{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# CondDensityEstimator Hyperparameter Tuning with Optuna\n",
    "\n",
    "In this notebook, we will use Optuna to tune the hyperparameters of a \n",
    "conditional density estimator (CDE). [Optuna](https://optuna.org/) is a \n",
    "framework-agnostic hyperparameter tuning package. Given a range of \n",
    "hyperparameters, it traverses this space in a smart way, exploring parameter\n",
    "values that are more likely to maximize/minimize your objective. \n",
    "\n",
    "In this notebook, we will also take advantage of an early-stopping protocol, \n",
    "where Optuna looks at the validation loss at each epoch of network training and\n",
    "stops the trial if it is unlikely to beat the best set of hyperparameters\n",
    "found so far.\n",
    "\n",
    "Before starting, please make sure you have `keras` and `optuna` installed\n",
    "in your environment.\n",
    "<!--- TODO: figure out what the requirement versions are --> "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "import visual_bars.generate_visual_bars_data as vbd\n",
    "from cfl.experiment import Experiment\n",
    "\n",
    "import optuna\n",
    "from optuna.integration import TFKerasPruningCallback\n",
    "from optuna.trial import TrialState\n",
    "\n",
    "from keras import Sequential\n",
    "from keras.layers import Dense, Dropout"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load Data\n",
    "We will be using the visual bars example data for this tutorial. To learn \n",
    "more about this included dataset, please refer to the visual bars background\n",
    "[page](https://cfl.readthedocs.io/en/latest/more_info/Visual_Bars_data.html).\n",
    "If you would like to see a standard example of running CFL on this dataset\n",
    "without hyperparameter tuning, please refer to the main CFL code \n",
    "[tutorial](https://cfl.readthedocs.io/en/latest/examples/cfl_code_intro.html).\n",
    "\n",
    "Here, we create a dataset of 10000 samples, where the cause is a 10x10 image\n",
    "and the effect is a binary 1D variable. For sake of simplicity, we will be\n",
    "tuning a standard feed-forward network (as opposed to a convolutional \n",
    "neural network), so flatten the images down to vectors of shape (1,100). \n",
    "\n",
    "Since we want to evaluate our models on the same subset of the data across\n",
    "all hyperparameter combinations, we generate in-sample and out-of-sample\n",
    "indices to pass to CFL. If these are not provided, CFL generates this\n",
    "split randomly, and it would look different across every trial. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(20000, 100)\n",
      "(20000, 1)\n"
     ]
    }
   ],
   "source": [
    "# create visual bars data \n",
    "n_samples = 20000 \n",
    "im_shape = (10, 10) \n",
    "noise_lvl= 0.03\n",
    "random_state = 180\n",
    "\n",
    "vb_data = vbd.VisualBarsData(n_samples=n_samples, im_shape=im_shape, \n",
    "                             noise_lvl=noise_lvl, set_random_seed=random_state)\n",
    "\n",
    "# retrieve the images and the target \n",
    "X = vb_data.getImages()\n",
    "X = np.reshape(X, (X.shape[0], np.product(X.shape[1:]))) # flatten images\n",
    "Y = vb_data.getTarget()\n",
    "Y = np.expand_dims(Y, -1)\n",
    "print(X.shape)\n",
    "print(Y.shape)\n",
    "\n",
    "# define train and validation sets to remain constant across tuning\n",
    "in_sample_idx, out_sample_idx = train_test_split(np.arange(X.shape[0]),\n",
    "                                                 train_size=0.75,\n",
    "                                                 random_state=42)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Make a CDE network\n",
    "\n",
    "Next, we need to outline the architecture of our neural network. We use\n",
    "Optuna's `trial.suggest_xxx` methods to determine the number of layers, number\n",
    "of units in each layer, and the amount of dropout after each layer in the\n",
    "network. For more information on how to use Optuna to suggest hyperparameters,\n",
    "please refer to their [documentation](https://optuna.org/)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def build_network(trial):\n",
    "    \n",
    "    # vary number of layers\n",
    "    n_layers = trial.suggest_int(\"n_layers\", 1, 5)\n",
    "    model = Sequential()\n",
    "    \n",
    "    for i in range(n_layers):\n",
    "        # vary size of each layer\n",
    "        num_hidden = trial.suggest_int(\"n_units_l{}\".format(i), 4, 512, log=True)\n",
    "        model.add(Dense(num_hidden, activation=\"relu\"))\n",
    "\n",
    "        # vary amount of dropout for each layer\n",
    "        dropout = trial.suggest_float(\"dropout_l{}\".format(i), 0.0, 0.5)\n",
    "        model.add(Dropout(rate=dropout))\n",
    "\n",
    "    # add output layer with size of target    \n",
    "    model.add(Dense(Y.shape[1], activation=\"linear\"))\n",
    "\n",
    "    return model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Define the tuning objective\n",
    "\n",
    "Optuna takes in an `objective` function to perform tuning over. An objective\n",
    "function should, assuming some set of hyperparameters generated by the\n",
    "current optuna trial, 1) build our model of interest, 2) run the model, and\n",
    "3) evaluate the model and return some metric of how well it did.\n",
    "\n",
    "Here, we define the parameters for a CFL Experiment that contains one Block - \n",
    "a CDE. Specifically, we use a 'CondExpDIY' CDE because 1) we want to estimate\n",
    "the conditional expectation of P(Y|X) instead of the full distribution for\n",
    "computational esse, and 2) we want to provide our own model (defined above) \n",
    "instead of using a pre-defined model that CFL supplies.\n",
    "\n",
    "Note that some of the parameters in `CDE_params` ('lr' and 'batch_size') are \n",
    "also defined by Optuna `trial.suggest_xxx` methods and will be optimized in\n",
    "the same way as the network architecutre parameters.\n",
    "\n",
    "Lastly, once the CDE has been trained, we pull the final validation loss\n",
    "from the packaged cfl results and return it as our evaluation of the current\n",
    "set of hyperparameters."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "def objective(trial):\n",
    "\n",
    "    # define the parameters for a CFL Experiment\n",
    "    data_info = {'X_dims': X.shape, 'Y_dims': Y.shape, 'Y_type': 'categorical'}\n",
    "\n",
    "    CDE_params = {\n",
    "        'model'        : 'CondExpDIY',\n",
    "        'model_params' : {\n",
    "            'optuna_callback' : TFKerasPruningCallback(trial, \"val_loss\"),\n",
    "            'optuna_trial'    : trial,\n",
    "            'build_network'     : build_network,\n",
    "\n",
    "            'batch_size'      : trial.suggest_int(\"batch_size\", 16, 128),\n",
    "            'n_epochs'        : 100,\n",
    "            'optimizer'       : 'adam',\n",
    "            'opt_config'      : {'lr' : trial.suggest_float(\"lr\", 1e-5, 1e-3, log=True)},\n",
    "            'loss'            : 'mean_squared_error',\n",
    "            'best'            : True, \n",
    "            'verbose'         : 0,\n",
    "            'early_stopping'  : True,\n",
    "        }\n",
    "    }\n",
    "\n",
    "    block_names = ['CondDensityEstimator']\n",
    "    block_params = [CDE_params]\n",
    "\n",
    "    my_exp = Experiment(X_train=X, \n",
    "                        Y_train=Y, \n",
    "                        in_sample_idx=in_sample_idx,\n",
    "                        out_sample_idx=out_sample_idx,\n",
    "                        data_info=data_info, \n",
    "                        block_names=block_names, \n",
    "                        block_params=block_params, \n",
    "                        verbose=0,\n",
    "                        results_path=None) # we don't want to save 100's of results\n",
    "\n",
    "    train_results = my_exp.train()     \n",
    "    score = train_results['CondDensityEstimator']['val_loss'][-1]\n",
    "    return score"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Run an Optuna study\n",
    "\n",
    "Finally, we are ready to optimize our hyperparameters. We first create\n",
    "a 'study' which will minimize the score returned by `objective` above, and\n",
    "will prune training during trials where the validation loss isn't dropping\n",
    "at a promising rate.\n",
    "\n",
    "We call `study.optimize` to try 4 different samples of hyperparameters from\n",
    "the space we described with the `trial.suggest_xxx` methods. In practice,\n",
    "you will want to try many more hyperparameter combinations (which you can\n",
    "set through the `n_trials` argument below).\n",
    "\n",
    "By setting `show_plot` to `True` in `CDE_params`, we can visualize how the \n",
    "loss varies across different hyperparameter combinations."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[32m[I 2022-04-01 15:43:06,847]\u001b[0m A new study created in memory with name: no-name-d8aa3c6d-62dc-47ae-91ac-aee562ace5b9\u001b[0m\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Block: verbose not specified in input, defaulting to 1\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a4c2a95f0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a4c2a95f0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0xf16b61ef0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0xf16b61ef0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_predict_function.<locals>.predict_function at 0x1a4e4b9440> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_predict_function.<locals>.predict_function at 0x1a4e4b9440> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[32m[I 2022-04-01 15:43:47,458]\u001b[0m Trial 0 finished with value: 0.3172536790370941 and parameters: {'batch_size': 23, 'lr': 0.00010775525177725577, 'n_layers': 5, 'n_units_l0': 11, 'dropout_l0': 0.3903148662273171, 'n_units_l1': 181, 'dropout_l1': 0.11695978734553025, 'n_units_l2': 195, 'dropout_l2': 0.44569404315026095, 'n_units_l3': 4, 'dropout_l3': 0.46439788790141745, 'n_units_l4': 92, 'dropout_l4': 0.08699934386147978}. Best is trial 0 with value: 0.3172536790370941.\u001b[0m\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Block: verbose not specified in input, defaulting to 1\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a4c9b3dd0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a4c9b3dd0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x1a4ca09320> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x1a4ca09320> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_predict_function.<locals>.predict_function at 0x1a4c4aecb0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_predict_function.<locals>.predict_function at 0x1a4c4aecb0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[32m[I 2022-04-01 15:44:29,006]\u001b[0m Trial 1 finished with value: 0.1596677303314209 and parameters: {'batch_size': 108, 'lr': 2.1381673250042515e-05, 'n_layers': 5, 'n_units_l0': 17, 'dropout_l0': 0.04523916844676085, 'n_units_l1': 222, 'dropout_l1': 0.14298986554926724, 'n_units_l2': 129, 'dropout_l2': 0.10457664459125571, 'n_units_l3': 180, 'dropout_l3': 0.2014939561454271, 'n_units_l4': 193, 'dropout_l4': 0.24951141600328575}. Best is trial 1 with value: 0.1596677303314209.\u001b[0m\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Block: verbose not specified in input, defaulting to 1\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x10b935560> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x10b935560> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x10a93e200> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x10a93e200> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_predict_function.<locals>.predict_function at 0x10a93e830> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_predict_function.<locals>.predict_function at 0x10a93e830> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[32m[I 2022-04-01 15:44:54,499]\u001b[0m Trial 2 finished with value: 0.1538044661283493 and parameters: {'batch_size': 44, 'lr': 0.00028887978890040927, 'n_layers': 3, 'n_units_l0': 266, 'dropout_l0': 0.27350843620215715, 'n_units_l1': 7, 'dropout_l1': 0.240795868909433, 'n_units_l2': 10, 'dropout_l2': 0.2586753115734922}. Best is trial 2 with value: 0.1538044661283493.\u001b[0m\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Block: verbose not specified in input, defaulting to 1\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a4f4558c0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a4f4558c0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x1a4f3a4950> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x1a4f3a4950> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_predict_function.<locals>.predict_function at 0x1a4d3e69e0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_predict_function.<locals>.predict_function at 0x1a4d3e69e0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[32m[I 2022-04-01 15:45:30,844]\u001b[0m Trial 3 finished with value: 0.2826230823993683 and parameters: {'batch_size': 24, 'lr': 9.12116430942231e-05, 'n_layers': 4, 'n_units_l0': 60, 'dropout_l0': 0.3345292372662158, 'n_units_l1': 142, 'dropout_l1': 0.37366260933900974, 'n_units_l2': 152, 'dropout_l2': 0.23042070465277703, 'n_units_l3': 385, 'dropout_l3': 0.2786944105090963}. Best is trial 2 with value: 0.1538044661283493.\u001b[0m\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Block: verbose not specified in input, defaulting to 1\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a4c97b440> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a4c97b440> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x1088197a0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x1088197a0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_predict_function.<locals>.predict_function at 0x1a4fdfcdd0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_predict_function.<locals>.predict_function at 0x1a4fdfcdd0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[32m[I 2022-04-01 15:45:49,520]\u001b[0m Trial 4 finished with value: 0.14809727668762207 and parameters: {'batch_size': 63, 'lr': 0.000354432971720819, 'n_layers': 3, 'n_units_l0': 93, 'dropout_l0': 0.022670093448453554, 'n_units_l1': 8, 'dropout_l1': 0.06405717592558907, 'n_units_l2': 42, 'dropout_l2': 0.19487686263996457}. Best is trial 4 with value: 0.14809727668762207.\u001b[0m\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Block: verbose not specified in input, defaulting to 1\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a4c3ff680> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a4c3ff680> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x108819cb0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x108819cb0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAbcAAAE0CAYAAACvs32dAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/d3fzzAAAACXBIWXMAAAsTAAALEwEAmpwYAABakklEQVR4nO3deXxU1fn48c+TnSwk7IQdRVYXNlERBbXuWrWoWC2KWq1btXWp+NNaW/Ur7jvWpYpWrVWs+9LiAoooCoooIMi+hjUhLNnz/P44d5JJmExmJpOZkDzv1+u+LjP33HvPzR3mmXPuWURVMcYYY5qThHhnwBhjjIk2C27GGGOaHQtuxhhjmh0LbsYYY5odC27GGGOaHQtuxhhjmh0LbnsxEeklItrApVcM8zimkc7hO/6Exji+qSYiY8L97NTaJ6KlkS+r0dT6/N8W7/y0JBbcjDGmFvvBtPdLincGTIOsArLq2HYE8L7378uAl+pItyvamTKmls+p+3N6HvB3798neWmNaTALbnsxdcPL7Ay0TUSK/F6WqGrAdI1NVVcC0sjnaNTjm4ZR1Qrq/pyW+L0sitfn1DQ/Vi1pjDGm2bHg1gKJyAT/B/Ui0l1EHhaRJSKy29uW45d+fxH5fyIyXUQ2ikiZiBSKyHwRuV9Eugc5V9AGJSKy0v9hu4j8UkT+JyKbRaRYRBaLyJ0i0jrIOep8PuLlWUVkivf6CBF5U0TyRKRERFaIyKMi0qmev5mIyEUi8qV37YUiMkdEfi8iibX/puESkTQROVFEnvD+roXe33mTiHwkIpeISEqQ/ad455/uvT5QRF4UkbXeda710uwbQl5OFZFpIrJNRHaJyI8icquIpEdybdEgIn29+7RQRHZ4n9PFIvKIiPQIsp+IyK9F5H0R2SAipSKyXUSWetf4J//9fZ8Xv0M8F6CBy5hGvNRA13CqiLwhIuu9e7nFy+cVIpIcZL+wrt1vvwxv2yzvM1DmnXORiLzmfRYzGveqo0BVbWmGCzAGUG+ZUGvbBL9thwLb/F77lhwv7UEBttVeCoFj68hHL790YwJsX+ltuw14MMg55gGZdZwj4HV626Z726YAfwQq6jj+KqBrHcdPBt4MkrdpwCW+1xHer2DX7lu+9N2XAPtP8dJMB8YBxXUcYxtwQJB83B/k/POBM/xe94rC59T/s7jH58NLcx1QFiRfu4BTAuyXWM998y3XB/i8BFsC5jOEz/9tYf5tUoFX68nL90CXaFy7t18usCSE/YY3xvdWNBcruZmpuOchFwDdgc7AqbgvR3Af5NnAjcCRQF+gHdAf1xhgDq6xwKsi0rkB+RgP/AF4BjjYO8cA4Elv+0HA/2vA8Y/EfXG/DYwC2gP7ALfjrrEHcG8d+/4fcJr373eBkd7+g7xjHgNMbEDewN2D13B/04Nx96ITMAz4G5CP+yHyRD3H6QM8j7tnxwEdvWP9ASgF2lD9N61BRH4LXOu9/AY4FugA7If78dEfd70xIyJXAvfh2ge8CxyP+4x2wF3f50A68JqIHFBr9wlU37cncH+/Lrgv8BG4z/z7uMDpcyI1G79c5r32X2LV6OVR4Czv328Bh+M+dwNxn9UK4EDgvQCl+gmEf+0A9+Du927gT8D+3jl74RqpXQvMxf2fadriHV1taZyF0EtuW4BuDThPEjDTO9ZfA2zv5XeuMQG2r/TbfnMd53jb276hju2hlNwUeKqO/R/xtpcArWtt6w6Ue9vfAiTA/jf6nUMb6X4e4OWjEtg3wPYpfnn4AEgKkOZavzT9a21L8z4LiislpwfY/3xq/nrvFYXr8v8sjqm1LZfqEuiTQT5/n3pp3q217XXv/TciyFedn6kwj+P/+b8tjP2G+O33Uh2fuyv80lwdjWsHtnr7XdMYn+NYLlZyM/eq6tpId1bVcuBl7+WxDcjHGmBSHdue89adJcjzvXr4fokGO34KroTo71xcFQ/Adep9A9RyHy7/jUZVfwC+xbU8Paae5Nd496W2KX7/PrjWtlNwpWWAiaq6O0AeXsCV1GPlMlzV3Fbg6kAJvOu81Xt5kvg9K6a6Nfi6xspgI7rYW5cCfwj0uVPVybhqSYDf1toc6bXvzX+zGiy4mfdCSSQiZ4jIqyKyzGtk4D96xONesn4NyMc0dU3GA1ns9+9Iqz6/UtWCCI4/0lsvVNWlgXb28h3S3zEYEWkrIjdKdcOd0lp/Z19ACvZ3Xq6qS+rI5zZgs/ey9nWO8ta7cM8Q6/Kfei4jmn7hrT8HkkUkM9AC/OSlE1w1rs933voiERkvIqkxync0+O7HdFXdHCTdVG+9f63AHum1+/a7S0RGi8he283G+rmZ5cE2ei3k/oN71lGf7AbkY32Qbf6liEhb7NV5fFXd7fd/uPbxe3nrxQT3Uz3bgxKRQ3HVrx1CSB7s7xzs7wjVf8u6rvPnID8yABbVc/xo8gXx04EdIe7j//d7EFft2RN4AXhCRL7AVaNPB2bVc63x1NNbL6wn3QJvLbjnxgXe60iv/U+4HxN9vHQbReQz771PVHVBgH2aJCu5tXCBqp9quZ/qwPY8rvqqD+4hs+8B+xXe9sQ99g5dqF8ykf6SjPT4vibP9Y3kEnHnY3HdHN7EfTFvBm4CDgO6AjlU/52/8HYJ9qM00uvM9Nb1XUcsO1lH8mMpzfcPVd2OazzxKK5BTgauEcrfgM+ANSJyjYg0xe9BX6OW+v7e/kG/qiFMpNeuql97+72BqxLthGvU8gjwo4jME5ETIr2oWLKSm6mTV2qb4L28W1UDtggUkbRA7zcTvqBWX7+ezHq2B3Mm7kukEjiqrl/HIlLXEFbR4PsSre86GnKd4dqJC+73qeoNkRxAVTcBV4vIH4HBuB8NRwEn4BqsPIQrtf6xwbmNrh24aw/nftQo3UZ67ar6PfAr7///Ibiq+WNxrSUPAt4XkbGq+kb4lxU7TfEXi2k6+lP9S/hfQdLVboLdnKzy1n3rSdeQ542DvfX8IIEtJYQ8NMRKb72fiAQrgQ9oxDzU5qsyH9LQA6lqharOVdXHVHUsrhXsTG/z70WkXZDd42Gltx5YT7pB3lqB1YESRHrtqrpbVT9V1TtVdQyu28FmXKn/LyFfSZxYcDPB+D+EDviF5z3QPz0muYkPX1XgIKljdA8vGJzcgHP4/s7BgspZ+FW5NQLfl10GwVu9/qoR81Db/7z1aBHpHc0De41rHvBeJrLnD4dyv23x4Lsfo0WkfZB0Z3rrH4M0mKohhGuva78FVP/IjeWPnIhYcDPBrPD792m1N3otqR7BdQxurl6m+jnWfXW0HrsW9zA/Ur4SygAR2eOLRkS6Anc34PiheBfX5B5gkgQYaktEzgeGN3I+/D2O63uYBPzT+yFVJxHpV+t1/3qO7/9jZWutbb7XXULIZ2P4h7dOpToQ1SAiv6O61P90rW1hX7s37FZ9XW18+9X+ezU5FtxMnVQ1D5jhvbxJRP4iIv1EpJ2IHAm8A1xI/S269lqqugZ42Ht5OvCmiBzqNdvvLyL34vrnBW11Wo/XcQE0CTfaxBkikisi3cSNl/kV7gfEqiDHaBBVLaZ6lJWDgE9F5Bfevd5XRG7FjR6zos6DRD9Pa3Ejq4AbnWOeiPxO3DiTOd7faKSIXCcis6luFu/zgYh85W0/TEQ6e/dtkIjcDNzhpfs2QPeJud76Am/fTBFJ8pZIGzV18z47wZYh3rXPozpgjReR12t97iZR3QVnHnuOOhPJtXcAlovIWyJyoYgcICLtRaSTd4znqK6hCPaYommIdy9yWxpnIcQRSkI4Tn+qR64ItLwCXFTX8QhjbMkgeajvGKGMUDKlnusMdowUXCCv62/wEXCp9+/yCO/XDUGOXwSMDXYt+I0tWc95gv69CT625A/EZ2zJS6l7rEz/ZW4d1xpsWQH0DXDO44PsEzCfIXx2Q1lW+u0b6tiSe4yJGsm1h5HXaUBGQ+99Yy9WcjNBqepPuI6xz+BGLSjDPVT+BBivqufgWvk1W6paiquWvQQ3ZuNOXMu073BVkidQ/dws1P5Ytc9xL25Mz09wA1GX4L6gngUOVtXXI7+CsPJxHfBL4GNcn6nduL5tt+PGJ8yPRT5q5ekp3DigdwJf4wZ/rsD9rRfiAvtpuNKdv+NxI5u8ibuGfNyztC245vDXAftrgE7vqvpf3H39ANhE9TO4mFHVElU9G3dtbwF5uP9/23D5vwr32Qg0mkgk174a9zf8C+7+L8O1Fi4F1uL6YY4DjlPVJj/JsXgR2xjTACLyMO7L5EdVbc6tR43ZK1jJzZgG8p7B+J5FzA2W1hgTGxbcjKmHiKTU01LvT1S3Ins1BlkyxtTDRigxpn5tge9F5EncHFhLcR1Z++FGb5/gpfsC+DAeGTTG1BT2MzcR+Q+Qp6pX1JvYmGZA3CSsG+pJtgA4vo6H+8aYGIskuJUAb6rquMbJUvPSvn177dWrV7yzYRpAVdm2bRvbt2+nqKiIsrIyKioqSEpKolWrVuTk5NC+fXsSEqyW35homTt37hZVDWWWjIAiqZZcByRHesKWplevXsyZE8v5HY0xZu8nIg0atCCSn5rvAkcEGp7HGGOMaQoiCW634TqZTg1hHDJjjDEm5iKplrwP9/D8FOBnEfkWN+ZdUYC0qqoXNyB/xhhjTNgiCW4TcOOLgRtz71BvCURxTaWNMcaYmIkkuF0Y9VwYY4wxURR2cFPV5xsjI8YYY0y0WMccY4wxzU6Dht8SkRTcdChdvbfW4eZUKm1oxlq6xXk7eOf79Vw0qjdtM1LinR1jjNmrRFRyE5FkEbkTN8/RTODf3jIT2Cwid4iIdfRugBVbdvLYp0vJ214c76wYY8xeJ+ySm4gk4jpy/wI3eOwGYLn3795ALnATcLCInKSqFdHLbsuRkepuza7SmM+RaOKsvLy8ariv8nK7/2bvlpSURHZ2Nm3btiUpKXZj9UdypkuBY4ElwB9UtcYo6CJyPPAQLvhdAvy9gXlskaqCW4l9ubUklZWVrFmzhtTUVHr06EFKSgpuujhj9j6qSmlpKVu3bmXNmjX07NkzZmOwRnKW83FTjx9TO7BB1fTsv8BNT39Bw7LXcmVWBTcr+LYk+fn5JCUlkZubS2pqqgU2s1cTEVJTU8nNzSUpKYn8/PyYnTuS4DYQ+DTY1B7etk+9tCYCVnJrmXbu3ElOTo4FNdOsiAg5OTns2rUrZueMJLgl40pl9dmNzR4QsYyURAB2WnBrUYqLi0lPtzHJTfOTnp5OUVGgURobRyTBbRVuVoA626d720Z5aU0ErOTWMlVWVtq8cKZZSkhIoLKyMnbni2Cft3EtIp8XkZzaG0UkG3jWS/NWg3LXgiUnJpCSlMBOay3Z4liVpGmOYv25jqS15D3Ar4GzgRNF5B1gBW6Q5H2AU4EsYK2X1kQoMzXJSm7GGBOBSMaW3CoiRwMvA8OB86ieJcAXmr8BzlXVbVHJZQuVkZporSWNMSYCEfWoU9WlwAgRGQWMxg2/JbjS2gxVnRm9LLZcGSlJ1qDEGGMiEPYzNxF5QERuBVDVmap6p6peoaqXe/+2wBYlVi1pTNMxYcIERIQpU6bE/Ny33XYbIsJtt90W83PvrSJpUPJ74KBoZ8TsKcOCmzF76NWrFyLCypUr450V04RFEtzyAPvGjYHMVKuWNKapuOuuu1i0aBFnnHFGvLNiQhDJM7ePgGNFJElV7Zu3EVmDEmOajtzcXHJzc+OdDROiSEpufwFaAX8XkYwo58f4sWpJY6pNmTIFEWHVKjc2RO/evRGRqmXlypVVaSZMmMDWrVu5+uqr6d27NykpKZx++ulVx3r99de56KKLGDRoEDk5OaSlpdGnTx+uvPJK1qxZE/D8dT1z838etnHjRn73u9/RrVs3UlNT6d27NxMnTqS4uHGnrnrvvfc48cQTad++PSkpKXTv3p0LLriARYsWBUy/fv16rrrqKvr06UNaWhrp6en06NGDE044gaeeemqP9K+88gpHH300bdu2JTk5mfbt23PAAQdw5ZVXsmzZska9tkhFUnKbAHwAXAicJiLTcCORBBpXRVX19siz17Jlpiaxq7QcVbWOvabF69OnDxdccAFTp05l165djB07lszMzKrt/v/esmULBx98MNu3b+eII45g+PDhtGvXrmr7uHHjSEtLY+DAgfziF7+gpKSEefPmMXnyZF599VW++OIL+vbtG1b+1qxZw7Bhw1BVRo4cSWFhITNnzuTuu+9m4cKFvP322w3/IwRw0003MWnSJBISEhg1ahRdu3Zl/vz5vPDCC7z66qtMnTqVk08+uSr9hg0bGDZsGHl5efTs2ZMTTjiB1NRU1q1bx1dffcXKlSu59NJLq9Lfdttt/PWvfyU5OZmRI0fSpUsXCgoKWLlyJZMnT+aII45g3333bZRra4hIgtttuH5tArQDzgmQxrddAQtuEcpITaJSoaisgvSU2M2DZJqmv76zgIXrC+OdjbAM7NKav5w6KCrHGjVqFKNGjWL69Ons2rWL++67j169egVM+95773HccccxdepUsrKy9tj+8ssvc8opp9QYx7O8vJy//vWv3HHHHVxzzTV88MEHYeXv2Wef5be//S2PP/44KSludMJFixYxYsQI3nnnHb744gsOP/zwsI5Zn/fff59JkyaRkZHB+++/z5FHHlm17d577+VPf/oT5513HkuWLKFjx44APP300+Tl5fG73/2OJ554osYP55KSEmbPnl3j9T333ENmZiZz587dI+D//PPPMZ2jLRyR5OpvVHfaNo3IN77kzpJyC27GhCE5OZknn3wyYGADOPvss/d4Lykpidtvv51nn32W//3vf+zYsaPO/QPp3r07jzzySFVgAxgwYADjx4/niSee4OOPP456cLv//vsBuOaaa2oENoAbbriB119/ndmzZ/P0009z8803A7Bx40YATjjhhD1qhFJTU2scp7CwkKKiIg466KCAJdn99tsvqtcTTZGMUHJbI+TDBJCZ6mYG2FVS4QY0My1atEpALcHQoUPrLNX5LFmyhA8//JClS5eyc+fOqkF9y8vLqaysZOnSpQwZMiTkcx599NG0atVqj/f79+8PuOdc0VReXs4XX3wBuOeBgVx44YXMnj2b6dOnVwW3ESNGMHnyZG688UYAjj32WDIyAjef6NChA7169eL777/nuuuu45JLLqm6nqYu7OAmIt8Cy1X1zEbIj/GTkWIzAxgTiZ49e9a5rby8nCuuuIJnnnkG1boroQoLw6sC7tGjR8D3W7duDRD1RiVbt26lpKSEhISEOq/X9yxs3brq6TfHjx/P//73P15++WXOOOMMEhMT2X///TnyyCM555xzGDlyZI1jvPDCC4wdO5YHHniABx54gA4dOnDooYdy/PHH85vf/Ibs7OyoXle0RNJasj9QFu2MmD1l+lVLGmNCF6gE5fPwww/z9NNPk5ubyyuvvMLq1aspLi5GVVFVDjvsMICggS+QWE9V5J+/uhqcBbqGhIQEXnrpJX744QfuuusuTjzxRFavXs2jjz7K4YcfzsUXX1wj/RFHHMHKlSv597//zeWXX06XLl149913q1pbfvfdd9G9sCiJdD63zHpTmQazOd2Mib7XXnsNgCeffJJx48bRvXt3UlNTq7YvXbo0XlkLS/v27UlNTaWysrLO0VpWrFgBQNeuXffYtv/++zNx4kTeeecdtmzZwjvvvENWVlbVM0d/6enpnH322UyePJl58+axbt06xo0bx5YtW7jyyiujfm3REElwex04UkTaRzszpqYMK7kZswdfg43y8sj+X2zb5iYr6d69+x7bpk2bxubNmyPPXAwlJSVVNVB54YUXAqbx9ckbM2ZM0GMlJCRwyimncNpppwHw/fffB02fm5vLnXfeGVLaeIkkuN0J/AT8V0QOiXJ+jJ/MqpKbjVJijI+vFFJXB+X6+BpEPPHEEzVmhl62bBmXXXZZwzMYQ9deey0ADz30UFXjEp8HHniAL7/8kuzsbH77299Wvf/CCy/w7bff7nGsrVu38uWXXwLVzyxXrVrFM888E/D54zvvvFMjbVMTSfvy94AK4GBglohsJHgn7mMakL8WLaOqtaSV3IzxOeOMM5g+fTrnnXcexx13HDk5OQDcfffdIe1/00038eGHH/Lkk0/y6aefMmTIELZt28aMGTM47LDD6Ny5M7NmzWrEK4iek08+mRtvvJG7776bI488kiOOOIIuXbrwww8/8OOPP5KWlsaLL75Ip06dqvb5z3/+wwUXXEDXrl0ZPHgwOTk5bN26lc8//5xdu3ZxxBFHVI2fmZ+fzyWXXMKVV17J4MGD6d27N5WVlSxcuJAFCxaQnJzMPfc0zTmpIwluY/z+LUBnbwnE+sM1gK+1pFVLGlPtqquuorCwkJdeeol3332XkpISAG655ZaQ9j/ssMP4+uuvueWWW5gzZw5vvfUWvXv35uabb+bGG2/k+OOPb8zsR92kSZMYNWoUjz32GN988w2zZs2iY8eOjB8/nokTJzJw4MAa6a+77jp69erFrFmzmDNnDvn5+bRv356hQ4cyYcIEzjvvPJKTkwHX2vLBBx9k+vTpLFiwgAULFpCQkEDXrl259NJLueaaa/Y4flMh4bYIEpHR4aRX1RlhnaCZGT58uM6ZMyfi/Qfe+iHnjujBLac0zQ+Qia5FixYxYMCAeGfDmEYRzudbROaq6vBIzxVJJ+4WHaxiLcMbX9IYY0zoYtsxw4TNzelmDUqMMSYcEQ9YKCIJwInAYUAHYLaqPutt6wC0AZapatjfzCJyLnA5cCCQiGud+RzwhKpWBtvX7xjJwJHAScDhQE/cQM+bgS+Bx1R1eh37TgEuCHL4xaoakzFo3JxuVnIzprmYNGkSP/30U0hpR40aVaOlowldRMFNRIYCrwD7Uj36fzLwrJfkl8BTwOnAO2Ee+3HgCqAY+Bg3GsoxwGPAMSJyVogBczQwzft3HjAX2AUMBMYCY0XkdlW9NcgxvgAC9ejcEMq1RENGis3GbUxz8uGHHzJjRuhPdyy4RSaSsSV74oJGG1y3gBlA7bagrwOPE2ZwE5GxuMCWBxypqj9773cCPgXOAK4CHg7hcJVePh5W1c9rnWcc8BLwZxH5VFU/reMYz6jqlFDz3xgyU5PIK2zciQ6NMbEzffr0eGehRYjkmdvNuMB2laqeqqr31U6gqgXAIlxfuHDc5K1v9AU273gbcdWUABO9KtGgVPUTVT2zdmDztv0bmOK9/E2YeYwpm43bGGPCF0lwOx5YpKqT60m3BsgN9aAi0g0YBpQCr9Xe7rXSXIfrU3doyLmtm2+0z25ROFajybAGJcYYE7ZInrl1Ar4KIV0x4c1C5ps4aYGqBhrtBOAboKuXtqFDCPhm2Qv2/OwoETkQN1D0RmAmMC3URi3RkGkNSowxJmyRBLcduABXn97AljCO29tbrwqSZnWttBERkc7ABO/l60GSnh/gvYUico6q/tCQPIQqIzWJorIKKiqVxITA01oYY4ypKZJqye+A4SJSZ5WjiPQDBgNfh3Fc3zQ6u4Kk2emtI56XWkSSgBeBbOBjVQ3U4GUecDUwyMtXF+AU4Htca8uPRGTPOSSqz3GpiMwRkTkNHWG8avBk68htjDEhiyS4PQukAy+JSLvaG0WkNa4bQALVXQNC4SuWNPZ4lH/HdS1YQx2NSVT1IVV9VFUXquouVd2gqu8BI3BVsh2pbvwSaP+nVHW4qg7v0KFDgzJrc7oZY0z4wg5uqvoK8AZuAOXlIvK2t+lQEfk3sAI4AnhVVd8N49A7vHWwiVB923YESVMnEXkYuBjX1eAYVc0LZ39VLQXu8l6eFEkewmXBzRhjwhfp8FvjcH3bknDVdQD9gbNwAeghYHyYx1zprYNNDuSbXXBlkDQBicj9uKrGzbjA9nM9u9TFN7RAndWS0ZTpTXtjLSaNMSZ0EY1QoqrluP5mdwNHAfvghslaA3ykqpsiOKyvaf4gEWlVR4vJg2ulDYmI3ANcC2wFjlXVhRHkz8dXFbszaKoo8U17YyU3Y4wJXcRjSwKoaj7wn1DTi8hpwEGq+rcAx1ojIt8CQ3ElwBdq7Tsa1yctDzc2ZKjnnATcAOTjAltD50Q/21t/08DjhMRXLWlDcBljTOhiPSvA6cBfgmz3Pc+6W0T6+N4UkY6Ar9P4JP9+ZiJyl4j8JCJ3UYuI3A7cCBTgAlu9JT4RGSwip4hIYq33k0TkWlzVJsCD9R0rGjLtmZsxMTdlyhREhAkTJkR8jOnTpyMijBkzJmr5MqFrUMkt2lR1qog8gRtq6wcR+YjqgZNbA2/iBlD2lwv0o9ZoKCLyS8A3Ne9S4PciAfuJ/aSqk/xe98I1mNkmIkuAtbiuBwfgugRU4oYH+29kVxkea1BijDHha1LBDUBVrxCRmcCVuJH9fVPePEsYU94Abf3+PdxbApkB+Ae373EDM4/ANW4ZguuesBY37c7jqjo3xDw0WGZVtaQ1KDHGmFA1ueAGoKovAy+HmHYC1aON+L8/herBkcM59wrgD+Hu11jSkhNIECu5GWNMOGwm7iZORLzBky24mZbtp59+QkTo2LEjZWVlAdNUVFTQuXNnRIQFCxYAMHv2bG644QaGDx9Op06dSElJoUuXLpx55pl89VUow+Q2jgULFnD++efTvXt3UlNTad++PSeddBIffPBBwPTFxcVMmjSJoUOHkpmZSWpqKrm5uRx22GHccsstFBfXnBrr66+/5qyzzqJr164kJyeTnZ1Nnz59OPfcc/nkk09icYlxZcFtL5Bp094YQ//+/TnkkEPYvHkz77//fsA0H374IRs3bmT48OEMGjQIgJtvvpkHH3yQsrIyRowYwS9/+UvatWvH66+/zqhRo3jttT0mIWl0b7/9NsOGDeOf//wn2dnZjB07loEDB/Lf//6Xk046iT//+c810ldWVnLyySdz0003sXz5ckaPHl21z5o1a7jzzjspKCioSj9t2jRGjRrF1KlT6dixI2eccQZHH300bdq0YerUqbz66qsxvuLYa5LVkqamjNQkG1vSwAcTIS8m43VHT+cD4MRJ9acL0YQJE5g9ezbPP/88p5122h7bn3/++ap0Ptdffz0vvfQSnTrVHO/9nXfeYezYsVx22WWcfPLJpKenRy2fweTl5TF+/HhKSkq4//77ufbaa6u2TZ8+nZNPPpk77riDUaNGcfzxxwMwc+ZMPvnkE4YOHcpnn31GRkZG1T6qyqxZs2jdunXVe3fddRdlZWW8/PLL/PrXv65x/q1bt7Jy5crGvcgmwEpuewGb080Y55xzziEtLY13332XLVtqTjqSn5/P22+/TUpKSo0v9BNOOGGPwAZw6qmnctZZZ7Ft2zY+/fTTRs+7z9NPP01hYSEjR46sEdgAxowZw1VXXQXAffdVzwO9ceNGAI444ogagQ3co4vDDz+8RnD2pT/xxBP3OH+7du0YNmxYdC6mCbOS217A5nQzQFRLQHurnJwcTj/9dF555RVefvllrr766qptr7zyCiUlJZx55pm0bdu2xn5btmzh3Xff5ccff6SgoIDycvf/6ccffwRgyZIlnHzyyTG5hhkzZgDU2Yfuoosu4p577mHmzJlUVFSQmJjI0KFDSUxM5B//+Ad9+/Zl7NixAQO2z4gRI1i4cCHnnnsuN998M4ceeiiJiYl1pm+OrOS2F8hIsWduxvj4goKvCtInUJUkwJNPPknPnj258MILuf/++/nHP/7B888/z/PPP8/8+fMBKCwsbPR8+6xbtw6A3r0DT0vZu3dvEhISKC4uZuvWrQDsu+++PPjgg5SWlnLllVfSuXNn9t13X8aPH8/UqVOpqKhZs3PXXXcxePBgPvjgA0aNGkV2djajR4/mr3/9K8uXL2/cC2wiYh3ctlI94agJUaa1ljSmyrHHHku3bt349ttv+eEH9wxy8eLFzJ49m86dO3PCCSdUpZ0zZw6XX345ZWVl3Hvvvfz000/s3LmTyspKVJWbbnIzV6k29kxb1XznqmNQiTr9/ve/Z9WqVTzxxBOcd955VFRU8OKLL3LWWWcxfPjwGgG6c+fOzJ07l48//piJEycydOhQZs+ezW233Ua/fv149tlwZiPbO8U0uKnq9araoFm0W6IMay1pTJWEhATGj3eTjkyZMqXG+je/+U2N6repU6eiqlx99dVcf/319OvXj4yMjKrAsnTp0pjmHaBbt24AdZagVq5cSWVlJWlpaXtUr3bu3JnLLruMF198kZUrVzJv3jwOOOAA5s2bx6RJNautExISOProo7nrrrv47LPP2Lp1K5MmTaK8vJwrr7wypqXVeKg3uInI+Q1ZYnERzZ0LbtagxBgfX9XjSy+9RGlpKS+++GKN9322bdsGQPfu3alt8+bNTJs2rVHzGcjo0aMBeOGFFwJuf+655wAYNWoUSUnBm0UcdNBBXHPNNQB8/33wMeEzMjK48cYb6datG8XFxSxevDjcrO9VQmlQMoXIZscWb7/Ad9CELDM1kdKKSkrLK0lJssekxvTt25eRI0cya9YsbrjhBtauXVujb5tP//79ARdILr74YjIz3XzHO3bs4KKLLqrRNyxWLrnkEu69915mzpzJI488UqNRzGeffcajjz4KwHXXXVf1/ieffEJxcTHHHXdcjYBXUVFR1eevZ8/qqTDvu+8+xo0bt0dQnzNnDhs2bCAhIaGqBNlchRLcXmDP4NYG+KX3/nyqJw/tBRzo/ftt3DQzpoF8gyfvLi0nJSklzrkxpmmYMGECs2bN4pFHHql6XduFF17IQw89xLfffss+++zDqFGjUFU+++wzUlJSuOiii2L+/Klz587885//ZNy4cVxzzTU888wz7L///qxfv57PP/+cyspKbrnllhrPDufPn88f//hHsrOzGTp0KLm5uezevZvZs2ezYcMGOnfuzI033liV/o477uCGG25gwIABDBgwgNTUVNasWcOsWbOorKxk4sSJ5ObmBspes1FvcPPGbqwiIm2B2cAs4HJV/aHW9v1x09MMAg6JWk5bMP853XLSLbgZA1QFh6Kioj36tvm0adOGOXPm8Oc//5lp06bx3nvv0bFjR371q1/xt7/9jSeffDIOOYfTTjuNOXPmcPfdd/PJJ58wdepUsrKyOO644/j973/PSSedVCP9qaeeSkFBAZ999hlLly5l1qxZZGZm0qNHDy677DIuv/xyOnToUJX+8ccfZ9q0acyZM4dPP/2UoqIicnNzOfXUU7niiis47rjjYn3JMSfhthISkceBccA+qhrwiaSIZAPLgFdV9YoG53IvNnz4cJ0zZ06DjvH+Dxu44qVv+e8fjqRf56wo5cw0RYsWLWLAgAHxzoYxjSKcz7eIzFXVumZzqVckD3BOBT6tK7ABqOp24FPglEgzZqrZbNzGGBOeSIJbR9wca/VJBDrUm8rUKzPV/bmtO4AxxoQmkuG31gJHiUg7Vd0aKIGItAeOBtY3JHPGSU+x2biNiac333yTN998M6S07du3rzEupImPSILbv4GbgI9E5GpV/dx/o4iMws1knQU83vAsmkyrljQmrubNm7fHcF916dmzpwW3JiCSask7gTnAQcB0EVktIjO8ZRUwAxgCfOulNQ3ke+ZmJTdj4uO2225DVUNaWsJ0MnuDsIObqu4GxgAPAruAbsAR3tId2I0ruY3x0poGyvA9cyu1UUqMMSYUEU154wWt60TkZmAYLsABrAPmqmpRlPJngNSkRJITxaoljTEmRA2az01Vi4EvopQXE4QNnmyMMaFr8GSlItIH1+R/q6ouaXiWTCAZKTbtTUuhqmFPh2JMUxfLaYUgwilvRCRJRG4VkY3AYmAmMNFv+wQRmeUNxWWiINNKbi1CYmIiZWVl8c6GMVFXVlYW09nAww5uIpIEvA/8BcgBFuFmAPA3BzgUGNvA/BlPRmqiTXvTAmRlZTX7ebZMy1RYWEhWVuyGD4yk5HYV8AvgY6CXqu5ROlPVH3EzBTT/0TljJMNm424R2rZtS35+Plu2bKG0tDTmVTnGRJOqUlpaypYtW8jPz99j8tXGFMkzt/HAVuBsVS0Ikm4FsF8kmTJ7ykxNIm97cbyzYRpZamoqPXr0YNu2baxcuZKKCiutm71bYmIiWVlZ9OjRg9TU1JidN5Lg1g+YXk9gA9gIjIzg+CYAay3ZcqSmppKbm9vs59sypjFFUi2pQGUI6ToDVtSIkkyrljTGmJBFEtxWAAeJSJ37ikgr3IzciyLNmKkpIzWRXaUV9gzGGGNCEElwexs3Isn1QdLcCLQB3ookU2ZPGalJVFQqJeWhFJqNMaZliyS4PQDkAXeJyMsi8ivv/fYicqKIPAv8GVgNTI5SPls8mxnAGGNCF3aDElXdJiIn4Epl5wDjcM/hTvYWAdYAp6rqjijmtUXL8JvTrX1m7FocGWPM3ijSgZN/EJGBwIXAicA+uJm31wAfAE+p6q6o5dJUTXtjJTdjjKlf2MFNRH4JlKnqB8AT3mIaWWbVnG7W78kYY+oTyTO3N4A/RDkfph5Vc7pZyc0YY+oVSXDbBmyJdkZMcNagxBhjQhdJcPsasNH+YywjtbpBiTHGmOAiCW53A4NE5OJoZ8bUzRqUGGNM6CKdrPTvwFMicibuGdwqoChQQlX9LMJzGD8ZKb5nbtagxBhj6hNJcJuO69cmwPEEn9ZGIzyHqSUpMYG05AR2lVrJzRhj6hNJ4PkMF7RMjNngycYYE5pIRigZ0wj5MCGwaW+MMSY0kTQoMXGSkWLBzRhjQmHBbS9i1ZLGGBOaBjX2EJEMoA/QGtfAZA/WWjJ6MlIT2bKzNN7ZMMaYJi+i4CYifYCHcS0lg5X+rLVkFGWkJrFq6+54Z8MYY5q8SAZO7gbMAtoD671jdAS+xJXiOuCC2pdAWdRyaqxa0hhjQhTJM7eJuMB2u6p2w01xo6p6uKp2wvV9WwGUErwPnAmTtZY0xpjQRBLcjsfN2/bXQBtVdZqXZiTwp8izZmrLSE1iV2kFlZXWzdAYY4KJJLh1A+apaqX3uhJARJJ9CVR1GTAD+HWDc2iqZHrT3uwusyG4jDEmmEiCWzFQ4vd6p7fuWCvdNqB3JJkygdnMAMYYE5pIgts6oIff66Xe+jDfGyIiwBBge+RZM7XZnG7GGBOaSJrpfw2cKSJpqloMfOi9/6CI7ALWApcD+wHvRSebBtwIJWAlN2OMqU8kJbf3gFbAKQCq+jPwD6Ar8C4wD7gM1w3glqjk0gA2p5sxxoQq7OCmqq+rarKqTvV7+3LgBlypbinwDjBGVedHkikROVdEPheR7SKyU0TmiMiVIhJyfkUkWUSOEZH7ReQrEdkgIqUisk5EporImFjkI5oyq565WYMSY4wJJiqjh6hqBXC/tzSIiDwOXIFruPIxrgR4DPAYcIyInOWdrz6jgWnev/OAucAuYCAwFhgrIrer6q2NnI+oyUj1TVhqJTdjjAmmSQ2cLCJjcQElDzhQVU9R1TNwz+8WAWcAV4V4uErgdeBIVc31jjVOVQ8AzgEqgD+LyFGNnI+oyUxzv0W2F9nAL8YYE0yTCm7ATd76Ru9ZHgCquhFX9QkwMZRqQVX9RFXPVNXPA2z7NzDFe/mbxsxHNHXITKVtRgrz11ojVGOMCSaSsSU/CSO5quoxIR63GzAMN2zXawEONENE1uEarhyKG9+yIb7z1t3inI+QiQgjerXl65VbY3VKY4zZK0XyzG1MCGkUNwVOOONEDfHWC1S1qI403+CCyhAaHlT289Yb4pyPsIzo3ZYPF+SxvqCILjmtYnlqY4zZa0QS3PZ4RuVJAHoCJ+MabNxNdR+4UPhGM1kVJM3qWmkjIiKdgQney9fjlY9IjOjdFoBvVm7jtMFdY316Y4zZK4Qd3FR1Rj1JpojIFcADwNR60vrL9Na7gqTxDfWVFcZxaxCRJOBFIBv4WFXfiXY+RORS4FKAHj16BEoSsQG5rclKTWL2CgtuxhhTl0ZpEKGqk4GVwG1h7Oabybuxh7z/O65J/xoCNyZpcD5U9SlVHa6qwzt06BDpYQJKTBCG92rD1yu2RfW4xhjTnDRma78fcNPehGqHt84Mksa3bUeQNHUSkYeBi3FN/I9R1bx45KOhRvRux9JNO9mys6T+xMYY0wI1ZnDrjBumK1QrvXXPIGm610obMhG5H7ga2IwLbD/XkbRR8xENvuduc1Za6c0YYwJplOAmIufgSm0/hbGbr2n+IBGpKygeXCttqPm5B7gW2Aocq6oL45GPaDmgazZpyQnMtqpJY4wJKJJ+bs8G2ZwJ9AcGea8fCfW4qrpGRL4FhgJnAS/UOu9oXJ+0PODLMPI7CTfuZT4usH0fj3xEU0pSAkO623M3Y4ypSyRdASaEkGYH8DdVnRLmse/CdZy+W0RmqepSABHpCEz20kzymwUcEbkLNxzWG6p6k//BROR24EagABfYQi1phZ2PWBvRuy2PfPIzhcVltE5Lrn8HY4xpQSIJbhcG2VaKm8z0myAdoOukqlNF5AncEFc/iMhHVA9Y3Bp4Ezdwsb9coJ+3riIiv6R6yp2lwO/dHKp7+ElVJ0UhHzF1SO+2qMLcVfkc1a/2JOjGGNOyRdLP7fnGyIjf8a8QkZnAlbiR/RNxz+6eBZ4Io7TU1u/fw70lkBnApNpvRjEfjWJIjzYkJQhfr9hmwc0YY2oR1cbuVtayDR8+XOfMmdMox/7V5C8QEV6/PJweF8YY0/SJyFxVratQUq+mNiuACcOI3u2Yv7aAolKbvNQYY/w19qwAtYU8S4Cp3yG92/L3Gcv4bk0+I/dtH+/sGGNMk9GQWQF89Zm1W2nU9b7/NhMFw3q1QQS+XrHNgpsxxviJdFaA04FrgLm4QYhXett64cZrHAY8jGtVaBpJ67RkBua2tv5uxhhTSyTBTYGrgOtV9YEA2x8WkT8C9wBvhjCLgGmAEb3b8q+vV1NaXklKkj1CNcYYiKxByS3AwjoCGwCq+iCwELg50oyZ0BzSuy3FZZX8sG57vLNijDFNRiTBbThuxP/6/ED1GIymkQzv5brzWdWkMcZUiyS4pRB8xHyfnkRW7WnC0D4zlX6dsvjgxw1Yn0VjjHEiCW7fAyNF5KS6EojIibhZAYIOUmyiY8LhvZi/djszl26Jd1aMMaZJiCS43Ytr5v+GiPxDRI4Wkd7ecpSIPEN1K8n7opVRU7dfDe1K59ZpPP7p0nhnxRhjmoSwg5uqvglM9PadAEzDDUy8FPgIuAg3DuPNXlrTyFKTErnkyH34avk25q6yZ2/GGBNR23FVvQfXWGQKsBw3G0ApsAJ4DhhRe6R907h+PaI7bTNSeOwTK70ZY0zEDT5UdR5wcfSyYhoiPSWJiw7vxX3/W8KC9dsZ1CU73lkyxpi4sV6/zcj4w3qRlZrE5E+XxTsrxhgTV1ELbiKSICK/FZFHReR6EcmK1rFNaLJbJTP+sJ68/+MGlm3eGe/sGGNM3IQd3ERkoojsFpExtTa9BzyJm9zzbuBLEclocA5NWC4a1ZvUpASemG6lN2NMyxVJye14oBA3gzUAInKc9/464A7ga2AAruWkiaH2mamcc3AP3vxuHWvzd8c7O8YYExeRBLc+uLEl/YfDGIsbUPkcVb0VOBrIB85teBZNuC49ch9E4KnPlsc7K8YYExeRBLd2wIZa740C8lR1FoCqFgGzcFPgmBjrktOKM4d14+XZq63fmzGmRYokuClQ9SxNRLKB/sAXtdJtB3IizplpkIknDqBLTiuuevk7tu0qjXd2jDEmpiIJbiuAQ0TEt+8puOG4ZtZK1wGwwQ7jJLtVMpPPG8rWnaX84d/zqKy0QZWNMS1HJMHtbaATbmzJq3FjTVYAb/kSiIgAQ3CB0MTJ/l2zufXUgXy2ZDOTp9vIJcaYliOS4HY3sAg4FXgI6Azcp6qr/NKMwpXcapfmTIydd0gPThvchQemLWHWMitIG2NahkgGTt6Om7D0AuBPwBhVvalWsnbAw8ArDc6haRAR4f/OOIDe7TO4+l/z2LSjON5ZMsaYRhfpwMlFqvpPVb1PVT8LsP1NVf2jqs73f19EThORWyPNrIlMRmoSk88bxs6SMq7+13eUllfGO0vGGNOoYj225OnAX2J8TgP065zF/51xAF8t38ZlL86luKwi3lkyxphGYwMntyC/GtqNO8/Yn08Xb+KCZ79mR3FZvLNkjDGNwoJbC3PeIT15aNxg5q7K57xnZpNvfeCMMc2QBbcW6LTBXXly/DB+ytvB2U9+ycZCa2RijGleLLi1UMcM6MTzF45gfUERZ/59Ft+tzo93lowxJmosuLVgh+3bjpcvOZSi0grOmDyL8f+YzdcrbCxKY8zez4JbC3dQ9xym33AUE0/sz6INhZz95JeMe/JLvli6hZoTPxhjzN7DgltTpQrr5sbkVJmpSVw2el8+/9PR/PmUgazYsovznpnNta9+b2NSGmP2Shbcmqrv/glPHw15P8bslK1SErl4VG8++9NRXDFmX974bh0PfrQkZuc3xphoseDWVPU7CRKSYP6/Y37qtOREbji+H+OGd+fRT5by+ty1Mc+DMcY0RKyD21ZgdYzPuXfKaA99fgE/TIXK2I8mIiLcfvr+jNy3HRP/M5/Zy7fGPA/GGBOpmAY3Vb1eVXvH8px7tQPHwY71sDI+kyukJCXwxHnD6NE2nd+9OJcVW3bFJR/GGBOupEh3FJGuwFFAFyCtjmSqqrdHeo4Wr9+JkJLlqib3GR2XLGSnJ/PshIM5Y/IsLpryDf+5fCRtMlLikhdjjAmVhNvc25uI9CHgCqpLflIrmXrvqaomNjCPe7Xhw4frnDlzIj/Am1fCwrfg+iWQkh69jIVpzsptnPv0bHq2S+f0IV0Z068DA3Nb4z4OxhgTXSIyV1WHR7p/JCW3G4DfA5XAh8BPQGGkGTD1OPBsmPciLPkA9h8bt2wM79WWyecN5cGPlnDvfxdz738X0zErldF9OzCmX0dG7dee7FbJccufMcb4i6TktgjYBzhGVW2m7Xo0uORWWQEP7g+5B8K5sW85GcimwmJmLNnM9MWb+eznzewoLicxQRjSPYfRfTswul8H9u+STUKCleqMMZFpaMktkuBWDHyhqsdEetKWpMHBDeB/f4avJsN1i10ryiakvKKS79YU8NmSzcxYspn5a7cD0DYjhZMPyOWs4d04oGu2VV8aY8ISj+CWB3yiqudGetKWJCrBbeMCeGIknHQfjLgkOhlrJFt2ljDz5y18tGgj0xZupKS8kn6dsjhreDdOH9KV9pmp8c6iMWYvEI/g9jJwsKruF+lJW5KoBDeAJw6H5Fbw248afqwY2V5Uxrvz1/PanLXMW1NAUoJwzoju3HLyQNKSW3Q7I2NMPRoa3CLp5/ZnoIOI/DnSk5oIHHg2rP0Gti6Ld05Clt0qmfMO6cmbVx7OtD8eya9H9ODFr1Zz+uNfsHTTznhnzxjTjEVScjsfGAJcDXwNfIAbdaQyUHpVfaGBedyrRa3ktn0dPDgIxkx0y15q+uJNXPvq9xSXVXDH6fvzq6Hd4p0lY0wTFI9qyUqq+7Hh/btO1s8tSsEN4PlToWANXP0d7MUNNPK2F3P1K9/x9YptnDWsG387bX9SkxIoKa+kuKyCorIKKlXp1DqN5EQb/tSYlige/dxeoJ6AZhrJgePgrSth9VfQ87B45yZinbPTePm3h/Dwxz/z2KdLeeO7dZQHmFonQSA3uxVd27SiW5tW9GybwTkjutOpdV0D4hhjjBN2yc2EJ6olt+JCeHQoJKbCRR9CTvfoHDeOvly2lelLNpGalEir5ETSkhNIS05EgPUFRazNL2JN/m7W5heRV1hMTqtk7jvrII4Z0CneWTfGNKKYV0ua8EQ1uAFsmA9TTnH93S76EDI7Ru/YTdyyzTu56uXvWLShkIsO782NJ/YjNalF13ob02zFo7WkiafcA+G812DHBvjnGVCUH+8cxcy+HTJ544qRTBjZi2e/WMHYJ2YFnKlAVbEfbca0bA0quYnIAKAv0Jo9B08GrLVk1EtuPss+hZfPhtyDYPybkJoZ/XM0Yf9bkMefXp9PWXklI/u0Z/vuMvJ3l5K/u4ztRaVkpCbxiwGdOGFQZ0bt19761Rmzl4lLtaSIjASeAgYES4bNCtB4wQ1g0bvw6vnQ63A49zVIblkNLdYXFHHLmz+yNn83OekptElPpk16Cm0yUsjbXsxHizayo7icjJRExvTvyPGDOjOmXwdap9kAz8Y0dfHoCtAfmAOkA7OAzkBv4BWgD64PXCLwFrBdVS+MNHPNQaMGN4DvX4E3fgd9T4Cz/wlJNteaT2l5JV8t38oHP+YxbWEeW3aWkpwoHLpPO34xoBPHDOhItzbxm0bIGFO3eAS3KcD5wO9U9WkReQ4431dC86oqn8cFv8NUdUfYmRI5F7gcOBAXKH8CngOeUNWAncXrOE4/4ATgYGA4rgpVgLNUdWo913hBkEMvVtX+oeSh0YMbwJxn4d0/Qv9T4KwpkGglk9oqKpVvV+fz0aKNfLRwI8s2u2d1/TtnsX/XbHq3z2Cf9hn07pBBr3YZVo1pTJzFo5/bGOBnVX060EZVXSQipwBLcUN1/Smcg4vI47iJUIuBj4Ey4BjgMeAYETlLVStCPNzlwDXhnL+WL3DXUduGBhwz+oZfBBXl8MEN8PrFMPZZSIx4kvVmKTFBOLhXWw7u1ZabThzA8s07+XjRJqYv2cRnSzYzde7aqrQi0K9TFqP7dWB03w4M79mWlCRre2XM3iSSb8DOwHt+rysARCRVVUsAVHWTiMwAziCM4CYiY3GBLQ84UlV/9t7vBHzqHe8q4OEQD/kjcC+uGnUu8A9gdKj5AZ5R1SlhpI+fQy6FynL4702QcCmc8ZQFuCD26ZDJPh0yueTIfQDYWVLOyi27WLFlF8s272T28m08O3MFT85YTkZKIoft256De7WhS04ruuSkkZvdio5ZqSTZCCrGNEmRfPvtpGbLSN8s3LnASr/3i4CuYR77Jm99oy+wAajqRhG5HJgOTBSRR0OpnlTVZ/xfN/s5xQ67wgW4aX8GSYQz/g4JVr0WiszUJPbvms3+XbOr3ttZUs6Xy7YyY8kmpi/ezEeLNtbYJ0GgW5t0TjkwlzOHdWOfDi2rxaoxTVkkwW0t4D80xk/e+ijcczFEJBk4BNgc6kFFpBswDCgFXqu9XVVniMg6XMA8FNeYxdR2+NVQWQYf/w0qSuDY26FNz3jnaq+UmZrEsQM7cezATqgqhcXl5G0vZv32IvK2F7OhoIgf1m3nyc+WM3n6Mob1bMNZw7px8oG5ZFmLTGPiKpLg9gVwoYi0VtVCXBVlBfCgiKThgt8lQDdcC8pQDfHWC1S1qI403+CC2xBiE9yOEpEDgUxgIzATmBZOo5a4OOI6kAT45E7XXeCAM2HUH6FjsJ4bJhgRIbtVMtmtkunXOavGtk2Fxbzx3Tpem7uWif/5gVvfXkCb9GSSExNISUogxVtnpibRJiOFtl53hbbpyWSmJZOUICQlCkkJCSQlCG0zUxjSPaf51zQY04giCW7/AY7DNSx5W1XXichduMYjj3lpBCgAbg7juL299aogaVbXStvYzg/w3kIROUdVf4hRHiIz6o9uoOUvH4c5z8H8f0O/k+Hwa6DbwZBgz4qipWPrNH43el8uPXIfvl+7nQ9+2EDB7jLKKiopqaikrLyS0opKdhSXs2h9Ifm7SykoKiNYQ+URvdtyy8kDOLBbTsyuw5jmJOzgpqofA/vVeu8vIjIfOBNoi6uqfEhVV4ZxaN8Diz3HU6rmm+EyK0iaaJiHa4DyMS7YtgaGAncCBwEfichQVV0XaGcRuRS4FKBHjx6NnNUgWneB4+90JbnZT8Lsv8Pi9yAtB3qOrF46HwTF22HjD27syrwfYNNC6H4IHP9/La5zeKREhMHdcxjcPafetBWVSmFRGTtLyimrqKSiUimrUMorK/l+TQEPffQzv3zsC04f3IXrj+9X1R+volKZv7aA6Ys3M3PpFipV6dE2ne5t0t26bTr7dsygY5bdM9OyNZmBk0XkZuAO4EVVHV9HmjuB/wc8paq/i+Ac03GtJYP2cwuyfwowA/fM73FVvaq+fWLSzy1UJTvhp3dh5UxYNQu2ebN6J6ZARWl1utZdoe0+sPJz6DIUxr0I2eG2DTINsaO4jL/PWMYzn69AgXNH9CB/dymfLdlM/u4yROCgbjlkpCayettu1hcUU+E3bVCX7DQO8gLt4O457NMhky07S9iwvYj1BcVs2F7E1p2l9OmYybCebRjUJdu6O5gmJR793BqLr7N3sCZnvm1hdwyPBlUt9apg3wJOikceGiQ1Ew46xy0AO/Jg1Rew7lvI7ASdD4DOB0JGO7d90btu9JOnRrvRT/biOeT2NllpydxwfH/OO6Qn9/13MVNmraRdRgpH9evI6H4dOHK/DrTJqB6Npqyikg0FxazetpvFG3cwb00B89bk88GPeQGPn5jgniG+8s0aAFKTEhjcPYfhvdrQt1MW3dqk061NKzpkppKQYM/+zN4n4pKbiGQDvwEOAzoAH6vqPd62fkBP4PMgjUNqH++XuKDxnaoOrSPNf3B93X6vqo8FSlPPOabTgJKbd4y+wGKgVFVT60vfpEpukdj0E7zyayhYDSfeAwdfHO8ctUjbd5eRlZYUdqDZurOE79cWsGrrbjpmpdE5O40uOWl0yHR99DYVFjNnVT7frNzG3FX5LFhfWKMEmJKYQG5OGn07ZfGLAR05un8nOmTV+7E3psHiUnITkROAl4AcvAGSAf/nT0O87ecC/w7xsN9560Ei0qqOoHhwrbTx4BVrqp7/NW8d+8Mln8Lrv4X3rnVVlUN+A72OrHscy8INsPYbaNMLOg2yvnZRkJ0eWdeCdpmpHN2/7oldO7ZO46QDcjnpgFwAikorWJO/m3X5RawtKGJdfhHrCor4dlU+0xZuROQHBnfP4RcDOnHcwE7s16mxH38bE5mwg5uI7I9rMZkETAY+Y88A9jauE/dpAbYFpKprRORbXMONs4AaU+WIyGhc94I84Mtw8x1FZ3vrb+KYh9hqlQPn/humT3KtLxe8AanZ0Pd4GHAKdBsB6+bA8hmwYgZsWVK9b1o29DgMeh7uZi/ofJCNnNKEtUpJpG+nLPrWClqqyk95O/ho4UY+WrSRe/+7mHv/u5gBua351ZCunDa4Cx1bWyMW03REMnDyy8A44AxVfdt7rxKYoqoX+aWbA7RS1UFhHPtMXAfuPOAIVV3qvd8RN/zWQOAPqvqw3z534aoq31DVm/Y8ao3jT6eeakkRGYwLoh/4j2EpIknA1bjhvBKAE1T1v/Vd015fLVlbWTEsnw6L3oHF70PRtuptyemu9WXv0dDjUNi2AlbNhJVfVDdeScuGfY6CPr+APse4Fp2BqLpBHk2TtLGwmA9+2MAb363j+7XbSRA4vE97Tj2oC+kpiRQWlVNYXMaO4jIKi8rZXlS2x1JWXklKUkJ1f8CkBNpmpHDa4C788qAu1hG+hYvHrADrgfX+J60juL0BjFbVtmEefzJuwONi4COqB05uDbwJnFkr6EzBjeD/vKpOqHWsobjSpc9AXDeCn4Gqb2VVPdRvn9OBN7ztS3Cd0rOAA4AuQCVwk+/5Yn2aXXDzV1EOq2fBhu9dq8puB9ddVbkjz7XSXP4pLP3YzSQO0HGg61xeVAC7t7pguTsfyotctWa7/aD9ftC+L7TtDcWFsH0tbF/jrde6WRBad/GWrm6dlu2GIqsod+vKMhd8e4+O7rRAFoRZtnknb363jje+W8fa/JpPExIThNZpSbRulUxOq2Raex3hs1slk5KUQGl5JWUVlZR6fQGXbdrF4o07SE9J5NQDu/DrQ3pwULds69DeAsUjuJXgSknn+L0XKLi9Dpykqq3CzpSb8uZKXEDxTXnzLAGmvKknuI3BlfiCUtWq/zki0hs3k8AIXKOYdrhnimuBz3FdAOaGei3NOrhFShU2LYKlH8HSaVCwBtLbQqu2bp3ezgWsbStg61K3+HdVAEhMhexurotCRTkUrnMBs3a62rK6uEGmh02AVm0iv4YN8+Hjv7qq2APPhsP/AB36Bk5buB5+fB1KdnjX2A7S27h1dnfIaB95PpqQykplyaYdJIjQOi2Z1q2SaJWcGFZgUlW+X7udf81ezTvz17O7tIJ+nbI4oFs2XXNa0bVNK7p569SkRBcYK1yALK9QdpaUU7C7jAJvVvaColJQGN2vAyN6tbWBrvci8QhuecBSVR3l916g4DYPyFHVXpFmrjmw4BYFlRWutWb+CtcB3RcQan9pqrrSX+E6V8JLTIaEJLckJrtjfPWEey6YnAFDx8Mhl7kSYai2LXfDmv041eVlv+NcFW15MQz8peswn3uQq75d/D7MewmWfQJaSXXbq1pad4MugyF3sFt3GVrdHaMF21Fcxtvfr+fteetZtXU3G3cUBx3VJZAUL5iVVlTSJj2ZYwd24oT9OzNy3/ZRnbMvb3sx0xdv4si+HeiSE/bv+b1P6W7Imw/r5rpl82L3Y7PjgOramHb7NWgAiHgEt7eB44H9/aakqRHcRORgYDbwL1U9L9LMNQcW3JqgDfPhq8nww1TQCldVeeA41zgmNUDrv8pK2PozfP00zH0OEpLdDAwjr3aNbXZuhtlPuO0lhdD9UNj8ExQXuGrSg34Ng8911axFBV7V61a3bFsBG+bB+u9cCRUAcaPDDDjV5alNr7qvpbjQfbFsXuS6bWxe5N73PdfsOCDyatOifPjpfVj4lgveg89zATw5Pl/epeWV5G0vZm2Ba81ZVqHeMzshJdE9u2uVkkhOejJt0lPISU+mVXIiRWUVfLZkMx/8mMcnizaxo6ScjJREDuiWzf5dsr3ZIFrTu30miQlCcVkFm3eUsHVXKQXbtpAupQzYrw9Zrfaszv5+TQH/mLmC93/YQHmly8+FI3txxZg+ZKdUulqJ5FaQlVv/kHdNrYq7qMA9K9++zv1g9D0G2LrMjWDkezrUuptrVV243jUmqyx370uC+7F39C0RnT4ewe144APgB+BsVV3sH9xEZB9cf7WBuGduMyPNXHNgwa0JK1xfPe5mwSpIagX9T3bVjInJsOYbWPs1rJ3jApUkwrALYPSNkNV5z+MVb4dvnoF5/3KltyHnucAZaleI4u0u8K6aBT+944ZBA9e5vu+J7t8782DHRm+dBzv9puFJauWqRstLXHAFVw3b5xjXWjUlwyvNJrsWqwnJbnSaxCRvneK+kFZ/CQvedA2HKssgu4e7Bl/JefC5rlq3Qz9XQt34owvO679z1c2ZHaFdH/es1PfMNCHJ/Q2LCqrXWumO1yrHPSNNy3Hn37LYfXlu+skdr2A17DsGhl0IXQN0gS0rdgF47nMubU5P94PAt+R0h/T2kNGekuQsvlyez8eLNvHDuu0s2lBISXklCVTSPrmU/WQtfSt+5sCE5Rwoy9k3wT0b3qWpbEjqSlFmT5I67EdZZi5fLdvC2m07SU8ShnbPYlDHVFYvX0zltuX0SthEF7YgvpJ6YoqrcWjT0+UvuVX1/du50d3T8mL3bLnjAOg00JWAOvRz91USvMAnbl1W5H5IFRe6dckOv/VOb+17zy9dcaHbN6e7d4/6uqXdvrBrs/vMbfjelcoKVtf8Oye1cqWzNj2hyxDoOszVMmT5dTUpL3UBcdMit3QfAfsdG9rnv5aYBzfvpA8Dv8fVsSwABuH6uW3A9XFLAh5Q1esjzVhzYcFtL6AKa752QW7Bf1yJBQCBDv2h+8Guu8M+oyEnhmOFblvhhktb9I7LnwhkdHCjyWR1dus2vbxqoP7uS9MXSLevdQ13ln0My6ZDyfbwzp3TAwaeDoNOd19gqrDyM5g7xY1cU1kGbXq7hj2+X+rp7d2X8q4t7td9RUnDrj8ly33RZ3Z0Vbtlu13V7fCLYP+xsGuT+3Hy3YuuNNx2H3eftq+B/JWutFGbJLhnrWk5UFGGlu5ES3eRUCuvu1M7sqv9AZR3HspOSWfH+iUkbFtGTtEaurGRJKljYpD0duzO7Mm8XW34ens2hWldGdA+iQ5lebQr30Db0g20Kd1AkpZRlNqekrQOVGR0RDM7kZaaRvauFSRuXuSuIWLiaiBSsyAl0/vh0BpSW7t1UpoLXFuWuM9Ydfs8t2+7fatHK+rQz3u23d393WJYsoxLcPNOfBlwK25mbn9bgdtV9ZFIM9WcWHDby5SXumdykgDdhrsvhqagZKf7Uoqkj2BFuXtWWFECFWVeK9IyF6Aqyl0jnMoy915FqQvoXYbU/UW2c7N7lrhmtgs+XYa4pXXX6n0qK9wX9BavQZBWeiW0nOp1QqJXktteXZqrLHMliQ793Zeq73jF22H+qzDnWVeqS053wU4Sof9JMPxir5TsV/VXVuzyULAadm+rWR1clO9KUykZ7lgpmZCS7gJ216F1dlGprFSW5uWzfesGhvZsR2Jikvus+J7r+lXZzlq2hUc+/pl1BUWUV1QPjF1Roewuq6gxEoyPCHTJbsWgdnBw+kb6J+eRlaykpySSniS0SkmgVXICiSnpkNoaSWuNpraG1CyS07NJSGvtrinUIFRe6n4IbF3qGjh1GuSG6WsC4hbcvJMnAIOBfXCtGtcAX6tqecQHbWYsuBkTRb5S9vf/cs+xho6vu69kE6aq7CqtcH3+vFadW3aWsnLLLpZv3smKLbtYvnkXO0pC/ypNEMhJT6FNejJtM1Jok55Cx9apHNg1hyE9cti3Q+ZeNU5oXIObqZ8FN2NMJFSV/N1lbNtVyvaiUq+LQxkFRW6uQFVQtKoFaVFpBfm7S92yq4z83aWsyy+qCpBZqUkc1D2Hg7pn0zUnnQ5ZqXTISqVjVirtM1Ob3KwQzWlWAGOMMR4RoW1GCm0z9mylGarKSmX5ll18tzrfmymigL/PWB6wSjQzNYmstKSqPoqt01wJMDc7jc7ZrcjNSSM3O42uOa3qHT2msLiMz5dsoUfbdA7oFp+q/YiDm4h0xw1l1QWoqzODqurtkZ7DGGNM5BIShD4dM+nTMZOzhncH3PRIW3aWsHmHWzZ564Ld3nBp3pBpeYXFLFhfyKYdxfjHQhHo1ymLg3u15eDebRnRqy2dWqeyfMsuPlm0iU9+2sQ3K7dRXqlccFjPuAW3SLoCJAGPAb/F9UrFb+2j3nuqqi16SHirljTG7M3KKyrZvLOE9QXF5G0vZummncxZ5aZI2l3qWlpmt0pme1EZ4ALf0QM6ckz/jgzunhPxqDDxqJa8DbgUKAfex43T2DKmfzHGmBYmKTGB3OxW5GbX7LxfXlHJwg2FfL1iG4vzdnBgt2yO6t+Rbm3S45TTmiIJbuOBXcDhqjo/yvkxxhizF0hKTODAbjkc2C0n3lkJKJLyYkdghgU2Y4wxTVUkwW010MChB4wxxpjGE0lwewUYLSJNoxu7McYYU0skwe3/gMXAeyJSxwRWxhhjTPyE3aBEVUtE5DjgS2CBiKzCTeQZaCRRVdVjGphHY4wxJixhBzcRaQ9Mw80EILhxJfepI7mN7WWMMSbmIukKMAk4CFc1+XdgKdbPzRhjTBMSSXA7GTdv26GqGuYkUcYYY0zji2T4rZ3AB6p6VuNkqXkRkc3Aqgh3bw9siWJ2TPTZPWr67B41fYHuUU9V7RDpASMpuS0CsiI9YUvTkJsjInMaMraaaXx2j5o+u0dNX2Pco0i6AjwOjLFuAMYYY5qqsIObqk4BHgKmi8jFItIt2pkyxhhjGiKSrgAVfi+f8t6rK7mqqk2IGrmn4p0BUy+7R02f3aOmL+r3KJIGJYE6a9dJVZvW3OXGGGOavbCDmzHGGNPUWanKGGNMs2PBrQkSkXNF5HMR2S4iO0VkjohcKSJ2vxqZiCSLyDEicr+IfCUiG0SkVETWichUERlTz/527+JARP5PRNRbrg+Szu5PjIlIKxH5k4h8IyIFIrJbRFaIyGsicngd+zT4Plm1ZBMjIo8DVwDFwMdAGXAMrm/hG8BZqlpR9xFMQ4jIL3BjpwLkAXNxM88PBPb33r9dVW8NsK/duzgQkYNxA7kn4Ma7vUFV7wuQzu5PjIlIb+B/QB9gE/AVbj7QXsBg4G+qeketfaJzn1TVliayAGNxg01vAPbze78TsNDbdk2889mcF+BoYCpwRIBt44By7z4cZfcu/guQCiwA1nlffApcHyCd3Z/Y35sM3NjDCvwNSK61vR3Qt7HuU9z/ALbUuLFzvJt3foBto/1uekK889pSF+AZ7z78w+5d/Bfgbu9veyowJUhws/sT+3tzl/d3fT6MfaJ2n6yeuYnwOsMPA0qB12pvV9UZuF+nnYFDY5s74+c7b101eIHdu/gQkUOA64CXVfWdIOns/sSYiKQAl3gvJ4W4T1TvkwW3pmOIt16gqkV1pPmmVloTe/t56w1+79m9izERSQOeB7YB19ST3O5P7A3DVTuuUdVFIjLSa/TzpIj8VUQOC7BPVO+TjR7SdPT21sFmEFhdK62JIRHpDEzwXr7ut8nuXezdCfQDzlHV+kb8t/sTewd4659FZApwQa3tt4rI68B4v0AW1ftkJbemI9Nb7wqSxjcprM3KEGMikgS8CGQDH9eqBrN7F0MiMhL4A/Cmqv47hF3s/sReW299JHA+cB+uxWQb4DRc9eJY3ED8PlG9Txbcmg7fAJ3WN6Np+juuOfIa4De1ttm9ixERaQU8BxTimouHtJu3tvsTO77YkoRrfHWDqi5T1QJVfRs4HXc/LhCRfby0Ub1PFtyajh3eOjNIGt+2HUHSmCgTkYeBi3H93o5R1bxaSezexc7/AX2Ba1V1Q32JPXZ/Ys//7/h07Y2qOgfXhzQBGFNrn6jcJ3vm1nSs9NY9g6TpXiutaWQicj9wNbAZF9h+DpBspbe2e9f4zgAqcb/4az/H6e+tLxeRU4Clqvpb7P7Ew0q/f6+oI80KYDiu9aP/PlG5Txbcmg5fE/NBItKqjtZCB9dKaxqRiNwDXAtsBY5V1YV1JLV7F1sJuD5PddnHW3K813Z/Yu9bv3+3w/04rK29t/Y9R4vqfbJqySZCVdfgPhApwFm1t4vIaFzfqjzcUEOmEYnIJOAGIB8X2L6vK63du9hR1V6qKoEWXNcAcMNviaoO9vax+xNjqroOmO29PKb2dhFpAwz1Xs7x9onqfbLg1rTc5a3vFpE+vjdFpCMw2Xs5SVXDmlPPhEdEbgduBApwgS2UX/N275o2uz+xd6e3vlVEBvve9PooPoFreTyXmoEqavfJBk5uYkRkMnA5btDQj6geNLQ18CZwptrgro1GRH4JvOW9nIMbtzCQn1S1xsgLdu/iy68/VV0DJ9v9iTERuRe4HjfqyGxcFf8IoAuuO8BRtZ9jR+0+xXv8MVsCjq92LvAFrrnzLtyvmyuxce9i8befgGuKXN8y3e5d01oIMrak3Z+43pczgE9wVfwlwM/A/UCHxrxPVnIzxhjT7NgzN2OMMc2OBTdjjDHNjgU3Y4wxzY4FN2OMMc2OBTdjjDHNjgU3Y4wxzY4FN2OMMc2OBTdjYkBEVoqIhrCMiXdeQyEit3n5vS3eeTEmEJsVwJjY+i9u4Ne6BNtmjAmRBTdjYmuSqk6PdyaMae6sWtIYY0yzY8HNmCZIRHp5z7RWikiSiEwUkUUiUiwiG0XkeRHpEWT/QSLygoisEZESEdkiIu+LyIn1nPd4EfmPiKwXkVIRyRORL0TkRhFpVcc+nUTkSRFZ651rhYhM8qY2qZ02UUQuE5FZIrLdO8dGEflWRO4XkQ7h/7WM2ZMFN2Oavn8DfwVW46b8KAXOB74RkX61E3vT9swFxgPbgdeBhcDxwPvefHW19xEReQL4EDeK+zpvv++B7sAkoFOAvHX3znUKbl6u6UBH3Hx4rwZI/w/cXF6DcVOgTPXOkY2b9XzfoH8JY0IV7+kQbLGlJSzAStx0LGNCTN+L6ul1NgID/balAP/0tn1da7/OuICmwLW1to3BTR+iwPG1tv3Rez8POLTWNgGOArL93rvNL39PAyl+2wYAO7xth/u939N7bzXQKcA1DwY6xvte2dI8Fiu5GRNbnwbpBlBQxz63q+pC3wtVLQWuwgWxg0XkcL+0l+AmdZylqg/4H0RdQ5bHvJfX+94XkSTg/3kvJ6jqV7X2U1X9VFW3B8jbGuBqL0++9ItwwRfcJJM+Hb31t6q6sfaBVHWeqm4KcA5jwmatJY2JrWBdAXbX8f6Ltd9Q1e0i8i5wHq5E9oW3abS3nlLHsZ4F/gSMEpFEdTMaDwfaA2tV9cP6LqCWT1S1KMD7P3nrLrXe2wGcLCL/D3hJVVeFeT5jQmLBzZjYCrcrQIGqFtSxbaW37ub3XldvvaKOfVYAlUAa0A7YhKsuBFgcRr58VtfxfqG3rmpUoqo7ROQiXIC9E7hTRNbhntW9B7yiqsUR5MGYPVi1pDF7P/X7twR4rzFVhpNYVacCPYAJuCC3EzgTeA74SUS6RzuDpmWy4GZM05YjItl1bOvlrdf7vbfWW+8TZJ8EoBjY5r3nqxrco+VlY1DVAlV9XlUvVtX+QB/gU1wJ8u5Y5ME0fxbcjGn6zqv9hhfwTvFeTvfbNMNbn1/HsS701jNVtdz791xgC9BNRI5vWFbDp6rLcNWUAAfF+vymebLgZkzTd6uIDPC9EJFk4GFc37C5qjrTL+3TuEYbo0Tkav+DiMiRwO+9l/f73lfVMuAu7+VzIjKi1n4iImOClCBDIiJDRGRcHZ3BT/XW1sDERIU1KDEmtiaKyIQg219W1f/5vV6NK1nNE5FPcM3/D8M9t9pCrRKaquaJyHhcx++HReS3wI+4VotH4H7Q3hGgVeSDuP5pvwW+EpE5wFKgLTAQ11m7t3f+SPUEXgF2i8i3uG4EKcAQXDXqDuDWBhzfmCoW3IyJrfqq/eYB/sFNgbOBibgRR3riWiK+CPxZVVfWPoCqviUiw3GjhByNa7Cxwzvuo6r6foB9FLhERN4CLgNG4DpVbwN+Bh6l4TMWfAXchOuu0B8YhhttZQ2uJPmodQ0w0SLuM22MaUpEpBeu2f4qVe0V39wYs/exZ27GGGOaHQtuxhhjmh0LbsYYY5ode+ZmjDGm2bGSmzHGmGbHgpsxxphmx4KbMcaYZseCmzHGmGbHgpsxxphm5/8DaanGrg68hO8AAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_predict_function.<locals>.predict_function at 0x108819950> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_predict_function.<locals>.predict_function at 0x108819950> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[32m[I 2022-04-01 15:46:22,849]\u001b[0m Trial 5 finished with value: 0.14239360392093658 and parameters: {'batch_size': 50, 'lr': 0.00014083659085561683, 'n_layers': 2, 'n_units_l0': 80, 'dropout_l0': 0.21122109285535867, 'n_units_l1': 37, 'dropout_l1': 0.2252598778382316}. Best is trial 5 with value: 0.14239360392093658.\u001b[0m\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Block: verbose not specified in input, defaulting to 1\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a4e896440> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a4e896440> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x1a4f372680> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x1a4f372680> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[32m[I 2022-04-01 15:46:26,013]\u001b[0m Trial 6 pruned. Trial was pruned at epoch 0.\u001b[0m\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Block: verbose not specified in input, defaulting to 1\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a4c4e67a0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a4c4e67a0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x1a4c559950> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x1a4c559950> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[32m[I 2022-04-01 15:46:27,296]\u001b[0m Trial 7 pruned. Trial was pruned at epoch 0.\u001b[0m\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Block: verbose not specified in input, defaulting to 1\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a4d39fd40> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a4d39fd40> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x1a4e92b440> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x1a4e92b440> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[32m[I 2022-04-01 15:46:29,354]\u001b[0m Trial 8 pruned. Trial was pruned at epoch 0.\u001b[0m\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Block: verbose not specified in input, defaulting to 1\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a4efa1c20> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a4efa1c20> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x1a4f292290> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x1a4f292290> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[32m[I 2022-04-01 15:46:32,370]\u001b[0m Trial 9 pruned. Trial was pruned at epoch 1.\u001b[0m\n"
     ]
    }
   ],
   "source": [
    "study = optuna.create_study(direction=\"minimize\", pruner=optuna.pruners.MedianPruner())\n",
    "study.optimize(objective, n_trials=10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Print out `study` statistics and optimal hyperparameters found"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Study statistics: \n",
      "  Number of finished trials:  10\n",
      "  Number of pruned trials:  4\n",
      "  Number of complete trials:  6\n",
      "Best trial:\n",
      "  Value:  0.14239360392093658\n",
      "  Params: \n",
      "    batch_size: 50\n",
      "    lr: 0.00014083659085561683\n",
      "    n_layers: 2\n",
      "    n_units_l0: 80\n",
      "    dropout_l0: 0.21122109285535867\n",
      "    n_units_l1: 37\n",
      "    dropout_l1: 0.2252598778382316\n"
     ]
    }
   ],
   "source": [
    "pruned_trials = study.get_trials(deepcopy=False, states=[TrialState.PRUNED])\n",
    "complete_trials = study.get_trials(deepcopy=False, states=[TrialState.COMPLETE])\n",
    "print(\"Study statistics: \")\n",
    "print(\"  Number of finished trials: \", len(study.trials))\n",
    "print(\"  Number of pruned trials: \", len(pruned_trials))\n",
    "print(\"  Number of complete trials: \", len(complete_trials))\n",
    "\n",
    "print(\"Best trial:\")\n",
    "trial = study.best_trial\n",
    "\n",
    "print(\"  Value: \", trial.value)\n",
    "\n",
    "print(\"  Params: \")\n",
    "for key, value in trial.params.items():\n",
    "    print(\"    {}: {}\".format(key, value))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Rerun CFL with the optimized hyperparameters\n",
    "\n",
    "Now, we run CFL with a CDE that is parameterized by the optimal hyperparameters \n",
    "printed out in the previous cell, and a CauseClusterer that uses the\n",
    "expected conditional probability estimated in the previous step to construct\n",
    "macrovariable states."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "All results from this run will be saved to demo_results/experiment0003\n",
      "Block: verbose not specified in input, defaulting to 1\n",
      "#################### Beginning CFL Experiment training. ####################\n",
      "Beginning CondDensityEstimator training...\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a54916d40> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x1a54916d40> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x1a54cb10e0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_test_function.<locals>.test_function at 0x1a54cb10e0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAa4AAAE0CAYAAAB0CNe/AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/d3fzzAAAACXBIWXMAAAsTAAALEwEAmpwYAABa6klEQVR4nO2dd3xUVfr/389MKiQk9ITepKoUsSEI6tp11cW2uii669r1u5ZFv5Z1V/2Cfe1rWUVX/akLa8O2WEARRUERpClK6KGHmj7P749zJ5kMk0kymTTyvF+v+zq595x77nPvncxnzjnPeY6oKoZhGIbRVPA1tAGGYRiGURNMuAzDMIwmhQmXYRiG0aQw4TIMwzCaFCZchmEYRpPChMswDMNoUphwNWFEpIeIaC23HvVo45g6ukaw/vF1Ub9RjoiMqelnJ+ycmLY6vq06I+zzf0dD27OvYMJlGIYRhv0YatwkNLQBRq1YCaRXkjcKeM/7+zLg5UrK7Y63UYYRxudU/jk9H/iH9/dJXlnDiIoJVxNGXdiTXZHyRCQ/ZLdQVSOWq2tUNQeQOr5GndZv1A5VLaXyz2lhyG5+Q31OjaaFdRUahmEYTQoTrmaIiIwPHfQWka4i8rCI/Cgie7y8zJDy+4vI/4rIDBHZICLFIrJDRBaIyAMi0jXKtaI6Z4hITujAtYj8WkT+KyKbRKRARJaJyN0i0irKNSodj/BsVhGZ7O2PEpE3RSRXRApFZIWIPCoiHat4ZiIiF4vIl9697xCRuSJytYj4w59pTRGRFBE5UUSe9J7rDu85bxSRj0TkEhFJinL+ZO/6M7z9A0XkJRFZ493nGq9M72rYcqqITBeRrSKyW0R+EJHbRaRFLPcWD0Skr/eeFovITu9zukxEHhGRblHOExH5rYi8JyLrRaRIRLaLyHLvHv8cen7w8xJSxfMRnEXG1OGtRrqHU0XkDRFZ573LzZ6dV4hIYpTzanTvIee19PJme5+BYu+aS0Tk395nsWXd3nUVqKpt++AGjAHU28aH5Y0PyTsM2BqyH9wyvbKDI+SFbzuAYyuxo0dIuTER8nO8vDuAh6JcYz6QVsk1It6nlzfDy5sM/AkoraT+lUDnSupPBN6MYtt04JLgfozvK9q9B7cvg+8lwvmTvTIzgHOAgkrq2AocEMWOB6JcfwFwRsh+jzh8TkM/i3t9Prwy1wPFUezaDZwS4Tx/Fe8tuN0Q4fMSbYtoZzU+/3fU8NkkA69XYcv3QKd43Lt3XjbwYzXOG14X31vV3azFZUzBjT9cCHQFsoBTcV984D6kc4AJwJFAX6At0B83sD4XN/D+uohk1cKOccD/AM8CB3vXGAA85eUPBv63FvUfiftSfhsYCbQDegF34u6xG3BfJef+H3Ca9/c0YIR3/iCvzmOAm2phG7h38G/cMz0Y9y46AgcBfwO24X5kPFlFPX2AF3Dv7Digg1fX/wBFQGvKn2kFROQPwHXe7jfAsUB7YD/cD4v+uPutN0TkSuB+3Hj8NOB43Ge0Pe7+PgdaAP8WkQPCTh9P+Xt7Evf8OuG+nA/Bfebfw4likBOp6EhymbcfutWXA8mjwFne328BR+A+dwNxn9VS4EDg3Qit8fHU/N4B7sW97z3An4H9vWv2wDl8XQfMw/3PNBwNqZq21d1G9Vtcm4EutbhOAjDLq+uvEfJ7hFxrTIT8nJD8Wyq5xtte/vpK8qvT4lLg6UrOf8TLLwRaheV1BUq8/LcAiXD+hJBraB29zwM8OwJA7wj5k0NseB9IiFDmupAy/cPyUrzPguJaty0inH8BFX9194jDfYV+FseE5WVT3nJ8Ksrn71OvzLSwvKne8TdisKvSz1QN6wn9/N9Rg/OGhpz3ciWfuytCylwTj3sHtnjnXVsXn+N4bdbiMu5T1TWxnqyqJcAr3u6xtbBjNTCpkrznvTRLooynVUHwF2S0+pNwLbtQzsN1uwBcr95/dxj34+yvM1R1IfAtzkPzmCqKX+u9l3Amh/x9cFjeKbhWLsBNqrongg0v4lrY9cVluO6yLcA1kQp493m7t3uShIzNUu41vbauDKxDfu+lRcD/RPrcqeoTuK5CgD+EZcd6703imZlwGe9Wp5CInCEir4vIz96AfWhUg8e9Yv1qYcd0dW7TkVgW8nes3ZFfqWpeDPWP8NLFqro80sme3dV6jtEQkTYiMkHKnWCKwp5zUGyiPedfVPXHSuzcCmzydsPvc6SX7saN2VXGf6q4jXjyKy/9HEgUkbRIG7DUKye4rtUg33npxSIyTkSS68nueBB8HzNUdVOUclO8dP8w0Y713oPnTRSR0SLSKKea2Dwu45domZ4n2X9wYwtVkVELO9ZFyQv99R+rZ1ul9avqnpD/z/D6e3jpMqKztIr8qIjIYbgu0fbVKB7tOUd7jlD+LCu7z5+i/IAAWFJF/fEkKNCnAzureU7o83sI1xXZHXgReFJEvsB1bc8AZldxrw1Jdy9dXEW5RV4quHHaPG8/1nv/M+6HQh+v3AYR+cw79omqLopwTr1jLa5mTqQuoTAeoFy0XsB1KfXBDdgGB6uv8PL9e51dfar7BRLrL8BY6w+6/VYVYSTmibPiXP3fxH3pbgJuBg4HOgOZlD/nL7xTov3gjPU+07y0qvuozwnCsfwQSgn+oarbcY4Ij+KcW1riHDr+BnwGrBaRa0WkMX4PBh1EqnreoYJe5lQS672r6tfeeW/guik74hxEHgF+EJH5InJCrDcVL6zFZVSK19oa7+3eo6oRPedEJCXS8X2EoGBVNW8lrYr8aJyJ+4IIAEdV9qtWRCoLmxQPgl+QVd1Hbe6zpuzCCff9qnpjLBWo6kbgGhH5EzAE94PgKOAEnPPH33GtzT/V2tr4shN37zV5HxVapbHeu6p+D/zG+/8/FNddfizOq3Aw8J6IjFXVN2p+W/GhMf7SMBoP/Sn/Bfv/opQLd0Pel1jppX2rKFeb8b0hXrogimglVcOG2pDjpfuJSLSW84A6tCGcYDf20NpWpKqlqjpPVR9T1bE4b9FZXvbVItI2yukNQY6XDqyi3CAvVWBVpAKx3ruq7lHVT1X1blUdg3O934Rrrf+l2ndSB5hwGdEIHdCN+GXmDY6fXi/WNAzB7rlBUknUCe+L/uRaXCP4nKMJxlmEdIPVAcEvspZE9w79TR3aEM5/vXS0iPSMZ8Weo8qD3q6fvX8UlITkNQTB9zFaRNpFKXeml/4QxfmoAtW498rOW0T5D9j6/AGzFyZcRjRWhPx9Wnim53H0CG5S677KK5SPG91fiZfVdbiB8VgJtiwGiMheXyIi0hm4pxb1V4dpOLdzgEkSIbyTiFwADK9jO0J5HDe3LgH4l/cjqVJEpF/Yfv8q6g/9IbIlLC+436kadtYF//TSZMpFpgIicinlrfVnwvJqfO9eqKeqppsEzwt/XvWKCZdRKaqaC8z0dm8Wkb+ISD8RaSsiRwLvABdRtedTk0VVVwMPe7unA2+KyGGe63p/EbkPN/8sqndmFUzFiWMCLgrCGSKSLSJdxMVf/Ar342BllDpqhaoWUB79YzDwqYj8ynvXvUXkdlxUkxWVVhJ/m9bgIn6AixoxX0QuFRe3MNN7RiNE5HoRmUO5a3iQ90XkKy//cBHJ8t7bIBG5BbjLK/dthCkE87z0Qu/cNBFJ8LZYHYS6eJ+daNtQ797nUy5G40RkatjnbhLl01Dms3c0lFjuvT3wi4i8JSIXicgBItJORDp6dTxPec9CtKGDuqehZ0DbVjcb1YycUY16+lMeUSHS9ipwcWX1UYNYhVFsqKqO6kTOmFzFfUarIwkn0pU9g4+AP3p/l8T4vm6MUn8+MDbavRASq7CK60R93kSPVbiQholV+Ecqj70Yus2r5F6jbSuAvhGueXyUcyLaWY3PbnW2nJBzqxurcK8Ym7Hcew1snQ60rO27r81mLS4jKqq6FDep81ncbPpi3ADtJ8A4VT0X5w23z6KqRbiu0ktwMQB34Ty4vsN1E55A+ThVdecbhV/jPlyMyE9wQYsLcV8+zwEHq+rU2O+gRnZcD/wa+Bg3J2gPbu7Wnbh4d9vqw44wm57GxZW8G/gaFyi4FPesF+NE+zRcqyyU43ERN97E3cM23NjVZpxL+PXA/hphwraqfoh7r+8DGykf86o3VLVQVc/G3dtbQC7u/28rzv6rcJ+NSFEuYrn3Vbhn+Bfc+/8Z51VbBKzBzTM8BzhOVRt0AVrxlNYwjFogIg/jvih+UNV92cvSMBoca3EZRi3xxjyCff/zopU1DKP2mHAZRhWISFIVHm1/ptzb6vV6MMkwmjUWOcMwqqYN8L2IPIVbw2g5bhJmP1wU7/FeuS+ADxrCQMNoTtR4jEtE/gPkquoVVRY2jH0AcQtkrq+i2CLg+EoGyg3DiCOxCFch8KaqnlM3Ju1btGvXTnv06NHQZhi1QFXZunUr27dvJz8/n+LiYkpLS0lISCA1NZXMzEzatWuHz2c974YRL+bNm7dZVSOulhBLV+FaILF2JjUfevTowdy59bn2nmEYRtNHRCqdcB/LT8RpwKhIIWEMwzAMo66JRbjuwE2QnFKNuFaGYRiGEVdi6Sq8HzcQfQrwk4h8i4uhlh+hrKrq72thn2EYhmFUIBbhGo+LVwUuhtth3hYJxbkLG4ZhGEZciEW4Loq7FYZhGIZRTWosXKr6Ql0YYhiGYRjVwSaeGIZhGE2KWoV8EpEk3JIXnb1Da3Fr4hTV1rDmzrLcnUxbsI7xI3rQNi256hMMwzCaCTG1uEQkUUTuxq1TMwt4zdtmAZtE5C4RsUnKteCXTbt49JPlbNxZ2NCmGIZhNCpq3OISET9uEvKvcIFG1+OWLRegJ5AN3AwcLCInqWpp/MxtPqQk+gEoKLbH19woKSkpCzFVUlLv6xcaRlxJSEggIyODNm3akJAQn7jusdTyR+BY4Efgf1S1QjRsETke+DtO2C4B/lFLG5slQeHKN+FqVgQCAVavXk1ycjLdunUjKSkJt9yXYTQ9VJWioiK2bNnC6tWr6d69e1xiesZSwwW45ZyPCRctz9APcaK1B7iwduY1X1KTnHAVFgca2BKjPtm2bRsJCQlkZ2eTnJxsomU0aUSE5ORksrOzSUhIYNu2bXGpNxbhGgh8Gm35Bi/vU69sjRGR80TkcxHZLiK7RGSuiFwpIjWyV0SuFpHXRWSJiGwRkWIR2SQiH4nI76SKb4V42RELKYnuEtbial7s2rWLzMxMEyxjn0JEyMzMZPfu3XGpL5Yv4ERca6oq9hBDFHkReRx4GRgOfA5MB/oCj+HiI/prUN0E4HRcOKrZwFTcIoBHA/8C3qhMhOJsR41JDXYVFplwNScKCgpo0cLiVxv7Hi1atCA/P1JkwJoTyxjXSlx0+KTK3N49N/mRXtlqIyJjgSuAXOBIVf3JO94R14I7A7gKeLiaVZ4LfKeqFWReRAYBHwOn4bozn69jO2pMmXNGiQlXcyIQCNi6XsY+ic/nIxCIz9BHLP8hb+M8B18QkczwTBHJAJ7zyrxVw7pv9tIJQbEAUNUNwOXe7k3V7apT1VnhouUdXwQ87u0eW9d2xEKKtbiaLdZNaOyLxPNzHUuL617gt8DZwIki8g6wAhdQtxdwKpAOrPHKVgsR6YKbzFwE/Ds8X1Vnisha3GTnw3Bdf7Uh6Gdc0MB2RCTV3OENwzAiEkuswi0icjTwCm7853zKo8UHJfUb4DxV3VqDqod66SJVrawj9BucYAylFoIhIj2By7zddxrKjmgk+gWfQIF5FRqGYVQgptlgqrocOERERgKjcV/igmtlzVTVWTFU29NLo42LrQorWy1E5CKcnYlAF2AErpt0oqq+UV921AQRITXRb16FhmEYYdR4jEZEHhSR26FsDOluVb1CVS/3/o5FtADSvDSav+QuL02vYd1H4JwwzgOO9I7dBvytLuwQkT96rvNzN23aVENTy0lJ9FtXoWE0EsaPH4+IMHny5Hq/9h133IGIcMcdd9T7tRsjsTgXXA0MjrchlHczatRSMaCqf1BVAVoAg3CRPe4AvhKRTvG2Q1WfVtXhqjq8ffv2sVZDirW4DGMvevTogYiQk5PT0KYYDUQswpVLuWNDPNnppWlRygTzdkYpUymqmq+qi1X1Rpzn4GDcvKx6taO6pCZZi8swGgsTJ05kyZIlnHHGGQ1tSrMnFuH6CDhCROITLbGcHC/tHqVM17CytSE4d+vUsEj29W1HpaQk+sw5wzAaCdnZ2fTv35+MjIyGNqXZE4tw/QVIBf4hIi3jaMt3XjpIRFIrKXNwWNnakIdrOSYAbRrQjkpJTfTbPC7D8Jg8eTIiwsqVzm+qZ8+eiEjZlpOTU1Zm/PjxbNmyhWuuuYaePXuSlJTE6aefXlbX1KlTufjiixk0aBCZmZmkpKTQp08frrzySlavXh3x+pWNcYWOP23YsIFLL72ULl26kJycTM+ePbnpppsoKCiIWGe8ePfddznxxBNp164dSUlJdO3alQsvvJAlS5ZELL9u3Tquuuoq+vTpQ0pKCi1atKBbt26ccMIJPP3003uVf/XVVzn66KNp06YNiYmJtGvXjgMOOIArr7ySn3/+uU7vLRKxtJrGA+8DFwGnich0nAdeJNdxVdU7q1Opqq4WkW+BYcBZwIuh+SIyGucRmAt8GYPd4RyJu/88YHMD2lEpKYl+dhXashaGAdCnTx8uvPBCpkyZwu7duxk7dixpaeU9+qF/b968mYMPPpjt27czatQohg8fTtu2bcvyzznnHFJSUhg4cCC/+tWvKCwsZP78+TzxxBO8/vrrfPHFF/Tt27dG9q1evZqDDjoIVWXEiBHs2LGDWbNmcc8997B48WLefvvt2j+ECNx8881MmjQJn8/HyJEj6dy5MwsWLODFF1/k9ddfZ8qUKZx88sll5devX89BBx1Ebm4u3bt354QTTiA5OZm1a9fy1VdfkZOTwx//+Mey8nfccQd//etfSUxMZMSIEXTq1Im8vDxycnJ44oknGDVqFL17966Te6uMWITrDpzjggBtcWGVwgnmK1At4fKYiJv0e4+IzPbc7hGRDsATXplJqlrWfyYiE3EhmN5Q1ZtDjo8CugFTVLXCaowicgTwT2/3nxHWDKuxHXVBSqKfTbaQpOHx13cWsXjdjoY2o0YM7NSKv5w6KC51jRw5kpEjRzJjxgx2797N/fffT48ePSKWfffddznuuOOYMmUK6el7O/++8sornHLKKRXiQpaUlPDXv/6Vu+66i2uvvZb333+/RvY999xz/OEPf+Dxxx8nKSkJgCVLlnDIIYfwzjvv8MUXX3DEEUfUqM6qeO+995g0aRItW7bkvffe48gjjyzLu++++/jzn//M+eefz48//kiHDh0AeOaZZ8jNzeXSSy/lySefrBDRorCwkDlz5lTYv/fee0lLS2PevHl7iflPP/0UtzW2akIsV/wbdeD5B6CqU0TkSVxYpYUi8hFQDBwDtALeZG9nimygn5eG0hs3jvWY14LKxbmv96Y8av27OLf4eNgRd1LNHd4wYiIxMZGnnnoqomgBnH322XsdS0hI4M477+S5557jv//9Lzt37qz0/Eh07dqVRx55pEy0AAYMGMC4ceN48skn+fjjj+MuXA888AAA1157bQXRArjxxhuZOnUqc+bM4ZlnnuGWW24BYMOGDQCccMIJe4VhSk5OrlDPjh07yM/PZ/DgwRFboPvtt19c76e6xBI54446sCO0/itEZBZwJW7SsB9Yiot/+GQNWjkzca29Ubio7iNwrcBcXJT4l1T1zXqwI2bMOcMIJV4tl+bAsGHDKm2NBfnxxx/54IMPWL58Obt27SoLAFtSUkIgEGD58uUMHTo0ah2hHH300aSm7j0s3r9/f8CNK8WTkpISvvjiC8CNv0XioosuYs6cOcyYMaNMuA455BCeeOIJJkyYAMCxxx5Ly5aR3RXat29Pjx49+P7777n++uu55JJLyu6nIamxcHmtl19U9cw6sAcAVX0FF1KqOmXH48bdwo+vAG6vLzvqAoucYRix0b175U7BJSUlXHHFFTz77LOoVt55tGNHzbplu3XrFvF4q1atAOLuoLFlyxYKCwvx+XyV3m9w7Gnt2vLlE8eNG8d///tfXnnlFc444wz8fj/7778/Rx55JOeeey4jRoyoUMeLL77I2LFjefDBB3nwwQdp3749hx12GMcffzy/+93vGsTLMhavwv64bjOjjkmxeVyGERORWj5BHn74YZ555hmys7N59dVXWbVqFQUFBagqqsrhhx8OEFXUIlHfy9GE2ldZ5PVI9+Dz+Xj55ZdZuHAhEydO5MQTT2TVqlU8+uijHHHEEfz+97+vUH7UqFHk5OTw2muvcfnll9OpUyemTZtW5pX43Xd16lwdkVie9EqiT8414kRKgp/CkgCBQJ0MKRpGs+Tf/3aLPjz11FOcc845dO3aleTk5LL85cuXN5RpNaJdu3YkJycTCAQqjSKyYsUKADp37rxX3v77789NN93EO++8w+bNm3nnnXdIT08vG+MLpUWLFpx99tk88cQTzJ8/n7Vr13LOOeewefNmrrzyyrjfW1XEIlxTgSNFpF28jTEqkppki0kaRjhB54eSktimimzd6hat6Nq1615506dPpzbxReuThISEMmePF198MWKZ4JyzMWPGRK3L5/NxyimncNpppwHw/fffRy2fnZ3N3XffXa2ydUEswnU3zknhQxE5NM72GCGkJLjXYw4ahlFOsPVQ2eTaqgg6Fzz55JMVVuT9+eefueyyyyo7rVFy3XXXAfD3v/+9zFEjyIMPPsiXX35JRkYGf/jDH8qOv/jii3z77bd71bVlyxa+/NJNTQ2Oma1cuZJnn3024njfO++8U6FsfRKLO/y7QCkuesRsEdlA9AnIx9TCvmZNsMVlDhqGUc4ZZ5zBjBkzOP/88znuuOPIzMwE4J577qnW+TfffDMffPABTz31FJ9++ilDhw5l69atzJw5k8MPP5ysrCxmz66TZfbizsknn8yECRO45557OPLIIxk1ahSdOnVi4cKF/PDDD6SkpPDSSy/RsWPHsnP+85//cOGFF9K5c2eGDBlCZmYmW7Zs4fPPP2f37t2MGjWqLB7jtm3buOSSS7jyyisZMmQIPXv2JBAIsHjxYhYtWkRiYiL33lvt9YLjRizCNSbkbwGyvC0SNjhTC1K8VZAt7JNhlHPVVVexY8cOXn75ZaZNm0ZhoZukf+utt1br/MMPP5yvv/6aW2+9lblz5/LWW2/Rs2dPbrnlFiZMmMDxxx9fl+bHnUmTJjFy5Egee+wxvvnmG2bPnk2HDh0YN24cN910EwMHDqxQ/vrrr6dHjx7Mnj2buXPnsm3bNtq1a8ewYcMYP348559/PomJLnxr7969eeihh5gxYwaLFi1i0aJF+Hw+OnfuzB//+EeuvfbaveqvD6SmnjNeyKNqo6oza3SBfYzhw4fr3LlzYzr3w0W5XPqveUy7eiT7d7bAns2BJUuWMGDAgIY2wzDqhJp8vkVknqoOj5QXywTkZi1E9Umq1+Iyl3jDMIxy6nfigVEjyrwKzTnDMAyjjJijI4qIDzgROBxoD8xR1ee8vPZAa+DnCAFsjWqSkmDOGYaxrzFp0iSWLl1arbIjR46s4BFoOGISLhEZBryKC1gbjAKfiIvjB/Br4GngdOCdWlvZTElNcg1iEy7D2Hf44IMPmDmz+iMuJlx7E0uswu7AdFyL6l1cMNtwf8ipwOOYcNWK5AQb4zKMfY0ZM2Y0tAlNnljGuG7BidZVqnqqqt4fXkBV84AllK8UbMRA+RiXCZdhGEaQWITreGCJqj5RRbnV7L1GllEDzKvQMAxjb2IRro7AD9UoV4BbuNGIkfIJyOZVaBiGESQW4dqJE6+q6AlsjqF+w8PvE5L8PnPOMAzDCCEW4foOGC4ilXYDikg/YAjwdYx2GR5uFWQTLsMwjCCxCNdzQAvgZRFpG54pIq1wrvA+yt3jjRhJSbTFJA3DMEKpsXCp6qvAG7hgu7+IyNte1mEi8hqwAhgFvK6q0+JlaHMl1VZBNgzDqECsIZ/Owc3dSgBO8Y71B87CrY78d2BcbY0zXPQMG+MyDMMoJ6bIGapaAtwkIvcARwG9AD/OBf4jVd0YPxObNylJfvItVqFhGEYZMccqBFDVbcB/qlteRE4DBqvq32pz3eZEqjlnGIZhVKC+o8OfDvylnq/ZpDHnDMOofyZPnoyIMH78+JjrmDFjBiLCmDFj4maX4bBlTRo5qYl+WwHZMAwjBBOuRk5Kop+CEhMuwzCMICZcjZyURL+FfDIMwwjBhKuRk5rop9DGuAyDpUuXIiJ06NCB4uLiiGVKS0vJyspCRFi0aBEAc+bM4cYbb2T48OF07NiRpKQkOnXqxJlnnslXX31Vn7dQgUWLFnHBBRfQtWtXkpOTadeuHSeddBLvv/9+xPIFBQVMmjSJYcOGkZaWRnJyMtnZ2Rx++OHceuutFBQUVCj/9ddfc9ZZZ9G5c2cSExPJyMigT58+nHfeeXzyySf1cYt1hglXIycl0WIVGgZA//79OfTQQ9m0aRPvvfdexDIffPABGzZsYPjw4QwaNAiAW265hYceeoji4mIOOeQQfv3rX9O2bVumTp3KyJEj+fe//12ftwHA22+/zUEHHcS//vUvMjIyGDt2LAMHDuTDDz/kpJNO4rbbbqtQPhAIcPLJJ3PzzTfzyy+/MHr06LJzVq9ezd13301eXl5Z+enTpzNy5EimTJlChw4dOOOMMzj66KNp3bo1U6ZM4fXXX6/nO44vtXKHN+qe1EQ/JQGluDRAot9+ZzRr3r8Jchc2tBU1I+sAOHFS3KobP348c+bM4YUXXuC0007bK/+FF14oKxfkhhtu4OWXX6Zjx4qxwd955x3Gjh3LZZddxsknn0yLFi3iZmc0cnNzGTduHIWFhTzwwANcd911ZXkzZszg5JNP5q677mLkyJEcf/zxAMyaNYtPPvmEYcOG8dlnn9GyZcuyc1SV2bNn06pVq7JjEydOpLi4mFdeeYXf/va3Fa6/ZcsWcnJy6vYm6xj7Jmzk2GKShlHOueeeS0pKCtOmTWPz5oqLT2zbto23336bpKSkCl/WJ5xwwl6iBXDqqady1llnsXXrVj799NM6tz3IM888w44dOxgxYkQF0QIYM2YMV111FQD331++Ru+GDRsAGDVqVAXRAhARjjjiiArCGyx/4okn7nX9tm3bctBBB8XnZhoIa3E1cpKDa3IVl5KektjA1hgNShxbLk2VzMxMTj/9dF599VVeeeUVrrnmmrK8V199lcLCQs4880zatGlT4bzNmzczbdo0fvjhB/Ly8igpKQHghx/c0oI//vgjJ598cr3cw8yZMwEqnSN28cUXc++99zJr1ixKS0vx+/0MGzYMv9/PP//5T/r27cvYsWMjinGQQw45hMWLF3Peeedxyy23cNhhh+H3++vidhoEa3E1coKrIBda2CfDAMq/8IPdgkEidRMCPPXUU3Tv3p2LLrqIBx54gH/+85+88MILvPDCCyxYsACAHTt21LndQdauXQtAz549I+b37NkTn89HQUEBW7ZsAaB379489NBDFBUVceWVV5KVlUXv3r0ZN24cU6ZMobS0Yo/MxIkTGTJkCO+//z4jR44kIyOD0aNH89e//pVffvmlbm+wHqhv4doCrKrnazZpUhLdKzIHDcNwHHvssXTp0oVvv/2WhQvdmN+yZcuYM2cOWVlZnHDCCWVl586dy+WXX05xcTH33XcfS5cuZdeuXQQCAVSVm2++GXDjRPVF8FoiUqPzrr76alauXMmTTz7J+eefT2lpKS+99BJnnXUWw4cPryC+WVlZzJs3j48//pibbrqJYcOGMWfOHO644w769evHc8817RWn6lW4VPUGVY38M8OISLDFZdEzDMPh8/kYN84tPjF58uQK6e9+97sKXWJTpkxBVbnmmmu44YYb6NevHy1btiwTjeXLl9er7QBdunQBqLTlk5OTQyAQICUlZa8uz6ysLC677DJeeuklcnJymD9/PgcccADz589n0qSKXck+n4+jjz6aiRMn8tlnn7FlyxYmTZpESUkJV155Zb22MuNNlcIlIhfUZquPm9iXCQqXOWcYRjnB7sCXX36ZoqIiXnrppQrHg2zduhWArl277lXHpk2bmD59ep3aGYnRo0cD8OKLL0bMf/755wEYOXIkCQnR3RAGDx7MtddeC8D3338ftWzLli2ZMGECXbp0oaCggGXLltXU9EZDdZwzJgOxtKPFOy/y2zGqRahzhmEYjr59+zJixAhmz57NjTfeyJo1ayrM3QrSv39/wInE73//e9LS0gDYuXMnF198cYW5T/XFJZdcwn333cesWbN45JFHKjiYfPbZZzz66KMAXH/99WXHP/nkEwoKCjjuuOMqiFlpaWnZnLbu3buXHb///vs555xz9hLsuXPnsn79enw+X1nLrylSHeF6kb2FqzXwa+/4AiDHO94DOND7+21gW60tbOaUt7jMOcMwQhk/fjyzZ8/mkUceKdsP56KLLuLvf/873377Lb169WLkyJGoKp999hlJSUlcfPHF9T7ek5WVxb/+9S/OOeccrr32Wp599ln2339/1q1bx+eff04gEODWW2+tMFa3YMEC/vSnP5GRkcGwYcPIzs5mz549zJkzh/Xr15OVlcWECRPKyt91113ceOONDBgwgAEDBpCcnMzq1auZPXs2gUCAm266iezs7Hq973hSpXCp6vjQfRFpA8wBZgOXq+rCsPz9gSeAQcChcbO0mRJ0zrCuQsOoSPCLPz8/f6+5W0Fat27N3Llzue2225g+fTrvvvsuHTp04De/+Q1/+9vfeOqppxrAcjjttNOYO3cu99xzD5988glTpkwhPT2d4447jquvvpqTTjqpQvlTTz2VvLw8PvvsM5YvX87s2bNJS0ujW7duXHbZZVx++eW0b9++rPzjjz/O9OnTmTt3Lp9++in5+flkZ2dz6qmncsUVV3DcccfV9y3HFampN42IPA6cA/RS1YijeyKSAfwMvK6qV9TayibM8OHDde7cuTGfv357PodP/ISJvzmA3x7SLY6WGY2RJUuWMGDAgIY2wzDqhJp8vkVknqoOj5QXi1fhqcCnlYkWgKpuBz4FTomhfkTkPBH5XES2i8guEZkrIleKSLXtFZFEETlGRB4Qka9EZL2IFInIWhGZIiJjopw7WUQ0yrY0lvuKBXPOMAzDqEgskTM6ANWZgu0H2ldZKgyvRXcFUAB8DBQDxwCPAceIyFmqWp1v8dFA0GUoF5gH7AYGAmOBsSJyp6reHqWOL4BI/rLrq3Mv8SDFnDMMwzAqEItwrQGOEpG2qrolUgERaQccDayrScUiMhYnWrnAkar6k3e8I64FdwZwFfBwNaoLAFOBh1X187DrnAO8DNwmIp+qamWByp5V1ck1uYd4k5zgjXHZPC7DaDDefPNN3nzzzWqVbdeuXYU4g0b8iUW4XgNuBj4SkWsiiMJInLCkA4/XsO6bvXRCULQAVHWDiFwOzABuEpFHVTWqm52qfgJEXHRGVV8TkWOB3wO/w4lio0RESE30U1BiXoWG0VDMnz9/rxBTldG9e3cTrjomljGuu4G5wGBghoisEpGZ3rYSmAkMBb71ylYLEekCHAQUAXstkKOqM4G1QBZwWAx2h/Odlzb6yQwpiT6LnGEYDcgdd9yBqlZra+pLhjQFatziUtU9nmPDncAluC/+0C//3cCzwK2quqcGVQ/10kWqml9JmW+Azl7Z2TWoOxL7eWm08aqjRORAIA3YAMwCplfV2os3qYl+c84wDMPwiGlZE0+QrheRW3CtpKBwrQXmRRGeaARjGK6MUiYYoLdW8Q5FJAsY7+1OjVI0UsiqxSJybvj8tbokJdFvzhmGYRgetVqPS1ULcJ538SDNS3dHKbPLS9NjvYiIJAAvARnAx6r6ToRi83FeiB/jhLQVMAzX9TkYN743TFXXxmpHTUixFpdhGEYZtV5IUkT64Nzet6jqj7Wpykvren2Bf+Dc61fjHDP2QlX/HnZoN/CuiEzHjeEdhnMkuSrS+SLyR+CPAN261X7ScGqS30I+NSNUtcZLXhhGYyeeS8fEtKyJiCSIyO0isgFYhhv7uSkkf7yIzPbCP1WXnV6aFqVMMG9nlDKVIiIP4zwJc4FjVDW3JuerahEw0ds9KUq5p1V1uKoODw3DEispiT7rKmwm+P1+iouLG9oMw4g7xcXFcVuFucbC5XW1vQf8BcgEllDeWgoyF9cqGVuDqnO8tHuUMsFQxzlRykRERB4ArgE24UTrpypOqYxg1IzOMZ5fY8w5o/mQnp7epNdJMozK2LFjB+npMY/yVCCWFtdVwK9w4z89VHWvVpWq/oATl5pEcgy6pw8SkdRKyhwcVrZaiMi9wHW4FZiPVdXFNTk/jLZeuitqqTiSbM4ZzYY2bdqwbds2Nm/eTFFRUb2uzGsY8UZVKSoqYvPmzWzbtm2vhTFjJZYxrnE4AThbVfOilFtBuct5lajqahH5FucEcRZh63iJyGic92Iu8GV16xWRScCNuCVWjlXV6KutVc3ZXvpNLeupNqmJfouc0UxITk6mW7dubN26lZycHEpL7b0bTRu/3096ejrdunUjOTk5LnXGIlz9gBlViBa4eU8jalj3RNzk43tEZLaqLgcQkQ64pVIAJoXOoxKRibhQUG+o6s2hlYnIncAEIA8nWlW21ERkCE4g3w+Nieh1kV7jbQAP1fDeYsYiZzQvkpOTyc7ObtLrJRlGXRKLcCkuDmBVZOEC5Va/YtUpIvIkcDmwUEQ+ojzIbivgTVyw3VCycWJa4b9cRH4N3OrtLgeursRTa6mqTgrZ7wG8AWwVkR9xsRnTgQOATrh7n6CqH9bk3mqDRc4wDMMoJxbhWgEMFhFfZREkvDGqA3GOGzVCVa8QkVnAlbgI736cQ8RzwJM1iFoR2pk63NsiMRMIFa7vcbEWD8E5igzFifUa4HngcVWdV00b4oJrcZWam7RhGAaxCdfbuDlMNwD3VlJmAtAaeCsWo1T1FeCVapYdT3kUjNDjk4HJMVx7BfA/NT2vLklJ8qMKhSWBsmVODMMwmiuxeBU+iHOQmCgir4jIb7zj7UTkRBF5DrgNF57picoqMapPSoItJmkYhhEkliC7W0XkBFxr6lzgHFxX2sneJrioFKeqakwThY2KpCYFhcscNAzDMGINsrtQRAYCFwEnAr1wY1GrgfeBp1U1WsxBowakJLqGsc3lMgzDiEG4PG+9YlV9H3jS24w6JNUb1zLPQsMwjNjGuN6gkTkv7OsEHTIKSky4DMMwYhGurcDmeBtiVE6ZcFmLyzAMIybh+hqoSdR3o5akWovLMAyjjFiE6x5cINzfx9sYIzIpZWNc5lVoGIYR60KS/wCeFpEzcWNeK4H8SAVV9bMYr2F4lDlnmFehYRhGTMI1AzdvS4Djib50icZ4DSOElCTXMLYJyIZhGLGJymc4QTLqiTLnDBMuwzCMmCJnjKkDO4wopJpwGYZhlBGLc4ZRzyT6fST4xMa4DMMwMOFqMqQk+s2r0DAMg1o6TohIS6APbpHHiAtFmVdhfEjx1uQyDMNo7sQkXCLSB7fY4nFEb7WZV2GcSEn0WeQMwzAMYguy2wWYDbQD1nl1dAC+xLW+2uME60ugOG6WNnNSrcVlGIYBxDbGdRNOtO5U1S64ZUxUVY9Q1Y64uV0rgCKiz/EyakBqkt+iwxuGYRCbcB2PW3frr5EyVXW6V2YE8OfYTTNCSUnwm1ehYRgGsQlXF2C+qgZd3AIAIpIYLKCqPwMzgd/W2kIDgJQkv62AbBiGQWzCVQAUhuzv8tIOYeW2Aj1jMcrYm5QEn01ANgzDIDbhWgt0C9lf7qWHBw+IiABDge2xm2aEkppkXYWGYRgQm6v618CZIpKiqgXAB97xh0RkN7AGuBzYD3g3PmYaqYl+a3EZhmEQW4vrXSAVOAVAVX8C/gl0BqYB84HLcK7wt8bFSsOLnGHCZRiGEUuQ3alAYtjhy4FlwJlAG2ApMFFVF9TaQgMIRs4w5wzDMIy4RLVQ1VLgAW8z6oDURD9FJQFKA4rfFzG6lmEYRrPAguw2EVISbTFJwzAMMOFqMqQm2ZpchmEYEFuswk9qUFxV9ZiaXsPYm5QEJ1zmEm8YRnMnljGuMdUoo7hlTjSG+o0IpJS1uMxBwzCM5k0swnVUJcd9QHfgZGAscA/lc7yMWpKaaF2FhmEYEJs7/MwqikwWkSuAB4EpMVll7EXQOcO6Cg3DaO7UiXOGqj4B5AB31EX9zRFrcRmGYTjq0qtwIW5pEyMOpHjCZdEzDMNo7tSlcGXhQkMZcSAoXBY9wzCM5k6dCJeInItrbS2ti/qbI2XzuKzFZRhGMyeWeVzPRclOA/oDg7z9R2IxytiblARzzjAMw4DY3OHHV6PMTuBvqjo5hvqNCFjkDMMwDEcswnVRlLwi3EKT36hqfmwmGZGwyBmGYRiOWOZxvVAXhhjR8fmEpASfCZdhGM2eRhlkV0TOE5HPRWS7iOwSkbkicqWIVNteEUkUkWNE5AER+UpE1otIkYisFZEpIjKmPuyIJ6mJfgot5JNhGM2cRidcIvI48DIwHPgcmA70BR4DpoiIv5pVjQY+Aq7DhaKaB7wBbMWFpPpURP5WD3bEjZREn83jMgyj2VPX0eHDiRotXkTGAlcAucCRqvqTd7wj8ClwBnAV8HA1rhUApgIPq+rnYdc5BydKt4nIp6r6aR3aETdSE/0UlJhwGYbRvKlNdPhg5Pfw5XgrOx6aVxk3e+mEoFgAqOoGEbkcmAHcJCKPqmrUPjNV/QSIKLKq+pqIHAv8HvgdTozqxI54kpLotxaXYRjNnlijw58OXIvrfnsJF5cQoAdOCA7CtUberG6lItLFO68I+Hd4vqrOFJG1QGfgMGB2DLaH8p2XdmlgO6pNSqLfnDMMw2j2xCJciusmu0FVH4yQ/7CI/Am4F3izGtHkgwz10kVRXOm/wQnGUGovGPt56foGtqPamHOGYRhGbM4ZtwKLKxEtAFT1IWAxcEsN6u3ppSujlFkVVjYmRCSL8onUUxvKjpqSmmQtLsMwjFiEazgu8ntVLAQOrkG9aV66O0qZXV6aXoN6KyAiCbjuzQzgY1V9J952iMgfPdf5uZs2bYrV1L1ISfRZ5AzDMJo9sQhXEs69vCq6U7OuyKAzR1UOHLXlH8AxwGrceFzc7VDVp1V1uKoOb9++fazV7IWNcRmGYcQmXN8DI0TkpMoKiMiJuOjw39eg3p1emhalTDBvZ5QylSIiD+M8CXOBY1Q1tyHsiJWURL+1uAzDaPbEIlz34Volb4jIP0XkaBHp6W1HicizlHsT3l+DenO8NFprrmtY2WojIg8A1wCbcKL1UyVF69SO2pCa6KfAnDMMw2jmxBKr8E0RuQm4G+fgMD6siOAm/96iqm/WoOqge/ogEUmtxKPv4LCy1UJE7sVF0NgCHKuqixvCjtqSal2FhmEYsYV8UtV7cV/ek4FfcHOeioAVwPPAIao6qYZ1rga+xY2hnRWeLyKjcXOucoEvq1uviEwCbgS24UQravdlXdkRD1ISfZQGlOJSa3UZhtF8iTlWoarOV9Xfq+p+qprqbX28Y9/GWO1EL71HRPoED4pIB+AJb3dSaLQKEZkoIktFZCJhiMidwAQgDyda1W0h1diO+iAl0YVH3F1YUp+XNQzDaFTEMgG5zlDVKSLyJHA5sFBEPgKKcV6ArXBjZ4+FnZYN9PPSMkTk17g5ZwDLgatFIkWhYml46zBGO+qc/TtnADB98QbOGt61itKGYRj7JnETLm+pj4uBwbjJu0+pao297lT1ChGZBVyJi/DuB5YCzwFP1qCV0ybk7+HeFomZwF7dmnG0I24c2rMN+3VI419frTThMgyj2SKqNZuu5Dlm3A6cpKozQo6/DxyHc85QYAlwqKpGm8i7zzN8+HCdO3du3Op78cscbn9rEW9eeQRDumbGrV7DMIzGhIjMU9WIDY5YxriOB3bgWirBCxznHV8L3AV8DQzAtcCMOHLG0M60TPLz4pc5DW2KYRhGgxCLcPXBxSoMbaqNxbWyzlXV24GjcV5859XeRCOU9JREfjOsC9MWrGfr7qKGNscwDKPeiUW42rJ3RPWRQK6qzgbw5j7Nxi1zYsSZcYd3p6gkwGvfrG5oUwzDMOqdWIRLgZbBHRHJAPoDX4SV2w5kxmyZUSl9O6ZzWK82vPTVSkoDdR3a0TAMo3ERi3CtAA71vAgBTsE5ZMwKK9ce2FwL24woXHB4D9bm5fPp0o0NbYphGEa9EotwvQ10xMUqvAYXu7AUeCtYQNyEqaE4kTPqgGMHdqRjq2Re/CrasmGGYRj7HrEI1z04V/dTgb8DWcD9qhr6DToS1+IKb4UZcSLR7+O3h3Tjsx83sWJzs55xYBhGM6PGwqWq23GTeS8E/gyMUdWbw4q1BR4GXq21hUalnHdINxJ8wkvW6jIMoxkRU+QMz2vwX1Hy36R8aZMyROQ0YLCq/i2W6xoV6dAqheP3z+L1uau5fExv2qUlN7RJhmEYdU7MQXZj5HTgL/V8zX2aP/1qPwqLA/ztnWgrtRiGYew71LdwGXGmT4d0rjyqD29/v45Plm5oaHMMwzDqHBOufYDLx/SmX8d0bnnjB3YWFDe0OYZhGHWKCdc+QFKCj0ljDyB3RwH3fbisoc0xDMOoU0y49hGGdmvN+BE9+NdXK5mbs7WhzTEMw6gzTLj2IW44rh+dMlKZMHUBhSWlDW2OYRhGnWDCtQ/RMjmB//vNAfy8aTePfPxTQ5tjGIZRJ8RtBWSjcTC6b3vOPKgLj3/6MwDXH9sPn08a2CrDMIz4YcK1DzLxNweQ6Pfx+Kc/88um3Txw9mBaJNmrNgxj38C6CvdBEv0+/u+M/bn15AF8sCiXs5/6ktztBQ1tlmEYRlyob+HaAqyq52s2S0SEP4zqxbMXDGfFpt2c9vgsfli7vaHNMgzDqDX1KlyqeoOq9qzPazZ3jhnQkSmXjyDB5+Psp75kxjJbv8swjKaNqMa2gq6IdAaOAjoBKZUUU1W9M0bb9gmGDx+uc+fObWgz2LijgPHPf8OPG3Zy75kH8pthXRraJMMwjEoRkXmqOjxSXo1H7L1FIv8OXEF5iy3cbU29Ywo0a+FqLHRolcJrlx7Gpf+ax3Wvf8/GnYVcemQv3Os0DMNoOsTianYjcDUQAD4AlgI74mmUUTekpyTy/EUHc8O/FzDp/aVs2FHAbScPNHd5wzCaFLEI10VAMXCMqtoKx02M5AQ/D58zhPZpyTz3xQoWr9vBpaN7MaZvBxMwwzCaBLEIV09glolW08XnE247ZQC92rfk0U9+4uLJc+nRtgUXjujBmQd1IT0lsaFNNAzDqJQaO2eISC7wiaqeVzcm7Vs0FueMyiguDfD+D7k8/8UKvluVR1pyAuMO785lo3uTkWoCZhhGwxDNOSMW4XoFOFhV94uHcfs6jV24Qpm/Oo9nP/+FaQvWk9kikauO6sO4w7uTnOBvaNMMw2hmRBOuWOZx3Qa0F5HbameW0dgY0jWTx84bxrSrR3JA5wzuencJxzwwk7fmryUQiG3ahGEYRryJpcV1ATAUuAb4GngfFw0jEKm8qr5YSxubNE2pxRXOZz9uYuL7S1myfgf9OqZzxVG9OeXATvjNicMwjDom3l2FAcrnaeH9XSmq2qz7mWIWrrzV8OVjcNxd4G+4saZAQHlnwToe+2Q5P23cRY+2Lbh8TG/OGNqFpAQLdWkYRt0Q1wnIwItUIVZGHMhdAHP+AWkdYNT1DWaGzyecNqQzpx7Yif8uzuWxT5czYepCHv7oJ357SDdOH9qZrm1aNJh9hmE0P2IO+WRUj1p1Fb42Dn78EK74Etr2jq9hMaKqfPbTZp6a+TOzf94CwKE92zD2oC6cuH+WudIbhhEX4tpVaNSMWgnXjvXw+KHQaTBc8DY0svBMa7bt4c3v1jL127Ws2Lyb5AQfg7tkMrhrBgd2yWRI10y6tE4tCytVUhogv7gUwATOMIyomHA1ILV2zpj7HEz7E5z2OAz9XfwMiyOqyner85j2/Xq+W72NRet2UFTifHXSk11vdEFJKcWl5Z+1g7q35uQDsjnpgGyyMiqL0WwYRnOlzoRLRAYAfYFW7B1oFzCvwloLVyAAk0+GjYvhqm/cmFcjp6gkwI8bdjJ/dR4/bdiJ3+cjJdFHaqKf1CQ/OwtK+HBRLktzdyICB3dvw4kHZHFk3/b0atfSAv8ahhF/4RKREcDTwIBoxXDLmphXYW3d4Tf9CP84AgacCmc+Fx/DGgHLN+7ivYXreXfBepZt2AlAp4wUjujTjpH7teOQnm3omJ5iMRQNoxkSb3f4/sBcoAUwG8jCxS98FeiDm+PlB94CtqvqRbGb3vSJ2zyuGffAjP+D8/4NfY+rfX2NjJVbdjNr+WZm/bSZ2T9vYXt+MQCJfiErI4XsjFQ6Z6bSs11LDu7RhqHdMklJbNa/iQxjnybewjUZuAC4VFWfEZHngQuCLSuv+/AFnLAdrqo7a2N8UyduwlVSBE+NgsKdcPaL0CXi+9wnKA0oP6zdzoI1eazbXsC6vHzW5xWwNi+fddvzUYUkv4/BXTM4tGdbjurfnmHdWlsXo2HsQ8RbuHKAQlXt5+1XEC7vWAdgOfAPVf1zrIbvC8Q1csa67+D//RZ25sJB4+GY26FFm/jU3UTYvqeYuSu3MmfFVub8soUf1u2gNKD065jO+Yd144yhnc1j0TD2AeItXAXAu6o61tt/FrdGVwtVLQwp9w7QP5ZgvCJyHnA5cCCu23Ep8DzwpKpGDC1VST39gBOAg4HhOEcSAc5S1SlRzpsMXBil6mWq2r86NsQ95FPhTpgxCb56ElIz4dg7Ych5jc5Vvr7YWVDMuwvW89KclfywdgctkvycNqQzQ7pmsLOghB35xezw0vbpyQzpmsngrplkZ6RYC80wGjHxjpyxi4oehMHVj7OBnJDj+UDnmlYuIo8DVwAFwMd4i1YCjwHHiMhZqlpazeouB66tqQ0hfIFrOYazvhZ11o7kdDj+bhh8Lky7Dt66Aua/AmOfhVbZDWZWQ5Geksi5h3TjnIO7smDNdl76aiVvfLeG//f1KsDpeXpyAukpiWzaWUhRqfvd08ETsb4d0+nWpgXd2rage9sW5gxiGE2AWIRrDdA1ZH+plx6FaxUhIonAocCmmlQsImNxopULHKmqP3nHOwKfAmcAVwEPV7PKH4D7cM4k84B/AqNrYNKzqjq5BuXrj6wD4OIP4bsX4YOb4enRcNYL0P3whrasQRARBnutqdtPHUjenmIyWiSSlpRQJkSFJaUsWb+T+au2MX91Ht+v2c7HSzdSGhL5PjnBx4jebTl1cCeOHdjRuh0NoxESi3B9AVwkIq1UdQfwLlAKPCQiKThhuwTogvM0rAk3e+mEoGgBqOoGEbkcmAHcJCKPVqfLUFWfDd3f57qGfD431tXlEHjtfHjhFDhhEhz8h2bbdQiuFRZJcJIT/Azp6iJ6BCkuDbAuL5+VW/awcuseft64i+mLN3Dd69+TlODjqH7tOfnATgzMbkXnzFRSk8yT0TAamliE6z/AccAY4G1VXSsiE3HrdD3mlREgD7ilupWKSBfgIKAI+Hd4vqrOFJG1uO7Hw3Cu+AZAx4FwyafwxqXw3g2w9ls45UFITG1oyxo9iX4f3du2pHvblmXHbj9lIN+tzuOd79fx3sL1fLhoQ1lem5ZJdM5MpVNmCm3TkmnbMok23pbZIonwXsa05AQGd8m07kfDiCM1Fi5V/RjYL+zYX0RkAXAm0AbXffh3Vc2pQdVDvXSRquZXUuYbnHANpX6E6ygRORBIAzYAs4DpNXEQqTdSM+Hc/wcz74GZk2Ddt3DULdD/FNcyM6qNzycc1L01B3VvzW2nDGTBmjxWbd3Dmm35rM3LZ+22fH7ZtJu5OdvYtqeIqtbY7N62BWcP78pZB3WhQ6u9w1uVlAYoVbWVpg2jmsTS4oqIqk4Fptaiip5eujJKmVVhZeuaCyIcWywi56rqwnqyofr4fHDUzW6O1wc3wevjoOMBMGYC9DvZBCwG/D5haLfWDO3WOmJ+aUDZnl/M1t2FbM8vJtxJd9XWPbz2zWru+3AZD07/kWP6d2DUfu1Ysy2fnzftZsXmXazaugcRYWSfdhw7sCPH9O8QUeAMw3DETbjiQJqX7o5SZpeXptexLfNxzhwf44S0FTAMuBsYDHwkIsNUdW2kk0Xkj8AfAbp161bHpkZgv2Oh11Hww1T47F547XdOwPY/w+UHAqClECiFVp2gx0ho26dZj4vFit8nZV2FkRjeow2/GdaFXzbt4rVvVjNl3hr+u3gDSX4fPdq1oE+HNI4blEV+USkfLdnAJ0s3AjC4ayaj92vH0O6tGdIlk9aV1G8YzZGYg+yKSAbwO+BwoD3wsare6+X1A7oDn0fp9guv7xbgLuAlVR1XSZm7gf8FnlbVS2OweQbOqzDqPK4o5ycBM3FjbI+r6lVVnRP3eVw1pbSkXMC2RPLs90jrCN2PgB5HQLfDoX1/8FnXVbwpKgmwaVchWa1S8IeNe6kqP27YxUdLNvDfxRtYuCavrBuyZ7uWDPWWifH7fPh94Pf5SPAJHVol0z+rFb3atyTRb61qY98g3vO4EJETgJeBTLxgukBo62Ool38e8Fo1qw2GhkqLUiaY1yBhpFS1yHNEeQs4qSFsqDH+BBh8Dhx4NhTtBl+CEyTxuxbW1l8g53PI+QJyZsGi/7jzktKg01DX7djlYGi7n5snllzXjd19m6QEH50zIzvNiAj9stLpl5XOlUf1YXdhCQvWbOe71dv4blUen/20mc27CiOeCy6uY+/2afTPSqdz61TapSWHbEmkJvlJ8vtI9PtITPCR5PeRlGBCZzQ9aixcIrI/zrMwAXgC+Iy9xelt3ATk0yLkVUaOl3aPUiY4fywnSpm6JjhvrcaTqxsUEUiO8JugbW+3HTQeVGHbClj9Daz5BtbOhdmPQqCkvHxyK0jPdl2MHQZA1oGQPRja9XUiacSNlskJHN67LYf3blvheCCglASUgCrFpQHWbMtnWe5OlubuZFnuDr5esZUNCworzE+rjK5tUjmgcwaDOmVwQOcMBmS3IrNForXcjEZNLN80/wskA2eo6tsAIlJBnFR1j4gsxY0HVZfvvHSQiKRW0sV4cFjZhiD4LbIraqmmiAi06eW2wee4Y8X5kLsQtuXAjnVu27kOtq+Buc9DifeaElKg4yDX3dj7KNfdGOqOv2cr/PwJ/PgBrP/eteA6DYVOQyB7CKS1d4GEd2+EXRtg5wY3Dpd1IGR2s/G3EHw+IcnrZkxJ9DMgO5EB2a0qlAkElLz8YjbvKmTzzkI27y6ioLiU4tIAxSUBSgLK7sJSftywk4Vrt/PewtwK5yf6hdREPy2SEkhN8uP3CX4Rl/qEjNREjurfgRP3z6JTJS1Iw6grYhGuMcB3QdGKwmpqEKVCVVeLyLc4J4izgAoLUIrIaNyk5lzgy5oYHGfO9tJvGtCG+iMxFboe4rZwSkvcuNn67yF3Aayd52Iozn7ECVm3w5worf4aVn8FGoAW7Vz34+ZlsOzd8rqSW0Hhjr2vAdCiLXQaBp2HuTp7jAJ/NSNaqDr7lk6DxBbQ51cu6kh1hXDHetj8oxPZlFZVl28k+EKcRvp2rLp7d/ueYhat286yDTvZVVDCnuJS8ou8rbiU0oBSEghQGoDSQIB1eQXcOW0xd05bzNBumZy0fzaj+7WnXVoyrVISSAhpsW3eVciS9TtYsn4Hi9ftYFdhKe3Tk8u3tGT6dGhJ7/Zp+16QAKNOiCXIbiHwhqqeG3IsAExW1YtDjk0FTlLVav8cE5EzcZOPc4FRqrrcO94BF/JpIPA/qvpwyDkTcaGg3lDVm/eutUL9M6jCOUNEhuAE8v3QmIgikgBcgwsh5QNOUNUPq7qnBnfOqG8Kd8HKL+CXGfDzp7BpiROKvie4rdOwcrf8gh1O8NZ9B3mrnKild3SOImkd3Mjp+u9g7XduXtqmpU78UltD/5Nh4BnQa/TeIqYKG36AH/4Di95w3Z/idy04cPX3+RX0PhrSs1xXaGkJBIqhpAA2LoX1851du7zJxwmpMOgMGDbOtSbtC5ZfNu3i/R9yeW/hehatq/ijIz0lgcwWiRQWB9i4s3xcLqtVCpktEtm8q5Atu4sqTB9on57MYb3acngv1z3ao20LE7JmTLyjw+cCy1V1ZMixSMI1H8hU1R41rP8JXHDcAuAjyoPstgLeBM4ME5TJuEjuL6jq+LC6huHG4YIMxLnS/wRsDR5U1cNCzjkdeMPL/xEXwiodOADoBASAm4MelFXR7IQrnJJCSEiOT12Fu2DFTFj0Jix7H4p2QkomdNwfine7/KLdLoJ+0U4nVj2PhP1/4yZilxS67srl011asL2SCwm07+dai52GQpuesOw9WDjV1du2Dxx4ruvCTGrpbWmuRZae5VqP1fnCDZRC3konlP5EN4WhiY4Trtyym+9W5ZG3p4hte4rZnl9M3p4i/D4fA7LTGZDdigHZrSpMGygpDbB1dxEbdxbyw9rtfPnLFr78eUuZ0LVukUj/rFYMaa8cHfiKfnmfk9C+F/7hF5HcaVBEOwIBpbAkQEqir2mIXnE+bFvpfrClRp4r2KgoLnA/BIv2uP/rxFSXJqQ6x62E+E3biLdwvQ0cD+wfEgS3gnCJyMHAHOD/qer5MRh8HnAlTiyCy5o8R4RlTaoQrjG4llpUVLXsEy4iPXER5Q/BOYq0xf32XwN8jnODn1fde2n2wlVXFBc48Vn8lmutJaeVC0hSmhOeAadCy3aRzy8tgdzvncj5Ep1w+BJc2rpnZEeWot3uet++CKui9FYnpZU7sKR1cPWK37U0xefq2bQUNv/kWnhBMrrC8Ith2AUV7Q51mtFS5xTTrh8ktah43UAAdq53Yih+F00lJdOlCcnuS3LHOldmx3rYvQkyOkP7AW5cM1w0SwrdWOauDeBPhsQU90WV2KL8WddUHAIBKNrlWrmBUi/1nH/Eh4qPldsKmJuzjaKfZ9I79wOGFn5DkpSyOtCeDrKNZClhrvbj7YTj+Dp1FHsCSewpKmFPUSl7itxv2iS/j7ZpSWUelR3SU+ifnc6BXTIZ1KlVzVfP3r4GVs/xur2/dt3anYZC54PclnWAezalxe657tro0sIdbuy2tNA9z5JC2LHWvfstP0HeatzXi7g6eoxy8yq7H159ISsucF3vGxa5bVuOezct2rg6Ulu7rvv8rbB7M+zZ4tKiXe7zGNx8fve/kJjqPluJ3la0yw0JbPkFtgftrYSkdGjR2nXvp7aBA850yy7FQLyF63jgfWAhcLaqLgsVLhHphXMXHwiMVtVZMVm9j2DCtY+yewsU5Ll/6qLdbivY7onCOvfltGOd+/IKTvjWgPuyTkiB9n3dXLn2/Zxw7FwP3zwDKz5zIrH/b5xABb8od28MM0CgdQ9XJlDivqy2rXRfkJFISKkokuH4k5zDTJuezua8Vc6maCSkOqeaNK97t0Vb1+pMznC/vlNaQWkRbPnZTbvYshy2rqjcxkikZ6ODfsO6bqeysLQ7O7ZtoHPOm/Rf9x/aFqxijy+Ndcm92ZXcgfyU9hSmdqQ4pS2BPXn4d28gOT+XlkWbSS/egi9QTBIlJEgpqb5SkqQUH6X4NIBP3d8AxZJMsS+ZEn8Kpf5UUkp307LILXQR8KdQlDWUxJZt8OfOd+8Z3I+T5HTI31b1PSWled68+0G7/dyPhq0r3LSUNd+UvydfQkVhEZ/7YeVPdu8rIcn7UZNT3g2ekOI+F0V7nC1FYTOH/EmuS75lW9czoOo+l8HPZ2mR+4FTnO8+08X57gdLm96up6FtH2d7citnZ3ArLnCf//ytThj3eOngc+HQGk+5BeIsXF6FDwNX46R3ETAIN49rPW4OVwLwoKreEJPF+xAmXEaN2LjUCdj3rzpRbN0Tuh4K3Q51qwAkJMPGxa7cxsWwaZlrKbXu6b6wWveA1t6Mkvw89+VVkOe+VFIyXSsw2Bps0db9gt641I1FblzqWmst20Nmd9cVmtnNdWOVlkDxHu9LbY9rqZa1LDZ66WavhREmkP4kZ1/bPtC2lxM5X6L3C9+bV4hU/AINBKBDf+elGmkivKqbd/j9q7D15/JWZKgo+hIgLQvSs9C0DhSQzNYCZUu+smlPgC35AQLi5jSq+Muu4ystxFdaQEJpAYmBAorxMz/Qh28D+7FEu1FCAskJPkb2acepvYQx6avJ3LLA3XvLDgRatmcrrVhTlE5hQhqtW6XRulUrWrdKIyEpJXpLtaTQOTmt+soJR/B5BJ9JaVHFTQPuuXYc5LrM2/Sq+LxKi91noHiPawElpzeZ8dm4C5dX6WXA7UBWWNYW4E5VfSSmivcxTLiMmCjc5UQirX1DW1JzSoqcsBVud12WGV3qJwqLqvuS3r3JiXTL9rWOz6mq7CosYdvuYrbsLmTr7iK27i5i0bodTF+8gbV5bjrIkK6ZdG6dyi9e/MmC4r3jcItA25au+zLoTdk+3U0QT/RL2dy8koCiCkO7ZXJIjzYVPDSbE3UiXF7FPmAI0As3FrUa+FpVS6Kd15ww4TKMfRNVZWnuTj5esoHpSzaSt6eIXu1a0qt9Gr3bp9GrfUuSEnxs2llYtm3cWcjmXeX7m3YVUlRS+WITbVsmcfz+WZy0fzaH9dpbxFSVHQUlrN+ez7q8fNblFbA9v5jsjBS6t21BtzYtaZeWVC1HlZJSN7+vxuN/dUSdCZdRNSZchmFURlB4AgHF5xMSvAnexaUBZv20mXcXrueTpRvZU1RKq5QE0lMSvfl0SnGpUlhSGrF1F0rLJD/d2rZkvw5pbuuYRp8O6bRM9vP96vKQYgvW5FEaUIZ3b8OYfu0Z068DfTs23Nw6E64GxITLMIzaUFBcyoxlm5j540aKSpQEn5DgdyKX6PfRoVUy2RmpdPIWOM1ITWRdXgGrtu52K3tv2cOKzbtZvnFXWddmKIl+YVCnDIZ2yyTR7+OzHzexNNc5dXTKSGFo99Z0ykihY6sUsjNSycpIpk+HdDJSqxkEIEbqaoyrK24ybyegssWDVFXvjOkC+wgmXIZhNBZ2F5bw86Zd/LRhF7sKSzigSwYDs/eeHrB+ez4zl21ixrJNLM3dwfrtBRSGdGn6xC29M6pPO0b1bc+Qrplxj28Zb3f4BOAx4A+4yPCEpEG8iQmoqjaODtMGwoTLMIymjqpbMHX99gJytxfw3eo8Zv20ifmr3dI7ackJdEhPpjgQoKTUOZiUlAa4cEQP/udXfWO6ZryXNbkDt0hiCfAeLgrFvhdw1jAMwwDckjuZLZLIbJHEgOxWHNW/A9cd25ft+cV8+fNmZi3fTN6eYhL9vpCuTB8Dw4I/x4tYhGscbpXiI1R1QZztMQzDMJoIGamJnLB/Nifsn12v142lU7IDMNNEyzAMw2gIYhGuVUANYrYYhmEYRvyIRbheBUaLSIQopIZhGIZRt8QiXP8HLAPeFZHY3EUMwzAMI0Zq7JyhqoUichxuFeJFIrISt+RHpOnbqqrH1NJGwzAMwyijxsIlIu2A6biI8IKLU9irkuIWlsMwDMOIK7G4w08CBuO6C/8BLMfmcRmGYRj1RCzCdTJu3a3DVLWytc8NwzAMo06IJeTTLuB9VT2rbkzatxCRTcDKGE9vB2yOozlG7Ni7aBzYe2gc1Md76K6qEReki6XFtQRIr509zYfKHnx1EJG5lcXqMuoXexeNA3sPjYOGfg+xuMM/DowxV3jDMAyjIaixcKnqZODvwAwR+b2IdIm3UYZhGIZRGbG4w5eG7D7tHausuKpqLN2RhuPphjbAKMPeRePA3kPjoEHfQyzOGdHXiQ5DVeO7uphhGIbRrIl5BWTDMAzDaAisNWQYhmE0KUy4GiEicp6IfC4i20Vkl4jMFZErRcTeVxwQkUQROUZEHhCRr0RkvYgUichaEZkiImOqON/eTx0iIv8nIuptN0QpZ+8hzohIqoj8WUS+EZE8EdkjIitE5N8ickQl59T/e1BV2xrRhptuoEA+MA14A9jhHfsP4G9oG5v6BvzKe56KiwIzDXgNWBhy/G/2fhrk3RwMlOCCditwg72Henv2PYGfvGe4AXgLeB34GigCbm0s76HBH5ZtFT4EY0O+TPcLOd4RWOzlXdvQdjb1DTgamAKMipB3jvfFqcBR9n7q9b0kA4uAtd4XYEThsvdQJ8++JS7urAJ/AxLD8tsCfRvLe2jwB2ZbhQ/CXO9lXxAhb3TIh8TX0LbuyxvwrPes/2nvp16f+z3eMzwVmBxFuOw9xP/ZT/Se2ws1OKfB3oP1BTcSvIncB+Ga5P8Oz1fVmbhfolnAYfVrXbPjOy8tm1xv76duEZFDgeuBV1T1nSjl7D3EGRFJAi7xdidV85wGfQ8mXI2HoV66SFXzKynzTVhZo27Yz0vXhxyz91NHiEgK8AKwFbi2iuL2HuLPQbiuwNWqukRERngOMk+JyF9F5PAI5zToe7CoFo2Hnl4aLZL8qrCyRpwRkSxgvLc7NSTL3k/dcTfQDzhXVauKOG7vIf4c4KU/ichk4MKw/NtFZCowLkSkGvQ9WIur8ZDmpbujlAku2GnR+esAEUkAXgIygI/Duqzs/dQBIjIC+B/gTVV9rRqn2HuIP2289EjgAuB+oA/QGjgN1+U3FudBGKRB34MJV+MhGPDRQpk0HP8AjgFWA78Ly7P3E2dEJBV4Huc+fUV1T/NSew/xI6gDCTiHpBtV9WdVzVPVt4HTcc/7QhHp5ZVt0PdgwtV42OmlaVHKBPN2RiljxICIPAz8HsgFjlHV3LAi9n7iz/8BfYHrVHV9VYU97D3En9Dn9Ex4pqrOBebh9GJM2DkN8h5sjKvxkOOl3aOU6RpW1ogDIvIAcA2wCSdaP0UoluOl9n7ixxm4icYXikj4uEp/L71cRE4BlqvqH7D3UBfkhPy9opIyK4DhOC/B0HMa5D2YcDUegi7Yg0QktRJPnYPDyhq1RETuBa4DtgDHquriSora+6kbfLg5P5XRy9syvX17D/Hn25C/2+J+wIXTzkuD41YN+h6sq7CRoKqrcR+gJOCs8HwRGY2bV5QLfFm/1u2biMgk4EZgG060vq+srL2f+KOqPVRVIm0493iAG71jQ7xz7D3EGVVdC8zxdo8JzxeR1sAwb3eud06DvgcTrsbFRC+9R0T6BA+KSAfgCW93kqrWaE00Y29E5E5gApCHE63q/Cq099M4sPcQf+720ttFZEjwoDfH7kmcp+08KopQg70HW4+rkSEiTwCXAwXAR0Ax7ldQK+BN4ExVLa20AqNKROTXuACi4H5BLqqk6FJVrRBJwN5P/RAyn+hGVb0/Qr69hzgjIvcBN+CiYczBdZ8fAnTCucQfFT7+22DvoaFjZNkWMQbYecAXODfh3bhfOldisdfi9XzHUx4FPto2w95Pg72jyUSJDm/voc6e+xnAJ7ju80JctPgHgPaN6T1Yi8swDMNoUtgYl2EYhtGkMOEyDMMwmhQmXIZhGEaTwoTLMAzDaFKYcBmGYRhNChMuwzAMo0lhwmUYhmE0KUy4DKMeEJEcEdFqbGMa2tbqICJ3ePbe0dC2GM0Piw5vGPXLh7jAo5URLc8wDEy4DKO+maSqMxraCMNoylhXoWEYhtGkMOEyjEaIiPTwxpByRCRBRG4SkSUiUiAiG0TkBRHpFuX8QSLyooisFpFCEdksIu+JyIlVXPd4EfmPiKwTkSIRyRWRL0RkgoikVnJORxF5SkTWeNdaISKTvCUxwsv6ReQyEZktItu9a2wQkW9F5AERaV/zp2U0N0y4DKPx8xrwV2AVbqmIIuAC4BsR6Rde2Fu2ZR4wDtgOTAUWA8cD73lrkYWfIyLyJPABLkL4Wu+873FLsE8COkawrat3rVNwazXNADrg1jp7PUL5f+LWdxqCWzpjineNDNxK1L2jPgnDAFvWxDbb6mMDcnDLdIypZvkelC+vsgEYGJKXBPzLy/s67LwsnFgpcF1Y3hjcshMKHB+W9yfveC5wWFieAEcBGSHH7gix7xkgKSRvALDTyzsi5Hh379gqoGOEex4CdGjod2Vb49+sxWUY9cunUVzh8yo5505VXRzcUdUi4CqcQB0sIkeElL0Et4jfbFV9MLQSdU4hj3m7NwSPi0gC8L/e7nhV/SrsPFXVT1V1ewTbVgPXeDYFyy/BCStUXAq+g5d+q6obwitS1fmqujHCNQyjAuZVaBj1SzR3+D2VHH8p/ICqbheRacD5uJbUF17WaC+dXEldzwF/BkaKiF/d6rTDgXbAGlX9oKobCOMTVc2PcHypl3YKO7YTOFlE/hd4WVVX1vB6hmHCZRj1TE3d4fNUNa+SvBwv7RJyrLOXrqjknBVAAEgB2gIbcV14AMtqYFeQVZUc3+GlZQ4aqrpTRC7GiefdwN0ishY3NvYu8KqqFsRgg9HMsK5Cw2j6hC5jLhGO1SWBmhRW1SlAN2A8TsB2AWcCzwNLRaRrvA009j1MuAyjcZMpIhmV5PXw0nUhx9Z4aa8o5/iAAmCrdyzYXbeXh2JdoKp5qvqCqv5eVfsDfYBPcS2/e+rDBqNpY8JlGI2f88MPeGJ2irc7IyRrppdeUEldF3npLFUt8f6eB2wGuojI8bUzteao6s+4rkOAwfV9faPpYcJlGI2f20VkQHBHRBKBh3Fzn+ap6qyQss/gHCBGisg1oZWIyJHA1d7uA8HjqloMTPR2nxeRQ8LOExEZE6XlVy1EZKiInFPJROZTvdScNYwqMecMw6hfbhKR8VHyX1HV/4bsr8K1iOaLyCc4F/jDceNEmwlrWalqroiMw01aflhE/gD8gPPuG4X7sXpXBO/Bh3Dzr/4AfCUic4HlQBtgIG6icU/v+rHSHXgV2CMi3+Jc6ZOAobiuzZ3A7bWo32gmmHAZRv1SVVfcfCBUuBQ4G7gJFwmjO85j7yXgNlXNCa9AVd8SkeG46BVH45wfdnr1Pqqq70U4R4FLROQt4DLgENyE4K3AT8Cj1D5y/VfAzTiX/f7AQbgoIKtxLcBHzT3eqA7iPq+GYTQmRKQHznV9par2aFhrDKNxYWNchmEYRpPChMswDMNoUphwGYZhGE0KG+MyDMMwmhTW4jIMwzCaFCZchmEYRpPChMswDMNoUphwGYZhGE0KEy7DMAyjSfH/AZuX4Z4JxUQcAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:AutoGraph could not transform <function Model.make_predict_function.<locals>.predict_function at 0x1a549625f0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <function Model.make_predict_function.<locals>.predict_function at 0x1a549625f0> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: 'arguments' object has no attribute 'posonlyargs'\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "CondDensityEstimator training complete.\n",
      "Beginning CauseClusterer training...\n",
      "CauseClusterer training complete.\n",
      "Experiment training complete.\n"
     ]
    }
   ],
   "source": [
    "# so our optimized model becomes:\n",
    "\n",
    "def build_optimized_model():\n",
    "\n",
    "    # vary number of layers\n",
    "    n_layers = study.best_trial.params['n_layers']\n",
    "    model = Sequential()\n",
    "    for i in range(n_layers):\n",
    "        # vary size of each layer\n",
    "        num_hidden = study.best_trial.params[f'n_units_l{i}']\n",
    "        model.add(Dense(num_hidden, activation=\"relu\"))\n",
    "\n",
    "        # vary amount of dropout for each layer\n",
    "        dropout = study.best_trial.params[f'dropout_l{i}']\n",
    "        model.add(Dropout(rate=dropout))\n",
    "\n",
    "    # add output layer with size of target    \n",
    "    model.add(Dense(Y.shape[1], activation=\"linear\"))\n",
    "    return model\n",
    "\n",
    "data_info = {'X_dims': X.shape, 'Y_dims': Y.shape, 'Y_type': 'categorical'}\n",
    "\n",
    "CDE_params = {\n",
    "    'model'        : 'CondExpDIY',\n",
    "    'model_params' : {\n",
    "        'build_network'   : build_optimized_model,\n",
    "        'batch_size'      : study.best_trial.params['batch_size'],\n",
    "        'n_epochs'        : 100,\n",
    "        'optimizer'       : 'adam',\n",
    "        'opt_config'      : {'lr' : study.best_trial.params['lr']},\n",
    "        'loss'            : 'mean_squared_error',\n",
    "        'best'            : True, \n",
    "        'verbose'         : 0,\n",
    "        'early_stopping'  : True\n",
    "    }\n",
    "}\n",
    "\n",
    "cause_cluster_params = {'model' : 'KMeans', \n",
    "                        'model_params' : {'n_clusters' : 4}, \n",
    "                        'verbose' : 0}\n",
    "\n",
    "block_names = ['CondDensityEstimator', 'CauseClusterer']\n",
    "block_params = [CDE_params, cause_cluster_params]\n",
    "\n",
    "my_exp = Experiment(X_train=X, Y_train=Y, data_info=data_info, \n",
    "                    in_sample_idx=in_sample_idx, out_sample_idx=out_sample_idx,\n",
    "                    block_names=block_names, block_params=block_params, \n",
    "                    results_path='demo_results')\n",
    "\n",
    "train_results = my_exp.train()     "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualize the conditional probability learned by CFL with the optimized hyperparameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(<Figure size 432x288 with 1 Axes>,\n",
       " <AxesSubplot:title={'center':'Sample of predicted P(Y|X) values after CDE training\\nColored by ground truth'}, xlabel='Sample', ylabel='Expectation of Target'>)"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from cfl.visualization.cde_diagnostic import pyx_scatter\n",
    "pyx_scatter(my_exp, vb_data.getGroundTruth())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualize the macrovariable states that CFL finds with this set of hyperparameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 576x864 with 24 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import visual_bars.visual_bars_vis as vis\n",
    "\n",
    "vis.viewImagesAndLabels(X, im_shape=im_shape, n_examples=6, \n",
    "                        x_lbls=train_results['CauseClusterer']['x_lbls'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "a3e2eae926291f18d1afe7b431ab5303b6e9ec8df44d60cf0ff75f0c781aceac"
  },
  "kernelspec": {
   "display_name": "Python 3.7.9 64-bit ('cfl_env': conda)",
   "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": "3.7.9"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}