAgdYgm02XNhkV3wNp5MPrn0KEPDJgE371bvWrsqn+p5tDzTLhigSbzbVwanPXXg5gQzoNobAZ3TqJDvIet+/LpkRxXb79GSmI0W/bls69KL4iqxHkiePLyTJ6+YiiTh1auNeULo4WG5UA0BJ9m0pCIL5+JqSHVaBtC0JzUIhIG/An4uZ/dEcBJwBTv6wUicmrVSc65J5xzmc65zPbt21fdbbQExs1QU9MnM6FdLxj9Mx0feAm4Uljz7/K5338Cr90Mx50KP3wGIqOhy0jI/jg4a68H/VMTGdIliThPaJXZCARhYcLp/TQnoqp5qTY6JcawZe9hreRah/M3IjyMk3t3qGaGSk2KKcvlaIyiif7waQ4NMWG1jY2iuNSxL7/+1WgbQiAFRC5QUQynecd8xAP9gfdEZCMwHHjV66jOAT5wzu10zh0C3gCGBHCtRqgS4YELZkGHfjDhUd0GLdWRmql9KgDy8zS6KbEzXPzP8nnpo7Rt6jHan+K8gZ2Yf/2oIy7N3dwoFxBxdcwsJyUxmh0HCth5sKBGE1N98GkRgdIgfMdtiInJ1wAqUI7qQAqI5UAPEekmIlHAJcCrvp3OuTznXDvnXLpzLh34BDjPObcCWAxkiEgrr8N6LPBFANdqhDKdBsP1H0PnKtEyAy+BbWthy+fw+k9hXy5MfEprRfnoOkpf6zIzFfhve2k0LSOOa8uMM3tzweDqBe9qolOSJstl7zp0VAJiaLrmWXSID0xApe+4DTEx+Sq6BsoPETAB4ZwrBm5EH/ZfAnOdc+tE5F4ROa+Oz+5BzU/LgSxglR8/hWHUTv+JEBYJL/8E1r4EJ98BnYdWnpMyAKLiazYzFRfC6z+DB7rA95/6n2M0GeFhwrVjj2vQr3hfJdmSUlerD6IuLj6hMw9dNIA+Hetv3moIPhNWQ3pntA1wuY2AGjadc2+g5qGKY7+uYe64Ktuz0VBXwzgyWrXRXtpfvQ7po7XzXVXCI6DLif4jmQ7sgLmXa5vVcA989Bfo4icyqrQENmepU3zjhzoW01r/0k+CjIsa93sZDaJia9Wj0SBiosLrbJR0NJw/OJX46OoZ07URaBOTeb6M5s3w6yEvBy54vOYS5V1HwTv3aEnxWG/Xsi2fw/OTtQ/3xKdg53p4/0Et89G+Z/lnv14E/7lW+18gkNJfu+ht/0L9GlnPQa+ztF+GERRSGklABJrEmMhKHePqQ9sAm5hMQBjNm/RR8JP365hzkr5mf6Rd8/ZvhdkTtQXqlYuh0yDVJj56BJY9Cuf9Vecf3AmvXA/xHeGsP0D3ceUCBuDbd2D2herf6Hla5XNueF8771VtlmQ0OvHRkcR7IthfUFypF0RzICYyHE9EWMDqMVktJsPoOAgiYtQPUVIM866CwgMwdZ4KB4C49jBwskZFHdiuYwtv15apE59SM1Js5Z7JdB2l2sS3VcqBlBTDS1ep8PjkscB/P6NMiziWNYgjQUTLlQeqoqsJCMOIiNIIqI0fwXv3Q/ZSOOfPmnBXkRE3Qkkh/PcJ+HqhJuaNuRWSq1WQUSKjodsY+ObNytVnN34IB3dAu56a0f3u71pMddpg0dGbdd3cBARoJNPug9U74jUGJiAMA9TMtG2NlhMfcrmGyFal3fHQ+2xY/qSGzXbo59/xXZHjx8OejZWbHa2bD1FxcPUSGHwZfPAQLPg5lJY26lcyyumY0Dw1CFBH9e5DgSlV17wMcoZxpPjyIZIz4Mxa2puOvFmjoiQcJj+v2kdt9PiBvn77lgqY4kL44lUVNJ44OO9vGm310SPax3vU9Mb5PkYlOiapgKjaLKg58Ktz+nKU1dRrxASEYQB0PlHNRYOnQmQtReC6nAgnXKHlPToNrvu4rdOhbQ8tSz78OtiwBPL3ao4GgAiMv0e1jLfvgc7D9Rz14bXpkJervhKjVs7O6Eje4aJmWfSwIWVHGooJCMMAzYc45a76zT33kYYdu8dpapYqPKQJe9FJ0P3k8v0iqklsWQ3zroRrP1StojZ2fQerntX3+fsqZ4cb1eiRHM9vzu0X7GWEHOaDMIxA02M8lBSos/qrBdD3vOqmqehEuOhpLV/+n+vqdlp//DctRuhKYdN/A7d2o0VjAsIwAo0v3PWtX2v4rM+8VJXUIXD672D9InWCF+z3P2//VsiaAwMugbCIY7qfhRHamIAwjEAT4YFuY2FvNsR20LIfNTHsGhh+A6z8Jzw6FNbOr65NLJupSXbjblc/yDFcrtwIbcwHYRhNQY/xsH4h9Du/5pIfoP6IM+6H/heqFjFvmvoazngAOvTWkh4r/gH9LoQ23aHrSFj2f1B0uHbnejOmqKiInJwc8vPz657cgomOjiYtLY3IyPpHcpmAMIymoPe58NlsOGFa/eanZWqexPInYcn9MGuk9uKO8KiZypd/0XWUhsjmrIButWgmzZicnBzi4+NJT0+3vhk14Jxj165d5OTk0K1bt3p/zgSEYTQF8clwzXsN+0x4BAy/Vst4LPmdCgtXCj1O16KAoOG5iJqZKgqIt36t5qyRNzbSF0BzOOrK+wgC+fn5JhzqQERo27YtDW3NbD4IwzjWiW2npT9+8qHmafzg3vJ9MUkqLCo6qnNWqlbx5p0aVtsYfDYHHugMn/+77rlBwIRD3RzJNTINwjBChZT+MGFm9fGuo2DlM+W/8N9/UHtRtO0Br9yoNZ9SMnTutnXw5l1QUgRtuqkfo+MgdaKH1fB7cdsXWgrEOXj5GvWh9L8wcN/TOGYwDcIwQp2uI6H4sCbabf4MvlmshQUnzdb8ihemaAXa9x+Gx8dqr4viAvjqDXj7bvjX+fBoplaWzd9X+diFB+HfV4AnHm74VDO9X/qxlgsxyoiLq3+P7FDCNAjDCHW6jNTX7I9g06eaqT3sGs2unjQbnj4THhkIRYc0+umsP0BsW/1Mfp6WAfn0cW9l2d/CgEmQeaVqLG/cqs2SLv+PahxT5mqvjHnT9DhDLq89KssIaUxAGEaoE9dezUirnoXd38HJd5aX3kjLhHP/Ch88DON/ow2RKhKdqE7wjIsgdyX89+8abbXiKejQVzvjjblNmyGBahJT5sHzl8Drt8Ans+CUO6HPeRqi6w/nat7nj9xV8OrNMOhSGHF9gy7FPa+t44vN++qe2AD6dkqod5kO5xy33XYbCxcuRES46667mDRpElu2bGHSpEns27eP4uJiZs2axciRI7nqqqtYsWIFIsKVV17JT39aR3XgJsYEhGE0B7qO1OQ6TyKc+JPK+wZN1r+6SD0BLjgBTr8fVj+vxzt+PIy9vfK86AS4YgF8+Rq8e5/27U4bptpFTOvyeaWl6rPY8RVMW6jCpTac00itxXeoj+Ttu6HXGeonCRHmz59PVlYWq1evZufOnQwdOpQxY8bw3HPPcfrpp3PnnXdSUlLCoUOHyMrKIjc3l7Vr1wKwd+/eIK++OiYgDKM50HWUPtCHX6dawdHQqg2MuEH/akJEa0r1Plv7br/+U3jxMpg6vzwU9v0HYI036umNW+GCWrrn7d0Eb/0K1r2sYbzjfwNPnQ5v3AZT/l1vDSTYBfmWLl3K5MmTCQ8PJzk5mbFjx7J8+XKGDh3KlVdeSVFREeeffz6DBg2ie/fubNiwgZtuuomzzz6b0047re4TNDHmpDaM5kDvc+CUX8HIm5r2vGHhMOQyja7a+CG8drNqAute1miqQVNUA1n9PGQ9X/mzh/eqUHv6LPhLf3V8j78HJr8Ayf3g5F9qH42vXm/a7xQAxowZwwcffEBqaipXXHEFzz77LK1bt2b16tWMGzeOxx57jB//+MfBXmY1TIMwjOZAVCsY84vgnX/gJO1p8d79KjTWvKRmp3P+rM2VNi7VUNm0oeoz+WSW1pQq2KfhuCffBQMu1v4ZPob9RPMvFs6A406BqNja11BSDPs3a52qUm+l29i20KptIL95JUaPHs3jjz/Oj370I3bv3s0HH3zAww8/THZ2NmlpaVx99dUUFBSwatUqzjrrLKKiopg4cSK9evVi6tSpTbbO+mICwjCMxmHsbSokPpsNCakaQRXhbdBz4d/hsVEw5yJtmHR4D/Q5V0uGdBri34QUHgFn/xGePsPrZL+79vPn74VDuyAiWoVSSaFWvo1p0zAn+VFwwQUXsGzZMgYOHIiI8NBDD5GSksIzzzzDww8/TGRkJHFxcTz77LPk5uYybdo0Sr2tZn//+983yRobgrhm0iw9MzPTrVixItjLMIyWTXEhLP2zPvyT+1be9/UieOFSOP5UOPmO+nXkA3j5Olg7D25ZA/Ep1XZ/+eWX9OnTR4VTwX5I7q8C4eBOyNsE7XqphmWUX6sKiMhK51ymv/kB9UGIyBki8rWIfCsiM2qZN1FEnIhkVhnvIiIHRCSIurNhGPUmIkrLkFcVDqARSb/cpE7n+goHUNNZSRH894ma5zgHBQcgKq5cW4hO0tf8Yy86KFQImIAQkXBgJnAm0BeYLCLV7hoRiQemA5/6OcyfgIWBWqNhGE1MXX4Ef7Q9DvqcA8uf0sxuf5QUQmkReCpkNIdHQFS8OsOrWkry81To1ERxIezbAlvXqmbSQgmkBjEM+NY5t8E5Vwi8AEzwM++3wINApWLuInI+8D9gXQDXaBhGKDDiJtUEPpvjf3/hAX2NqlLyIiZJ270WHy4fK9gPuzdoe9eqlJbqvu3r4MBWwKkwcaWN8jVCjUAKiFRgU4XtHO9YGSIyBOjsnFtQZTwOuB24J4DrMwwjVOhyokZAfTITSkuq7y84oI7piOjK476ckMNeM5NzkJer7/Pz/GgWe3U8tr1mkid2VuFQeKhxv0+IELQ8CBEJQ01IP/ez+27gz865A3Uc4xoRWSEiKxpa59wwjBBj5E1q7vlqQfV9hQfUvFQ1Wik8UrUKn5np0G7VJqLi1SxVXFB5fn6e9vlOSNUILJ9GUlhDf/D6UFLsX6iFAIEMc80FOlfYTvOO+YgH+gPveeuUpwCvish5wInARSLyEJAElIpIvnPu0YoncM49ATwBGsUUqC9iGMYxQO9zNE9i2aOaxe2jtFgfwrHt/X8uJgnycrRY4f7NENkKkrqoGakgDyK9WkdpqeZlxLQuFzThEdrKteCAPrFqwjkVPMUF3r/88veuRIVN+z61h9uWFJXPLy1RzSU6UYVckAikgFgO9BCRbqhguAS41LfTOZcHtPNti8h7wC+ccyuA0RXG7wYOVBUOhmG0MMLCYfj1sPA27aDX1VvFtrgACK/uf/AR7RUQu/+nwqRNd422iojR8uZxyTqvcL/3oZxU+fNR8XBwhz60/VWuLS2G3RsraxlhkSoUfLWpDu1U7SSmyrEP71GtpuiwOtmrUrBfq+gGiYCZmJxzxcCNwGLgS2Cuc26diNzr1RIMwzAaxuCpEN8JXrsFirxxLcUF6n+IjPH/GZ+ZqbRIH9i+SKroBDVNlRbrdn6eHsdTRdB44gHnP4KqKB92rNfjJKRC+96QMkBLpbfrAUmdITENwj3qFHeuvHdEUT7syVZtwxMPCals3B9B/x9MUW2jVTtvtFXhUV+2IyWgmdTOuTeAN6qM/bqGueNqGL+70RdmGEZoEhULE/6mPSmW3Aen3acCIipFzTcLZ8DWNdU/V+o130S2AvH+LnYlanaKiFa/Q+FBfa3q6E7pBwMvVSHgK6MO3mio/+l52x5fXbD4ENHyInk5lYXMvlxdS7ue5WYkzyEdi4zWzxzaCQd3QULHmq9JQ8upNwAr1mcYRmhx/Hg4YRp8/Ch88Ur1/Ad/hEWqcJEKjzwJA6Tc3o+roflRmAqWggompOICDYcNj2TGn55l5pPPlO26++67ue+++zj11FMZMmQIGRkZvPLmUhU+vtDa/Dz1d8Qn1+hjyC+Gab/4HRknjmHw4MFRb/STAAAJA0lEQVQsWbIEgHXr1jFs2DAGDRrEgAED+CbrIw5uXs/ZZ5/NwIED6d+/Py+++GLd17EeWC0mwzBCj9N+C9+9o+1Px88u9z+c+UDDjrNno/ohWrXR0hwpGf6FxL4tmhdRWqxmqLwcHW9zHJMmX8ott9zCDTdoefS5c+eyePFibr75ZhISEti5cyfDhw/nvFUfID4BkZerZqeaHOvAzJkzkcgY1rwzl6+2FXDahB+yfv16HnvsMaZPn86UKVMozNtOyc7veOPtd+nUsSMLFmiEV15eXsOuQw2YBmEYRujhiYfzZ6l9XsJq9j/UeZwENTUd3KnHrKl9qk9DKTiouRIF+yC+I0REMXjwYLZv387mzZtZvXo1rVu3JiUlhTvuuIMBAwYwfvx4cnNz2ba/FBDVVkoKIDG1skZThaVLlzL18isg3EPvtDZ07dqV9evXM2LECO6//34evP8+std+QkxCGzJOHMdbb7/N7bffzocffkhi4lH2BPFiAsIwjNAk/STtH+GJr/VBWytlPgVXe6OlqFhAVDjk5ahAqvDr/+KLL2bevHm8+OKLTJo0iTlz5rBjxw5WrlxJVlYWycnJ5BcVl5ce98SrcKoLEYhtB0UHy7K5L730Ul79z3xi3CHOuuwm3s3aSM9evVm1ahUZGRncdddd3HvvvUd2PapgAsIwjNDlpFuOroNeWES5eaq240iYN+Fut5qZErtUcgxPmjSJF154gXnz5nHxxReTl5dHhw4diIyMZMmSJWRnZ+vEuA6AQEJanY7l0aNHM2fOHGjVhvUbNvF99kZ6dW7HhrXL6Z4k3Dzth0yYMIHP133J5s2badWqFVOnTuXWW29l1apVR35NKmA+CMMwWjbxKZqHUFdCmidOcx1i21crH96vXz/2799PamoqHTt2ZMqUKZx77rlkZGSQmZlJ7969dWKERwVDZLSfE1Tm+uuv57rrriNj4GAixPHPP/0Gz+FtzH3uX/zrpQVEemJI6ZTGHb+6m+XLl3PrrbcSFhZGZGQks2bNOtKrUQnrB2EYRkjjr8dBQCgphAPb1fdQk68iULhSrTArotqMhENYww1ADe0HYRqEYRhGfQiP0qS3YODLjWhiTEAYhmE0MWvWrOGyyy6rNObxePj0U39tcYKHCQjDMEIe5xzSRH2nG4OMjAyysrKa9JxH4k6wKCbDMEKa6Ohodu3adUQPwJaCc45du3YRHd0wM5VpEIZhhDRpaWnk5ORgPWFqJzo6mrS0hvlQTEAYhhHSREZG0q1b8EpiN2fMxGQYhmH4xQSEYRiG4RcTEIZhGIZfmk0mtYjsALKP4hDtgJ2NtJzmgF2Pytj1qI5dk8qE6vXo6pzzW3e82QiIo0VEVtSUbt4SsetRGbse1bFrUpnmeD3MxGQYhmH4xQSEYRiG4RcTEOU8EewFHGPY9aiMXY/q2DWpTLO7HuaDMAzDMPxiGoRhGIbhFxMQhmEYhl9avIAQkTNE5GsR+VZEZgR7PcFARDqLyBIR+UJE1onIdO94GxF5S0S+8b62DvZamxIRCReRz0Tkde92NxH51HuvvCgiUcFeY1MhIkkiMk9EvhKRL0VkhN0f8lPv/5e1IvK8iEQ3t3ukRQsIEQkHZgJnAn2BySLSN7irCgrFwM+dc32B4cAN3uswA3jHOdcDeMe73ZKYDnxZYftB4M/OueOBPcBVQVlVcHgEWOSc6w0MRK9Li70/RCQVuBnIdM71B8KBS2hm90iLFhDAMOBb59wG51wh8AIwIchranKcc1ucc6u87/ej//lT0WvxjHfaM8D5wVlh0yMiacDZwJPebQFOAeZ5p7SY6yEiicAY4CkA51yhc24vLfj+8BIBxIhIBNAK2EIzu0dauoBIBTZV2M7xjrVYRCQdGAx8CiQ757Z4d20FkoO0rGDwF+A2oNS73RbY65wr9m63pHulG7ADeNprcntSRGJpwfeHcy4X+APwPSoY8oCVNLN7pKULCKMCIhIHvATc4pzbV3Gf03joFhETLSLnANudcyuDvZZjhAhgCDDLOTcYOEgVc1JLuj8AvP6WCajw7ATEAmcEdVEBoKULiFygc4XtNO9Yi0NEIlHhMMc5N987vE1EOnr3dwS2B2t9Tcwo4DwR2YiaHU9BbfBJXnMCtKx7JQfIcc596t2ehwqMlnp/AIwH/uec2+GcKwLmo/dNs7pHWrqAWA708EYeRKFOpleDvKYmx2tffwr40jn3pwq7XgV+5H3/I+CVpl5bMHDO/dI5l+acS0fviXedc1OAJcBF3mkt6XpsBTaJSC/v0KnAF7TQ+8PL98BwEWnl/f/juybN6h5p8ZnUInIWam8OB/7hnPtdkJfU5IjIScCHwBrKbe53oH6IuUAXtJT6D51zu4OyyCAhIuOAXzjnzhGR7qhG0Qb4DJjqnCsI5vqaChEZhDrso4ANwDT0B2aLvT9E5B5gEhoF+BnwY9Tn0GzukRYvIAzDMAz/tHQTk2EYhlEDJiAMwzAMv5iAMAzDMPxiAsIwDMPwiwkIwzAMwy8mIAyjAYhIiYhkVfhrtAJ1IpIuImsb63iGcbRE1D3FMIwKHHbODQr2IgyjKTANwjAaARHZKCIPicgaEfmviBzvHU8XkXdF5HMReUdEunjHk0XkZRFZ7f0b6T1UuIj83dtn4E0RiQnalzJaPCYgDKNhxFQxMU2qsC/POZcBPIpm5wP8DXjGOTcAmAP81Tv+V+B959xAtK7ROu94D2Cmc64fsBeYGODvYxg1YpnUhtEAROSAcy7Oz/hG4BTn3AZv4cOtzrm2IrIT6OicK/KOb3HOtRORHUBaxTIM3lLrb3kb8CAitwORzrn7Av/NDKM6pkEYRuPhanjfECrW7SnB/IRGEDEBYRiNx6QKr8u87z9GK8ICTEGLIoK26LwOynpfJzbVIg2jvtivE8NoGDEiklVhe5Fzzhfq2lpEPke1gMnesZvQTmy3ol3ZpnnHpwNPiMhVqKZwHdqZzDCOGcwHYRiNgNcHkemc2xnstRhGY2EmJsMwDMMvpkEYhmEYfjENwjAMw/CLCQjDMAzDLyYgDMMwDL+YgDAMwzD8YgLCMAzD8Mv/A5hGeXSYjf48AAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 392
        },
        "id": "1i2nDc0WOwBU",
        "outputId": "dc7b0b54-4202-4cfb-cb7f-74b3b2f32777"
      },
      "source": [
        "prediction = siamese_network.predict(test_ds)\n",
        "\n",
        "classes=[0,1,2,3,4,5,6]\n",
        "con_mat  = tf.math.confusion_matrix(\n",
        "    test_labels, prediction, num_classes=None, weights=None, dtype=tf.dtypes.int32,\n",
        "    name=None\n",
        ").numpy()\n",
        "con_mat_norm = np.around(con_mat.astype('float') / con_mat.sum(axis=1)[:, np.newaxis], decimals=2)\n",
        "\n",
        "con_mat_df = pd.DataFrame(con_mat_norm,\n",
        "                     index = classes, \n",
        "                     columns = classes)\n",
        "\n",
        "\n",
        "import seaborn as sns\n",
        "figure = plt.figure(figsize=(8, 8))\n",
        "sns.heatmap(con_mat_df, annot=True,cmap=plt.cm.Blues)\n",
        "plt.tight_layout()\n",
        "plt.ylabel('True label')\n",
        "plt.xlabel('Predicted label')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "error",
          "ename": "ValueError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mValueError\u001b[0m                                Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-65-19cd557d6956>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mprediction\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0msiamese_network\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpredict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtest_ds\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0mclasses\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m3\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m4\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m5\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m6\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m con_mat  = tf.math.confusion_matrix(\n\u001b[1;32m      5\u001b[0m     \u001b[0mtest_labels\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mprediction\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnum_classes\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mweights\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mtf\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdtypes\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mint32\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/training.py\u001b[0m in \u001b[0;36m_method_wrapper\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m    128\u001b[0m       raise ValueError('{} is not supported in multi-worker mode.'.format(\n\u001b[1;32m    129\u001b[0m           method.__name__))\n\u001b[0;32m--> 130\u001b[0;31m     \u001b[0;32mreturn\u001b[0m \u001b[0mmethod\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    131\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    132\u001b[0m   return tf_decorator.make_decorator(\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/training.py\u001b[0m in \u001b[0;36mpredict\u001b[0;34m(self, x, batch_size, verbose, steps, callbacks, max_queue_size, workers, use_multiprocessing)\u001b[0m\n\u001b[1;32m   1577\u001b[0m           \u001b[0muse_multiprocessing\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0muse_multiprocessing\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1578\u001b[0m           \u001b[0mmodel\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1579\u001b[0;31m           steps_per_execution=self._steps_per_execution)\n\u001b[0m\u001b[1;32m   1580\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1581\u001b[0m       \u001b[0;31m# Container that configures and calls `tf.keras.Callback`s.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/data_adapter.py\u001b[0m in \u001b[0;36m__init__\u001b[0;34m(self, x, y, sample_weight, batch_size, steps_per_epoch, initial_epoch, epochs, shuffle, class_weight, max_queue_size, workers, use_multiprocessing, model, steps_per_execution)\u001b[0m\n\u001b[1;32m   1115\u001b[0m         \u001b[0muse_multiprocessing\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0muse_multiprocessing\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1116\u001b[0m         \u001b[0mdistribution_strategy\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mds_context\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget_strategy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1117\u001b[0;31m         model=model)\n\u001b[0m\u001b[1;32m   1118\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1119\u001b[0m     \u001b[0mstrategy\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mds_context\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget_strategy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/data_adapter.py\u001b[0m in \u001b[0;36m__init__\u001b[0;34m(self, x, y, sample_weights, steps, **kwargs)\u001b[0m\n\u001b[1;32m    698\u001b[0m     \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_user_steps\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0msteps\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    699\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 700\u001b[0;31m     \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_validate_args\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msample_weights\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msteps\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    701\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    702\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0mget_dataset\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/data_adapter.py\u001b[0m in \u001b[0;36m_validate_args\u001b[0;34m(self, y, sample_weights, steps)\u001b[0m\n\u001b[1;32m    745\u001b[0m       \u001b[0;32mif\u001b[0m \u001b[0msize\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0mcardinality\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mINFINITE\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0msteps\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    746\u001b[0m         raise ValueError(\n\u001b[0;32m--> 747\u001b[0;31m             \u001b[0;34m\"When providing an infinite dataset, you must specify \"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    748\u001b[0m             \u001b[0;34m\"the number of steps to run (if you did not intend to \"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    749\u001b[0m             \u001b[0;34m\"create an infinite dataset, make sure to not call \"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mValueError\u001b[0m: When providing an infinite dataset, you must specify the number of steps to run (if you did not intend to create an infinite dataset, make sure to not call `repeat()` on the dataset)."
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "FWHeC1kyENvP",
        "outputId": "4f275660-8f87-4c25-a600-ab9171d2a642"
      },
      "source": [
        "loss, accuracy, RMSE = siamese_network.evaluate(test_ds,steps= 32)\n",
        "print(\"Loss: {:1.2}\".format(loss))\n",
        "print(\"RMSE: {:1.2}\".format (RMSE))\n",
        "print(\"Accuracy: {:2.2%}\".format(accuracy))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "32/32 [==============================] - 6s 195ms/step - loss: 0.4804 - accuracy: 0.8125 - root_mean_squared_error: 0.3900\n",
            "Loss: 0.48\n",
            "RMSE: 0.39\n",
            "Accuracy: 81.25%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "y--vQot9Ng_D"
      },
      "source": [
        "from keras.models import load_model\n",
        "\n",
        "siamese_network.save('/content/drive/MyDrive/Capstone/progress2/project/siamese_network.h5')  # creates a HDF5 file 'my_model.h5'"
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}