exercise-deep-neural-networks.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "db33e3ac",
   "metadata": {
    "papermill": {
     "duration": 0.006154,
     "end_time": "2023-06-30T15:59:43.999512",
     "exception": false,
     "start_time": "2023-06-30T15:59:43.993358",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "**This notebook is an exercise in the [Intro to Deep Learning](https://www.kaggle.com/learn/intro-to-deep-learning) course.  You can reference the tutorial at [this link](https://www.kaggle.com/ryanholbrook/deep-neural-networks).**\n",
    "\n",
    "---\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "acd20f8e",
   "metadata": {
    "papermill": {
     "duration": 0.005205,
     "end_time": "2023-06-30T15:59:44.010554",
     "exception": false,
     "start_time": "2023-06-30T15:59:44.005349",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Introduction #\n",
    "\n",
    "In the tutorial, we saw how to build deep neural networks by stacking layers inside a `Sequential` model. By adding an *activation function* after the hidden layers, we gave the network the ability to learn more complex (non-linear) relationships in the data.\n",
    "\n",
    "In these exercises, you'll build a neural network with several hidden layers and then explore some activation functions beyond ReLU. Run this next cell to set everything up!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "218399ba",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-06-30T15:59:44.024279Z",
     "iopub.status.busy": "2023-06-30T15:59:44.023864Z",
     "iopub.status.idle": "2023-06-30T15:59:53.804607Z",
     "shell.execute_reply": "2023-06-30T15:59:53.803323Z"
    },
    "papermill": {
     "duration": 9.791853,
     "end_time": "2023-06-30T15:59:53.808291",
     "exception": false,
     "start_time": "2023-06-30T15:59:44.016438",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.10/site-packages/tensorflow_io/python/ops/__init__.py:98: UserWarning: unable to load libtensorflow_io_plugins.so: unable to open file: libtensorflow_io_plugins.so, from paths: ['/opt/conda/lib/python3.10/site-packages/tensorflow_io/python/ops/libtensorflow_io_plugins.so']\n",
      "caused by: ['/opt/conda/lib/python3.10/site-packages/tensorflow_io/python/ops/libtensorflow_io_plugins.so: undefined symbol: _ZN3tsl6StatusC1EN10tensorflow5error4CodeESt17basic_string_viewIcSt11char_traitsIcEENS_14SourceLocationE']\n",
      "  warnings.warn(f\"unable to load libtensorflow_io_plugins.so: {e}\")\n",
      "/opt/conda/lib/python3.10/site-packages/tensorflow_io/python/ops/__init__.py:104: UserWarning: file system plugins are not loaded: unable to open file: libtensorflow_io.so, from paths: ['/opt/conda/lib/python3.10/site-packages/tensorflow_io/python/ops/libtensorflow_io.so']\n",
      "caused by: ['/opt/conda/lib/python3.10/site-packages/tensorflow_io/python/ops/libtensorflow_io.so: undefined symbol: _ZTVN10tensorflow13GcsFileSystemE']\n",
      "  warnings.warn(f\"file system plugins are not loaded: {e}\")\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "\n",
    "# Setup plotting\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "plt.style.use('Solarize_Light2')\n",
    "# Set Matplotlib defaults\n",
    "plt.rc('figure', autolayout=True)\n",
    "plt.rc('axes', labelweight='bold', labelsize='large',\n",
    "       titleweight='bold', titlesize=18, titlepad=10)\n",
    "\n",
    "# Setup feedback system\n",
    "from learntools.core import binder\n",
    "binder.bind(globals())\n",
    "from learntools.deep_learning_intro.ex2 import *"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5756d9e7",
   "metadata": {
    "papermill": {
     "duration": 0.005357,
     "end_time": "2023-06-30T15:59:53.819537",
     "exception": false,
     "start_time": "2023-06-30T15:59:53.814180",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "In the *Concrete* dataset, your task is to predict the compressive strength of concrete manufactured according to various recipes.\n",
    "\n",
    "Run the next code cell without changes to load the dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "0cbe67d9",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-06-30T15:59:53.833620Z",
     "iopub.status.busy": "2023-06-30T15:59:53.832772Z",
     "iopub.status.idle": "2023-06-30T15:59:53.897653Z",
     "shell.execute_reply": "2023-06-30T15:59:53.896634Z"
    },
    "papermill": {
     "duration": 0.074724,
     "end_time": "2023-06-30T15:59:53.900050",
     "exception": false,
     "start_time": "2023-06-30T15:59:53.825326",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Cement</th>\n",
       "      <th>BlastFurnaceSlag</th>\n",
       "      <th>FlyAsh</th>\n",
       "      <th>Water</th>\n",
       "      <th>Superplasticizer</th>\n",
       "      <th>CoarseAggregate</th>\n",
       "      <th>FineAggregate</th>\n",
       "      <th>Age</th>\n",
       "      <th>CompressiveStrength</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>540.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>162.0</td>\n",
       "      <td>2.5</td>\n",
       "      <td>1040.0</td>\n",
       "      <td>676.0</td>\n",
       "      <td>28</td>\n",
       "      <td>79.99</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>540.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>162.0</td>\n",
       "      <td>2.5</td>\n",
       "      <td>1055.0</td>\n",
       "      <td>676.0</td>\n",
       "      <td>28</td>\n",
       "      <td>61.89</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>332.5</td>\n",
       "      <td>142.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>228.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>932.0</td>\n",
       "      <td>594.0</td>\n",
       "      <td>270</td>\n",
       "      <td>40.27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>332.5</td>\n",
       "      <td>142.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>228.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>932.0</td>\n",
       "      <td>594.0</td>\n",
       "      <td>365</td>\n",
       "      <td>41.05</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>198.6</td>\n",
       "      <td>132.4</td>\n",
       "      <td>0.0</td>\n",
       "      <td>192.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>978.4</td>\n",
       "      <td>825.5</td>\n",
       "      <td>360</td>\n",
       "      <td>44.30</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Cement  BlastFurnaceSlag  FlyAsh  Water  Superplasticizer  CoarseAggregate  \\\n",
       "0   540.0               0.0     0.0  162.0               2.5           1040.0   \n",
       "1   540.0               0.0     0.0  162.0               2.5           1055.0   \n",
       "2   332.5             142.5     0.0  228.0               0.0            932.0   \n",
       "3   332.5             142.5     0.0  228.0               0.0            932.0   \n",
       "4   198.6             132.4     0.0  192.0               0.0            978.4   \n",
       "\n",
       "   FineAggregate  Age  CompressiveStrength  \n",
       "0          676.0   28                79.99  \n",
       "1          676.0   28                61.89  \n",
       "2          594.0  270                40.27  \n",
       "3          594.0  365                41.05  \n",
       "4          825.5  360                44.30  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "concrete = pd.read_csv('../input/dl-course-data/concrete.csv')\n",
    "concrete.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "12d0e4af",
   "metadata": {
    "papermill": {
     "duration": 0.005983,
     "end_time": "2023-06-30T15:59:53.912109",
     "exception": false,
     "start_time": "2023-06-30T15:59:53.906126",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# 1) Input Shape #\n",
    "\n",
    "The target for this task is the column `'CompressiveStrength'`. The remaining columns are the features we'll use as inputs.\n",
    "\n",
    "What would be the input shape for this dataset?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "b60ed7e5",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-06-30T15:59:53.928152Z",
     "iopub.status.busy": "2023-06-30T15:59:53.927473Z",
     "iopub.status.idle": "2023-06-30T15:59:53.936923Z",
     "shell.execute_reply": "2023-06-30T15:59:53.935629Z"
    },
    "lines_to_next_cell": 2,
    "papermill": {
     "duration": 0.019303,
     "end_time": "2023-06-30T15:59:53.939273",
     "exception": false,
     "start_time": "2023-06-30T15:59:53.919970",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "application/javascript": [
       "parent.postMessage({\"jupyterEvent\": \"custom.exercise_interaction\", \"data\": {\"outcomeType\": 1, \"valueTowardsCompletion\": 0.3333333333333333, \"interactionType\": 1, \"questionType\": 2, \"questionId\": \"1_Q1\", \"learnToolsVersion\": \"0.3.4\", \"failureMessage\": \"\", \"exceptionClass\": \"\", \"trace\": \"\"}}, \"*\")"
      ],
      "text/plain": [
       "<IPython.core.display.Javascript object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "<span style=\"color:#33cc33\">Correct</span>"
      ],
      "text/plain": [
       "Correct"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# YOUR CODE HERE\n",
    "input_shape = [8]\n",
    "\n",
    "# Check your answer\n",
    "q_1.check()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "c658684d",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-06-30T15:59:53.953612Z",
     "iopub.status.busy": "2023-06-30T15:59:53.953231Z",
     "iopub.status.idle": "2023-06-30T15:59:53.958079Z",
     "shell.execute_reply": "2023-06-30T15:59:53.956781Z"
    },
    "papermill": {
     "duration": 0.014786,
     "end_time": "2023-06-30T15:59:53.960353",
     "exception": false,
     "start_time": "2023-06-30T15:59:53.945567",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# Lines below will give you a hint or solution code\n",
    "#q_1.hint()\n",
    "#q_1.solution()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cd930431",
   "metadata": {
    "papermill": {
     "duration": 0.006036,
     "end_time": "2023-06-30T15:59:53.972695",
     "exception": false,
     "start_time": "2023-06-30T15:59:53.966659",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# 2) Define a Model with Hidden Layers #\n",
    "\n",
    "Now create a model with three hidden layers, each having 512 units and the ReLU activation.  Be sure to include an output layer of one unit and no activation, and also `input_shape` as an argument to the first layer."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "367417af",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-06-30T15:59:53.987398Z",
     "iopub.status.busy": "2023-06-30T15:59:53.986997Z",
     "iopub.status.idle": "2023-06-30T15:59:54.223482Z",
     "shell.execute_reply": "2023-06-30T15:59:54.222181Z"
    },
    "lines_to_next_cell": 0,
    "papermill": {
     "duration": 0.246955,
     "end_time": "2023-06-30T15:59:54.226004",
     "exception": false,
     "start_time": "2023-06-30T15:59:53.979049",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "application/javascript": [
       "parent.postMessage({\"jupyterEvent\": \"custom.exercise_interaction\", \"data\": {\"outcomeType\": 1, \"valueTowardsCompletion\": 0.3333333333333333, \"interactionType\": 1, \"questionType\": 2, \"questionId\": \"2_Q2\", \"learnToolsVersion\": \"0.3.4\", \"failureMessage\": \"\", \"exceptionClass\": \"\", \"trace\": \"\"}}, \"*\")"
      ],
      "text/plain": [
       "<IPython.core.display.Javascript object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "<span style=\"color:#33cc33\">Correct</span>"
      ],
      "text/plain": [
       "Correct"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from tensorflow import keras\n",
    "from tensorflow.keras import layers\n",
    "\n",
    "# YOUR CODE HERE\n",
    "model = keras.Sequential([\n",
    "#     creating 3 hidden layers\n",
    "    layers.Dense(units=512, activation='relu', input_shape=input_shape),\n",
    "    layers.Dense(units=512, activation='relu'),\n",
    "    layers.Dense(units=512, activation='relu'),\n",
    "#     creating output layer\n",
    "    layers.Dense(units=1),\n",
    "])\n",
    "\n",
    "# Check your answer\n",
    "q_2.check()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "65abb3ad",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-06-30T15:59:54.241410Z",
     "iopub.status.busy": "2023-06-30T15:59:54.240975Z",
     "iopub.status.idle": "2023-06-30T15:59:54.246067Z",
     "shell.execute_reply": "2023-06-30T15:59:54.244886Z"
    },
    "papermill": {
     "duration": 0.015678,
     "end_time": "2023-06-30T15:59:54.248486",
     "exception": false,
     "start_time": "2023-06-30T15:59:54.232808",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# Lines below will give you a hint or solution code\n",
    "#q_2.hint()\n",
    "#q_2.solution()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a83656c1",
   "metadata": {
    "papermill": {
     "duration": 0.006826,
     "end_time": "2023-06-30T15:59:54.262151",
     "exception": false,
     "start_time": "2023-06-30T15:59:54.255325",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# 3) Activation Layers #\n",
    "\n",
    "Let's explore activations functions some.\n",
    "\n",
    "The usual way of attaching an activation function to a `Dense` layer is to include it as part of the definition with the `activation` argument. Sometimes though you'll want to put some other layer between the `Dense` layer and its activation function. (We'll see an example of this in Lesson 5 with *batch normalization*.) In this case, we can define the activation in its own `Activation` layer, like so:\n",
    "\n",
    "```\n",
    "layers.Dense(units=8),\n",
    "layers.Activation('relu')\n",
    "```\n",
    "\n",
    "This is completely equivalent to the ordinary way: `layers.Dense(units=8, activation='relu')`.\n",
    "\n",
    "Rewrite the following model so that each activation is in its own `Activation` layer."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "3c6f28cd",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-06-30T15:59:54.277565Z",
     "iopub.status.busy": "2023-06-30T15:59:54.277138Z",
     "iopub.status.idle": "2023-06-30T15:59:54.339621Z",
     "shell.execute_reply": "2023-06-30T15:59:54.338345Z"
    },
    "lines_to_next_cell": 0,
    "papermill": {
     "duration": 0.073129,
     "end_time": "2023-06-30T15:59:54.342246",
     "exception": false,
     "start_time": "2023-06-30T15:59:54.269117",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "application/javascript": [
       "parent.postMessage({\"jupyterEvent\": \"custom.exercise_interaction\", \"data\": {\"outcomeType\": 1, \"valueTowardsCompletion\": 0.3333333333333333, \"interactionType\": 1, \"questionType\": 2, \"questionId\": \"3_Q3\", \"learnToolsVersion\": \"0.3.4\", \"failureMessage\": \"\", \"exceptionClass\": \"\", \"trace\": \"\"}}, \"*\")"
      ],
      "text/plain": [
       "<IPython.core.display.Javascript object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "<span style=\"color:#33cc33\">Correct</span>"
      ],
      "text/plain": [
       "Correct"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "### YOUR CODE HERE: rewrite this to use activation layers\n",
    "model = keras.Sequential([\n",
    "    layers.Dense( 32, input_shape=[8]),\n",
    "    layers.Activation('relu'),\n",
    "    layers.Dense(32),\n",
    "    layers.Activation('relu'),\n",
    "    layers.Dense(1),\n",
    "])\n",
    "\n",
    "# Check your answer\n",
    "q_3.check()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "beff3aa4",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-06-30T15:59:54.358506Z",
     "iopub.status.busy": "2023-06-30T15:59:54.358026Z",
     "iopub.status.idle": "2023-06-30T15:59:54.363051Z",
     "shell.execute_reply": "2023-06-30T15:59:54.361934Z"
    },
    "papermill": {
     "duration": 0.01602,
     "end_time": "2023-06-30T15:59:54.365415",
     "exception": false,
     "start_time": "2023-06-30T15:59:54.349395",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# Lines below will give you a hint or solution code\n",
    "#q_3.hint()\n",
    "#q_3.solution()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7639d728",
   "metadata": {
    "papermill": {
     "duration": 0.006761,
     "end_time": "2023-06-30T15:59:54.379402",
     "exception": false,
     "start_time": "2023-06-30T15:59:54.372641",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Optional: Alternatives to ReLU #\n",
    "\n",
    "There is a whole family of variants of the `'relu'` activation -- `'elu'`, `'selu'`, and `'swish'`, among others -- all of which you can use in Keras. Sometimes one activation will perform better than another on a given task, so you could consider experimenting with activations as you develop a model. The ReLU activation tends to do well on most problems, so it's a good one to start with.\n",
    "\n",
    "Let's look at the graphs of some of these. Change the activation from `'relu'` to one of the others named above. Then run the cell to see the graph. (Check out the [documentation](https://www.tensorflow.org/api_docs/python/tf/keras/activations) for more ideas.)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "0806111e",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-06-30T15:59:54.395052Z",
     "iopub.status.busy": "2023-06-30T15:59:54.394619Z",
     "iopub.status.idle": "2023-06-30T15:59:54.932888Z",
     "shell.execute_reply": "2023-06-30T15:59:54.931645Z"
    },
    "lines_to_next_cell": 0,
    "papermill": {
     "duration": 0.549277,
     "end_time": "2023-06-30T15:59:54.935635",
     "exception": false,
     "start_time": "2023-06-30T15:59:54.386358",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAnYAAAHWCAYAAAD6oMSKAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAABXQklEQVR4nO3dd3gU570+/Htmd7Xaol5ABYEkeu/VgLEx4AI2jmvcEyd2TnKccpKT105+8YlPknNOmhOnObETO4mduHcINsUGDKaDRJFAIIGQhFCXtmnbPO8fI620IIEkdrTa3ftzXbp2ZnZm/Ozj1XDrOzPPSH7nWQEiIiIiinhyuBtARERERKHBYEdEREQUJRjsiIiIiKIEgx0RERFRlGCwIyIiIooSDHZEREREUYLBjoiIiChKMNgRERERRQl9uBsQLRRFCUxLkhTGlhAREVEkE6Lr2RGy3L8aHINdCNns9nA3gYiIiKJEgtXa7214KpYihiRJSExIYEVUI+xf7bGPtcc+1h77eGhjsAuRzi84v+jakSQp8EOhx/7VHvtYe+xj7bGPtXclmYLBjoiIiChKMNgRERERRQkGOyIiIqIowWBHREREFCUY7IiIiIiiBIMdERERUZRgsCMiIiKKEgx2RERERFGCwY6IiIgoSjDYEREREUUJfbgb0F9llbX46LMjqKxtRKvdhUdvW4bp40ZecpsTZ2rxxqa9qKlvRnKCGSvmT8aSWeOD1jlQehrvbT2IhmYb0lMScPPSmZgx/tL7JSIiIhpKIq5i5/b4kDssFXetnN+n9RtabPjtq5swekQmvvfwGqxaOBWvfrQHB0pPB9Ypr6rD829txfzJhfj+w2swf3Ihnnv7E1RU12v0KYiIiIhCL+IqdpNH52Ly6Nw+r7/twHGkJlpwx4p5AICs9GScOdeAjbuOYub4UQCAzXuOYUJ+NlYtmgoAWJWejBOVtdi85xgeXrs05J+BiIiISAsRV7Hrr/KqekwoyA5aNrEgB2fONcDvV9R1qntep7yqbtDaSURERHSlIq5i119tDhcSLaagZYlWExRFwO5sR1KCGW32HtaxmNDmcPW6X6/PD5/fH5gXQgAAZFkOTFNoybIc9Eqhxf7VHvtYe+xj7cVCHwsh4FOAdj/g8qqv7T6g3Sfg9kP98amvHj/g9gt4AtMITHsUAW/HtFdRfzx+wKdAXa6o/x1fx3udr5+fqMODkwfW9qgPdgAgXTAfCF7d3pAuWEng0uFsw85irNteFJg3GvR46tE1sJjNkC7cGYWUxWwOdxOiGvtXe+xj7bGPtTfU+lgIAacXaHML2DwCdo+AzS1g9wrY3eq83Svg8Ag4vALOzlcv4PQKuHxCffUCLp8auMLFI3QA/JddrydRH+wSLSa0XlB5sznaIcsSrKZ4dR2rCa32i9e5sIrX3aqFU7F83qTAvBACXo8bDqeTFTuNyLIMi9kMh9MJRQnjb1yUYv9qj32sPfax9gajj71+gRY30NwOtLQL9dUt0OoG2txAq1ugzQO0tgM2j4DNA9g8gD+C/vmVAMTpAL2s/hg6fvQyYNYNLNQBMRDsCnIzUFx2NmhZSUUNRmalQ6dTy8gFORkoqagJCmol5TUoyM3sdb8GvQ4GvS4w3xnsFEVhsNOYoig8YGuI/as99rH22Mfa628fCyHg8AL1LqDeCdQ7BRqcQGM70OQCGtsFmlxAU7sa0gaTTgIsBiBeD5j0wa/xeiBeJ8GoU6eNOsCoB+I7XuPkjteOdeJkwKBTQ5tBVl+7T3eGN73c+9k9SZIGnCUiLti1e7yob2oLzDe02HG2thEWkxGpSVa8/fF+tNiceGjNYgDAkpnj8Mm+Ury+cQ+umjEW5VX12HGoDF9cuySwj2vmTsQv/vYvfLjzMKaNHYGiE2dRcroG37n/hkH/fERERJHIpwjUOYFzduCcQ+C8Ax0/ArUOoM4JuHyh/W9KABLiAGuc+tr5Y42TYDWoYc0ap75aDBIsBsBsAMz6jteOaYMuei6hirhgd+ZcA55+6cPA/Bub9gIA5k8txIOrF6PV7kRTqz3wfnpyAr5253K8vnEPtu4vRZLVjDtXzA0MdQIAhbmZ+OLapXhv6wG8t/UgMlIS8KW1VyM/J2OwPhYREdGQ5/ELnGsTqK/z4HidH5WtAjV2gRo7UOu48lOhZj2QagJS4oEUI5AcLyElHkg2AsnxQJJRQmIckGgEkoxqiJN5XXsQye88y/OGISCEQJvNdkXlU7o0WZaRYLXCZrfzFIsG2L/aYx9rj30cGg6vQEULcLpVoKIVKG8RqGwDzjkAZQD/xMXrgGEWIMMMZJglZJiAdDOQbpKQbgLSTEBqPGAyMKQBXadiExMS+n1DZsRV7IiIiCg0FKFW2040AWXNAmXNAieb1epbf5gNQI4VyLYCOQkShluA4Rb1dZgFSIwDR4wYJAx2REREMUAIgXN24FijQEkjcKxB4Hgz4PT2bXuzAchLAPKSJIxJN2JYvBc5VoHcBAa3oYTBjoiIKAq5/WqAK64TKKoTONoAtLgvv53FABQkAwVJEvKTgfyO1wyTGt7U090m2Ox+nu4eghjsiIiIooDLK1BUD+yvVYNcSaP6FINLGWYGxqYCY1MljEmRMDYFyLKy+hbJGOyIiIgikE9Rq3D7agX2nRM43IBLPi0hMQ6YmA5MTJMwMV3ChDQgzcQAF20Y7IiIiCJEc7vAZ9UCO6qB3TUC9ktcHzciAZiWKWFaJjA1Q0JeIitxsYDBjoiIaAiraBX4+IzAjmqBYw3o9UnmuQnA7OESZg8HZgyTWI2LUQx2REREQ4gQAuUtwJZKgS1n1HHkepIQB8zLkjA3C5idJSHbyiBHDHZERERDQrVN4F/lAhtPC5xp63mdwmRgYY6ERTkSJmdc+nmjFJsY7IiIiMLE5hHYfEYNdEV1Pa8zNQO4ZqSEpSMkZLEqR5fBYEdERDSIhBA4cB54p0xga6WA54I7WSUA0zOBZSMlXJ0nIdPMMEd9x2BHREQ0CGwegfWnBN4uEzjdw3Vzo5KAGwokrMqXkGlhmKOBYbAjIiLS0MlmgVdLBD46LeD2B7+XbARW5Eu4oUDCuFQOR0JXjsGOiIgoxIQQ2FcLvHxMwa6ai9+fkQncOlbC0jwJcTqGOQodBjsiIqIQ8SnqECUvHxM43hT8nsUAXF8g4daxEgqSGeZIGwx2REREV8inCHxYIfCXYoFqe/B7WRbg7okSbiqUYDYw0JG2GOyIiIgGyK+o4879uVjgrC34vfGpwD2TJCzLkzjeHA0aBjsiIqJ+UoQ6/tzzRRcPJjxnOPDgFBkzh/FmCBp8DHZERET9cKhO4Jl9Co41Bi+fNQz40jQZ04cxzFH4MNgRERH1QbVN4HcHFGypDF4+LRP48jQZs4Yz0FH4MdgRERFdgt0j8MJhgddKBbzdnhJRmAz8+ywZ87J4ypWGDgY7IiKiHoiO6+ie3ifQ6OpanhIPPDJdwupCCTreFEFDDIMdERHRBWpsAj/bo+CzboMLx8nqsCX3T5JgiWOgo6GJwY6IiKiDTxH4Z4l6t2v3x39dlQt8a46MbCsDHQ1tDHZEREQAypoFntqhoKy5a1mGGfiPOTKWjuB1dBQZGOyIiCim+RX1EWB/KhLwddwcIQG4fbyER6bxtCtFFgY7IiKKWdU2gR/uUFBc37WsMBn43gIZE9MZ6CjyMNgREVHMEULg3ZMCv94n4PKpyyQA906S8KVpEuJ0DHUUmRjsiIgoptg9Aj/5LHig4Wwr8INFMqZnMtBRZGOwIyKimFHWJPD4NgVVtq5la0ZL+PpsCRYDQx1FPgY7IiKKekIIvH9S4Bd7u4YxSYgDvr9AxtI8BjqKHgx2REQU1VxegZ/tEVhfLgLLxqcCP1kiIzuBoY6iC4MdERFFrVqHwLc/VnCy29h0nxsr4bHZEoy8QYKiEIMdERFFpcP1Av/5iYLmdnXepAceny9hRb4c3oYRaYjBjoiIos6GcgU/2qHA0zHgcG4C8LNlMvKTWKWj6BaRwe6TfaXYuOsIWu1OZGek4Pbr5mJM3rAe133x/e3YVXzqouVZ6cl48pFbAAA7i8rwtw92XLTOb757Lwz6iOwiIqKYpAiBX+1y4Y/7ux70OmsY8JOlMpKMDHUU/SIutew7VoHXN+7B3avmo3BEJrYfOI7fvrIRTz5yC1KTrBetf+d187B22azAvKII/Oj59zBzwsig9eKNBvzw0bVByxjqiIgiR7tP4KmdCrac8QWW3TxawnfmSdDLDHUUGyIuuWzafRSLpo/BVTPGAgDuWDEPx8prsPXA8aAA18kUHwcT4gLzh46fgdPlxsJpY4LWkwAkWc2atp2IiLRh86g3SRTVqfOyBHx9loQ7xkuQJIY6ih0RFex8fj8qzzVi5YIpQcsnFGSjvKquT/vYcagM4/OzkXZBdc/t8eGJ37wORQiMGJaK1UtnIG94WsjaTkRE2mh0CXx9c9edr2YD8JMlOszPDm+7iMIhooKd3emGIgQSraag5YkWE9rsrstu32pz4uipanzhliVBy4enJ+GB1VchJzMFLrcXW/Yew8/+uh7f/9LNGJaa2OO+vD4/fP6uaziEUMdHkmU5ME2hJcty0CuFFvtXe+zj0KtqE/j3jX5U29X5lHjgudUJGGX1QFGU8DYuSvF7rD1JkuDvljH6I6KCXacLi+pCiIsX9uCz4pMwxcdh+ri8oOUFOZkoyMkMzBeOyMRPnn8Pn+wtwZ0r5/W4rw07i7Fue1Fg3mjQ46lH18BiNrPsrzGLmafMtcT+1R77ODRKG3z48gY7Gjr+rs9OkPH8Givyk3WI0H/eIgq/x9oRQqDNZrv8ij2IqG++1WyELElovaA6Z3O2I9Fi6mUrlRACO4rKMG9KIfQ63SXXlSUJI7PTUdfU1us6qxZOxfJ5k4L27/W44XA6WbHTiCzLsJjNcDid/EtcA+xf7bGPQ6eoTsE3N/lh96rzBcnAM8tlZMa5AbCPtcTvsfaupEAUUcFOr9MhLysNJRU1mDG+667WkooaTBubd4ktgROVtahvtmHRBTdN9EQIgarzTcjJSOl1HYNeB4NeF7SN1+OGoigMdhpTFIUHEw2xf7XHPr4yh+oEvrlZgavj5tcpGcDPl8lIMopAv7KPtcc+1k7MBDsAWD5vEl54dztGZqWjIDcD2w+eQHOrA0tmjgMAvP3xfrTYnHhozeKg7XYeKkN+djpyMi8Oax9sO4T8nAxkpiai3e3Bx3tLcPZ8E+5aOX9QPhMREfXNhaFuXhbwv0tlmAy8BIYIiMBgN3tiPuxON9Z9eghtdheyM1LwtbuWB+5ybbU70dRqD9rG1e7BgdIzuGNFz9fLOd0evLx+J9ocLpiMcRgxPBXfvu965OdkaP55iIiob4rqBL7VLdTNzwb+72qZz3wl6kbyO8/yvGEIdF7oKEkST8VqRJZlJFitsNntLP9rgP2rPfbxwBV1VOqclwl17GPtsY+115klEhMS+n1alvcqExHRkFZ8Qaibl8VKHVFvGOyIiGjIKm0U+MaWrlA3l6GO6JIY7IiIaEiqbOuo1HUMaTI3C/jp1TLi9Qx1RL1hsCMioiGn3inw9U0Kmt3q/NQMhjqivmCwIyKiIcXmEfjGZgXnHOp8YbI6Th1DHdHlMdgREdGQ0e4T+PbHCk61qPPDLcCvrpWRaGSoI+oLBjsiIhoSfIrA97YpKKpT51OM6mPCMswMdUR9xWBHRERhJ4TAL/YK7KhW580G4OlrZeQlMtQR9QeDHRERhd1rpQJvn1AHdzfIwE+XyhifxlBH1F8MdkREFFY7qwV+vb/riT1PLJAwO4uhjmggGOyIiChsTjYLfH+7AqUj1z04WcL1BfyniWig+NtDRERh0eRS74DtHID4mjzgy9NZqSO6Egx2REQ06Nx+ge9uVVDbMVbdhDTgB4tkyP184DkRBWOwIyKiQSWEwP98JnC4Xp3PMPOpEkShwmBHRESD6o3jAhsq1Ivq4nXqUyU4Vh1RaDDYERHRoCmuE/jVvq47YH+wSMa4VIY6olBhsCMiokHR6FKfLOHvyHX3TJRwzUiGOqJQYrAjIiLN+RSB729TUO9S52cOA74yg6GOKNQY7IiISHO/PyhwsOMZsBkm4L8Xy9DLDHZEocZgR0REmtpyRuAfx9Tzr3oZ+MlSGWkmhjoiLTDYERGRZirbBH60UwnMf32WhCkZDHVEWmGwIyIiTXj86uPCnD51fmW+hNvGMdQRaYnBjoiINPGHgwInmtTpkYnA/zdPgsQnSxBpisGOiIhC7rNqgX+WqNfVGWTgqcUyTAaGOiKtMdgREVFINboEnup2Xd3XZkochJhokDDYERFRyChC4KkdCprb1flFOcAd4xnqiAYLgx0REYXMP0sEdp9Tp9NMwPcXyryujmgQMdgREVFIlDYK/OGgel2dBODJRTJS4hnqiAYTgx0REV2xdp/Af+1Q4Ou4tO7eSRLmZjHUEQ02BjsiIrpifyoSON2qTo9PBR6ZzlBHFA4MdkREdEUO1Qn8s+ORYXGyegqWz4ElCg8GOyIiGjCXV31kmOiYf2S6hPxkhjqicGGwIyKiAfvdQYEqmzo9NQO4awJDHVE4MdgREdGA7D0n8MZxtVZn1KlDm+h4CpYorBjsiIio3+wegR9/Fvx0ibxEhjqicGOwIyKifvv1PoFahzo9axjwuXEMdURDgT7cDRiIT/aVYuOuI2i1O5GdkYLbr5uLMXnDelz3+JlzePqlDy9a/l+P3ILh6cmB+QOlp/He1oNoaLYhPSUBNy+diRnjR2r1EYiIItaecwLvn1JPwZoNwPcWypD5dAmiISHigt2+YxV4feMe3L1qPgpHZGL7geP47Ssb8eQjtyA1ydrrdj98dC3ijYbAfII5PjBdXlWH59/aijVLZ2D6uDwcOl6J597+BN+5/wbk52Ro+nmIiCJJu0/g/3Z1nYL995kSsq0MdURDRcSdit20+ygWTR+Dq2aMRVZ6Mu5YMQ8piRZsPXD8ktslWOKRZDUHfmS566Nv3nMME/KzsWrRVAxPT8aqRVMxflQWNu85pvXHISKKKM8XC1Tb1enpmcDNYxjqiIaSiKrY+fx+VJ5rxMoFU4KWTyjIRnlV3SW3/fHz78Pr8yMrIxk3LJqKcaOyAu+VV9fj2rkTg9afWJCDLZcIdl6fHz6/PzAvhHpaQpblwDSFVmcY7x7KKXTYv9qL9D4+3tg1ELFBBp5YqIdeN7SCXaT3cSRgH2tPkiT4u2WM/oioYGd3uqEIgUSrKWh5osWENrurx22SrGbcc8NCjByeBp/fj12HT+FXL3+Ib923CmPyhgMA2uwuJFp62Kej530CwIadxVi3vSgwbzTo8dSja2AxmyHxWhNNWczmcDchqrF/tReJfexTBP53vQ3+jr9bH50djyk5pktvFEaR2MeRhn2sHSEE2my2AW0bUcGu04WxSQhx8cIOw9OSMDwtKTBfkJuJ5jYHNu46Ggh2AHBhFhO4dNVt1cKpWD5vUlAbvB43HE4nK3YakWUZFrMZDqcTiqJcfgPqF/av9iK5j18+6sfRerXNBcnAXWN9sNnt4W1UDyK5jyMF+1h7V1IgiqhgZzUbIUsSWi+oztmc7RdV3C4lPycDe46UB+YTraaL9+m49D4Neh0Mel1gvjPYKYrCYKcxRVF4MNEQ+1d7kdbHNTaBPx5U2ysBeHy+DJ0koChD91gXaX0cidjH2rmSYBdRJ8j1Oh3ystJQUlETtLykogYFuZl93s/Z801I6nY6tyAn4+J9lvdvn0RE0UgIgZ/uUdDecbnPrWMlTMng5SZEQ1VEBTsAWD5vEnYcKsOOQ2U419CC1zbuQXOrA0tmjgMAvP3xfrzw3vbA+pv3HMWh42dwvqkNNfXNePvj/ThYegZXz54QWOeauRNRUl6DD3ceRm1DCz7ceRglp2suuqGCiCjWbDotsKvj794MM/BvMxjqiIayiDoVCwCzJ+bD7nRj3aeH0GZ3ITsjBV+7aznSOsawa7U70dTadd2Hz6/gzc370GJzwqDXITsjGV+9czmmjM4NrFOYm4kvrl2K97YewHtbDyIjJQFfWns1x7Ajopjm8Ar8en/X6dbvzJVhiWOwIxrKJL/z7NC9SCKCdN7BIkkSr7HTiCzLSLBaYbPbeV2HBti/2ou0Pn5mv4J/dAxvclUu8PNlustsEX6R1seRiH2svc4skZiQ0O/r7SLuVCwREWmvvEXg1RI11MXJwDdn858LokjA31QiIgoihMDP9yiBMevunywhJ4GnYIkiAYMdEREF2Xha4MB5dTrHCtw7iaGOKFIw2BERUYDDI/BMtxsmvjVHRryewY4oUjDYERFRwPPFAg0d47VflQssymWoI4okDHZERAQAONUs8FqpWq0z6oBv8YYJoojD31oiIoIQAr/cG3zDRDZvmCCKOAx2RESErWeB/bxhgijiMdgREcU4j1/gN/u7Bpr991kyjDoGO6JIxGBHRBTjXisVqO54EuPMYcDSEeFtDxENHIMdEVEMa3QJ/OWwemGdBPUJE/19hBERDR0MdkREMey5IgGnV51eM1rCmFSGOqJIxmBHRBSjypoE3jupVuvMBuDL0xnqiCIdgx0RUQwSQuBX+xQoHcObPDRFQpqJwY4o0jHYERHFoO1VwcOb3DmeoY4oGjDYERHFGK9f4Jluw5t8bZaMOA5vQhQVGOyIiGLMWycEqmzq9IxhwNUc3oQoajDYERHFELuna3gTAPj6LA5vQhRNGOyIiGLI348KtLrV6RWjJIxPY6gjiiYMdkREMaLOKfBKiVqtM8jAoxzehCjqMNgREcWI54oE3H51+nPjJGQnMNgRRRsGOyKiGHCqWWDdKbVaZzWo49YRUfRhsCMiigG/P9g1GPH9kyUkGRnsiKIRgx0RUZTbXyuwo1qdzjQDd3AwYqKoxWBHRBTFhBD43YGuwYi/PF1CvJ7BjihaMdgREUWxzWcEjjWq04XJwPX5DHVE0YzBjogoSvkUgT8e6hqM+KszZehkBjuiaMZgR0QUpdafEjjb7dFhC7LD2x4i0h6DHRFRFHL7BZ4v7qrWfWU6Hx1GFAsY7IiIotDbJwTqnOr0ohxgaiZDHVEsYLAjIooyDq/AXw93Vesenc5DPVGs4G87EVGUebVEoNmtTl83SsKYVFbriGIFgx0RURRpdQu8fEyt1ukk4EvTGOqIYgmDHRFRFPn7UQGHV52+qVBCXiKDHVEsYbAjIooS9U6B10vVal2cDHxhKkMdUazRh7sBA/HJvlJs3HUErXYnsjNScPt1czEmb1iP6x4sPYOtB0pRdb4JPp+CrIxk3LR4OiYV5gTW2VlUhr99sOOibX/z3Xth0EdkFxFRDHrhsIDbr07fOk7CMAuDHVGsibjUsu9YBV7fuAd3r5qPwhGZ2H7gOH77ykY8+cgtSE2yXrR+WWUtJuRn45arZ8EUH4fPisrw+9c247sP3Yi84WmB9eKNBvzw0bVB2zLUEVGkqLELvHdSrdaZ9cADkxnqiGJRxCWXTbuPYtH0MbhqxlgAwB0r5uFYeQ22HjiOtctmXbT+HSvmBc3fsmwWik6cxeGys0HBTgKQZDVr2nYiIq28eFjAp6jTd06QkBLPYEcUiyIq2Pn8flSea8TKBVOClk8oyEZ5VV2f9qEIgXaPF+Z4Y9Byt8eHJ37zOhQhMGJYKlYvnREU/IiIhqoqm8C6U2q1zmoA7p7AUEcUqyIq2NmdbihCINFqClqeaDGhze7q0z427ToKj9eHWRNHBZYNT0/CA6uvQk5mClxuL7bsPYaf/XU9vv+lmzEsNbHH/Xh9fvj8/sC8EOpBVZblwDSFlizLQa8UWuxf7WnVxy8e9sHfcdj5/CQZySZdSPcfSfg91h77WHuSJMHfLWP0R0QFu04X/i0qhLh4YQ/2Hi3HB9sP4Su3X4NES1c4LMjJREFOZmC+cEQmfvL8e/hkbwnuXDmvp11hw85irNteFJg3GvR46tE1sJjNfB6jxixmnjLXEvtXe6Hs49MtfqwvbwMAJBklPDw7EQlGHoP4PdYe+1g7Qgi02WwD2jaigp3VbIQsSWi9oDpnc7YHBbWe7DtWgb99sANfvvVqTMjPvuS6siRhZHY66prael1n1cKpWD5vUmBeCAGvxw2H08mKnUZkWYbFbIbD6YSiKOFuTtRh/2pPiz5+5jMflM5q3UQJ8Dpg84Zk1xGJ32PtsY+1dyUFoogKdnqdDnlZaSipqMGM8SMDy0sqajBtbF6v2+09Wo6/fbADX7xlCaaMGXHZ/44QAlXnm5CTkdLrOga9Dga9Lmgbr8cNRVEY7DSmKAoPJhpi/2ovVH1c0SrwYYV6vEkyAreNA//fdeD3WHvsY+3ETLADgOXzJuGFd7djZFY6CnIzsP3gCTS3OrBk5jgAwNsf70eLzYmH1iwGoIa6F97bjjuum4f8nAy02p0AgDi9Hqb4OADAB9sOIT8nA5mpiWh3e/Dx3hKcPd+Eu1bOD8+HJCLqg78Ui0C17t5JEiwGnoIlinURF+xmT8yH3enGuk8Poc3uQnZGCr5213KkdYxh12p3oqnVHlh/24HjUBSBVz7chVc+3BVYPn9qIR5crYY/p9uDl9fvRJvDBZMxDiOGp+Lb912P/JyMwf1wRER9dKpZYNNpNdWlGIHbxjLUEREg+Z1ned4wBDovdJQkiadiNSLLMhKsVtjsdpb/NcD+1V4o+/iJrX5sqVSnH5sl4fMTeYciwO/xYGAfa68zSyQmJPT7tCyPBEREEaasWQRCXZoJuJXVOiLqwGBHRBRh/lLcVSW5f5KEeD2DHRGpGOyIiCLIyWaBj7tV624ew1BHRF0Y7IiIIsgLh7uu4b2P1ToiugCDHRFRhChvEdhyRg12qfHALazWEdEFGOyIiCLEC4cFOut197BaR0Q9YLAjIooAFa3B49bdymodEfWAwY6IKAK8UBxcrTPxKRNE1AMGOyKiIe5Mq8Cmjmvrko0ct46IejegYFdWWYuyytpQt4WIiHrw4pGuZ8J+fqIEM6t1RNSLAQW7X/59A55+6cMe3/vKj1/Ev/3kr1fUKCIiUlW2CXxYoaa6xDjgc+MY6oiodwM+FdvT01A7n5HKJ6USEYXGXy+o1llYrSOiS9D3dcWmVjsaW+1By8oqz6N7jKs63wwA0Mk88BARXakam8CG8q5q3e2s1hHRZfQ52O0sOol1nxZ1LRACv3xpw8UrShKSEyyhaBsRUUz7+1EBf8ffzndOkGCJY7Ajokvrc7ADAAhx6fkOC6YWDrQ9REQEoM4h8MEp9RhrNrBaR0R90+dglzssFfOnjgYA7Co+CUgS5k/pCnASAIvZiIKcTMwYPzLkDSUiiiUvHRPwKur07eMkJBoZ7Ijo8voc7KaPy8P0cXkAOoIdgAdWX6VNq4iIYlijS+DdMrVaF68D7prAUEdEfdO/U7EdfvS120LdDiIi6vCPYwJuvzq9dqyElHgGOyLqmwEFu6aOu2ObLrhLtrsxecMH1iIiohjW0i7w1gm1WhcnA/dMZKgjor4bULD75d83AFLvBxsJwO+feGCgbSIiilmvlgq4fOr0mjES0s0MdkTUdwMKdgB6vSMWAMQlQh8REfXM5hF4rVQ9tupl4N5JPJYSUf8MKNh13h3bSVEU1DXbcLq6HsY4A2byrlgion57vVTA4VWnbyiQMNzCYEdE/TOgYNfb3bD7S07j+bc+QeGIYVfUKCKiWOP0CrxSolbrdBJw/2SGOiLqvwE/K7YnsyaMgtFowJa9x0K5WyKiqPf2CYE2jzp93SgJuQkMdkTUfwO/xu4CHq8Ph09Wwe32oq6pLVS7JSKKem6/wD86qnUSgAemMNQR0cAMKNh95Sd/7f1NSUJqIp8VS0TUV++fFGh0qdNX5wH5SQx2RDQwA6vYXeKOWABYMmv8gHZLRBRrfIrAS0e7jqkPTgnpFTJEFGMGFOxSkiyQEPwXpUGvQ1qSFfOnFmLOpIKQNI6IKNptqBCodajTC7KBcams1hHRwA0o2P3ka7eHuh1ERDHHrwj87QirdUQUOld8FGl3e9HQYkO72xuK9hARxYwtlQKVHfeazRwGTMtktY6IrsyA74o9VVWHNzfvw+nqegiod3Ll52Tgc9fORkFuZuhaSEQUhYQQ+CurdUQUYgM6kpw4U4unX9qAiup6CCEAISCEQHlVHX750gaUVdaGup1ERFHl0yrgZLM6PTENmDM8vO0hougwoGD39sf74fcrgBCQZRmJVhNkWd2V36/g7S37Q9pIIqJoIoTAi0eUwPyDU2RIfMY2EYXAgE7FVp1vAiQJS2aOw+eunY04gx4erw9vbNqL7QeO4+z5plC3k4goauytFTjaoE4XJgNX5Ya1OUQURQZUsTMZ4wAAt1w9E3EGNRvGGfRYu2wWAMAcbwxR84iIos8Lxd2qdZMlyKzWEVGIDKhiN2vCKHyyvxTNNidM8XGB5c02JwBgzqT80LSuF5/sK8XGXUfQanciOyMFt183F2PyhvW6/okztXhj017U1DcjOcGMFfMnXzSI8oHS03hv60E0NNuQnpKAm5fOxIzxIzX9HEQUew7W+rC/Vr1pYkQCcM1IhjoiCp0BVewmFGTDajbi2Te24NODJ3CsvAafHjyBP765BcmJFkzIz0JZZW3gJ5T2HavA6xv34PpFU/G9h9dg9IhM/PaVjWhqtfe4fkOLDb99dRNGj8jE9x5eg1ULp+LVj/bgQOnpwDrlVXV4/q2tmD+5EN9/eA3mTy7Ec29/gorq+pC2nYjoT/vbA9P3T5agkxnsiCh0BlSx+8NrmwFJgt3pxsv/+uyi93/76ubAtATg9088MOAGXmjT7qNYNH0MrpoxFgBwx4p5OFZeg60HjgdOBXe37cBxpCZacMeKeQCArPRknDnXgI27jmLm+FEAgM17jmFCfjZWLZoKAFiVnowTlbXYvOcYHl67NGRtJ6LYdqJJ4JPTPgDAMDOwKp+hjohCa+ADJ3UMc3K5n0s/VbZ/fH4/Ks81YkJ+dtDyCQXZKK+q63Gb8qp6TCgIXn9iQQ7OnGtQ7+wFUF7d8zq97ZOIaCBePOwPTN8zSYJBx2BHRKE1oIrdvKmFFz0rdjDYnW4oQiDRagpanmgxoc3u6nGbNocLiZYL1reaoCgCdmc7khLMaLP3sI7FhDZHz/sEAK/PD5+/6yAthBphZVkOTFNodQ6p0/lKocX+1daZVoHNp9VjQ2o8cMtYHWSehg05fo+1xz7WniRJ8HfLGP0xoGD34OrFA/qPhcqFh0IhxMULL7f+BW9ceFPa5WqNG3YWY932osC80aDHU4+ugcVs5nhUGrOYzeFuQlRj/2rjn7sdgaPKgzNMyEiOD2t7oh2/x9pjH2tHCIE2m21A2w4o2P3t/U8BScL9Ny266L1dxScBScL8KYUDatClWM1GyJKE1guqczZn+0UVt06JFhNaL6i82RztkGUJVpN6YE20mi7ep6P3fQLAqoVTsXzepMC8EAJejxsOp5MVO43IsgyL2QyH0wlFUS6/AfUL+1c75+wC7x5Xr61LNEpYU+CHzd7zDV90Zfg91h77WHtXUiAaULD7rCO89RTs/vr+p5A0CnZ6nQ55WWkoqagJGoqkpKIG08bm9bhNQW4GisvOBi0rqajByKx06HRqGbkgJwMlFTVBQa2kvOaSz7w16HUw6HWB+c5gpygKg53GFEXhwURD7N/Q+/sRBf6Ow8I9U4ww6/3sY43xe6w99rF2riTYhfQEucer/kWqZaxZPm8Sdhwqw45DZTjX0ILXNu5Bc6sDS2aOA6A+7uyF97YH1l8ycxyaWh14feMenGtoCWx73fyuEHfN3IkoKa/BhzsPo7ahBR/uPIyS0zW4du5EDT8JEcWCRpfA+yfVo6JJD9w3jQO4E5F2+lyxO3S8EsUnKoOW/e39T4Pm65rbAADGuAEVAvtk9sR82J1urPv0ENrsLmRnpOBrdy1HWpIVANBqdwaNaZeenICv3bkcr2/cg637S5FkNePOFXMDQ50AQGFuJr64dine23oA7209iIyUBHxp7dXIz8nQ7HMQUWz4Z4mAu+Ma6FvHykiJl2HjWVgi0ojkd57tU4Htg22HsG77IfUug8DNBz2XCkdlp+O7D94YqjZGhM4LHSVJ4qlYjciyjASrFTa7neV/DbB/Q6/VLbD2LQVOH2CQgXc+p0dBZgL7WEP8HmuPfay9ziyRmJDQ79Oy/S+tCdEV6HoIMFZLPG695uKBgomIYs3rpQJO9QoV3DRaQoaZd8wTkbb6HOwWTBuNsSOHQ0DgVy99CEgSvnnvqsD7kgRYTEZkpCQG3VRARBSLHF6B10rVP351EnDfRIY6ItJen4NdWpI1cB3b6LxhkCQJY0cO16xhRESR7O0TAm0edXplvoTsBAY7ItLegO5y+I/7rg91O4iIoka7T+Afx9RqnQTg/skMdUQ0OAY+QPGl9DLGHRFRLHj/pEBTuzq9LA8YlcRgR0SD44oGKO5Rx80VDHZEFIu8foGXjnbdWPbAFD5Pk4gGz8AHnOOQHkREF9lQIXDeqU4vzAHGpbJaR0SDZ0DBrvvdsID6WJG6Zhs27z6KVrsLD6y+KiSNIyKKJH5F4G9Huv7ofXAyq3VENLgGFOx6uht2fD4wdcwIPPGb11F6+lzQs1yJiGLBljMCZ23q9MxhwNRMVuuIaHCF9M/JRKsJBoMOB0vPhHK3RERDniIEXuxereO1dUQUBgOq2O0qPhk0LwB4vD4cPVUNj8fX78dfEBFFuk+rgFMt6vTENGAOh/kkojAYULD76/uf9n5XrCRhxLDUK2kTEVFEEULghcNdz8x8cIrMP3CJKCxCfles1RyP26+bO+DdEhFFmt3ngJJGdXpMCrA4N7ztIaLYNaBgd8PiaZAQ/NeoQa9DWrIVkwtzEW80hKRxRERDnRACfynuqtY9xGodEYXRgILdjYun4+y5RjS22QEAaYlWjMhKg8yDGRHFmAPngeJ6dXpUEnB1XnjbQ0Sxrd/Bbuv+Uqzbfgg2pztoeYLZiBsXT8fSWeND1jgioqEu6Nq6yRL/wCWisOpXsHtz815s2n1MnbngGjubox2vfLgbDS02fO7aOSFrIBHRUFVcJ7CvVp3OTQCWj2KoI6Lw6vNASyfPnsemXUfVmY5QZ4qPgyk+rmslIbBp9zGcPHs+pI0kIhqKulfrHpgsQS8z2BFRePW5Yrd1/3EA6k0SNy6ehqumj4XFZAQAOFxu7DhUhnXbD8Hj82Pr/uMYPWKYNi0mIhoCShsFPqtRp4dbgOsLGOqIKPz6HOzKq+oAScI9NyzAvMmFQe9ZTEasWDAZSQkmvPjudpyqYsWOiKJb92rdfZNYrSOioaHPp2LbHC4AwIxxvT8DtvM9m6P9CptFRDR0nWwW2HpWnc4wATeNZqgjoqGhz8FOJ6urery+XtfpfE/mX65EFMVePNx189i9kyQYdTzmEdHQ0Odgl5pkAQBs2XOs13W27C0BAKQlWa+wWUREQ1NFi8DmM2qwS4kHbh7DUEdEQ0efr7EbNyoL5+pb8K+dh1Fd34yF08YgPTkBANDQYsPOojIUl1UBkoTx+dmaNZiIKJxeOCzQWa+7d6KEeD2DHRENHX0OdtfOnYgdh8rg9flRXFalhrgLCYE4gx7XzJkQyjYSEQ0Jp1sFNp5WY12yEbh1LEMdEQ0tfT4Vm56cgPtvWgSdLKnj2HX/AQAhoNPJuH/1VYFKHhFRNHmxW7Xu8xMlmAwMdkQ0tPTryROzJ+YjMyUR6z4tQunpGng86s0ScQYdJuRn44ZF05CXlaZJQ4mIwqmyTeCjjmpdkhG4bRxDHRENPf1+VmxeVhq+cvs1UISAw+mGgIDVHM/nIxJRVHvxsIDSUa77/AQJZlbriGgI6new6yRLEhIs8aFsCxHRkHS2TeDDCjXVJcaxWkdEQ1efr7EjIopVfz0i4O+o1t01QYIljsGOiIYmBjsiokuotgn8q1xNdQlxwB3jGeqIaOhisCMiuoQXu1Xr7hwvwcpqHRENYQx2RES9qLYJrD+lpjqrAbhzAkMdEQ1tDHZERL144XC3at0ECQms1hHREMdgR0TUg8q24Gvr7mK1jogiAIMdEVEPulfr7ma1jogixIDHsQsHh8uN1z7ajaKyswCAaWNG4M6V82CON/a4vt+v4N2tB3DkZBUaWuwwGQ0Yn5+NtctmITnBHFjvF3//F8oqzwdtO3viKDy89mqtPgoRDWGnW4PHrbuTd8ISUYSIqGD3l3e2odnmwGN3XQcAeGn9Trzw7nZ89c7lPa7v8fpQWduIG66ahtxhqXC2u/H6R3vw+9c244kvrg5a96rpY7F66fTAfJw+orqGiELoL8VdT5m4dxLHrSOiyBExp2LPNbTgaHk17rtxEQpyM1GQm4l7b1yIwyerUNvY2uM2pvg4fOPzKzF7Yj6GpyWhICcTd66cj8raRjS12oPWjTPokGQ1B35M8XGD8bGIaIgpbxHY2PFM2GQ+E5aIIkzElKXKq+phMhqQn5MRWFaQkwmT0YDyqjoMT0vq035cbg8k4KLgtudoOXYfKUeixYRJhTm4afF0xBsNve7H6/PD5/cH5oVQ/yGQZTkwTaEly3LQK4UW+1f152IfOn+D758sw2rUhWzf7GPtsY+1xz7WniRJ8HfLGP0RMcGuzeFCgsV00fIEiwltdlef9uH1+fD2lv2YM7kAJmNXsJs7uRDpyVYkWkyoqW/GOx8fQFVdE77x+ZW97mvDzmKs214UmDca9Hjq0TWwmM2QJP6FryWL2Xz5lWjAYrl/jzf4sPmMDQCQbpLw4KxEmAyh/32O5T4eLOxj7bGPtSOEQJvNNqBtwx7s3t92MCgg9eTxh24CAPR0eBVC9PzGBfx+Bc+/vRVCCNy9an7Qe4tnjA1M52SmIDM1Ef/zlw9Qea4ReVlpPe5v1cKpWD5vUlA7vB43HE4nK3YakWUZFrMZDqcTiqKEuzlRh/0L/OozX2D6/skSfG4HbO7Q7Z99rD32sfbYx9q7kgJR2IPdstkTMGdi/iXXSUu2oqquGW2Oiytzdmc7Enuo5HXn9yv401ufoKHFjm/eszKoWteTvOFp0Mky6prbeg12Br0OBn3XKZrOYKcoCoOdxhRF4cFEQ7Hav6WNAp9Uqr+7GSZgzWho1g+x2seDiX2sPfaxdiI62FnN8bCa4y+7XkFuBlxuLyqq6wPX2VVU18Pl9qIgN7PX7TpDXX1zG755z6o+/bdq6lvgVxQkWS8dGIkoejx7qOsfqPsnS4jX85IKIoo8EXPlY1Z6MiYV5OCl9TtRXl2H8uo6vLR+J6aMzg26ceLJZ9/CwdIzAAC/ouCPb36MynMN+MLNS6AIBa12J1rtzsCND/XNbVi3/RDO1DSgocWGwyer8Nxbn2DEsFQUXiIwElH0OHheYFeNOj3cAtw8hqGOiCJT2Ct2/fGFW5bg1Y9245l/bAQATB07AnetnBe0zvnGNrjcHgBAc5sDxR2DGf/o+feC1vvmvSsxbmQWdDodSk+fw5a9JXB7vEhJtGDy6FzctHg67/ghigFCiKBq3cPTJMTpGOyIKDJJfudZXhAWAp13sEiSxGvsNCLLMhKsVtjsdl7XoYFY7d+d1QLf2qJ+3pGJwMurZehlbYJdrPbxYGIfa499rL3OLJGYkNDv6+1YkiKimKUIgWcPdv3D9Mh07UIdEdFgYLAjopi15YzAiWZ1enwqsCwvvO0hIrpSDHZEFJN8isCfiroum3h0hszBxYko4jHYEVFMWn9KoLJNnZ6RCczLCm97iIhCgcGOiGKO2y/w52JW64go+jDYEVHMeeeEwHmnOr0oB5iWyVBHRNGBwY6IYordI/DC4a5q3SPTeRgkoujBIxoRxZS/HxVocavTK0ZJGJvKah0RRQ8GOyKKGXVOgVdK1GqdXgYenc5QR0TRhcGOiGLGc0UCbvUx0bhtnITsBAY7IoouDHZEFBPKWwTWnVKrdVYD8NAUhjoiij4MdkQUE353QIHScc/E/ZMlJBkZ7Igo+jDYEVHUO3BeYEe1Op1pBu4Yz1BHRNGJwY6IopoQAr/drwTmvzxdQryewY6IohODHRFFtc1nBI41qtOFycD1+Qx1RBS9GOyIKGp5/QLPHuoajPirM2XoZAY7IopeDHZEFLVePy5QZVOnZw0DFmSHtz1ERFpjsCOiqNTcLvCXYrVaJwF4bLYMSWK1joiiG4MdEUWl54oE7F51+qbREsbx0WFEFAMY7Igo6pxsFninTK3WmfV8dBgRxQ4GOyKKKkII/Gpf12DED06RkGZisCOi2MBgR0RR5dMqYF+tOp1lAe6cwFBHRLGDwY6IoobXL/BMt8GI/32WDKOOwY6IYgeDHRFFjTeOC5ztGN5keiawLC+87SEiGmwMdkQUFVraBf7cbXiTb3B4EyKKQQx2RBQVnj3UNbzJjYUSxqcx1BFR7GGwI6KId7RB4N3O4U0MHN6EiGIXgx0RRTS/IvDzPQo6nwj7pakS0s0MdkQUmxjsiCiivX9SoKRRnS5IAm4fz1BHRLGLwY6IIlZLu8DvD4rA/LfnydDLDHZEFLsY7IgoYv3hoECbR51emS9h5jCGOiKKbQx2RBSRjtQLvHey64aJf5/JUEdExGBHRBHnwhsmvjyNN0wQEQEMdkQUgd47KVDapE4XJgO3jWOoIyICGOyIKMI0ui64YWIub5ggIuqkD3cD+sPhcuO1j3ajqOwsAGDamBG4c+U8mOONvW7z4vvbsav4VNCy/Ox0fPehmwLzXp8fb27ei71HK+D1+TF+VBbuXjUfKYkWbT4IEQ3YL/cK2DpumLi+QMIM3jBBRBQQUcHuL+9sQ7PNgcfuug4A8NL6nXjh3e346p3LL7ndpIIc3L96UWBer9MFvf/6xj0oLjuLh9cuhcVkxJub9uJ3r27CE19cDVlmUZNoqPi0SmDzGbVal2QEHpvFUEdE1F3EpJZzDS04Wl6N+25chILcTBTkZuLeGxfi8Mkq1Da2XnJbvV5GktUc+LGYuip8rnYPdhwqw23L52BCfjbyhqfhoZuXoLq+BSUV57T+WETURw6vwE93K4H5b8yWkBLPYEdE1F3EVOzKq+phMhqQn5MRWFaQkwmT0YDyqjoMT0vqddsTZ2rxnadfgSk+DmPyhuHmq2ci0WICAJypbYRfUTAhPzuwfnKCGdkZySivqsOkwhztPhQR9dmzBwXqnOr03CxgVT5DHRHRhSIm2LU5XEjoCGPdJVhMaLO7et1ucmEuZo0fhdQkKxpb7Hhv6wH86uUP8fgXVsOg16HN7oJeJwdV8QAg0WJCm6P3/Xp9fvj8/sC8EOrpIVmWA9MUWp2nxXl6XBtDuX8P1yt447j6e2XUAY8v0EOni7xgN5T7OFqwj7XHPtaeJEnwd8sY/RH2YPf+toNYt73okus83nGjQ0+HcSFEz290mD0xPzCdk5mCkVlpeOK3b+DIySrMGD+y1+0ELh3ONuwsDmq30aDHU4+ugcVshiRF3j84kcRiNoe7CVFtqPWv1y/wv7vaAr+Rj80zYXxWfFjbdKWGWh9HI/ax9tjH2hFCoM1mG9C2YQ92y2ZPwJxu4asnaclWVNU191hBszvbA6dV+yIpwYzUJAvqmtoAAIlWE3x+BQ6XO6hqZ3O0ozAns9f9rFo4FcvnTQrMCyHg9bjhcDpZsdOILMuwmM1wOJ1QFOXyG1C/DNX+faHYj7ImtT3jUoG1hV7Y7L4wt2pghmofRxP2sfbYx9q7kgJR2IOd1RwPq/nyf30X5GbA5faioro+cJ1dRXU9XG4vCnJ7D2AXsjvb0dzmQJJVDYMjh6dBJ8soqagJVPdabU7U1Lfg1mtm97ofg14Hg77r7trOYKcoCoOdxhRF4cFEQ0Opf8+0Cvy5SG2LLAGPz5chQ0BRIvt3bCj1cbRiH2uPfaydiA52fZWVnoxJBTl4af1O3HPDAgDAy+s/w5TRuUE3Tjz57Fu45epZmDF+JNo9Xnyw7RBmjh+JRKsJja12vPvxAVjN8Zg+Tj0Na4qPw6LpY/Dmpr2wmowwm4x4c/M+5GQkY0J+Vlg+KxEBPkXgqZ0KPB3/btw1QcL4NF7mQER0KRET7ADgC7cswasf7cYz/9gIAJg6dgTuWjkvaJ3zjW1wudXRS2VJQk19M3YfPgVnuwdJVhPGjhqOh2+9GvFGQ2Cb26+bA1mW8NzbW+Hx+jB+VBYeuHM5LwwlCqOXjwocbVCnRySoz4MlIqJLk/zOs5F9TmOI6LzQUZIknorViCzLSLBaYbPbWf7XwFDq35PNAg+uV+BT1FOwf1wpY0pG5Ae7odTH0Yp9rD32sfY6s0RiQkK/T8uyJEVEQ4rXL/DUDjXUAcA9E6WoCHVERIOBwY6IhpQXjwicaFanC5KAL/EULBFRnzHYEdGQUdIo8OJh9VIGnQT8YJGMuAgciJiIKFwY7IhoSHB3nIL1d1yi+tAU3gVLRNRfDHZENCT88ZBARas6PS4VeHAKQx0RUX8x2BFR2O2qEfjHMbVUZ5CBHyyUoZcZ7IiI+ovBjojCqtEl8MMdXUMmfG2mhMIUhjoiooFgsCOisFGEwH/vVNDcrs4vzAHuGM9QR0Q0UAx2RBQ2r5QI7KpRp9NMwPcXyFf0jEQioljHYEdEYVHaKPD7g11PafnBQhmpJoY6IqIrwWBHRIPO6RX4f9u7ni5x7yQJ87IZ6oiIrhSDHRENKiEEfr5H4KxNnZ+YBjzCp0sQEYUEgx0RDap3TwqsL1dPwZr1wFOLZRj4dAkiopBgsCOiQXOsQeAXe7quq/vufAm5CQx1REShwmBHRIOipV3giW0KvB3X1d0+TsLKfB6CiIhCiUdVItKcXxH4wacKah3q/JQM4LFZrNQREYUagx0Rae65IoE959Tp1HjgJ0t4XR0RkRYY7IhIU9vOCrx4RL2uTicBP1oiI8PMUEdEpAUGOyLSTGWbwFPdngP71ZkSZg5jqCMi0gqDHRFpotUt8B9bFNi96vw1ecDdExjqiIi0xGBHRCHnUwSe2KoEBiEuTAa+t5DPgSUi0hqDHRGFlBACP90tsP+8Op8SD/x8mQyLgaGOiEhrDHZEFFL/LBF476R6s0ScDPzsahlZVoY6IqLBwGBHRCGz/azAb/Z3PVni+wslTM5gqCMiGiwMdkQUEmVN6iDEnbHui1MlrOCTJYiIBhWPukR0xWrsAt/cosDlU+evGyXh4ams1BERDTYGOyK6Ik0uga9vUtDgUucnpQPfWyDxDlgiojBgsCOiAXN41Epd57AmIxOBXyyTEa9nqCMiCgcGOyIaELdf4D8/UXC8SZ3PNAO/Xi4jOZ6hjogoXBjsiKjf/IrAf32qBMaqS4wDfn2tjOEWhjoionBisCOifhFC4Gd7BD6uVOfjdcAvr5GRn8xQR0QUbgx2RNRnQgj8Yq/AO2XqoCY6CfifpTLHqiMiGiIY7IioT4QQ+OVegTeOq6FOAvCDRRIW5DDUERENFQx2RHRZQgg8vU/g9W6h7v8tlLCSAxATEQ0pPCoT0SUJIfCrfQKvlXaFuu8vlHBDIQ8fRERDjT7cDegPh8uN1z7ajaKyswCAaWNG4M6V82CON/a6zaM/frHH5bdeMxsrFkwGAPzi7/9CWeX5oPdnTxyFh9deHYpmE0UsIQR+vU/g1QtC3Y0MdUREQ1JEBbu/vLMNzTYHHrvrOgDAS+t34oV3t+Ordy7vdZv/+/odQfNHT1Xj7x/swIzxI4OWXzV9LFYvnR6Yj9NHVNcQhZxfUW+UeOtEV6j7HkMdEdGQFjHp5VxDC46WV+O7D96I/JwMAMC9Ny7ET19cj9rGVgxPS+pxuySrOWi+6EQlxo7KQkZKQtDyOIPuonWJYpXHL/DDHQKbz3QLdQsk3MRQR0Q0pEVMsCuvqofJaAiEOgAoyMmEyWhAeVVdr8Guuza7C4dPVuHB1Ysvem/P0XLsPlKORIsJkwpzcNPi6Yg3GkL6GYgigdMr8J0tCvbWqvM6Sb37lTdKEBENfRET7NocLiRYTBctT7CY0GZ39Wkfnx0+ifg4A2aMzwtaPndyIdKTrUi0mFBT34x3Pj6AqromfOPzK3vdl9fnh8/vD8wLoVY2ZFkOTFNoybIc9EqhJcsyml0K/u0jP441qMuMOuD/rtZhYS77PBT4HdYe+1h77GPtSZIEf7eM0R9hD3bvbzuIdduLLrnO4w/dBEA9HXQhIUTPb/RgZ1EZ5k4ugOGC6+cWzxgbmM7JTEFmaiL+5y8foPJcI/Ky0nrc14adxUHtNhr0eOrRNbCYzZAkjuulJYuZp8y1cM6m4OH3bShvVv8wSTJK+MNNVswYHvbDRNThd1h77GPtsY+1I4RAm802oG3DfsReNnsC5kzMv+Q6aclWVNU1o81xcWXO7mxHYg+VvAuVVZ7H+cY2fGnt1ZddN294GnSyjLrmtl6D3aqFU7F83qTAvBACXo8bDqeTFTuNyLIMi9kMh9MJRVHC3ZyocqxBwbe3+NHQ8SuWYQKeuU6HQms7bPbwti2a8DusPfax9tjH2ruSAlHYg53VHA+rOf6y6xXkZsDl9qKiuj5wnV1FdT1cbi8KcjMvu/2OohPIG56G3GGpl123pr4FfkVBkrX3wGjQ62DQ6wLzncFOURQGO40pisKDSQhtOq3gv3cKuDuq/iMSgV9fKyPbKqAo/C5rgd9h7bGPtcc+1s6VBLuIOUGelZ6MSQU5eGn9TpRX16G8ug4vrd+JKaNzg26cePLZt3Cw9EzQti63BwdKzuCq6WMu2m99cxvWbT+EMzUNaGix4fDJKjz31icYMSwVhX0IjESRSgiBPxcr+P72rlA3K0uP56/XI9vKywmIiCJR2Ct2/fGFW5bg1Y9245l/bAQATB07AnetnBe0zvnGNrjcnqBl+45WQAiBOZMKLtqnTqdD6elz2LK3BG6PFymJFkwenYubFk/nhaEUtdp9Aj/+TGDj6a6K3E2FEn58nRVul4OVOiKiCCX5nWd5BA+BzgsdJUniqViNyLKMBKsVNrud5f8rcN4h8PhWBcca1XkJwFdnSrhvsg6JCQnsXw3xO6w99rH22Mfa68wSiQkJ/T4tG1EVOyK6MjurBX64Q0GrW5036YEfXiVjyQiJd3MTEUUBBjuiGOBXBJ4rEnjxSFc1OcsC/PRqGWNSGeiIiKIFgx1RlGt0Cfxgu4L957uWLc4F/t9CGYlGhjoiomjCYEcUxfaeE/ivHQoaO8an00nAV2ZIuGciT70SEUUjBjuiKNTuE/jdAYHXj3edes0wAf+9WMb0YQx0RETRisGOKMoca1BvkDjT1rVsznD1JolUE0MdEVE0Y7AjihI+ReDFwwIvHBbwdxTqjDrg32ZIuH28BJmnXomIoh6DHVEUKG0U+N9dCkqbupaNTwWevEpGfhIDHRFRrGCwI4pgDq/Ac4cEXjsu0PmwCJ0EPDhFwkNTJOhlhjoioljCYEcUobadFfj5HgV1zq5lhcnAEwtkTEpnoCMiikUMdkQRptom8Ot9CrZVdS0z6oAvTpXw+Yms0hERxTIGO6IIYfOoN0e8Virg7fZ4xvnZwHfmyshJYKAjIop1DHZEQ5xPEXinTH0kWOczXgEgzQR8Y7aE5SM52DAREakY7IiGKCEEtp0F/nBIwenWruVxMnD3RAn3T5ZgMTDQERFRFwY7oiFGCIEd1cCfihScaAp+b8UoCV+ZISHLykBHREQXY7AjGiKEEPisBni+SMGxxuD3pmQAX58lY3IGAx0REfWOwY4ozPyKwNazwEtHLw50Y1OBL0+TsSgHvI6OiIgui8GOKEzafQLrTgn8s0Sgyhb83ugU4EtTZSwZwUBHRER9x2BHNMgaXQJvnxB447hAizv4vcJk4ItTZVydBz7blYiI+o3BjmgQCCGwvxZ464SCrWcBvwh+f/Zw4N5JMuZlsUJHREQDx2BHpKFWt8C/ytUK3Zm24Pd0EnDtSPVpEePTGOaIiOjKMdgRhZjXr97dur5cwadVgE8Jfj8lHlgzWsItYzhsCRERhRaDHVEICCFwtAHYUCGw8XTwEyI6zRgG3DpWwtUjJBh0DHRERBR6DHZEA6QIgSP1wJZKgY/PCJx3XrxOmkkdVHjNGAn5SQxzRESkLQY7on7w+gUO1gHbzwp8UilQ77p4HaMOWDpCwvUFEuZkAXqZgY6IiAYHgx3RZTQ4BXZWC+yoFth7DnD6Ll5HJwFzstSbIZblSbDGMcwREdHgY7AjuoDTK3CoDthXK7DvnMCJ5p7XM8jA3CzgmpESFudKSDQyzBERUXgx2FHMc3rVGx8O1QnsrxU40nDxnaydko3AghwJi3KA+dmszBER0dDCYEcxRQiBeidwpAEorhMoqhc40XTxgMGdJABjUoFFORIW5UiYkAboeM0cERENUQx2FNVa3QKljcCxRoFjDQIljUBDDzc8dJebAMwZLmFOloRZw4EknmIlIqIIwWBHUcGvCFTbgZPNwIkmgbJmgbJmoK6HIUguVJAETMuUMDUTmJ7JQYOJiChyMdhRRPH4BU63CpxuEShvEShvAU63qo/rcvsvv73VAExIAyamS5iaIWFyBityREQUPRjsaMhp9wmcswM1dqDaLlBlAyrbBM62+XHO0QKll+vhLmQxAGNTgDGp6rVxk9Il5CYAssQgR0RE0YnBjgaVEAJtHqDOAdQ6gPNOob46gFqHGugudw3chXSSel3cqCRgdIqEMSkSxqYAWVZAYogjIqIYwmBHIeFTBJrbgaZ2oMkFNLYLNLnU6XoXUO8UaHABDU7A08tQIpdjMQCjknXIsSgYkagGufwkCXmJQByfvUpERMRgR8GEEHB6AZsXsHnUnza3endpqxto8wCtbqC1XaDZDbS0Ay1udb1QSDMB2VYg2yohx6pO5ySo4S3dLCMxIQE2ux2KMsB0SEREFMUiKtit/7QIR05W4ez5Juh1Mp7+9j2X3UYIgQ+2H8KnB0/A2e7BqOx03L1qPrIzUgLreH1+vLl5L/YerYDX58f4UVm4e9V8pCRatPw4IeNTBNp9gKv7j7dr2ukVcHjVR2E5veqPwwvYvQIOT+c0YPeor329hq2/EuOADHPHj0nCMAsw3AJkWiT11QzE63uvvPG0KhER0aVFVLDz+xXMnDAKBbmZ2HHoRJ+2+eizI9i8+xgeWH0VMlMT8a8dRfj1Pz7CDx+9FfFGAwDg9Y17UFx2Fg+vXQqLyYg3N+3F717dhCe+uBqyLPerjVU2AbdPwKuoTy/wKoDXj8C8xy8Cyzzd3vN0vLr96nTnj9sv4PYDbp/6XnvHa+e0y9f74LqDwWIAUuLVJzIkxwNp8RJSTWrlLTVeQmo8kG4G0k2XDm1ERER05SIq2K1eOgMAsLOorE/rCyGwec8xXL9oKmaMHwkAeGD1Yvznr17BnqPlWDJzHFztHuw4VIaHbl6MCfnZAICHbl6Cx3/zOkoqzmFSYU6/2njfB344vf3aJOz0sjoMiMUAJMR1/5FgjQOSjECiEUiMkwLTSUY1zPHaNiIioqEjooJdfzW02NHmcGFCQXZgmUGvw5i84SivqsOSmeNwprYRfkUJhDoASE4wIzsjGeVVdb0GO6/PD5+/a+A0IbQvm8kSYNSpP/H6jh+dWgnrnDfpAZNegtnQNW82ABaDBHPHdOe8NQ6wxgFxcmSc5uysnva3ikp9w/7VHvtYe+xj7bGPtSdJEvz+PgzO2oOoDnZtDnXcjESLKWh5osWEpja7uo7dBb1OhsVkvGidzu17smFnMdZtLwrMGw16PPXoGtww2gAFEgyyWs3Sd7waZMCg6z4tIa5jvvurUS91hDcJRr26XA1t6naREMC0ZjGbw92EqMb+1R77WHvsY+2xj7UjhECbzTagbcMe7N7fdjAoIPXk8Yduwsjs9AH/Ny6MQgKXr65dbp1VC6di+bxJXesLAa/Hje/Ok7pV70JQxRMAvIDbC7ivfG8RTZZlWMxmOJxO3hWrAfav9tjH2mMfa499rL0rKeKEPdgtmz0BcybmX3KdtGTrgPbdWalrdbiQlND1l4XN0R54L9Fqgs+vwOFyB1XtbI52FOZk9rpvg14Hg14XmO8MdoqiDMpp2VimKAoPJhpi/2qPfaw99rH22MfaiehgZzXHw2qO12Tf6clWJFpMKKmoQd7wNACAz+9HWWUt1l4zGwAwcngadLKMkooazO4ImK02J2rqW3BrxzpEREREkSDswa4/mlrtcLjcaG5zQBECZ2sbAQAZqYmIj1OHLnny2bdwy9WzMGP8SEiShGvnTsSGHcXITElEZmoiNuwsRpxBj7mTCgAApvg4LJo+Bm9u2guryQizyYg3N+9DTkYyJuRnhe2zEhEREfVXRAW797YdxK7iU4H5H//5fQDAN+9diXEj1RB2vrENLnfXYxBWLJgMj8+Hf27YBWe7G/k5GXjs7hWBMewA4Pbr5kCWJTz39lZ4vD6MH5WFB+5czjt+iIiIKKJIfudZXhAWAp13sEiSxGvsNCLLMhKsVj5STCPsX+2xj7XHPtYe+1h7nVkiMSGh39fbsSRFREREFCUY7IiIiIiiBIMdERERUZRgsCMiIiKKEgx2RERERFGCwY6IiIgoSjDYEREREUUJBjsiIiKiKMFgR0RERBQlGOyIiIiIogSDXYh0PkaMjxPTjhAi8EOhx/7VHvtYe+xj7bGPtXclmYLPig0Rt8eL9nZXuJtBREREUSI+3gRjnKFf2+g1akvM8fn9+PGf1+PJR2+BOd4Y7uZEJZfbg8efeR3/89jtMBnjwt2cqMP+1R77WHvsY+2xj7XnbHfjh8++gycfvQVGMNiFhSzLaHO2Q5ZlSJIU7uZEJUmS4Pb6IEkS+1gD7F/tsY+1xz7WHvtYe90zRb+31aA9RERERBQGDHZEREREUYLBLkT0Oh1uXDwNep0u3E2JWuxjbbF/tcc+1h77WHvsY+1dSR/zrlgiIiKiKMGKHREREVGUYLAjIiIiihIMdkRERERRguPYaeD3r23G2fNNsDlcMMcbMSE/C2uvmY3kBHO4mxYVGlpsWP9pMY6fPoc2hwtJVjPmTS7A9VdN5cW8IbT+0yIcOVmFs+eboNfJePrb94S7SRHvk32l2LjrCFrtTmRnpOD26+ZiTN6wcDcrKpRV1uKjz46gsrYRrXYXHr1tGaaPGxnuZkWNDTuKcfD4GdQ2tiJOr0dBbgbWXjMbw9OSwt20qLF1fym2HTiOxhY7ACArIxk3XjUNk0fn9ms/DHYaGDtyOFYtmoIkqxktNife3LQXf3rzY/zngzeGu2lR4XxjK4QQuOeGBchISURNfTNeWr8Tbq8Pty2fE+7mRQ2/X8HMCaNQkJuJHYdOhLs5EW/fsQq8vnEP7l41H4UjMrH9wHH89pWNePKRW5CaZA138yKe2+ND7rBULJw2Bn988+NwNyfqnKisxdJZ4zEqOx2KIvDuJwfwzD8+wpOP3NLvR15Rz1ISLLhl2SxkpiQAAD4rPoU/vL4F33t4NbIzUvq8HwY7DSyfNykwnZZkxcqFU/Ds61vg9yvQ6Xj2+0pNKszFpMKuv2AyUhJwvrEN2w6UMtiF0OqlMwAAO4vKwtyS6LBp91Esmj4GV80YCwC4Y8U8HCuvwdYDx7F22awwty7yTR6d2+/KBvXdY3evCJq//6ar8J1fvYLK2kaMyRseplZFl6ljRwTN37JsJrYdKEVFdX2/gh1ThsYcLjf2HClHQW4mQ52GXG4Pn9FLQ5bP70fluUZMyM8OWj6hIBvlVXVhahXRwLncHgDgcVcjiqJg79FyeLw+5Odk9mtbVuw08taWffhkX2nH/5QMfPWOa8PdpKhV39yGj/eV4LZrWa2jocnudEMRAolWU9DyRIsJbXZXmFpFNDBCCLyxaS9Gj8hETmbfK0l0edV1zfjpi+vg9flhjNPjkduuQXZGcr/2wWDXR+9vO4h124suuc7jD92EkdnpAIAV8ydj0bQxaGx1YN32Q3jxvU/x1Tuv5QOTL6G/fQwALTYnnvnnRswaPypwiot6N5A+ptC58LdfCHHxQqIh7pUPd6Oqrgnfuf+GcDcl6gxLS8T3Hl4DV7sHB46fwV/f345v3Xt9v8Idg10fLZs9AXMm5l9ynbTkrgugreZ4WM3xGJaWhKz0JDz+m9dRUV2Pgtz+lVRjSX/7uMXmxNMvbUBBbibuuXGh1s2LCv3tYwoNq9kIWZLQekF1zuZsR6LF1MtWREPPKx/uQvGJSvzH/dcjJdES7uZEHb1Oh8zURADAyOx0nKlpwMd7j+GeG/r+bxyDXR91BrWB6Hxmm9fvD12DolB/+ri5zYGnX/4QecPT8MBNiyCzEtonV/I9poHT63TIy0pDSUUNZozvGoKjpKIG08bmhbFlRH0jhMArH+7GoeOV+NZ9q5CenBDuJsUEgf5nBwa7EKuorsfpmgaMHpEJs8mIhmYb3t92EBkpCSjo5wWQ1LMWmxO/fGkDUpOs+Ny1s2FztgfeS7JyrMBQaWq1w+Fyo7nNAUUInK1tBABkpCYinsMb9NvyeZPwwrvbMTIrHQW5Gdh+8ASaWx1YMnNcuJsWFdo9XtQ3tQXmG1rsOFvbCIvJyOFkQuCfG3Zh79FyfOX2axEfp0er3QkAMBnjEGdglAiFdz7ej0mFuUhJNMPt8WHv0QqcOFOLf7/run7tR/I7z4rLr0Z9VV3XjNc+2o2quma4PV4kWc2YVJiD6xdNZdk6RHYWleFvH+zo8b1nv/fg4DYmir34/nbsKj510fJv3rsS40ZmhaFFke+TfaX4aNdhtNldHQMUz+FQESFy/Mw5PP3Shxctnz+1EA+uXhyGFkWXR3/8Yo/L779pERZOGzO4jYlSf/tgB0pP16DN7oLJGIeczBSsWDAFEwuyL79xNwx2RERERFGCA6sRERERRQkGOyIiIqIowWBHREREFCUY7IiIiIiiBIMdERERUZRgsCMiIiKKEgx2RERERFGCwY6IiIgoSvA5IEREF3h/20Gs214EYOg+uWBnURkaW+0AgGvnToQ53hjmFhHRUMBgR0QUgT4rPomyyvMAgAVTRzPYEREAnoolIiIiihqs2BER9UH307P3XL8AzTYndhWfhM3Zjqz0ZNyybFbQw7p/8fd/BSpq333wRnyyvxRHTlbB4/WhMDcTty2fg9xhqYH1Ox+ynppkwU++dnuP+/nRVz+Hxlb7RQ+7//7v3gxM/+irn0N6ckJoPzwRRQxW7IiI+untj/dj/adFaGpzwOvzo7K2Eb9/bRMaWmw9rv/71zZj9+FTcLjc8Pr8KD19Dj//+79wvqltkFtORNGOFTsion5SFIHH7l6B/Ox0/GPDLuw9Wg6fX8G+YxVYtXDqReunJVvx3YduRJxBj5fX70TRibNod3vx/taDeHjt0n79t8eNzMKz33vwokoeq3REBLBiR0TUb4umj8HEgmyY4uMwZ1J+YHlji73H9dcsnYH05AQkWky49ZrZgeUlFTWat5WIYguDHRFRPw1PTw5MGw1dJz68fn+P66cmWXucdrjcUBQleGURPKsoFywgIroEBjsion7SyVLXjNT7ep2aWu09TltMRsiyehjW69RXj88XeF8Rotfr9iSpD/9hIoo5DHZERBp7f9shNLTYYHO0460t+wLLJ+R33UWblqxW8uxON8qr6yCEwObdR9Fqd/W4T4upa9y6qvPNEIKVPSLizRNERJprbnMEDUkCAPFGA1YvnRGYnz9lNN795AAA4GcvrkdcnB5ujw9xBj08Xh8uVJibiYOlZwAAz76xBcDFQ6UQUexhxY6ISGNfue0aLJw6GhaTEQa9DuNGZeHb912PYamJgXVWzJ+MVQunIjXJAp1OxvC0JDx62zUYmZXW4z6XzhqHpbPGIznBzNOyRBQg+Z1nWb8nIgoxDkdCROHAih0RERFRlGCwIyIiIooSPBVLREREFCVYsSMiIiKKEgx2RERERFGCwY6IiIgoSjDYEREREUUJBjsiIiKiKMFgR0RERBQlGOyIiIiIogSDHREREVGUYLAjIiIiihL/Pwo9+TflMcK4AAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# YOUR CODE HERE: Change 'relu' to 'elu', 'selu', 'swish'... or something else\n",
    "activation_layer = layers.Activation('tanh')\n",
    "\n",
    "x = tf.linspace(-3.0, 3.0, 100)\n",
    "y = activation_layer(x) # once created, a layer is callable just like a function\n",
    "\n",
    "plt.figure(dpi=100)\n",
    "plt.plot(x, y)\n",
    "plt.xlim(-3, 3)\n",
    "plt.xlabel(\"Input\")\n",
    "plt.ylabel(\"Output\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "95676fe7",
   "metadata": {
    "papermill": {
     "duration": 0.007238,
     "end_time": "2023-06-30T15:59:54.950633",
     "exception": false,
     "start_time": "2023-06-30T15:59:54.943395",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Keep Going #\n",
    "\n",
    "Now move on to Lesson 3 and [**learn how to train neural networks**](https://www.kaggle.com/ryanholbrook/stochastic-gradient-descent) with stochastic gradient descent."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2add7fc5",
   "metadata": {
    "papermill": {
     "duration": 0.007256,
     "end_time": "2023-06-30T15:59:54.965931",
     "exception": false,
     "start_time": "2023-06-30T15:59:54.958675",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "---\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "*Have questions or comments? Visit the [course discussion forum](https://www.kaggle.com/learn/intro-to-deep-learning/discussion) to chat with other learners.*"
   ]
  }
 ],
 "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": "3.10.10"
  },
  "papermill": {
   "default_parameters": {},
   "duration": 26.722385,
   "end_time": "2023-06-30T15:59:58.522844",
   "environment_variables": {},
   "exception": null,
   "input_path": "__notebook__.ipynb",
   "output_path": "__notebook__.ipynb",
   "parameters": {},
   "start_time": "2023-06-30T15:59:31.800459",
   "version": "2.4.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}