A1-T1-cnn.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "id": "9PUXkk5s8Cdp"
   },
   "outputs": [],
   "source": [
    "import keras\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import tensorflow as tf\n",
    "fashion_mnist = keras.datasets.fashion_mnist\n",
    "cifar10=keras.datasets.cifar10\n",
    "(X_train_full, y_train_full), (X_test, y_test) = cifar10.load_data()\n",
    "#create the validation set\n",
    "X_valid, X_train = X_train_full[:6000] / 255.0, X_train_full[6000:] / 255.0\n",
    "y_valid, y_train = y_train_full[:6000], y_train_full[6000:]\n",
    "#add label for class\n",
    "class_names = [\"airplane\",\"automobile\",\"bird\",\"cat\",\"deer\",\"dog\",\"frog\",\"horse\",\"ship\",\"truck\"\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "cypxdEQxWe8k",
    "outputId": "95afc9e2-bd48-476d-d6f9-a638e18b479e"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/caroldong/anaconda3/lib/python3.11/site-packages/keras/src/layers/convolutional/base_conv.py:107: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.\n",
      "  super().__init__(activity_regularizer=activity_regularizer, **kwargs)\n",
      "2024-11-14 09:30:09.842561: I metal_plugin/src/device/metal_device.cc:1154] Metal device set to: Apple M1\n",
      "2024-11-14 09:30:09.842584: I metal_plugin/src/device/metal_device.cc:296] systemMemory: 8.00 GB\n",
      "2024-11-14 09:30:09.842589: I metal_plugin/src/device/metal_device.cc:313] maxCacheSize: 2.67 GB\n",
      "2024-11-14 09:30:09.842605: I tensorflow/core/common_runtime/pluggable_device/pluggable_device_factory.cc:305] Could not identify NUMA node of platform GPU ID 0, defaulting to 0. Your kernel may not have been built with NUMA support.\n",
      "2024-11-14 09:30:09.842615: I tensorflow/core/common_runtime/pluggable_device/pluggable_device_factory.cc:271] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 0 MB memory) -> physical PluggableDevice (device: 0, name: METAL, pci bus id: <undefined>)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/30\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-11-14 09:30:12.139397: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:117] Plugin optimizer for device_type GPU is enabled.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m42s\u001b[0m 30ms/step - accuracy: 0.2052 - loss: 2.0824 - val_accuracy: 0.4320 - val_loss: 1.5520\n",
      "Epoch 2/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m39s\u001b[0m 29ms/step - accuracy: 0.4619 - loss: 1.4773 - val_accuracy: 0.5380 - val_loss: 1.2815\n",
      "Epoch 3/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m39s\u001b[0m 29ms/step - accuracy: 0.5417 - loss: 1.2810 - val_accuracy: 0.5935 - val_loss: 1.1470\n",
      "Epoch 4/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.6060 - loss: 1.1171 - val_accuracy: 0.6338 - val_loss: 1.0670\n",
      "Epoch 5/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.6547 - loss: 0.9852 - val_accuracy: 0.6512 - val_loss: 1.0224\n",
      "Epoch 6/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.6939 - loss: 0.8782 - val_accuracy: 0.6563 - val_loss: 0.9927\n",
      "Epoch 7/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.7316 - loss: 0.7661 - val_accuracy: 0.6688 - val_loss: 0.9723\n",
      "Epoch 8/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.7619 - loss: 0.6804 - val_accuracy: 0.6698 - val_loss: 1.0191\n",
      "Epoch 9/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.7885 - loss: 0.6085 - val_accuracy: 0.6732 - val_loss: 1.0042\n",
      "Epoch 10/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.8129 - loss: 0.5378 - val_accuracy: 0.6685 - val_loss: 1.0302\n",
      "Epoch 11/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.8311 - loss: 0.4815 - val_accuracy: 0.6672 - val_loss: 1.1237\n",
      "Epoch 12/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.8528 - loss: 0.4291 - val_accuracy: 0.6765 - val_loss: 1.1186\n",
      "Epoch 13/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.8677 - loss: 0.3763 - val_accuracy: 0.6697 - val_loss: 1.1856\n",
      "Epoch 14/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.8802 - loss: 0.3472 - val_accuracy: 0.6642 - val_loss: 1.2246\n",
      "Epoch 15/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 29ms/step - accuracy: 0.8877 - loss: 0.3220 - val_accuracy: 0.6657 - val_loss: 1.2703\n",
      "Epoch 16/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m42s\u001b[0m 30ms/step - accuracy: 0.8992 - loss: 0.2946 - val_accuracy: 0.6792 - val_loss: 1.3384\n",
      "Epoch 17/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m43s\u001b[0m 31ms/step - accuracy: 0.9085 - loss: 0.2670 - val_accuracy: 0.6698 - val_loss: 1.3799\n",
      "Epoch 18/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m43s\u001b[0m 32ms/step - accuracy: 0.9166 - loss: 0.2435 - val_accuracy: 0.6675 - val_loss: 1.4871\n",
      "Epoch 19/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m45s\u001b[0m 32ms/step - accuracy: 0.9211 - loss: 0.2325 - val_accuracy: 0.6650 - val_loss: 1.4562\n",
      "Epoch 20/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m42s\u001b[0m 30ms/step - accuracy: 0.9207 - loss: 0.2324 - val_accuracy: 0.6657 - val_loss: 1.5309\n",
      "Epoch 21/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m43s\u001b[0m 31ms/step - accuracy: 0.9267 - loss: 0.2171 - val_accuracy: 0.6740 - val_loss: 1.5753\n",
      "Epoch 22/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.9349 - loss: 0.1958 - val_accuracy: 0.6683 - val_loss: 1.5319\n",
      "Epoch 23/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.9335 - loss: 0.1960 - val_accuracy: 0.6648 - val_loss: 1.6685\n",
      "Epoch 24/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m39s\u001b[0m 28ms/step - accuracy: 0.9354 - loss: 0.1912 - val_accuracy: 0.6673 - val_loss: 1.6665\n",
      "Epoch 25/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m39s\u001b[0m 29ms/step - accuracy: 0.9421 - loss: 0.1842 - val_accuracy: 0.6733 - val_loss: 1.5857\n",
      "Epoch 26/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m39s\u001b[0m 29ms/step - accuracy: 0.9405 - loss: 0.1752 - val_accuracy: 0.6678 - val_loss: 1.7306\n",
      "Epoch 27/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.9490 - loss: 0.1579 - val_accuracy: 0.6607 - val_loss: 1.7126\n",
      "Epoch 28/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m39s\u001b[0m 29ms/step - accuracy: 0.9469 - loss: 0.1662 - val_accuracy: 0.6663 - val_loss: 1.7408\n",
      "Epoch 29/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m42s\u001b[0m 31ms/step - accuracy: 0.9508 - loss: 0.1531 - val_accuracy: 0.6680 - val_loss: 1.7375\n",
      "Epoch 30/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.9489 - loss: 0.1566 - val_accuracy: 0.6637 - val_loss: 1.6716\n"
     ]
    }
   ],
   "source": [
    "kinit='glorot_normal'\n",
    "lr=0.001\n",
    "dr=0.1\n",
    "#a convolutional neural network described in Ch. 14, p. 447.\n",
    "model = keras.models.Sequential([\n",
    "        keras.layers.Conv2D(64, 7, activation=\"relu\",kernel_initializer=kinit, padding=\"same\",\n",
    "                            input_shape=[32, 32, 3]),\n",
    "        keras.layers.MaxPooling2D(2),\n",
    "        keras.layers.Conv2D(128, 3, activation=\"relu\",kernel_initializer=kinit, padding=\"same\"),\n",
    "        keras.layers.Conv2D(128, 3, activation=\"relu\",kernel_initializer=kinit, padding=\"same\"),\n",
    "        keras.layers.MaxPooling2D(2),\n",
    "        keras.layers.Conv2D(256, 3, activation=\"relu\",kernel_initializer=kinit, padding=\"same\"),\n",
    "        keras.layers.Conv2D(256, 3, activation=\"relu\",kernel_initializer=kinit, padding=\"same\"),\n",
    "        keras.layers.MaxPooling2D(2),\n",
    "        keras.layers.Flatten(),\n",
    "        keras.layers.Dense(128, activation=\"relu\",kernel_initializer=kinit),\n",
    "        keras.layers.Dropout(dr),\n",
    "        keras.layers.Dense(64, activation=\"relu\",kernel_initializer=kinit),\n",
    "        keras.layers.Dropout(dr),\n",
    "        keras.layers.Dense(10, activation=\"softmax\")\n",
    "])\n",
    "opt=keras.optimizers.Adam(learning_rate=lr)\n",
    "\n",
    "model.compile(loss=\"sparse_categorical_crossentropy\",\n",
    "                  optimizer=opt,\n",
    "                  metrics=[\"accuracy\"])\n",
    "\n",
    "history = model.fit(X_train, y_train, epochs=30,\n",
    "                    validation_data=(X_valid, y_valid))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "id": "-d6e-GYUY-kn"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.0, 1.0)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "pd.DataFrame(history.history).plot(figsize=(8, 5))\n",
    "plt.grid(True)\n",
    "plt.gca().set_ylim(0, 1) # set the vertical range to [0-1] plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "id": "qIu3ScKzZBb5"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[1m313/313\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m3s\u001b[0m 8ms/step - accuracy: 0.5309 - loss: 461.5630\n",
      "\u001b[1m1/1\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 1s/step\n",
      "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
      " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
      " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]]\n",
      "['dog' 'ship' 'truck']\n",
      "[['cat']\n",
      " ['ship']\n",
      " ['ship']]\n"
     ]
    }
   ],
   "source": [
    "model.evaluate(X_test, y_test)\n",
    "\n",
    "X_new = X_test[:3]\n",
    "y_proba = model.predict(X_new)# get prediction probabilities\n",
    "print(y_proba.round(2))\n",
    "\n",
    "y_pred = np.argmax(y_proba, axis=-1) # get predicted class index by finding the index of maximum probability along the last axis\n",
    "print(np.array(class_names)[y_pred])\n",
    "print(np.array(class_names)[y_test[:3]])\n",
    "#The predict_classes method was deprecated in TensorFlow 2.6 and has been removed in later versions.\n",
    "#Instead, you should use model.predict followed by np.argmax to get the predicted class indices."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/caroldong/anaconda3/lib/python3.11/site-packages/keras/src/layers/convolutional/base_conv.py:107: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.\n",
      "  super().__init__(activity_regularizer=activity_regularizer, **kwargs)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m44s\u001b[0m 31ms/step - accuracy: 0.2527 - loss: 1.9800 - val_accuracy: 0.4625 - val_loss: 1.4793\n",
      "Epoch 2/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.4645 - loss: 1.4651 - val_accuracy: 0.5382 - val_loss: 1.3037\n",
      "Epoch 3/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.5516 - loss: 1.2541 - val_accuracy: 0.6078 - val_loss: 1.1219\n",
      "Epoch 4/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.6196 - loss: 1.0715 - val_accuracy: 0.6338 - val_loss: 1.0595\n",
      "Epoch 5/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.6777 - loss: 0.9181 - val_accuracy: 0.6503 - val_loss: 1.0471\n",
      "Epoch 6/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.7138 - loss: 0.8135 - val_accuracy: 0.6478 - val_loss: 1.0510\n",
      "Epoch 7/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.7499 - loss: 0.7150 - val_accuracy: 0.6787 - val_loss: 0.9429\n",
      "Epoch 8/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.7827 - loss: 0.6180 - val_accuracy: 0.6822 - val_loss: 0.9845\n",
      "Epoch 9/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.8088 - loss: 0.5423 - val_accuracy: 0.6837 - val_loss: 0.9832\n",
      "Epoch 10/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.8367 - loss: 0.4586 - val_accuracy: 0.6815 - val_loss: 1.0311\n",
      "Epoch 11/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.8573 - loss: 0.4009 - val_accuracy: 0.6767 - val_loss: 1.1094\n",
      "Epoch 12/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.8709 - loss: 0.3663 - val_accuracy: 0.6858 - val_loss: 1.1524\n",
      "Epoch 13/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.8851 - loss: 0.3310 - val_accuracy: 0.6808 - val_loss: 1.1947\n",
      "Epoch 14/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.8982 - loss: 0.2958 - val_accuracy: 0.6793 - val_loss: 1.3230\n",
      "Epoch 15/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.9052 - loss: 0.2642 - val_accuracy: 0.6828 - val_loss: 1.4080\n",
      "Epoch 16/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m44s\u001b[0m 32ms/step - accuracy: 0.9131 - loss: 0.2505 - val_accuracy: 0.6860 - val_loss: 1.3663\n",
      "Epoch 17/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.9206 - loss: 0.2372 - val_accuracy: 0.6672 - val_loss: 1.4625\n",
      "Epoch 18/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.9258 - loss: 0.2171 - val_accuracy: 0.6788 - val_loss: 1.4024\n",
      "Epoch 19/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.9286 - loss: 0.2094 - val_accuracy: 0.6747 - val_loss: 1.5231\n",
      "Epoch 20/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.9297 - loss: 0.2093 - val_accuracy: 0.6697 - val_loss: 1.5703\n",
      "Epoch 21/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.9392 - loss: 0.1849 - val_accuracy: 0.6707 - val_loss: 1.5063\n",
      "Epoch 22/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.9372 - loss: 0.1852 - val_accuracy: 0.6695 - val_loss: 1.5657\n",
      "Epoch 23/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.9420 - loss: 0.1761 - val_accuracy: 0.6730 - val_loss: 1.5642\n",
      "Epoch 24/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.9462 - loss: 0.1610 - val_accuracy: 0.6708 - val_loss: 1.6410\n",
      "Epoch 25/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.9437 - loss: 0.1660 - val_accuracy: 0.6763 - val_loss: 1.6281\n",
      "Epoch 26/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.9468 - loss: 0.1574 - val_accuracy: 0.6688 - val_loss: 1.6046\n",
      "Epoch 27/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.9506 - loss: 0.1546 - val_accuracy: 0.6855 - val_loss: 1.6882\n",
      "Epoch 28/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m42s\u001b[0m 30ms/step - accuracy: 0.9535 - loss: 0.1394 - val_accuracy: 0.6752 - val_loss: 1.7870\n",
      "Epoch 29/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m44s\u001b[0m 32ms/step - accuracy: 0.9521 - loss: 0.1476 - val_accuracy: 0.6722 - val_loss: 1.8541\n",
      "Epoch 30/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m43s\u001b[0m 31ms/step - accuracy: 0.9585 - loss: 0.1273 - val_accuracy: 0.6713 - val_loss: 1.8116\n"
     ]
    }
   ],
   "source": [
    "kinit='glorot_uniform'\n",
    "lr=0.001\n",
    "dr=0.1\n",
    "#a convolutional neural network described in Ch. 14, p. 447.\n",
    "model = keras.models.Sequential([\n",
    "        keras.layers.Conv2D(64, 7, activation=\"relu\",kernel_initializer=kinit, padding=\"same\",\n",
    "                            input_shape=[32, 32, 3]),\n",
    "        keras.layers.MaxPooling2D(2),\n",
    "        keras.layers.Conv2D(128, 3, activation=\"relu\",kernel_initializer=kinit, padding=\"same\"),\n",
    "        keras.layers.Conv2D(128, 3, activation=\"relu\",kernel_initializer=kinit, padding=\"same\"),\n",
    "        keras.layers.MaxPooling2D(2),\n",
    "        keras.layers.Conv2D(256, 3, activation=\"relu\",kernel_initializer=kinit, padding=\"same\"),\n",
    "        keras.layers.Conv2D(256, 3, activation=\"relu\",kernel_initializer=kinit, padding=\"same\"),\n",
    "        keras.layers.MaxPooling2D(2),\n",
    "        keras.layers.Flatten(),\n",
    "        keras.layers.Dense(128, activation=\"relu\",kernel_initializer=kinit),\n",
    "        keras.layers.Dropout(dr),\n",
    "        keras.layers.Dense(64, activation=\"relu\",kernel_initializer=kinit),\n",
    "        keras.layers.Dropout(dr),\n",
    "        keras.layers.Dense(10, activation=\"softmax\")\n",
    "])\n",
    "opt=keras.optimizers.Adam(learning_rate=lr)\n",
    "\n",
    "model.compile(loss=\"sparse_categorical_crossentropy\",\n",
    "                  optimizer=opt,\n",
    "                  metrics=[\"accuracy\"])\n",
    "\n",
    "history = model.fit(X_train, y_train, epochs=30,\n",
    "                    validation_data=(X_valid, y_valid))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.0, 1.0)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAp8AAAGyCAYAAACiMq99AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjkuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8hTgPZAAAACXBIWXMAAA9hAAAPYQGoP6dpAACEqUlEQVR4nOzdd3gU5d7G8e+29E5IQiD03hEE6UpVFMFyLKBiPyrYOCriUdDXo1iO2JVjb2BXUEEEUbCAiPTeIbQESCA92c3uvn9MshAJkIQkk3J/rmuu3Z2d3fltnizcmZnneSxer9eLiIiIiEglsJpdgIiIiIjUHgqfIiIiIlJpFD5FREREpNIofIqIiIhIpVH4FBEREZFKo/ApIiIiIpVG4VNEREREKo3Cp4iIiIhUGoVPEREREak0Cp8iIiIiUmlKHT5/+eUXhg8fTnx8PBaLhZkzZ572NQsXLuSss87C39+f5s2b895775WhVBERERGp7kodPrOysujUqROvvvpqibbfuXMnF154Ieeddx6rVq3innvu4eabb+aHH34odbEiIiIiUr1ZvF6vt8wvtlj4+uuvGTly5Em3mTBhArNnz2bdunW+dVdddRVHjx5l7ty5Zd21iIiIiFRD9orewZIlSxg0aFCRdUOHDuWee+456Wvy8vLIy8vzPfZ4PKSmplKnTh0sFktFlSoiIiIiZeT1esnIyCA+Ph6r9eQn1ys8fCYlJREbG1tkXWxsLOnp6eTk5BAYGHjCa6ZMmcJjjz1W0aWJiIiISDnbs2cPDRo0OOnzFR4+y2LixImMHz/e9zgtLY2GDRuyc+dOQkNDK3z/LpeLn3/+mfPOOw+Hw1Hh+6uy3C7s/+uNJfsQ+SP+h7fFkErbdUnaIPOnnzj40L+xhoXRcObXWIOCKq2+iuL1eDj83HNkfPkVAHXuvpvwq68ypRZ9D8ynNqga1A7mUxuYryRtkJGRQZMmTU6b1So8fMbFxZGcnFxkXXJyMmFhYcUe9QTw9/fH39//hPVRUVGEhYVVSJ3Hc7lcBAUFUadOHf2S9xgFi1+CHbPgnKsrbbclaYOoSy8l/623ce7eje2nn4gaM6bS6qsI3vx8Dvz7YbyzZhFitxP32KNEXnGFafXoe2A+tUHVoHYwn9rAfCVpg8L1p7tEssLH+ezZsycLFiwosm7+/Pn07Nmzonct5aHLtcbt1nmQkWRuLX9jsdmIuulGAFLefQ+v02lyRWXndTrZN/5fpM2aBTYb8c88Y2rwFBERqSilDp+ZmZmsWrWKVatWAcZQSqtWrSIxMREwTplfd911vu1vu+02duzYwQMPPMCmTZt47bXX+Oyzz7j33nvL5xNIxarbEhJ6gNcNq2aYXc0JwkeOxF63LvlJSaTNnmN2OWXiyclhz9hxZMybh8XhoMFLLxI+/CKzyxIREakQpT7t/tdff3Heeef5HhdemzlmzBjee+89Dhw44AuiAE2aNGH27Nnce++9vPjiizRo0IC33nqLoUOHlkP5Uim6XAt7lsLKj6DPvVCFRhyw+vkRNeY6Dv73OVLeeovwERdjOUUPuzPlzsjg4LP/JW/rViwOBxY/v2O3vvsOLA6/glvjOevx2/3tNvW998letgxLQAANXn2FkN69K6x+ERERs5U6fJ577rmcamjQ4mYvOvfcc1m5cmVpdyVVRbuR8P0ESN0OiUugUS+zKyoi4qqrOPy/N3Bu307mwoWEDhhQIftx7trFnjvG4tyxo9zf2xoSQsL/phHUtWu5v7eIiEhVUiV7u0sV4x8K7S8xjnyu+LDKhU9bSAiRV11FyptvkvLGm4Scd165jwebtXgxe+8djyctDXtcHDH/Go/FZsPrcuFxOvG6XHidTrxOF15Xwe3x633Pn7jOGhJC3XvvJbB9u3KtWUREpCpS+JSS6XKdET43zIQLnoaAih91oDSirruW1PffJ2fVKnKWLyeoW7dyeV+v18uRj6aT/NRT4HYT2KkTDV55GXvduuXy/iIiIrVNhfd2lxoioTvUaQGubFj/ldnVnMBety7hl1wCwOE33yyX9/Q6nSRNmkzyE0+A2034iBE0/OB9BU8REZEzoPApJWOxwFkFwy6t/MjcWk6izo03gNVK1qJfyN28+YzeKz81ld033sjRzz8Hi4WYBx6g3lNTsBYz/qyIiIiUnMKnlFynq8Fig73L4OAms6s5gV+jRoQONWZhSnnr7TK/T+7mzey6/B/k/LXc6Ag07XXq3HhDuV9HKiIiUhspfErJhcRAy/ON+ys/NLeWk6hz880ApM+Zg3PvvlK/PuPHH9l19Shc+/fjaNSQxp9+Qkj//uVdpoiISK2l8CmlU3jqffUnkF/1ZhQKbNeO4F69wO0m9d13S/w6r9fL4ddfZ++4O/FmZxPU8xyafPop/s2aVWC1IiIitY/Cp5RO88EQEgvZh2HLXLOrKVadW4yjn0e//JL81NTTbu/JyWH/v/7FoRdfAiDymmto+MYb2CIiKrJMERGRWknhU0rHZjeu/YQqe+o96JxzCGjXDm9uLkc+OnXnKFdSEruvuZb0Od+D3U7c/z1G3MP/xuJwVFK1IiIitYvCp5Rel4JT79t+hPT95tZSDIvFQp1bbgEgdfoMPFlZxW6Xs2oVO//xD3LXr8cWGUmjd98h8oorKrNUERGRWkfhU0ovujk07AVeD6yaYXY1xQodPAi/Ro3wpKVx5PPPT3g+bdYsdl83Bvehw/i3bEnjzz8j6OyzTahURESkdtEMR1I2Xa6BxMXGmJ99xoO1av0dY7HZiLrpRpImTSb1vfeJGjUKi58fXrebg1Onkvr2OwCEDBxI/NNPYwsJNrliERGRkvN6vWTk5XM0y0VqtpMj2U6OZjtJzXIV3Do5mu2if8u6XHF2gtnlFqHwKWXTbiR8PwGO7DRCaOM+Zld0gvARIzj08svkJyWR9t1sQocMZv+/7iNz0SIA6tz2T+redReWKhacRUSkZvB6veR7vLgLluPvG489eDyQ7/H4ns9xuTmS5eRItqvgtmApCJnHB8x8j/e0NYQHORQ+pYbwC4b2l8KK92HFh1UyfFr9/akzZgwH//scKf/7Hylvv41z+3Ys/v7Ue/IJwi+80OwSRUSkAnm9XjLz8klOz+NgRi4H0/NITs/laI4LV76HfI8Xl9tDvtuLy2Pc5ns8uNxe8t1/f75g3fHbuj24igmUhfdLkA3PWKDDRmSQg4ggP6KC/YgIchTc+hEV5KBd/fCKL6KUFD6l7Lpca4TPDbNg2DMQUPV+wSOuvJLD0/6Hc/duAOyxsTR45RUCO7Q3uTIRESmrk4XKgxnHbg+m55KcnkeOy212ucWyWy3YrBbfrbFYCXBYiQzyIzLYj8ggh3E/yI+o4BMDZmSQHwEOm9kfpdQUPqXsGnSDuq3h0CZY9yV0u9Hsik5gCw0l6vrrOfzKKwR07EiDV17GERNjdlkiIpXO7fGS5cwnKy+frDx3wW0+mXn5ZDnzycxz48z34PZ4cHvA7TGO/Hk8xZ86LvqcB7e34DVuLx6vsY3FYsFqMUYhsQBWiwWr9W+PC5+3GI+L287r9bBxu5WPDizjUEZeqUNlqL+dmDB/YkIDiA3zJzLYDz+bFbvNgt1qxWGzYLdZsVstOArWO2wF663HbgvX263G9sevLwySVovluMdWbBYLNttxz1ktWK21e7pmhU8pO4vFOPo579/GqfcqGD4BosfeQch55xLQogUWPz+zyxERKbMcp5uk9FySC5aD6Xlk5OUXDZJ5+WQ5jw+Xxv2qegSw5KyQcqTImsJQGRsWQEyocVu34LZwXUyYP0F+ijtViVpDzkzHK+HHybB/BSSvh9h2Zld0AovFQmC7qleXiEghj8dLSpaT5PRcktJyfQHz7/fTc/PPeF92q4Vgfzsh/naC/Gy++8H+Nvzttr+dBi48LWzFZgWb1eo7clfcaePjn7NYLOAFL8a1jx5vwTWQ3qKPvV4v3uMfU/DYc+xxfr6bfbu20r97F+IjgxUqqzm1mpyZkLrQ6gLY+K0x7NL5U8yuSESk1NweLylZeSdcO5icnsfB9BySk618d3QVfnZbkcBV5PTqCYGs6OnYwtvs445eJqXl+q5bdLlL1jsl0GEjLtw4fRwbFkBogN0IkH7244KknSB/m3Hf71i4DPa342+3GsGwGnG5XMyZs4VhHeJwaAa6ak/hU85cl2uN8Ln6Exj0KNj9za5IRAQwjp4dyXae0DElOaMwWBq3hzLzcJ+ya7KVdUcOVmitFgtEh/gTV3DKOC7cuB8TFkBcWEBB4AwgLMBe7cKjyPEUPuXMNRsIofUg4wBs/t4YA1REpBy4PUav5sJrGTNyj93PzM0no+A2y1n0uSPZTg6W8oiitSD8HbtW0Di6GBVkZ8O6dbRp1x4sVl8Hm3yPF7e7aAcc3/rCx+4T1/vbrcSGFwTKsADf/bqh/jhsGndYaj6FTzlzNjt0HgW/PgfL3lL4FJHTysrLZ8ehLLYfymTbwUx2HM7kSJbL6HV9XKgsr04ydYL9fGEytqDHc0zBEcbC09d1gv2wFxP+XC4Xcw6tZVj3BJ3yFSkHCp9SPrreAL+/CLt+hb3LoUFXsysSEZN5vV4OZeax/WAW2w5lsv1gJtsLbven5ZbqvRw2C6EBDt/1jKH+dkICjl3jGBpgL/JcWKDd1+M5OsQfP7uOKIpUFQqfUj4iEqDDFbB6Bvw2Fa6abnZFIlJJ8t0e9hzJYfvBTF/ILLw9Ve/s6BA/mtYNoVndEJrVDaZuqH9BiHQQ7G8j1N9REDCNXtgiUjMofEr56XMPrP4YNn0HBzdBTGuzKxKRcuLM97D/aA57jmSzJzWHxNRsElOz2HYwk12Hs3G6PcW+zmqBhKggmtUNoXmMETKbx4TQNDqEyGCNuytSGyl8Svmp2wpaX2iEz99fhEteN7siESkhj8c4Rb4nNZs9R7JJTCkMmtnsPZLDgbScU85THeCw0jQ6hGYxITSvG0KzGCNkNq4TXC2n/xORiqPwKeWr73gjfK79DM57yDgdLyJVQkaui90p2ew97ujl8QEzL7/4o5eFAhxWEiKDSIgKIiEy0DiiWRA260cE1vopA0WkZBQ+pXzV7wpN+sPORbD4ZRj2jNkVidQ6uS432w5msiU5g83JGWxJymBLcib7juac8nVWC9QLD6RhVBAJUYHHgmbB47oh/hpfUkTOmMKnlL8+9xrhc8UH0P8BCI42uyKRGinf7WFXSrYRMpOMZUtyBrtSsk56ijwq2M935LJhYbCMDKJhVBD1IgI0zqSIVDiFTyl/Tc+F+C6wfyX88ToMfMTsikSqNY/Hy57U7CJHMjcnG73JT9bRJzzQQau4UFrFhtKy8DY2hIggdfIREXMpfEr5s1igz3j47FpY9ib0vhsCwsyuSqTK8Hi8ZOTlk5bt4miOk6PZLo7muEjLdpKW4/I9PpKVx7a9Nh5a/hNZzuIHWw/ys9EiNpRWsSG0jA31Bc66oTpFLiJVk8KnVIzWF0GdFpCyFZa/awRQkRruUEYeKxOPsOdIzrEgeVyYTMt2cjTHRXqO65Q9x4uyAG78bFaa1g2mVVyoETILgqY6+ohIdaPwKRXDajXG/Zw1Fpa8Ct3/CY4As6sSKTf5bg+bkjJYkXiE5buPsCLxCHtST92h5++C/GxEBDoIC3QQEeQgItCPiCAH4QX3Q/ys7Ny0ln8M7UvzuHBdjykiNYLCp1ScDlfAz09C+j5j5qNuN5pdkUiZpWTmsSLxKCsSj7Bi9xHW7E07Yd5xiwVaxYbSIjaUiIJAGR7oICLIr8jj8ILb083aY8wpvobmMSEKniJSYyh8SsWx+0GvO2Hug8ag812uA5t+5aTqy3d72JycwYrEo6wsOKq5KyX7hO3CAux0aRjJWQ0jOatRBJ0SIggLcJhQsYhI9aEkIBXrrOtg0TNwZBdsmAkdLje7IpETHMlysnLPEVbsNo5srtpzlOxiOvi0iAnxBc2ujSJpGh2i6y1FREpJ4VMqll8w9LgNFj4Jv70A7S8zzk2KmCgt28UfO1NYsj2FP3aksCkp44RtQv3tdG4YURA2I+mcEEF4oI5qioicKYVPqXjdb4HFL0HyWtg6H1oOMbsiqWUycl0s25XKku0pLNmRwvr96Xj/1tu8ad1guhYEzbMaRtI8JgSbjmqKiJQ7hU+peEFR0PV6WPIK/DZV4VMqXLYzn792HWHJjhQWb09h3b403H8b26hZ3WB6NqtDr2bR9GgSRZ0Qf5OqFRGpXRQ+pXL0HAd/vgGJS2D3EmjU0+yKpAbJdblZsdsIm0u2p7B671Fc7qJhs1GdIHo2rUPPZnXo2bQOMWEa+ktExAwKn1I5wupBp6thxfvw2/MKn3JG8vLdrN6TVnAa/TArEo/izC86zWT9iEBf0OzZrA7xEYEmVSsiIsdT+JTK0/tuWPkhbP0BktZBXHuzK5JqIjk9lxUFQx6tSDzK2n1pJ4TNmFB/ejUrPLIZTUJUoKaXFBGpghQ+pfLUaQZtR8D6r42jn5e/bXZFUgW53B427E/3Bc0Vu4+w7+iJMwfVCfbjnOOObDaNDlbYFBGpBhQ+pXL1udcIn+u/ggEPQ1QTsysSkx3KyCsImsdmDsr721FNqwVaxYVx1nFDHzWuE6SwKSJSDSl8SuWq1wmaD4JtPxrDL130vNkVSSUq6XzoEUEOuiQcC5qdEiII8dc/VyIiNYH+NZfK1+deI3yunA79H4TQWLMrkgqUkpnHnLUH+H5dEisTjxY7H3rLmFDOahRBl4aRBTMH6RS6iEhNpfApla9Rb2jQHfb+CX+8CoP/z+yKpJyl5biYtz6Jb1bvZ/H2lCJjbIb65kM3pqjUfOgiIrWLwqdUPosF+o6Hj6+CZe9An/EQGGF2VXKGcpxuFmxK5ptV+1m4+RBO97HrNjs2CGd4x3jObVWXZnU1H7qISG2m8CnmaDEUYtrCwQ2w7C3od5/ZFUkZOPM9/LLlEN+u2c/8DclkO4+dUm8RE8LFneK5qFM8TaKDTaxSRESqEoVPMYfVCr3vga9vhT9eh3PuAL8gs6uSEnB7vPyxI4VvV+/n+3VJpOW4fM8lRAUyvGM8F3eOp1VsqK7bFBGREyh8innaXwY//weOJsLKj6DHrWZXJCfh9XpZkXiUb1fvZ/baAxzKyPM9VzfUn4s61uPiTvF0TohQ4BQRkVNS+BTz2OzQ6y6Yc58x7FK3G8CmjidVhdfrZV8WPDtvC7PXJhcZ6D080MGwDnEM7xRPjyZ1sOkaThERKSGFTzFXl2tg0dOQtgfWfQmdrjK7olrP5fYwZ+0B3vxlB+v224FdAAT52RjSNpaLO8fTp3ld/OxWU+sUEZHqSeFTzOUIhHNuhwX/Z0y52eEK43pQqXRp2S4+XpbIe7/vIik9FwCbxcuA1rGM7NKAAa1jCPSzmVyliIhUdwqfYr6zb4bfXoBDm2DL99D6QrMrqlV2Hc7i3d938vnyvb7e6tEh/lzTI4G6aZu4YkRnHA5dDiEiIuVD4VPMFxAOZ99kHPn8dSq0GmZ2RTWe1+tl2a4jvPXrDuZvTMZbMAZ867hQburThIs7x2P1epgzZ5O5hYqISI2j8ClVwzl3GEMu7fsLdv0KDXqaXVGN5HJ7+H5dEm/9uoM1e9N8689tVZeb+zSld/M6vt7qLpfnZG8jIiJSZgqfUjWExBidj5a9ZRwBvUrhszyl5bj4tOB6zv1pxvWcfnYrl51Vnxt7N6FFbKjJFYqISG2h8ClVR6874a93YftPcGCV2dXUCIkp2by7eCefLdtDlu96Tj+uPacx15zTkDoh/iZXKCIitY3Cp1QdkY2NgefXfoZt8UsQeLnZFVVLxoDwR3jr1538sD4JT8H1nK1iQ7mpbxMu7hRPgEO91kVExBwKn1K19LkX1n6GZdO3hLTpbXY11Uq+28Pc9Um89etOVu056lvfv2Vdbu7bhD7NozX7kIiImE7hU6qW2LbQ8gIsW76nefJs4CazK6rysp35fLZsD2//vpM9qcYsRH52K5d2qc+NfZrQUtdziohIFaLwKVVPn3thy/ckHPkd99FEqNvM7IqqpMOZeby/eBcf/rGbo9kuAKKC/bj2nEZc27MR0bqeU0REqiCFT6l6GvbA07gf1l2/wMIn4B/vmF1RlbLjUCZv/rqTL1fsxZlvDIfUqE4QN/dtyuVnNdAsRCIiUqUpfEqV5B4wGes7A7Gu/xJ63wnxXcwuyXTLd6fyv0VFB4XvnBDBP/s1ZUi7OGxWXc8pIiJVn8KnVE31OrEnshcJRxbDvEdgzLdQCzvLeDxe5m9M5o1fdrB89xHf+kFtYri1XzPObhypTkQiIlKtWMvyoldffZXGjRsTEBBAjx49+PPPP0+5/QsvvECrVq0IDAwkISGBe++9l9zc3DIVLLXHxvjL8dr8jBmPts43u5xKletyM2NpIoOmLuKfHy5n+e4j+NmsXNGtAfPv7cdbY86me5MoBU8REal2Sn3k89NPP2X8+PFMmzaNHj168MILLzB06FA2b95MTEzMCdvPmDGDBx98kHfeeYdevXqxZcsWrr/+eiwWC1OnTi2XDyE1U45fNJ6zb8H2x6swfxI0GwC2mn2w/mi2k4/+2M17i3dxONMJQGiAnWvOacQNvRoTExZgcoUiIiJnptT/k0+dOpVbbrmFG264AYBp06Yxe/Zs3nnnHR588METtl+8eDG9e/dm1KhRADRu3Jirr76apUuXnmHpUht4et2LbdV0OLQRVs+As64zu6QKsSc1m7d/28lnf+0hu2AmovjwAG7s04SrujckxL9mh24REak9SvU/mtPpZPny5UycONG3zmq1MmjQIJYsWVLsa3r16sVHH33En3/+Sffu3dmxYwdz5szh2muvPel+8vLyyMvL8z1OT08HwOVy4XK5SlNymRTuozL2JcXztYE9GGuf8dh+nIT3p/+Q3+pi8As2ubrys35/Om/+tou565NxF0xF1DoulJv7NGZY+1gcNivgNeV3Ud8D86kNqga1g/nUBuYrSRuUtH0sXm9hv9nT279/P/Xr12fx4sX07NnTt/6BBx5g0aJFJz2a+dJLL3Hffffh9XrJz8/ntttu4/XXXz/pfh599FEee+yxE9bPmDGDoKCgkpYrNYTV42LAxgkEOw+zsd5lbIkbYXZJZ8TrhS1pFn7cb2FL2rHLrluGexgY76VVuLc29q0SEZFqLjs7m1GjRpGWlkZYWNhJt6vwc3kLFy7kySef5LXXXqNHjx5s27aNu+++m8cff5xHHnmk2NdMnDiR8ePH+x6np6eTkJDAkCFDTvlhyovL5WL+/PkMHjwYh8NR4fuTE/29DSxN8mHmP2mdMpfmVzwOISdeX1zV5bs9/LDhIG/+tpP1+zMAsFktXNAulpv7NKZdfMX/bpeGvgfmUxtUDWoH86kNzFeSNig8U306pQqf0dHR2Gw2kpOTi6xPTk4mLi6u2Nc88sgjXHvttdx8880AdOjQgaysLG699Vb+/e9/Y7We2OHe398ff/8TZ2dxOByV+ktX2fuTE/naoOMV8Oc0LPtX4lg8FS58zuzSSizH6eaL5Xt489edJKZmAxDgsHLV2Q25qU8TEqKq9tF8fQ/MpzaoGtQO5lMbmO9UbVDStilV+PTz86Nr164sWLCAkSNHAuDxeFiwYAHjxo0r9jXZ2dknBEybzZiBpRRn/KW2s1ph8OPw/kXw17vQ4zaIbmF2Vad0JMvJhwU911OzjJ7rkUEOxvRqzHU9GxMV7GdyhSIiIpWv1Kfdx48fz5gxY+jWrRvdu3fnhRdeICsry9f7/brrrqN+/fpMmTIFgOHDhzN16lS6dOniO+3+yCOPMHz4cF8IFSmRJn2h5fmwZS78+ChcNd3sioq194jRc/2TP/eQ4zJ6rjeIDOSWvk25oluCpr8UEZFardTh88orr+TQoUNMmjSJpKQkOnfuzNy5c4mNjQUgMTGxyJHOhx9+GIvFwsMPP8y+ffuoW7cuw4cP54knnii/TyG1x6DHYOs82PQd7F4CjXqe/jWVZOOBdP63aDvfrjng67netl4Y/+zflAs71MNuK9OcDiIiIjVKmTocjRs37qSn2RcuXFh0B3Y7kydPZvLkyWXZlUhRMa2hy7Ww4n2Y/wjcNN/UaTe9Xi9/7Ehl2qLtLNpyyLe+d/M63Na/GX2aR2sWIhERkeNo5Gqpfs57CNZ+DnuXwYZZ0G5kpZfg9niZtz6JaYu2s3pvGgBWC1zQoR639WtGhwbhlV6TiIhIdaDwKdVPaBz0uhMWPQ0LHoNWw8BeOZ133B4vXy7fy+uLtrPzcBYA/nYr/+jWgFv6NqVRnZozAL6IiEhFUPiU6qnXnUav99QdsPxd6PHPCt/l6j1HeWTWOtYUHOkMD3RwXc9GjOnVmOiQE4cGExERkRMpfEr15B8K5z4Is8cbR0A7XQUBFXOq+0iWk2d+2MwnyxLxeiHU386dA5szukcjgjXnuoiISKnof06pvs4aA3+8Dilb4bcXYFD5dmpze7x8umwPz/ywiaPZxny1l55Vn4kXtKFuqI50ioiIlIXCp1RfNjsMfgw+GQV/vAZn3wzh9cvlrVfvOcqkWet8nYlax4Xy+Mj2nN04qlzeX0REpLZS+JTqrdUwaNgLEhfDz0/AyNfO6O2KO8U+fkhLrj2nkcbpFBERKQf631SqN4sFhjxu3F81A5LWleltPB4vH/+ZyHnPLeTjP43geWmX+iy4rz839G6i4CkiIlJOdORTqr8G3aDdJbD+a5g/Ca79qlQvL+4U+/+NaE/3JjrFLiIiUt4UPqVmGDgJNn4H2xfA9p+g2YDTvuRIlpNn5232HekM9bdz7+CWXNdTp9hFREQqisKn1AxRTY0OR0tfN45+NjkXrMUHSI/Hy6d/7eGZuZs4UtiLvUt9HhzWmpjQgMqrWUREpBZS+JSao9/9sGo6JK2FtZ8ZY3/+zZq9R3lk1npW7zkKQKvYUP5vRDt6NK1TycWKiIjUTgqfUnME14G+4+HHR2HB49B2BDgCATia7eTZHzYzo+AUe8hxp9gdOsUuIiJSaRQ+pWbpcRv8+Rak74Wl/8Pb+25mrtrH/327wXeK/ZIu9Zl4QWtiwnSKXUREpLIpfErN4giEAQ/DzNvw/Poc927pwKwtuYBOsYuIiFQFOt8oNY63wz84GtYKa146HXe8icNm4b4hLfnurj4KniIiIiZT+JQa5WBGLrd8tIpxhy8F4Dr7fOZe14BxA1ro2k4REZEqQP8bS43g9XqZuXIfg6f+wo8bk1lq6cjuiHNwkE+zNc+bXZ6IiIgUUPiUau9gRi63fricez5dRVqOi3bxYXwzrg+NrvovYIH1X8He5WaXKSIiIih8SjXm9XqZtWofQ57/hfkbknHYLIwf3JKZY3vTpl4YxHWATlcbG897GLxecwsWERER9XaX6ulgRi4Pf72OeRuSAWgXH8Z//9HJCJ3HG/Bv48hn4mLYPAdaX2hCtSIiIlJIRz6lWjn+aOe8DcnYrRbuHXTc0c6/C28A59xu3J/3MOQ7K7dgERERKUJHPqXaOJSRx8Mz1/LDeuNoZ9t6xtHOtvHFhM7j9f0XrJwOqTtg6TTofVclVCsiIiLF0ZFPqfIKj3YOfn4RP6w/drRz1rjepw+eAP6hMGiycf+XZyHzYMUWLCIiIiel8ClV2uHMPG7/aAV3f7KKo9ku2tYzerLfPaiU43Z2GgX1OkNeOvz0eIXVKyIiIqem8ClVktfrZcVhC8NeXszc9UnYrRbuGdSi5Ec7/85qhQueNu6v+BD2ryrXekVERKRkFD6lyknNcnLXp2t4f6uNI9ku2tQLY9a43twzqOWZzVLU8BxofznghbkPauglEREREyh8SpXy8+aDDH3hF+auT8Zq8XLneU2ZNbY37eLDy2cHgx8DeyAkLoH1X5fPe4qIiEiJKXxKlZDjdPPIzHXc8O4yDmXk0axuMOPbu7lrQHP87OX4axreAPrca9yfPwmc2eX33iIiInJaCp9iutV7jnLhS7/y4R+7Abi+V2Nm3n4OCSEVtMNed0JYA0jbA4tfrqCdiIiISHEUPsU0+W4PLy3YymWvL2bH4Sxiw/z54MbuPHpxOwIctorbsV8QDPk/4/5vz0Pa3orbl4iIiBSh8Cmm2HU4i3/8bwlT528h3+Plwg71+OGefvRrWbdyCmh3KTTsCfk58OOjlbNPERERUfiUyuX1evn4z0SGvfQrKxOPEupv5/krO/HKqC5EBPlVXiEWC5z/FGCBtZ9D4tLK27eIiEgtpvApleZwZh63fPAXE79aS7bTzTlNo5h7bz8u6dIAi8VS+QXFd4Yu1xj3504Aj6fyaxAREallFD6lUvy4IZmhz//CjxsP4mez8tCw1sy4+RzqRwSaW9jASeAXCvtXwuqPza1FRESkFlD4lAqVlZfPxK/WcPMHf5GS5aR1XCizxvXm1n7NsFpNONr5dyEx0P9+4/6CxyAvw9x6REREaji72QVIzbV89xHGf7aK3SnZWCxwS9+mjB/csmJ7spdFj9tg+XuQugN+fQ4GPWp2RTVCWl4aW45sYcuRLWxO3cyOtB0E2AKIDoomJjCG6MBo6gbVpW5gXd9tkCPI7LIrjNvjJtOVSYYzo8iS7kw37rtOXJeel87R9KNsWrGJkS1G0iqqldkfQ6Ra23ZkG/N3z2dv5l5aRrakXZ12tK3Ttkb/21MVKXxKuXMVDKH06s/b8HghPjyA567oTM9mdcwurXh2fxj6JHx8FSx5Fc66DqKaAuB0O0l3ppOWl+Zb0p3pZOdn43Q7cbqd5LnzjPue4+4Xrvc4fY+LbHvc9m6PG3+7P4H2QILsQQTaA0++OE7+XOFrw/3DiQyIxGqpnBMb+Z58EtMTjZB5ZLMvbCZnJ5f6vYIdwb4wGh0YTd3AusQExfjuF4bUYEewKdcJFwbIwt8D3++FM430vHTSnMd+R/4eMjNdmWXe70ebPuKjTR/RJqoNFze7mGFNhxEVEFWOn0ykZvJ6vWw9upV5u+Yxf/d8dqTtOGEbCxaahjelXXQ72tVpR7vodrSOao2/zd+EimsHhU8pV9sPZXLvp6tYszcNgEu61OfRi9sRHugwta5sVzZH8476gsLxQTI9L520vKOkNWlDek4Kad9dSVpgOOnOdHLycyqlPqfTSYaz/E752yw26gTUIToo2hfc6gTWoW6gEeoKl7pBdUv1D2zh0czNqZt9YXP70e3kufOK3b5+SH1aRrakVVQrmkU0I9+Tz+HswxzMOXjsNucwB7MPkpOfQ5YriyxXFrvSd52yjkB7IBH+Efjb/PGz+eFn9cPP5nfsse24x8c957A58Lf5G/etDt/2dqudTGemL1D+/fb4QOnFW5qmKLb2UEcooX4nLmF+YUUeB1mD+HXpryTXSeaXfb+wMXUjG1M38txfz9G3QV9GNB9Bv/r9cNgq//vl9rg5kneEcL9wU/YvcjJer5fNRzb7Aufx/544rA56xfeiTZ02bEndwvqU9SRnJ7M9bTvb07bzzfZvALBb7DSPbO4Lo+3rtKd5ZHMcVv2ulweFTykXXq+Xj/7YzRNzNpLr8hAe6OCJS9pzUcd4U+s6knuEF1a8wMxtM/F4S9CbPTAAvLmQnetbZcFCmH8Y4X7hhPuHE+YXRpAj6JTB5+/h5u9B6PjtbRYbue5ccvJzyHHlGLclWLLzs0/YPjs/mwxnBm6vm4M5BzmYc/C0HznUEUp00IkBNdIvkjXONWxftZ1tadvYcmTLSY9mBtoDaRHZglaRrXxhs3lEc0L9QkvcVlmuLA5lH+JQziHfbWEwPZxz2Lc+05Xp+7xmCbIH+X4Xwv2P/V4U/p6E+oUS5h9GmCPshHBZmqDmcrlI9Uvl/r73k+XOYs7OOXyz/RvWp6zn5z0/8/Oen4n0j2RY02Fc3Oxi2kS1qbAjwlmuLNYcWsPKgytZeXAlaw6tITvfmJ42xBFCVEAUkQGRRAZEGvf9j7tfuN7fuB9gD6iQGqVq8Hq9pDvTyXRlEhMUUymBzev1sjF1oy9wJmYk+p7zs/rRq34vhjQawrkJ557w79LhnMOsP7ye9SnrWXd4HetT1pOam8qm1E1sSt3El1u/9L1Pq6hWRQJpk/Am2KxV7FKyakDhU86Y1+vlsW838N7iXQD0bRHNs5d3Ii7cvP9gPF4PX2/9mudXPE9annEU1s/qR4R/hBEKjgsN4X7hx8Llhu8I2zqP8PBGhF31GeFBUYQ4QirtFHZ5cHlcpOSkkJKT4gtth3MO+5ZDOYdIyUnhUPYhnB6nca1hWgY703YW/4Ybij6sH1LfCJlRLX1hs0FogzP+GQU7ggkOD6ZxeONTbpeTn8Ph7MMczTta5LKG4y9zONXlD0We8xj38z35BDuCT/574R9+7PfGpCN9EQERjGozilFtRrHtyDa+2f4N3+74lsM5h5m+cTrTN06nRWQLRjQbwYVNLyQ6MPqM9peUleQLmqsOrmLzkc0n/QMu05VJpiuzyH/4pxJoDyQqIOpYMPU3QmrhHz/HXwsc7Ag+o88h5cvj9XAk9wjJ2ckkZyUbt9nJHMw+WORx4R+HdoudBqENaBTWyLc0DmtMo7BGxATFnNEfS16vl/Up65m3ex7zdxnXcRbyt/nTp34fhjQaQr8G/QjxO/l8zdGB0fRP6E//hP6+903KSmJ9StFAmuHMYO3htaw9vBY2G68NtAfSOqo1sUGxhPiFEOoIJcQvhBBHiO+PziL3/UIItgfX+sCq8ClnxOv18n/fGcHTYoF/D2vDjb2bmNqTfXPqZh7/43FWH1oNQMvIljx8zsN0iely+hc3HAxbz4KkzbBlPnS/pYKrLX8Oq4O44DjiguNOuZ3X6yXDlcHh7MPFh9TsQyQdSqJbk260qdOGllEtaRHR4pT/iFeGQHsgCWEJJJBgah1mah7ZnPHdxnPXWXexZP8SZm2fxc+JP7P1yFb++9d/eX758/Sp34eLm13MuQnn4mc79QQObo+brUe3FgmbB7IOnLBdfHA8nWM60yWmC11iutAsohlZrixSc1M5knuEI7lHSM0z7qfmphZZX/hcviefnPwc9mXuY1/mvtN+1iB70LEwGlj3WIe1427rBtYlxBFiznjB5cjr9ZKTn1PsZR8nXFecl+47uuhvM64ZD7AHnPqa8b9dH17cNeRZriwjSB4fLv8WMl0eV4k+j91qJ9+Tz670XcVeShNoDyw2lDYKa0S4f/hJf0ZrDq3xHeHcn7Xf91yALYC+Dfr6AmdZOxFZLBbqhdSjXkg9BjUa5Nvvnow9RiA9vJ51KevYmLKR7PxsVh5cWep9BDuCTwioheE1yBGEzWLDZrVht9qxW+zYrDZsluIf+26tthO3tdqJDowmIbRq/Xtp8Xq9Z3YBUyVIT08nPDyctLQ0wsLCKnx/LpeLOXPmMGzYMBwOXd9xMl6vl//M3sjbvxlHzJ6+rANXnt2wXN67LG2Q6czk1VWvMmPTDDxeD0H2IMZ2HsuoNqOwW0vxd9afb8Kc+yAwEu5cAUG1s2OHvgfmK00bpOWl8cOuH5i1fRZrDq3xrQ/zC+OCJhcwotkI2ke3x2KxkO3KZs3hglPoyStZc3gNWa6sIu9ns9hoGdmSs2LPMgJn3S7EBsee0ecp/IPHF0ZziwbVlJyUE64DLqlAe2CRjmmhfqG4PW5cHhf5nnzf4vIWfZzvyT9hm3xPPvneY+vdHjd4IMAvAIfVgd1qx2F14LA5jNvC5e+Pj1vne43VgcvjOna9+d+uQS9psDOTBQt1AusQGxRrLMHGbUxQDHHBcb77fjY/DmYfZFf6Lnan7TZu03ezO303+zL34fa6T7qPSP9IGocfC6NxAXF8u/xbdth2kJSd5Nsu0B5Ivwb9GNJoCH3q96nUXutuj5vd6bvZmLqRI7lHyHBlkOnMLDKqxfGPM12ZJ70+viJd2epKHj7n4TN+n5L8e1TSvKYjn1ImXq+XJ+ccC55TLi2/4FmWWn7Y/QPP/vms7xrHIY2G8MDZD5TtP8uuN8Cyt+HQRlj0NFzwdDlXLFL+wv3DuaLVFVzR6gp2pO3g2+3f8s32bziYfZBPN3/Kp5s/pWl4U/xt/mw5suWE//iDHcF0qtvJd2SzY3THcv+P3GKxGNfG+oXRKKzRabcv7XXAezL2sCdjT7nWfDyn01lh7308u9Ve5Brz4q4rLlwX7AjG6XYWf124K7vE15Dn5Ofg8XqwWWzUDap7QrD03QbFUjewbokvPSk8C3NOvXOKrHe5XezN3OsLo75gmrabgzkHOZJ3hCMHjxR7VDHIHkT/hP4MaTSE3vV7E2g3Z7ISm9VG04imNI1oWuLXON1OXxDNdGb6hljLdB4LqDn5OcYfPV43bo8bt9f4I+qExwX3C//IOtn2VXFkDIVPKTWv18tT32/izV+N4PnEJe25urs5wXN3+m6eXPoki/cvBqBhaEMe6vEQvev3Lvub2uxwwVPwwQjjKGjXGyCmdTlVLFLxmoY35e6z7mZc53EsTVrKrG2zWJC4oMgwM/WC6xU5hd4iokWVuw6ttNcBH8o55BtJIdOV6TvaWHiqsvA05PGLw+o46XN2qx2HxYHH4+Gnn36id7/eeK1e39HSwiXfk4/L7Sqy7oTHx21ntVqJ8I844briwoAZaA+s9EsIvF4vTo/Td9q2ojlsDpqEN6FJeJMTnstyZZGYnlgklCamJ2JJszCm5xj6JvSttp3W/Gx+1AmsQ53AKjr0YCVR+JRS8Xq9PD13M//7xfhP7PGR7Rnd4/RHMMpbbn4ub697m7fXvo3L48LP6sfNHW7mxg43ls/YbE3PhdYXwabv4IeJcM1XUM2vJ5Pax2a10Su+F73ie5HhzGDR3kXYLDa6xHQ57TXB1YnvOuCwirmuzeVyEWWLokl4kxp7CYrFYqky41oGO4JpU6cNbeq08a0rPOV7boNzcdhrZhvUJgqfUmJer5dnf9jMtEXbAfi/Ee249pzKD56/7v2VKX9O8Z1e6x3fm4d6PETDsHI++jrkcdg6D7b/BFt+gFbnl+/7i1SiUL9QLmp6kdlliIgofErJeL1enpu3hdcWGsHz0eFtua5n40qtISkriWeWPcP83fMBiAmKYcLZExjcaHDFnKKKagrn3AG/v2Ac/Ww2AOyn7jUsIiIip1Z9Bi8UUz3/41Ze+XkbAJMuasv1vU+8TqeiuDwu3lv3HhfPvJj5u+djs9gY03YM34z8hiGNh1TstVH97oOQWGPe96XTKm4/IiIitYSOfMppvfDjFl5asBWAhy9sw419ig+eh7IP4fF6CHYEE+QIKpeB2VceXMmUv6aw7agRfDvX7czD5zxMq6hWZ/zeJeIfCgMnwayxsOgZ6HQVhMRUzr5FRERqIIVPOaUXf9zKCz8eC5439z02pITX62X70e3M3z2febvn+QJiocLBjAvD6PH3gx3BBNmDfPeD7QXbFDy2e+18lf0VK35cAUCEfwTju45nRPMRlT/bUKdRRq/3A6vgp8fh4pcrd/8iIiI1iMKnnNTLC7by/I9bAHhoWGtu7tsUr9fLptRNzN89n/m75xeZtcJqsWLB4hs/sHDsuJTclDOq47IWl3HPWfcQERBxRu9TZlYrXPAMvDMEVnwI3W6C+M7m1CIiIlLNKXxKsV79eRvPzTeC54TzW9GzTTZTl089Yf5ch9VB7/jeDG48mP4N+hPmF0aeO4/s/GyyXFlku7J99//+ONtVcJuf7buflX9svV+uH48OeJRu8d1M+ikcp2EPaH85rPsC5j4IN3yvoZdERETKQOFTTvDawm08+8NGbIGJnNNhH18ffpHX5hyb5znAFkCf+n0Y3Ggw/Rr0O2Gu7wB7AAH2gDOaVaFwTLdOdTuV+T3K3eDHYNNsSFwC67+G9peaXZGIiEi1o/ApPm6Pm0fnfctnG+cQ3HwdVkcGazKM5wLtgfRv0J/BjQZX+vy5VUZ4A+hzLyx8EuY9DM0HQkC42VWJiIhUKwqftZzL42JZ0jJ+3P0j322bR44nDb+CA5YhjhDOTTiXwY0G0yu+V7Wdzqxc9boTVs+AI7tg9n1w2ZtmVyQiIlKtKHzWUodzDvPOunf4Zvs3pOWl+dZ784NoFXYOd/e8jHPqnYOfTYOqF+EXBJe+Ce+cD2s/g5ZDocPlZlclIiJSbSh81jJpeWm8u+5dZmyaQU5+DgCB1nDSUlqRn9GBO3oM4V9D2ppcZRWX0B363Q+LnoLvxkNCD4iomDmlRUREahqFz1oiw5nBRxs+4oMNH5DpygSgQ3QHmjku4cOfAgArdw1ozr2DW5pbaHXR737Y9iPs+wu+vg3GfANWm9lViYiIVHkKnzVctiubGZtm8N7693yn11tFtmJcl3Ek7m3MpFnrARh3nhE8K3SqyprEZodL34BpfWH3b7D4JaMzkoiIiJySwmcNlefO47PNn/HW2rdIzU0FoGl4U+7ofAeDGw1m0ZbDPPrNMgDuOLcZ/xqi4FlqdZrBBU/DN+Pgpyeg6XkafF5EROQ0FD5rGJfbxdfbvuZ/a/7HweyDACSEJnB7p9sZ1mQYNquNbQczuWvGSjxeuLJbAvcPbaXgWVZdroGtP8DGb+GrW+DWRUanJBERESmWwmcNke/J57sd3zFt9TT2Ze4DIC44jts63sbFzS/GYXUAcDTbyc3vLyMjL5+zG0fy+Mj2Cp5nwmKB4S/BnmVweAvMfwQufM7sqkRERKoshc9qzuP1MHfnXF5f/bpvnvXowGhu6XALl7e8vMhQSfluD+NmrGRXSjb1IwJ5/Zqu+NmtJlVegwRFwSWvw4eXwLK3oMVQaDnE7KpERESqJIXPasrr9fJT4k+8suoVth3dBkCkfyQ3dbiJK1pdQaA98ITX/Gf2Rn7bdpggPxtvXteN6BD/yi675mo2AM65A/54DWbdAbcvgZC6ZlclIiJS5Sh8VjNer5df9/3KKytfYWPqRgBCHaFc3/56RrcZTbAjuNjXffxnIu8t3gXA1Cs60zY+rLJKrj0GToYdC+HgBqMT0tWfGKflRURExKdM51xfffVVGjduTEBAAD169ODPP/885fZHjx5l7Nix1KtXD39/f1q2bMmcOXPKVHBtlpaXxg0/3MDYBWPZmLqRIHsQt3a8le8v+55bO9560uC5dEcKj8xcB8D4wS05v31cZZZdezgCjNmPbH6wZS789Y7ZFYmIiFQ5pT7y+emnnzJ+/HimTZtGjx49eOGFFxg6dCibN28mJibmhO2dTieDBw8mJiaGL774gvr167N7924iIiLKo/5aw+v1Mun3SSxPXo6/zZ+rW1/NDe1vICog6pSv25Oaze3TV5Dv8XJRx3rcOaB5JVVcS8W1h0GPwg8PwQ//hib9ILqF2VWJiIhUGaUOn1OnTuWWW27hhhtuAGDatGnMnj2bd955hwcffPCE7d955x1SU1NZvHgxDofR47px48ZnVnUt9Nnmz/hpz084rA4+uOAD2tY5/RSYmXn53PLBX6RmOelQP5xnL++knu2VocftsHWecQr+y5vhpvlg9zvty0RERGqDUoVPp9PJ8uXLmThxom+d1Wpl0KBBLFmypNjXfPPNN/Ts2ZOxY8cya9Ys6taty6hRo5gwYQI2W/HTEebl5ZGXl+d7nJ6eDoDL5cLlcpWm5DIp3Edl7Kskth3dxrN/PQvAXZ3vokVYi9PW5vF4uefjVWxKyqBuiB+vXt0Ju8WDy+WpjJLPWFVrg1K76GXsb/bDcmAV7p+ewHPew2ZXVGrVvg1qALVB1aB2MJ/awHwlaYOSto/F6/V6S7rj/fv3U79+fRYvXkzPnj196x944AEWLVrE0qVLT3hN69at2bVrF6NHj+aOO+5g27Zt3HHHHdx1111Mnjy52P08+uijPPbYYyesnzFjBkFBtWsAb5fXxesZr3PQc5CW9pZcG3xtiY5efpdoZf4+K3aLlzvbuWkcWgnFShH1ji6j+86X8WLh9xYTSQlpbXZJIiIiFSY7O5tRo0aRlpZGWNjJOzZXeG93j8dDTEwMb7zxBjabja5du7Jv3z6effbZk4bPiRMnMn78eN/j9PR0EhISGDJkyCk/THlxuVzMnz+fwYMH+y4VMMuUZVM4mHaQOgF1eG3Ya6e9xhPg2zUHmL9krfH6SzswsnN8RZdZ7qpSG5TdMDzfpmBdM4PeyR+Qf/EvEFB9RhmoGW1QvakNqga1g/nUBuYrSRsUnqk+nVKFz+joaGw2G8nJyUXWJycnExdXfA/qevXq4XA4ipxib9OmDUlJSTidTvz8TrwWzt/fH3//E8egdDgclfpLV9n7+7sFiQv4fOvnADzZ50liQ2NP+5rVe44y8ev1APyzX1P+cXajCq2xopndBmfswmdgzxIsR3bimPcgXPam2RWVWrVvgxpAbVA1qB3MpzYw36naoKRtU6qhlvz8/OjatSsLFizwrfN4PCxYsKDIafjj9e7dm23btuHxHLvWcMuWLdSrV6/Y4CmGpKwkJi82jgzf0O4GetXvddrXJKfncuuHf5GX72FA6xgeOF+neU3nH2oMv2SxwdrPYO0XZlckIiJiqlKP8zl+/HjefPNN3n//fTZu3Mjtt99OVlaWr/f7ddddV6RD0u23305qaip33303W7ZsYfbs2Tz55JOMHTu2/D5FDeP2uJn460TS8tJoV6cdd3a587SvyXW5ufXD5SSn59E8JoQXr+qMzaqe7VVCwtnQ737j/nfj4egec+sRERExUamv+bzyyis5dOgQkyZNIikpic6dOzN37lxiY41TwomJiVitxzJtQkICP/zwA/feey8dO3akfv363H333UyYMKH8PkUN8/a6t/kr+S+C7EE83e9pHLZTH8b2er08+OUaVu85SkSQg7eu60ZogE5LVCn97oftC2DvMvj6NhjzDViLH+1BRESkJitTh6Nx48Yxbty4Yp9buHDhCet69uzJH3/8UZZd1TqrDq7itVWvAfDvc/5No7DTX7P5+qLtzFy1H5vVwmujzqJxdPEzHYmJbHa49A14vQ/s/g0WvwR97jW7KhERkUpXpuk1pWKkO9OZ8MsE3F43Fza9kOFNh5/2NT9uSObZHzYD8OjwtvRqHl3RZUpZRTWFC5427v/0BOxfZWo5IiIiZlD4rCK8Xi+PL3mc/Vn7aRDSgId7PHza8Tw3J2Vw9ycr8XphdI+GXNuzceUUK2XX5RpoMxw8LvjqFnBmm12RiIhIpVL4rCJmbpvJ3F1zsVvsPN3vaUL8Qk65fWqWk5s/WEaW0805TaN49OJ2lVSpnBGLBYa/BKH14PAWmP+I2RWJiIhUKoXPKmBn2k6m/DkFgLFdxtKxbsdTbu9ye7hj+nL2pOaQEBXIa6O74rCpKauNoCgYaVzXy7K3YMsP5tYjIiJSiZRYTOZ0O5nwywRy8nPoUa8HN7a/8bSvefSb9fyxI5VgPxtvjzmbqGCNl1rtNBsA59xh3J81FjIPmVuPiIhIJVH4NNkLK15gY+pGIvwjeLLPk1gtp26SD5fsYvrSRCwWePGqLrSM1aTt1dbAyRDTDrIOwczb4LiJGERERGoqhU8T/br3Vz7c8CEA/+n9H2KCYk65/Zq9R3n02w0APDC0NYPann66TanCHAFw+dtgD4RtP8KSl82uSEREpMIpfJrkcM5hHv79YQBGtR5F/4T+p9ze7fHy8Mx1uD1eLuxQj9v6N62MMqWixbSBC54y7i/4P9izzNx6REREKpjCpwk8Xg8P/foQqbmptIxsyfhu40/7mhl/JrJmbxqh/nYmX9z2tMMwSTVy1hhodyl48uHLGyHnqNkViYiIVBiFTxN8sP4DlhxYQoAtgGf7PYu/zf+U2x/OzOPZuZsA+NeQlsSEBlRGmVJZLBYY/gJENoajifDtXeD1ml2ViIhIhVD4rGTrD6/nxRUvAvBA9wdoGnH60+dPfb+J9Nx82sWHcc05p59uU6qhgHC4/B2w2mHDLFj+rtkViYiIVAiFz0qU5crigV8eIN+bz+BGg7m8xeWnfc2yXal8sXwvAI+PbI9d43nWXPW7wqBHjftzJ0LyelPLERERqQhKMpXoyaVPkpiRSFxwHJN7Tj7tdZv5bg+PzFwHwNXdEzirYWRllClmOmcstBgC+bnw+fXgzDK7IhERkXKl8FlJvtvxHd9s/warxcpTfZ8i3D/8tK95b/EuNiVlEBnk4IGhrSuhSjGd1QojXz82/eb3D5hdkYiISLlS+KwEezL28J8//gPAPzv+k66xXU/7mqS0XJ6fvwWACee3JlKzGNUewdFw6ZtgscLKj2DN52ZXJCIiUm4UPiuYy+Niwi8TyHJlcVbMWdza8dYSve4/szeQ5XTTpWEEV3RLqOAqpcpp0hf6FRz1/O4eSNluajkiIiLlReGzgr2+6nXWHl5LqF8oT/V9CrvVftrX/Lb1MN+tOYDVAv8Z2R6rVWN61kr97odGvcGZCV/cAPl5ZlckIiJyxhQ+K9DaQ2t5e93bAEzuOZl6IfVO+5q8fDeTZhmdjK7r2Zh28ae/NlRqKJvdOP0eGAUHVsP8yWZXJCIicsYUPitInjuPR35/BI/XwwVNLmBo46Elet1bv+5kx+EsokP8GT+kZQVXKVVeeH2jAxLA0tdh0xxz6xERETlDCp8V5PVVr7M9bTt1AurwUPeHSvSaPanZvPzTVgAevrANYQGOiixRqotW5xtDMAHMugPS9ppbj4iIyBlQ+KwAaw+t5d31xgw1j/R8hIiAiBK97rFvN5Dr8nBO0yhGdI6vwAql2hk0Gep1hpwj8OXN4M43uyIREZEyUfgsZ3nuPB7+/WE8Xg/DmgxjYMOBJXrdgo3J/LgxGbvVwuMj2p92AHqpZez+xvSbfqGQuAQWPW12RSIiImWi8FnOXlv1GjvSdlAnoA4Tu08s0WtynG4mf2NMpXhT3ya0iA2tyBKluqrTDIa/YNz/5VnYscjUckRERMpC4bMcrT20lvfWvweU7nT7awu3sfdIDvXCA7hrQIuKK1Cqvw6XQ5drAS98dQtkHjK7IhERkVJR+CwnZT3dvvNwFv9btAOAycPbEux/+nFApZa74BmIbgWZyTDzNvB4zK5IRESkxBQ+y0lZTrd7vV4mzVqH0+2hf8u6DG0XV8FVSo3gFwT/eA/sAbDtR1jystkViYiIlJjCZzlYc2iN73T7pJ6TSny6fc7aJH7dehg/u5XHLm6nTkZScrFt4fynjPsL/g/2/mVuPSIiIiWk8HmGjh9M/sKmFzKg4YASvS4zL5/Hv9sAwO39m9E4Orgiy5SaqOv10HYkePKN6TdzjppckIiIyOkpfJ6h40+3P3j2gyV+3Ys/biEpPZeGUUHcfm6zCqxQaiyLBS5+CSIawtFE+PZu8HrNrkpEROSUFD7PQFlPt29OyuCd33cB8NiIdgQ4bBVToNR8AeFw+XtgtcOGmbD8XbMrEhEROSWFzzI6vnd7aU63e71eHpm5DrfHy9B2sZzXKqaCK5Uar0FXGDjZuD93IiSvN7ceERGRU1D4LKNXV73KzrSdRAdGl7h3O8BXK/bx565UAh02Jg1vV4EVSq3Scxw0Hwz5ufDZGMhNM7siERGRYil8lsGaQ2t4f/37AEw6ZxLh/uElel1ajosp328E4K6BLagfEVhhNUotY7XCJdMgNB5StsJXt2r8TxERqZIUPkvp76fbz2t4Xolf+9y8zRzOdNI8JoSb+jSpwCqlVgqOhqumg80ftsyFhU+aXZGIiMgJFD5Lqayn29fuTePDP3YD8H8j2uFn149eKkD9s4we8GDM/75hlrn1iIiI/I0SUCmsPrS6TKfbPR4vD89ah9cLIzrH06tZdEWWKbVdp6vgnLHG/a9vVwckERGpUhQ+S+j4weQvanpRqU63f7JsD6v3HCXU386/h7WpwCpFCgz+P2jSH1xZ8PHVkJ1qdkUiIiKAwmeJHX+6/cHuJR9MPiUzj6fnbgLg3sEtiQkLqKgSRY6x2Y353yMawdHdxgxI7nyzqxIREVH4LImynm4HeHHBVtJyXLSpF8Z1PRtVVIkiJwqKgqtmgCMIdiyEHyebXZGIiIjC5+mcyen2vHw3X6/cB8C/h7XBbtOPWypZXHsY+Zpxf8krsPoTc+sREZFaT2noNF5dWbbT7QA/bzpERm4+9cID6NWsTgVVKHIa7S6Bvv8y7n9zF+xbYW49IiJSqyl8nsKaw2t4f4Nxun1yz8mlOt0O8M1q46jnxZ3isVot5V6fSImd929oMRTcefDpNZB50OyKRESkllL4PAmX18WjfzyKx+theNPhnJtwbqlen57r4seNxn/wIzrXr4AKRUrBaoPL3oQ6zSF9H3x2HeQ7za5KRERqIYXPk1iQu4Bd6buIDoxmQvcJpX79D+uScOZ7aBETQpt6oRVQoUgpBYTDVR+DfxgkLoG5pbuMREREpDwofBZjzeE1/J73O1C20+0As1btB2Bkl/pYLDrlLlVE3ZZw6ZuABf56G5a/Z3ZFIiJSyyh8/k1ufi6P/vEoXrxc2PjCUp9uBziYnsvi7YcB43pPkSql1fkw4N/G/dn3QeIf5tYjIiK1isLn3yxLWkZiRiKhllDu73Z/md7j2zUH8Hiha6NIEqKCyrlCkXLQ9z5oczF4XPDptZC+3+yKRESkllD4/Ju+Dfry7uB3uTzocsL8wsr0Ht+sMnq5j+iso55SRVksMPJ1iGkHWQfhk9HgyjW7KhERqQUUPovRIboDzRzNyvTanYezWL03DZvVwrAO9cq5MpFy5B8CV02HgAjYvwK+uxe8XrOrEhGRGk7hs5zNKjjq2bdFNNEh/iZXI3IaUU2MOeAtVlg9A5ZOM7siERGp4RQ+y5HX6/X1ctcpd6k2mp0Hgx837v/wb9ixyNx6RESkRlP4LEdr96Wx83AWAQ4rQ9rGmV2OSMn1HAsdrwSvGz6/Ho7sMrsiERGpoRQ+y9HMlcZRz8Ft4wj2t5tcjUgpWCww/EWo1xlyUo0OSM4ss6sSEZEaSOGznLg9Xr5dUzCwvE65S3XkCDQ6IAXXheR12L67Sx2QRESk3Cl8lpMl21M4lJFHRJCDvi3qml2OSNmEN4ArPgCrHevGWbRI/s7sikREpIZR+Cwnhb3cL+xQDz+7fqxSjTXqBRc8DUCbA19gWfOpyQWJiEhNopRUDnJdbuauSwJgROf6JlcjUg663YS7641Y8GL/diwse9vsikREpIZQ+CwHP286SEZePvUjAunWKNLsckTOnMWCZ+hT7Kg72Hg8ezwsftncmkREpEZQ+CwHMwtOuQ/vFI/VajG5GpFyYrGytv41uHvdbTye9zAsfEqdkERE5IwofJ6htBwXP286BMDILurlLjWMxYLnvEdgwMPG44VTYP4kBVARESkzhc8zNHfdAZxuD61iQ2kdF2Z2OSIVo9/9MHSKcX/xSzDnPvB4zK1JRESqJYXPM+SbTlNHPaWm63mHMRA9Flj2FswaC+58s6sSEZFqRuHzDCSl5bJkRwoAF3dS+JRaoOv1cOkbYLHB6hnw5U2Q7zS7KhERqUYUPs/Ad2v24/XC2Y0jaRAZZHY5IpWj4xVwxftgdcCGmfDpNeDKNbsqERGpJhQ+z0BhL/eLNban1DZthsPVn4A9ALb+ADP+AXmZZlclIiLVgMJnGW07mMm6fenYrRYu7FDP7HJEKl+LQXDNl+AXAjt/gY8uhdw0s6sSEZEqrkzh89VXX6Vx48YEBATQo0cP/vzzzxK97pNPPsFisTBy5Miy7LZK+abgqGe/lnWJCvYzuRoRkzTuA9fNgoBw2LMU3h8OWSlmVyUiIlVYqcPnp59+yvjx45k8eTIrVqygU6dODB06lIMHD57ydbt27eK+++6jb9++ZS62qvB6vcws7OXeWR2NpJZr0A2unw1B0XBgNbx3IWQkmV2ViIhUUaUOn1OnTuWWW27hhhtuoG3btkybNo2goCDeeeedk77G7XYzevRoHnvsMZo2bXpGBVcFq/YcJTE1myA/G4Pbxppdjoj54jrADd9DaD04tBHevQCO7jG7KhERqYLspdnY6XSyfPlyJk6c6FtntVoZNGgQS5YsOenr/u///o+YmBhuuukmfv3119PuJy8vj7y8PN/j9PR0AFwuFy6XqzQll0nhPk62r69W7AVgUOsYHBZvpdRU25yuDaTilboNIprAtd9in34pltQdeN85n/zRX0FU9f+D0yz6HlQNagfzqQ3MV5I2KGn7lCp8Hj58GLfbTWxs0aN9sbGxbNq0qdjX/Pbbb7z99tusWrWqxPuZMmUKjz322Anr582bR1BQ5Q1pNH/+/BPWub3w9XIbYCHOuZc5c3R0pyIV1wZSuUrbBgENxtM79ylC0vfifnMwi5tPICOwQQVVVzvoe1A1qB3MpzYw36naIDs7u0TvUarwWVoZGRlce+21vPnmm0RHR5f4dRMnTmT8+PG+x+np6SQkJDBkyBDCwip+CkuXy8X8+fMZPHgwDoejyHO/bj1M5h8riAxycM9Vg3DYNGBARThVG0jlOKM2yByC9+PLCTi4gfN2/5f8qz+Hep0qptAaTN+DqkHtYD61gflK0gaFZ6pPp1ThMzo6GpvNRnJycpH1ycnJxMXFnbD99u3b2bVrF8OHD/et8xTMB22329m8eTPNmjU74XX+/v74+/ufsN7hcFTqL11x+/tunfHZL+oYT1DAiTVK+arsNpcTlakNIusbnZA+ugzL/hU4pl8Co7+Ahj0qpsgaTt+DqkHtYD61gflO1QYlbZtSHbbz8/Oja9euLFiwwLfO4/GwYMECevbsecL2rVu3Zu3ataxatcq3XHzxxZx33nmsWrWKhISE0uzedDlONz+sM3rxjtRc7iKnFhRlDMPUsBfkpcOHl8CmOWZXJSIiJiv1affx48czZswYunXrRvfu3XnhhRfIysrihhtuAOC6666jfv36TJkyhYCAANq3b1/k9REREQAnrK8OFmxKJsvppkFkIGc1jDS7HJGqLyDMGIj+09Gw/Sf45GpoeT6cP0UdkUREaqlSh88rr7ySQ4cOMWnSJJKSkujcuTNz5871dUJKTEzEaq2Z10HOXHlsbE+LxWJyNSLVhF+QMRXnT4/DH6/DlrlGEO11J/T9F/gFm12hiIhUojJ1OBo3bhzjxo0r9rmFCxee8rXvvfdeWXZpuqPZThZtMQbSH6G53EVKx+4PQ/4DXa6DuROM8Pnrc7D6ExjyOLS7FPQHnYhIrVAzD1FWgDlrk3C5vbSOC6VlbKjZ5YhUT3VbwjVfwZXTIaIhpO+DL240puVMXm92dSIiUgkUPktoVsFc7iO76KinyBmxWKDNRTD2Tzh3ItgDYNevMK0vfD8Bco6aXaGIiFQghc8S2H80h6U7UwEY3km93EXKhSMQzn3QCKFthoPXDUunwctdYcUHUDAsm4iI1CwKnyXw7Wqjo1H3JlHUjwg0uRqRGiayEVz5EVw7E6JbQfZh+OZOeGsg7F1udnUiIlLOFD5LYOaqY73cRaSCNDsPbv8dhjwBfqGwfwW8NQBmjoXMg2ZXJyIi5UTh8zS2JGew8UA6DpuFYe3rmV2OSM1mc0CvcXDncug0yli36iPjVPyS18DtMrc+ERE5Ywqfp1HY0ah/y7pEBvuZXI1ILREaC5e8DjfNh3qdjRmSfphodErascjs6kRE5AwofJ6C1+tllu+Uu3q5i1S6hO5wy08w/EUIqgOHNsIHF8NnY+DoHrOrExGRMlD4PIWVe9LYeySHYD8bg9rEml2OSO1ktUHX641T8d1vBYsVNsyEV7vDH9PUK15EpJpR+DyFb9ccAGBouzgC/WwmVyNSywVGwrBn4Z+/QsNe4Mo2Zkt69wI4vM3s6kREpIQUPk/C7YHZa5MAuFi93EWqjrj2cP1suHAq+IXAnj9gWm/4/SXwuM2uTkRETkPh8yQ2p1k4ku2iTrAffZpHm12OiBzPaoWzb4I7/oBmAyA/F+Y/Am8PhoObzK5OREROQeHzJP46bAHgoo71sNv0YxKpkiISjLniL34F/MNh33L4X1/45b8alklEpIpSqipGtjOftalG+ByhudxFqjaLBc66Fsb+AS3PB7cTfnrcmCEpaa3Z1YmIyN8ofBZjwaZDOD0WEiID6ZIQYXY5IlISYfFw9SdwyRsQEAEHVsMb58LPUyDfaXZ1IiJSQOGzGIW93Id3rIfFYjG5GhEpMYsFOl0JY/+ENsPBkw+LnjJC6P6VZlcnIiIofJ4gNcvJr1tTABjeMc7kakSkTEJj4YoP4R/vQVA0HFwPbw6EHx8DV67Z1YmI1GoKn3+zKyWL6BA/GgR7aR4TYnY5IlJWFgu0uwTGLoX2l4HXDb9Nhf/1gz3LzK5ORKTWUvj8m7MaRrLoX/24tbXGCxSpEYKj4fJ34MrpEBILhzcbQzL98G9wZptdnYhIraPwWQyr1UK4n9lViEi5anORMS5op1GAF5a8YgxOv3ux2ZWJiNQqCp8iUnsERcElr8OozyGsPqTuMKbnnHM/5GWaXZ2ISK2g8CkitU/LIXDHEjhrjPH4zzdgWh9I/MPcukREagGFTxGpnQLC4eKX4NqZEJ4AR3YaR0HnT4b8PLOrExGpsRQ+RaR2a3Ye3P47dB4NXg/8/gK8cZ5mRxIRqSAKnyIiAeEw8jWjR3zhuKBvnAe/TgWPRr4QESlPCp8iIoUKe8S3vgg8LljwmHEqPmW72ZWJiNQYCp8iIscLqQtXfgQjXwf/MNiz1OiMtOxt8HrNrk5EpNpT+BQR+TuLBTqPgtsXQ5N+4MqG2ePho8sgfb/Z1YmIVGsKnyIiJxORANfOgvOfBnsAbF8Ar50Da78wuzIRkWpL4VNE5FSsVjjnNvjnLxDfBXLT4Mub4PPrITvV7OpERKodhU8RkZKo2wpumg/nTgSLDdZ/bRwF3TLP7MpERKoVhU8RkZKyOeDcB+HmHyG6JWQmw4x/wLd3a3pOEZESUvgUESmt+mcZp+HPGWs8Xv4eTOsNu5eYWpaISHWg8CkiUhaOQDj/SRjzbcH0nLuMMUHnPaLpOUVETkHhU0TkTDTpVzA95zWAFxa/BM+3h09Gwy//he0/Qc4Rs6sUEaky7GYXICJS7QWEw8hXofUw4/rPrIOw6TtjKRTVFOLPMk7Zx58F9TqCX7B5NYuImEThU0SkvLS+EJoNgP0rYd9y2LcC9q8wTsmn7jCWdQVjhFqsULcN1O9yLJTGtAO7n6kfQUSkoil8ioiUJ0cgNOplLIWyU40Qum9lwe0KyEyCg+uNZeVHxnY2f4hrf1wY7QhejzmfQ0Skgih8iohUtKAoaD7IWAql7z92ZHTfCuNoae7RgiOmy2EZOIBh1gBsOQOh1QXQYgiExpr1KUREyoXCp4iIGcLijaXNRcZjr9c4Lb9/pS+Ueg+sxuHKhs2zjQWMWZZanm8E0XqdjRmYRESqEYVPEZGqwGKBOs2MpcPlAOTn5bD4q2n0icnGtn2+EUwLl4VTICTWCKEth0LTc8E/1NzPICJSAgqfIiJVldXO0aCmePoNwzbw35CRDFvnwdYfYPvPxgxLKz80FpsfNO4DLYYaYTSqidnVi4gUS+FTRKS6CI2Fs641lvw82P27Mbf8lrlwZKcxpuj2n2DuBIhuBS2HGKfoE3oYU4OKiFQBCp8iItWR3d8Y1qnZADh/CqRsM0Lolh8gcQkc3mwsi18G/3BoPtAIoi2HQGCk2dWLSC2m8CkiUt1ZLBDdwlh63Qk5R40joFvnGUt2Cqz/ylgcQdB5FPS4HaKbm125iNRCCp8iIjVNYAS0v9RYPG6j9/yWucaMS4c2wbK3YNnbxpHQnndA475GgBURqQQao0NEpCaz2iDhbBj4CNzxB4z5FlpeAHhhy/fw/nD4X19Y/QnkO82uVkRqAYVPEZHawmKBJv1g1CcwbjmcfTPYAyFpLXz9T3ihA/zyX2NGJhGRCqLwKSJSG0U3hwufg/EbYOBkCK1nTPn50+MwtS18dy8c3mp2lSJSAyl8iojUZkFR0Hc83L0GLnkD6nWC/Bz46x14pRtMvwJ2LDRmYBIRKQcKnyIiAnY/6HQl3LoIrp8NrS4ELMaA9h+MgGl9YOV0Y3xREZEzUGN6u3s8HpzO8rlY3uVyYbfbyc3Nxe12l8t7SulUVBv4+flh1VzYIidnsRgzJTXuAynbYek0WPkRJK+DWXfAj49C91ug240QHG12tSJSDdWI8Ol0Otm5cycej6dc3s/r9RIXF8eePXuwaPgRU1RUG1itVpo0aYKfn1+5vadIjVWnGQx7Fs57CJa/D0v/Bxn74ecn4NfnoOOV0O0GiO9idqUiUo1U+/Dp9Xo5cOAANpuNhISEcjmq5fF4yMzMJCQkREfJTFIRbeDxeNi/fz8HDhygYcOG+sNCpKQCI6HPPdBzLGyYBUtegf0rYcX7xhLXEbqOgQ7/gIBws6sVkSqu2ofP/Px8srOziY+PJygoqFzes/AUfkBAgMKnSSqqDerWrcv+/fvJz8/H4dBc1yKlYnNAh8uh/WXGFJ7L3oaN30DSGpj9L5j3CLS7BM4aAwndNXC9iBSr2ofPwusBdRpVSqLw98Ttdit8ipSVxQKNehlLdqoxQP2K943Zk1ZNN5a6rY0Q2ukqo0e9iEiBGnNYT6dQpST0eyJSzoKijCk67/gDbpwHnUcbA9cf2gQ/TITnWsEXN8KORVBO1+WLSPVW7Y98iohIFWCxQMMexnL+FFj7hXE09MBqWPelsUQ2gbOuMwJqaKzZFYuISWrMkU8REakiAsLh7Jvgn78Y44Z2uxH8QuHITljwGDzfFj4ZDVvng0fD2YnUNjryKSIiFSe+s7EM+Q+s/9oYsmnvn7DpO2MJawBdrjGWiASzqxWRSqDwKT4ul0udcESkYvgFHwuZBzfCig9g9ceQvhcWPQWLnoa6rSCqacHS5Nj98ASw2sz+BCJSTnTa3URz586lT58+REREUKdOHS666CK2b9/ue37v3r1cffXVREVFERwcTLdu3Vi6dKnv+W+//Zazzz6bgIAAoqOjueSSS3zPWSwWZs6cWWR/ERERvPfeewDs2rULi8XCp59+Sv/+/QkICGD69OmkpKRw9dVXU79+fYKCgujQoQMff/xxkffxeDw888wzNG/eHH9/fxo2bMgTTzwBwIABAxg3blyR7Q8dOoSfnx8LFiwojx+biFR3MW2M60LHb4LL3obGfQGv0Ulp8xxjHNHZ/4IPL4EXO8F/YuHlrsY8898/CEvfgK0/GjMwuV1mfxoRKaUad+TT6/WS4zqza4g8Hg85Tjd2Z36pxpgMdNhK1Zs6KyuL8ePH07FjRzIzM5k0aRKXXHIJq1atIjs7m/79+1O/fn2++eYb4uLiWLFihW8Wp9mzZ3PJJZfw73//mw8++ACn08mcOXNK/VkffPBBnnvuObp06UJAQAC5ubl07dqVCRMmEBYWxuzZs7n22mtp1qwZ3bt3B2DixIm8+eabPP/88/Tp04cDBw6wadMmAG6++WbGjRvHc889h7+/PwAfffQR9evXZ8CAAaWuT0RqMEeAMW5oh8shfb8RPlN3QOrOgtuC++48SNlmLH9nsUFEw+OOmBYs0S2MW41wIVLl1LjwmeNy03bSD6bse8P/DSXIr+Q/0ssuu6zI43feeYe6deuyYcMGFi9ezKFDh1i2bBlRUcYYec2bN/dt+8QTT3DVVVfx2GOP+dZ16tSp1DXfc889XHrppUXW3Xfffb77d955Jz/88AOfffYZ3bt3JyMjgxdffJFXXnmFMWPGANCsWTP69OkDwKWXXsq4ceOYNWsWV1xxBQDvvfce119/vYY5EpGTC4s3lmZ/+yPV44H0fceF0R1FA2p+jtGR6chO2P63syuBkVC/GzQoWOp3NdaJiKlqXPisTrZu3cqkSZNYunQphw8f9h3VTExMZNWqVXTp0sUXPP9u1apV3HLLLWdcQ7du3Yo8drvdPPnkk3z22Wfs27cPp9NJXl6eb/aojRs3kpeXx8CBA4t9v4CAAK699lreeecdrrjiClasWMG6dev45ptvzrhWEamFrFajI1JEAjTtX/Q5rxcykooJptvh0BbIOQLb5htLoTotoMHZ0KCrcRvTDmz6r1CkMtW4b1ygw8aG/xt6Ru/h8XjISM8gNCy01KfdS2P48OE0atSIN998k/j4eDweD+3bt8fpdBIYGHjqfZ3meYvFgtfrLbLO5Trx2qjg4OAij5999llefPFFXnjhBTp06EBwcDD33HMPTqezRPsF49R7586d2bt3L++++y4DBgygUaNGp32diEipWCwQVs9YGvcu+ly+E5LXwd6/YN9fsHeZEUxTthrL6hnGdvZAiO9yLIzW7wbh9Sv/s4jUImUKn6+++irPPvssSUlJdOrUiZdfftl3PeDfvfnmm3zwwQesW7cOgK5du/Lkk0+edPszZbFYSnXquzgej4d8PxtBfvYKm9s9JSWFzZs38+abb9K3b18AfvvtN9/zHTt25K233iI1NbXYo58dO3ZkwYIF3HDDDcW+f926dTlw4IDv8datW8nOzj5tXb///jsjRozgmmuuAYyfxZYtW2jbti0ALVq0IDAwkAULFnDzzTcX+x4dOnSgW7duvPnmm8yYMYNXXnnltPsVESlXdj+of5axcKuxLisF9i03gui+v2DvcshLg8TFxlIoNP5YGG1wNtRtZ8pHEKmpSp3SPv30U8aPH8+0adPo0aMHL7zwAkOHDmXz5s3ExMScsP3ChQu5+uqr6dWrFwEBATz99NMMGTKE9evXU79+7f3rMjIykjp16vDGG29Qr149EhMTefDBB33PX3311Tz55JOMHDmSKVOmUK9ePVauXEl8fDw9e/Zk8uTJDBw4kGbNmnHVVVeRn5/PnDlzmDBhAmD0On/llVfo2bMnbrebCRMmlGgYpRYtWvDFF1+wePFiIiMjmTp1KsnJyb7wGRAQwIQJE3jggQfw8/Ojd+/eHDp0iPXr13PTTTf53qew41FwcHCRXvgiIqYJrgMthxgLGNeTpmw7Lowug+T1kLEfNu6Hjd8CYLdYucAaiH1HHQgIA//QYpbi1v9tnSPYuIxApJYrdficOnUqt9xyi++I27Rp05g9ezbvvPNOkfBUaPr06UUev/XWW3z55ZcsWLCA6667roxlV39Wq5VPPvmEu+66i/bt29OqVSteeuklzj33XAD8/PyYN28e//rXvxg2bBj5+fm0bduWV199FYBzzz2Xzz//nMcff5ynnnqKsLAw+vXr53v/5557jhtuuIG+ffsSHx/Piy++yPLly09b18MPP8yOHTsYOnQoQUFB3HrrrYwcOZK0tDTfNo888gh2u51Jkyaxf/9+6tWrx2233Vbkfa6++mruuecerr76agICAsrhJyYiUs6sVqjb0li6jDbWObNg/6rjAulfWDIO4OfOgrQsSDvlO56GxQihIbEQ2xZi20NMW4htBxGNFEyl1ihV+HQ6nSxfvpyJEyf61lmtVgYNGsSSJUtK9B7Z2dm4XK6TdqQByMvLIy8vz/c4PT0dMK5Z/Pt1iy6XC6/Xi8fj8XXYOVOF10oWvm9FGTBggO9yhEJutzFMlMfjISEhgc8+++yE1xXWNHLkSEaOHFnsc3FxcXz//fdFnktNTfVt07BhwyL7KhQREcFXX31VbL3Hbzdx4sQivwd/f/7gwYPk5uZyww03lOlnWFFt4PF48Hq9uFwubDYNWn0qhd+14q4VlsqhNjCBxQ/qdzeWAq6j+/ljwWx6deuA3Z0DeRmQl4HFmXHsfl4GODMgL/OE58jLwOJ1A17ISzeWlK2wYZZvH16/YLx12+CNaQsx7fDGtMFbty0ERlT+z6AK0nfBfCVpg5K2T6nC5+HDh3G73cTGxhZZHxsb6xvn8XQmTJhAfHw8gwYNOuk2U6ZMKTKEUKF58+b5el0XstvtxMXFkZmZ6esUU14yMjLK9f1qA5fLRWpqKo888gjdunWjefPmvj8eyqK828DpdJKTk8Mvv/xCfn5+ub53TTV//vzTbyQVSm1QBQTW54f1qQUP7EBkwVLAAvgXLH/n9WLzOrG7c7G7swlyHiYsZy9huYmE5ewlNHcfNmcWln0FnaOOk+OIIi0wgfTABNIDGpAe2JDMgDi8lhrXX7hE9F0w36naoCR9S6CSe7s/9dRTfPLJJyxcuPCUp2InTpzI+PHjfY/T09NJSEhgyJAhhIWFFdk2NzeXPXv2EBISUm6nd71eLxkZGYSGhmpsylJauHAhAwcOpGXLlnz22WcntFdJVVQb5ObmEhgYSL9+/XQ5wGm4XC7mz5/P4MGDNe2qSdQGVUNFt4PH7cKTugPLoQ1YkjdgObgey6GNWNL2EOhKJdCVSlz6at/2XqsDolvijWmLN7Yd3oa98cZ1rNFTkOq7YL6StEFJDzaVKnxGR0djs9lITk4usj45OZm4uLhTvva///0vTz31FD/++CMdO3Y85bb+/v6+2XGO53A4TvjAbrcbi8WC1Wott57phad5C99XSm7AgAEnDPFUFhXVBlarFYvFUuzvkhRPPyvzqQ2qhgprB4cD4tsby/FyjsLBjXBwvdERKnkDJK83TukfXI/l4HpY97mxbWAkND0Xmp4Hzc4zZn2qgfRdMN+p2qCkbVOq8Onn50fXrl1ZsGCB71pDj8fDggULTpjP+3jPPPMMTzzxBD/88MMJg5qLiIhIMQIjoFFPYynk9cLRRDi4wRjHdN9K2PWrMaD++q+NBaBO84IgOgAa9zF66YtUEaU+7T5+/HjGjBlDt27d6N69Oy+88AJZWVm+3u/XXXcd9evXZ8qUKQA8/fTTTJo0iRkzZtC4cWOSkpIACAkJISQkpBw/ioiISA1nsUBkI2NpdYGxzp1vjF+642fY/pMxsH7KNmNZ9iZY7cZ4pYVhNL6LZnUSU5X6t+/KK6/k0KFDTJo0iaSkJDp37szcuXN9nZASExOLnCZ9/fXXcTqdXH755UXeZ/LkyTz66KNnVr2IiEhtZ7NDwx7Gcu6DkJsGO389FkZTd0DiEmNZ+CT4h0PTfsfCaFQTsz+B1DJl+tNn3LhxJz3NvnDhwiKPd+3aVZZdiIiISFkEhEObi4wF4Mgu2P6zEUZ3LDTC6cZvfYPoE9n4uCDaFDz5xxa3q+C+Gzyu49ad4rEn3zga6/WAzQF2f7AHgM3PuF94a/cHm//p19n8TPpBSkXRcXcREZGaLLIxdLvBWDxuYxD97T8ZYXTPUiOcLn/XWKoou82fAfYobHlfQb1OENfBWEJijUsRpFpR+BQREaktrLaCeeu7Qv/7jUHwd/1uhNGdiyA71bhG1GY3bq12sDqM19kcRR9b7cetsx/3uOA5ixXcTsh3Qn5uwf28427zjNsi6wq29RQdh9niziPUfQA2zDSWQsF1jwXR2ILbOs11TWsVp9Yxybnnnkvnzp154YUXzC5FRERqK/9QaHW+sVQlHs+xcOp24spJZ9ncT+jRKBDbwQ2QtNaYJSrrkBGct/907LX2AGPa0sJQGtfBmMLUP9S8zyNFKHyKiIhI1WK1gjUQHIHGY/9IDoV1wNNzGLbCsSSd2cY4qMlrjTCatBaS1oErC/avMJbjRTWF2PYQ1xFi2kB4fQiNN46eakzvSqXwKSIiItWPX9CxSwgKeTxwZOdxYbRgydhv9PpP3QEbvyn6PlY7hNYzlrB4Yzn+fuFje3Fzp0pZKHxWAUeOHOHuu+/m22+/JS8vj/79+/PSSy/RokULAHbv3s24ceP47bffcDqdNG7cmGeffZZhw4Zx5MgRxo0bx7x588jMzKRBgwY89NBDvnFXRUREag2rFeo0M5Z2I4+tz0opeoT08BZIPwCZycb1pWl7jOVUguoUBNF4CKsHYfULQmo9CIgAv2BwBIFfiBGM7QHqDHUSNS98er3gKtnE9ifl8Rjv4bSV7lC8I6hMv2jXX389W7du5ZtvviEsLIwJEyYwbNgwNmzYgMPhYOzYsTidTn755ReCg4PZsGGDb4D+Rx55hA0bNvD9998THR3Ntm3byMnJKXUNIiIiNVZwnYLpR88tut6dbwTQ9P3G0dH0giXjQNH7+bmQnWIsSWtLtk+LFRzBRij1CyoIp8c/DjkxsPoFQ3CM0aM/vEGNDa81L3y6suHJ+DN6CysQUZYXPrTf+MUphcLQ+fvvv9OrVy8Apk+fTkJCAjNnzuQf//gHiYmJXHbZZXTo0AGApk2b+l6fmJhIly5dfNOWNm7cuCyVi4iI1D42u3HtZ3j9k2/j9RrTl/pC6T7jqGn6PuNxRhLkpYMzy7gONb/gAJDXA84MYymLoDpGCK3X2biN7wwRjWpEIK154bOa2bhxI3a7nR49evjW1alTh1atWrFx40YA7rrrLm6//XbmzZvHoEGDuOyyy+jYsSMAt99+O5dddhkrVqxgyJAhjBw50hdiRURE5AxZLBAUZSxx7U+/vcddcPY0q+jiyjoWUJ2ZxW/jzIK0RKMjVXbKiT35AyKOBdHCYBrVtNoF0poXPh1BxhHIM+DxeEjPyCAsNLTIVKEl2ncFuPnmmxk6dCizZ89m3rx5TJkyheeee44777yTCy64gN27dzNnzhzmz5/PwIEDGTt2LP/9738rpBYRERE5BavNGNbpTIZ2cuXCwfVwYLUxKcCBVZC8AXKPGuOx7lx0bFv/cKjXsSCUdjFuo5pV6R78NS98WiylPvV9Ao8HHG7jfSq48dq0aUN+fj5Lly71HbFMSUlh8+bNtG3b1rddQkICt912G7fddhsTJ07kzTff5M477wSgbt26jBkzhjFjxtC3b1/uv/9+hU8REZHqyhEA9bsaS6F8JxzcYATR/auMYJq8HvLSYNevxlLIL8QYUiq+szFtaovBlfwBTq3mhc9qpkWLFowYMYJbbrmF//3vf4SGhvLggw9Sv359RowYAcA999zDBRdcQMuWLTly5Ag///wzbdq0AWDSpEl07dqVdu3akZeXx3fffed7TkRERGoIu58RJuM7Q2EmdbuMU/QHVh8LpcnrjNP6iYuNxZmp8Cknevfdd7n77ru56KKLcDqd9OvXjzlz5uAoGEjX7XYzduxY9u7dS1hYGOeffz7PP/88AH5+fkycOJFdu3YRGBhI3759+eSTT8z8OCIiIlIZbI6CU+4dgWuNde58OLz52NHRpv3NrLBYCp8mWbhwoe9+ZGQkH3zwwUm3ffnll0/63MMPP8zDDz9cnqWJiIhIdWWzG9OJxraDLqPNrqZYVfdqVBERERGpcRQ+RURERKTSKHyKiIiISKVR+BQRERGRSqPwKSIiIiKVRuFTRERERCqNwqeIiIiIVBqFTxERERGpNAqfIiIiIlJpFD6rscaNG/PCCy+YXYaIiIhIiSl8ioiIiEilUfgUU7jdbjwej9lliIiISCVT+DTJG2+8QXx8/AkBbMSIEdx4441s376dESNGEBsbS0hICGeffTY//vhjmfc3depUOnToQHBwMAkJCdxxxx1kZmYW2eb333/n3HPPJSgoiMjISIYOHcqRI0cA8Hg8PPPMMzRv3hx/f38aNmzIE088AcDChQuxWCwcPXrU916rVq3CYrGwa9cuAN577z0iIiL45ptvaNu2Lf7+/iQmJrJs2TIGDx5MdHQ04eHh9O/fnxUrVhSp6+jRo/zzn/8kNjaWgIAA2rdvz3fffUdWVhZhYWF88cUXRbafOXMmwcHBZGRklPnnJSIiIhWjxoVPr9dLtiv7jJec/JxSv8br9Za4zn/84x+kpKTw888/+9alpqYyd+5cRo8eTWZmJsOGDWPBggWsXLmS888/n+HDh5OYmFimn4vVauWll15i/fr1vP/++/z000888MADvudXrVrFwIEDadu2LUuWLOG3335j+PDhuN1uACZOnMhTTz3FI488woYNG5gxYwaxsbGlqiE7O5unn36at956i/Xr1xMTE0NGRgZjxozht99+448//qBFixYMGzbMFxw9Hg8XXHABv//+Ox999BEbNmzgqaeewmazERwczFVXXcW7775bZD/vvvsul19+OaGhoWX6WYmIiEjFsZtdQHnLyc+hx4wepux76ailBDmCSrRtZGQkF1xwATNmzGDgwIEAfPHFF0RHR3PeeedhtVrp1KmTb/vHH3+cr7/+mm+++YZx48aVurZ77rnHd79x48b85z//4bbbbuO1114D4JlnnqFbt26+xwDt2rUDICMjgxdffJFXXnmFMWPGANCsWTP69OlTqhpcLhevvfZakc81YMCAItu88cYbREREsGjRIvr168ePP/7In3/+ycaNG2nZsiUATZs29W1/880306tXLw4cOEC9evU4ePAgc+bMOaOjxCIiIlJxatyRz+pk9OjRfPnll+Tl5QEwffp0rrrqKqxWK5mZmdx33320adOGiIgIQkJC2LhxY5mPfP74448MHDiQ+vXrExoayrXXXktKSgrZ2dnAsSOfxdm4cSN5eXknfb6k/Pz86NixY5F1ycnJ3HLLLbRo0YLw8HDCwsLIzMxkz549AKxevZoGDRr4guffde/enXbt2vH+++8D8NFHH9GoUSP69et3RrWKiIhIxahxRz4D7YEsHbX0jN7D4/GQkZFBaGgoVmvJ83mgPbBU+xk+fDher5fZs2dz9tln8+uvv/L8888DcN999zF//nz++9//0rx5cwIDA7n88stxOp2l2gfArl27uOiii7j99tt54okniIqK4rfffuOmm27C6XQSFBREYODJaz/Vc4DvZ3T8ZQcul6vY97FYLEXWjRkzhpSUFF588UUaNWqEv78/PXv29H3O0+0bjKOfr776Kg8++CDvvvsuN9xwwwn7ERERkaqhxoVPi8VS4lPfJ+PxeMi35xPkCCpV+CytgIAALr30UqZPn862bdto1aoVZ511FmB0/rn++uu55JJLAMjMzPR13imt5cuX4/F4eO6553yf57PPPiuyTceOHVmwYAGPPfbYCa9v0aIFgYGBLFiwgJtvvvmE5+vWrQvAgQMHiIyMBIwjqSXx+++/89prrzFs2DAA9uzZw+HDh33Pd+jQgb1797Jly5aTHv285ppreOCBB3jppZfYsGGD79IAERERqXp02t1ko0ePZvbs2bzzzjuMHj3at75FixZ89dVXrFq1itWrVzNq1KgyD03UvHlzXC4XL7/8Mjt27ODDDz9k2rRpRbaZOHEiy5Yt44477mDNmjVs2rSJ119/ncOHDxMQEMCECRN44IEH+OCDD9i+fTt//PEHb7/9tu/9ExISePTRR9m6dSuzZ8/mueeeK1FtLVq04MMPP2Tjxo0sXbqU0aNHFzna2b9/f/r168dll13G/Pnz2blzJ99//z1z5871bRMZGcmll17K/fffz5AhQ2jQoEGZfk4iIiJS8RQ+TTZgwACioqLYvHkzo0aN8q2fOnUqkZGR9OrVi+HDhzN06FDfUdHS6tSpE1OnTuXpp5+mffv2TJ8+nSlTphTZpmXLlsybN4/Vq1fTvXt3evbsyaxZs7DbjYPjjzzyCP/617+YNGkSbdq04corr+TgwYMAOBwOPv74YzZt2kTHjh15+umn+c9//lOi2t5++22OHDnCWWedxbXXXstdd91FTExMkW2+/PJLzj77bK6++mratm3LAw884OuFX6jwEoIbb7yxTD8jERERqRwWb2nGBzJJeno64eHhpKWlERYWVuS53Nxcdu7cSZMmTQgICCiX/Xk8HtLT0wkLC6vQ0+5ycqVtgw8//JB7772X/fv34+fnd9LtKuL3paZyuVzMmTOHYcOG4XA4zC6nVlIbVA1qB/OpDcxXkjY4VV47Xo275lNql+zsbA4cOMBTTz3FP//5z1MGTxERETGfDuvVANOnTyckJKTYpXCszprqmWeeoXXr1sTFxTFx4kSzyxEREZHT0JHPGuDiiy+mR4/iB9av6acnHn30UR599FGzyxAREZESUvisAUJDQzWVpIiIiFQLOu0uIiIiIpVG4VNEREREKo3Cp4iIiIhUGoVPEREREak0Cp8iIiIiUmkUPquxxo0b88ILL5RoW4vFwsyZMyu0HhEREZHTUfgUERERkUqj8CkiIiIilUbh0yRvvPEG8fHxeDyeIutHjBjBjTfeyPbt2xkxYgSxsbGEhIRw9tln8+OPP5bb/teuXcuAAQMIDAykTp063HrrrWRmZvqeX7hwId27dyc4OJiIiAh69+7N7t27AVi9ejXnnXceoaGhhIWF0bVrV/76669yq01ERERqrhoXPr1eL57s7DNfcnJK/Rqv11viOv/xj3+QkpLCzz//7FuXmprK3LlzGT16NJmZmQwbNowFCxawcuVKzj//fIYPH05iYuIZ/4yysrIYOnQokZGRLFu2jM8//5wff/yRcePGAZCfn8/IkSPp378/a9asYcmSJdx6661YLBYARo8eTYMGDVi2bBnLly/nwQcfrPHTeIqIiEj5qHHTa3pzcth8Vtdyea/kUm7fasVyLEFBJdo2MjKSCy64gBkzZjBw4EAAvvjiC6KjoznvvPOwWq106tTJt/3jjz/O119/zTfffOMLiWU1Y8YMcnNz+eCDDwgODgbglVdeYfjw4Tz99NM4HA7S0tK46KKLaNasGQBt2rTxvT4xMZH777+f1q1bA9CiRYszqkdERERqjxp35LM6GT16NF9++SV5eXkATJ8+nauuugqr1UpmZib33Xcfbdq0ISIigpCQEDZu3FguRz43btxIp06dfMEToHfv3ng8HjZv3kxUVBTXX389Q4cOZfjw4bz44oscOHDAt+348eO5+eabGTRoEE899RTbt28/45pERESkdqhxRz4tgYG0WrH8jN7D4/GQnpFBWGgoVmvJ87klMLBU+xk+fDher5fZs2dz9tln8+uvv/L8888DcN999zF//nz++9//0rx5cwIDA7n88stxOp2l2kdZvfvuu9x1113MnTuXTz/9lIcffpj58+dzzjnn8OijjzJq1Chmz57N999/z+TJk/nkk0+45JJLKqU2ERERqb5qXvi0WEp86vukPB6s+flYg4JKFT5LKyAggEsvvZTp06ezbds2WrVqxVlnnQXA77//zvXXX+8LdJmZmezatatc9tumTRvee+89srKyfEc/f//9d6xWK61atfJt16VLF7p06cLEiRPp2bMnM2bM4JxzzgGgZcuWtGzZknvvvZerr76ad999V+FTRERETkun3U02evRoZs+ezTvvvMPo0aN961u0aMFXX33FqlWrWL16NaNGjTqhZ/yZ7DMgIIAxY8awbt06fv75Z+68806uvfZaYmNj2blzJxMnTmTJkiXs3r2befPmsXXrVtq0aUNOTg7jxo1j4cKF7N69m99//51ly5YVuSZURERE5GRq3JHP6mbAgAFERUWxefNmRo0a5Vs/depUbrzxRnr16kV0dDQTJkwgPT29XPYZFBTEDz/8wN13383ZZ59NUFAQl112GVOnTvU9v2nTJt5//31SUlKoV68eY8eO5Z///Cf5+fmkpKRw3XXXkZycTHR0NJdeeimPPfZYudQmIiIiNZvCp8msViv79+8/YX3jxo356aefiqwbO3ZskcelOQ3/92GgOnTocML7F4qNjeXrr78u9jk/Pz8+/vjjEu9XRERE5Hg67S4iIiIilUbhswaYPn06ISEhxS7t2rUzuzwRERERH512rwEuvvhievToUexzmnlIREREqhKFzxogNDSU0NBQs8sQEREROS2ddhcRERGRSlNjwuffe3OLFEe/JyIiIuaq9qfdHQ4HFouFQ4cOUbduXSwWyxm/p8fjwel0kpubW6EzHMnJVUQbeL1eDh06hMVi0bWwIiIiJqn24dNms9GgQQP27t1bbtNPer1ecnJyCAwMLJcwK6VXUW1gsVho0KABNput3N5TRERESq7ah0+AkJAQWrRogcvlKpf3c7lc/PLLL/Tr109HyExSUW3gcDgUPEVERExUI8InGEdAyytU2Gw28vPzCQgIUPg0idpARESkZirTxXSvvvoqjRs3JiAggB49evDnn3+ecvvPP/+c1q1bExAQQIcOHZgzZ06ZihURERGR6q3U4fPTTz9l/PjxTJ48mRUrVtCpUyeGDh3KwYMHi91+8eLFXH311dx0002sXLmSkSNHMnLkSNatW3fGxYuIiIhI9VLq8Dl16lRuueUWbrjhBtq2bcu0adMICgrinXfeKXb7F198kfPPP5/777+fNm3a8Pjjj3PWWWfxyiuvnHHxIiIiIlK9lOqaT6fTyfLly5k4caJvndVqZdCgQSxZsqTY1yxZsoTx48cXWTd06FBmzpx50v3k5eWRl5fne5yWlgZAampquXUqOhWXy0V2djYpKSm63tAkagPzqQ3MpzaoGtQO5lMbmK8kbZCRkQGcfkztUoXPw4cP43a7iY2NLbI+NjaWTZs2FfuapKSkYrdPSko66X6mTJnCY489dsL6Jk2alKZcEREREalkGRkZhIeHn/T5KtnbfeLEiUWOlno8HlJTU6lTp06ljLuZnp5OQkICe/bsISwsrML3JydSG5hPbWA+tUHVoHYwn9rAfCVpA6/XS0ZGBvHx8ad8r1KFz+joaGw2G8nJyUXWJycnExcXV+xr4uLiSrU9gL+/P/7+/kXWRURElKbUchEWFqZfcpOpDcynNjCf2qBqUDuYT21gvtO1wamOeBYqVYcjPz8/unbtyoIFC3zrPB4PCxYsoGfPnsW+pmfPnkW2B5g/f/5JtxcRERGRmqvUp93Hjx/PmDFj6NatG927d+eFF14gKyuLG264AYDrrruO+vXrM2XKFADuvvtu+vfvz3PPPceFF17IJ598wl9//cUbb7xRvp9ERERERKq8UofPK6+8kkOHDjFp0iSSkpLo3Lkzc+fO9XUqSkxMxGo9dkC1V69ezJgxg4cffpiHHnqIFi1aMHPmTNq3b19+n6Kc+fv7M3ny5BNO/UvlURuYT21gPrVB1aB2MJ/awHzl2QYW7+n6w4uIiIiIlJMyTa8pIiIiIlIWCp8iIiIiUmkUPkVERESk0ih8ioiIiEilUfj8m1dffZXGjRsTEBBAjx49+PPPP80uqVZ59NFHsVgsRZbWrVubXVaN9ssvvzB8+HDi4+OxWCzMnDmzyPNer5dJkyZRr149AgMDGTRoEFu3bjWn2BrqdG1w/fXXn/C9OP/8880ptoaaMmUKZ599NqGhocTExDBy5Eg2b95cZJvc3FzGjh1LnTp1CAkJ4bLLLjthEhUpu5K0wbnnnnvCd+G2224zqeKa5/XXX6djx46+geR79uzJ999/73u+vL4DCp/H+fTTTxk/fjyTJ09mxYoVdOrUiaFDh3Lw4EGzS6tV2rVrx4EDB3zLb7/9ZnZJNVpWVhadOnXi1VdfLfb5Z555hpdeeolp06axdOlSgoODGTp0KLm5uZVcac11ujYAOP/884t8Lz7++ONKrLDmW7RoEWPHjuWPP/5g/vz5uFwuhgwZQlZWlm+be++9l2+//ZbPP/+cRYsWsX//fi699FITq65ZStIGALfcckuR78IzzzxjUsU1T4MGDXjqqadYvnw5f/31FwMGDGDEiBGsX78eKMfvgFd8unfv7h07dqzvsdvt9sbHx3unTJliYlW1y+TJk72dOnUyu4xaC/B+/fXXvscej8cbFxfnffbZZ33rjh496vX39/d+/PHHJlRY8/29Dbxer3fMmDHeESNGmFJPbXXw4EEv4F20aJHX6zV+7x0Oh/fzzz/3bbNx40Yv4F2yZIlZZdZof28Dr9fr7d+/v/fuu+82r6haKDIy0vvWW2+V63dARz4LOJ1Oli9fzqBBg3zrrFYrgwYNYsmSJSZWVvts3bqV+Ph4mjZtyujRo0lMTDS7pFpr586dJCUlFflehIeH06NHD30vKtnChQuJiYmhVatW3H777aSkpJhdUo2WlpYGQFRUFADLly/H5XIV+S60bt2ahg0b6rtQQf7eBoWmT59OdHQ07du3Z+LEiWRnZ5tRXo3ndrv55JNPyMrKomfPnuX6HSj1DEc11eHDh3G73b6ZmgrFxsayadMmk6qqfXr06MF7771Hq1atOHDgAI899hh9+/Zl3bp1hIaGml1erZOUlARQ7Pei8DmpeOeffz6XXnopTZo0Yfv27Tz00ENccMEFLFmyBJvNZnZ5NY7H4+Gee+6hd+/evtn4kpKS8PPzIyIiosi2+i5UjOLaAGDUqFE0atSI+Ph41qxZw4QJE9i8eTNfffWVidXWLGvXrqVnz57k5uYSEhLC119/Tdu2bVm1alW5fQcUPqVKueCCC3z3O3bsSI8ePWjUqBGfffYZN910k4mViZjnqquu8t3v0KEDHTt2pFmzZixcuJCBAweaWFnNNHbsWNatW6frzU10sja49dZbffc7dOhAvXr1GDhwINu3b6dZs2aVXWaN1KpVK1atWkVaWhpffPEFY8aMYdGiReW6D512LxAdHY3NZjuh11ZycjJxcXEmVSURERG0bNmSbdu2mV1KrVT4u6/vRdXStGlToqOj9b2oAOPGjeO7777j559/pkGDBr71cXFxOJ1Ojh49WmR7fRfK38naoDg9evQA0HehHPn5+dG8eXO6du3KlClT6NSpEy+++GK5fgcUPgv4+fnRtWtXFixY4Fvn8XhYsGABPXv2NLGy2i0zM5Pt27dTr149s0uplZo0aUJcXFyR70V6ejpLly7V98JEe/fuJSUlRd+LcuT1ehk3bhxff/01P/30/+3cPUsjcRTF4bNIJihocCT4BjMEYmujBG3SBBQFQbEIVkHFRhE0qGARRPADCH4Au+kttVgtLYNdwCCkUFvxDRvvFguysrIubHaGnf09kGqmOMzlwiH5Z74qk8m8uz40NKREIvFuF2q1mhqNBrvQJJ/N4CPValWS2IW/6PX1VS8vL03dAX52/0G5XFapVNLw8LByuZz29/f1+Pio+fn5qKP9NzY2NjQ1NSXf93V9fa2dnR21tLRobm4u6mix9fDw8O5bg6urK1WrVbmuK8/ztLa2pr29PQ0MDCiTyahSqaivr0/T09PRhY6ZX83AdV3t7u5qdnZWPT09qtfr2traUjab1fj4eISp42VlZUVBEOjo6Ejt7e1vZ9hSqZRaW1uVSqW0uLiocrks13XV0dGh1dVVjY6OamRkJOL08fDZDOr1uoIg0OTkpLq6unRxcaH19XXl83kNDg5GnD4etre3NTExIc/zdH9/ryAIdHZ2puPj4+buQHP/kP/vOzg4MM/zzHEcy+Vydn5+HnWk/0qxWLTe3l5zHMf6+/utWCza5eVl1LFi7fT01CT99CmVSmb2/XVLlUrFuru7LZlMWqFQsFqtFm3omPnVDJ6enmxsbMzS6bQlEgnzfd+Wlpbs9vY26tix8tHzl2SHh4dv9zw/P9vy8rJ1dnZaW1ubzczM2M3NTXShY+azGTQaDcvn8+a6riWTSctms7a5uWl3d3fRBo+RhYUF833fHMexdDpthULBTk5O3q43awe+mJn9aVMGAAAAfgdnPgEAABAayicAAABCQ/kEAABAaCifAAAACA3lEwAAAKGhfAIAACA0lE8AAACEhvIJAACA0FA+AQAAEBrKJwAAAEJD+QQAAEBoKJ8AAAAIzTeZ9ZSlVbZwJwAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 800x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "pd.DataFrame(history.history).plot(figsize=(8, 5))\n",
    "plt.grid(True)\n",
    "plt.gca().set_ylim(0, 1) # set the vertical range to [0-1] plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[1m313/313\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m3s\u001b[0m 8ms/step - accuracy: 0.5026 - loss: 453.7460\n",
      "\u001b[1m1/1\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 115ms/step\n",
      "[[0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
      " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
      " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]]\n",
      "['cat' 'ship' 'ship']\n",
      "[['cat']\n",
      " ['ship']\n",
      " ['ship']]\n"
     ]
    }
   ],
   "source": [
    "model.evaluate(X_test, y_test)\n",
    "\n",
    "X_new = X_test[:3]\n",
    "y_proba = model.predict(X_new)# get prediction probabilities\n",
    "print(y_proba.round(2))\n",
    "\n",
    "y_pred = np.argmax(y_proba, axis=-1) # get predicted class index by finding the index of maximum probability along the last axis\n",
    "print(np.array(class_names)[y_pred])\n",
    "print(np.array(class_names)[y_test[:3]])\n",
    "#The predict_classes method was deprecated in TensorFlow 2.6 and has been removed in later versions.\n",
    "#Instead, you should use model.predict followed by np.argmax to get the predicted class indices."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/caroldong/anaconda3/lib/python3.11/site-packages/keras/src/layers/convolutional/base_conv.py:107: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.\n",
      "  super().__init__(activity_regularizer=activity_regularizer, **kwargs)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m43s\u001b[0m 30ms/step - accuracy: 0.3149 - loss: 1.8372 - val_accuracy: 0.5305 - val_loss: 1.3117\n",
      "Epoch 2/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.5469 - loss: 1.2736 - val_accuracy: 0.6277 - val_loss: 1.0631\n",
      "Epoch 3/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.6426 - loss: 1.0196 - val_accuracy: 0.6767 - val_loss: 0.9493\n",
      "Epoch 4/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.7010 - loss: 0.8597 - val_accuracy: 0.6697 - val_loss: 0.9730\n",
      "Epoch 5/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.7454 - loss: 0.7405 - val_accuracy: 0.7103 - val_loss: 0.8589\n",
      "Epoch 6/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 29ms/step - accuracy: 0.7777 - loss: 0.6405 - val_accuracy: 0.7103 - val_loss: 0.8640\n",
      "Epoch 7/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.8131 - loss: 0.5360 - val_accuracy: 0.7198 - val_loss: 0.8740\n",
      "Epoch 8/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.8432 - loss: 0.4564 - val_accuracy: 0.7135 - val_loss: 0.9140\n",
      "Epoch 9/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.8661 - loss: 0.3955 - val_accuracy: 0.7135 - val_loss: 0.9929\n",
      "Epoch 10/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.8726 - loss: 0.3669 - val_accuracy: 0.7117 - val_loss: 1.0158\n",
      "Epoch 11/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.8896 - loss: 0.3192 - val_accuracy: 0.7088 - val_loss: 1.1317\n",
      "Epoch 12/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.9070 - loss: 0.2766 - val_accuracy: 0.7187 - val_loss: 1.0851\n",
      "Epoch 13/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.9161 - loss: 0.2480 - val_accuracy: 0.7183 - val_loss: 1.2022\n",
      "Epoch 14/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.9187 - loss: 0.2430 - val_accuracy: 0.7062 - val_loss: 1.2013\n",
      "Epoch 15/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.9261 - loss: 0.2221 - val_accuracy: 0.7082 - val_loss: 1.2085\n",
      "Epoch 16/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.9305 - loss: 0.2028 - val_accuracy: 0.7173 - val_loss: 1.2646\n",
      "Epoch 17/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.9321 - loss: 0.2038 - val_accuracy: 0.7055 - val_loss: 1.2500\n",
      "Epoch 18/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.9416 - loss: 0.1825 - val_accuracy: 0.7023 - val_loss: 1.3375\n",
      "Epoch 19/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.9380 - loss: 0.1873 - val_accuracy: 0.7077 - val_loss: 1.3760\n",
      "Epoch 20/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.9452 - loss: 0.1698 - val_accuracy: 0.7107 - val_loss: 1.3339\n",
      "Epoch 21/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.9482 - loss: 0.1681 - val_accuracy: 0.7020 - val_loss: 1.4338\n",
      "Epoch 22/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.9444 - loss: 0.1702 - val_accuracy: 0.7035 - val_loss: 1.4093\n",
      "Epoch 23/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.9496 - loss: 0.1525 - val_accuracy: 0.7145 - val_loss: 1.4983\n",
      "Epoch 24/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.9490 - loss: 0.1600 - val_accuracy: 0.7163 - val_loss: 1.5000\n",
      "Epoch 25/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.9532 - loss: 0.1462 - val_accuracy: 0.7078 - val_loss: 1.5134\n",
      "Epoch 26/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.9569 - loss: 0.1401 - val_accuracy: 0.7125 - val_loss: 1.5424\n",
      "Epoch 27/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.9552 - loss: 0.1441 - val_accuracy: 0.6950 - val_loss: 1.6248\n",
      "Epoch 28/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m40s\u001b[0m 29ms/step - accuracy: 0.9546 - loss: 0.1479 - val_accuracy: 0.7120 - val_loss: 1.4308\n",
      "Epoch 29/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m41s\u001b[0m 30ms/step - accuracy: 0.9545 - loss: 0.1443 - val_accuracy: 0.7152 - val_loss: 1.7316\n",
      "Epoch 30/30\n",
      "\u001b[1m1375/1375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m42s\u001b[0m 30ms/step - accuracy: 0.9598 - loss: 0.1345 - val_accuracy: 0.7105 - val_loss: 1.6997\n"
     ]
    }
   ],
   "source": [
    "kinit='he_uniform'\n",
    "lr=0.001\n",
    "dr=0.1\n",
    "#a convolutional neural network described in Ch. 14, p. 447.\n",
    "model = keras.models.Sequential([\n",
    "        keras.layers.Conv2D(64, 7, activation=\"relu\",kernel_initializer=kinit, padding=\"same\",\n",
    "                            input_shape=[32, 32,3]),\n",
    "        keras.layers.MaxPooling2D(2),\n",
    "        keras.layers.Conv2D(128, 3, activation=\"relu\",kernel_initializer=kinit, padding=\"same\"),\n",
    "        keras.layers.Conv2D(128, 3, activation=\"relu\",kernel_initializer=kinit, padding=\"same\"),\n",
    "        keras.layers.MaxPooling2D(2),\n",
    "        keras.layers.Conv2D(256, 3, activation=\"relu\",kernel_initializer=kinit, padding=\"same\"),\n",
    "        keras.layers.Conv2D(256, 3, activation=\"relu\",kernel_initializer=kinit, padding=\"same\"),\n",
    "        keras.layers.MaxPooling2D(2),\n",
    "        keras.layers.Flatten(),\n",
    "        keras.layers.Dense(128, activation=\"relu\",kernel_initializer=kinit),\n",
    "        keras.layers.Dropout(dr),\n",
    "        keras.layers.Dense(64, activation=\"relu\",kernel_initializer=kinit),\n",
    "        keras.layers.Dropout(dr),\n",
    "        keras.layers.Dense(10, activation=\"softmax\")\n",
    "])\n",
    "opt=keras.optimizers.Adam(learning_rate=lr)\n",
    "\n",
    "model.compile(loss=\"sparse_categorical_crossentropy\",\n",
    "                  optimizer=opt,\n",
    "                  metrics=[\"accuracy\"])\n",
    "\n",
    "history = model.fit(X_train, y_train, epochs=30,\n",
    "                    validation_data=(X_valid, y_valid))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.0, 1.0)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "pd.DataFrame(history.history).plot(figsize=(8, 5))\n",
    "plt.grid(True)\n",
    "plt.gca().set_ylim(0, 1) # set the vertical range to [0-1] plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[1m313/313\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m3s\u001b[0m 8ms/step - accuracy: 0.6185 - loss: 393.4829\n",
      "\u001b[1m1/1\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 126ms/step\n",
      "[[0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
      " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
      " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]]\n",
      "['cat' 'ship' 'ship']\n",
      "[['cat']\n",
      " ['ship']\n",
      " ['ship']]\n"
     ]
    }
   ],
   "source": [
    "model.evaluate(X_test, y_test)\n",
    "\n",
    "X_new = X_test[:3]\n",
    "y_proba = model.predict(X_new)# get prediction probabilities\n",
    "print(y_proba.round(2))\n",
    "\n",
    "y_pred = np.argmax(y_proba, axis=-1) # get predicted class index by finding the index of maximum probability along the last axis\n",
    "print(np.array(class_names)[y_pred])\n",
    "print(np.array(class_names)[y_test[:3]])\n",
    "#The predict_classes method was deprecated in TensorFlow 2.6 and has been removed in later versions.\n",
    "#Instead, you should use model.predict followed by np.argmax to get the predicted class indices."
   ]
  }
 ],
 "metadata": {
  "colab": {
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}