Skip to content

Instantly share code, notes, and snippets.

@agrhn

agrhn/autoencoder_tensorflow.ipynb

Created January 24, 2021 19:53
Show Gist options

  • Select an option

    Learn more about clone URLs
  • Save agrhn/575c63da0c8174714f6759ee305af66f to your computer and use it in GitHub Desktop.

Select an option

Learn more about clone URLs
Save agrhn/575c63da0c8174714f6759ee305af66f to your computer and use it in GitHub Desktop.
Download ZIP
Autoencoder_TensorFlow.ipynb
Display the source blob
Display the rendered blob
Raw
autoencoder_tensorflow.ipynb
{
"nbformat": 4,
"nbformat_minor": 0,
"metadata": {
"colab": {
"name": "Autoencoder_TensorFlow.ipynb",
"provenance": [],
"collapsed_sections": [],
"authorship_tag": "ABX9TyPVHfOfCW2PjQmruCiyyhxr",
"include_colab_link": true
},
"kernelspec": {
"name": "python3",
"display_name": "Python 3"
},
"accelerator": "GPU"
},
"cells": [
{
"cell_type": "markdown",
"metadata": {
"id": "view-in-github",
"colab_type": "text"
},
"source": [
"<a href=\"https://colab.research.google.com/gist/mrgrhn/575c63da0c8174714f6759ee305af66f/autoencoder_tensorflow.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
]
},
{
"cell_type": "code",
"metadata": {
"id": "bj7Q8dC7eC9i"
},
"source": [
"import tensorflow as tf\r\n",
"import matplotlib.pyplot as plt\r\n",
"from tensorflow.keras import datasets, layers, models, losses, Model\r\n",
"from random import randint\r\n",
"import numpy as np"
],
"execution_count": 1,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "FTVtr6gheH4d"
},
"source": [
"(x_train, _), (x_test, _)=tf.keras.datasets.mnist.load_data()"
],
"execution_count": 2,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "C0A0gF3dVN2G",
"outputId": "9315d09b-0ca0-4f1d-c0b7-1de08037416c"
},
"source": [
"hidden_size = 100\r\n",
"latent_size = 20\r\n",
"\r\n",
"input_layer = layers.Input(shape = x_train.shape[1:])\r\n",
"flattened = layers.Flatten()(input_layer)\r\n",
"hidden = layers.Dense(hidden_size, activation = 'relu')(flattened)\r\n",
"latent = layers.Dense(latent_size, activation = 'relu')(hidden)\r\n",
"encoder = Model(inputs = input_layer, outputs = latent, name = 'encoder')\r\n",
"encoder.summary()"
],
"execution_count": 3,
"outputs": [
{
"output_type": "stream",
"text": [
"Model: \"encoder\"\n",
"_________________________________________________________________\n",
"Layer (type) Output Shape Param # \n",
"=================================================================\n",
"input_1 (InputLayer) [(None, 28, 28)] 0 \n",
"_________________________________________________________________\n",
"flatten (Flatten) (None, 784) 0 \n",
"_________________________________________________________________\n",
"dense (Dense) (None, 100) 78500 \n",
"_________________________________________________________________\n",
"dense_1 (Dense) (None, 20) 2020 \n",
"=================================================================\n",
"Total params: 80,520\n",
"Trainable params: 80,520\n",
"Non-trainable params: 0\n",
"_________________________________________________________________\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "code",
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "rojD--dpW5HW",
"outputId": "408b98b3-30db-414e-c16a-36d3f5a5ea6d"
},
"source": [
"input_layer_decoder = layers.Input(shape = encoder.output.shape)\r\n",
"upsampled = layers.Dense(hidden_size, activation = 'relu')(input_layer_decoder)\r\n",
"upsampled = layers.Dense(encoder.layers[1].output_shape[-1], activation = 'relu')(upsampled)\r\n",
"constructed = layers.Reshape(x_train.shape[1:])(upsampled)\r\n",
"decoder = Model(inputs = input_layer_decoder, outputs = constructed, name= 'decoder')\r\n",
"decoder.summary()"
],
"execution_count": 4,
"outputs": [
{
"output_type": "stream",
"text": [
"Model: \"decoder\"\n",
"_________________________________________________________________\n",
"Layer (type) Output Shape Param # \n",
"=================================================================\n",
"input_2 (InputLayer) [(None, None, 20)] 0 \n",
"_________________________________________________________________\n",
"dense_2 (Dense) (None, None, 100) 2100 \n",
"_________________________________________________________________\n",
"dense_3 (Dense) (None, None, 784) 79184 \n",
"_________________________________________________________________\n",
"reshape (Reshape) (None, 28, 28) 0 \n",
"=================================================================\n",
"Total params: 81,284\n",
"Trainable params: 81,284\n",
"Non-trainable params: 0\n",
"_________________________________________________________________\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "code",
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "4f--N3yLZlBE",
"outputId": "2c8449c6-d5e3-4b5a-d347-7370043fc322"
},
"source": [
"autoencoder = Model(inputs = encoder.input, outputs = decoder(encoder.output))\r\n",
"autoencoder.summary()"
],
"execution_count": 5,
"outputs": [
{
"output_type": "stream",
"text": [
"WARNING:tensorflow:Model was constructed with shape (None, None, 20) for input KerasTensor(type_spec=TensorSpec(shape=(None, None, 20), dtype=tf.float32, name='input_2'), name='input_2', description=\"created by layer 'input_2'\"), but it was called on an input with incompatible shape (None, 20).\n",
"Model: \"model\"\n",
"_________________________________________________________________\n",
"Layer (type) Output Shape Param # \n",
"=================================================================\n",
"input_1 (InputLayer) [(None, 28, 28)] 0 \n",
"_________________________________________________________________\n",
"flatten (Flatten) (None, 784) 0 \n",
"_________________________________________________________________\n",
"dense (Dense) (None, 100) 78500 \n",
"_________________________________________________________________\n",
"dense_1 (Dense) (None, 20) 2020 \n",
"_________________________________________________________________\n",
"decoder (Functional) (None, 28, 28) 81284 \n",
"=================================================================\n",
"Total params: 161,804\n",
"Trainable params: 161,804\n",
"Non-trainable params: 0\n",
"_________________________________________________________________\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "code",
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "aYH9867pfPC1",
"outputId": "cf83e527-7ea0-41d1-bb27-6d4fbbd64653"
},
"source": [
"autoencoder.compile(optimizer='adam', loss=losses.MeanSquaredError())\r\n",
"history = autoencoder.fit(x_train, x_train, epochs=50, batch_size=64, validation_data = (x_test, x_test))"
],
"execution_count": 6,
"outputs": [
{
"output_type": "stream",
"text": [
"Epoch 1/50\n",
"WARNING:tensorflow:Model was constructed with shape (None, None, 20) for input KerasTensor(type_spec=TensorSpec(shape=(None, None, 20), dtype=tf.float32, name='input_2'), name='input_2', description=\"created by layer 'input_2'\"), but it was called on an input with incompatible shape (None, 20).\n",
"WARNING:tensorflow:Model was constructed with shape (None, None, 20) for input KerasTensor(type_spec=TensorSpec(shape=(None, None, 20), dtype=tf.float32, name='input_2'), name='input_2', description=\"created by layer 'input_2'\"), but it was called on an input with incompatible shape (None, 20).\n",
"936/938 [============================>.] - ETA: 0s - loss: 3087.8979WARNING:tensorflow:Model was constructed with shape (None, None, 20) for input KerasTensor(type_spec=TensorSpec(shape=(None, None, 20), dtype=tf.float32, name='input_2'), name='input_2', description=\"created by layer 'input_2'\"), but it was called on an input with incompatible shape (None, 20).\n",
"938/938 [==============================] - 3s 2ms/step - loss: 3085.7667 - val_loss: 1981.6154\n",
"Epoch 2/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1917.1781 - val_loss: 1790.9906\n",
"Epoch 3/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1792.8246 - val_loss: 1738.5490\n",
"Epoch 4/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1745.4821 - val_loss: 1706.6228\n",
"Epoch 5/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1717.2978 - val_loss: 1680.4696\n",
"Epoch 6/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1692.8290 - val_loss: 1666.8123\n",
"Epoch 7/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1672.1409 - val_loss: 1651.9363\n",
"Epoch 8/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1660.4536 - val_loss: 1636.8737\n",
"Epoch 9/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1648.2435 - val_loss: 1622.4740\n",
"Epoch 10/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1637.0548 - val_loss: 1596.4133\n",
"Epoch 11/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1595.9997 - val_loss: 1584.0935\n",
"Epoch 12/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1590.5362 - val_loss: 1576.9849\n",
"Epoch 13/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1587.5259 - val_loss: 1558.0955\n",
"Epoch 14/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1555.8398 - val_loss: 1547.5657\n",
"Epoch 15/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1558.6432 - val_loss: 1544.5577\n",
"Epoch 16/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1547.6599 - val_loss: 1541.0756\n",
"Epoch 17/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1545.8276 - val_loss: 1533.2150\n",
"Epoch 18/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1543.5184 - val_loss: 1530.2543\n",
"Epoch 19/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1533.9707 - val_loss: 1524.8423\n",
"Epoch 20/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1525.1622 - val_loss: 1517.8408\n",
"Epoch 21/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1525.7032 - val_loss: 1519.9493\n",
"Epoch 22/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1522.4636 - val_loss: 1507.9938\n",
"Epoch 23/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1519.2834 - val_loss: 1497.2184\n",
"Epoch 24/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1504.4597 - val_loss: 1496.0609\n",
"Epoch 25/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1499.6125 - val_loss: 1492.7795\n",
"Epoch 26/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1497.3636 - val_loss: 1488.9916\n",
"Epoch 27/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1490.1915 - val_loss: 1497.2924\n",
"Epoch 28/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1489.1811 - val_loss: 1483.2697\n",
"Epoch 29/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1482.8142 - val_loss: 1485.7706\n",
"Epoch 30/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1481.0900 - val_loss: 1475.7570\n",
"Epoch 31/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1479.1094 - val_loss: 1482.9922\n",
"Epoch 32/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1476.7044 - val_loss: 1478.1715\n",
"Epoch 33/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1479.9351 - val_loss: 1472.2162\n",
"Epoch 34/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1473.5275 - val_loss: 1479.4594\n",
"Epoch 35/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1476.8302 - val_loss: 1472.4565\n",
"Epoch 36/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1468.4903 - val_loss: 1463.4948\n",
"Epoch 37/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1465.0616 - val_loss: 1469.3004\n",
"Epoch 38/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1469.4806 - val_loss: 1469.9330\n",
"Epoch 39/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1463.6156 - val_loss: 1463.9084\n",
"Epoch 40/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1463.5214 - val_loss: 1466.7897\n",
"Epoch 41/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1461.9914 - val_loss: 1457.1173\n",
"Epoch 42/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1462.0404 - val_loss: 1463.0872\n",
"Epoch 43/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1462.4493 - val_loss: 1453.1805\n",
"Epoch 44/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1457.9046 - val_loss: 1461.2170\n",
"Epoch 45/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1451.4781 - val_loss: 1455.3942\n",
"Epoch 46/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1450.4941 - val_loss: 1460.7539\n",
"Epoch 47/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1456.5284 - val_loss: 1454.6064\n",
"Epoch 48/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1454.4282 - val_loss: 1456.6366\n",
"Epoch 49/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1447.7401 - val_loss: 1457.3942\n",
"Epoch 50/50\n",
"938/938 [==============================] - 2s 2ms/step - loss: 1456.5116 - val_loss: 1454.8712\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "code",
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/",
"height": 910
},
"id": "GxoP7pe9gr3q",
"outputId": "c2b88c08-3fed-4495-9080-eb36a835bf1c"
},
"source": [
"fig, axs = plt.subplots(figsize=(15,15))\r\n",
"\r\n",
"axs.plot(history.history['loss'])\r\n",
"axs.plot(history.history['val_loss'])\r\n",
"axs.title.set_text('Training Loss vs Validation Loss')\r\n",
"axs.set_xlabel('Epochs')\r\n",
"axs.set_ylabel('Loss')\r\n",
"axs.legend(['Train','Val'])"
],
"execution_count": 7,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"<matplotlib.legend.Legend at 0x7fd78602c898>"
]
},
"metadata": {
"tags": []
},
"execution_count": 7
},
{
"output_type": "display_data",
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 1080x1080 with 1 Axes>"
]
},
"metadata": {
"tags": [],
"needs_background": "light"
}
}
]
},
{
"cell_type": "code",
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/",
"height": 863
},
"id": "o9sziaCdzLWM",
"outputId": "a02f4609-14ce-4e55-8e05-8fd0bcaad94a"
},
"source": [
"fig, axs = plt.subplots(3,2,figsize=(10,15))\r\n",
"for i in range(3):\r\n",
" sample1 = x_train[randint(0,x_train.shape[0])]\r\n",
" axs[i][0].imshow(sample1, cmap = 'gray')\r\n",
" axs[i][1].imshow(autoencoder(np.expand_dims(sample1,0))[0], cmap = 'gray')"
],
"execution_count": 22,
"outputs": [
{
"output_type": "display_data",
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 720x1080 with 6 Axes>"
]
},
"metadata": {
"tags": [],
"needs_background": "light"
}
}
]
},
{
"cell_type": "code",
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/",
"height": 587
},
"id": "sLEyKWkplFkv",
"outputId": "696b9750-41f6-4923-aeff-83a9081f6ea7"
},
"source": [
"sample1_idx = randint(0,x_train.shape[0])\r\n",
"sample1 = x_train[sample1_idx]\r\n",
"\r\n",
"sample2_idx = randint(0,x_train.shape[0])\r\n",
"sample2 = x_train[sample2_idx]\r\n",
"\r\n",
"latent1 = encoder(np.expand_dims(sample1,0))\r\n",
"latent2 = encoder(np.expand_dims(sample2,0))\r\n",
"\r\n",
"weights = np.arange(0.05,0.95,0.15)\r\n",
"fig, axs = plt.subplots(2,4,figsize=(20,10))\r\n",
"\r\n",
"axs = axs.ravel()\r\n",
"\r\n",
"axs[0].imshow(sample2, cmap = 'gray')\r\n",
"axs[-1].imshow(sample1, cmap = 'gray')\r\n",
"for i in range(6):\r\n",
" latent = latent1*weights[i] + latent2*(1-weights[i])\r\n",
" constructed = decoder(latent)\r\n",
" axs[i+1].imshow(constructed[0], cmap='gray')\r\n"
],
"execution_count": 11,
"outputs": [
{
"output_type": "display_data",
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAABH4AAAI6CAYAAABRrI9ZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzdedTddX0n8M8XsgBZIAESwpKErViWYYsUUZS6tICtgDootS2tnaGeUz1tdZyqraPTaT0zpY5Ly7FNyxIZRRgpi1V0BKRULELCsATCTiAJIWFtElmyfecPHjoB8vw+N8/6e355vc7h5Mn9vP3ejzfcT+7z4XfvU2qtAQAAAED37DDaDQAAAAAwPCx+AAAAADrK4gcAAACgoyx+AAAAADrK4gcAAACgoyx+AAAAADpq3EjeWSnFz46Hbniq1rrnaDcxUGYRdIZZBLSBWQS0Qb+zaFBX/JRSTi6l3FdKebCU8qnBnAWMKY+OdgNbMotgu9WqWRRhHsF2yiwC2qDfWTTgxU8pZceIOC8iTomIQyPirFLKoQM9D2AgzCKgLcwjoA3MIuC1BnPFz3ER8WCt9eFa6/qI+FZEnDY0bQH0zCwC2sI8AtrALAJeZTCLn30iYtkWv1/ed9urlFLOKaUsLKUsHMR9AfTHLALaIp1HZhEwAswi4FWG/cOda63zI2J+hA8OA0aPWQS0gVkEtIFZBNuXwVzxsyIi9tvi9/v23QYwkswioC3MI6ANzCLgVQaz+Lk1Ig4upexfSpkQER+MiKuHpi2AnplFQFuYR0AbmEXAqwz4rV611o2llI9GxA8iYseIuKDWeveQdQbQA7MIaAvzCGgDswh4rVLryL2l0/tHoTMW1VrnjXYTA2UWQWeYRUAbmEVAG/Q7iwbzVi8AAAAAWsziBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjxo12A3TLv/t3/y7NnHrqqWnmoIMOSjOzZ89urK9atSo94x/+4R/SzBVXXJFmgJG19957N9bnzZuXnrHffvulmT333DPNTJo0qbH+3HPPpWc88MADaeayyy5LM8DImjt3bmP9rW9966DPiIjYfffd08wOOzT/99yHH344PeOqq65KM72cA4ysn/u5n2usv+Md70jPyF5bRUSsWbMmzSxcuLCx/qMf/Sg9g6Hnih8AAACAjhrUFT+llKURsTYiNkXExlpr/p9YAYaBeQS0gVkEtIFZBGxpKN7q9Yu11qeG4ByAwTKPgDYwi4A2MIuAiPBWLwAAAIDOGuzip0bE/ymlLCqlnLO1QCnlnFLKwlJK86c8AQxO4zwyi4ARYhYBbWAWAf9msG/1ekutdUUpZUZE/LCUcm+t9cYtA7XW+RExPyKilFIHeX8A/WmcR2YRMELMIqANzCLg3wzqip9a64q+X1dHxBURcdxQNAWwrcwjoA3MIqANzCJgSwNe/JRSJpVSprzydUT8UkQsHqrGAHplHgFtYBYBbWAWAa81mLd6zYyIK0opr5zzzVrr94ekK4bcrFmz0sxv/MZvpJmzzjqrsX7QQQelZ0yaNCnN1DoyV5z+2q/9Wpq54IILGut/9Ed/lJ7xzDPP9NwTA2IejRGHH354mvnlX/7lNPOmN72psT5Us2jjxo1pZvz48YPuZeHC/CMWLrvsssb6XnvtlZ7xxBNPpBkGxSwaI+bNy3+y9amnnppm3vzmNzfW3/CGN6RnTJ06Nc308rpol112aazvtNNO6Rk33HBDmjnppJPSDKPOLBojTjjhhDTzlre8ZdCZo48+Oj1j5syZaaaX72mWLVvWWH/jG9+YnnHsscemmUWLFqUZ/r8BL35qrQ9HxJFD2AvAgJhHQBuYRUAbmEXAa/lx7gAAAAAdZfEDAAAA0FEWPwAAAAAdZfEDAAAA0FEWPwAAAAAdZfEDAAAA0FEWPwAAAAAdNW60G9jeTZkyJc28/e1vTzMzZ85srJ977rnpGZMnT04z26Pf+Z3faaw//fTT6Rmf+tSnhqodGDWzZ89OM2eddVZj/bjjjkvPOPjgg9PMtGnTGuvjx49Pz3jyySfTzJo1a9LMpEmTGuv33XdfesYhhxySZm666abG+pVXXpme0cvfBdB2xxxzTJr5rd/6rcb6L/zCL6RnHHDAAWlm1113baz3Mot6eR3x0ksvpZnsdVytNT2jlJJmvvrVrzbW//AP/zA9Y9OmTWkG2u70009PM+95z3sa67285tlrr73STDaL9txzz/SMDRs2pJnsNU9E/prmoYceSs848MAD0wzbxhU/AAAAAB1l8QMAAADQURY/AAAAAB1l8QMAAADQURY/AAAAAB1l8QMAAADQURY/AAAAAB1Vaq0jd2eljNydjRFf/vKX08zHPvaxNFNKaayP1J/zo48+mmauvfbaNHP55ZenmVWrVjXWP/e5z6VnvOc970kz2WO7adOm9Ixf//VfTzOXXnppmmmRRbXWeaPdxECZRa93zDHHpJk/+IM/SDNHH310Y33ChAnpGatXr04zL730UmP9ne98Z3pGLzPimWeeSTPZjJg6dWp6xiGHHJJmDj300Mb6Y489lp4xf/78NHPNNdekmRYxizrmrW99a5r54z/+40Gf08ssWrFiRZp54YUXGuu9PLfPO++8NLN27do0M2nSpMb6TjvtlJ4xefLkNPPiiy821q+++ur0jCuuuCLNjDFmUcf8yq/8Spr5wz/8wzTz5je/ubHey/dojzzySJrZsGFDY/3II49Mz/j+97+fZp5//vk0k31v1MsZ9957b5r55je/2Vjv5fvSDup3FrniBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOqrUWkfuzkoZuTtriTlz5jTW77zzzvSMyZMnp5lSSmN9qP6c3/3udzfWf/KTn6RnrFmzZkh6yey8885p5k/+5E/SzKc//enGei+P7e23355mzj777Mb64sWL0zNG0KJa67zRbmKgtsdZtOeeezbWzzvvvPSMY445Js1s3LixsX7rrbemZ9xzzz1p5vHHH2+s9/J8eeihh9LMc889l2aGwr//9/8+zZxxxhmN9b322is947bbbkszf/3Xf91YX7p0aXrGCDKLOubSSy9NM+985zvTTPba6ZZbbknPuOuuu9JM9nwYqpm3atWqNJPZcccd08zJJ5+cZo4//vjGei+vi37wgx+kmZtuuinNtIhZ1DFf+MIX0syZZ56ZZmbOnNlY7+Xv5V6+X3zyyScb67383b1s2bI008trp+XLlzfWe/nedu+9904zL774YmO9Za9XRkq/s8gVPwAAAAAdZfEDAAAA0FEWPwAAAAAdZfEDAAAA0FEWPwAAAAAdZfEDAAAA0FEWPwAAAAAdZfEDAAAA0FHjRruBsWz8+PFp5rOf/WxjfcqUKUPSSymlsX7TTTelZ3zoQx9KM4899ljPPY22F154Ic388R//cZqZOXNmY/3DH/5wesbRRx+dZs4+++zG+ic/+cn0DOhP9u/p4Ycfnp6xZs2aNPMv//IvjfWrr746PeO2225LM08++WSaGUt+9KMfpZn//b//d2P9uuuuS8846KCD0syb3vSmxvrSpUvTM6A/v/zLv9xY33XXXdMzHn744UFnvve976VnLFy4MM3cfffdaaYtNm3alGZ+8pOfpJnvfve7jfVLLrkkPWPy5MlpJnsd18vfFTBQS5YsSTO9fH+18847N9b/+Z//OT1j0aJFaeaee+5prD/33HPpGSOll9eTvWQyp5xySppZu3Ztmrnrrrsa6//6r//ac0+jKb3ip5RyQSlldSll8Ra3TS+l/LCU8kDfr9OGt00A8whoB7MIaAOzCOhVL2/1uigiTn7NbZ+KiOtqrQdHxHV9vwcYbheFeQSMvovCLAJG30VhFgE9SBc/tdYbI+KZ19x8WkQs6Pt6QUScPsR9AbyOeQS0gVkEtIFZBPRqoB/uPLPWurLv6yciovlDUACGj3kEtIFZBLSBWQS8zqA/3LnWWksptb96KeWciDhnsPcDkGmaR2YRMFLMIqANzCLgFQO94mdVKWVWRETfr6v7C9Za59da59Va5w3wvgCa9DSPzCJgmJlFQBuYRcDrDHTxc3VEvPKzp8+OiKuGph2AbWYeAW1gFgFtYBYBr5O+1auUcklEnBQRe5RSlkfE5yLiv0fEZaWU34mIRyPizOFssq0+/OEPp5nf/u3fbqzX2u+75LbJJz7xicb6eeedl56xfv36IemlaxYvXtxYH6o/w1NPPbWx/slPfnJI7mcsM4+27pBDDkkz8+Y1/we9iRMnpmcsWrQozXz7299urP/oRz9Kz9gePfXUU2nm13/91xvrzz33XHrGDjvk/71n0qRJaWZ7ZxYNXDZrHnnkkfSMXmbR9ddf31i/7rrr0jO2R88++2ya+ehHP9pYnzx5cnrGfvvtl2bmzJnTWL/tttvSM7rOLBo+F198cZpZsmRJmsn+Tn300UfTM5YuXZpmtkfvete7GuvHHntsesb06dPTzMEHH9xYv+mmm9Iz7r///jQz3NLFT631rH5K7xjiXgAamUdAG5hFQBuYRUCvBvpWLwAAAABazuIHAAAAoKMsfgAAAAA6yuIHAAAAoKMsfgAAAAA6yuIHAAAAoKMsfgAAAAA6atxoNzCWvfe97x2R+7n22mvTzJe+9KUR6GT7dOutt47I/TzwwAMjcj90z5ve9KY0s9deezXWly9fnp5xyy23pJmbbropzTAwzz33XGP9xRdfTM+YMGFCmtl555177gm2VTZrbrjhhvSMhx56KM0sXLiw15bYwuzZs9PMhg0bGuvr169Pz9hhh/y/PWd/b8FoG4o508vfubvvvnuaefrppwfdy1CZOHFiY33OnDnpGb08/2fNmtVY7+Vx22OPPdLMEUcc0Vi/8MIL0zPawBU/AAAAAB1l8QMAAADQURY/AAAAAB1l8QMAAADQURY/AAAAAB1l8QMAAADQURY/AAAAAB1l8QMAAADQUeNGu4G22mWXXdLMgQceOOj7Wb9+fZr5xje+Mej7YeBeeOGFxnovf4YTJkxIMwcffHDPPcGW9tlnnzRz4oknNtYvvvji9IwHHnggzfTyfGBg1q5dO6h6RMRHPvKRoWoHBuS2225rrJdS0jMWLVo0VO10yowZMxrrvbwWeeyxx9LMfffd11g/6KCD0jM++clPphnYHkydOjXNTJs2Lc08/fTTQ9FO6rjjjkszxxxzTGN93bp16Rn/63/9rzSzZs2axnovM+/v//7v00xXuOIHAAAAoKMsfgAAAAA6yuIHAAAAoKMsfgAAAAA6yuIHAAAAoKMsfgAAAAA6yuIHAAAAoKPGjXYDbXX00Uenmf3333/Q9/Pwww+nma9//euDvh8GbuPGjYOqR0RMmDBhqNphOzNjxow0s2nTpjTzrW99q7H+0EMPpWc88sgjaYbh80//9E+N9be//e3pGVdeeWWaOf3003vuCbbVzjvv3FjvZZ7tsssuaeb555/vuaexIHvcIiImTpzYWF+2bNmQ9HLDDTc01vfbb7/0jAULFqSZs88+u9eWYMz62c9+lmbWrl07Ap305m1ve9ugz/jud787BJ1E3H777Y311atXp2f8l//yX9LMn/7pnzbWjz/++PSMm2++Oc0MN1f8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR40b7QbGslJKK85geB144IGN9UmTJg3J/Vx00UVDcg7dMnXq1DQzc+bMNLPDDs17/ieffDI945FHHkkzDJ8pU6Y01p9//vn0jNNPPz3NnHbaaY31q666Kj0D+nPSSSc11nv5O/XRRx9NM708H8aSF154Ic0sW7ZsBDrJ9fLnc/bZZ6eZD3zgA431Sy+9tOeeoK3WrVs3Ivcza9asNDNv3rw088ADD6SZK6+8sqeehtvjjz+eZv70T/80zbz//e9vrH/729/uuafR5IofAAAAgI5KFz+llAtKKatLKYu3uO3zpZQVpZTb+/45dXjbBLZ3ZhHQFuYR0AZmEdCrXq74uSgiTt7K7V+qtR7V98/3hrYtgNe5KMwioB0uCvMIGH0XhVkE9CBd/NRab4yIZ0agF4B+mUVAW5hHQBuYRUCvBvMZPx8tpdzZd4nhtP5CpZRzSikLSykLB3FfAP0xi4C2SOeRWQSMALMIeJWBLn6+FhEHRsRREbEyIr7YX7DWOr/WOq/Wmn9MOMC2MYuAtuhpHplFwDAzi4DXGdDip9a6qta6qda6OSL+LiKOG9q2AHJmEdAW5hHQBmYRsDUDWvyUUmZt8dszImJxf1mA4WIWAW1hHgFtYBYBWzMuC5RSLomIkyJij1LK8oj4XEScVEo5KiJqRCyNiN8dxh5bq9baijMYXu9+97sb60P1Z/jiiy8OyTldtb3Oor333jvNzJo1K81MmTKlsf7lL3+5554YHZMmTWqs77rrrukZCxYsSDMXXXRRry1tt7bXeZQ544wz0sz73//+xvqHPvSh9Ixjjz02zSxatCjNMDx23333NPO3f/u3aeb73//+ULTTaWbR6Hrb296WZm699dbG+vPPPz9U7TSaPn16mrn//vvTzH333TcU7bTGf/gP/yHNrFixYgQ6GX7p4qfWetZWbj5/GHoB6JdZBLSFeQS0gVkE9GowP9ULAAAAgBaz+AEAAADoKIsfAAAAgI6y+AEAAADoKIsfAAAAgI6y+AEAAADoKIsfAAAAgI4aN9oNQNu9853vHPQZjz/+eJq56KKLBn0/dM8uu+ySZmbMmJFm3v72tzfW/+t//a/pGZ///OfTTK01zTAwTzzxRGP9gAMOSM9YtWpVmnnsscd67gm2dOKJJ6aZD33oQ43122+/PT3ji1/8YppZtGhRmmF4vPGNb0wz69atSzMrV64cinZgQObPn59mnn/++TQzblzzt9vXXXddzz0NRi/fizz77LMj0MnIOfzww9PMmjVr0syKFSuGop1R54ofAAAAgI6y+AEAAADoKIsfAAAAgI6y+AEAAADoKIsfAAAAgI6y+AEAAADoKIsfAAAAgI4aN9oNtNXSpUvTzJNPPplm9txzzyHohuHypS99Kc3Mnj27sV5rTc+49tpr08zatWvTDNufZ599Ns28/e1vTzMPPvhgY/1f/uVf0jN6+XedgTnllFPSzHvf+97G+qRJk9Izli1blmYeeuihNANb8/GPfzzNbNiwobH+4osvpmdcfPHFPffEtjnssMPSzAc/+MHG+qxZs9IzFi1alGZuvvnmNAMD9elPf7qx/ra3vS0945BDDhmqdoZdL68nx5rjjjuusb7PPvukZ9x///1p5s477+y5pzZzxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHTUuNFuoK1WrFiRZi6++OI08/GPf7yxPnv27PSM973vfWnm8ssvTzPbm9122y3NnHrqqYO+n1tvvTXNfOITnxj0/bB9+ulPf5pm/uqv/irNTJw4sbF+xBFH9NwT2+bII49MM6ecckqaOfjggxvrJ510UnrGCSeckGZgoHr593jp0qWN9QMOOGBIennDG97QWL/33nuH5H7Gknnz5qWZX/u1X0sz2WPby2urQw89NM3AcLrlllsa6x/4wAeG5H6+8IUvDKoeEbFu3boh6aUtDjzwwDTTy7yaPHlyY/3888/vuaftgSt+AAAAADrK4gcAAACgoyx+AAAAADrK4gcAAACgoyx+AAAAADrK4gcAAACgoyx+AAAAADrK4gcAAACgo8aNdgNj2ZNPPplmSimN9UmTJqVn/Pmf/3maufzyy9PMWDJ+/Pg0c9xxxzXW//Ef/zE9Y9ddd00z2Z/hpZdemp7xzDPPpBkYqLvuuivNfOxjH2us11rTM77zne+kmQsuuKCxftVVV6VnbN68Oc20yfHHH99Yv/nmm9MzjjjiiDSzdu3annvqz09+8pNBnwH9ueaaa9LMwQcf3Fhfv359esaSJUvSzCWXXNJYv/DCC9Mzli1blmZGyi677JJmfvVXf7Wx3svrlS996UtpZtWqVWkmc8899wz6DBiM6667rrF+xx13pGc8+uijaeaBBx5orJ988snpGd/+9rfTzFDIvueJiDj66KMHnTn//PPTM37zN38zzdx7771phv8vveKnlLJfKeVHpZR7Sil3l1J+v+/26aWUH5ZSHuj7ddrwtwtsr8wioA3MIqANzCJgW/TyVq+NEfGJWuuhEXF8RPxeKeXQiPhURFxXaz04Iq7r+z3AcDGLgDYwi4A2MIuAnqWLn1rrylrrbX1fr42IJRGxT0ScFhEL+mILIuL04WoSwCwC2sAsAtrALAK2xTZ9xk8pZW5EHB0RP42ImbXWlX2lJyJiZj//m3Mi4pyBtwjwamYR0AZmEdAGZhGQ6fmnepVSJkfE5RHxB7XWNVvW6sufCrrVTwattc6vtc6rtc4bVKcAYRYB7WAWAW1gFgG96GnxU0oZHy8PlG/UWv+h7+ZVpZRZffVZEbF6eFoEeJlZBLSBWQS0gVkE9KqXn+pVIuL8iFhSa/2fW5Sujoiz+74+OyLyn9ELMEBmEdAGZhHQBmYRsC3Ky1cANgRKeUtE/HNE3BURm/tu/ky8/B7SyyJidkQ8GhFn1lqfSc5qvrMxZu7cuWlmyZIljfUJEyYMSS/3339/Y/2b3/xmesbXvva1Iekl8+EPfzjNHHnkkWnmgx/84FC0k1qwYEFj/SMf+Uh6xvr164eqnbZYNNKXBptFg3PppZc21s8888z0jGeffTbNXH/99Y31K6+8Mj1j4cKFg+5l3Lj8I+zmzJmTZubNy/81//mf//nG+hve8Ib0jJdeeinNfO9732usf/WrX03P6CCzaIzZc889G+vf+c530jOy51xExNKlSxvrV1xxRXpGNs8iIp544onG+ubNmxvrEb39/znxxBPTzCGHHNJY7+V168qVK9PMxRdf3Fj/xje+kZ7RQWZRxxx//PFp5gtf+EKaOeCAAxrr1157bXpGL5lsFu24447pGYcddliaOfzww9PMTjvt1FjfsGFDesY//dM/pZmvf/3raWY71O8sSl8Z11p/HBGln/I7BtMVQK/MIqANzCKgDcwiYFv0/OHOAAAAAIwtFj8AAAAAHWXxAwAAANBRFj8AAAAAHWXxAwAAANBRFj8AAAAAHWXxAwAAANBRpdY6cndWysjdWUu8//3vb6xffPHF6RkTJkxIM6WUxvpI/jlnsl4jRq7f888/P8382Z/9WWP9scceG6p2xpJFtdZ5o93EQG2Ps+iggw5qrH/lK19JzzjiiCPSzOzZsxvr3/nOd9IzHn/88TSzfv36xvr48ePTM6ZNm5Zm9t133zTz0ksvNdbvvvvu9IyPfexjaYatMos6Zt68/I/zM5/5TJp573vf21hftGhRekYvs+jpp59urG/cuDE9Y+bMmWnm4IMPTjObNm1qrN9yyy3pGb/927+dZtgqs2g7dPzxx6eZm2++ubF+/fXXp2e8+OKLaWbNmjWN9V5eF+2xxx5pZtKkSWlm5cqVjfVeXgvOnz8/zbBV/c4iV/wAAAAAdJTFDwAAAEBHWfwAAAAAdJTFDwAAAEBHWfwAAAAAdJTFDwAAAEBHWfwAAAAAdJTFDwAAAEBHlVrryN1ZKSN3Z2PEmWeemWY+97nPpZmf//mfb6yP5J9zppSSZjZu3JhmHnjggcb63/zN36Rn9JLZsGFDmtkOLaq1zhvtJgbKLHq9Aw88MM18+MMfTjMnnHBCY32nnXZKz+gls8suuzTWe5kh69atSzO/8Au/kGb+03/6T431Sy65JD1jxYoVaYatMou2QzNmzEgzv//7v99Ynzcv/9dm7ty5aSabV+vXr0/P6CVz2GGHpZns9eKFF16YnvHYY4+lGbbKLGJAstcQERHveMc70sysWbMa6zvvvHN6xgsvvJBmjjrqqDRz1llnNdZ7eV3EgPU7i1zxAwAAANBRFj8AAAAAHWXxAwAAANBRFj8AAAAAHWXxAwAAANBRFj8AAAAAHWXxAwAAANBRpdY6cndWysjd2XbmhBNOaKy/+93vTs946aWX0sw73vGOxvr999+fntGLSy+9NM1ce+21Q3JfDMiiWuu80W5ioMyigRk3blyaede73tVYv+aaa9Iz/uqv/irNfOxjH2usf/7zn0/PWLlyZZq544470szNN9+cZhg2ZhHD5rOf/Wya+W//7b811v/yL/8yPWP16tVp5pZbbkkzN9xwQ5ph2JhFDJu3vOUtaebHP/5xY/0//+f/nJ6xZs2aNHP99denmaH6fpAB6XcWueIHAAAAoKMsfgAAAAA6yuIHAAAAoKMsfgAAAAA6yuIHAAAAoKMsfgAAAAA6yuIHAAAAoKMsfgAAAAA6qtRaR+7OShm5OwOG06Ja67zRbmKgzCLoDLMIaAOzCGiDfmdResVPKWW/UsqPSin3lFLuLqX8ft/tny+lrCil3N73z6lD3TXAK8wioA3MIqANzCJgW4zrIbMxIj5Ra72tlDIlIhaVUn7YV/tSrfUvh689gH9jFgFtYBYBbWAWAT1LFz+11pURsbLv67WllCURsc9wNwawJbMIaAOzCGgDswjYFtv04c6llLkRcXRE/LTvpo+WUu4spVxQSpk2xL0BbJVZBLSBWQS0gVkEZHpe/JRSJkfE5RHxB7XWNRHxtYg4MCKOipe3zV/s5393TillYSll4RD0C2znzCKgDcwioA3MIqAXPf1Ur1LK+Ij4x4j4Qa31f26lPjci/rHWenYqeSgAACAASURBVHhyjk+Mh24YlZ9eYRYBr2EWAW1gFgFtMKif6lUi4vyIWLLlQCmlzNoidkZELB5slwD9MYuANjCLgDYwi4Bt0ctP9XpzRPxGRNxVSrm977bPRMRZpZSjIqJGxNKI+N1h6RDgZWYR0AZmEdAGZhHQs57e6jVkd+YyQuiKUbmkeaiYRdAZZhHQBmYR0AYDf6sXAAAAAGOTxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR40b4ft7KiIe3eL3e/TdNlaMpX71OnzGUr/D1eucYThzJJlFI0evw2cs9WsWbd1rZ1GEP9fhMpZ6jRhb/erVLBpteh0+Y6lfvTbMolJrHYb7600pZWGtdd6oNbCNxlK/eh0+Y6nfsdTraBprj9NY6levw2cs9TuWeh1tY+mx0uvwGUv96rWbxtJjpdfhM5b61Wszb/UCAAAA6CiLHwAAAICOGu3Fz/xRvv9tNZb61evwGUv9jqVeR9NYe5zGUr96HT5jqd+x1OtoG0uPlV6Hz1jqV6/dNJYeK70On7HUr14bjOpn/AAAAAAwfEb7ih8AAAAAhonFDwAAAEBHjdrip5RycinlvlLKg6WUT41WH70opSwtpdxVSrm9lLJwtPt5rVLKBaWU1aWUxVvcNr2U8sNSygN9v04bzR5f0U+vny+lrOh7fG8vpZw6mj2+opSyXynlR6WUe0opd5dSfr/v9tY9tg29tvKxbROzaOiYRcPDLNo+mEVDxywaHmNpFkWYRwM1lmZRRLvnkVk0PMyiAfYxGp/xU0rZMSLuj4h3RcTyiLg1Is6qtd4z4s30oJSyNCLm1VqfGu1etqaU8taIWBcRX6+1Ht53219ExDO11v/eN7Sn1Vr/aDT77Otra71+PiLW1Vr/cjR7e61SyqyImFVrva2UMiUiFkXE6RHxW9Gyx7ah1zOjhY9tW5hFQ8ssGh5mUfeZRUPLLBoeY2kWRZhHAzHWZlFEu+eRWTQ8zKKBGa0rfo6LiAdrrQ/XWtdHxLci4rRR6mXMq7XeGBHPvObm0yJiQd/XC+Llf7lGXT+9tlKtdWWt9ba+r9dGxJKI2Cda+Ng29Eozs2gImUXDwyzaLphFQ8gsGh5jaRZFmEcDZBYNIbNoeJhFAzNai599ImLZFr9fHu0exDUi/k8pZVEp5ZzRbqZHM2utK/u+fiIiZo5mMz34aCnlzr7LDFtxWd6WSilzI+LoiPhptPyxfU2vES1/bEeZWTT8Wv182YpWP1/Mos4yi4Zfq58vW9Hq58tYmkUR5tE2GGuzKGLszaPWP19eo9XPFbOodz7cuTdvqbUeExGnRMTv9V0KN2bUl9/PN/Lv6evd1yLiwIg4KiJWRsQXR7edVyulTI6IyyPiD2qta7aste2x3UqvrX5s2WZm0fBq9fPFLKJFzKLh1erny1iaRRHm0XZgzM6jNj5fXqPVzxWzaNuM1uJnRUTst8Xv9+27rZVqrSv6fl0dEVfEy5dBtt2qvvcTvvK+wtWj3E+/aq2raq2baq2bI+LvokWPbyllfLz8BP1GrfUf+m5u5WO7tV7b/Ni2hFk0/Fr5fNmaNj9fzKLOM4uGXyufL1vT5ufLWJpFEebRAIypWRQxJudRa58vr9Xm54pZtO1Ga/Fza0QcXErZv5QyISI+GBFXj1IvjUopk/o+hClKKZMi4pciYnHz/6oVro6Is/u+PjsirhrFXhq98gTtc0a05PEtpZSIOD8iltRa/+cWpdY9tv312tbHtkXMouHXuudLf9r6fDGLtgtm0fBr3fOlP219voylWRRhHg3QmJlFEWN2HrXy+bI1bX2umEUD7KOOwk/1iogoL/+4si9HxI4RcUGt9c9HpZFEKeWAeHl7HBExLiK+2bZeSymXRMRJEbFHRKyKiM9FxJURcVlEzI6IRyPizFrrqH9gVz+9nhQvX+JWI2JpRPzuFu/PHDWllLdExD9HxF0Rsbnv5s/Ey+/JbNVj29DrWdHCx7ZNzKKhYxYND7No+2AWDR2zaHiMpVkUYR4N1FiZRRHtn0dm0fAwiwbYx2gtfgAAAAAYXj7cGQAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjxo3knZVS6kjeHzBsnqq17jnaTQyUWQSdYRYBbWAWAW3Q7ywa1BU/pZSTSyn3lVIeLKV8ajBnAWPKo6PdwJbMIthutWoWRZhHsJ0yi4A26HcWDXjxU0rZMSLOi4hTIuLQiDirlHLoQM8DGAizCGgL8whoA7MIeK3BXPFzXEQ8WGt9uNa6PiK+FRGnDU1bAD0zi4C2MI+ANjCLgFcZzOJnn4hYtsXvl/fdBjCSzCKgLcwjoA3MIuBVhv3DnUsp50TEOcN9PwBNzCKgDcwioA3MIti+DGbxsyIi9tvi9/v23fYqtdb5ETE/wifGA8PCLALaIp1HZhEwAswi4FUG81avWyPi4FLK/qWUCRHxwYi4emjaAuiZWQS0hXkEtIFZBLzKgK/4qbVuLKV8NCJ+EBE7RsQFtda7h6wzgB6YRUBbmEdAG5hFwGuVWkfuyj6XEUJnLKq1zhvtJgbKLILOMIuANjCLgDbodxYN5q1eAAAAALSYxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHTUuNFuAAC21fjx4xvrGzZsGKFOgK6aM2dOmjnxxBMb608//XR6xjXXXNNzT8D2Z+rUqWnm2Wefbaz/j//xP9IzPvOZz/TcE2OPK34AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOsriBwAAAKCjLH4AAAAAOmrcaDdAe8yZMyfN/OIv/mJj/dhjj03PqLWmmeuvvz7NXHnllWkGaJcpU6akmV/5lV9JMyeccEJjfdddd03PuO2229LMl7/85TQDdNM73/nONPMf/+N/bKxPmjQpPePwww9PM+eee26aAbZf2fdXxxxzTHrGhAkT0sz69et77ol2GdTip5SyNCLWRsSmiNhYa503FE0BbCvzCGgDswhoA7MI2NJQXPHzi7XWp4bgHIDBMo+ANjCLgDYwi4CI8Bk/AAAAAJ012MVPjYj/U0pZVEo5ZygaAhgg8whoA7MIaAOzCPg3g32r11tqrStKKTMi4oellHtrrTduGegbNIYNMNwa55FZBIwQswhoA7MI+DeDuuKn1rqi79fVEXFFRBy3lcz8Wus8HygGDKdsHplFwEgwi4A2MIuALQ148VNKmVRKmfLK1xHxSxGxeKgaA+iVeQS0gVkEtIFZBLzWYN7qNTMiriilvHLON2ut3x+SrgC2jXkEtIFZBLSBWQS8yoAXP7XWhyPiyCHshWH0vve9L8286U1vSjPHHfe6d/O9ymGHHZaeseuuu6aZefPyq05nzJjRWJ8/f356Bt1gHo0d73//+9PMu971rjRzxBFHNNZ322239Iz9998/zSxbtizNXH755WmG7YNZ1C377rtvmtlrr70a67Nnz07POOWUU9LMueeem2bgFWZRt6xbty7NnHfeeY31j370o+kZn/3sZ4ckQzv5ce4AAAAAHWXxAwAAANBRFj8AAAAAHWXxAwAAANBRFj8AAAAAHWXxAwAAANBRFj8AAAAAHWXxAwAAANBR40a7AXLTpk1LM7/3e7/XWD/hhBPSMw466KA0M3Xq1Mb69OnT0zM2b96cZubMmZNmfvVXf7Wxvuuuu6ZnnHvuuWkG6N1JJ53UWD/xxBPTM4455pg0k82iiRMnpmdMmjQpzRx22GFp5vLLL08zQLtMmTIlzey+++5pZsOGDY31Ukp6RjbPInp7jfbggw+mGWDs6eV7p3Xr1g36fnqZRYxdrvgBAAAA6CiLHwAAAICOsvgBAAAA6CiLHwAAAICOsvgBAAAA6CiLHwAAAICOsvgBAAAA6Khxo90AuWOPPTbNzJs3r7F+xBFHpGds2LBh0JlSSnrGkiVL0szq1avTzDPPPNNYnzx5cnrGG97whjRz7733phngZTvvvHNjfd99903PGDcu/6tp4sSJjfXdd989PePmm29OM5MmTUoz2RwxQ6B9Nm/enGaeeuqpNFNrbaz38rro//7f/5tm5s6dm2YefPDBNAPQn15eo2WvvyIiXnrppaFohyHmih8AAACAjrL4AQAAAOgoix8AAACAjrL4AQAAAOgoix8AAACAjrL4AQAAAOgoix8AAACAjrL4AQAAAOiocaPdALmNGzemmbVr1zbW77vvvvSMTZs2DTpzwQUXpGd84xvfSDNLlixJM3fccUdj/dlnn03PePrpp9MM0LvHH3+8sf7www+nZ+y2225pZsKECY31hQsXpmfccsstaebmm29OM/fee2+aAdrlZz/7WZpZvXp1mtlpp50a67XW9Iwf/vCHaSZ7nQd014477phmZs2aNej7Oe2009LM7rvvnmay14KMDlf8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR40b7QbI3XDDDWlm//33b6zPnj07PWPz5s1p5vHHH2+sP/jgg+kZP/7xj9PMpk2b0gzQPnfccUdjfeHChekZu+22W5p58sknG+sPPPBAesa1116bZq677ro0A3TTSy+9NOgzVq5cmWYeffTRNLNu3bpB9wKMTZMmTUozv/mbvzkCnTCWpVf8lFIuKKWsLqUs3uK26aWUH5ZSHuj7ddrwtglgHgHtYBYBbWAWAb3q5a1eF0XEya+57VMRcV2t9eCIuK7v9wDD7aIwj4DRd1GYRcDouyjMIqAH6eKn1npjRDzzmptPi4gFfV8viIjTh7gvgNcxj4A2MIuANjCLgF4N9DN+ZtZaX3nT8hMRMbO/YCnlnIg4Z4D3A5DpaR6ZRcAwM4uANjCLgNcZ9Ic711prKaU21OdHxPyIiKYcwGA1zSOzCBgpZhHQBmYR8IqB/jj3VaWUWRERfb+uHrqWALaJeQS0gVkEtIFZBLzOQBc/V0fE2X1fnx0RVw1NOwDbzDwC2sAsAtrALAJeJ32rVynlkog4KSL2KKUsj4jPRcR/j4jLSim/ExGPRsSZw9kkuQsvvLCx/uY3vzk9Y8qUKWnmueeea6zffffd6RmbNm1KM2PJnnvumWZ6+f+8cePGxnovfz4rVqxIM2OZeTT2XX/99Wlm2rT8J8/OnNnvR8tFRMTSpUvTM5YtW5ZmXnzxxTTD9scs2j787Gc/SzOrVzdfTDF+/Pj0jOzv/4iI3XffPc2w/TGLgF6li59a61n9lN4xxL0ANDKPgDYwi4A2MIuAXg30rV4AAAAAtJzFDwAAAEBHWfwAAAAAdJTFDwAAAEBHWfwAAAAAdJTFDwAAAEBHWfwAAAAAdNS40W6AkbF06dI0c9RRR6WZmTNnNtZvvvnmXlsatMMOO6yxPn78+PSMHXbId5877bRTY33OnDnpGb30MnXq1Mb6X//1X6dnQNs9/PDDaeapp55KM7Nnz26sT5o0KT3jgAMOSDN77713msk89thjaebBBx9MMy+99NKgewF6t3nz5jSTzZrsNURExPTp04ckA3TTe97zntFugQ5wxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHTUuNFugJGxYcOGNDN9+vQ088wzzzTWd9555/SMY445Js2ccMIJaWa33XZrrN9///3pGQsWLEgz++67b2N9hx3y/emPf/zjNAO8bObMmWlmxowZjfWhmnkTJ05MM//6r/+aZjJ33333oM8AhtbcuXPTzKxZsxrrmzZtSs/o5bVTKSXNAN103HHHjXYLdIArfgAAAAA6yuIHAAAAoKMsfgAAAAA6yuIHAAAAoKMsfgAAAAA6yuIHAAAAoKMsfgAAAAA6atxoN8DI2GOPPdJMrTXNfPe7322sH3/88ekZN910U5r5yEc+kmYeeuihxvq3vvWt9IxeLF++fFD1iIjDDz88zSxevLix/ta3vjU948Ybb0wzMJoOO+ywNLPPPvukmQ984AON9a985SvpGR//+MfTzF/8xV+kmey5e80116RnAO1zwAEHpJnddtutsX7vvfemZ7znPe9JMyeddFKaAbqplDLozA475Nd7bN68ueeeGHtc8QMAAADQURY/AAAAAB1l8QMAAADQURY/AAAAAB1l8QMAAADQURY/AAAAAB1l8QMAAADQURY/AAAAAB01brQbYGSMG5f/Udda08zpp5/eWN9rr73SM0499dQ0c9NNN6WZv/mbv0kzbbF48eI0s88++zTWb7zxxqFqB0bN3Llz08zUqVPTzPnnn99Yf/bZZ9MzPv3pT6eZq666Ks30Mq+Adsn+zo2ImDFjRpq5//77G+uPP/54esZnP/vZNHPuueemGaCbevkeLcts3rw5PeNv//Zv08wTTzyRZmin9IqfUsoFpZTVpZTFW9z2+VLKilLK7X3/5N/JAwyCWQS0hXkEtIFZBPSql7d6XRQRJ2/l9i/VWo/q++d7Q9sWwOtcFGYR0A4XhXkEjL6LwiwCepAufmqtN0bEMyPQC0C/zCKgLcwjoA3MIqBXg/lw54+WUu7su8RwWn+hUso5pZSFpZSFg7gvgP6YRUBbpPPILAJGgFkEvMpAFz9fi4gDI+KoiFgZEV/sL1hrnV9rnVdrnTfA+wLoj1kEtEVP88gsAoaZWQS8zoAWP7XWVbXWTbXWzRHxdxFx3NC2BZAzi4C2MI+ANjCLgK0Z0OKnlDJri9+eERH5z6oGGGJmEdAW5hHQBmYRsDXjskAp5ZKIOCki9iilLI+Iz0XESaWUoyKiRsTSiPjdYewRwCwCWsM8AtrALAJ6lS5+aq1nbeXm84ehF/oxZ86cNDNtWr+faRsRETvuuGN6xurVq9PMuHHN/8pMmTIlPeORRx5JMxdeeGGa6Zrx48ePdgutZhaNDbvsssug6hG9zYjnn3++sT59+vT0jCeeeCLN3HTTTWmG7Y95NPb18tpqxYoVaSb7u3vTpk3pGU899VSaefHFF9MM2x+ziKG0YcOGNLN58+YR6IThMJif6gUAAABAi1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHSUxQ8AAABAR1n8AAAAAHTUuNFuYHs3ceLENHPSSSelmQkTJjTW77jjjvSMJUuWpJmZM2c21p9//vn0jAsvvDDNbI+WLl062i3AoB166KGN9fXr16dnLF68OM3ssccejfWDDjooPcMsgu3XunXr0kwvfy/vtddejfXsdVNExNe+9rU0AwCD4YofAAAAgI6y+AEAAADoKIsfAAAAgI6y+AEAAADoKIsfAAAAgI6y+AEAAADoKIsfAAAAgI4aN9oNbO/e+MY3ppkFCxaMQCe9Wb58eWN948aNI9QJ0EYLFy4cVD0i4vDDD08z+++/f2P92GOPTc8444wz0swVV1yRZoCx584770wzvbxGmz59emP9537u59Iz/uRP/iTN/Nmf/VmaAYD+uOIHAAAAoKMsfgAAAAA6yuIHAAAAoKMsfgAAAAA6yuIHAAAAoKMsfgAAAAA6yuIHAAAAoKMsfgAAAAA6atxoN7C9e+6554bknPPPP7+xftlll6Vn/OAHPxh0H8uXL08ze++9d5p5/PHHB90LMPLe/e53N9Z32CH/7w133313mtm8eXNjfcOGDekZxxxzTJpZsmRJmrn33nvTDNAukydPTjMTJkxIM9OnT2+s77rrrukZRx55ZJqZOXNmmlm1alWaAdqllznTy/MfMq74AQAAAOgoix8AAACAjrL4AQAAAOgoix8AAACAjrL4AQAAAOgoix8AAACAjrL4AQAAAOgoix8AAACAjhqXBUop+0XE1yNiZkTUiJhfa/1KKWV6RFwaEXMjYmlEnFlrfXb4Wu2mxYsXp5mvfvWraWb27NmN9UMPPTQ947rrrkszmzZtaqzvv//+6RmHHHJImtlll13SzPr16xvrCxYsSM8YP358mtmwYUOaYfiZRWPD1KlTG+sTJ05Mz8jmWUTE3LlzG+vTpk1LzzjwwAPTTDbzIiL+/u//vrG+fPny9AzGDrOoG2bOnJlm5syZk2ZmzJjRWO/ldUZ2RkRvr51WrVqVZugOs6gb9txzzzTzvve9bwQ6oet6ueJnY0R8otZ6aEQcHxG/V0o5NCI+FRHX1VoPjojr+n4PMFzMIqANzCKgDcwioGfp4qfWurLWelvf12sjYklE7BMRp0XEK5dULIiI04erSQCzCGgDswhoA7MI2Bbb9Bk/pZS5EXF0RPw0ImbWWlf2lZ6Ily8zBBh2ZhHQBmYR0AZmEZBJP+PnFaWUyRH/r737DfWrvu8A/v7QRI2tQkWQ6DKrYyIyYzWxT1pGJW5sfdIVVOaDomxowVVa2IMNn0wYhTI6t0cZtCh04DYKutU8CQsi+/NAMYrG1OAsM2FXnFGqtEnQufjdg/uz3IV7f+d3b+4vv3NOXi+Q/O45n3t+H77x+87lw/mdmyeSfLu19vOq+uW51lqrqrbG992f5P6zbRQgkUVAP8gioA9kETCLme74qaqtWQ6Ux1trT04Ov11V2yfntyc5vtr3tta+31rb3VrbvRkNA+cvWQT0gSwC+kAWAbPqHPzU8tj40SRHWmuPrDj1VJJ7Jq/vSfLjzW8PYJksAvpAFgF9IIuA9Zjlo15fTPL1JK9U1UuTYw8l+W6SH1XVHyY5luSu+bQIkEQWAf0gi4A+kEXAzDoHP621f09Sa5zes7ntsJp9+/Z11lx99dVTz1933XWd17jvvvs6a06fPj31/JVXXtl5jZ07d3bWbNu2rbPm0KFDU8/fdtttndd45plnOmvoB1k0DK+//vrU87fcckvnNXbt2tVZ05Vp11xzTec1rr322s6arsxLkv379089v7S01HkNhkMWjcMsP2dcdtllnTUXXnjh1PMffvhh5zUuuOCCzppLLrmks4bziywah/fee6+z5umnn+6s2bPHXznTreu3egEAAAAwHAY/AAAAACNl8AMAAAAwUgY/AAAAACNl8AMAAAAwUgY/AAAAACNl8AMAAAAwUgY/AAAAACO1ZdEN0O3AgQNnXbN3797Oa9x+++2dNdu2bZt6fseOHZ3XuPzyyztrlpaWOmtOnz499fzzzz/feQ1gcx08eHDq+e3bt3de48Ybb+ysueGGG6aev+mmmzqvcdFFF3XWnDx5srPmxIkTnTVAvxw9erSzZsuW7h+Tu34W2bp1a+c1Tp061Vlz7NixzhpgeGbZ/6+99lpnzZ49e866l3379p31Negvd/wAAAAAjJTBDwAAAMBIGfwAAAAAjJTBDwAAAMBIGfwAAAAAjJTBDwAAAMBIGfwAAAAAjJTBDwAAAMBIbVl0A5wbe/fu7ax58MEHO2tuvvnmqecvvfTSzmscP368s+bWW2/trLn++uunnj9x4kTnNYBza9++fZ01u3bt6qzZsmX6P18nT57svMaRI0c6a+68887OGmB4ZvkZ4Y033uiseeedd6ae/+ijjzqv8fLLL3fWHD58uLMGGKcDBw501jzwwANn/T7PPffcWV+D/nLHDwAAAMBIGfwAAAAAjJTBDwAAAMBIGfwAAAAAjJTBDwAAAMBIGfwAAAAAjJTBDwAAAMBIVWvt3L1Z1bl7M9bt4osv7qw5derU1PP33ntv5zXef//9zpr9+/d31nzwwQedNczNC6213YtuYqNk0fjdcccdnTVLS0udNc8+++xmtMP8yCJ6befOnZ01hw4dOgedMGeyCOiDNbPIHT8AAAAAI2XwAwAAADBSBj8AAAAAI2XwAwAAADBSBj8AAAAAI2XwAwAAADBSBj8AAAAAI2XwAwAAADBS1Vo7d29Wde7eDJinF1pruxfdxEbJIhgNWQT0gSwC+mDNLOq846eqdlTVM1X1alX9pKq+NTn+cFW9WVUvTf77ymZ3DfAJWQT0gSwC+kAWAeuxZYaa/03yx621F6vqkiQvVNWBybm/aq19b37tAfySLAL6QBYBfSCLgJl1Dn5aa28leWvy+hdVdSTJVfNuDGAlWQT0gSwC+kAWAeuxroc7V9Xnktyc5LnJoW9W1aGqeqyqPrvG99xfVQer6uBZdQowIYuAPpBFQB/IIqDLzA93rqrPJPmXJN9prT1ZVVckeTdJS/LnSba31v6g4xoeHAbjsLCHGMoiYAVZBPSBLAL6YOMPd06Sqtqa5Ikkj7fWnkyS1trbrbXTrbWPk/wgyRc2q1uA1cgioA9kEdAHsgiY1Sy/1auSPJrkSGvtkRXHt68o+1qSw5vfHsAyWQT0gSwC+kAWAesxy2/1NYZbSgAABexJREFU+mKSryd5papemhx7KMndVfX5LN9GeDTJN+bSIcAyWQT0gSwC+kAWATOb+Rk/m/JmPj8KY7Gwz7JvBlkEoyGLgD6QRUAfnN0zfgAAAAAYHoMfAAAAgJEy+AEAAAAYKYMfAAAAgJEy+AEAAAAYKYMfAAAAgJEy+AEAAAAYKYMfAAAAgJEy+AEAAAAYKYMfAAAAgJEy+AEAAAAYKYMfAAAAgJEy+AEAAAAYKYMfAAAAgJEy+AEAAAAYKYMfAAAAgJHaco7f790kx1Z8ffnk2FAMqV+9zs+Q+p1Xr1fP4Zrnkiw6d/Q6P0PqVxat7swsSvy9zsuQek2G1a9eZdGi6XV+htSvXqdkUbXW5vB+s6mqg6213QtrYJ2G1K9e52dI/Q6p10Ua2joNqV+9zs+Q+h1Sr4s2pLXS6/wMqV+9jtOQ1kqv8zOkfvU6nY96AQAAAIyUwQ8AAADASC168PP9Bb//eg2pX73Oz5D6HVKvizS0dRpSv3qdnyH1O6ReF21Ia6XX+RlSv3odpyGtlV7nZ0j96nWKhT7jBwAAAID5WfQdPwAAAADMycIGP1X1O1X1WlX9tKr+dFF9zKKqjlbVK1X1UlUdXHQ/Z6qqx6rqeFUdXnHssqo6UFWvT/787CJ7/MQavT5cVW9O1velqvrKInv8RFXtqKpnqurVqvpJVX1rcrx3azul116ubZ/Ios0ji+ZDFp0fZNHmkUXzMaQsSuTRRg0pi5J+55Esmg9ZtME+FvFRr6r6VJL/SPJbSZaSPJ/k7tbaq+e8mRlU1dEku1tr7y66l9VU1W8mOZHkb1trvzE59hdJftZa++4ktD/bWvuTRfY56Wu1Xh9OcqK19r1F9namqtqeZHtr7cWquiTJC0l+L8m96dnaTun1rvRwbftCFm0uWTQfsmj8ZNHmkkXzMaQsSuTRRgwti5J+55Esmg9ZtDGLuuPnC0l+2lr7z9ba/yT5hyRfXVAvg9da+9ckPzvj8FeT/HDy+odZ/p9r4dbotZdaa2+11l6cvP5FkiNJrkoP13ZKr0wnizaRLJoPWXRekEWbSBbNx5CyKJFHGySLNpEsmg9ZtDGLGvxcleS/Vny9lH4HcUvyz1X1QlXdv+hmZnRFa+2tyev/TnLFIpuZwTer6tDkNsNe3Ja3UlV9LsnNSZ5Lz9f2jF6Tnq/tgsmi+ev1fllFr/eLLBotWTR/vd4vq+j1fhlSFiXyaB2GlkXJ8PKo9/vlDL3eK7Jodh7uPJsvtdZuSfK7Sf5ocivcYLTlz/P1+de3/U2SX0vy+SRvJfnLxbbz/1XVZ5I8keTbrbWfrzzXt7Vdpddery3rJovmq9f7RRbRI7Jovnq9X4aURYk8Og8MNo/6uF/O0Ou9IovWZ1GDnzeT7Fjx9a9MjvVSa+3NyZ/Hk/xjlm+D7Lu3J58n/ORzhccX3M+aWmtvt9ZOt9Y+TvKD9Gh9q2prljfo4621JyeHe7m2q/Xa57XtCVk0f73cL6vp836RRaMni+avl/tlNX3eL0PKokQebcCgsigZZB71dr+cqc97RRat36IGP88n+fWquqaqLkjy+0meWlAvU1XVpycPYUpVfTrJbyc5PP27euGpJPdMXt+T5McL7GWqTzboxNfSk/WtqkryaJIjrbVHVpzq3dqu1Wtf17ZHZNH89W6/rKWv+0UWnRdk0fz1br+spa/7ZUhZlMijDRpMFiWDzaNe7pfV9HWvyKIN9tEW8Fu9kqSWf13ZXyf5VJLHWmvfWUgjHarq2ixPj5NkS5K/61uvVfX3Sb6c5PIkbyf5syT/lORHSX41ybEkd7XWFv7ArjV6/XKWb3FrSY4m+caKz2cuTFV9Kcm/JXklyceTww9l+TOZvVrbKb3enR6ubZ/Ios0ji+ZDFp0fZNHmkUXzMaQsSuTRRg0li5L+55Esmg9ZtME+FjX4AQAAAGC+PNwZAAAAYKQMfgAAAABGyuAHAAAAYKQMfgAAAABGyuAHAAAAYKQMfgAAAABGyuAHAAAAYKQMfgAAAABG6v8A14co9NiCMB4AAAAASUVORK5CYII=\n",
"text/plain": [
"<Figure size 1440x720 with 8 Axes>"
]
},
"metadata": {
"tags": [],
"needs_background": "light"
}
}
]
}
]
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment