--- a
+++ b/1. Applying AI to 2D Medical Imaging Data/11. Evaluating Your Model Exercise/solution.ipynb
@@ -0,0 +1,652 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Using TensorFlow backend.\n"
+     ]
+    }
+   ],
+   "source": [
+    "import numpy as np \n",
+    "import pandas as pd \n",
+    "import os\n",
+    "from glob import glob\n",
+    "%matplotlib inline\n",
+    "import matplotlib.pyplot as plt\n",
+    "import tensorflow as tf\n",
+    "from skimage import io\n",
+    "\n",
+    "from keras.preprocessing.image import ImageDataGenerator\n",
+    "from keras.layers import GlobalAveragePooling2D, Dense, Dropout, Flatten, Conv2D, MaxPooling2D\n",
+    "from keras.models import Sequential, Model\n",
+    "from keras.applications.vgg16 import VGG16\n",
+    "from keras.applications.resnet import ResNet50 \n",
+    "from keras.optimizers import Adam\n",
+    "from keras.callbacks import ModelCheckpoint, LearningRateScheduler, EarlyStopping, ReduceLROnPlateau"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "train_df = pd.read_csv('train.csv')\n",
+    "valid_df = pd.read_csv('test.csv')"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Setting up the image augmentation from last Lesson: "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "IMG_SIZE = (224, 224)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Found 20 validated image filenames belonging to 2 classes.\n",
+      "Found 6 validated image filenames belonging to 2 classes.\n"
+     ]
+    }
+   ],
+   "source": [
+    "train_idg = ImageDataGenerator(rescale=1. / 255.0,\n",
+    "                              horizontal_flip = True, \n",
+    "                              vertical_flip = False, \n",
+    "                              height_shift_range= 0.1, \n",
+    "                              width_shift_range=0.1, \n",
+    "                              rotation_range=20, \n",
+    "                              shear_range = 0.1,\n",
+    "                              zoom_range=0.1)\n",
+    "\n",
+    "train_gen = train_idg.flow_from_dataframe(dataframe=train_df, \n",
+    "                                         directory=None, \n",
+    "                                         x_col = 'img_path',\n",
+    "                                         y_col = 'class',\n",
+    "                                         class_mode = 'binary',\n",
+    "                                         target_size = IMG_SIZE, \n",
+    "                                         batch_size = 9\n",
+    "                                         )\n",
+    "\n",
+    "# Note that the validation data should not be augmented! We only want to do some basic intensity rescaling here\n",
+    "val_idg = ImageDataGenerator(rescale=1. / 255.0\n",
+    "                                 )\n",
+    "\n",
+    "val_gen = val_idg.flow_from_dataframe(dataframe=valid_df, \n",
+    "                                         directory=None, \n",
+    "                                         x_col = 'img_path',\n",
+    "                                         y_col = 'class',\n",
+    "                                         class_mode = 'binary',\n",
+    "                                         target_size = IMG_SIZE, \n",
+    "                                         batch_size = 6) ## We've only been provided with 6 validation images"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "## Pull a single large batch of random validation data for testing after each epoch\n",
+    "testX, testY = val_gen.next()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Now we'll load in VGG16 with pre-trained ImageNet weights: "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Downloading data from https://github.com/fchollet/deep-learning-models/releases/download/v0.1/vgg16_weights_tf_dim_ordering_tf_kernels.h5\n",
+      "553467904/553467096 [==============================] - 7s 0us/step\n"
+     ]
+    }
+   ],
+   "source": [
+    "model = VGG16(include_top=True, weights='imagenet')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "transfer_layer = model.get_layer('block5_pool')\n",
+    "vgg_model = Model(inputs=model.input,\n",
+    "                   outputs=transfer_layer.output)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "## Now, choose which layers of VGG16 we actually want to fine-tune (if any)\n",
+    "## Here, we'll freeze all but the last convolutional layer\n",
+    "for layer in vgg_model.layers[0:17]:\n",
+    "    layer.trainable = False"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "new_model = Sequential()\n",
+    "\n",
+    "# Add the convolutional part of the VGG16 model from above.\n",
+    "new_model.add(vgg_model)\n",
+    "\n",
+    "# Flatten the output of the VGG16 model because it is from a\n",
+    "# convolutional layer.\n",
+    "new_model.add(Flatten())\n",
+    "\n",
+    "# Add a dropout-layer which may prevent overfitting and\n",
+    "# improve generalization ability to unseen data e.g. the test-set.\n",
+    "new_model.add(Dropout(0.5))\n",
+    "\n",
+    "# Add a dense (aka. fully-connected) layer.\n",
+    "# This is for combining features that the VGG16 model has\n",
+    "# recognized in the image.\n",
+    "new_model.add(Dense(1024, activation='relu'))\n",
+    "\n",
+    "# Add a dropout-layer which may prevent overfitting and\n",
+    "# improve generalization ability to unseen data e.g. the test-set.\n",
+    "new_model.add(Dropout(0.5))\n",
+    "\n",
+    "# Add a dense (aka. fully-connected) layer.\n",
+    "# This is for combining features that the VGG16 model has\n",
+    "# recognized in the image.\n",
+    "new_model.add(Dense(512, activation='relu'))\n",
+    "\n",
+    "# Add a dropout-layer which may prevent overfitting and\n",
+    "# improve generalization ability to unseen data e.g. the test-set.\n",
+    "new_model.add(Dropout(0.5))\n",
+    "\n",
+    "# Add a dense (aka. fully-connected) layer.\n",
+    "# This is for combining features that the VGG16 model has\n",
+    "# recognized in the image.\n",
+    "new_model.add(Dense(256, activation='relu'))\n",
+    "\n",
+    "# Add a dense (aka. fully-connected) layer.\n",
+    "# Change the activation function to sigmoid \n",
+    "# so output of the last layer is in the range of [0,1] \n",
+    "new_model.add(Dense(1, activation='sigmoid'))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Model: \"sequential_1\"\n",
+      "_________________________________________________________________\n",
+      "Layer (type)                 Output Shape              Param #   \n",
+      "=================================================================\n",
+      "model_1 (Model)              (None, 7, 7, 512)         14714688  \n",
+      "_________________________________________________________________\n",
+      "flatten_1 (Flatten)          (None, 25088)             0         \n",
+      "_________________________________________________________________\n",
+      "dropout_1 (Dropout)          (None, 25088)             0         \n",
+      "_________________________________________________________________\n",
+      "dense_1 (Dense)              (None, 1024)              25691136  \n",
+      "_________________________________________________________________\n",
+      "dropout_2 (Dropout)          (None, 1024)              0         \n",
+      "_________________________________________________________________\n",
+      "dense_2 (Dense)              (None, 512)               524800    \n",
+      "_________________________________________________________________\n",
+      "dropout_3 (Dropout)          (None, 512)               0         \n",
+      "_________________________________________________________________\n",
+      "dense_3 (Dense)              (None, 256)               131328    \n",
+      "_________________________________________________________________\n",
+      "dense_4 (Dense)              (None, 1)                 257       \n",
+      "=================================================================\n",
+      "Total params: 41,062,209\n",
+      "Trainable params: 28,707,329\n",
+      "Non-trainable params: 12,354,880\n",
+      "_________________________________________________________________\n"
+     ]
+    }
+   ],
+   "source": [
+    "new_model.summary()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "## Set our optimizer, loss function, and learning rate\n",
+    "optimizer = Adam(lr=1e-4)\n",
+    "loss = 'binary_crossentropy'\n",
+    "metrics = ['binary_accuracy']"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "new_model.compile(optimizer=optimizer, loss=loss, metrics=metrics)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 13,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Epoch 1/10\n",
+      "3/3 [==============================] - 6s 2s/step - loss: 0.7557 - binary_accuracy: 0.6500 - val_loss: 0.7366 - val_binary_accuracy: 0.5000\n",
+      "Epoch 2/10\n",
+      "3/3 [==============================] - 0s 162ms/step - loss: 0.8036 - binary_accuracy: 0.5500 - val_loss: 0.6993 - val_binary_accuracy: 0.5000\n",
+      "Epoch 3/10\n",
+      "3/3 [==============================] - 2s 506ms/step - loss: 1.0279 - binary_accuracy: 0.4000 - val_loss: 0.6848 - val_binary_accuracy: 0.6667\n",
+      "Epoch 4/10\n",
+      "3/3 [==============================] - 0s 153ms/step - loss: 0.6307 - binary_accuracy: 0.6000 - val_loss: 0.6639 - val_binary_accuracy: 0.3333\n",
+      "Epoch 5/10\n",
+      "3/3 [==============================] - 1s 195ms/step - loss: 0.5893 - binary_accuracy: 0.6500 - val_loss: 0.6538 - val_binary_accuracy: 0.5000\n",
+      "Epoch 6/10\n",
+      "3/3 [==============================] - 0s 162ms/step - loss: 1.0570 - binary_accuracy: 0.5000 - val_loss: 0.6682 - val_binary_accuracy: 0.5000\n",
+      "Epoch 7/10\n",
+      "3/3 [==============================] - 1s 168ms/step - loss: 0.8354 - binary_accuracy: 0.5500 - val_loss: 0.6473 - val_binary_accuracy: 0.5000\n",
+      "Epoch 8/10\n",
+      "3/3 [==============================] - 1s 198ms/step - loss: 0.7465 - binary_accuracy: 0.6500 - val_loss: 0.6107 - val_binary_accuracy: 0.8333\n",
+      "Epoch 9/10\n",
+      "3/3 [==============================] - 1s 172ms/step - loss: 0.6981 - binary_accuracy: 0.6000 - val_loss: 0.5994 - val_binary_accuracy: 0.8333\n",
+      "Epoch 10/10\n",
+      "3/3 [==============================] - 1s 207ms/step - loss: 0.6567 - binary_accuracy: 0.6500 - val_loss: 0.5964 - val_binary_accuracy: 0.6667\n"
+     ]
+    }
+   ],
+   "source": [
+    "## Run for 10 epochs to see if any learning occurs:\n",
+    "history = new_model.fit_generator(train_gen, \n",
+    "                                  validation_data = (testX, testY), \n",
+    "                                  epochs = 10)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 14,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Define a function here that will plot loss, val_loss, binary_accuracy, and val_binary_accuracy over all of \n",
+    "# your epochs: \n",
+    "def plot_history(history):\n",
+    "    N = len(history.history[\"loss\"])\n",
+    "    plt.style.use(\"ggplot\")\n",
+    "    plt.figure()\n",
+    "    plt.plot(np.arange(0, N), history.history[\"loss\"], label=\"train_loss\")\n",
+    "    plt.plot(np.arange(0, N), history.history[\"val_loss\"], label=\"val_loss\")\n",
+    "    plt.plot(np.arange(0, N), history.history[\"binary_accuracy\"], label=\"train_acc\")\n",
+    "    plt.plot(np.arange(0, N), history.history[\"val_binary_accuracy\"], label=\"val_acc\")\n",
+    "    plt.title(\"Training Loss and Accuracy on Dataset\")\n",
+    "    plt.xlabel(\"Epoch #\")\n",
+    "    plt.ylabel(\"Loss/Accuracy\")\n",
+    "    plt.legend(loc=\"lower left\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 15,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEaCAYAAAD+E0veAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjAsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8GearUAAAgAElEQVR4nOydd1gU1/rHv2e20nuRjggoCApiTeyKsUaj0cQkxqumaMq9Xn+5icYk3iTmGk1VY2LUmESTaIox1ajEXlBAsKBSFBDpRTos7M77+2N1daUtuA2Yz/PwwM7MmfPOYWa/c877nvMyIiIICAgICAgA4ExtgICAgICA+SCIgoCAgICABkEUBAQEBAQ0CKIgICAgIKBBEAUBAQEBAQ2CKAgICAgIaBBEwQw4dOgQGGO4fv16m8oxxrB9+3YDWdV1GTFiBBYsWGBqMwQETIIgCm2AMdbij5+fX7vOO2TIEOTl5cHDw6NN5fLy8jBjxox21dlWBAFqmueffx4ikQhr1641tSmdmhUrVmieM5FIBHt7e/Tr1w//+c9/kJ2d3ebzjRkzBnPnztW/oTrQo0cPrFixwiR164IgCm0gLy9P8/PLL78AAE6fPq3ZFhcXp3V8fX29TueVSqVwd3cHx7Xt3+Hu7g65XN6mMgL6o6amBtu3b8eyZcvw+eefm9ocALrfcx0RPz8/5OXl4fr16zh16hReeuklHD58GKGhoThx4oSpzes8kEC7OHr0KAGgjIwMzTYA9PHHH9Ojjz5Ktra2NGPGDCIiWrZsGfXs2ZMsLCzIy8uLnnnmGSorK9OUO3jwIAGg7Oxsrc/79u2joUOHkoWFBfXq1Yv++usvLRsA0LZt27Q+f/LJJ/T444+TtbU1eXl50bvvvqtVpri4mGbMmEGWlpbk6upKy5cvpzlz5tDo0aNbvN6767qbL7/8knr16kVSqZQ8PT3p1VdfpYaGBq32GjJkCFlbW5O1tTWFh4drXc/KlSvJ39+fpFIpOTs7U3R0NNXU1DRb3zfffEMDBgwgW1tbcnJyogkTJlBKSopmf0ZGBgGgnTt30qRJk8jCwoL8/f3p66+/1jpPZmYmjRs3juRyOXl7e9PatWtp+PDhNH/+/Bbbg4joiy++oIiICKqrqyMHBwc6fvx4o2N27NhBkZGRJJPJyNHRkR544AEqLS3V7F+/fr2m3VxcXGj69Omafb6+vvTWW29pnW/+/Pk0fPhwzefhw4fTvHnzaPny5eTu7k7Ozs46tQ8RUUFBAc2dO5dcXV1JJpNRUFAQbdmyhVQqFfn7+9PKlSu1jq+qqiIbGxvaunVrs21y+fJlmjBhAllZWZGVlRVNmjSJ0tLSNPu3bt1KIpGIjh07RhEREWRhYUFRUVEUHx/ffEMT0RtvvEEBAQGNttfX19OgQYOoR48epFKpiIjo6tWrNG3aNOrWrRtZWFhQ7969tf7vTz75JAHQ+jl48CARtf6slpeX09y5c8nNzY2kUil5eXnR4sWLtWxau3YtBQcHk0wmox49etDbb7+teRaGDx/eqO47v0PMAUEU2klzouDo6Ehr166l9PR0zUP41ltv0ZEjRygjI4NiYmIoODiY5syZoynXnCiEh4fTnj17KDU1lZ544gmys7OjGzduaNV3tyi4urrS559/Tunp6fTxxx8TADpw4IDmmMmTJ1NgYCAdOHCALly4QHPnziVbW9t7EoXff/+dOI6jd955h1JSUmjHjh1kb29Py5cvJyIipVJJDg4OtHjxYkpNTaXU1FTatWsXHTlyhIiIfvrpJ7KxsaFff/2VsrKyKDExkT788MMWReGLL76g3377jdLT0+nMmTM0efJk6tGjBykUCiK6LQr+/v60c+dOSktLo5dffplEIhGlpqYSERHP8xQREUFRUVEUGxtLiYmJNGbMGLKxsdFJFAYOHEgff/wxEREtXLiQnnzyyUY2isVievPNNyk5OZnOnj1LH330ERUVFRER0euvv05WVla0bt06SklJoYSEBC0R0FUUrK2t6ZlnnqHk5GQ6d+6cTu1TU1NDPXv2pIiICNq/fz9duXKF9u7dS9999x0REb3zzjvUvXt34nleU9fmzZvJzs6Oqqurm2yPmpoa8vHxoVGjRlF8fDzFx8fTiBEjKCAgQFPv1q1biTFGQ4cOpSNHjtClS5do7Nix1L17d62XiLtpThSIiH744QcCQHFxcUREdO7cOVq/fj2dPXuW0tPTae3atSQSiTTPQVlZGQ0dOpRmzpxJeXl5lJeXp7GvtWf1hRdeoPDwcIqNjaWsrCw6fvw4ff7551p2+vj40K5du+jq1av0xx9/kLe3t+ZZKCkpIT8/P1qyZImmbqVS2ex1mwJBFNpJc6Iwb968Vsvu2rWLpFKp5s2mOVH46aefNGXy8vIIgNbbdVOi8MILL2jVFRwcTK+88goREaWmphIAiomJ0eyvr68nLy+vexKF+++/nx5++GGtbR999BHJ5XJSKBRUWlqq9TZ2Nx988AEFBgZSfX19iza0RElJCQGgY8eOEdFtUXj//fc1xzQ0NJCVlRV99tlnRES0f/9+AqD1Bl1YWEhyubxVUUhKSiKJREKFhYVERHTq1CmysLDQEm1vb2967rnnmixfVVVFcrmc1qxZ02wduopCYGCg5l5qjrvbZ/PmzSSTyTT33N3k5+eTRCKh/fv3a7YNGjSIFi1a1GwdmzdvJgsLC43o3TqPXC6nr776iojUogCAEhISNMecPHmSANDly5ebPXdLonDp0iVNr7A5pkyZQgsWLNB8Hj16dCMRb4q7n9UpU6Y0W666uposLCxoz549Wtu/+uorsrOz03wOCAigN954o9W6TYXgU9AzAwYMaLRt165dGDZsGDw8PGBtbY3HHnsM9fX1yM/Pb/Fcffv21fzt7u4OkUiEgoICncsAgKenp6bMxYsXAQCDBg3S7JdIJIiKimr5olohOTkZw4YN09o2fPhw1NXV4cqVK3BwcMCCBQswbtw4jB8/HqtWrUJKSorm2JkzZ6KhoQG+vr6YO3cutm3bhsrKyhbrTEpKwrRp0+Dv7w8bGxv4+PgAALKysrSOu7M9xGIx3NzctNrD2dkZQUFBmmNcXFwQHBzc6jVv3LgREyZMgIuLCwD1/93f31/jjC8sLER2djaio6ObLJ+cnIy6urpm97eFfv36NfJHtdY+CQkJCAkJgZeXV5PndHNzw4MPPohNmzZp7I2NjcVTTz3VrB3JyckICQmBs7Oz1nmCg4ORnJys2cYYQ58+fTSfPT09AaDVe7s56OaanowxAGpfzyuvvILQ0FA4OjrC2toaf/75Z6N7oylae1YXLVqEH3/8Eb1798Y///lP7NmzBzzPa66/trYW06dPh7W1tebnmWeeQXl5OYqKitp1fcZGEAU9Y2VlpfX51KlTePjhhzFs2DD8/PPPOHPmDD777DMArTsFpVJpo223bkBdyzDGGpW59fDok7vPefeDumnTJiQkJGDs2LE4fPgwevfujY0bNwJQfylcvnwZX3zxBVxdXfHWW28hODi42aiSmpoaREdHgzGGL774AqdPn0ZcXBwYY43atKX2IKJ2tUV1dTW++eYb/PrrrxCLxZqfS5cuNXI4t3b+lvZzHKdpx1s0NDQ0Ou7ue07X9mnNtmeffRa7d+9GUVERNm3ahP79+zd66dDleu5uZ47jIBKJGpVp7d5ujgsXLgAAAgICAAAvvfQStm/fjtdffx0HDx5EUlISJkyY0OrzpsuzOm7cOFy7dg2vvvoq6urq8Pjjj2PUqFFQqVQa+3/44QckJSVpfs6fP4+0tDQ4Ojq26/qMjSAKBubYsWNwdnbG22+/jYEDByIoKKjN8xH0RUhICADg5MmTmm1KpRIJCQn3dN7Q0FAcPnxYa9uRI0dgYWGB7t27a7b17t0b//73v7Fnzx7Mnz9f6wtUJpPhgQcewOrVq3H+/HnU1NRg9+7dTdZ36dIlFBUVYeXKlRg5ciR69eqFGzduNPoC1cXuoqIipKWlabYVFxcjNTW1xXI7duyASCTC2bNntR7+o0ePat6oXV1d4eXlhb179zZ5jpCQEMjl8mb3A4Crqytyc3O1tiUmJrZ6Xbq0T79+/ZCcnNzivThq1Cj4+Pjg888/x7Zt21rsJQDq9kxOTkZxcbFmW0FBAVJTUxEaGtqq3e2hoaEBH3zwAYKCgjSCdeTIETz22GOYNWsW+vTpg+7duzf6n0qlUqhUKq1tuj6rjo6OePTRR7Fx40b88ccfOHz4MC5evIjQ0FDI5XJcvXoVPXr0aPRzSwibqtucEJvagM5OcHAwioqKsGXLFowcORLHjh3Dhg0bTGJLYGAgJk+ejOeeew4bN26Ei4sL3n//fVRUVOj0xnzt2jUkJSVpbfPw8MDSpUsxefJkrFq1Cg899BCSkpKwYsUKLFmyBFKpFOnp6di0aRMmT54Mb29v5Obm4ujRo4iMjAQAbNmyBTzPY8CAAbC3t8fff/+NyspKjYjdja+vL2QyGdatW4clS5YgMzMTr7zySpvf+kePHo0+ffrg8ccfx7p16yCVSvHyyy9DLG75sdi4cSOmTZuGsLCwRvvuu+8+fP755xg0aBDeeOMNLFy4EG5ubpgxYwZ4nsfBgwfxyCOPwNnZGUuWLMGKFStgYWGBsWPHora2Fn/++SeWLl0KQB1Lv2HDBkybNg2+vr747LPPkJWV1eobpy7t8+ijj2L16tWYMmUKVq9ejYCAAFy9ehXFxcWYNWsWAPUb/NNPP43ly5dDKpXi0UcfbbHe2bNn480338SsWbOwZs0aEBH+7//+D56enppz3gsqlUozjFNeXo7ExER8+OGHuHz5Mvbu3asZQgsODsYvv/yiGcb54IMPkJubCzc3N825/P39cfDgQVy5cgV2dnaws7PT6Vl99dVX0a9fP4SGhoLjOHzzzTewtraGj48PrK2tsWzZMixbtgwAMHbsWCiVSpw/fx6JiYl49913NXUfP34c165dg6WlJRwdHdscjm5QTOfO6Ng052huyhm7fPlycnV1JUtLSxo/fjx9++23WmWbczTf7QQUiURa4YB319dU/Xc71IqLi2n69OlkYWFBLi4u9Nprr9GMGTNo0qRJLV4v7gqju/Xzv//9j4jUIak9e/YkiURCHh4etGzZMk00SW5uLk2bNo08PT1JKpVSt27daMGCBZpQv59++okGDx5M9vb2ZGFhQaGhobR58+YW7fnhhx+oR48eJJPJqG/fvnTo0CGt9rnlaD569KhWubudfBkZGTR27FiSyWTk6elJH330UYshqYmJiY0c/neyfv16srS01Fzb9u3bKTw8nKRSKTk6OtKECRM0zmie5+mjjz6ioKAgkkgk5OrqqgljJiKqqKigxx9/nOzt7cnFxYXeeOONJh3NTdnaWvsQqYMXnnjiCXJyciKZTEbBwcGNwk2LiopIIpHQ008/3eT13s3ly5dp/PjxmpDUiRMnNhmSeifZ2dktBiIQqR3Nt+45xhjZ2tpSREQEvfTSS42ek2vXrlF0dDRZWlqSu7s7vf766zRv3jytdrty5QoNHTqUrKystOpu7Vl98803KTQ0lKysrMjW1paGDRvW6B7bvHkz9enTh2QyGdnb29OAAQNow4YNmv1xcXEUGRlJcrncLENSGZGQea0ro1Kp0LNnT0yZMgXvv/++qc0RMDNuDYvEx8ejX79+pjZHwAgIw0ddjCNHjqCwsBARERGorKzEhx9+iMzMTJNN+RcwTxQKBXJycrB06VIMHz5cEIQuhCAKXQyVSoW3334b6enpkEgk6N27Nw4ePNjk+LhA1+W7777DvHnzEBoaih9//NHU5ggYEWH4SEBAQEBAgxm5vAUEBAQETI0gCgICAgICGjq8T+HuyT264uzsrDXJpqsjtIc2QnvcRmgLbTpDe7SUu0XoKQgICAgIaBBEQUBAQEBAgyAKAgICAgIaBFEQEBAQENAgiIKAgICAgAZBFAQEBAQENAiiICAgICCgocPPUxC4d+jKZdTnWQDdfE1tioCAgIkRRKGLQ2Wl4Ne+iXKpFHj3CzBzSvYhICBgdIRvgC4MEYHf9glQUwW+rBS4dsXUJgkICJgYo/QUNmzYgDNnzsDOzq7JRC5EhK1btyIxMREymQyLFi3Syu0rYBjoxAHgXBzYhIdBe34EnYsH8ws0tVkCAgImxCg9hREjRmjyljZFYmIi8vPzsXbtWjz99NPYvHmzMczq0lBpMWjnJiAwBOzBxyAJDAGdjze1WQICAibGKKIQEhICa2vrZvfHx8dj2LBhYIwhKCgI1dXVuHHjhjFM65IQEfiv1gEqFbi5/wTjOEijhgCZaaAKod0FBLoyZuFoLi0thbOzs+azk5MTSktL4eDg0OjYmJgYxMTEAABWrVqlVa4tiMXidpft6NTs243Ki4mweWoJLEPUGdf4gcNR/e0mWGemwmLURBNbaHq68v1xN0JbaNPZ28MsRKGp5G+MsSaPHTNmDMaMGaP53N4lbDvD8rftgYoLwH+xDugZjuqooai52QZO3v6AvRMqThxEdfhAE1tperrq/dEUQlto0xnaw+yXznZyctJq5JKSkiZ7CQL3BvE8+C/XAgC4uS9qhZ8yxsDC+gEXk0BKpalMFBAQMDFmIQpRUVE4cuQIiAipqamwtLQURMEA0OE9QMp5sJnzwJxcG+1n4VFAbQ2QftEE1gkICJgDRhk++uijj3Dx4kVUVlbi2WefxcyZM6G8+TYaHR2NiIgInDlzBi+++CKkUikWLVpkDLO6FFSYC/rxSyA0AmxodNMH9ewDiMWgc3FgPcONap+AgIB5YBRR+Ne//tXifsYYFixYYAxTuiTEq8BvXQuIxODmvNCsv4bJLYCgMHVo6sz5RrZSQEDAHDCL4SMBw0J//w6kXwR7ZAGYY8tREyw8CsjPARXmGck6AQEBc0IQhU4O5V8H/bwN6DMAbPCoVo9nYVHqcsJENgGBLokgCp0Y4lXgv/gIkEjBPb6o2WGjO2Gu3QB3T9A5QRQEBLoigih0YmjvbiAjFWz2M2D2jjqXY2FRQOp5UF2tAa0TEBAwRwRR6KRQThbo12+AyMFgA4a1qSwLiwKUSuDyWQNZJyAgYK4IotAJIaUS/NaPAbkluMcW6jRspEVgCCC3EIaQBAS6IIIodELorx+BrHS1H8HWvs3lmVgChESAzic0uQSJgOEhRR3433eCKspMbYpAF0MQhU4GXbsK+n0nWP+hYP2GtPs8LDwKKCsBsjP0aJ2ArtCpQ6BfvgH/6f9AygZTmyPQhRBEoRNBygbwWz8CrG3BZj9zT+divfupzymEppoESowF5BZA+iXQN58JPTYBoyGIQieCft8JXM9UDxtZ297TuZidA+DbQxAFE0A11cClc2DDxoFNmAk6th904A9TmyXQRRBEoZNAmWmgPT+CDR4J1lc/S1+z8CjgagqoskIv5xPQDbqQAKiUYBGDwB6cDfQdCPp+M+hikqlNE+gCCKLQCaCGevUkNVt7sEee0tt5WVh/gAiUnKC3cwroQGIsYGMHdA8G4zhw8xcD7l7gN64GFeaa2jqBTo4gCp0A+vU7IC8b3JMvgFk2n/a0zfgGALb2gBCaajSooR50PgGs70AwTgQAYHJLcM8vBxgDv34lqLbGxFYKdGYEUejg0JXLoL0/gw2N1jiH9QXjOLDe/UDJZ0AqlV7PLdAMl88BilqwiMFam5mLO7hnXwYKcsBvfh/EC/8PAcMgiEIHhhQK9SQ1Byewh+cZpA4WHgXUVANXLhvk/ALaaKKOmshnwXqGq4cHz8WBdn9jAusEugKCKHRgaPd2oCBHnVrTwtIwlfTqC4hEQhSSESBeBUo6BRYWBSaRNHkMGzEBbNg40J4fwZ86bGQLBboCgih0UCg1GfT3r+oviV59DFYPs7QCeoQIomAMrqQAleVAxKBmD2GMgT36NBAYAvpqHSgzzYgGCnQFBFHogFBdLfgvPwac3cCmP2nw+lh4FJCTBSopMnhdXRlKPAmIxa36hphYAm7hUsDWHvwn74DKSo1koUBXQBCFDgjt+gooLlAPG8ktDF4fC+uvrvd8nMHr6qoQkdqf0LOPTkOBzMYO3HOvAjVV6qUwGuqNYKVAV0AQhQ4GXToLOvgn2OjJYEG9jVOpuyfg4i6smmpIcjKB4gKwFoaO7oZ5+4Obt1g9wXD7p8JSGAJ6QWysipKSkrB161bwPI/Ro0dj6tSpWvurqqrw6aefoqCgABKJBAsXLoSPj4+xzOsQUG0N+C/XAq4eYFOfMFq9jDGw8P6go3tB9QowqcxodXcV6EwswBhY3wFtKsf6DQGb/Ajotx2Alx/Y2AcNZKFAV8EoPQWe57FlyxYsW7YMH374IY4fP47r169rHfPzzz/Dz88P7733Hp5//nl8+eWXxjCtQ0E/fAHcKAE3719gMuN+MbOwKKC+Hkg5b9R6uwqUGAsE9ASzdWhzWTbpESByMOiHraALZwxgnUBXwiiikJ6eDnd3d7i5uUEsFmPIkCGIi9Men75+/TrCwsIAAJ6enigqKkJZmbCW/C3oQgLo6D6w6AfBAnoa34Cg3oBMDjon+BX0DRXlA9cz2jR0dCeM48D941+Apw/4z9eA8nP0bKFAV8Iow0elpaVwcnLSfHZyckJamnYona+vL06dOoWePXsiPT0dRUVFKC0thb29dpKYmJgYxMTEAABWrVoFZ2fndtkkFovbXdbY8NWVKNm+ASJvfzjNe9Egwze6tEdZn/5oSD4DJyentmdz62AY8/6oPhGDKgCOI8dDfA91ql57HyUvzQf77H9wfHcTOCsbvdjXkZ4VY9DZ28MootCUA+zuL5WpU6fiyy+/xEsvvQQfHx/4+/uD4xp3ZMaMGYMxY8ZoPhcXF7fLJmdn53aXNTb8Fx+BbpSAe+YVlFRUAqjUex26tAcfHA46fRTF586Aefrq3QZzwpj3h+pYDODpizKJHLiXOjkJ2DP/geqD11C0aim4F17TrJ90L3SkZ8UYdIb28PDwaHafUYaPnJycUFJSovlcUlICBwftsVNLS0ssWrQIa9aswfPPP4+Kigq4uroawzyzhs6eBp08APbADDD/QJPawsKi1DYJUUh6gyrKgPTLjdY6ai8sqLc6wdKFM6BdX+vlnAJdC6OIQkBAAPLy8lBYWAilUokTJ04gKipK65jq6moolUoAwN9//41evXrB0tJASzd0EKiqAvy2T9RRJZNnmdocMAcnwNtfmK+gR+jsaYB4sAj95MAAAG7YA2AjJoD2/gz+5EG9nVega2CU4SORSIR58+Zh5cqV4HkeI0eOhLe3N/bt2wcAiI6ORk5ODtavXw+O4+Dl5YVnn33WGKaZNfTd50BVBbgX3wATN70WjrFhYf1Bf/0Iqq4Cs9LjMt1dFEqMBZxcAe/uej0vm7UAlJcN+no9yM0DrHuwXs8v0Hkx2jyFyMhIREZGam2Ljo7W/B0UFIS1a9cayxyzhxJOgE4fAZsyG8xHv18Y9wILjwL9+T0o+QzYgGGmNqdDQ3U1wKWzYCPG691xz8RicM+8DP6dJeA3/A/cq++re3oCAq0gzGg2Q6iyHPw3nwI+AWDjZ5jaHG38AwFrW0BYIO/euXAGUDaA9W1fKGprMBtbdXKeulrwG94B1SsMUo9A50IQBTODiMBv/xSorVZPUhMbrTOnE4wTgfWOVM+bEBK93BOUGKsW2MBeBquDefqCW7AYyExTDyUJS2EItIIgCmYGxR0FzpxQDxuZa9hnWBRQVQlkCMs2txdSNoDOx4P1GaCXsNGWYH0HgU19HHTqMGjvLoPWJdDxEUTBjKDyG6BvNwL+QWDR00xtTrOw0EiA44TQ1Hvh8nmgtqbds5jbCpvwMFjU/aBdXwu5MQRaRBAFM4GI1OGn9Qpw//gXmMiwb4/3ArOyBgJ6CqGp9wAlxgIyORDS1yj1McbA5r4IePuD3/QeKC/bKPUKdDwEUTAT6ORB4OxpsKmPg3XzMrU5rcLC+gPZGaAbJa0fLKAF8Tzo7CmgdySYRGq0eplMDm7Rq4BYAn7926DqKqPVLdBxEETBDKDSYtCOTUCPXmBjJpvaHJ1g4TdnNwtDEW0nIxUov2GwqKOWYE4u4BYtBUqKwH++GqQSggUEtBFEwcSoh43WAyoluH/80+BOR73h4QM4ugii0A4o8SQgEmmE1diwHiFgjy8ELiaBftxqEhsEzBdBFEwMHdsPXDgDNv1JMNfmF6kyN9SJd6KAS2dBDQ2mNqfDoEm7GRwOZmm6GeHc/WPBRk8GxfwK/niMyewQMD8EUTAhVFwA+n4LEBwGNmKCqc1pMywsClDUAakXTG1KxyE3GyjM0+taR+2FPTwP6NUHtH0DKP2Sqc0RMBMEUTARdO0q+DVLAQDc3BfBmlgm3OwJDgckUmEIqQ1Q4kkAAOtrBqIgEoF75j+AgzP4T/8HKi0ytUkCZkAH/Cbq+NCZE+DffRngCdz/vQPm7GZqk9oFk8mAnuGgc3HCTFkdoaRTQPdgMHvzWIeIWdmol8KoV4D/5B2QwvyWwiCeB+VfV2eoEzA45rWGQieHiEB/7AT98i3gHwRu0TIwe0dTm3VPsLAodU+hIAdwN/9QWlNCJUVAVjrYQ0+a2hQtmIcPuAX/B/6Tt0FfrQWe+j+TZdYjIqC4AJSZDmSlqX9fu4KGujqU2DmDe+olMFs7k9h2C1tbW5PWb2gEUTASpFCAvloLijsKNmgE2JznjRqjbihYeBToW3XiHSaIQotQUiwAGG0Wc1tgffqDTZsD2vUV4OkLNnGmweskIuBGMZCZDspKvykE6UD1zcyCYjHg3R1s4AgcUUlxuawK+Guvwe1qDTc3N0yfPh1iM1uXTF90zqsyM+hGCfhPVgLXroA99CTYAw91mhzHzMkV8PBR9xaip5raHLOGEmOBbt5g7p6mNqVJ2AMPAdczQbu3gzx99D6Pgspv3BSAmz2AzDSgsly9UyRSi1HkYMCvB5hvIODpAyaWoKGhAVc2b0Z3KzlCLpwEmzQLLKCnXm3TlYqKChw6dAgnTpzAsGGdc+l4QRQMDF1NAb/hHaCuDtxzr4L1GWBqk/QOC4sCxfwCqq0Bs+ja2fKag6oqgLRksAemm9qUZrJOeqkAACAASURBVGGMAU8+DyrIAb/5Q3BLV7d7UUaqrACy7ugBZKYBZTdnvzMO8PBWR6/59QDz7QF4+zfbc87IyEBDQwP6TpiAbteTgb92gHtzg8nutdraWpw6dQp+fn7w8fExiQ2GRBAFA8LHHgJ9tQ6wdwS3+E3zXfX0HmHhUerVNy8mAf2GmNocs4TOxgE8b5ZDR3fCpDJwi5apk/N8shLcsvcAZ+cWy1BNFZB15aYApAGZ6UBJ4e0D3DzBgnvf7gF4+4PJLXS2KSUlBVZWVvD08QHmPA/+fy+Bdn0N9phpsjOOHTsWqamp2L9/P2bPng0LC92vpSMgiIIBIJ4H7d4G2vMTENQb3LOvgNl0YudUQC/A0gp0Lg5MEIUmoaRYwMEZ8O1halNahTk6g1u4FPx7y8B/9i7o7U80+6iuFrh2FZSVrh4KykwDCnNvF3Z2A/MLBEZOUPcAfALALK3abUttbS2ysrLQt29fdU/GPxBs9CTQ37+BBg4H62G4XBTNIZVKMW7cOHz//fc4cOAAJkyY0GmGgwFBFPQO1dWA3/yBenG7odFgs58xm/zKhoKJRGChkaDz8SCe75hzLgwIKeqA5ESw+8d2mC8PFtAT7InnQFs/Rvl7r4HnOPUwUP514Fb48U2RY0NGgfkFAr4BYNb6ffm5cuUKeJ5HUFDQbdsefAx05iT4bZ+Ae+1Dkzxfrq6uGDRoEE6cOIHLly+jVy/ji5OhEERBj1BRvtqhnJcN9sjTYKMmdpgvgXsmPAqIOwpkXVGn7BS4TfIZoKHe7IeO7oYbMhp8zjUo9v0M2NgBfoFgUfdBLQA9wOwcDG5DSkoKHBwc4OLiotnG5BbgHl8Ifu2boL9+Apv0iMHtaIrIyEhkZmbi0KFD8PDwgJ2daUNl9YXOovDVV19h+PDh8PPza1dFSUlJ2Lp1K3iex+jRozF1qnakSk1NDdauXYuSkhKoVCpMnjwZI0eObFddpoBSL4D/dBXAq8D98w2wkAhTm2RUWGg/EGOg83FggihoQYmnAEtrIKi3qU1pM2zGXDg/9hRKFA1Gf8GprKxETk4OBg4c2KhuFhYF1n8o6I/vQVH3myQcmuM4REdH49tvv8W+ffswffp0cJ2gl6zzFahUKqxcuRJLlizB7t27UVKi+zr6PM9jy5YtWLZsGT788EMcP34c169f1zrmr7/+gpeXF9asWYMVK1bg66+/hlKp1P1KTAh/dB/4D14DrG3ALX2vywkCoE4Sj+7BQja2uyClEnTutHoegBknTmoOxhg4W3uT9HjT0tTpXoODg5vczx5ZAEhl4Ld9AuJ5Y5qmwdbWFiNGjEBeXh4SEhJMYoO+0VkU5s2bh40bN2L27NnIzMzE4sWL8dZbb+Hw4cOoq6trsWx6ejrc3d3h5uYGsViMIUOGIC5OO2sXYwx1dXUgItTV1cHa2trsVZdUKvA7NoG+Xg/0DAe3dI3ZxqAbAxYWpQ5DLL9halPMh9QLQE01WMRgU1vS4UhJSYGbmxvs7e2b3M9sHcBm/ANITQaZcKXX4OBgBAYG4tSpUygoKDCZHfqCUTsXrcnOzsbatWtx7do1SKVS3HfffZg5cyYcHRsv2xAbG4ukpCQ8+6w6hOzIkSNIS0vD/PnzNcfU1tZi9erVyMnJQW1tLRYvXozIyMhG54qJiUFMjPoGWLVqFerr69tjPsRi8T31RPiqCpS/9xrqz8bBcvIsWD/5HJio47po7rU9AKAhIxWl/54L2+eXwWL0JD1ZZhr00R4AUPH5+6j9+3e4fr0HTCbXg2XGR19t0RaKioqwbt06PPDAAxgypPmINiLCjddfgDIjDU7rvoXIwfBrSjXVHrW1tfjkk08gkUiwcOFCSKXmvVpBS/a16VuspqYGsbGxOHr0KLKysjBw4EDMnz8fzs7O+P333/HOO+/gvffea1SuKd25uzt69uxZ+Pr64vXXX0dBQQHeeust9OzZE5aW2hNUxowZgzFjxmg+FxcXt+USNDg7O7e7LOXngF//NlBcADbneSiGRkNxo6xd5zIX7qU9bkHWDoC9EypOHEJ1n47lVL0bvbQHz4M/eQgIiUBJZRVQ2THTX+qjLdrKqVOnAACenp6t1k2zngL990WUbHhXveqrgWmuPUaPHo2ff/4Zv/zyi9n7Qz08ms/dorMovP/++zh79ix69eqFsWPHon///pBIboeCzZkzB3Pnzm2yrJOTk5YPoqSkBA4O2pELBw8exNSpU8EYg7u7O1xdXZGbm4sePcwrrpsuJoLfuBrgROD+/RZYUKipTTIbGGNgYf1AcUdByoZOH4rbKlnpQFkJWMQTprakQ0FESElJgbe3N6ysWp/jwNw9wSbNUi/PMWgkWJ/+RrCyMd7e3oiIiEBiYiL8/Pzg7+9vEjvuFZ0H7QMDA7F27VosXboUQ4YM0RIEQO2J37RpU5NlAwICkJeXh8LCQiiVSpw4cQJRUdqpCJ2dnXH+/HkAQFlZGXJzc+Hq6trW6zEYRAT+79/Af/xfwMEZ3KvvC4LQBCw8CqirBdIumtoUk0OJsQDHmexLqqNSWFiI8vJyrbkJrcHGTQM8fcF/+ymorsaA1rXM4MGD4eTkhJiYGNTUmM6Oe0FnUQgPD280jlZcXIzMzEzNZ5lM1mRZkUiEefPmYeXKlVi8eDEGDx4Mb29v7Nu3D/v27QMATJ8+HampqViyZAneeustPPbYY2azRC0pG0DbPgHt2ASE9wf3yrsdNgeCwenZBxCLhcQ7uCkKQb3BrGxMbUqHIiUlBRzHtWmUgIkl4J54DrhRAtr9jQGtaxmxWIxx48ZBoVDg77//7pB5RnQePlq3bh3+8x/t8TqlUon169c36Ue4m8jIyEaO4+joaM3fjo6OWL58ua7mGA2qrAD/2f+A1GSwCQ+DPfiYMGO3BZjcAggKU4vCzPmtF+ikUN51IP862MiOl2bVlPA8j9TUVPj7+zf7ktkcLKAn2IgJoAO/q5fA8Ne9p6FPnJ2dcd999+Ho0aNITk5G794da36Kzt9uxcXFcHPTfjt2d3dHUVHnTeFH1zPBr/w3cDUVbMEScNOeEARBB1h4FJCfAyrMM7UpJkOTO8EM0m52JHJyclBTU9OmoaM7YdOeAOwcwX+1DmTCeU59+/aFt7c3jhw5grKyjhWEovM3nKOjI65evaq17erVq40cxp0FSjoFftXLgFIJ7j//AzdwuKlN6jCwMLW/qCsPIVFirHopCEeX1g8W0JCSkgKJRNJuJy2zsAT32LNAThZo/249W9cGOxjDmDFjIBKJsHfvXqhUKpPZ0lZ0FoWJEydizZo12LNnD86cOYM9e/bgvffew6RJHTse/W6ICPyeH9U5ENw91Q5lE3VDOyrMtRvg7tllZzfTjRIgI7XDrXVkapRKJdLT09GjR497ymrG+g4EIoeAftsBunMFVyNjY2ODUaNGoaCgoNFkXXNG55YfM2YMrKyscODAAZSUlMDJyQlz5szBoEGd58anhnrQV+tApw6D9R8KNvdFMGnbxjUF1LCwKNDBP0B1tW1aO78zQEnqGHsWKcxibgtZWVmor69v99DRnXCPPg3+0lnw2zaoQ8dNtDBlYGAgMjIyEBcXB19fX3Tr1s0kdrSFNsnx4MGDMXhw57zRqaxU3TvISAWb+rjaqdxVVjg1ACwsCrT/F+DyWUDPaR3NHUo8Cbh5AkLO6jaRkpICCwsLeHt73/O5mL0j2PQnQds3gE4cALtvtB4sbB/Dhw9Hbm4u9u7di9mzZ5v9bOc2eU3LysoQHx+PgwcP4sCBA5qfjg5lpYNfuQTIvQZu4VJwE2cKgnCvBIYAcosuN4RE1VVA6gWwiEHCPdQGFAoFMjIyEBQUpLc1z9jQaKBHCOiHL0AVpnP2ymQyjB07FpWVlThy5IjJ7NAVnXsKp0+fxrp169CtWzdkZ2fD29sb2dnZ6NmzJ0aNGmVIGw0KH3cM9OVHgLUduJffBfPumLMQzQ0mlgAhEaDzCSCiLvMFSefjAJVK8Ce0katXr0KlUull6OgWjOPAzXkO/Jv/BO3cAvbUEr2du614enqiX79+iI+Ph5+fn9mt1HAnOkvyzp07sWjRIqxevRpyuRyrV6/G008/3WGnchPPg//lG9DnqwGfHmqHsiAIeoWFR6mTtWdnmNoUo0GJsYCdI+An5JRoCykpKbC1tYW7u7tez8u6eYONfxh0+jDovGmXth44cCBcXFxw4MABVFdXm9SWlmjTPIW7/QnDhw/vEN2huyFFHcrXLAf9vhPsvtFqR5Rt08vzCrQf1rsfgK4Tmkr1CuDCGbCIgcJ8ljZQU1OD7OxsBAUFGaRHycbPALp5g//mU3VqVBMhEokwbtw4KJVK7N+/32xnO+t859ra2momYbi4uCA1NRUFBQXgTZTc4l6g+ONQnD4CNnM+2JMvgkm6+MJtBoLZOQC+PbqMKOBiElCvEIaO2khaWhqIqNlkOvcKk9xcAqOkEPSL6ZbAANTzve6//35cu3YN586dM6ktzaGzT2H06NG4fPkyBg0ahIkTJ+K///0vGGMdcp4CGzIKDuGRKLfpnBPvzAkWHgX6fSeoskKdna0TQ4mxgIVVh0y7aUpSUlLg7OwMJyfD5UJggSFgwx8AxfymXgLD13Rj+mFhYcjIyMCxY8fg5eVl0OtuDzr3FKZMmaKZkzB8+HB8/PHHWLVqFR55xDRJs++FynoeseQIhbLj9XI6GiysP0AESu4cqQqbg1QqddrN8ChhyfA2UF5ejvz8fL06mJuDPfQkYGsP/uv1IBPOML4121kikZjlbGedRIHneTzxxBNoaGjQbHN2doaXV8eMwz59vRKv/HYJc35Kw7tHc3AkswI1Deb1j+k0+AYAtvZAZw9NTb8IVFUKaTfbSGpqKgAYRxQsrcDNfhq4dhUU86vB62sJKysrjB49GsXFxYiNjTWpLXej0/ARx3Hw8PBAZWVlk+k2OxrD/ezg7+6Efck5OJVdiRPXKiHmGPq4W2KQtw0GeFnDXt5xU2uaE4zjwHr3AyXFglSqDpm8XhcoMRYQS4DQCFOb0mG4lUzHw8PDeMvkRwwG+g4E/foNKHIwmIt+o53aQkBAAEJDQ5GQkABfX1+zeckWrVixYoUuB9bW1mLHjh2QSCSorKxEUVERCgsLUVhYaNJkOJWVlW0uI+IYgjycEOrAYUovR0R0s4KlRISLRbU4mFGBXy+X4lx+NWoaeDhaiGEl7ZxfZHdiaWlpuKQgvAp08iBYrz5gTuaTOKkl2tIeRATavgHo0QvcENPNnDUUhro3SkpKcPr0afTr16/RCsyGgjEG1iMEdHgPKDsDbNCINkc86bM9vLy8kJaWhitXriAkJOSe1nxqCzY2zef40NmCW8lwfvjhB63tjDGsX7++naaZHo4x9HKxRC8XS8yNcEHGDQVir1ci9loVNicUYnNCIQIc5RjsbY3B3jbwshPWQmozIRGASAQ6Fw/WGZ2w164CpUVgUx41tSUdilvJdAIDjTungzk6gz00B/TtRtCpQ2CDTJdPWSqVYty4cfjhhx9w6NAhjBs3rtUyRAScOQl4+oAZYCkVnUXhk08+0Xvl5gZjDN0d5ejuKMfscBfkVtQjNrsSJ7Mrsf1sMbafLYaXrRSDvG0w2NsGAY6yLjNT915gFpZAYKg6NHXGXFObo3co8STAOLDwAaY2pcNAREhNTYWPjw8sLIy/YCIbPh506jBo5xZQaD+TRsa5u7tjwIABOHXqFPz9/Vv0r1B+DvjvPgcuJoKNmAD22LN6t0eYYdMCHrZSPBTqhDUP+OGLaQF4OsoNjhZi7LpYgiV/ZeKp3VewOb4AyQU1UPHmORHFXGBhUUDuNVBJoalN0TuUGAsEhnT6kFt9kpeXh8rKSqM4mJuCcZx67kJtDeiHLSax4U769+8PNzc3HDx4sMkhcVIowP+8Dfx/XwAyUsAeeQrskacMYovOPYWFCxc2u+/TTz/VizHmjJOlBBODHTAx2AEVChXirlfiZHYV/korw28pN2AnE2GAl3qIKdzdEhKRoLd3wsKj1AuTnYvvVCkqqSAXyL0GNmuBqU3pUKSkpEAsFqN79+4ms4F5+oI98BDoj+9Bg0aAhZguSIDjOIwbNw7fffcd9u/fj2nTpoExph4qSjoFfscm9RDloJFgM+aqJ4YaCJ1F4YUXXtD6fOPGDfz555+477779G6UuWMrE2F0gD1GB9ijpkGFxNxqnMyuxLGsSuy/Ug5LCYcoD2sM8rFGZDdrWEgEgYCbJ+Dirh5C6kyicCvtpjCLWWdUKhXS0tLQvXt3ky8jzSbOBMUfB7/9U3BvrANrY15ofWJvb49hw4bh77//RmJiIiK83MF/twm4kAB4+oJ76R2j+OR0FoWQkJBG20JDQ7Fy5UpMmND6Q56UlIStW7eC53mMHj0aU6dO1dr/66+/4ujRowDU8yKuX7+OLVu2wNraWlcTTYKlRIT7fG1xn68tGlQ8zubX4GR2JU5fr8KRrApIRQx9u1lhsLcN+ntaw0bW+SOZmoIxBhbeH3RkL0ihMOnDp08oMRbw6d5hoqrMgezsbNTV1Zls6OhOmEQK7onnwL+3DPTbd2Am9nmFhIQg48oVnDh2DJ5X4+CsqgebNR9sxEQwI0Um3VMtYrEYhYWtjxHzPI8tW7Zg+fLlcHJywtKlSxEVFaUVlztlyhRMmTIFABAfH48//vjD7AXhbiQiDlGe1ojytIaKJ1wqqsXJ7ErE3hQJjgG93Swx2NsGA72s4WTZtWa+srAo0N+/ASnngPD+pjbnnqGyUuBqihB11EZSUlIgk8ng6+tralMAACy4N9jQaND+3aABw8B8TDekhbOnMSL2D+xw6oH9AVGYNXs2JE7GzfOtsyjs3LlT67NCoVB3cSJaH4dLT0+Hu7u7JhZ5yJAhiIuLa3ayxvHjxzv8sJSIY+jtZonebpZY0M8V6aV1iM2uwsnsSmyMK8DGuAIEO8sR6WENd2sJ3KwkcLGWwEEuhojrpBFNQb0BmRx0Ph6sM4jC2dMAkTCLuQ00NDTg6tWrCA4OhsiMJjKy6XNBZ0+D/3o9uGVrwDjj2kZF+Wq/wbk4WHj4YMzAAfj1zFmcTL6EYcPMVBRKSkq0PstkMkyaNAnDhg1rtWxpaanWok9OTk5IS0tr8liFQoGkpCTMnz+/yf0xMTGIiYkBAKxatQrOzs66XoIWYrG43WXbg4sLMDgYWAwgs7QGh9NLcPhKCb47V6xtF8fgZiODu40M3WzlcLeVwd325t82MjhbyyA2gGgYqz3K+vRHQ/IZODk5mXU4ry7tcSM5ASp3TziFR5r1tdwr+rw3zp8/j4aGBvTv39+oz1+rODuj7uklKH/vNVieOgSrybOaPVSf7UEKBap3b0f1T9vARGJYzX0elhNnwk0sRr5YhtOnTyM8PNyoSXl0FoVFixa1u5Km1g1v7iFKSEhAcHBws0NHY8aMwZgxYzSfi4uLmzyuNZydndtd9l6xBjCxuwUmdveCQsmjsLoBRdUNKKhS/y6sbkBhtQLHS6pxo1apVZZjgLOlBK5WYrhaS+BqJYGLlfq3q5UEzlaSdomGsdqDDw4HnT6K4nNnwDzNY/igKVprD6qpBn8uAWz05EYvTJ0Nfd4b8fHxsLKygpWVlcmev+agoHAgLApV32xETVBYs34ifbUHnYtT9w6K8sH6DwV7eB5qHZxQezNFQb9+/ZCWloaffvoJs2fP1ut8Dg8Pj2b36SwKu3fvRu/evbUUKz09HcnJyXjwwQdbLOvk5KT14JSUlMDBoemQquPHj+P+++/X1awOj0zMwdtOBu9mZkrXq3gUVSs1YnGncJzNr0FpjRJ3Si7HAEcLsUYkXK21RcPFSmzScFkWFgUC1KGpZiwKrUHn4wGVUog6agN1dXXIyspCnz599JaHWZ8wxsA9thD8G8+po5FefN0gPUAqyge/czNw9jTg7qVO8tWrT6PjJBIJxo0bh++//x4HDx7E+PHjjdIj1VkU/vzzTzzwwANa27y8vLBmzZpWRSEgIAB5eXkoLCyEo6MjTpw4gRdffLHRcTU1Nbh48WKj8NeujFTEwdNWCk/bpkP3GlSE4pqGJkXjYlENjmQpcee8OgbAwUIMFyu1H0MtGmL0qhXDnilhZ+CFAJmDE+Dtr85lPH66QesyKImx6tVfuxsmMUxnJD09HTzPGyyZjj5gTi5gUx8H7dwMijsKNqD14XFdoYZ60N5doD9/BDhOPd9g9OQWl1p3dXXFoEGDcOLECVy+fBm9evXSmz3NofM3gFKpbLRYk1gsRn19fatlRSIR5s2bh5UrV4LneYwcORLe3t6a9ZSio6MBAKdPn0afPn0gl8vbcg1dGomIoZuNFN1smhYNJU8oqWlAUbVSPSxVdWt4qgEpJbU4fq0CKgJwugAA4CAXwddBDj97GXztZfCzl8HbTqrX3gUL6w/660dQdRWYVceKMANuPtwXzoANHCak3WwDKSkpcHBwgIuLcR2nbYWNmqheAmPHJlBoBJhV84vH6QpdSFAvT1GYBxZ1P9jD88AcdfNLREZGIjMzE4cOHYKHhwfs7Ozu2Z6W0FkUunfvjr1792LixImabfv27dN5RmJkZCQiIyO1tt0Sg1uMGDECI0aM0NUkAR0Qcwxu1lK4WTctGiqeUFqrRBWzwLmsQmSW1SGrTIE/UmrQcLOLwTHAy1Z6UyTk8HNQC4azpbhd3VkWHgX683tQ8hm9vokZjUtnAUWtMHTUBiorK5GTk4OBAweavVOecSJwc54H//Zi0A9bweY2HtXQFSopVA8VJcYCbp7gFv+3zTOnOY5DdHQ0vv32W+zbtw/Tp0836PCbzqLw5JNP4u2338aRI0fg5uaGgoIClJWV4bXXXjOYcQKGR8QxuFhJ0MvZHv6Wt53aKp6QW1mPzBsKZJUpkFmmQEpxLY5m3V6XxUrKwddOBj+H22LhYydrfQa3fyBgbQucjwc6oChQYiwgtwCCw01tSofhVrShOQ8d3Qnz9gcbNw205yf1Ehg92/a/poYG0L6fQX9+D4CBPTQHbMyD7c4Hb2trixEjRmDfvn1ISEhA//6GC+nWWRS8vb3x8ccfIyEhASUlJRg4cCD69esnDPV0UkQc0zjAh96xvbpehWs3RSKzTIHMGwocvFqBWmWZ5hh3a4m6V+Eg0/Qu3K0lmvkXjBOB9Y4EXUgA8Sqjx4TfC8SrQGdPg4VFtfsB74qkpKTAzc0N9vb2pjZFZ9ikR9RLYGz7BNwba8Gkus3Cp+RE8N9uBApzgcgh4GbOB9PDBLTg4GBkZGTg1KlT8PHxMVgOCp1FobS0FFKpVGtSWVVVFUpLSztFNrauDM/z4Hnd8lVbSUXo5WqJXq6Wmm1EhMLqBmSWKZB1Qy0WWWUKxOVUaZzcUhGD7x1+Cp+AwfBJOA37jDQgoKchLsswpF8GKsvVGbz0TG0Df0dIcgPK65Rws5YaxK9jTEpLS1FUVIShQ4e2frAZwaQy9RIYH7wG+uN7sGlPtHg8lRaB37kFOHMCcPUA988VYL0jWyzTJnsYw8iRI5Gbm4u9e/fi0UcfhcQALyY6i8KaNWuwcOFCrfkDpaWl+Oyzz/DOO+/o3TAB40BE+PXXX8FxnGaZkbbC2G2/xUCv2045hZJHdnk9ssrqND2LuOtViLlSDsABuG8FHE/VwzcrG343exZuVhLYycWwk4tgKeHMbvxZnXZT3K6HvapehUKtuSh3/q1EpaL5POEcAzxt1QLhZy/X9MTa69cxJsbMw6xvWK8+YENGq6OG+g8F8/JrdAwpG0D7fwX9vgMAgU19HCx6mkF6knK5HNHR0fj5559x8uRJnSYPtxWdRSE3Nxc+Pj5a23x8fJCTk6N3owSMR2JiIq5duwYAKCsr02v3Xibm0MNJjh5O2kOMZbVKZJYpcPXnXbjGbJBl3we/pdRAeVdOCjHHYCcXwV4ugp1MLRS3BMNeLoad7PZnO7kIUgO/SRORelXUXn3ViYPu2lehUDX6or9TBGoatHtjMpHan+NmLUGgk4VmIqLbzbkltjIR8ivrNcN0WeUKpBTXaft1JFyjoTofeyksJeYxJHcrD7O3tzesrKxMbU67YA//A3Q+Xr0Exivvau2jS2fBf/sZkJ8D9B0EbtZ8MGfDphb19vbGyJEjG30f6wudRcHW1hb5+flwd7+d6Do/P7/FXJ8C5k1RURFOnDgBLy8v5OTkICUlBQMHDjR4vfYWYvS1ECPc3xK06wtwj2wFb+eInMp6FFc3oLxOhXKFEuV1KpTVqVBep/47u1yBcoUK9aqmExpZSji1QMjuEI6bgnH3NmupqE1rTPFEuHElAwX1MhQFjUJxckmjmeiKu+yylHCaiYOhrhaNJhLaykStvuV72cngZSfD/XfM87vTr3MrCEDt17ktOrf8OrcE426/jrEoLCxEeXk5oqKijFqvPmHWtmCzFoA2vw86tAeYORdUWqzODxJ/DHBxV090CzPeNYaFhRns3DqLwsiRI/H+++/jkUcegZubG/Lz87Fz506MGjXKYMYJGA6lUom9e/dCLpdj/PjxiImJQWpqKgYMGGC04QgWHgXa9RXofDxEw8bBx04dvdQSRIQ6JamFQqFC2U3BKNf8VqFMoUR+VQNSimtRoVChqaR4HANsZCLYy+4QjpuCYSMVQXW1BpmF5Zov/qJqpTpEN/J5oARASRFspBxcrSXwspMi0sPqjlnj6kmB1lLDvK235NfJuiMAoCm/jo8mWuy2f8fWgBMWb+VhNubaPYaADRgGOnkAtGsbqngl+N3fAjwP9uBssHEPgUlMmxdCn+h8N0ydOhVisRjbtm1DSUkJnJycMGrUKEyePNmQ9hkEnidkpFfB2o7MfjzWUJw4cQKlpaV48MEHYWFhgfDwcPzyyy8oKiqCq6uRcgN4+ACOLuolI4a1nrAcUPsvLCQMFhIp3HXopPJEqFKoUKa4SzhucmvUXgAAIABJREFU/X2zR3KltA7ldSpUN/Bwghg3oISNXARXKwn8HeQY6CWB85HdcGH1cH/yKbhYic1miAbQ9usMuMuvc72iHpk3mvLrqHGwEN/0VdzuWXjZ3rtjm+d5pKamwt/fH7J7zJ9RX89DUUuwsTNNm2uWwFjxAqp3fgH0GQBu1gIwF/fWC+sZIsKNEhVkMgYrG/23h86icMsReaczkud5JCYmNpqUZu5kZ9TjXHw5fAOk6B1pAa6zLlXdDNeuXUNSUhL69OmjWdM+JCQEv/32G1JSUowmCurEO1GgEwdADfUGedviGIOtXKx+G26hF6JsIFzPqkdGmgJVFTy8u1uhT9RtJy4V5YNP/h3s4Xng7DtOgiCZmEOAoxwBjk37dTLL6jS9ijv9OiIGeNmqRSLMuw697Fmz63M1R05ODmpqau7ZwVxRpsLpo1WorSE4OIvgHyhDNy+J0Z9bdnOYyNbKCpVexs+5oFIRcq81ICNNgfIbKvgGSBEeZdl6wTbSrn5jVlYWDh8+jGPHjoHneWzevFnfdhkUn+5SEC/D+TNlqKnm0W+wFSTSriEMtbW12L9/PxwcHLTCiy0sLODn54fU1FTcd999RluwjIVFqcdpUy4Aegzf05XqShUy0uuRnaGAsgGwcxDBy0+C7KvVsLGVIyBY/WVKiZ0r7eYtv07fbredv8q7JixmldXhUlENjmRVAAA8bKQY5G2NgV42CHKWg2ull52SkgKJRAJ/f/9221mY34CEE9UQiRiCe8uRnVGPMydrILdg8A2QwTdACpnceKG6LDgMMmdnVBpxhdfaGh5ZVxTIulKPegXB2pZDWD8LePkaZshKZ1GoqKjA0aNHcfjwYWRlZYExhn/84x8d0qfAGEPUYGcwToHzCbU4/nclBgyzhqVVx4wD1xUiwoEDB1BbW4vJkyc3WssqKCgIV69eRW5ubrMJkPROcDggkaoT7xhJFIgIRQVKZKQqUJinBGOAh7cEfoEyODipu+Mcp8Sls9WwdxDDyVWsjjry8jPJcIGxEHOsSb+OSmaDv85dw+nrlfjlUil2XSyFvVyEgV7qDILh7paNhpqUSiXS09MREBDQ6D7Tlcx0BS6cqYWNLYcBw6xhYckhsJcMhflKZKQpkHKhDmkX6+DhI4F/oAz2jsZJV2kMiAilxSpkpimQd70BRICbhxj+QTI4uxo2DLnVVoyNjcWhQ4dw9uxZeHp64v7778dLL72EV199FYMGDTLI5Alj4Rsgg6U1h/jj1TgWU4n+91vBwanz3Fh3c/nyZVy5cgVDhgxpcojI398fEokEKSkpRhMFJpMBPcNB5+NBjzxl0Jtd2UDIzlQPEVVX8pDJGYJCZfANkEFuof2lNnSUK3bvzELCyWoMHUyQpl8Cm9R84pXOjJuNDBODHTAx2AFV9Sok5FTh1PUqHM6swN70sv9v78zDoyrvxf85s89kXwiBLJCEZFjCFgJBFmUJQUQvSLHWqq2Fa21rr639uYDaYq/FpV6r9V7b2hbXUmtbaxUtBQJiZIewb5MEQgiQAEnINmtmzvn9MWQkZg/JTJb38zw8ZHLOec933syc7/t+VwwaFZOGBpEVH8ykuGCCdWpKSkpwuVxdKmuhyArHDzk4XeAkZoiGSTcEodE2ZsNLDB6qZfBQLXW13odm6RkX5840EBGlZniqnqHxWlTqvrnz97gVzp91UVzoorbag1YrkZymZ/gIHaZg//hT2n0CvvzyywQHB/Pwww8zZcoUf8jkVwYN1jIjO4Q9eVZ2fFbPxCwTQxP6TyRBIzU1Nb4qi635gLRaLcnJyRQVFXHTTTd1eYXXWaRxmV5nc/l5GNL9yqi+7urDo9iF2w3hkWomZpkYkqBF3crDQ6dXM3l6EF9sqiN/Wx1ZqNBM6B+mo+shWKfmpqQwbkoKw+WROVJuY/e5enafq2P72TrUEowdbGJY5VH0BgMJCQmdGt/tVti/y8rF826SUnWMntC6zy8kVM3YSSZGjjVSesbFmUInB3bZOG740rT0VWXfW7FZvzQRNbgUQsJUjMs0EjdMh0bjXwWnfvrpp59u64RBgwZRXV3NunXrOHDgAC6Xi5iYGHJzc5k3b17Aax/V1dW1f1ILmEwmbDYbAHq9irhELZWX3JwucKFWQ0R0+zHkfQVZlvnkk0+w2+0sXry4xb9Z43xoNBqOHz/O4MGDW22E1O2EhKPkfgzRMUgp3VMvXlEULpW7ObrfzrEDDmqqPQyJ1zJ+sglzupHQcHWbjkqTyYRHdmAKVlFcosIdHEXMvN5f4bMnuPa7ci1qlcTQUB2T44NZNCqSjCHBBOvVFF6qI+j8AUo0sWyqMFHtcBOiU7ebl+Gwy+z63ErlZQ/pE42Y040dmm+1WiIiSsPwVB3hURocNpmzp68GDdR5MBhVGE3dpxxam4/OoigKlZc9HD9o53C+naoKDzGxWsZmGhk1zkh4pKbHnOlt5Ze1uxRsLGd9+fJlPv/8c/7973/zzjvvAN5s2BtvvLFXdlHqLHqDihtmB3Nwj40Thx1Y62TGTjL22W3oteTn51NWVkZOTg6hoaFtnpuQkIDBYMBisXS4LPr1IkUNgrhhKIf3Qc7t1zVWg0vxrRqt9V4TkTndQGJy11aNQ2PcVJVu5ExCDpFnG4jrIedeX0clSYwcZGTkICNTDJVsOiszMX0UB2ph7aEK1h6qYEiIlqz4EKbGB5MWbWySSFdzxRth1NCgMGVGEIOHdt4sLUkSg4doGTxE22R3eL6kgfDIq6alNnaH/sLtVjh/NdKtrkZGq5MYYdYzbIS+V/g1O2wfGDRoEEuXLmXp0qWcPHmSzz//nLfffpv33nuP119/vSdl9BtqtUTGVBNBwQ4Kjzu9kUnTTeh0gf9DdZWLFy+ye/du0tLSOmTfVavVpKamcuLECVwuFzqdfx6C0thMlE3/RLFZkUydL4dwrX3Z44aIKDXmdJM3dPE6HgLKkf2MLHiP2pE3cWgvhISpCQ3vPfkJvRGLxUJoaCh33WDmm5JEpa2BPee8fohPLFX880QVYQY1U+KCmZoQwhBFy6HddrRaielzQgiLuP75DQ5Rk57xpWmpuNDJwd02jh+UGJaiY/iI5n6knsZm9XCmyMXZ014TUWi4ivGTjcQl6lD72UTUFu0qhcOHDzN69Ogm9uWRI0cycuRIli1bxt69e3tUQH8jSRIjxxoJClZzaJ+N7bn1TLkxiCA/OXm6k4aGBjZs2IDJZGLWrFkdNn2YzWaOHDnC6dOnGTnSPxVMpbGZKP/+AE4chEnT278Ar0PyUrmb0wVOKi66Uano/kiUg7tQBQeTMTuSL3Jt7NtuZea8kAETwtxZbDYbpaWlTJo0yfd5izJpWZAWwYK0CKwuD/svWNl9ro5tJXWcP93AVHUITo1MtFmNupvD7jVaiaRUr6P28kU3ZwqdFB53UnTCyZAE72clIqrnTMWKolB5yU1xoYvyCw1IQGyclqQ0PZG91ETd7jdn3bp1/PrXv8ZsNvu6pzWWytZqtUybNq3HhQwECUk6jEGNkUn1TJ4RRGR034pM2rZtG9XV1dx+++2d8v0MGTKEkJAQLBaL35QCKSMhOAT5jZdhxxZvv4KxmS3WoW9wyZwtdnGm0IXNKmMwSpjHGhiW3L0x64q7wRsqO2k6xiAtk6YFsfOzeg7ssTJ5elCv/EIHmsLCQhRFaXVXGqRTM3N4KNMTQziy38bZUw3YTR42uK5Qsc+NKr+c9MEmpsaHMCU+mEFB3RPdKEkSMbFaYmK1WOs9nCl0cbbYyYWzDYRFeBPihiZ2n2nJ7VY4d9WMWVcro9NLjBipZ/gIfbf6N3qCdp9yTz75JE6nkyNHjnDgwAE+/PBDTCYTEydOJCMjg7S0tH7hU2iJ6BgNM7KD2ZNnZedn9UyYYuozNuXi4mKOHDlCRkZGpyNAJEkiLS2N/fv3Y7PZMJm6P2uy2T3ValQPP4OybSPK4X0oh/eiAMQNQ0qfhDQuk7roNM6cdnPujAuPByKj1YwaZyK2p7JbTx4Guw3patRR1CANo8cbOHbQQdFJJ6mjRIOpr2KxWIiOjiYqKqrVc9wNCvk7rVwqc5Ns1jN6nIGlUiRFlQ52n6tnV2kdv993kd/vu0hKpN6XDzEsXN8tijgoWM2YiUbM6QZfFvvBPTaOH5JITNZd14O7UeGUFrtoaFAIi1AzYYqRoYm6gPsyOkqHlr56vZ7MzExfpcOzZ89y4MAB3nvvPS5cuMCYMWNYuHAhqampPSpsIAgOUTMjO5i9263s32XDWi+TOrp7Ppw9hc1mIzc3l+joaKZO7VoYpdlsJj8/n6KiIsaN80/bSSkxGemb30O5S4Hyc95yxYfzuXjwLGcup1IZaUOluBkaXE3S+GjCE3q2Qq9yYDfojTB6vO93SWl6rlR6OHnEQXiEmkGxfTdPp7upqamhvLy8TeuB3SazJ6+eulpvIMfwEd5EOQlIizaSFm3k3gmDOF/rYve5OnaX1vPe4Qr+fLiCUL2aKJOGCIOGcKOGSKOGCKOaCIOGCOOX/wyajj3QNVqJ4SO8oasVl7wJcUUnnJw66SQ23mta6oiJR1EUKi56r794wZsMOSTeayLqSdNUT9Ele0hiYiKJiYksWrQIm83GoUOHsNvtbV5z8OBB3nzzTWRZZu7cuSxevLjZOceOHeOtt97C4/EQEhLCz3/+866I1+3o9Cqm3hTM4b02LEcdWOs8jJts6pWaX1EUNm/ejMvlYsmSJV3ONWhc7RUUFPhNKTQiSRKuyDjOJg7ijGsudquMQdOA2b6fhKN/R1d5Dj6W8Awb4TMzMSwFqRt3rIosoxzajZSe0aQmkyRJjJ9soramjv27bMycF9IrIkZ6A+0106mucrPnCysej8KUG4OIaUOhxoXqWDI6iiWjo7hid7PnXD1FVXau2D1csbspqXFSbXfTUhV1o0Z1VUGov1QWVxVHpO+1mpCrIbKSJDFosJZBg7XY6r90BpeVNhAariIpVd+iM9jd4DUReUNfvSai1NH6gDixu5MOPzGOHj1KTEwMMTExXLlyhbVr16JWq7nrrru44Ya2WxPKssyaNWt46qmniIqKYuXKlWRmZjbJmrVarfzxj3/kySefJDo6mpqamjZG9D9qtcSELBNBId70eputnsnTg9Dpe9cf/9ixYxQXFzNz5sw2t/AdIS0tjZ07d1JbW9tuKGt3UVvtobjQybkSF7IHogapGT3eRGycFpVqDoo8C0qLvRnQR/ahfPIXlHXvQUgYUvokGJuJNGYCkim43Xu1RUPBMai5Ai3UOtJopS8T23ZYmTYnuFcuEPxJYzOdoUOHtvhZKTvn4sAuGzq9xNSbQjoVwRVh1DA/NZz5NG0AJSsKdU6vkrjiuPr/1X9VdjfVDjenqxxU2T043M3bzWpUEH7tLsNwVZEEaQifpEZVrab2vIdDe+0cP+RgWLKOYSP01GpcHN1vo/SMC3eDNxlyQpapV4S7dgcdVgpr1qzhySefBPDlKajVal5//XUef/zxNq8tKioiNjbW12h62rRp7N27t4lS2LZtG1lZWURHRwMQFhbWuXfiByRJIm2MgaBgFQf32Nh2NTIpuAfK13aFK1eukJeXR0JCAhMmTLju8RqVQkFBQY83SVEUhQO7bJw/24BKDfHDdCSl6ps9PCSVyrsrGJYCt96JUleLciwfjuSjHNoDO7egqFQwYtSXu4ihiZ3ewjt354Fa02rjlOBQ74Ng33Ybxw7Ye6RaZV+isrKSqqoqZs2a1eT3iqJw2uLk+CEH4ZFqpswM6rZgAJUkXe2BoWF4O+faG2SqHVeVxVWlca0yudTYf8Ph4aubjyGSjnGuIFwnZQpPOgDvgtVq8mCL9nDRBKeq7GhrJLQqCY1KQqv+yv8qCY26hePX/F6rVn35O5WERkVATE8dVgpVVVVER0fj8Xg4dOgQv/nNb9BoNDzwwAMduvbaVWtUVBSFhYVNzikrK8PtdvP0009jt9u55ZZbuOmmm5qNlZubS25uLgDPP/+8T4l0Fo1G0+Vro6Mhdqidzf8qY8dmK3MWDCE2ztilsboLj8fDBx98gFar5c477+z0yr6l+YiOjiYhIYFTp05x8803d6e4zThdWMf5szWMGR/O+MwI9IYOKtroaEhKhlvvQPG4aSg4jit/B878nbg/eBvlg7dRDRqMLmMa+knT0I3NQDK0/bdSFIXK3Xnoxk0iInFYq+dFR4PTXsGR/dUkDAsndZR/dlP+piPflf3796NSqcjKyvK13ZQ9Cru+uIzlmIPhKcHMzI5B00F7f0/QkXALt0fmir2BSmsDlVYXlTYXFVYXVVYXxbUutFXgkmXOqJxYZQ+uCgW3R8blUWjwyM0UyvWiU0to1Crv/yrv/1q1Cq1a4tYxsdwxYWg337ETSsFoNFJdXU1paSnx8fEYDAbcbjdut7vdaxWl+VR9VQN6PB6Ki4v56U9/isvl4qmnniI1NZWhQ5u+6ezsbLKzs32vK7pYwjY6OrrL1wKotTB9bhB78qz8++PzjM80kZAUuMikXbt2cf78eRYsWIDL5er0e2ttPpKTk/n888+xWCzXbY5qDbdbYfcXtd7QwDSFuvor1NV3cbBBQ+HmpXDzUlRXKr3O6iP52Leux77hQ9BoYeRYpPRMpLGTkGKGNBtCOV+CXFaKPOfWducxMUWh7JyGHZ9fQqWxERbRt8KWO0J73xVFUTh06BCJiYnY7XbsdjsNLoV9O6xUXHQzYpSekWPVVFdX+VHqriMB0WqIDgVCNXgfk1/uBNuaD4+s0CAruD1X/5cVGpr9LOOWocEjt3L8y5/dXx3vmuOKy97lZ9hXn6vX0uFP8M0338zKlStxu93cd999gLfqZlxcXLvXRkVFUVlZ6XtdWVnZrK5OVFQUISEhGAwGDAYDo0aNoqSkpE3hA01QsJrp2cHkb7dxcI8Na70Hc7rB71u+srIy9u7dy6hRo7o9Aiw1NZW8vDwsFkuP5aQUnXDgsCtMusGI1I2hpVJEFNKN8+HG+SgNDVB4DOVIvtcX8Zffo/wFiI3zKohxmZA6Gkmj9fZOkCSkCe33q1apJDJuMJG3sY59223MnBfc6/xMPU1ZWRl1dXU+36LN6mF3nhVrncz4yUYSk/tOU6LrRa2SvOU7+vDaoFPtOKdMmYJKpSI21ltTPjIyku9973vtXpuSkkJZWRmXLl0iMjKSHTt28NBDDzU5JzMzkzfeeAOPx+Orxb5w4cJOvh3/o9OpyLopiCP77BQe99bbmTDFf5FJLpeLDRs2EBISwo033tjt45tMJhISEigoKOCGG27odoVnrfdw6qSTuGFaIgf13DdJ0mph9ASk0RPgzuUoly58qSC2/gsl96Mvw0/PnUGbNgY5PLJDY+sNKjKnBbH9s3oO7LYxZebASmyzWCxoNBqSk5O5UumNMFJkmHpTENGDRchuX6NT38JrV+1Hjx5FpVIxevTodq9Tq9UsW7aM1atXI8sys2fPJiEhgY0bNwKQk5NDfHw8EyZM4JFHHkGlUjFnzhwSExM7+XYCg0olMW6ykaAQFScOO7BbvRnQ/ugIlZeXR11dHV/72teuuw9ua5jNZjZt2kR5eTlDhjQ3t1wPxw86kFQwerx/fTJSzFCkuUNh7m0oTgecPOyLaKKqAsNtd9KZOpgR0RrSJxo5km+n4JgTc/rASGzzeDwUFhaSlJRExUU4sLseg0HFlBuDCAntHQEYgs7RYaWwatUq7rrrLkaOHMk///lPPv30U1QqFfPnz2fJkiXtXt9YIuNacnJymrz+ag/ovoQkSYwYZcAUrOLA7i8jk3ryi1FUVMTx48eZPHlyj5rZkpOTUavVWCyWblUKl8obKD/fwMhxhoDGdUt6A4yfgjR+itf/VXUZ4wgztitXOjXOsBQdVyrdFBzzRtp0pdJnX6O0tBSHw0GIKYn8HTYiotXeBdEAM6H1Jzr8lystLfUlpWzevJlVq1axevVqNm3a1GPC9UWGJuiYNjsYj0dhW24dly829Mh96uvr2bJlCzExMT3e/Eiv15OUlERhYSGy3DzeuyvIssKx/XZMwSqS03qPzVmSJKSoGCR155W5JEmMnWQiNFzFgV1eH1N/x3LSgkatp6psEHGJWm6YFSwUQh+nw3+9xgii8vJyAOLj44mOjsZqtfaMZH2YiCgNM7JDMBpV7P7cSskpZ7eOrygKubm5uN1u5s+fj7oLD7DOYjabsdvtlJaWdst4jVmg6RON/SLhpxGNRiJzujckc992Gx53dwcp9h6sVieFRacw6hIxp5uYOLV3ZvkLOkeHlYLZbOaNN97g3XffZfLkyYBXQbTVwWcgYwpSMX1uCNGDNRzeZ+fEIXuLobld4fDhw5w9e5YZM2b4rTvasGHD0Ol0WCyW6x7L6ZApOOYgZoiGmCF9OEyjFYKC1UycaqK22sOR/O77u/cmrPUeNn5yEll2M2HiyA53SRP0fjqsFB588EFMJhPDhg3j61//OgAXLlzglltu6THh+jpancSUmUEMS9FRdNJJ/g4b7utcOVZWVrJt2zaGDx/O2LFju0nS9tFoNIwYMYJTp051KDelLU4eduDxwJgJ/fdBMniolrQxekrPeOvo9CcqL7v5YlM9VTWnMRqDGJ/ReoKfoO/R4WVaSEgI3/zmN5v8rrUG8IIvUakkxk4yEhyi4thBB3abNzKpK45Vj8fDhg0b0Ol0ZGdn+/2BajabOX78OMXFxV3Oh6iudHO22EXKSD3B/Tw6JW20gSuVHo7utxMariYiqu/vis6VuDi0x4bW0IDddZ7x48f329L5A5UO/zXdbjd//etf+eEPf8jdd9/ND3/4Q/76179e96pxICBJEslmA5NnBFFX42Fbbh211Z13Qu7atYuKigrmzp3rlx4HXyUuLo6goKAum5AUReHoATt6g0Tq6P4fsimpvO1d9QaJfTusOB3d46QPBIqiYDnq4MAuGxFRamISypFluUMtXgV9iw4rhT/96U8cOXKE+++/nxdffJH777+fo0eP8qc//akn5etXxMZpmTYnGEWB7ZvruFTW8cikc+fOkZ+fT3p6OsnJyT0oZeuoVCpSU1M5c+YMTmfnnefnzjRwpdLDqHFGtNr+aTb6Kjq9iszpQbgcCvt32VDkvudf8HgUDuy2UXDMQfxwLVNvCubUqQIiIiIYNKh5ZzxB36bDSmHXrl089thjjB8/nqFDhzJ+/HgeeeQRdu7c2ZPy9TvCI72RSaZgFXu+sHKmqP2Hq9PpZNOmTYSHhzNz5kw/SNk6ZrMZWZYpKirq1HUNDQonDtuJiFITP7z/x+9fS3ikhrGTjFRcdHPyqCPQ4nQKm9XDho/Oc76kAfNYAxOmmLDa6jl//jxpaWn91ic0kOl0SKrg+jGaVEyfE8KgWA1H8u2Un297x7B161bq6+vJyclBqw3sAzUmJobw8HBfQ5WOUnjcgdOhkD6x/zqX2yIxWU9iso6iE07KzvVux7OiKFy+2MCebfVs/rSOiktOMm4wkTbaW9erscKxMB31Tzrs+brhhht44YUXWLp0qa9K4AcffNBugx1/oygKDocDWZbbfPhcvHixSyaQ7mRMhkRMuURdrRVjsAattrmOrqmpITo6mlGjRhEaGorN1pniCx3nq/OhKAoqlQqDoWmBv8b+zXv27MFqtfrKJLdFfa2H0wVOEpJ0hPcDZ2tXSc8wUnPFw8E9NkLC1L2mD0cj7gbF17O4/mqz+dRReiZOHoLdUe07z2KxMHjwYMLDw9sYTdBX6fA39J577uGDDz5gzZo1XLlyhcjISKZNm9brHM0OhwOtVttuG0qNRuOXpK/2MCYp1Nd6UCFhNKqaPIA9Hg9qtZqkpCQiIiJ6dIXd0ny43W4cDgdGY9O6RGazmT179lBQUMDEiRPbHLfRuaxWw6hx/d+53BZqtTexLW9jHfu2WZkxLwSNJvC7psZm82eLnbgbuNps3sTQRG8nsaBgDfarVq+qqiouX74ccDOmoOfosFLQaDTceeed3Hnnnb7fuVwu7r33Xu65554eEa4ryLLc5b7EgUCtljAFq7DWydisMqYgr2JQFIXa2loAQkNDA9OBSaNpcTcVERFBTExMh5TCpTI3l8vdjJ5g8EuBwN6OKUjFpBtM7PrcyuG9NiZONQXkb+s1Ebk5c22z+QRvs/q2ms2314dZ0Pe5rqdnb7QN90aZ2kOrVWEwgsMu49Io6A0SNpsNl8tFaGhoQJVca/OZlpbGtm3bqK6ubtWM4PF4dwnBod7m5wIvg2K1jBxr4OQRBxFRLpL8WPvJ3aBQerXZvLWTzeYb+zAnJCR0yGwo6JuIpVsvQW+Q0Gol7DYZh6MBq9WKXq/HYOidJpfG5LW2chZOFzix1XvrG6m6sXlOf2DEKD2Dh2o4dtBO1eWeN8HW13k4ut/Gpo9rOLrfjlYrMTHLRPZtoYwca+xQMuWlS5eoqakRu4R+TrtL0KNHj7Z6rLf5E/oykiRhDFLhqfVQW1uDJEkBMxt1hJCQEOLi4igoKGDKlCnN5LTbZAqPO4iN0zIodmCFoHYESfI+lPM21bNvh5Ubc0K6vXy4oihcKveaiC6VuZFUMNRnIur87tNisaBSqRgxYkS3yinoXbT7yfjtb3/b5vH2GnoPNGpqavjwww99LUs7yr333sv//d//IalUKIoHva5jCuHHP/4x2dnZ3HrrrV2UuOuYzWa2bNnC5cuXiYmJaXLsxGE7igyjJ/TOnU5vQKtTMXl6EF/k1pG/08oNs4K7ZUfV4PKaiM4UejsB6g0SaWMMDEvRdVnxyLJMYWEhw4cP77FmToLeQbtK4bXXXvOHHP2G2tpa3nnnnWZKoTGSqDXeffddnE4n1dXV6PVGFFmLw65gNPXOnQLAiBEj2Lp1KxaLpYlSqLrs5nxJA6mj9QQFBz7CqzcTGq5mfKaJA7ttnDjkYMzErnegq6v1cKbQSekZFx4plEuyAAAgAElEQVQ3RESpMaebGBKvRXWdJa3Pnz+P1WoVuQkDgL4TptMF5L/8AaW0uOVjVyN8OouUkITqG/e3evzZZ5+lpKSEefPmodVqMZlMDB48mGPHjrF161aWLVvGhQsXcDqdLF++3Be5lZWVxbvvvovD4eC//uu/yJiYSf7+fIYMieWtt95sFhbaEl988QXPPPMMHo+H8ePH89xzz6HX63n22WfZuHEjGo2GG2+8kZ/97GesW7eOl19+GZVKRWhoKB9//HGn58JgMDBs2DAKCgqYPn06KpUKRfY6lw1Gbyc6QfvED/d2bDtd4CQiSs3QRF2Hr1UUhUtlbooLnVwud6NSwdBEr4koPLL7vt4WiwWtVktSUlK3jSnonfRrpRAInnjiCSwWC5s2bWLHjh1861vfYsuWLb5+0y+99BIRERHY7XYWLlzILbfcQkREBLIsoygKwcHBFBcX89prr/Hfib/kRz/+Pp988il33LG0zfs6HA4efvhh3n//fVJSUnjooYd45513WLp0KevXrycvLw9JkqipqQHglVdeYe3atQwZMsT3u65gNpspLi7mwoULxMfHc7bYRc0VDxk3mHpFDH5fYcyEq4lte72JbSFhbe+wGlwyZ4tdnCl0YbPKGIwS5nSviai7Q3/dbjdFRUWkpKT0qXBvQdfo13/htlb0Go3GL47yCRMm+BQCwBtvvMH69esBbz+K4uJijEYjiqL4Kp8mJCSQnp6O7FEYM3osp0+XIstKm/bmU6dOkZiYSEpKCgB33HEHb7/9Nt/5znfQ6/U88sgjzJ07l+zsbAAyMzN5+OGHue2221iwYEGX319SUhJarfaqCWkoJ484iBykZmiCcC53BpVaYtI0b2Lb3u1WZs4LabFoYF2Nh+JCJ+fOuPB4ICJazchxV01EPRThVVhYiMvlEqajAYLflMLBgwd58803kWWZuXPnsnjx4ibHjx07xi9/+UufbTorK4ulS9teHfcFri1xvWPHDr744gvWrVuH0Whk6dKl2Gw26urqkCQJk8mEzWbzOfJUagmDUU31FQd225eJbS3RmilMo9Hw6aefsm3bNj766CPefPNN/va3v/HCCy+wf/9+Nm/eTE5ODlu2bCE0NLTT70+r1ZKcnExRURGDwqbgcg3c+kbXi9GkYtI0E7u2Wjm0x8akad7ENkVWuHjVRFRx0WsiikvUMTxV160motY4fPgwRqORhISEHr+XIPD4RSnIssyaNWt46qmniIqKYuXKlWRmZhIfH9/kvFGjRrFixQp/iNRjBAUFUV9f3+Kxuro6wsLCMBqNFBUVsX//fl8tI5Wq5Qe+Wq1Co5FocCk41QoGY8sP2xEjRlBaWkpxcTFJSUl88MEHTJ06FavVit1uZ+7cuWRkZDBjxgwAzpw5Q0ZGBhkZGWzatIkLFy50SSmA14RksVg4cfwMo0anEBbRrzegPUp0jJZR4wwcP+Sg4JgTjQaKi1zYr5qIRo41kJjc/Sai1nA6nVgsFkaPHi2a6QwQ/PLtLSoqIjY2lsGDBwMwbdo09u7d20wp+AuHw+ErIdETjB07lhtvvBG9Xk9UVBSXLl0CYMyYMdhsNmbNmsWwYcNIT0/H7Xa32+darfG29nTYZe/PLRTOMxgM/OpXv+KBBx7wOZrvvfdeqqurWbZsGU6nE0VRWLVqFQC/+MUvKC4uRlEUZsyYwZgxY/B4Ot/4ByA+Ph6N2oDNWczIsWO6NEZvob6+ng8//JD6+vqAVgaWPQpnLl59IXk7+EkSFJ4DtvpPDkVR8Hg8wnQ0gJAUP3zyd+3axcGDB/ne974HQF5eHoWFhSxfvtx3zrFjx3jppZeIiooiIiKCe++9t8Xtam5uLrm5uQA8//zzuFxNyxBfvHix3TjqhoYG7Pbe0VBdo9FgMrVf/0aRFaqvuFAUCIvQob7OEMOO4nQ6fcq8Nc6cqudvf/0Im+sUK1Y83mfj2GVZ5t133+Xs2bNkZWUFVinIClWXnQSHajEYAxvWGxYWRlZWljAJXsVf/sieRKdrPcLNLzuFlr5cX/2AJSUl8Zvf/AaDwcD+/ft58cUXefXVV5tdl52d7XOWAlRUVDQ57nQ6261+2pgt3Fv+sB1doZuCVdTXeqitcREc0rp/oSu09kF3Op3N5vha3G6FXXm1DI5OpqjUwt69exk5cmS3yeVPDh48yKlTp5g9ezazZ89u830PJBpL5Qu89If5GDp0aKvH/GIkjIqKorKy0ve6srKSiIiIJueYTCZfnZ+MjAw8Hk+Pmnj6Gk888QQ335zD1+5YwKLFN5OdncP7778faLE4ddKJ3aYwedpwQkJCuty/OdBUVlayfft2hg8fTnp6eqDFEQgChl92CikpKZSVlXHp0iUiIyPZsWMHDz30UJNzqqurCQsLQ5IkioqKkGW5XVv7QOLZZ5/1/Wy3yTgd3mikQGKzyhSddDA0QcugwVrS0tJ8zvNro656O263mw0bNqDT6cjOzhZmEsGAxi9KQa1Ws2zZMlavXo0sy8yePZuEhAQ2btwIQE5ODrt27WLjxo2o1Wp0Oh0//vGPxZezFQxGCY9bwmaVUasl1AFKEjt+0A7A6AnebGuz2Ux+fj5FRUWMGzcuIDJ1hd27d1NRUcGtt97ap5SZQNAT+C12sDH88VpycnJ8P998883cfPPN/hKnTyNJks+/YK2XCQ5V+b00dcXFBsrONWBON2A0eXcs0dHRREZGUlBQ0GeUwrlz58jPzyc9PZ3k5ORAiyMQBBwReNxHUakkTEFqZFnBbpX9GikjX61vZAxSkWJuGmlkNpu5cOFCn/AHOZ1ONm7cSHh4uGgvKRBcRSiFPoxGK2E0qWhoUHA6/KcUSk65qKuRGTPB0Mx01diApbFtY29m69atWK1W5s+fj1YrynIIBCCUQsBp7GDWEqWlpcyZM6fN63V6yZfY1tAgd7d4zXA6ZSxHHEQP1hAb1/xBGhYWRmxsbK9XChaLBYvFQlZWVrt5GALBQEIohT6OJEmYglSo1RK2ehmPp2d3DJYjDtzutusbmc1mKioqmoQh9ybq6ur47LPPiI2NJTMzM9DiCAS9in5dpOaP+y5SfMXR4jGpi/0UkiIM/Gdm6yvL1atXExcX52uy89JLLyFJErt27aKmpga3281jjz3G/PnzO3Vfh8PBypUrOXz4MGq1mlWrVjF9+nQsFgs/+clPcLlcuBtkXv316ySPGML3vvc9ysrKkGWZH/3oRyxatKjT7/Wr1FxxU3LK22i+rdLOqamp5OXlYbFYmDZt2nXftzuRZZmNGzeiKAo5OTmino9A8BX6tVIIBIsWLWLVqlU+pbBu3TrWrl3L/fffT0hICFVVVdx2223k5OR0KuT2rbfeAmDz5s0UFRVx11138cUXX/Duu++yfPlylixZgrXeQV2tmw0bthAbG8u7774L0C1OX0VROLLfjk4vYR7TdhkLk8lEQkICBQUF3HDDDb0qtPjAgQOcP3+e7OxswsPDAy2OQNDr6NdKoa0VfU/VL0lPT6eiooLy8nIqKysJCwsjJiaGp59+mt27dyNJEuXl5S32NW6LvXv38p3vfAfwVkSNj4/n9OnTTJo0iVdffZWysjIWLFjAkNjhJA8388ILq1m9ejXZ2dlkZWVd9/s6f7aBKxUexk82otW1v7o2m81s2rSJ8vJyhgwZct337w4uX77Mzp07SUlJYdSoUYEWRyDolYi9cw+wcOFCPv30Uz7++GMWLVrEP/7xDyorK1m/fj2bNm0iOjoap9PZqTFbM3XdfvvtvPnmmxgMBu6++27y9+9gRGoK7//lE1JTzTz33HO8/PLL1/V+3A0KJw7ZCYtQk5DUsVaRycnJqNXqXlP2ojFr2Wg0MmfOnF61exEIehNCKfQAixYt4qOPPuLTTz9l4cKF1NXVER0djVarZfv27Zw7d67TY2ZlZfHhhx8C3i5r58+fJyUlhZKSEoYNG8by5cuZN28eJ06coLbuEkEmAzfnLOa7332AI0eOXNf7KTzhwGFXSM/oePMcvV5PUlIShYWFyHLPR0W1x44dO6iqqiI7O7tD/a4FgoFKvzYfBQqz2YzVavX1kFiyZAnf/va3WbBgAWPGjGHEiBGdHvPb3/42K1asYO7cuajVal5++WX0ej0ff/wx//jHP9BoNMTExPDwww9z6NAhnnnmGRRFhVar4YUXnuvye2lokDltcRM/TEtkdOc+LmazmaKiIkpLSxk2bFiXZbheSkpKOHjwIOPHjw+oHAJBX8Av/RR6kgsXLjR53dFibP2hJnp7OB0ydpuM3qDylaJojdbm42xxDUf3K8y5JRSDsXMbS7fbzR//+EeSk5OblDTxJ3a7nT//+c/o9Xq+8Y1vdLjxfH8oj9xdiLloSn+Yj4CXzhYEBp1eQqeXcDpkGlydN+E0uGTsdoW00YZOKwTwKpoRI0Zw6tSpgChgRVHYsmULdrud+fPnd1ghCAQDGfEt6QWcOHGiWSlxvV7PJ598cl3jSpK3DIbHLWOzygSrpQ53bFMUBbtdQaOB4Sld76RmNps5fvw4xcXFbWZv9wQnTpzg1KlTTJ8+nUGDBvn13gJBX0UohV7AqFGj2LRpU4+MfW1FVdvViqodcRY7HQqyRyE8Un1drT/j4uIICgrCYrH4VSnU1NTw+eefExcXx8SJE/12X4GgryPMRwMAtdpbCsPj6VhFVVlWcDrkqwX3rq8/sEqlIjU1lTNnznQ6DLerNGYtS5IkspYFgk4ivi0DBK1OhcGowuVScDnbVgoOm4yi0K5zuqOYzWZkWaaoqKhbxmuP/Px8ysrKmD17tujeJxB0EqEUBhB6g4RGK2G3ybgbWlYMbreCy6WgN6iuy2x0LTExMYSFhfmlcurFixfZvXs3aWlpmM3mHr+fQNDfEEphANFYUVWlkrBZPchyU8WgKAp2m4ykAoOh+zJ+JUnCbDZTWlqK1WrttnG/SkNDAxs2bMBkMjFr1qweu49A0J8RSqGbqamp8RWv6wz33nsvNTU13S/QV1CpvI5nWQZbfVP/gsul4HErGI0qpG5u7+mP5jvbtm2jurqaefPmYTAYeuw+AkF/pl9HHx3db6O22tPisa6Wzg4NV5Oe0XpyXG1tLe+8846vSmojHo8Htbp1p21jRVN/oNF4dww2q4zDrmA0SciygsMmo9Z4m/Z0N5GRkQwaNIiCgoIeiQYqLi7myJEjZGRkkJCQ0O3jCwQDBb8phYMHD/Lmm28iyzJz585l8eLFLZ5XVFTEk08+ycMPP8zUqVP9JV638eyzz1JSUsK8efPQarWYTCYGDx7MsWPH2Lp1K8uWLePChQs4nU6WL1/OPffcA3hrG61fvx6r1co999zDlClT2LdvH7Gxsbzxxhut1utZu3Yta9euxeVykZSUxKuvvorRaOTy5cusWLGCkpISAJ577jkmT57M3/72N15//XUA0lJH8uzqV1BrQPHgcy73VLE4s9nsW813Z9lqm81Gbm4u0dHRffIzIxD0JvyiFGRZZs2aNTz11FNERUWxcuVKMjMziY+Pb3be2rVrmTBhQrfct60VfU+VuXjiiSewWCxs2rSJHTt28K1vfYstW7aQmJgIeJvuREREYLfbWbhwIbfccguRkZFNxiguLua1117jxRdf5IEHHuBf//oXX/va11q834IFC7j77rsBeOGFF3jvvfdYtmwZP/3pT5k6dSpr1qzB4/FgtVqxWCy8+uqrfPTRR0RGRlJVVYVaI10NU5XR6SU0mp6rHpqamsq2bdt8bTC7A0VRyM3NxeVysWTJEpG1LBBcJ37xKRQVFfmKw2k0GqZNm8bevXubnbd+/XqysrIIDQ31h1h+YcKECT6FAPDGG2+QnZ3NbbfdxoULFyguLm52TUJCAunp6QCMGzeO0tLSVse3WCzcfvvtzJ07lw8//NBXqnr79u1861vfAkCtVhMaGsr27dtZuHChTwlFRkYSFOT9CEiS1KVSFp0hJCSEuLg4CgoKumS6a4mjR49y5swZpk+fTlRUVLeMKRAMZPyyrKqqqmryhY2KiqKwsLDZOXv27GHVqlX89re/bXWs3NxccnNzAXj++eeJjo5ucvzixYsdXi32xKqy0W+g0WhQq9UEBQX57rN9+3a2bdvGv/71L0wmE7fffjtutxuNRoMkSajVatRqNXq93neNVqvF5XK1KuvDDz/M22+/zZgxY/jLX/7Cjh07fONpNJom1zXeo8lYGgiPUAMS6hZ2CXq9vtkcXw+TJk3i448/pqGhoc2iXB2hoqKCbdu2kZKSwpw5c7o1SU2j0XTr++7LiLloSn+fD78ohZZWhV+1W7/11lvcfffd7X6xs7Ozyc7O9r3+arVCp9PZpkO3kZ4yHxkMBurr63G73Xg8HhRF8d2nurqa0NBQdDodJ0+eJD8/H4/Hg9vtRlEUPB4PHo/XMd54jSzLyLLcqqz19fVERUVht9v5+9//TmxsLG63m+nTp/PGG29w//334/F4sNlsTJs2jeXLl7N8+XIiIyO5cuUKERERbc6H0+ns1oqQsbGxqFQqdu/ezcyZM7s8jsfj4W9/+xtqtZqbbrqJqqqqbpMR+kclzO5CzEVT+sN8tLUg84tSiIqKorKy0ve6srLS9zBq5NSpU/z6178GvBE8Bw4cQKVSMWXKFH+I2G1ERkYyefJk5syZg8FgaLKimDVrFu+++y7Z2dkkJyeTkZFx3fd79NFHufXWW4mPj2fkyJHU19cD8N///d889thj/OUvf0GlUvHcc8+RmZnJQw89xNKlS1GpVKSnp/PKK69ctwydwWAwMGzYMAoKCpg+fXqXV/d79+7l0qVLLFiwgODg4G6WUiAYuPiln4LH4+FHP/oRP/vZz4iMjGTlypU89NBDrYYOvvbaa0yaNKlDkSSin0L30Np8dHQ+O0NBQQH//ve/WbJkSbNgg45QVlbG3//+d0aOHMm8efO6VbZG+sNqsLsQc9GU/jAfAd8pqNVqli1bxurVq5FlmdmzZ5OQkMDGjRsBAtaARRAYkpKS0Gq1WCyWTisFl8vFhg0bCAkJ4cYbb+whCQWCgYvf4vcyMjKamUtaUwYPPvigP0TqUzzxxBPNIrb+8z//kzvvvDNAEnUdrVZLcnIyRUVF3HTTTZ1y+Ofl5VFXV8fXvvY19Pqu93kQCAQtI4K6+wjPPvtsoEXoVsxmMxaLhbNnz5KcnNyha4qKijh+/DiTJ0++7sglgUDQMqL2kSAgJCQkYDAYfHkV7VFfX8+WLVuIiYnpc8EHAkFfQigFQUBQq9WkpqZSXFyMy+Vq81xFUdi0aRNut5v58+d3KORYIBB0DaEUBAHDbDbjdrs5ffp0m+cdOnSI0tJSZs6c2SyUWSAQdC9CKQgCxpAhQwgJCWnThFRZWcn27dsZPny4r/SHQCDoOYRSCDD+bGbf25AkibS0NM6ePYvNZmt23O12s2HDBnQ6HdnZ2T1WvVUgEHxJv44+ysvL4/Llyy0e62o/hUGDBon4+G7EbDaTn59PUVER48aNa3Js9+7dVFRUcOutt3Z7Ap1AIGiZfq0UAsHq1auJi4vzNdl56aWXkCSJXbt2UVNTg9vt5rHHHmP+/PntjmW1WvnOd77T4nXX9kUYNWoU//u//9tqD4XeTFRUFJGRkRQUFDRRCufOnSM/P5/09PQOh6wKBILrp18rhbZW9D1V5mLRokWsWrXKpxTWrVvH2rVruf/++wkJCaGqqorbbruNnJycds0her2eNWvWNLuuoKCgSV+EK1euALTYQ6G309i/eefOndTW1hIaGorT6WTjxo2Eh4dfV9E8gUDQefq1UggE6enpVFRUUF5eTmVlJWFhYcTExPD000+ze/duJEmivLycy5cvExMT0+ZYiqLw/PPPN7vuq30RGiNytm/f7isq2NhDoS+QlpbGzp07KSgoIDMzk61bt2K1WrnjjjvQarWBFk8gGFAIpdADLFy4kE8//ZRLly6xaNEi/vGPf1BZWcn69evRarVkZWXhdDrbHae16xRF6VdO17CwMGJjYykoKPBFI02dOpXY2NhAiyYQDDhE9FEPsGjRIj766CM+/fRTFi5cSF1dHdHR0Wi1WrZv3865c+c6NE5r182YMYN169b5egg0mo9mzJjBO++8A3gr09bV1fXAu+sZzGYzFRUVbN68mdjYWDIzMwMtkkAwIBFKoQcwm81YrVZfC9IlS5Zw6NAhFixYwIcffsiIESM6NE5r15nNZl9fhOzsbH7+858D3h4KO3bsYO7cudx8880dLiHRG0hNTUWSJCRJIicnp1u7qAkEgo7jl34KPYnop9A9+LOfQmscPHiQsLAwkpKS/HK/tugPNfO7CzEXTekP8xHwfgoCQUeYMGFCoEUQCAY8Qin0Ak6cOMFDDz3U5Hd6vZ5PPvkkQBIJBIKBSr9TCn3RGjZq1Cg2bdoUaDFapC/Op0Ag6Dr9zpunUqmEr6CbcLvdwuErEAww+t1OwWAw4HA4cDqdbcby6/X6DuUKDBS+Oh+KoqBSqTAYDAGUSiAQ+Bu/KYWDBw/y5ptvIssyc+fOZfHixU2O7927l/fffx9JklCr1dx3332MHDmy0/eRJAmj0djuef0hgqA7EfMhEAjAT0pBlmXWrFnDU089RVRUFCtXriQzM5P4+HjfOWPHjiUzMxNJkigpKeHll1/mlVde8Yd4AoFAILiKXwzGRUVFvkQujUbDtGnT2Lt3b5NzDAaDz9zTnulHIBAIBD2DX3YKVVVVREVF+V5HRUVRWFjY7Lw9e/bw5z//mZqaGlauXOkP0QQCgUBwDX5RCi2FNba0E5gyZQpTpkzh+PHjvP/++/z0pz9tdk5ubi65ubkAPP/8821m5rXH9VzbHxHz0RQxH18i5qIp/Xk+/GI+ioqKorKy0ve6srKyzQbso0ePpry8nNra2mbHsrOzef7553n++eevS6YVK1Zc1/X9DTEfTRHz8SViLprS3+fDL0ohJSWFsrIyLl26hNvtZseOHc2qYJaXl/t2FKdPn8btdhMSEuIP8QQCgUBwFb+Yj9RqNcuWLWP16tXIsszs2bNJSEhg48aNAOTk5LBr1y7y8vJQq9XodDoefvhh4WwWCAQCP+O3PIWMjAwyMjKa/C4nJ8f38+LFi5vlLvQk2dnZfrtXX0DMR1PEfHyJmIum9Pf56POlswUCgUDQfYjCNgKBQCDwIZSCQCAQCHz0u4J4HaG9OkwDiYqKCl577TWqq6uRJIns7GxuueWWQIsVUGRZZsWKFURGRvb78MP2sFqt/O53v6O0tBRJkvj+979PWlpaoMUKCJ988glbtmxBkiQSEhL4wQ9+gE6nC7RY3c6AUwodqcM0kFCr1dx7770kJydjt9tZsWIF48aNG7DzAfCvf/2LuLg47HZ7oEUJOG+++SYTJkzg//2//4fb7R6wlYWrqqpYv349L7/8Mjqdjl/96lfs2LGDWbNmBVq0bmfAmY86UodpIBEREUFycjIARqORuLg4qqqqAixV4KisrGT//v3MnTs30KIEHJvNxokTJ5gzZw7g7eMdFBQUYKkChyzLuFwuPB4PLperzQTcvsyA2yl0tA7TQOTSpUsUFxczYsSIQIsSMN566y3uuecesUvA+3kIDQ3lN7/5DSUlJSQnJ3PfffcNyB4bkZGR3HbbbXz/+99Hp9Mxfvx4xo8fH2ixeoQBt1PoaB2mgYbD4eCll17ivvvuw2QyBVqcgJCfn09YWJhv5zTQ8Xg8FBcXk5OTwy9/+Uv0ej3//Oc/Ay1WQKivr2fv3r289tprvP766zgcDvLy8gItVo8w4JRCZ+swDQTcbjcvvfQSM2fOJCsrK9DiBAyLxcK+fft48MEHeeWVVzh69CivvvpqoMUKGFFRUURFRZGamgrA1KlTKS4uDrBUgeHIkSPExMQQGhqKRqMhKyuLgoKCQIvVIww489G1dZgiIyPZsWMHDz30UKDFChiKovC73/2OuLg4br311kCLE1C++c1v8s1vfhOAY8eOsW7dugH92QgPDycqKooLFy4wdOhQjhw5MmADEKKjoyksLMTpdKLT6Thy5AgpKSmBFqtHGHBKobU6TAMVi8VCXl4eiYmJPProowDcddddzUqSCAYmy5Yt49VXX8XtdhMTE8MPfvCDQIsUEFJTU5k6dSqPP/44arWa4cOH99tyF6LMhUAgEAh8DDifgkAgEAhaRygFgUAgEPgQSkEgEAgEPoRSEAgEAoEPoRQEAoFA4EMoBYHAT3z961+nvLw80GIIBG0y4PIUBAKABx98kOrqalSqL9dFs2bNYvny5QGUqmU2bNhAVVUVd911F6tWrWLZsmUMGzYs0GIJ+ilCKQgGLI8//jjjxo0LtBjtcvr0aTIyMpBlmXPnzg3YrGKBfxBKQSD4Clu3bmXz5s0kJSXx+eefExERwfLlyxk7dizgrbT7hz/8gZMnTxIcHMyiRYt82a2yLPPPf/6Tzz77jJqaGoYMGcKjjz5KdHQ0AIcPH+bZZ5+lrq6O6dOns3z58nYLMp4+fZqlS5dy4cIFYmJiUKvVPTsBggGNUAoCQQsUFhaSlZXFmjVr2LNnD//zP//Da6+9RnBwML/+9a9JSEjg9ddf58KFCzzzzDMMHjyYsWPH8sknn7B9+3ZWrlzJkCFDKCkpQa/X+8bdv38/zz33HHa7nccff5zMzEwmTJjQ7P4NDQ3cf//9KIqCw+Hg0Ucfxe12I8sy9913H//xH//BkiVL/DklggGCUAqCAcuLL77YZNV9zz33+Fb8YWFhLFy4EEmSmDZtGuvWrWP//v2MHj2akydPsmLFCnQ6HcOHD2fu3Lnk5eUxduxYNm/ezD333MPQoUMBGD58eJN7Ll68mKCgIIKCghgzZgxnzpxpUSlotVreeustNm/eTGlpKffddx+/+MUv+MY3vjGg+10Ieh6hFAQDlkcffbRVn0JkZGQTs86gQYOoqqriypUrBAcHYzQafVqA0kwAAAGzSURBVMeio6M5deoU4C3FPnjw4FbvGR4e7vtZr9fjcDhaPO+VV17h4MGDOJ1OtFotn332GQ6Hg6KiIoYMGcJzzz3XqfcqEHQUoRQEghaoqqpCURSfYqioqCAzM5OIiAjq6+ux2+0+xVBRUUFkZCTg7UFw8eJFEhMTr+v+P/7xj5Flme9+97v8/ve/Jz8/n507dw7oUt4C/yDyFASCFqipqWH9+vW43W527tzJ+fPnmThxItHR0ZjNZv785z/jcrkoKSnhs88+Y+bMmQDMnTuX999/n7KyMhRFoaSkhLq6ui7JcP78eQYPHoxKpaK4uLjf1u8X9C7ETkEwYHnhhRea5CmMGzfO11MiNTWVsrIyli9fTnh4OD/5yU8ICQkB4Ec/+hF/+MMfeOCBBwgODuaOO+7wmaFuvfVWGhoa+MUvfkFdXR1xcXE88sgjXZLv9OnTJCUl+X5etGjR9bxdgaBDiH4KAsFXaAxJfeaZZwItikDgd4T5SCAQCAQ+hFIQCAQCgQ9hPhIIBAKBD7FTEAgEAoEPoRQEAoFA4EMoBYFAIBD4EEpBIBAIBD6EUhAIBAKBj/8PVgSFsNx+V9MAAAAASUVORK5CYII=\n",
+      "text/plain": [
+       "<Figure size 432x288 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "plot_history(history)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Based on the plot above, it looks like our model stopped learning after 7 epochs. You can tell this by looking at the decline in val_loss from epochs 0-7. "
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Try a model with less dropout, same learning rate: "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "new_model = Sequential()\n",
+    "\n",
+    "# Add the convolutional part of the VGG16 model from above.\n",
+    "new_model.add(vgg_model)\n",
+    "\n",
+    "# Flatten the output of the VGG16 model because it is from a\n",
+    "# convolutional layer.\n",
+    "new_model.add(Flatten())\n",
+    "\n",
+    "# Add a dense (aka. fully-connected) layer.\n",
+    "# This is for combining features that the VGG16 model has\n",
+    "# recognized in the image.\n",
+    "new_model.add(Dense(1024, activation='relu'))\n",
+    "\n",
+    "# Add a dropout-layer which may prevent overfitting and\n",
+    "# improve generalization ability to unseen data e.g. the test-set.\n",
+    "new_model.add(Dropout(0.3))\n",
+    "\n",
+    "# Add a dense (aka. fully-connected) layer.\n",
+    "# This is for combining features that the VGG16 model has\n",
+    "# recognized in the image.\n",
+    "new_model.add(Dense(512, activation='relu'))\n",
+    "\n",
+    "# Add a dropout-layer which may prevent overfitting and\n",
+    "# improve generalization ability to unseen data e.g. the test-set.\n",
+    "new_model.add(Dropout(0.3))\n",
+    "\n",
+    "# Add a dense (aka. fully-connected) layer.\n",
+    "# This is for combining features that the VGG16 model has\n",
+    "# recognized in the image.\n",
+    "new_model.add(Dense(256, activation='relu'))\n",
+    "\n",
+    "# Add a dense (aka. fully-connected) layer.\n",
+    "# Change the activation function to sigmoid \n",
+    "# so output of the last layer is in the range of [0,1] \n",
+    "new_model.add(Dense(1, activation='sigmoid'))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 17,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "new_model.compile(optimizer=optimizer, loss=loss, metrics=metrics)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 18,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Epoch 1/10\n",
+      "3/3 [==============================] - 1s 417ms/step - loss: 0.8695 - binary_accuracy: 0.4000 - val_loss: 0.7596 - val_binary_accuracy: 0.5000\n",
+      "Epoch 2/10\n",
+      "3/3 [==============================] - 0s 163ms/step - loss: 0.7205 - binary_accuracy: 0.4500 - val_loss: 0.8358 - val_binary_accuracy: 0.5000\n",
+      "Epoch 3/10\n",
+      "3/3 [==============================] - 0s 158ms/step - loss: 0.7830 - binary_accuracy: 0.5000 - val_loss: 0.7132 - val_binary_accuracy: 0.5000\n",
+      "Epoch 4/10\n",
+      "3/3 [==============================] - 1s 168ms/step - loss: 0.6932 - binary_accuracy: 0.5000 - val_loss: 0.8323 - val_binary_accuracy: 0.5000\n",
+      "Epoch 5/10\n",
+      "3/3 [==============================] - 0s 153ms/step - loss: 0.7039 - binary_accuracy: 0.6500 - val_loss: 0.5906 - val_binary_accuracy: 0.5000\n",
+      "Epoch 6/10\n",
+      "3/3 [==============================] - 1s 174ms/step - loss: 0.4466 - binary_accuracy: 0.7000 - val_loss: 0.8360 - val_binary_accuracy: 0.5000\n",
+      "Epoch 7/10\n",
+      "3/3 [==============================] - 1s 185ms/step - loss: 0.7707 - binary_accuracy: 0.6500 - val_loss: 0.6338 - val_binary_accuracy: 0.5000\n",
+      "Epoch 8/10\n",
+      "3/3 [==============================] - 1s 197ms/step - loss: 0.4513 - binary_accuracy: 0.7500 - val_loss: 0.6204 - val_binary_accuracy: 0.6667\n",
+      "Epoch 9/10\n",
+      "3/3 [==============================] - 1s 197ms/step - loss: 0.7032 - binary_accuracy: 0.6500 - val_loss: 0.6882 - val_binary_accuracy: 0.6667\n",
+      "Epoch 10/10\n",
+      "3/3 [==============================] - 1s 192ms/step - loss: 0.5135 - binary_accuracy: 0.8500 - val_loss: 0.5230 - val_binary_accuracy: 0.8333\n"
+     ]
+    }
+   ],
+   "source": [
+    "history = new_model.fit_generator(train_gen, \n",
+    "                                  validation_data = (testX, testY), \n",
+    "                                  epochs = 10)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 19,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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\n",
+      "text/plain": [
+       "<Figure size 432x288 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "plot_history(history)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "With less dropout, it looks like the model got worse since the train_loss dropped dramatically but the val_loss increased."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Finally, try a model with the same amount of dropout as you initiall had, but a slower learning rate: "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 20,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "new_model = Sequential()\n",
+    "\n",
+    "# Add the convolutional part of the VGG16 model from above.\n",
+    "new_model.add(vgg_model)\n",
+    "\n",
+    "# Flatten the output of the VGG16 model because it is from a\n",
+    "# convolutional layer.\n",
+    "new_model.add(Flatten())\n",
+    "\n",
+    "# Add a dropout-layer which may prevent overfitting and\n",
+    "# improve generalization ability to unseen data e.g. the test-set.\n",
+    "new_model.add(Dropout(0.5))\n",
+    "\n",
+    "# Add a dense (aka. fully-connected) layer.\n",
+    "# This is for combining features that the VGG16 model has\n",
+    "# recognized in the image.\n",
+    "new_model.add(Dense(1024, activation='relu'))\n",
+    "\n",
+    "# Add a dropout-layer which may prevent overfitting and\n",
+    "# improve generalization ability to unseen data e.g. the test-set.\n",
+    "new_model.add(Dropout(0.5))\n",
+    "\n",
+    "# Add a dense (aka. fully-connected) layer.\n",
+    "# This is for combining features that the VGG16 model has\n",
+    "# recognized in the image.\n",
+    "new_model.add(Dense(512, activation='relu'))\n",
+    "\n",
+    "# Add a dropout-layer which may prevent overfitting and\n",
+    "# improve generalization ability to unseen data e.g. the test-set.\n",
+    "new_model.add(Dropout(0.5))\n",
+    "\n",
+    "# Add a dense (aka. fully-connected) layer.\n",
+    "# This is for combining features that the VGG16 model has\n",
+    "# recognized in the image.\n",
+    "new_model.add(Dense(256, activation='relu'))\n",
+    "\n",
+    "# Add a dense (aka. fully-connected) layer.\n",
+    "# Change the activation function to sigmoid \n",
+    "# so output of the last layer is in the range of [0,1] \n",
+    "new_model.add(Dense(1, activation='sigmoid'))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "## Set our optimizer, loss function, and learning rate\n",
+    "optimizer = Adam(lr=1e-6)\n",
+    "loss = 'binary_crossentropy'\n",
+    "metrics = ['binary_accuracy']"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 22,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "new_model.compile(optimizer=optimizer, loss=loss, metrics=metrics)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 23,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Epoch 1/10\n",
+      "3/3 [==============================] - 1s 406ms/step - loss: 1.1356 - binary_accuracy: 0.3500 - val_loss: 0.7164 - val_binary_accuracy: 0.5000\n",
+      "Epoch 2/10\n",
+      "3/3 [==============================] - 1s 176ms/step - loss: 0.8184 - binary_accuracy: 0.3000 - val_loss: 0.7161 - val_binary_accuracy: 0.5000\n",
+      "Epoch 3/10\n",
+      "3/3 [==============================] - 1s 196ms/step - loss: 1.0050 - binary_accuracy: 0.3500 - val_loss: 0.7158 - val_binary_accuracy: 0.5000\n",
+      "Epoch 4/10\n",
+      "3/3 [==============================] - 1s 173ms/step - loss: 0.8322 - binary_accuracy: 0.3500 - val_loss: 0.7155 - val_binary_accuracy: 0.5000\n",
+      "Epoch 5/10\n",
+      "3/3 [==============================] - 1s 180ms/step - loss: 0.9515 - binary_accuracy: 0.4500 - val_loss: 0.7155 - val_binary_accuracy: 0.5000\n",
+      "Epoch 6/10\n",
+      "3/3 [==============================] - 1s 183ms/step - loss: 0.7037 - binary_accuracy: 0.6500 - val_loss: 0.7154 - val_binary_accuracy: 0.5000\n",
+      "Epoch 7/10\n",
+      "3/3 [==============================] - 1s 203ms/step - loss: 0.7191 - binary_accuracy: 0.5500 - val_loss: 0.7154 - val_binary_accuracy: 0.5000\n",
+      "Epoch 8/10\n",
+      "3/3 [==============================] - 1s 208ms/step - loss: 0.6984 - binary_accuracy: 0.6500 - val_loss: 0.7153 - val_binary_accuracy: 0.5000\n",
+      "Epoch 9/10\n",
+      "3/3 [==============================] - 1s 191ms/step - loss: 0.8035 - binary_accuracy: 0.4500 - val_loss: 0.7151 - val_binary_accuracy: 0.5000\n",
+      "Epoch 10/10\n",
+      "3/3 [==============================] - 1s 191ms/step - loss: 0.8383 - binary_accuracy: 0.4000 - val_loss: 0.7150 - val_binary_accuracy: 0.5000\n"
+     ]
+    }
+   ],
+   "source": [
+    "history = new_model.fit_generator(train_gen, \n",
+    "                                  validation_data = (testX, testY), \n",
+    "                                  epochs = 10)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 24,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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\n",
+      "text/plain": [
+       "<Figure size 432x288 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "plot_history(history)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "The performance with this architecture and parameters is pretty unstable, and also isn't learning based on the flat val_loss curve. We see, however, that between epoch 3 and 8 the train_loss goes down, which indicates that while the model was learning _something,_ it was essentially just learning to overfit on the training data. "
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.7.6"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}